Datasets:
kilian-group
/
arxiv-classifier

Dataset card Data Studio Files Community
1
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/0111319
1
0111
2001-11-16T01:21:42
Magnetospheric Structure and Non-Thermal Emission of AXPs and SGRs
[ "astro-ph" ]
In the framework of the magnetar model for the Soft Gamma Repeaters and Anomalous X-ray Pulsars, we consider the structure of neutron star magnetospheres threaded by large-scale electrical currents. We construct self-similar, force-free equilibria under the assumption of axisymmetry and a power law dependence of magnetic field on radius, ${\bf B} \propto r^{-(2+p)}$. A large-scale twist of the field lines softens the radial dependence to $p < 1$, thereby accelerating the spindown torque with respect to a vacuum dipole. A magnetosphere with a strong twist ($B_\phi/B_\theta = O(1)$ at the equator) has an optical depth $\sim 1$ to resonant cyclotron scattering, independent of frequency (radius), surface magnetic field strength, or the charge/mass ratio of the scattering charge. We investigate the effects of the resonant Compton scattering by the charge carriers (both electrons and ions) on the emergent X-ray spectra and pulse profiles.
astro-ph
astro-ph
Neutron Stars in Supernova Remnants ASP Conference Series, Vol. 9999, 2002 P. O. Slane and B. M. Gaensler, eds. Magnetospheric Structure and Non-Thermal Emission of AXPs and SGRs M. Lyutikov1,2,3, C. Thompson4, S.R. Kulkarni5 In the framework of the magnetar model for the Soft Gamma Abstract. Repeaters and Anomalous X-ray Pulsars, we consider the structure of neutron star magnetospheres threaded by large-scale electrical currents. We construct self-similar, force-free equilibria under the assumption of axisymmetry and a power law dependence of magnetic field on radius, B ∝ r−(2+p). A large-scale twist of the field lines softens the radial de- pendence to p < 1, thereby accelerating the spindown torque with respect to a vacuum dipole. A magnetosphere with a strong twist (Bφ/Bθ = O(1) at the equator) has an optical depth ∼ 1 to resonant cyclotron scattering, independent of frequency (radius), surface magnetic field strength, or the charge/mass ratio of the scattering charge. We investigate the effects of the resonant Compton scattering by the charge carriers (both electrons and ions) on the emergent X-ray spectra and pulse profiles. 1. Nature of the AXPs and SGRs The Anomalous X-ray Pulsars and Soft Gamma Repeaters are neutron stars which share similar spin periods, P = 5 − −12 s, characteristic ages, P/ P = 3 × 103 − 4 × 105 yr, and X-ray luminosities, LX = 3 × 1034 − 1036 erg s−1, well in excess of the spin-down luminosities (Thompson et al. 2001). They are almost certainly young and isolated: no evidence for a binary stellar companion has yet been detected in any of these sources, and a few are convincingly associated with young supernova remnants. The overlap between the SGR and AXP sources in P and LX), and the observed variabil- a three-dimensional parameter space (P , 1Department of Physics, McGill University, Montr´eal, QC 2Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139 3CITA National Fellow 4Canadian Institute for Theoretical Astrophysics, 60 St. George St., Toronto, ON M5S 3H8 5California Institute of Technology, 105-24, Pasadena, CA 91125 1 2 Lyutikov, Thompson & Kulkarni ity in their X-ray output, provides circumstantial evidence that they share a common energy source: the decay of a very strong (∼> 1015 G) magnetic field. The persistent emission of the SGRs and AXPs has both a thermal and a non-thermal component. In the case of the SGRs, this emission becomes brighter after periods of bursting activity and involves a comparable release of energy to the outbursts (averaged over time). 2. Twisted Neutron Star Magnetospheres The magnetic fields of magnetars are most likely generated by a hydromagnetic dynamo as the star is born, and may be associated with rapid initial rotation (Duncan & Thompson 1992). A strong twist in a ∼ 1015 G magnetic field will relax at intervals as the field is transported through the deep interior of the neutron star. Even if the electrical current were initially confined to interior of the star, the Lorentz force would become strong enough to deform its crust, thereby twisting up the external magnetic field. In such a situation, a persistent current will flow through the magnetosphere, supported by emission of light ions (e.g. H, He, C) and electrons from the neutron star surface. The decay of this current outside of the star is an efficient mechanism for converting magnetic energy to X-rays, and for inducing rapid variations in the X-ray flux. To see how the persistent current will modify the structure of the mag- netosphere, we find axisymmetric force-free equilibria outside a (non-rotating) spherical surface, ∇ × B = α(P)B. These equilibria form a one-dimensional sequence labeled by the flux parameter P = P(R, θ), with poloidal magnetic field BP = ∇P × φ/R sin θ. As a major simplification we consider self-similar configurations P = P0(R/RNS)−pF (θ), B(R, θ) ∼ F (θ) × (R/RNS)−(2+p), (1) following Lynden-Bell & Boily (1994). The radial index p is uniquely determined by a single parameter C, which is related to the strength of the current: p(p + 1)F + sin2 θ ∂2F ∂(cos θ)2 = −CF 1+2/p. (2) Choosing a dipole field at a zero current, p(C) is determined by the three bound- ary conditions, Br(R, θ = π/2) = 0, Br(RNS, θ = 0) = const, Bφ(R, θ = 0) = 0. The index p is most conviently expressed as a function of the net twist ∆φN−S between the north and south magnetic poles (Fig. 1a). 3. Resonant Scattering The current-currying charges also provide a significant optical depth to resonant cyclotron scattering. For a particle of charge Ze and mass M , the resonant cross- section is σres(ω) = (π2Ze2/M c) (1 + cos2 θkB)δ(ω − ωc). The optical depth is determined by relating the particle density nZ to the twist in the magnetic field through (Ze)nZ vZ = ǫZ (c/4π)∇ × B. In our self-similar model, (cid:18) vZ c (cid:19) τres = πǫZ 8 (cid:16)1 + cos2 θkB(cid:17) (cid:20) F (θ) F (π/2)(cid:21)1/p ∆φN−S. (3) APS Conf. Ser. Style 3 Figure 1. (a) The radial index p as a function of the twist ∆φN−S. (b) The actual polar field inferred from spin parameters P and P , compared with the magnetic dipole formula. Thus the optical depth is τres ∼ (vZ /c)−1 for strong twists ∆φN−S ∼ 1. Re- markably it is independent of the mass and charge of the scatterers, the radius, or the resonant frequency. 4. Implications for X-ray Spectra and Pulse Profiles • Surface Heating. If ions and electrons supply the current, then the stellar surface is heated at the rate LX ≃ 3 × 1035 ǫion(cid:18) Bpole 1014 G(cid:19) (cid:18) Bφ Bθ(cid:19)θ=π/2 erg s−1. (4) (The electrons are electrostatically accelerated downward above the an- ode.) This is comparable to the observed luminosities of AXPs and SGRs. A global twist will decay on the timescale tdecay = EB − EB(dipole) LX = 30 ǫ−1 ion(cid:18) Bpole 1014 G(cid:19) ∆φN−S yr. (5) • Resonant Comptonization. Both ions and electrons can be expected to move mildly relativistically where they resonantly scatter 1 − 10 keV pho- tons. The product of the mean frequency shift h∆ω/ωi per scattering with the expected number of scatterings is O(1) when ∆φN−S ∼ 1 radian. Therefore, multiple scattering at the cyclotron resonance by the moving charge carriers will create a non-thermal spectral tail to the X-ray flux emerging from the surface. At optical depths ∼< 1, the hardness of the spectrum increases with the number of scattering and thus with the reso- nant optical depth. • Pulse Profiles. The emergent pulse profile is strongly modified by magne- tospheric scattering. Three effects enter here: the optical depth τres(θ) 4 Lyutikov, Thompson & Kulkarni is anisotropic, vanishing toward the magnetic poles; the resonant sur- face is aphserical; and the scattered radiation tends to be beamed along the magnetic field (in part because of the motion of the charge carriers). The pulse profile will be approximately frequency-independent in this self- similar model. • Ion Cyclotron Resonance. The ion component of the current will generate a comparable optical depth to the electron component in a self-similar mag- netosphere (at frequencies below the surface cyclotron frequency). Since the ion cyclotron resonance sits much closer to the star (at ∼ 10−20 km for 2− 10 keV photons), it is more sensitive to the presence of higher magnetic multiples. Thus, a combination of ion and electron scattering in a multi- polar magnetic field will produce an energy dependent total profile which is more complex at higher energies. An ion cyclotron emission line would be a clear observational signature of surface heating by magnetospheric charges. • SGR/AXP spindown. When twisted, the magnetic field drops off more slowly than ∼ R−3. Thus the real polar field inferred from P, P is smaller than the magnetic dipole formula would imply, while the braking index becomes n = 2p + 1 < 3 (Fig. 1b). Our model predicts that the spindown rate increases with the optical depth to resonant scattering, and hence with the hardness of the persistent X-ray spectrum. In fact the active SGRs 1806-20 and 1900+14 both have higher P and harder X-ray spectra than any AXP. The quiescent SGR 0525-66 has a softer spectrum. We predict that its spindown rate, when measured, will be intermediate between these sources and the AXPs. • Giant Flare Mechanism There are two generic possibilities for the produc- tion of the giant flares of the SGRs, in the framework of our model: (i) giant flares may result from a sudden change (unwinding) in the internal magnetic field, implanting a twist into the magnetosphere; or alternatively (ii) the twist may build up more gradually in the magnetosphere, leading to a sudden relaxation in close analogy with Solar flares. Measurements of hardening/softening of the persistent X-ray spectrum, changes in pulse profile, and changes in spindown rate provide a way of discriminating be- tween these possibilities. In particular, the simplified pulse profile observed in SGR 1900+14 after the 27 August 1998 giant flare (Woods et al. 2000) can be explained by an increase in the currrent at the radius of the electron cyclotron resonance. References Duncan, R.C. and Thompson, C., 1992, ApJ, 392, L9 Lynden-Bell, D. and Boily, C., 1994, MNRAS, 267, 146 Thompson, C., Lyutikov M., Kulkarni S., astro-ph/0110677 Woods, P., et al. 2001, ApJ, 552, 748
astro-ph/0702277
2
0702
2007-09-06T08:23:56
Directionality in the WMAP Polarization Data
[ "astro-ph" ]
Polarization is the next frontier of CMB analysis, but its signal is dominated over much of the sky by foregrounds which must be carefully removed. To determine the efficacy of this cleaning it is necessary to have sensitive tests for residual foreground contamination in polarization sky maps. The dominant Galactic foregrounds introduce a large-scale anisotropy on to the sky, so it makes sense to use a statistic sensitive to overall directionality for this purpose. Here we adapt the rapidly computable D statistic of Bunn and Scott to polarization data, and demonstrate its utility as a foreground monitor by applying it to the low resolution WMAP 3-yr sky maps. With a thorough simulation of the maps' noise properties, we find no evidence for contamination in the foreground cleaned sky maps.
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 1 -- 5 (2007) Printed 31 October 2018 (MN LATEX style file v2.2) Directionality in the WMAP Polarization Data Duncan Hanson1, Douglas Scott1⋆ and Emory F. Bunn2 1Department of Physics & Astronomy, University of British Columbia, Vancouver, B.C. V6T 1Z1, Canada 2Physics Department, University of Richmond, Richmond, VA 23173, USA Accepted 2007 July 3. Received 2007 June 27; in original form 2007 February 9 ABSTRACT Polarization is the next frontier of CMB analysis, but its signal is dominated over much of the sky by foregrounds which must be carefully removed. To determine the efficacy of this cleaning it is necessary to have sensitive tests for residual foreground contamination in polarization sky maps. The dominant Galactic foregrounds introduce a large-scale anisotropy on to the sky, so it makes sense to use a statistic sensitive to overall directionality for this purpose. Here we adapt the rapidly computable D statis- tic of Bunn and Scott to polarization data, and demonstrate its utility as a foreground monitor by applying it to the low resolution WMAP 3-yr sky maps. With a thorough simulation of the maps' noise properties, we find no evidence for contamination in the foreground cleaned sky maps. Key words: cosmic microwave background -- methods: numerical -- cosmology: ob- servations -- cosmology: theory -- large-scale structure 1 INTRODUCTION for example, Our progress in understanding the large scale make-up of the Universe has been propelled rapidly in the last 15 years by extremely detailed measurements of the Cosmic Microwave Background (CMB). Current measurements of the CMB temperature anisotropies have allowed us to place stringent constraints on some of our Universe's key pa- rameters and are proving to be a driving force in fun- damental physics (see e.g. Scott & Smoot 2006). Cosmic- variance places limits on the amount of information which can be extracted from the temperature anisotropies alone, however. The Planck satellite, is expected to reach this limit over the entire primary temperature anisotropy power spectrum (Planck Bluebook 2006). For- tunately, the polarization of the CMB anisotropies pro- vides us with a wealth of additional cosmological infor- mation which we have only just begun to probe. The benefits of CMB polarization data are numerous. Perhaps most excitingly, polarization data provide a chance to de- tect primordial gravitational waves, through the measure- ment of B-mode polarization (Seljak & Zaldarriaga 1997; Kamionkowski, Kosowsky & Stebbins 1997; Hu & White 1997; Kamionkowski & Kosowsky 1998). The rapid im- provements in CMB polarization data provided by experi- ments like PIQUE (Hedman et al. 2001), DASI (Kovac et al. 2002), CAPMAP (Barkats et al. 2005), BOOMERanG (Masi et al. 2006), CBI (Readhead et al. 2004), MAXIPOL (Wu et al. 2006), WMAP (Page et al. 2006), and upcoming ⋆ E-mail: [email protected] experiments like BICEP, CLOVER (Taylor 2006), QUAD, and the Planck satellite, ensure that polarization will be- come one of the most important tools we have for furthering our understanding of the Universe in the years to come. Unfortunately, the polarization signal is much more dif- ficult to measure than the temperature anisotropies. The magnitude of the CMB polarization anisotropies is at most 10 per cent that of the temperature anisotropies, making their detection alone a challenging task. This difficulty is being gradually overcome with improvements in detector technology. Another problem with the CMB polarization, however, is that its signal is dominated over most of the sky by foregrounds. In particular, over the wavelength range where the CMB is typically measured, synchrotron and non- spherical dust emission from within our Galaxy contaminate uncleaned polarization maps. It is not yet clear how well the problem of foreground contamination can be solved. Its reso- lution will require both a detailed knowledge of the structure and magnitude of the foregrounds, and an understanding of how to remove them from the polarization data without in- troducing unwanted systematics. Understandably, the issue of polarization foregrounds has received much attention in the literature. Several pa- pers have been written on the construction of templates for the emission (de Oliveira-Costa et al. 2003; Baccigalupi 2003; Page et al. 2006; Hansen et al. 2006), and a number of methods have been proposed for performing the foreground removal (Bouchet, Prunet & Sethi 1999; Hansen et al. 2006; Slosar, Seljak & Makarov 2004). Testing the efficacy of these methods is a crucial step in the analysis pipeline, to compare the abilities of the various removal techniques and ultimately 2 D. Hanson, D. Scott and E.F. Bunn to provide confidence in the accuracy of CMB polarization data. Tests for the existence of residual polarized foreground contamination have received less attention in the literature, however. The test used in the WMAP team's analysis, where foregrounds are cleaned using a template method in pixel space, is a simple χ2 statistic (Page et al. 2006). Unfortu- nately, this statistic cannot be used to make comparisons between different cleaning methods. The only currently pro- posed 'general' statistic which is capable of this application is based on the Bipolar Power Spectrum (BiPS) and has been developed by Basak, Hajian & Souradeep (2006). In this paper, we introduce and analyse another general method of testing for residual foreground contamination in CMB polarization maps, the D statistic of Bunn & Scott (2000). It possesses the virtues of having a simple interpreta- tion as a measure of directionality in a CMB map and of be- ing extremely rapid to compute. We argue that this statistic is well suited to the detection of foreground contamination, as both the synchrotron and dust foreground components have strong overall directionality on large scales. We then apply the D statistic to the WMAP 3-yr polarization maps to demonstrate its usefulness for real data. 2 THE D STATISTIC The D statistic was first presented by Bunn & Scott (2000); the D value for a sky map pixelized into N pixels is defined as D ≡ max n f (n) min n f (n) , (1) where n runs over the unit sphere, and with the function f (n) given by f (n) = N Xp=1 wp(n · gp)2 . (2) Here wp are weights chosen to compensate for noise and the effects of a cut sky, and gp is a local directionality vector calculated for the pixel p. The f (n) value is a measure of how much (or how little) the gp of a map tend to point in a given direction. The D statistic tells us how extreme these values are. Once D has been calculated for real sky data, we can compare its value to that found for simulated realizations of the CMB. Calculating D for a large number of these simulations gives us a histogram of expected values, and excess directionality in a CMB data-set appears as a value of D which is an outlier of this distribution. For sky maps of temperature, a reasonable quantity to use for gp is the local temperature gradient ∇Tp. For polar- ization maps, on the other hand, the polarization psuedo- vectors themselves are obvious candidates. We define gp to be a vector whose magnitude and direction are those of the polarization at point p. To express gp explicitly in terms of the Stokes parameters Q and U , recall that the polarization magnitude is P = pQ2 + U 2, while the polarization direction lies in the tangent plane to the celestial sphere at p and makes an angle with the meridian of (3) θ = 1 2 tan−1(cid:18) U Q(cid:19) . (4) Measuring the angle from the meridian is a WMAP con- vention (Page et al. 2006) which we adopt here, although it is not universally followed. We define the vector gp to have magnitude P and direction given by θ. Because the polarization is a spin-2 quantity, it is repre- sented by headless 'pseudo-vectors' rather than vectors, so θ can always be rotated by 180◦ without altering the polariza- tion. However, due to the quadratic definition of f (n), this ambiguity does not affect D. Moreover, because the func- tion f smoothly averages together polarization information over the whole sky, D is only sensitive to large scale (& 20◦) directionality. The weights wp must be chosen to satisfy the null hy- pothesis that for isotropically distributed gp no preferred direction will be found on average, even when parts of the sky have been removed. We can see how these weights may be obtained by casting f (n) in matrix form as f (n) = nT A n, where A is a 3 × 3 matrix defined by Aij = N Xp=1 wp gpi gpj . (5) (6) In order for no preferred direction to be found, we must have f (n) independent of n (on average). From equation 5 it can be seen that this constraint may be realized by choosing suitably normalized wp such that hAi is equal to the iden- tity matrix for isotropically distributed gp. This criterion ensures that f (n) will be approximately constant over the sky when the underlying data lack a preferred direction, al- though it does not guarantee that fluctuations of f about its ensemble average will be isotropic. As a result, D may still pick out preferred directions arising from the noise structure or sky coverage, even when the underlying signal is isotropic. We find that such departures from anisotropy are weak in practice; this point is discussed further in Section 4. Because A is symmetric, requiring hAi = I gives only 6 independent equations on the weights. To completely de- termine them for a large number of pixels, the additional constraint is imposed that Var(wp Pp ) be minimized, where Pp is the expected value of gp2 given by pixel noise and cosmological signal. For a large signal-to-noise ratio on an isotropic sky this is equivalent to minimizing the variance of the weights. When noise is not negligible, minimizing Var(wp Pp) rather than Var(wp) has the favourable effect of down-weighting noisy pixels. More explicit details on the calculation of the weights are given in Bunn & Scott (2000). Once the weights wp have been calculated, the simple quadratic definition of f (n) makes it possible to calculate D for a given data-set in O(N ) operations, where N is the number of pixels. This can be done by noting that the values of n which extremize f (n) are given by the eigenvectors of A. D follows immediately by taking the ratio of the largest and smallest eigenvalues. D is one of the simplest statistics which can be pro- posed for the identification of statistical anisotropy in a CMB map. To compare it to the BiPS method of Basak, Hajian & Souradeep (2006), we note that it is ul- timately a less powerful statistic, as BiPS can be used to Directionality in the WMAP Polarization Data 3 probe both large and small scales, whereas D is only sensi- tive to large scale anisotropies. The advantage of D, however, is that it can be calculated in O(N ) operations, an unbeat- able scaling. This speed makes D an excellent statistic for 'first look' determination of anisotropy in CMB maps, and the concomitant possible foreground contamination. 3 DIRECTIONALITY OF POLARIZATION FOREGROUNDS The two main sources of foreground contamination in CMB polarization maps at the ∼ 100 GHz frequencies where the CMB is usually probed are Galactic synchrotron and ther- mal dust emission. Polarized synchrotron emission results from the acceleration of cosmic-ray electrons as they orbit in the Galactic magnetic field. Polarized dust emission, on the other hand, is the result of non-spherical dust grains which tend to align their long axes perpendicular to the field; these dust grains preferentially emit thermal radiation polarized along their long axis (Davis & Greenstein 1951). The polarization of both the synchrotron and thermal dust emission has its origins in the anisotropy introduced by the Galactic magnetic field, and at CMB wavelengths (where Galactic Faraday rotation is negligible) these two compo- nents should both be polarized preferentially in the same direction. Thus, we expect that they have a common and distinctive directionality on the sky, dictated by the large scale structure of the Galactic magnetic field. To illustrate the signature of this directionality under application of the D statistic, we analyse the low resolution K-band polariza- tion map at 23 GHz and the dust polarization template of the WMAP team, masking both with the standard polar- ization mask (P06) to mimic the situation when analysing maps for foreground contamination. For the K-band map we find a preferred direction at a Galactic latitude 86.4◦, within 4◦ of the poles. For the dust emission template we find that the preferred direction lies at 79.1◦, 11◦ shy of the poles. Thus, the significant presence of synchrotron and dust foregrounds in a polarization map is expected to result in an outlying value of D, with a preferred direction in the vicinity of the Galactic poles. 4 ANALYSIS OF WMAP 3-YR POLARIZATION DATA The release of the WMAP 3-year data has provided us with a glimpse at the structure of CMB polarization. Here we analyse the Q and U polarization maps for foreground con- tamination using the D statistic. We have chosen to examine the WMAP low resolution 3rd year (v2) polarization maps1 at HEALPix2 resolution 4, corresponding to 3072 pixels on the entire sphere. These are the maps used for low ℓ analysis of the WMAP data, cor- responding to the angular scales which D effectively probes. At this resolution it is also feasible to accurately simulate realizations of the WMAP noise structure using the full 1 Available at http://lambda.gsfc.nasa.gov/ 2 See http://healpix.jpl.nasa.gov Figure 1. Directionality histograms for maps simulated with WMAP low resolution V-band noise levels, with (solid) and with- out (dot-dashed) cosmological signal added. The two are essen- tially indistinguishable. The dashed line shown at D = 1.37 is the value of D calculated for the V-band data themselves. noise covariance matrix, which becomes necessary for large angle analyses. In all of our simulations, we generate cosmological signals given by the 'WMAP +all ΛCDM' fit parameters (Spergel et al. 2006). The details of this choice have little effect on the outcome of our simulations. All of the polar- ization maps are strongly noise-dominated, and we find that histograms of D for simulated data with and without cos- mological signal are indistinguishable for all of the WMAP bands; this is illustrated in Fig. 1 for simulated V-band maps. Thus, we believe that, with an accurate treatment of the maps' noise properties, any excess directionality which we detect in the WMAP data can be attributed to residual foregrounds, and not a cosmological signal. For all of our simulations, we add correlated Gaussian noise to each unmasked map pixel based on the full QU co- variance matrices for the corresponding WMAP frequency band, as provided by the WMAP team. We generate this noise using the Cholesky decomposition of the covariance matrix. This method accounts for QQ, QU and UU corre- lations both within each pixel and with other pixels across the sky. For uncleaned maps, the D statistic produces the ex- pected results. For each band, we generate and analyze ∼ 2000 simulated skies. An example of this is shown in Fig. 1. We can use the number of standard deviations (at which the real data lie from the histogram mean) as an approximate measure of the significance with which we detect foreground contamination. This is illustrated in Fig. 2, which shows the expected dependence on radiometer band as the synchrotron emission decreases with frequency and the dust contribution increases, leading to an apparent minimum of foreground contamination near the V-band. We can see that use of the standard P06 polar mask results in a greater level of con- tamination than when we use the more stringent P02 mask. For the P06 mask, all of the preferred directions which are found for the real data lie within 10◦ of the Galactic poles, and for the P02 mask the preferred directions all lie within 25◦ of the poles, further demonstrating the foreground origin 4 D. Hanson, D. Scott and E.F. Bunn Figure 2. Significances of foreground contamination for the WMAP low resolution maps using noise based on the full co- variance matrix. The (✷) symbols are for the standard P06 mask (27 per cent sky cut), while the (△) symbols are for the more stringent P02 masking (50 per cent sky cut for our downgrad- ing scheme). The curve roughly follows that expected from the frequency dependence of contamination by synchrotron and dust foreground components. Figure 3. Directionality histograms for simulated realizations of the WMAP V-band (solid) and Q-band (dashed) 'low resolution foreground reduced' maps, using the full covariance matrix de- scription of the maps' noise properties. The vertical lines are the values of D calculated for the Q- and V-band data themselves. of the directionality excess. With P06 masking, we find that 98% of our V-band simulations have values of D less than the value calculated for the WMAP data. With P02 mask- ing, this result drops to 90%. The corresponding significance figures of merit are 2.5σ and 1.3σ respectively. In order to assess the efficacy of the foreground removal performed by the WMAP team we also analyse the low reso- lution Q- and V-band foreground reduced maps. These have had dust and synchrotron templates projected out to remove foreground contamination. The results of these analyses are shown in Fig. 3. From this figure it can be seen that the Q and V maps have similar noise properties under analy- sis with D, as the simulation histograms almost completely overlap. We find no evidence for residual foreground con- tamination in these maps, their D statistics both lying well within the range of probable values, and with the maximal D directions for the Q- and V-band data being 55◦ and 27◦ from the pole, respectively. Our significance figure of merit is 0.6σ for the Q-band and 0.16σ for the V-band maps. 31% of the simulated D values lie below the map value for the Q- band, and 60% of the simulated data lie below the V-band value. Thus, we find no evidence for foreground contamina- tion in these cleaned maps. It is interesting to note that if we had generated noise which only accounted for the intra-pixel or diagonal compo- nents of the noise covariance matrix then we would have found strong evidence for residual contamination. In the case of the V-band foreground reduced data, for example, we would have found some evidence for foreground contam- ination, with over 98% of simulated D values falling below that calculated for real data in the cleaned V-band maps (corresponding to a value of 2.5σ for our significance fig- ure). Accounting for inter -pixel noise correlations broadens the distribution of simulated D values. Indeed, the detailed noise structure of the WMAP data appears to be such that it introduces a degree of anisotropy into the maps. This can best be seen by looking at the histograms of preferred direc- tions found in a series of 30,000 V-band map simulations, given in Fig. 4. Differing methods of noise generation were utilized for three sets of simulations. 'Diagonal' noise was generated from only the QQ and U U elements of the covari- ance matrix, intra-pixel noise included QU terms, and the full covariance matrix treatment simulated inter-pixel corre- lations as well. Unlike the isotropic distribution of preferred directions which is found when accounting only for diagonal noise correlations, there is a tendency towards anisotropy for the intra-pixel and full covariance simulations, with a definite slight preference for directions in the vicinity of (l, b) = (150◦, 40◦) for the full covariance simulations. The statistics of these two dimensional histograms support this visual finding, with the diagonal, intrapixel, and full co- variance simulation methods having variance/mean values of 1.41, 1.66 and 3.80 respectively. For an isotropic sky, we expect the histogram values to follow a multinomial distribu- tion, and to have a variance/mean of 1±.08. We believe that the excess variance which is seen in the diagonal and intra- pixel simulation histograms can be attributed to the slight inherent bias possessed by D, as discussed in Section 2. The much higher excess variance seen in the inter-pixel simula- tions can be attributed to the sum of this bias and another, stronger bias given by the noise covariance structure of the map. This bias is independent of the specifics of the D statis- tic, and most likely can be detected in the future by other polarization statistics sensitive to anisotropy. The depen- dence of our results on the precise treatment of noise un- derscores the general need for care when dealing with large angle polarization data. Directionality in the WMAP Polarization Data 5 ination in data from the Planck satellite, as well as other large-angle polarization experiments. (a) Diagonal (b) Intrapixel (c) Full covariance matrix Figure 4. Histograms of preferred direction found for maps simu- lated with diagonal, intra-pixel, and full covariance matrix based noise (Mollweide projection, Galactic coordinates). The colour units are in number of simulations. The preferred directions cal- culated by D are headless, and so we have taken the convention of choosing them to be in the Northern Hemisphere. 6 ACKNOWLEDGMENTS The work carried out in this paper made use of the HEALPix (G´orski et al. 2005) package for pixelization. This research was supported by the Natural Sciences and Engineering Research Council of Canada and the Canadian Space Agency. We also acknowledge the contributions of Patrick Plettner for some related earlier work. REFERENCES Baccigalupi C., 2003, New Astron. Rev., 47, 1127 Barkats D., et al., 2005, ApJ, 619, L127 Basak S., Hajian A., Souradeep T., 2006, Phys. Rev. D, 74, 021301 Bouchet F.R., Prunet S., Sethi S.K., 1999, MNRAS, 302, 663 Bunn E.F., Scott D., 2000, MNRAS, 313, 331 Davis, L.J., Greenstein, J.L. 1951, ApJ, 114, 206 de Oliveira-Costa A., Tegmark M., O'Dell C., Keating B., Timbie P., Efstathiou G., Smoot G., 2003, Phys. Rev., D68, 083003 G´orski K.M., Hivon E., Banday A.J., Wandelt B.D., Hansen F.K., Reinecke M., Bartelman M., 2005, ApJ, 622, 759 Hansen F.K., Banday A.J., Eriksen H.K., G´orski K.M., Lilje P.B., 2006, ApJ, 648, 784 Hedman M., Barkats D., Gundersen J.O., Staggs S.T., Winstein B., 2001, ApJ, 548, L111 Hu, W., White M., 1997, NewA, 2, 323 Kamionkowski M., Kosowsky A., 1998, Phys. Rev. D, 57, 685 5 CONCLUSION Kamionkowski M., Kosowsky A., Stebbins, A., 1997, Phys. Rev. D, 55, 7368 Kovac J., Leitch E.M., Pryke C., Carlstrom J.E., Halverson N.W., Holzapfel W.L., 2002, Nature, 420, 772 Masi S., et al., 2006, A&A, 458, 687 Page L., et al., 2007, ApJS, 170, 335 Planck BlueBook 'The Scientific Programme', ESA- SCI(2006)1; astro-ph/0604069v1 Readhead A.C.S., et al., 2004, Science, 306, 836 Scott D., Smoot, G.F., 2006, in The Review of Particle Physics, Yao W.-M., et al., JPhysG, 33, 1 Seljak, U., Zaldarriaga, M., 1997, Phys. Rev. Lett., 78, 2054 Slosar, A., Seljak, U., Makarov, A., 2004, Phys. Rev. D, 69, 123003 Spergel D.N., et al., 2006, ApJS, 170, 377 Taylor, A., 2006, NewA Rev., 50, 993 Wu J.H.P., et al., 2007, ApJ, 665, 55 We have investigated the D statistic of Bunn & Scott (2000) in the context of CMB polarization data, finding that it is an excellent general tool for quickly assessing the magnitude of foreground contamination in polarization maps. Under anal- ysis with D, imperfectly removed foregrounds leave a tell- tale signature of excess directionality toward the Galactic poles. An analysis of the WMAP 3-year polarization data supports these statements. We find the expected dependence of foreground contamination on channel frequency in all of the WMAP low resolution maps, with the magnitude of this excess disappearing in the cleaned, 'foreground reduced' maps. In the future, a better understanding of the polariza- tion foregrounds will allow us to develop more sophisticated statistics to precisely evaluate the possibility of foreground contamination in cleaned maps. Given the current state of understanding, however, we feel that calculation of D is a useful technique for this purpose, as it is one of the sim- plest possible statistics which captures the directional signa- ture of the polarization foregrounds. It has the advantages of being simple and rapidly computable, and can be con- sidered complementary to other statistics such as the BiPS (Basak, Hajian & Souradeep 2006). Hence it should be one of the diagnostic tests ready to assess foreground contam-
0812.0824
2
0812
2009-01-26T20:29:39
He II absorption and the sawtooth spectrum of the cosmic far-UV background
[ "astro-ph" ]
Cosmic ultraviolet background radiation between 3 and 4 Ryd is reprocessed by resonant line absorption in the Lyman series of intergalactic He II. This process results in a sawtooth modulation of the radiation spectrum from the He II Lya frequency to the Lyman limit. The size of this modulation is a sensitive probe of the epoch of helium reionization and of the sources that keep the intergalactic medium (IGM) highly ionized. For large absorption opacities, the background intensity will peak at frequencies just above each resonance, go to zero at resonance, and fluctuate greatly just below resonance. The He II sawtooth modulation may be one of the missing ingredients needed in the modelling of the abundances of metal ions like C III and Si IV observed in the IGM at redshift 3.
astro-ph
astro-ph
ApJL, in press Preprint typeset using LATEX style emulateapj v. 08/22/09 HE II ABSORPTION AND THE SAWTOOTH SPECTRUM OF THE COSMIC FAR-UV BACKGROUND Piero Madau1 & Francesco Haardt2 ApJL, in press ABSTRACT Cosmic ultraviolet background radiation between 3 and 4 Ryd is reprocessed by resonant line ab- sorption in the Lyman series of intergalactic He II. This process results in a sawtooth modulation of the radiation spectrum from the He II Lyα frequency to the Lyman limit. The size of this modulation is a sensitive probe of the epoch of helium reionization and of the sources that keep the intergalactic medium (IGM) highly ionized. For large absorption opacities, the background intensity will peak at frequencies just above each resonance, go to zero at resonance, and fluctuate greatly just below reso- nance. The He II sawtooth modulation may be one of the missing ingredients needed in the modelling of the abundances of metal ions like C III and Si IV observed in the IGM at redshift 3. Subject headings: cosmology: theory -- diffuse radiation -- intergalactic medium -- quasars: general 1. INTRODUCTION The intensity and spectrum of the cosmic ultraviolet background are one of the most uncertain yet critically important astrophysical input parameters for cosmological simulations of the intergalactic medium (IGM) and early reionization. Theoretical models of such diffuse radiation field can help interpret quasar absorption-line data and derive information on the distribution of primordial baryons (traced by H I, He I, He II transitions) and of the nucleosynthetic products of star formation (C III, C IV, Si III, Si IV, O VI, etc.). Because of the high ionization threshold (54.4 eV) and small photoionization cross-section of He II, and of the rapid recombination rate of He III, the double ionization of helium is expected to be completed by hard UV-emitting quasars around the peak of their activity at z ≈ 3 (e.g. Madau et al. 1999; Sokasian et al. 2002), much later than the reionization of intergalactic H I and He I. Several observations have provided controversial evidence for a transition in some properties of the IGM around the predicted epoch of helium reionization, from the He II Gunn-Peterson trough measured in the spectra of z > 2.8 quasars (e.g. Jakobsen et al. 1994; Reimers et al. 1997; Heap et al. 2000; Smette et al. 2002), to the possible detection of an increase in the temperature of the IGM (e.g. Ricotti et al. 2000; Schaye et al. 2000; McDonald et al. 2001; Zaldarriaga et al. 2001; Theuns et al. 2002; Bernardi et al. 2003), to the claimed sharp evolution in the column density ratios of metal line absorbers (Songaila et al. 1998; Boksenberg et al. 2003; Agafonova et al. 2007). With the imminent installation of the Cosmic Origins Spectrograph on board of the Hubble Space Telescope, the quantity and quality of far-UV observations of the IGM will improve significantly. Numerical simulations of patchy He II reionization (Paschos et al. 2007; McQuinn et al. 2008) are already shedding new light on the nature of such late reheating process and its potential impact on observables. In this Letter we return to the theory of cosmological radiative transfer and to the atomic processes that shape the spectrum of the far-UV background. We address a hitherto unnoticed effect, resonant absorption by the He II Lyman series, and show that this process will produce a sawtooth modulation of the radiation spectrum between 3 and 4 Ryd. The size of this modulation depends sensitively on the abundance of He II in the IGM, and may in turn be a crucial factor in determining the abundance of metal ions like C III, Si III, and Si IV in the IGM. The analogous modulation between 0.75 and 1 Ryd from hydrogen line absorption was first studied by Haiman et al. (1997) in the limiting case of a fully neutral IGM. 2. SPECTRAL FILTERING BY THE IGM 9 0 0 2 n a J 6 2 ] h p - o r t s a [ 2 v 4 2 8 0 . 2 1 8 0 : v i X r a We treat the radiation field Jν (z) as a uniform, isotropic background, and include the reprocessing of UV radiation in a clumpy IGM (HM96). The specific intensity (in ergs cm−2 s−1 Hz−1 sr−1) at redshift zo and observed frequency νo is given by (1) Jνo (zo) = dt dz dz (1 + zo)3 (1 + z)3 ǫν(z)e−¯τ , c 4π Z ∞ zo where ν = νo(1 + z)/(1 + zo), (dt/dz) = [H(z)(1 + z)]−1, H(z) is the Hubble parameter, e−¯τ ≡ he−τ i is the average cosmic transmission over all lines of sight, and ǫν (in ergs s−1 Hz−1 Mpc−3) is the proper volume emissivity. The effective continuum (LyC) optical depth between zo and z from Poisson-distributed absorbers is ¯τ (νo) ≡ ¯τc = Z z zo dz ′Z ∞ 0 dNHI f (NHI, z ′)(1 − e−τc), (2) 1 Department of Astronomy & Astrophysics, University of California, Santa Cruz, CA 95064. 2 Dipartimento di Fisica e Matematica, Universit´a dell'Insubria, Via Valleggio 11, 22100 Como, Italy. 2 Sawtooth spectrum of the far-UV background where f (NHI, z ′) is the bivariate distribution of absorbers in redshift and column density along the line of sight, and τc is the LyC optical depth through an individual cloud of hydrogen and helium column densities NHI, NHeI, and NHeII. The effective line absorption optical depth is instead ¯τn(z) = (1 + z)νn c Z dNHI f (NHI, z)Wn, (3) where νn is the frequency of the 1s → np Lyman series transition (n > 2) and Wn is the rest equivalent width of the line expressed in wavelength units. 2.1. Resonant He II absorption The far-UV metagalactic flux has long been known to be partially suppressed by the He II and H I continuum opacity of the IGM (e.g. Miralda-Escude & Ostriker 1990; Madau 1992), but little attention has been given to He II line absorption. Photons passing through the He II Lyα resonance are scattered until they redshift out of resonance, without any net absorption: aside from the photoionization of H I and He I, the only other He II Lyα destruction mechanism, two-photon decay, is unimportant in the low density IGM at the redshifts of interest.3 This is not true, however, for photons passing through a He II Lyman series resonance between the Lyman limit at energy hνL = 4 Ryd and the He II Lyβ at hνβ = 3.56 Ryd. If the opacity of the IGM in the Lyman series lines is large, Lyβ and higher Lyman line photons will be absorbed and degraded via a radiative cascade rather than escaping by redshifting across the line width. The net result is a sawtooth modulation of the spectrum between 3 and 4 Ryd, and a large discontinuous step at the He II Lyα frequency, as we show below. Consider for example radiation observed at frequency νo < νβ and redshift zo. The resonant absorption cross-section is a narrow, strongly-peaked function, different lines dominate the opacity at different absorption redshifts, and the line and continuum transmission can be treated as independent random variables (e.g. Madau 1995). Photons emitted between zo and zβ = (1 + zo)(νβ/νo) − 1 can reach the observer without undergoing resonant absorption. Photons emitted between zβ and zγ = (1 + zo)(νγ/νo) − 1 pass instead through the He II Lyβ resonance at zβ and are absorbed. Photons emitted between zγ and zδ = (1 + zo)(νδ/νo) − 1 pass through both the He II Lyβ and the Lyγ resonances before reaching the observer. The background intensity can then be written as Jνo(zo) = I(zo, zβ) + I(zβ, zγ)e−¯τβ + I(zγ, zδ)e−¯τβ −¯τγ + .... + I(zL, ∞)e− Pn ¯τn , (4) where we denote with the symbol I(zi, zj) the right hand side of equation (1) integrated between zi and zj with ¯τ ≡ ¯τc. Here zL = (1+zo)(νL/νo)−1, the LyC opacity ¯τc in all I integrals except the last (where He II must be added) includes only H I and He I absorption, and ¯τβ, ¯τγ, ¯τδ, ... are the He II Lyman series effective opacities at redshift zβ, zγ, zδ, .... Equation (4) is easily generalized to higher frequencies, e.g. for νβ < νo < νγ the first two terms must be replaced by the integral I(zo, zγ). The effect of the sawtooth modulation is best depicted in the idealized case of negligible continuum absorption, ¯τc → 0, and large line opacity, ¯τn → ∞ (this is similar to the hydrogen case studied by Haiman et al. 1997). The ensuing radiation flux is shown in Figure 1, where we have also assumed for simplicity that the integrand in equation (1) is independent of redshift and the proper emissivity is ǫν ∝ ν0 =const. Note how only sources between the observer and the "screen" redshift zn = (1 + zo)(νn/νo) − 1 corresponding to the frequency of the nearest Lyman series line above νo are not blocked from view: the background flux peaks at frequencies just above each resonance, as the first integral in equation (4) extends over the largest redshift path, and goes to zero only at resonance. 2.2. He II Lyα riemission The usual assumption that each photon entering a Lyman series resonance causes a radiative cascade that terminates in a Lyα photon requires full l-mixing of the 2s − 2p levels (Seaton 1959). Collisions are infrequent in the low-density IGM, however, and most radiative cascades from an np state terminate instead in two-photon 2s → 1s emission (Hirata 2006). The fraction, fn, of decays that generates Lyα photons can be determined from the selection rules and the decay probabilities, and it is fn = (1, 0, 0.2609, 0.3078, 0.3259...) for n = (2, 3, 4, 5, 6...) (Pritchard & Furlanetto 2006). Without l-mixing, the quantum selection rules forbid a Lyβ photons from being converted into Lyα, while at large n the conversion fraction asymptotes to 0.36. Let now Jνγ (zα) be the background intensity measured just above the He II Lyγ resonance at redshift zα = (1 + zo)(να/νo) − 1. The flux that is absorbed and converted into Lyα is then f4 × Jνγ (zα)[1 − e−¯τγ (zα)]. The additional flux observed at frequency νo ≤ να and redshift zo from this process is then ∆Jνo (zo) = (cid:18) νo να(cid:19)3 e−¯τc(νo,zo,zα) [f4 × Jνγ (zα)(1 − e−¯τγ (zα))]. (5) When summing up over all Lyman series lines, the term in square brackets must be replaced by Pn>3[fn × Jνn(zα)(1 − e−¯τn(zα))].4 3 Another conversion process of He II Lyα photons, the O III Bowen fluorence mechanism (Kallman & McCray 1980), can be neglected in nearly primordial intergalactic gas. 4 Note that the analogous equation (7) of Haiman et al. (1997) for the reprocessing of hydrogen Lyman series radiation erroneously includes a term (νn/να) to account for the conversion of higher energy photons into Lyα. This factor is spurious since in the process the specific intensity is conserved. Madau & Haardt 3 Fig. 1. -- Left: The sawtooth modulation of the metagalactic flux between 3 and 4 Ryd produced by resonant absorption in the Lyman series of intergalactic He II. The background specific intensity was computed assuming negligible continuum absorption, large line opacities, an integrand in equation (1) independent of redshift, and a proper emissivity ǫν ∝ ν 0 =const. Right: The far-UV background intensity (in units of 10−22 ergs cm−2 s−1 Hz−1 sr−1) at redshift 3 from quasar sources, computed using the radiative transfer code CUBA. Several models are compared: 1) "HM", where resonant absorption from the Lyman series of singly-ionized helium was neglected (dotted line); 2) "HM+S", where the sawtooth modulation of He II was added (solid red line). Both these models assume photoionization equilibrium with a uniform radiation field following HM96, and yield a value He II/H I=35 in optically thin absorbers; and 3) "DR" (for "delayed reionization"), where the He II/H I ratios were artificially increased to 160 (dashed green line), 250 (dashed magenta line), and 530 (solid blue line). All 5 models assume the quasar emissivity and absorbers distribution given in eqs. (6) and (7) of the text. The positions of the ionization thresholds of different ions are indicated by tick marks. 3. UV METAGALACTIC FLUX We now compute the integrated far-UV background from quasar sources at redshift 3, including the effect of He II resonant absorption. We parameterize the recent determination of the comoving quasar emissivity at 1 Ryd by Hopkins et al. (2007) as εion(z) = (4 × 1024 ergs s−1 Hz−1 Mpc−3) (1 + z)4.68 e−0.28z e1.77z + 26.3 . (6) At z = 3, this yields a value that is 1.4 times lower than that used in Madau et al. (1999). Assuming a power- law far-UV spectrum with spectral index −1.57 (Telfer et al. 2002), the adopted proper emissivity becomes ǫν(z) = (1 + z)3 εion(z) (hν/1 Ryd)−1.57. To compute the effective opacity of the IGM we use the standard parameterization for the distribution of absorbers along the line of sight, f (NHI, z) = A N −β HI (1 + z)γ, (7) with (A, β, γ) = (1.4, 1.5, 2.9) over the column density range 1011 < NHI < 1017.2 cm−2, and (A, β, γ) = (5, 1.5, 1.5) for NHI > 1017.2 cm−2 (e.g. Kim et al. 1997; Hu et al. 1995; Meiksin & Madau 1993; Petitjean et al. 1993; Tytler 1987). Here, the normalization A is expressed in units of 107 cm−2(β−1). The high-column density distribution agrees with the results of Stengler-Larrea et al. (1995), while the low-column density distribution produces an H I Lyα effective opacity at z = 3 of 0.41 as in Faucher-Giguere et al. (2008). The detailed redshift evolution of the quasar emissivity and IGM opacity are not important for the problem at hand, since the measured -- at 1 Ryd -- and inferred -- at 4 Ryd -- absorption distances for ionizing radiation at redshift 3 are quite small, and the background flux in the relevant energy range is largely determined by local sources. We have used the cosmological radiative transfer code CUBA to follow the propagation of UV radiation through a partially ionized inhomogeneous medium (HM96). The code uses a multi-zoned approximation to model the physical conditions within absorbing systems and infer the amount of singly-ionized helium that is present along the line of sight from the well-measured NHI distribution. We include the reprocessing into He II Lyα and two-photon continumm from resonant absorption in the Lyman series (as detailed in the preceding section) as well as continuum absorption, and assume photoionization equilibrium with a uniform radiation field. The resulting far-UV background intensity at redshift 3 is shown in Figure 1 for model "HM", where resonant absorption from the Lyman series of He II is neglected, and model "HM+S", where the sawtooth modulation was added. Both these models yield a value He II/H I=35 in optically thin absorbers. In the HM+S case the effective Lyβ line opacity is 0.43, and the sawtooth modulation causes 4 Sawtooth spectrum of the far-UV background a small decrease in the metagalactic flux, by at most a factor of 2, relative to the old HM spectrum. We have also run 3 other representative cases, termed "DR" for "delayed reionization". These models ignore the patchy nature of the reionization process and assume that a larger fraction of intergalactic helium at redshift 3 is in He II: this is obtained by artificially increasing the He II/H I ratios computed by CUBA to 160, 250, and 530, respectively. Resonant absorption from the Lyman series now causes a reduction of the background intensity between 3 and 4 Ryd by as much as one dex (off-resonance) compared to the HM spectrum. 4. DISCUSSION Since the pioneering work of Chaffee et al. (1986) and Bergeron & Stasinska (1986), many studies have used obser- vations of intervening metal absorption systems to reconstruct the shape of the photoionizing radiation field at z ∼< 3. The many modifications to the HM96 background intensity that have been proposed include: 1) a stronger He II Lyα feature in order to match the observed Si IV/Si II abundance ratios (Levshakov et al. 2003); 2) a depression between 3 and 4 Ryd in order to enhance the predicted C II/C IV and C III/C IV ratios, incorrectly attributed by Agafonova et al. (2007) to a He II Lyα Gunn-Peterson effect; 3) a softer far-UV spectrum in order to predict [O/Si] values that are consistent with theoretical yields (Aguirre et al. 2008). A detailed modelling of the abundances of intergalactic metals is beyond the scope of this paper: here, we just want to point out that the sawtooth modulation, if as large as computed in the DR models, may provide a better match to the observations. At the top of the right panel of Figure 1 we have indicated the positions of the ionization thresholds of Si III (33.5 eV), Si IV (45.1 eV), C III (47.9 eV), C IV (64.5 eV), O II (35.1 eV), and O III (54.9 eV) ions. The reprocessing of Lyman series and Lyman continuum photons increases He II Lyα in the DR spectra by a factor of 1.7 compared to the HM case. The flux at the Si IV ionization threshold decreases by a factor of 2, boosting the predicted abundance of Si IV. An even larger boost is expected in the abundance of C III, whose ionization threshold lies exactly within the He II Lyβ deep absorption feature, and of O III, whose threshold lies just beyond the He II Lyman limit. The above results show that line absorption from the Lyman series of intergalactic helium may be an important, so far neglected, process shaping the spectrum of the cosmic radiation background above 3 Ryd. The large resonant cross- sections for far-UV light scattering make the sawtooth modulation a sensitive probe of the epoch of helium reionization and of the sources that keep the IGM highly ionized. The He II sawtooth may be one of the crucial missing ingredients in the modelling of the abundances of metal ions like C III and Si IV observed in the IGM at redshift 3. In the case of large line opacities, substantial fluctuations are expected in the far-UV background intensity near each resonance, as the first integral in equation (4) extends over a small absorption distance, and just a few quasars are expected to contribute to the local emissivity. Such fluctuations may cause large variations in e.g. the observed C III abundances. In future work, we intend to study such fluctuations and address how the contribution of star-forming galaxies to the background may affect the He II/H I ratio in the IGM and the predicted sawtooth modulation. Support for this work was provided by NASA through grants HST-AR-11268.01-A1 and NNX08AV68G (P.M.). We thank Steve Furlanetto, Zoltan Haiman, and Avery Meiksin for useful discussions. The spectra plotted in Figure 1 are available upon request. REFERENCES Agafonova, I. I., Levshakov, S. A., Reimers, D., Fechner, C., Levshakov, S. A., Agafonova, I. I., Centurin, M., & Molaro, P. Tytler, D., Simcoe, R. A., & Songaila, A. 2007, A&A, 461, 893 2003, A&A, 397, 851 Aguirre, A., Dow-Hygelund, C., Schaye, J., & Theuns, T. 2008, ApJ, 689, 851 Bergeron, J., & Stasinska, G. 1986, A&A, 169, 1 Bernardi, M., et al. 2003, AJ, 125, 32 Boksenberg, A., Sargent, W. L. W., & Rauch, M. 2003, preprint (arXiv:0307557) Madau, P. 1992, ApJ, 389, L1 Madau, P. 1995, ApJ, 441, 18 Madau, P., Haardt, F., & Rees, M. J. 1999, ApJ, 514, 648 Madau, P., & Meiksin, A. 1994, ApJ, 433, L53 McDonald, P., Miralda-Escud´e, J., Rauch, M., Sargent, W. L. W., Barlow, T. A., & Cen, R. 2001, ApJ, 562, 52 Chaffee, F. H., Jr., Foltz, C. B., Bechtold, J., & Weymann, R. J. McQuinn, M., Lidz, A., Zaldarriaga, M., Hernquist, L., Hopkins, 1986, 301, 116 Faucher-Giguere, C.-A., Prochaska, J. X., Lidz, A., Hernquist, L., & Zaldarriaga, M. 2008, ApJ, 681, 831 Haardt, F., & Madau, P. 1996, ApJ, 461, 20 (HM96) Haiman, Z., Rees, M. J., & Loeb, A. 1997, ApJ, 476, 458 Heap, S. R., Williger, G. M., Smette, A., Hubeny, I., Sahu, M., Jenkins, E. B., Tripp, T. M., & Winkler, J. N. 2000, ApJ, 534, 69 Hirata, C. M. 2006, MNRAS, 367, 259 Hopkins, P. F., Richards, G. T., & Hernquist, L. 2007, ApJ, 654, 731 Hu, E. M., Kim, T.-S., Cowie, L. L., Songaila, A., & Rauch, M. 1995, AJ, 110, 1526 Jakobsen, P., Boksenberg, A., Deharveng, J. M., Greenfield, P., Jedrzejewski, R., & Paresce, F. 1994, Nature, 370, 35 Kallman, T., & McCray, R. 1980, ApJ, 242, 615 Kim, T.-S., Hu, E. M., Cowie, L. L., & Songaila, A. 1997, AJ, 114, 1 P. F., Dutta, S., & Faucher-Giguere, C.-A. 2008, ApJ, submitted (arXiv:0807.2799) Meiksin, A., & Madau, P. 1993, ApJ, 412, 34 Miralda-Escud´e, J., & Ostriker, J. P. 1990, ApJ, 350, 1 Paschos, P., Norman, M. L., Bordner, J. O., & Harkness, R. 2007, submitted (arXiv:0711.1904) Petitjean, P., Webb, J. K., Rauch, M., Carswell, R. F., & Lanzetta, K. M. 1993, MNRAS, 262, 499 Pritchard, J. R., & Furlanetto, S. R. 2006, MNRAS, 367, 1057 Reimers, D., Kohler, S., Wisotzki, L., Groote, D., Rodriguez-Pascual, P., & Wamsteker, W. 1997, A&A, 327, 890 Ricotti, M., Gnedin, N. Y., & Shull, J. M. 2000, ApJ, 534, 41 Schaye, J., Theuns, T., Rauch, M., Efstathiou, G., & Sargent, W. L. W. 2000, MNRAS, 318, 817 Seaton, M. J. 1959, MNRAS, 119, 90 Smette, A., Heap, S. R., Williger, G. M., Tripp, T. M., Jenkins, E. B., & Songaila, A. 2002, ApJ, 564, 542 Sokasian, A., Abel, T., & Hernquist, L. 2002, MNRAS, 332, 601 Songaila, A. 1998, AJ, 115, 2184 Madau & Haardt 5 Stengler-Larrea, E., et al. 1995, ApJ, 444, 64 Telfer, R. C., Zheng, W., Kriss, G. A., & Davidsen, A. F. 2002, Tytler, D. 1987, ApJ, 321, 49 Zaldarriaga, M., Hui, L., & Tegmark, M., 2001, ApJ, 557, 519 ApJ, 565, 773 Theuns, T., Bernardi, M., Frieman, J., Hewett, P., Schaye, J., Sheth, R. K., & Subbarao, M. 2002, ApJ, 574, L111
astro-ph/0509664
2
0509
2005-12-21T10:53:36
RHESSI and SOHO/CDS Observations of Explosive Chromospheric Evaporation
[ "astro-ph" ]
Simultaneous observations of explosive chromospheric evaporation are presented using data from the Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI) and the Coronal Diagnostic Spectrometer (CDS) onboard SOHO. For the first time, co-spatial imaging and spectroscopy have been used to observe explosive evaporation within a hard X-ray emitting region. RHESSI X-ray images and spectra were used to determine the flux of non-thermal electrons accelerated during the impulsive phase of an M2.2 flare. Assuming a thick-target model, the injected electron spectrum was found to have a spectral index of ~7.3, a low energy cut-off of ~20 keV, and a resulting flux of >4x10^10 ergs cm^-2 s^-1. The dynamic response of the atmosphere was determined using CDS spectra, finding a mean upflow velocity of 230+/-38 km s^-1 in Fe XIX (592.23A), and associated downflows of 36+/-16 km s^-1 and 43+/-22 km s^-1 at chromospheric and transition region temperatures, respectively, relative to an averaged quiet-Sun spectra. The errors represent a 1 sigma dispersion. The properties of the accelerated electron spectrum and the corresponding evaporative velocities were found to be consistent with the predictions of theory.
astro-ph
astro-ph
RHESSI and SOHO/CDS Observations of Explosive Chromospheric Evaporation Ryan O. Milligan1,3, Peter T. Gallagher2,3,4, Mihalis Mathioudakis1, D. Shaun Bloomfield1, Francis P. Keenan1, and Richard A. Schwartz3,5 ABSTRACT Simultaneous observations of explosive chromospheric evaporation are presented using data from the Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI) and the Coronal Diagnostic Spectrometer (CDS) onboard SOHO. For the first time, co-spatial imaging and spec- troscopy have been used to observe explosive evaporation within a hard X-ray emitting region. RHESSI X-ray images and spectra were used to determine the flux of non-thermal electrons accel- erated during the impulsive phase of an M2.2 flare. Assuming a thick-target model, the injected electron spectrum was found to have a spectral index of ∼7.3, a low energy cut-off of ∼20 keV, and a resulting flux of ≥4×1010 ergs cm−2 s−1. The dynamic response of the atmosphere was deter- mined using CDS spectra, finding a mean upflow velocity of 230±38 km s−1 in Fe XIX (592.23 A), and associated downflows of 36±16 km s−1 and 43±22 km s−1 at chromospheric and transition region temperatures, respectively, relative to an averaged quiet-Sun spectra. The errors repre- sent a 1σ dispersion. The properties of the accelerated electron spectrum and the corresponding evaporative velocities were found to be consistent with the predictions of theory. Subject headings: Sun: atmospheric motions -- Sun: flares -- Sun: UV radiation -- Sun: X-rays, γ rays 1. INTRODUCTION Current solar flare models (Antiochos & Stur- rock 1978; Fisher, Canfield, & McClymont 1984, 1985a,b,c; Mariska, Emslie, & Li 1989) predict two types of chromospheric evaporation processes. "Gentle" evaporation occurs when the chromo- sphere is heated either directly by non-thermal electrons, or indirectly by thermal conduction. The chromospheric plasma subsequently loses en- ergy via a combination of radiation and low- velocity hydrodynamic expansion. "Explosive" evaporation takes place when the chromosphere is 1Department of Physics and Astronomy, Queen's Uni- versity Belfast, Belfast, BT7 1NN, Northern Ireland. 2School of Physics, Trinity College Dublin, Dublin 2, Ireland. 3Laboratory for Astronomy and Solar Physics, NASA Goddard Space Flight Center, Greenbelt, MD 20771, U.S.A. 4L-3 Communications GSI. 5Science Systems and Applications, Inc. unable to radiate energy at a sufficent rate and consequently expands at high velocities into the overlying flare loops. The overpressure of evapo- rated material also drives low-velocity downward motions into the underlying chromosphere, in a process known as chromospheric condensation. From a theoretical perspective, Fisher, Can- field, & McClymont (1985a) investigated the re- lationship between the flux of non-thermal elec- trons (F ) and the velocity response of the atmo- sphere for the two classes of evaporation. For gen- tle evaporation, non-thermal electron fluxes of ≤ 1010 ergs cm−2 s−1 were found to produce upflow velocities of tens of kilometres per second. In con- trast, explosive evaporation was found to be as- sociated with higher non-thermal electron fluxes (F ≥ 3 × 1010 ergs cm−2 s−1) which drive both upflows of hot material at velocities of several hundred kilometres per second and downflows of cooler material at tens of kilometres per second. Observationally, previous studies have identi- fied blue-shifted Soft X-Ray and EUV lines indica- 1 Lin et al. 2002) and CDS observations are com- bined for the first time to investigate the relation- ship between the non-thermal electron flux and the response of the solar atmosphere. In Section 2 the analysis techniques employed are described, while the results are presented in Section 3. Our Con- clusions are then given in Section 4. 2. OBSERVATIONS AND DATA ANAL- YSIS This study focuses on a GOES M2.2 flare, which began at 12:44 UT on 2003 June 10. The event was selected from a sample of approximately 50 flares jointly observed by RHESSI and CDS. The limited field of view, cadence, and operating schedule of CDS, coupled with RHESSI nighttime and South Atlantic Anomaly passes, make simultaneous ob- servations by the two instruments quite rare. 2.1. The Coronal Diagnostic Spectrome- ter (CDS) The CDS observations reported here were ob- tained with the FLARE AR observing sequence. FLARE AR contains five .4 A wide spectral win- dows centered on He I (584.33 A; log T = 4.5), O V (629.73 A; log T = 5.4), Mg X (624.94 A; log T = 6.1), Fe XVI (360.76 A; log T = 6.4), and Fe XIX (592.23 A; log T = 6.9). Each raster consists of 45 slit positions, each ∼15 seconds long, result- ing in an effective cadence of ∼11 minutes. The slit itself is 4′′×180′′ yielding a ∼180′′×180′′ field of view. A zoomed-in region of the He I, O V, and Fe XIX rasters from the impulsive phase of the flare is given in Figure 1. Also shown is the Extreme ultraviolet Imaging Telescope (EIT; De- laboudini`ere et al. 1995) 195 A passband image ob- tained at 12:48 UT. A series of subsequent EIT im- ages makes it clear that the He I and O V bright- enings come from a flare ribbon rather than two distinct footpoints as Figure 1 may suggest. The spectrum from each CDS pixel was fit- ted with a broadened Gaussian profile (Thomp- son 1999), for each of the five spectral windows. Velocity maps were created by measuring Doppler shifts relative to quiet-Sun spectra, which were as- sumed to be emitted by stationary plasma. Pre- liminary fits to the Fe XIX line during the im- pulsive phase of the flare revealed an asymmetric broadening beyond the instrumental resolution of Fig. 1. -- CDS images in the He I, O V, and Fe XIX emission lines observed during the implusive phase of the flare, with the corresponding EIT 195 A im- age. RHESSI 12 -- 25 keV (dashed) and 25 -- 60 keV (solid) contours are overlayed, each drawn at 10% of the peak intensity. tive of chromospheric evaporation. Using the Bent Crystal Spectrometer onboard the Solar Maxi- mum Mission, Antonucci & Dennis (1983) and Zarro & Lemen (1988) reported upflow velocities of 400 km s−1 and 350 km s−1, respectively, in Ca XIX lines (3.1 -- 3.2 A). More recently, Cza- ykowska et al. (1999), Teriaca et al. (2003), and Del Zanna et al. (2005) observed velocities of 140 -- 200 km s−1 in Fe XIX (592.23 A), using the Coro- nal Diagnostic Spectrometer (CDS; Harrison et al. 1995) onboard the Solar and Heliospheric Obser- vatory (SOHO). Simultaneous upflows and down- flows during a Hard X-Ray (HXR) burst indica- tive of explosive evaporation have been observed using CDS and Yohkoh/Hard X-Ray Telescope by Brosius & Phillips (2004). While these studies provided a measurement of the dynamic response of the flaring chromosphere, they were unable to provide a measurement of the flux of electrons re- sponsible for driving such motions, nor the spatial relationship between the two. In this Letter, simultaneous Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI; 2 Fig. 2. -- Sample of spectra from the spectral window centered on Fe XIX (592.23 A). The vertical short- dashed line shows the rest wavelength of the Fe XIX line at 592.32 A while the long-dashed line shows the rest wavelength of the Fe XII line at 592.62 A. Panel 'a' was obtained from a quite Sun region, panel 'b' shows a stationary Fe XIX line from a post-implusive phase flare kernel, while panel 'c' shows an Fe XIX line from the flare ribbon during the implusive phase. The dotted line indicates the stationary component while the triple-dot-dashed line indicates the blue-shifted component. CDS. The strongest blue asymmetries were found within the flare ribbon during the impulsive phase. Outside this area, and after the impulsive phase, the Fe XIX line was observed to have a width comparible to the instrumental width. Figure 2 shows a sample of spectra taken from the spec- tral window centered on the Fe XIX (592.23 A) emission line. Panel 'a' shows a spectrum from a quiet Sun area in which no Fe XIX emission was visible. Instead, a weak emission line was observed at 592.6 A which Del Zanna & Mason (2005) have identified as Fe XII. Panel 'b' shows a stationary Fe XIX emission line extracted from a bright region, but after the impulsive phase at ∼12:50 UT when no significant flows are expected. An emission line with a strong blue asymmetry is shown in panel 'c'. This was extracted from the flare ribbon during the impulsive phase. The best fit to this line was consistent with stationary and blue-shifted components, both with widths com- parible to the instrumental resolution. As Fe XIX is not observed in quiet-Sun spectra, and follow- ing from Teriaca et al. (2003), the Doppler veloc- ity was measured as the shift between these two components. A heliographic correction was also applied, due to the longitude of the observations and assuming purely radial flows, 2.2. The Reuven Ramaty High Energy Spectroscopic Imager (RHESSI) RHESSI is an imaging spectrometer capable of observing X- and γ-ray emission over a wide range of energies (∼3 keV -- 17 MeV). During this event the thin attenuators on RHESSI were in place thus limiting the energy range to &6 keV. Flare emission was not observed above ∼60 keV. The flare lightcurves are shown in the top panel of Fig- ure 3. Both the RHESSI images and spectra were obtained over a 64 second period from 12:47:34 -- 12:48:38 UT to coincide with the timerange over which CDS observed blue asymmetries in the Fe XIX line. This time interval lies within the impulsive 25 -- 60 keV HXR burst and is indicated by two vertical dotted lines in the top panel of Figure 3. RHESSI images in two energy bands (12 -- 25 and 25 -- 60 keV) were reconstructed using the Pixon algorithm (Hurford et al. 2002). Con- tours at 10% of the peak intensity in each band are overlayed on each EUV image in Figure 1. The RHESSI spectrum was fitted assuming an isothermal distribution at low energies, and thick- target emission at higher energies (bottom panel of Figure 3). A thick-target model was chosen over a thin-target model as it is believed that the den- 3 ǫ−δ electrons keV−1 s−1 is the thick-target elec- tron injection spectrum and δ is the associated spectral index (Brown 1971). Because of the steep- ness of the RHESSI spectrum at high energies, the non-thermal flux is quite sensitive to the value of the low energy cut-off. In order to put a con- straint on this value, the temperature of the ther- mal component was obtained by another indepen- dent method, i.e. the equivalent width of the Fe line complex at 6.7 keV (Phillips 2004). The value of the equivalent width of this line, which is quite sensitive to the temperature, was used to estimate the temperature of the thermal component. Hav- ing fixed this value, the entire RHESSI spectrum was fitted using a least-squares fit. 3. RESULTS The thick-target model fitted to the RHESSI spectrum in Figure 3 was consistent with an elec- tron distribution having ǫc ∼20 keV and δ ∼7.3. The break energy of 20 keV is consistent with earlier works (e.g. Holman 2003, Sui, Holman, & Dennis 2005). The total power in non-thermal electrons was therefore 1×1029 ergs s−1. Explor- ing the possible range of values for the break en- ergy for this flare would yield an electron power value of 4×1029 ergs s−1 for ǫc = 15.0 keV while ǫc = 25.0 keV would give a power value of 6×1028 ergs s−1. However, either of these break energies would give a worse χ2 value than ob- tained from the original fit. By comparison, the total thermal power for the same time interval was found to be 1.2×1028 ergs s−1. Using the reconstructed 25 -- 60 keV image, an upper-limit to source size was calculated to be 2.3×1018 cm2. This was found by summing over all pixels with counts greater than 10% of the peak value. This threshold was chosen to eliminate sources outside of the main HXR-emitting region, which were assumed to be unreal; the source area was not found to be highly sensitive to this value. For example, a threshold of 5% yielded an area of 3.2×1018 cm2 and 15% yielded 2×1018 cm2. This area was also confirmed using the Fourier mod- ulation profiles from each of RHESSI's nine de- tectors, which are sensitive to spatial scales from 2.2′′to 183′′. Assuming a filling factor of unity, the resulting non-thermal electron flux was calculated to be ≥4×1010 ergs cm−2 s−1. Fig. 3. -- Top panel: RHESSI lightcurves from the 6 -- 12, 12 -- 25, and 25 -- 60 keV bands. The dot- ted vertical lines indicate the time interval over which images and spectra were obtained to corre- spond to time when significant upflows were ob- served using CDS. Bottom panel: Portion of the RHESSI spectrum integrated over the timerange given above. The energy range 6 -- 60 keV (ver- tical dot-dash lines) was fitted with an isother- mal component (dotted curve) and a thick-target bremsstrahlung component (dashed curve). sity of the flare loop is insufficent to thermalise the electrons as they propagate to the chromo- sphere. The thick-target model is used in the vast majority of cases (e.g. Holman 2003, Veronig et al. 2005). Furthermore, in this flare the HXR source is clearly aligned with the He I ribbon as seen by CDS, which implies that the acceler- ated electrons are losing their energy in the dense chromosphere rather than in the coronal loops. The total power of non-thermal electrons above the low energy cut-off (ǫc) was calculated from P (ǫ ≥ ǫc) = R ∞ fe(ǫ)dǫ ergs s−1, where fe(ǫ) ∼ ǫc 4 Fig. 5. -- Plasma velocity as a function of temper- ature for the five lines observed using CDS. Pos- itive velocities indicate downflows, while negative values indicate upflows. upflows were evident once the HXRs begin to di- minish from time T2 onwards. By identifying each pixel in the Fe XIX map that required a two-component fit to the line pro- file, between times T1 and T2, the velocity was measured for the corresponding pixel in each of the five CDS rasters. Figure 5 shows the mean ve- locity as a function of temperature for each line us- ing the methods described in Section 2. The error bars represent a 1σ dispersion. At chromospheric and transition region temperatures, plasma ve- locities show red-shifts of 36±16 km s−1 and 43±22 km s−1, respectively, while the blue-shift observed in the 8 MK Fe XIX line corresponds to a velocity of 230±38 km s−1. No significant flows were observed in the Mg X and Fe XVI lines. The combination of high-velocity upflows and low- velocity downflows, together with a non-thermal electron flux of ≥4×1010 ergs cm−2 s−1 provides clear evidence for explosive chromospheric evapo- ration. 4. DISCUSSION AND CONCLUSIONS For the first time, co-spatial and co-temporal HXR and EUV observations of chromospheric evaporation are presented using RHESSI and SOHO/CDS. High upflows velocities (∼230 km s−1) were clearly observed in high-temperature Fe XIX emission during the impulsive phase of an M2.2 Fig. 4. -- Velocity maps in He I and Fe XIX. Downflows are indicated by red pixels while up- flows are indicated by blue pixels. The vertical dashed lines correspond to the times indicated by the vertical dashed lines in Figure 3 and the ar- row denotes the direction in which the CDS slit moves. Black regions in the Fe XIX map represent pixels where no significant Fe XIX emission was observed. RHESSI 25 -- 60 keV contours at 10% and 40% of the peak intensity are overlayed. Figure 4 shows velocity maps in the He I and Fe XIX lines. The He I map shows consistent downflows of 20 -- 50 km s−1 until the slit leaves the flaring region at ∼12:50 UT. A velocity map in O V showed a similar trend. However, the Fe XIX map shows strong upflow velocities of 190 -- 280 km s−1 during the HXR peak, indicated by 'T1' and 'T2' on Figures 3 and 4. No significant 5 flare, while much lower downflow velocities (∼40 km s−1) were observed in the cooler He I and O V lines. The value of the non-thermal electron flux (≥4×1010 ergs cm−2 s−1) and the resulting veloc- ity response are indicative of an explosive evapo- ration process occuring during this flare, as laid out in Fisher, Canfield, & McClymont (1985a) and Mariska, Emslie, & Li (1989). The combination of HXR and EUV observa- tions presented in this Letter have enabled us to obtain a greater understanding of the char- acteristics of chromospheric evaporation, a fun- damental process in solar flares. We have pre- sented the first detection of explosive mass mo- tions within HXR footpoints, and determined the flux of non-thermal electrons responsible for driv- ing such flows. This work has been supported by a Department of Employment and Learning (DEL) studentship and a Cooperative Award in Science and Technol- ogy (CAST). FPK is grateful to AWE Aldermas- ton for the award of a William Penny Fellowship. We would like to thank Drs. Brian Dennis, Joe Gurman, and Dominic Zarro at NASA Goddard Space Flight Center for their stimulating discus- sion and continued support. We would also like to thank the anonymous referee for their comments and suggestions which have greatly improved this Letter. SOHO is a project of international collabo- ration between the European Space Agency (ESA) and NASA. Del Zanna, G., Berlicki, A., Schmieder, B., & Ma- son, H. E. 2005, Sol. Phys., In Press Delaboudini`ere, J. -P., et al. 1995, Sol. Phys., 162, 291 Fisher, G. H., Canfield, R. C., & McClymont, A. N. 1984, ApJ, 281, L79 Fisher, G. H., Canfield, R. C., & McClymont, A. N. 1985a, ApJ, 289, 414 Fisher, G. H., Canfield, R. C., & McClymont, A. N. 1985b, ApJ, 289, 425 Fisher, G. H., Canfield, R. C., & McClymont, A. N. 1985c, ApJ, 289, 434 Harrison, R. A., et al. 1995, Sol. Phys., 162, 233 Holman, G. D. 2003, ApJ, 586, 606 Hurford, G. J., et al. 2002, Sol. Phys., 210, 61 Lin, R. P., et al. 2002, Sol. Phys., 210, 3 Mariska, J. T., Emslie, A., G., & Li, P. 1989, ApJ, 341, 1067 Phillips, K. J. H. 2004, ApJ, 605, 921 Sui, L., Holman, G. D., & Dennis, B. R. 2005, ApJ, 626, 1102 Teriaca, L., Falchi, A., Cauzzi, G., Falciani, R., Smaldone, L. A., & Andretta, V. 2003, ApJ, 588, 596 REFERENCES Thompson, W. T. 1999, CDS Software Note No. Antiochos, S. K., & Sturrock, P. A. 1978, ApJ, 220, 1137 Antonucci, E., & Dennis, B. R. 1983, Sol. Phys., 86, 67 Brosius, J. W., & Phillips, K. J. H. 2004, ApJ, 613, 580 Brown, J. C. 1971, Sol. Phys., 18, 489 Czaykowska, A., De Pontieu, B., Alexander, D., & Rank, G. 1999, ApJ, 521, L75 Del Zanna, G., & Mason, H. E. 2005, A&A, 433, 731 53 Veronig, A. M., Brown, J. C., Dennis, B. R., Schwartz, R. A., Sui, L., & Tolbert, A. K. 2005, ApJ, 621, 482 Zarro, D. M., & Lemen, J. R. 1988, ApJ, 329, 456 This 2-column preprint was prepared with the AAS LATEX macros v5.2. 6
astro-ph/9801309
1
9801
1998-01-29T22:06:59
The Stellar Content of Active Galaxies
[ "astro-ph" ]
We present results of a long-slit spectroscopic study of 39 active and 3 normal galaxies. Stellar absorption features, continuum colors and their radial variations are analyzed in an effort to characterize the stellar population in these galaxies and detect the presence of a featureless continuum underlying the starlight spectral component. Spatial variations of the equivalent widths of conspicuous absorption lines and continuum colors are detected in most galaxies. Star-forming rings, in particular, leave clear fingerprints in the equivalent widths and color profiles. We find that the stellar populations in the inner regions of active galaxies present a variety of characteristics, and cannot be represented by a single starlight template. Dilution of the stellar lines by an underlying featureless continuum is detected in most broad-lined objects, but little or no dilution is found for the most of the 20 type 2 Seyferts in the sample. Color gradients are also ubiquitous. In particular, all but one of the observed Seyfert 2s are redder at the nucleus than in its immediate vicinity. Possible consequences of these findings are outlined.
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 000–000 (0000) Printed 6 March 2018 (MN LATEX style file v1.4) The Stellar Content of Active Galaxies Roberto Cid Fernandes Jr.1‡, Thaisa Storchi Bergmann2⋆‡ and Henrique R. Schmitt2†‡ 1 Departamento de F´ısica, CFM - UFSC, Campus Universit´ario - Trindade, Caixa Postal: 476. CEP: 88040-900 Florian´opolis, SC, Brazil. 2 Instituto de F´ısica - UFRGS, Caixa Postal: 15051. CEP: 91501-970 Porto Alegre, RS, Brazil. 6 March 2018 ABSTRACT We present the results of a long-slit spectroscopic study of 39 active and 3 normal galaxies. Stellar absorption features, continuum colors and their radial variations are analyzed in an effort to characterize the stellar population in these galaxies and detect the presence of a featureless continuum underlying the starlight spectral component. Spatial variations of the equivalent widths of conspicuous absorption lines and con- tinuum colors are detected in most galaxies. Star-forming rings, in particular, leave clear fingerprints in the equivalent widths and color profiles. We find that the stellar populations in the inner regions of active galaxies present a variety of characteristics, and cannot be represented by a single starlight template. Dilution of the stellar lines by an underlying featureless continuum is detected in most broad-lined objects, but little or no dilution is found for the most of the 20 type 2 Seyferts in the sample. Color gradients are also ubiquitous. In particular, all but one of the observed Seyfert 2s are redder at the nucleus than in its immediate vicinity. Possible consequences of these findings are briefly outlined. Key words: galaxies:active; galaxies:nuclei; galaxies:Seyfert; galaxies:stellar content; galaxies:general 1 INTRODUCTION Starlight constitutes a substantial fraction of the light gath- ered in optical spectra of active galactic nuclei (AGN), par- ticularly in low luminosity objects such as LINERs and Seyfert galaxies, where the stellar component is often domi- nant. Quantifying and removing the contribution of the stel- lar population is one of the first and most critical steps in the analysis of AGN spectra, as the shape of the resulting continuum is strongly dependent on the adopted stellar pop- ulation spectrum. The most widely used technique for starlight subtrac- tion consists of using an appropriately chosen spectrum of a normal galaxy as a template for the stellar component. The template spectrum is scaled to match as closely as possi- ble the stellar absorption features in the observed spectrum, and the residual after its subtraction is taken as the stellar- free, pure AGN spectrum. This technique has been exten- ⋆ Visiting Astronomer, Cerro Tololo Interamerican Observa- tory, operated by the Association of Universities for Research in Astronomy, Inc., under contract with the National Science Foundation. † CNPq fellow. ‡ E-mail: [email protected] [email protected] (HRS). [email protected] (RCF); (TSB); c(cid:13) 0000 RAS sively applied since it was first introduced by Koski (1978), in his early study of Seyfert 2s (e.g., Stauffer 1982, Phillips, Charles & Baldwin 1983, Malkan & Filippenko 1983, Filip- penko & Sargent 1988, 1983, Miller & Goodrich 1990, Ho, Filippenko & Sargent 1993, Kay 1994, Tran 1995a). Alter- natively, the shape and strength of the starlight component can be estimated by means of stellar population-synthesis techniques. This approach was followed by Keel (1983), who built synthetic templates using a library of stellar spec- tra plus assumptions about the mass function and star- formation history. Synthetic templates can also be obtained combining a spectral library of star-clusters, a method first used by Bica (1988) in his analysis of the stellar populations of normal galaxies. Bonatto, Bica & Alloin (1989), Storchi- Bergmann, Bica & Pastoriza (1990) used this technique to determine the integrated stellar content of the bulge of ac- tive galaxies. The starlight evaluation procedure yields the stellar and AGN fluxes as a function of wavelength. The AGN spec- trum, of course, is not known a priori, but the starlight subtraction is usually done in a way to produce a smooth residual continuum. This assumption is based on the fact that the continuum is essentially featureless in objects like bright Seyfert 1s and QSOs, where the AGN component dominates. It is customary to characterize the relative in- tensity of the starlight and featureless continuum (hereafter 2 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt FC) components by their fractions (f⋆ and fFC) of the to- tal flux at a given wavelength. These fractions depend on the chosen template, the aperture and the actual contrast between the AGN and stellar spectral components. The ef- fect of the FC is to dilute the stellar absorption lines in the nuclear spectrum, and it is precisely the degree of dilution with respect to the starlight template which determines f⋆ and fFC. In this paper we explore a different approach to esti- mate the dilution of absorption lines caused by the presence of a FC. Our method makes use of the radial information available through high signal to noise long-slit spectroscopy of a large sample of objects. By studying the variation of the equivalent width (W) of conspicuous absorption lines as a function of distance to the nucleus, we are able to de- termine how much dilution (if any) occurs in the nucleus with respect to extra-nuclear regions, presumably not af- fected by the FC. In a way, this method amounts to using the off-nucleus spectrum of a galaxy as its own starlight template, a starlight evaluation technique which has been a sucessifully applied in some previous works (Fosbury et al. 1978, Schmitt, Storchi-Bergmann & Baldwin 1994, Winge et al. 1995, Storchi-Bergmann et al. 1997). Continuum color gradients are also examined, complementing the study of absorption line gradients and bringing in information about the reddening of the nuclear regions. A second goal of this paper is to provide an homoge- neous database for the study of the stellar populations in active galaxies. As discussed above, in previous works the interest in the stellar population has focused in removing its contribution from the nuclear spectrum. The Ws and contin- uum colors measured in this work allow a characterization of the global properties of the stellar populations as close as a few hundred of parsecs from the nucleus, offering the prospect of a comparative study between the stellar content of normal and active galaxies and between different types of AGN. This investigation touches upon a number of key is- sues in current AGN research, such as: (1) the question of starlight evaluation and subtraction techniques; (2) the char- acterization of the stellar population in active galaxies; and (3) the nature of the FC in Seyfert 2 galaxies and its impli- cations for unified models. In the current paper we focus on the presentation of the data, the analysis of the stellar pop- ulation features and their radial variations. Future commu- nications will explore the consequences and interpretation of our results in full detail. This paper is organized as follows. In §2 we present the sample galaxies and describe the observations. The method of analysis (continuum determination and W measurement) is discussed in §3. The results are presented in §4 and dis- cussed in §5. Finally, §6 summarizes our main results. 2 OBSERVATIONS The sample comprises 20 Seyfert 2s, 6 Seyfert 1s, 7 LINERs and 5 radio galaxies, as well as 4 'normal' nuclei. With the exception of the radio galaxies, all of them are bright nearby objects. Table 1 describes the sample properties. Long-slit spectra of these galaxies have been collected in three epochs using the same technique and instrumentation such that we have formed an homogeneous database, which can be used for a comparative study of the different kinds of galaxies. The observations were carried out using the Cassegrain Spectrograph with the 4m telescope of the Cerro Tololo Inter-American Observatory. The spectral range covered is ∼ 3500–7000 A, at a resolution of 5–8 A. A slit width cor- responding to 2′′ in the sky was oriented along the posi- tion angle (hereafter PA) which allowed the best coverage of the extended emission (although in this work we concen- trate on the properties of the continua and stellar absorp- tion features). A log of the observations is presented in Table 2, where we also list the parallactic angle, air mass during the observations, pixel size and spatial scale (evaluated with H0 = 75 km s−1 Mpc−1). It can be noticed that, only for a few galaxies, namely 3C33, CGCG420-015 and NGC 4303, the effect of differential refraction is important, displacing the light at 3700 A by ∼ 1′′ relative to the light at 5000 A, so that some distortion of the continuum slope can be expected in the individual extractions for these objects, if the continuum presents significant variations along the slit. The reduction of the spectra was performed using standard techniques in IRAF. After flux and wavelength calibration, the redshifts were determined from the narrow emission lines present and the spectra were converted to rest-frame units. The long-slit used corresponded to 5′ in the sky, and sky subtraction was performed by the fit of a polynomium to the outer regions of the frame and then subtracting its contribution to the hole frame using the task background in IRAF. The most extented galaxies, presenting enough S/N ratio to allow measure the spectral features up to approx- imately 100′′, were NGC1097, the most extended, followed by NGC1672. For these galaxies the sky level adopted may have been overestimated. In order to evaluate an upper limit to the effect, we have compared the flux level of the adopted sky to the flux of the inner regions. We have concluded that its value amounts to less than one percent when compared with the flux of the nucleus, but rising with distance from the nucleus. At approximately 25′′, its contribution is about 10%. Assuming that the 'sky' flux is actually galaxy flux, the flux level of the sky-subtracted spectrum will be under- estimated by at most 10 percent at 25′′ and more outwards. Thus the measurements of these two galaxies are subjected to this uncertainty farther than 25′′ from the nucleus. The effect should be negligible for the other less extended galax- ies. One-dimensional spectra were extracted from the two- dimensional frames. The number of extractions varied from 4 for PKS0745-19 to as many as 26 for NGC 1097, totaling 491 spectra for the whole sample. The spatial coverage ranged between ±3 and ±80′′ from the nucleus, while the width of the extractions was typically 2′′ in the brighter nuclear re- gions, gradually increasing towards the fainter outer regions to guarantee enough signal. Given that the seeing during the observations was typically 1.5′′, some of the narrow extrac- tions are slightly oversampled, resulting in a smoothing of the sharpest spectral variations. The S/N ratio at 5650 A ranges from ∼ 5 to 100, with an average of 40 for all extractions. Even in the blue region, where the noise is usually higher, the spectra are of excellent quality, with a mean S/N ratio of 25 at 4200 A. The high quality of our data has allowed the measurement of stellar population features in spectra extracted up to several kpc's c(cid:13) 0000 RAS, MNRAS 000, 000–000 NAME ALT. NAME α (1950) δ (1950) TYPE MORPH. v BT o E(B-V)G SAMPLE PROPERTIES The Stellar Content of Active Galaxies 3 2 SA(s)0/a NLRG E 4669 17230 5762 5161 5086 1193 4792 1244 2716 3980 741 920 19190 4939 1155 8811 4616 3603 10308 16320 35940 3910 2072 2164 8670 4047 7263 1486 4386 4772 3959 1128 1066 1368 997 1676 2473 4385 2459 4850 1767 1551 13.94 19.50 14.50 13.57 13.66 9.92 14.15 11.25 13.31 12.70 12.12 11.37 16.80 13.44 10.25 15.00 13.51 13.58 15.94 17.58 18.00 14.21 11.81 12.59 13.89 13.00 13.71 13.68 15.00 13.81 12.37 10.63 10.23 9.77 10.20 10.96 12.06 13.68 12.82 12.88 11.13 10.83 0.060 0.025 0.000 0.008 0.000 0.020 0.000 0.000 0.018 0.025 0.000 0.000 0.000 0.000 0.000 0.070 0.005 0.000 0.225 0.405 - 0.018 0.028 0.033 0.005 0.088 0.063 0.000 0.015 0.185 0.058 0.000 0.125 0.220 0.120 0.150 0.045 0.020 0.008 0.000 0.000 0.000 Mrk 348 3 C 33 NGC 526a Mrk 573 IC 1816 NGC 1097 ESO 417-G6 NGC 1326 Mrk 607 NGC 1358 NGC 1386 NGC 1433 Pks 0349-27 NGC 1598 NGC 1672 CGCG 420-015 ESO 362-G8 ESO 362-G18 PICTOR A Pks 0634-20 Pks 0745-19 Mrk 1210 NGC 2935 NGC 3081 Mrk 732 IRAS 11215-2806 MCG-05-27-013 NGC 4303 MCG-02-33-034 FAIRALL 316 NGC 5135 NGC 5248 NGC 5643 NGC 6221 NGC 6300 NGC 6814 IC 4889 NGC 6860 NGC 6890 NGC 7130 NGC 7213 NGC 7582 NGC 262 Pks 0106+13 ESO 352-IG66 UGC 1214 ESO 355-G25 ESO 416-G20 MCG-05-08-006 ESO 357-G026 NGC 1320 MCG-1-10-003 ESO 358-G35 ESO 249-G014 GSP 022 ESO 202-G26 ESO 118-G043 MCG-06-12-009 MCG-05-13-017 ESO 252-GA018 IRAS 06343-2032 UGC 4203 ESO 565-G23 ESO 499-IG31 IC 2637 UGC 7420 ESO 269-G12 ESO 444-G32 UGC 8616 ESO 272-G16 ESO 138-G03 ESO 101-G25 MCG-02-50-001 ESO 185-G14 ESO 143-G09 MCG-07-41-023 IC 5135 ESO 288-G43 ESO 291-G16 00 46 05 01 06 12 01 21 37 01 41 23 02 29 48 02 44 11 02 54 18 03 22 01 03 22 18 03 31 11 03 34 51 03 40 27 03 49 32 04 27 08 04 44 55 04 50 47 05 09 20 05 17 44 05 18 24 06 34 24 07 45 17 08 01 27 09 34 26 09 57 10 11 11 14 11 21 35 11 24 55 12 19 22 12 49 35 12 53 50 13 22 57 13 35 02 14 29 28 16 48 26 17 12 18 19 39 56 19 41 18 20 04 29 20 14 49 21 45 20 22 06 08 23 15 38 Lin Lin 2 2 1 1.9 2 2 Lin-1 2 Lin 2 2 2 Lin 31 41 04 13 02 33 -35 19 32 02 05 56 -36 53 29 -30 29 01 -32 23 00 -36 38 24 -03 13 03 -05 15 24 -36 09 47 -47 22 48 -27 53 29 -47 53 29 -59 20 18 03 58 47 -34 27 12 -32 42 30 -45 49 43 -20 32 19 -19 10 15 05 15 22 2 -20 54 12 Normal -22 35 06 2 1.5 09 51 33 2 -28 96 46 2 -28 59 00 Lin 04 45 03 -13 08 36 1 2 -46 39 18 -29 34 26 2-SB 09 08 23 Normal 2 -43 57 12 -59 08 00 SB 2 -62 45 54 -10 26 33 1 -54 27 54 Normal 1 -61 14 42 -44 57 48 2 2-SB -35 11 07 Lin-1 -47 24 45 -42 38 36 2-SB NLRG E BLRG SA00:pec NLRG E NLRG E S? SAB(s)b SAB(r)0/a E pec S0 pec SAB(rs)0+ SA:(r:)a SB(rl)b (R)SA0/a SB(rl)0/a Sa:sp SAB(r)0/a SB(s)0+ SB(rs)ab SAB(s)c SB(r)bc E/S0 S0? S0/a SB(r)a? SAB(rs)bc S0? SB(l)ab SAB(rs)bc SAB(rs)c SB(s)bc pec SB(rs)b SAB(rs)bc E SB(r)ab SB(r)ab Sa pec SA(s)00 SB(s)ab Table 1. Column (5) gives the activity class of the galaxy, where SB means Starburst. Column (7) gives the radial velocity relative to the local group (in km s−1), while column (8) lists the total blue magnitude of the galaxies and column (9) contains the foreground galactic value of E(B-V) (extracted from the Nasa Extragalactic Database). from the nuclei. There is no similar database with this qual- ity for so many galaxies in the literature. 3 METHOD OF ANALYSIS Fig. 1 shows representative spectra of four of the sam- ple galaxies for several positions along the slit. Notice that absorption features are well detected in all spectra, as well as variations of spectral characteristics with distance from the nuclei. Signatures of young stellar populations, in par- ticular, are clearly present in several of the spectra shown: over the whole inner 8′′ of Mrk 732, more conspicuously at 4′′ NE (r = −4′′ in Fig. 1); at ∼ 8′′ from the nucleus in IC 1816, and at ∼ 8–12′′ in NGC 1097 (corresponding to the locus of the nuclear ring; Storchi-Bergmann, Wilson & Baldwin 1996). Further examples of the spatially resolved spectra are shown in Fig. 45. c(cid:13) 0000 RAS, MNRAS 000, 000–000 Our main goal is to study the run of the stellar absorption features with distance from the nucleus. To this end, the analysis of each of the sample spectra consisted of (1) de- termining the 'pseudo'-continuum in selected pivot-points (3780, 4020, 4510, 4630, 5313, 5870 and 6080 A) and (2) measuring the equivalent widths of the CaII K and H, CN, G-band, MgI+MgH and NaI absorption lines integrating the flux with respect to the pseudo continuum. The continuum and W measurements followed the method outlined by Bica & Alloin (1986a,b), Bica (1988) and Bica, Alloin & Schmitt (1994), and subsequently used in several studies of both normal and emission line galaxies (e.g., Bonatto, Bica & Alloin 1989, Storchi-Bergmann, Bica 4 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt LOG OF OBSERVATIONS NAME DATE EXP. TIME (sec) PA [◦] AIR φ [◦] PIXEL SCALE COMMENTS MASS pc/arcsec Mrk 348 3 C 33 NGC 526a Mrk 573 IC 1816 NGC 1097 ESO 417-G6 NGC 1326 Mrk 607 NGC 1358 NGC 1386 NGC 1433 Pks 0349-27 NGC 1598 NGC 1672 CGCG 420-015 ESO 362-G8 ESO 362-G18 PICTOR A Pks 0634-20 Pks 0745-19 Mrk 1210 NGC 2935 NGC 3081 Mrk 732 IRAS 11215-2806 MCG-05-27-013 NGC 4303 MCG-02-33-034 FAIRALL 316 NGC 5135 NGC 5248 NGC 5643 NGC 6221 NGC 6300 NGC 6814 IC 4889 NGC 6860 NGC 6890 NGC 7130 NGC 7213 NGC 7582 6/7 Dec. 94 6/7 Dec. 94 5/6 Dec. 94 6/7 Dec. 94 6/7 Jan. 94 6/7 Jan. 94 6/7 Dec. 94 5/6 Jan. 94 6/7 Jan. 94 6/7 Jan. 94 6/7 Jan. 94 6/7 Dec. 94 5/6 Dec. 94 6/7 Jan 94 5/6 Jan 94 6/7 Jan 94 6/7 Jan. 94 6/7 Jan. 94 6/7 Dec. 94 5/6 Dec. 94 6/7 Dec. 94 5/6 Jan. 94 5/6 Jan. 94 28/29 May 92 6/7 Jan. 94 5/6 Jan. 94 6/7 Jan. 94 28/29 May 92 6/7 Jan. 94 6/7 Jan. 94 29/30 May 92 28/29 May 92 28/29 May 92 28/29 May 92 28/29 May 92 29/30 May 92 28/29 May 92 28/29 May 92 29/30 May 92 29/30 May 92 29/30 May 92 6/7 Jan. 94 1800 1800 3600 1800 1800 1800 3600 1800 900 900 1800 1200 3600 1800 1800 900 1800 1800 1800 3600 1800 1800 1800 1200 1800 900 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 900 1200 300 170 65 124 125 90 139 155 77 135 145 169 19 82 123 94 40 165 55 71 139 150 163 0 73 55 145 0 112 25 100 30 146 90 5 124 161 0 70 153 143 50 67 2.20 1.43 1.02 1.64 1.07 1.01 1.01 1.15 1.22 1.23 1.12 1.05 1.04 1.13 1.32 1.46 1.17 1.19 1.15 1.07 1.03 1.25 1.02 1.25 1.50 1.02 1.20 1.69 1.24 1.29 1.25 1.58 1.19 1.24 1.19 1.06 1.12 1.21 1.03 1.01 1.04 1.53 170 167 69 131 82 88 69 88 140 136 89 19 103 65 63 140 88 82 71 129 149 163 182 70 31 76 78 50 55 102 100 144 75 56 18 170 29 36 12 50 0 83 0.9 0.9 0.9 0.9 1 1 0.9 1 1 1 1 0.9 0.9 1 1 1 1 1 0.9 0.9 0.9 1 1 0.9 1 1 1 0.9 1 1 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 0.9 1 302 1114 372 334 328 77 310 80 176 257 48 59 1240 319 75 570 298 233 666 1055 2323 253 134 140 560 262 470 96 284 308 256 73 69 88 63 108 160 283 159 314 114 100 seeing 3′′ clouds clouds clouds clouds clouds clouds Table 2. Details of the observations. Column (4) lists the slit position angle (PA). The parallactic angle (φ) is listed in column (6). Column (7) lists the pixel size in arcsec, while column (8) lists the spatial scale in pc arcsec−1. & Pastoriza 1990, Jablonka, Alloin & Bica 1990, Storchi- Bergmann, Kinney & Challis 1995b, McQuade, Calzetti & Kinney 1995). This method consists of determining a pseudo-continuum in a few pivot-wavelengths and integrat- ing the flux difference with respect to this continuum in pre-defined wavelength windows (Table 3) to determine the Ws. The pivot wavelengths used in this work are based on those used by the above authors and were chosen to avoid, inasmuch as possible, regions of strong emission or absorp- tion features. The use of a compatible set of pivot points and wavelength windows is important because it will allow a detailed quantitative analysis of the stellar population- via synthesis techniques using Bica's (1988) spectral library of star clusters-to be presented in a forthcoming paper. Here we will use our measurements only to look for broad LINE WINDOWS CaII K CaII H+Hǫ CN G band MgI+MgH NaI 3908–3952 3952–3988 4150–4214 4284–4318 5156–5196 5880–5914 Table 3. Wavelength windows used to measure the equivalent widths of the absorption lines. trends in the properties of the stellar populations, and to identify signs of a nuclear FC (see below). The determination of the continuum has to be done c(cid:13) 0000 RAS, MNRAS 000, 000–000 The Stellar Content of Active Galaxies 5 Figure 1. Selected examples of the spatially resolved spectra of four of the sample galaxies. The nuclear spectra are indicated by thick lines. The angular distances from the nuclei are indicated, positive values corresponding to West (W, NW or SW) or South direction (in the case when the observed PA is 0◦). The spectra are normalized at 5870 A and shifted vertically by one unit to facilitate comparison. Residual sky emission lines have been ticked out. interactively, taking into account the flux level, the noise and minor wavelength calibration uncertainties as well as anomalies due to the presence of emission lines. The 5870 and 6080 A points, in particular, are sometimes buried un- derneath HeI 5876 and [Fe VII]6087 A emission lines (for instance, in the nuclear regions of IC 1816, Mrk 348 and NGC 3081). In such cases the placement of the continuum was guided by adjacent wavelength regions. In a few cases, it was also necessary to make 'cosmetic' corrections when a noise spike was present in the wavelength window used to compute the W. We emphasize, however, that such anoma- lies were very rare. In the majority (> 90%) of cases the excellent quality of the spectra allowed a precise determina- tion of the continua and Ws. Fig. 2 illustrates the application of the method to four of the sample spectra. The spectrum in Fig. 2a corresponds c(cid:13) 0000 RAS, MNRAS 000, 000–000 6 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt the numerous broad lines and intense non-stellar continuum complicate the analysis. This is illustrated in Fig. 2d, where we plot the nuclear spectrum of MCG -02-33-034. The 4510, 4630 and 5313 A points, in particular, are all underneath Fe II blends, making it impossible to determine them accu- rately. None of the absorption lines in Fig. 2d, with the possi- ble exception of Ca K is free of contamination. The Mg line, for instance, is filled up by FeII and [NI]5200 A emission, while the Na window is contamined by broad HeI 5876 A. It is clear that in extreme cases like this no reliable value for the Ws of the absorption lines can be derived with our or any other method. The presence of such a strong nuclear FC should appear as a dip at r = 0 in the radial profiles of the Ws, and this is indeed seen in Fig. 7. However, given the difficulties in defining a continuum, the depth of this dip is unreliable. At any rate, the only other galaxy in the present sample with a spectrum as complex as MCG -02-33-034 is the Broad Line Radio Galaxy Pictor A (Fig. 32). For less extreme Seyfert 1s like NGC 6814 (Fig. 2c), the continuum can be defined with little ambiguity. The four examples in Fig. 2 span essentially all types of spectra found in the sample. Overall, we found that the method works well for most spectra, with the exception of the nuclear regions of strong Seyfert 1s. The comparison of the Ws and continuum colors in the nuclear and outer re- gions is a useful tool to detect the presence of a FC, partic- ularly in Seyfert 2s and LINERs, where the FC, if present, is not as conspicuous as in type 1 objects. Indeed, one of our major goals is to verify whether Seyfert 2s and LINERs exhibit signs of such a nuclear FC, which should cause a di- lution of the absorption lines with respect to those measured outside the nucleus. 3.1 Consistency and Error Analysis The positioning of the continuum can be somewhat subjec- tive, particularly in the noisier spectra, given that it has to be carried out interactively. In order to verify whether our measurements are consistent with those of Bica (1988) we have measured the continuum fluxes and Ws of the 15 templates corresponding to his spectral groups E1–E8 and S1–S7. The difference between his and our continuum mea- surements is of the order of 1%. For the Ws the difference is typically ≈ 0.5 A and always smaller than 1 A, correspond- ing to an agreement at a 5–10% level. This compatibility is important, since it insures that we can apply stellar popu- lation characterization techniques similar to those employed by Bica. In the discussion of our results (Section 4) we will com- pare the Ws and continua of the galaxies in the present sample with templates S1 to S7 and E1 to E8 of Bica (1988) as a guide to obtain a broad characterization of the stellar populations. (As pointed out in the previous section, a full quantitative analysis of the stellar populations will be pre- sented in a forthcoming paper.) These templates correspond to increasing contributions of young components, from S1 and E1, to S7 and E8. Templates S1 to S3 and E1 to E6 are composed of old (> 5 × 109 yr) and intermediate age stars (5 × 108–109 yr), differing mostly on the contribution from stars of different metallicities. From S4 to S7 and E7 to E8 we have an increasing contribution from young stars (< 108 c(cid:13) 0000 RAS, MNRAS 000, 000–000 Figure 2. Illustration of the continuum determination procedure for the nuclear spectra of the LINER NGC 1598 (a), the Seyfert 2 Mrk 348 (b), and the Seyfert 1 galaxies NGC 6814 (c) and MCG -02-33-034 (d). The continuum pivot points are marked by the filled circles. Dashed vertical lines indicate the wavelength windows used to measure the equivalent widths of Ca K and H, CN, G band, Mg and Na lines. to the nucleus of the LINER NGC 1598. The spectrum is almost purely stellar, posing no difficulties to the determina- tion of the pseudo continuum. Note that the continuum runs well above the actual spectrum except in the pivot points. Fig. 2b shows the nuclear spectrum of Mrk 348, a type 2 Seyfert. Here, as for all Seyfert 2s in the sample, the con- tinuum can still be placed with little ambiguity, despite the presence of many emission lines. As in many other cases, the Ca H line is filled with [NeIII] and Hǫ emission in the central regions, but not in outer positions, an effect which must be taken into account when interpreting W(Ca H) spatial gra- dients. Note also that, as in Fig. 2d, the 5870 and 6080 A points are underneath HeI and [Fe VII] emission, respec- tively. This is not a problem, as the continuum can still be determined from the adjacent regions. Whilst for most LINERs, Seyfert 2s and normal galax- ies the placement of the continuum is straightforward, this is not the case in the nuclear regions of strong Seyfert 1s, where yr), which can reach as much as 65% for S7 and 20% for E8 of the flux at 5870 A. The Ws measured in this work are subjected to two different sources of error: (1) the noise present in the spectra, and (2) the uncertainty in the placement of the continuum. The error in the Ws due to noise was evaluated using standard propagation of errors. The noise level was com- puted from the rms dispersion in a 60–100 A region free of strong emission or absorption lines. Two windows, one in the 4200 A region and the other around 5650 A, were used for this purpose. The noise in the first window was used to compute the errors in the Ws of Ca K and H, CN and G-band, whereas the latter window was used for the Mg and Na lines. The resulting errors are typically 0.5 A for the 'blue' lines (Ca K to G-band), and 0.3 A for Mg and Na. The errors in the central regions are usually smaller than these average values, since the noise is lower there. The Ca K line in NGC 7213, for instance, has an error of ∼ 0.3 A in the central 5′′, gradually increasing to ∼ 0.7 A for r > 22′′. The uncertainties in the placement of the continuum are less straightforward to evaluate, since, as explained above, the continuum is not automatically defined, but determined interactively. To compute the error in the Ws associated with this source of uncertainty we have re-analyzed the spec- tra of 4 galaxies (42 extractions in total), defining upper and lower values of the continuum at the pivot-points. The mean difference between these values and the adopted one (all evaluated in the middle of the line-window), was taken as a measure of the error in the continuum flux, which was then propagated to the Ws. We found that the resulting errors are typically 30–50% larger than the errors due to noise alone. As expected, the uncertainty in the position- ing of the continuum increases with decreasing S/N. Since it is not practical to measure lower and upper continua for all 491 spectra, we have used this scaling between the S/N ratio and the continuum errors measured above to estimate the errors due to the positioning of the continuum for the remaining objects. The combined error in the Ws due to noise and contin- uum positioning, added in quadrature, are typically 0.5 A for Ca K, Ca H and the G-band, 0.4 A in Mg and Na, and 1.0 A for the CN line. The larger error in the CN reflects the fact that this feature is very shallow, being essentially a measure of the local height of the pseudo continuum above the actual spectrum. In fact, it is often difficult to identify the CN feature in the spectra (e.g., Fig. 2). Indeed, the CN band is only pronounced in very metal rich populations, such as templates S1 and E1 of Bica (1988). We nevertheless keep this line in our analysis, since our measured Ws are consis- tent with those found by Bica and in most cases its radial variations follow those of the other absorption lines. In the discussion of our results, we shall put more em- phasis on the Ca K, G-band and Mg features, since the Ca H line is often contaminated by emission and the CN line is subjected to larger errors. The Na line is not a good diag- nostic of stellar population since it is partially produced in the interstellar medium of the host galaxy. This line can, however, be used as an indication of the presence of dust (Bica & Alloin 1986, Bica et al. 1991). In the atribution of template types to the different spectra a larger weight is given to Ca K, since this is often the strongest line, less affected by errors. As discussed in section 4.1, sometimes c(cid:13) 0000 RAS, MNRAS 000, 000–000 The Stellar Content of Active Galaxies 7 different lines indicate different template types, which sim- ply indicates a different mixture of stellar populations than found in Bica's (1988) grid of spectral templates. The uncertainty in the determination of the Ws limits the degree of dilution measurable from our data. The effect of a FC contributing a fraction fFC of the total flux in a given wavelength is to decrease the W of an absorption line in this wavelength by this same factor: ∆W = fFCW. There- fore, the minimum amount of dilution that can be safely measured from the radial variations of a line with an W of, say, 10 A and an error of 0.5 A is ∼ 5%. From the results presented in the next section we estimate that this method of evaluating the contribution of a FC works well for fFC ∼ 10% for the present data set as a whole, though smaller lower limits are reached in the best spectra. > 4 RESULTS The main results of the data analysis are presented in Figs. 3–44, where we plot the radial variations of the Ws, colors and surface brightness for all the galaxies in the sam- ple. The upper panel shows the Ws of Ca K (solid line) and Ca H (dashed) against the distance from the nucleus (r). Similarly, the second panel shows the Ws of the G-band (solid) and CN (dashed), while the third panel shows Mg (solid) and Na (dashed). Negative values of the W indicate emission, as often found in the Ca H plots due to contam- ination by [NeIII] and Hǫ. The Mg and Na lines are also contaminated by emission in some cases. Apart from such cases (discussed in detail in Section 4.1), a dip in the W plots indicates dilution of the absorption lines by a FC or the presence of a young stellar population. The fourth panel shows the color profiles, mapped by the ratios of the 5870 to 4020 A (solid line) and 4630 to 3780 A (dashed line) pseudo- continuum fluxes. (Note that the left hand scale in this plot corresponds to F5870/F4020, whereas the values in right cor- respond to F4630/F3780.) Dips in the color plots indicate a blue continuum, whereas peaks indicate a red continuum, due to an old stellar population and/or reddening. The col- ors have been corrected by foreground galactic reddening using the values of E(B-V) in Table 1. This correction, be- sides being small, does not affect the shape of the curves. Finally, to give a measure of the flux level in each position the bottom panels show the run of the surface brightness in the 5870 A pseudo continuum with distance from the nu- cleus. In order to better illustrate the results presented in Figs. 3–44, Fig. 45 shows the spatially resolved spectra of five galaxies, representing the five classes of objects studied, zoomed into the wavelength regions of the absorption lines analysed in this work. This figure also serves as a further illustration of the determination of the pseudo-continuum and of the quality of the current database. We start our discussion of the results with the cases of NGC 1097, NGC 1433 and NGC 1672. The presence of star- forming rings in these galaxies serves as a good illustration of the consistency of our measurements and as a guide to the interpretation of the W and color gradients in other objects. NGC 1326, 2935, 4303 and 5248 also present star-forming rings, and much of the the discussion below also applies to these objects. 8 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt 15 10 5 15 10 5 10 8 6 4 2 1.5 1 0.6 0.4 0.2 0 3 2.5 2 1.5 -50 0 50 Figure 3. Radial variations of the equivalent widths (W), colors and 'surface brightness' for NGC 1097. The upper panel plots the Ws of CaII K (solid line) and CaII H (dashed), while the second panel shows G-band (solid), CN (dashed), and the third shows MgI+MgH (solid) and NaI (dashed). The fourth panel shows the radial variations of the continuum colors, mapped with the F5870/F4020 (solid line, left hand scale) and F4630/F3780 (dashed line, right hand scale) flux ratios. Red is up and blue is down in this plot. The bottom panel shows the run of the surface bright- ness at 5870 A (in units of erg cm−2 s−1 A−1per arcsec2) along the slit. NGC 1097, a nearby spiral galaxy, has a low luminosity nucleus, characterized by a LINER like emission line spec- trum, surrounded by a ring of HII regions (Sersic & Pastoriza 1965, 1967, Phillips et al. 1984). In addition to its LINER spectrum, the broad, double peaked, variable Balmer lines in the nucleus of NGC 1097 are a clear signature of the pres- ence of an active nucleus in this galaxy (Storchi-Bergmann, Baldwin & Wilson 1993; Fig. 1). The low Ws and blue col- ors typical of the young stellar populations of the ring are remarkably well depicted in Fig. 3 by the dips at r ∼ 10′′ in the radial profiles of the Ws and colors. Despite its ac- tive nucleus, the optical continuum shows little evidence of a blue featureless continuum of the kind found in type 1 Seyferts and QSOs. This is seen in Fig. 3, which shows that the Ws in the nucleus of NGC 1097 are not significantly di- luted with respect to their values inside or outside the ring. On the contrary, the CN and Mg lines seem to be stronger in the nucleus than outside it, while the Na line clearly peaks at r = 0. The Ca K and H lines, however, do present a cen- tral dip, albeit of a low amplitude. Comparing the nuclear spectrum with the r = −4 and +4′′ extractions we find vari- ations of 7 and 11% in the Ws of Ca K and H respectively (§5.2), which indicates that an AGN continuum, if present, contributes less than ∼ 10% of the light in this wavelength region. Considering that the error in these Ws is ∼ 5%, the evidence for a diluting FC in NGC 1097 is marginal. The Ws in NGC 1097 can be represented by a S2–S3 template at the nucleus, S6–S7 at the ring and S3 outwards. The continuum ratios are consistent with the spectral tem- plates inferred from the Ws, with the exception of the nu- clear region, where the continuum is redder, more typical of a S1 template. This could be interpreted as evidence of reddening of the nuclear regions, as is also indicated by the increase of W(Na) towards the nucleus (Storchi-Bergmann et al. 1995a). This difference in template types can be ac- counted for with E(B − V ) ≈ 0.15. Storchi-Bergmann, Kin- ney & Challis (1995b) obtained a large aperture spectrum of this object (10′′×20′′), including the nucleus and part of the starforming ring. Their measured Ws can be represented by a S5 template, which is comparable to our results in the inner 10′′. The IUE spectrum of NGC1097 is dominated by the ring, showing several absorption lines typical of young stars (Kinney et al. 1993). NGC 1433 has been classified as a Seyfert 2 by V´eron- Cetty & V´eron (1986), though our spectrum presents emis- sion line ratios more typical of LINERs. It also contains a star-forming ring ∼ 8′′ from its nucleus (Buta 1986). As in the case of NGC 1097, the ring leaves a clear finger print in the W and color profiles (Fig. 36). No dilution of the absorption lines or blueing of the continuum is seen in the nuclear spectrum. Both the Ws and continuum ratios indi- cate similar spectral templates: S3 at the nucleus, S5 at the ring and S3 outwards. As in NGC 1097, the general decrease of the Ws and bluer colors towards large distances seen in Fig. 36 correspond to the stellar population of the disk of the galaxy. NGC 1672 is a Sersic-Pastoriza galaxy, whose nucleus has already been classified as Starburst (Garcia-Vargas et al. 1990, Kinney et al. 1993), Seyfert 2 (Mouri et al. 1989) and LINER (D´ıaz 1985, Storchi-Bergmann et al. 1996b). Our nu- clear spectrum is consistent with a LINER classification. Its W and color profiles (Fig. 38) are very similar to those of the previous two galaxies. Both Ws and continuum ratios indi- cate a S4 template at the nucleus, S7 at the r = 5′′ ring, S4 outside the ring, decreasing to values typical of S7 template in the outer regions. The spatio-spectral variations in NGC 1672 are also seen in Fig. 45a. The correspondence between Figs. 38 and 45a is clear. The star forming ring, with its shallower ab- sorption lines and bluer continuum, is visible in the r = −8, +4 and +8′′ extractions plotted in Fig. 45a. The blueing of the spectra towards the outer regions of the galaxy is also illustrated. The +67′′ extraction, in particular, crosses an HII region. These three examples demonstrate that the method em- ployed in this study is able to detect variations of the stellar populations as a function of distance to the nucleus. Star- forming rings, in particular, leave unambiguous signatures in the W and color profiles. Since "dilution" of the metal lines by a young stellar population is detectable, we also expect to be able to mea- sure dilution by an AGN continuum. This is best illustrated in the case of Seyfert 1 objects, such as MCG-02-33-034 (Fig. 7), ESO 362-G18 (Fig. 5), NGC 6860 (Fig. 9), NGC 6814 (Fig. 8), and the Broad Line Radio Galaxy Pictor A (Fig. 32). c(cid:13) 0000 RAS, MNRAS 000, 000–000 15 10 5 0 12 10 8 8 6 4 2 2 1.8 1.6 1.4 0.1 0.08 0.06 0.04 0.02 0 14 12 10 8 6 4 12 10 8 6 4 2 6 4 2 0 -2 1.1 1 0.9 0.8 0.4 0.2 0 -10 -5 0 5 10 Figure 4. Same as Fig. 3. -10 0 10 The Stellar Content of Active Galaxies 9 12 10 8 6 4 2 10 8 6 4 2 7 6 5 4 3 2 1.2 1.1 1 0.9 0.8 0.3 0.2 0.1 0 15 10 5 0 10 5 0 10 5 0 1.8 1.6 1.4 1.2 1 0.8 0.5 0.4 0.3 0.2 0.1 0 2.4 2.2 2 1.8 1.6 1.4 1.4 1.3 1.2 1.1 1.5 1.4 1.3 1.2 1.1 2 1.5 1 -5 0 5 10 Figure 6. Same as Fig. 3. -15 -10 -5 0 5 Figure 5. Same as Fig. 3. Figure 7. Same as Fig. 3. c(cid:13) 0000 RAS, MNRAS 000, 000–000 10 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt 15 10 5 10 8 6 4 2 8 6 4 2 2 1.8 1.6 1.4 1.2 50 40 30 20 10 0 25 20 15 10 5 15 10 5 8 6 4 2 0 -2 1.4 1.3 1.2 1.1 0.15 0.1 0.05 0 10 5 0 10 9 8 7 6 4 2 1.6 1.5 1.4 1.3 1.2 0.06 0.04 0.02 2.4 2.2 2 1.8 1.6 1.4 -20 0 20 -5 0 5 Figure 8. Same as Fig. 3. Figure 10. Same as Fig. 3. 20 15 10 5 0 12 10 8 6 8 6 4 2 1.6 1.4 1.2 0.15 0.1 0.05 0 1.8 1.6 1.4 -10 0 10 -5 0 5 10 Figure 9. Same as Fig. 3. Figure 11. Same as Fig. 3. c(cid:13) 0000 RAS, MNRAS 000, 000–000 1.8 1.6 1.4 2.2 2 1.8 1.6 1.4 -10 0 10 Figure 12. Same as Fig. 3. The Stellar Content of Active Galaxies 11 20 15 10 12 10 8 6 8 6 4 2 1.8 1.6 1.4 0.3 0.2 0.1 0 25 20 15 10 20 18 16 14 12 10 10 8 6 4 2 1.9 1.8 1.7 1.6 0.4 0.3 0.2 0.1 0 2 1.8 1.6 1.4 1.2 2.2 2 1.8 1.6 -20 0 20 Figure 14. Same as Fig. 3. 2.2 2 1.8 1.6 2.8 2.6 2.4 2.2 2 15 10 5 0 10 8 6 8 6 4 2 1.6 1.4 1.2 1 0.3 0.2 0.1 0 14 12 10 8 6 12 10 8 6 8 6 4 2 1.7 1.6 1.5 1.4 0.2 0.15 0.1 0.05 0 -10 0 10 -10 0 10 Figure 13. Same as Fig. 3. Figure 15. Same as Fig. 3. c(cid:13) 0000 RAS, MNRAS 000, 000–000 12 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt 16 14 12 10 8 12 10 8 6 8 6 4 2.2 2 1.8 1.6 1.4 0.8 0.6 0.4 0.2 0 -40 15 10 5 0 12 10 8 8 6 4 2.2 2 1.8 1.6 1.4 0.2 0.15 0.1 0.05 12 10 8 8 7 6 5 7 6 5 4 3 1.3 1.2 1.1 1 0.8 0.6 0.4 0.2 0 2.5 2 -20 0 20 -10 -5 0 5 10 Figure 16. Same as Fig. 3. Figure 18. Same as Fig. 3. 10 5 0 10 8 6 8 6 4 2 0 -2 1.3 1.25 1.2 1.15 1.1 0.25 0.2 0.15 0.1 0.05 2.6 2.4 2.2 2 1.8 -5 0 5 -5 0 5 Figure 17. Same as Fig. 3. Figure 19. Same as Fig. 3. c(cid:13) 0000 RAS, MNRAS 000, 000–000 2.2 2.1 2 1.9 1.8 1.7 1.8 1.7 1.6 1.5 The Stellar Content of Active Galaxies 13 20 15 10 5 0 16 14 12 10 10 8 6 4 2.6 2.4 2.2 2 1.8 1.6 0.08 0.06 0.04 0.02 2.15 2.1 2.05 2 1.95 1.9 -5 0 5 10 -5 0 5 Figure 20. Same as Fig. 3. Figure 22. Same as Fig. 3. 20 18 16 14 12 15 14 13 12 11 10 12 10 8 6 4 1.9 1.8 1.7 1.6 0.2 0.15 0.1 0.05 2 1.9 1.8 1.7 2.6 2.4 2.2 2 1.8 2.6 2.4 2.2 2 1.8 14 12 10 8 6 10 8 6 8 6 4 1.8 1.7 1.6 1.5 0.08 0.06 0.04 0.02 20 18 16 14 12 10 12 10 8 6 6 4 2 1.5 1.45 1.4 1.35 1.3 0.2 0.15 0.1 0.05 -10 -5 0 5 -4 -2 0 2 4 Figure 21. Same as Fig. 3. Figure 23. Same as Fig. 3. c(cid:13) 0000 RAS, MNRAS 000, 000–000 14 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt 12 10 8 6 4 8 6 4 2 0 -2 6 4 2 1.2 1.1 1 0.9 15 10 5 15 10 5 0 10 8 6 4 8 6 4 2 1.8 1.6 1.4 1.2 40 30 20 10 0 -5 0 5 Figure 24. Same as Fig. 3. 18 16 14 12 10 8 6 4 2 6 4 2 3 2.5 2 1.5 1 25 20 15 10 15 10 5 8 6 4 2 0 -2 8 6 4 2 2 1.9 1.8 25 20 15 10 1.8 1.6 1.4 1.2 2.2 2 1.8 1.6 1.4 -10 -5 0 5 10 Figure 26. Same as Fig. 3. 2.5 2 1.5 2.2 2.1 2 1.9 -20 -10 0 10 -2 0 2 Figure 25. Same as Fig. 3. Figure 27. Same as Fig. 3. c(cid:13) 0000 RAS, MNRAS 000, 000–000 The Stellar Content of Active Galaxies 15 15 10 5 14 12 10 8 10 8 6 4 2 1.8 1.6 1.4 0.025 0.02 0.015 0.01 0.005 1.6 1.4 1.2 1 -10 -5 0 5 10 -5 0 5 Figure 28. Same as Fig. 3. Figure 30. Same as Fig. 3. 20 15 10 12 10 8 12 10 8 6 4 2 2 1.8 1.6 0.03 0.02 0.01 2 1.9 1.8 1.7 1.6 2.4 2.2 2 1.8 2.4 2.2 2 1.8 10 5 0 6 4 2 0 -2 6 4 2 1.2 1.1 1 0.9 40 30 20 10 15 10 5 12 10 8 6 4 10 8 6 4 1.7 1.6 1.5 1.4 0.5 0.4 0.3 0.2 0.1 -5 0 5 0 -5 0 5 Figure 29. Same as Fig. 3. Figure 31. Same as Fig. 3. c(cid:13) 0000 RAS, MNRAS 000, 000–000 16 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt 15 10 5 0 10 5 0 8 6 4 2 0 -2 1 0.9 0.8 0.7 0.05 0.04 0.03 0.02 0.01 0 18 16 14 12 10 12 11 10 9 10 8 6 4 1.4 1.2 1 0.04 0.03 0.02 0.01 14 12 10 8 6 4 14 12 10 8 6 10 8 6 4 2.1 2 1.9 0.015 0.01 0.005 1.4 1.2 1 0.8 -10 -5 0 5 -2 0 2 Figure 32. Same as Fig. 3. Figure 34. Same as Fig. 3. 15 10 5 12 10 8 6 8 6 4 2 2 1.8 1.6 1.4 1.2 1 1 0.8 0.6 0.4 0.2 0 2 1.8 1.6 1.4 -4 -2 0 2 4 6 -20 0 20 Figure 33. Same as Fig. 3. Figure 35. Same as Fig. 3. c(cid:13) 0000 RAS, MNRAS 000, 000–000 2.4 2.2 2 1.8 2.5 2 1.5 16 14 12 10 8 12 10 8 6 8 6 4 2 1.6 1.4 1.2 0.4 0.3 0.2 0.1 0 10 8 6 4 2 7 6 5 4 3 6 4 2 1.1 1 0.9 0.8 0.7 0.15 0.1 0.05 -40 -20 0 20 40 Figure 36. Same as Fig. 3. The Stellar Content of Active Galaxies 17 12 10 8 6 4 2 8 6 4 2 6 4 2 1.6 1.4 1.2 1 0.8 0.6 1 0.8 0.6 0.4 0.2 0 16 14 12 10 10 8 6 8 6 4 2 1.8 1.6 1.4 1.2 0.4 0.2 0 2 1.8 1.6 1.4 1.2 1 -50 0 50 Figure 38. Same as Fig. 3. 2 1.8 1.6 1.4 1.2 1 2.2 2 1.8 -10 -5 0 5 10 -10 0 10 20 Figure 37. Same as Fig. 3. Figure 39. Same as Fig. 3. c(cid:13) 0000 RAS, MNRAS 000, 000–000 18 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt -40 -20 0 20 40 Figure 40. Same as Fig. 3. 12 10 8 6 4 8 6 4 2 6 4 2 1.4 1.2 1 0.8 0.6 0.4 0.2 0 20 18 16 14 20 18 16 14 12 10 10 8 6 4 1.8 1.7 1.6 1.5 0.06 0.04 0.02 14 12 10 8 6 4 8 6 4 2 8 6 4 2 1.6 1.4 1.2 1 0.15 0.1 0.05 0 12 10 8 6 4 8 6 4 2 0 -2 6 4 2 0 -2 1.5 1.4 1.3 1.2 0.15 0.1 0.05 1.8 1.6 1.4 1.2 2.6 2.4 2.2 2 -50 0 50 Figure 42. Same as Fig. 3. 1.8 1.6 1.4 1.2 1.8 1.6 1.4 -10 0 10 -10 -5 0 5 10 Figure 41. Same as Fig. 3. Figure 43. Same as Fig. 3. c(cid:13) 0000 RAS, MNRAS 000, 000–000 15 10 5 12 10 8 6 10 8 6 4 1.9 1.8 1.7 1.6 100 80 60 40 20 0 2.6 2.4 2.2 2 1.8 1.6 -20 0 20 Figure 44. Same as Fig. 3. One of the clearest examples of dilution by an AGN continuum in our data is NGC 6814 (Figs. 8 and 45c). The high degree of variability both in the lines and continuum of NGC 6814 (Yee 1980, Sekiguchi & Menzies 1990) are an unequivocal signature of the presence of an AGN in this galaxy. The presence of a FC is reflected in our W profiles, which show substantial dilution in the nucleus. Comparing the values of the nuclear Ws with those between r =5 and 11′′ we obtain dilution factors of 47% for Ca K, 28% for CN, 53% for the G-band and 22% for Mg. The larger dilution in the G-band is due to broad Hγ entering the line window (Fig. 45c). The FC also affects the nuclear colors, which are bluer than in the neighboring regions (Fig. 8). The Ws and continuum ratios at 5′′ from the nucleus correspond to a S2– S3 template, changing to S5–S6 outwards. The attribution of a spectral template to the nucleus is not meaningful in this case given the obvious presence of a non-stellar continuum. An interesting case of a galaxy showing signs of dilution is that of the LINER/Seyfert 1 NGC 7213 (Phillips 1979, Filippenko & Halpern 1984). Our W profiles show a clear drop in the Ws of Ca K, CN and G-band, but not in Mg. (Ca H is also diluted, but mostly due to Hǫ.) This indicates that the FC is not strong enough in the 5100 A region to dilute the Mg line, but its contrast with the stellar component increases towards shorter wavelengths, resulting in a dilution of ∼ 31% of Ca K. This value is very similar to that obtained by Halpern & Filippenko (1984). Apart the nucleus of from their dilution at the NGC 7213, the Ws of Ca K and H, CN and G-band in- dicate a S2–S3 template at 5′′ and outwards. Mg has a W similar to that of a S1 template at the nucleus, gradually changing to S4 in the outer regions. The continuum ratios are similar to a S2 template in the inner 5′′ radius, decreas- ing to S3 in the outwards. The stellar population template obtained by Bonatto et al. (1989) is S2, similar to the values we obtained for the regions outside the nucleus. c(cid:13) 0000 RAS, MNRAS 000, 000–000 The Stellar Content of Active Galaxies 19 Figure 45. (cont.) In Fig. 45b we show the spatially resolved, zoomed spec- tra of NGC 7213. The correspondence between the Ws vari- ations in Fig. 44 and the features in the individual spectra is clearly seen. The dilution of Ca K and H, CN and G band is observed as shallower absorption lines in the nuclear spec- trum, while the lack of dilution in Mg can be noticed by the similar depths of the line in different extractions. 4.1 Individual Objects The five cases discussed above fully illustrate the potential of long slit spectroscopy as a tool to probe stellar population gradients and the presence of a featureless AGN-like contin- uum. In this section we provide a description of our results for the remaining objects in our sample, separated by ac- tivity class (Seyfert 1s, Seyfert 2s, Radio Galaxies, LINERs and Normal Galaxies). 4.1.1 Seyfert 1s NGC 526a: This is the brightest galaxy of a strongly inter- acting pair, which presents an optical emission line region elongated in the direction of the companion galaxy (NW- SE, Mulchaey et al. 1996). Both Ws and continuum ratios of this galaxy present a mild gradient (Fig. 4), but opposite to that expected from dilution by a blue continuum. The Ws are larger at the nucleus, indicating S3–S4 templates, decreasing outwards, with values of S4–S5 templates at 7′′ from the nucleus. The continuum ratios indicate the same templates outside the nucleus but a redder ratio at the nu- cleus, corresponding to a S1 template. 20 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt Figure 45. Spectra of five of the sample galaxies for several positions along the slit, zoomed into the regions of the absorption features discussed in this work. (a) NGC 1672, (b) NGC 7213 (c) NGC 6814, (d) Mrk 573 and (e) 3C33. The radial positions are indicated on the right side of the plots. ESO 362-G18: According to Mulchaey et al. (1996), this Seyfert 1 is a strongly perturbed galaxy, with the strongest [OIII] emission concentrated in the nucleus and showing an extension to the SE, suggestive of a conical mor- phology. The Ws and continuum ratios show a strong dilu- tion by a FC in the nuclear region (Fig. 5). At 5′′ and farther out, both Ws and continuum ratios indicate a S5 template. Mrk 732: The individual spectra (Fig. 1) show signa- tures of young stars to the NE of the nucleus (r < 0), where the Ws consistently indicate an S7 template. The nuclear Ws indicate a dilution when compared to the SW side (r > 0), which shows Ws of a S5 template (Fig. 6). The continuum ratios indicate the same templates. MCG-02-33-034: This galaxy was classified as Seyfert 1 by Osterbrock & De Robertis (1985), based on the ap- pearence of permitted lines broader than the forbidden ones and strong FeII emission. Mulchaey et al. (1996) show that it presents evidence of two nuclei, both emitting in [OIII] and Hα. We can see the dilution of Ws and color changes due to a blue FC in the nuclear region (Fig. 7). The Ws change to values of a S3 template at 4′′ and outwards, while the continuum ratios have values similar to a S4 at 4′′ and farther out. As noted in §3, though dilution clearly occurs, the complex nuclear spectrum (Fig. 2d) prevents an accu- rate determination of both Ws and continuum at r = 0 for this galaxy. NGC 6860: According to L´ıpari, Tsvetanov & Mac- chetto (1993) the Hα image of this galaxy shows bright emis- sion line regions associated with the nucleus and a circum- nuclear ring of star-formation. The emission line spectrum is typical of a Seyfert 1.5 and variable. The Ws (Fig. 9) are diluted by a FC at the nucleus. Their values change to those of a S2 template at 4′′ and outwards. The continuum ratios show a gradient in the opposite direction, decreasing from S4 in the inner 5′′ radius to S5 outwards, bluer than the values indicated by the Ws in the outer regions. 4.1.2 Seyfert 2s Mrk 348: This is a Seyfert 2 galaxy with evidence of broad Hα (De Robertis & Osterbrock 1986b), HeI 10830 A and Paβ lines (Ruiz, Rieke & Schmidt 1994). Koski (1978) observed this galaxy through a 2.7′′×4′′ aperture, obtaining a 14% contribution from a FC at 5000 A. Kay (1994) estimated a FC contribution of 35% to the flux at 4400 A, while Tran (1995a) finds 27% at 5500 A. Mrk 348 is one of the Seyfert 2 galaxies which had a hidden Seyfert 1 nucleus revealed by spectropolarimetry (Miller & Goodrich 1990). The Ws and continuum ratios of this galaxy are shown c(cid:13) 0000 RAS, MNRAS 000, 000–000 The Stellar Content of Active Galaxies 21 Figure 45. (cont.) in Fig. 10. The Ws have values similar to those of a S4 template, without any noticeable gradient in the inner 8′′. From Koski (1978), Kay (1994) and Tran (1995a) results, we would expect the Ws to be substantially diluted in the nu- clear region, but this is not detected. One possibility would be that the FC is extended, diluting all the Ws. However, the NLR of this galaxy is extended by less than 2′′ (Schmitt & Kinney 1996), and the FC would have to be extended by ≈ 8′′. The continuum ratios show a gradient, having values similar to those of a S4 template at the nucleus and be- coming bluer outwards, reaching values similar to those of S5–S6 templates. The continuum behavior of Mrk 348 is in- triguing, due to the fact that outside the nucleus it indicates bluer templates than the ones expect from the analisys of the Ws. This behavior is also seen in many other objects in the present sample, particularly Seyfert 2s. Mrk 573: Pogge & De Robertis (1993) found ex- cess near-UV emission, spatially extended along the [OIII] emission, interpreted as scattered nuclear continuum. Koski (1978) estimated that the FC contributes 12% of the light at 5000 A, observed through a 2.7′′×4.0′′ aperture, while Kay (1994) finds 20% FC at 4400 A within an aperture of ≈ 2 × 6′′. She also estimated, based on spectropolarimetry, that the FC is polarized by 5.6% in the 3200–6300 A range, in agreement with the values found by Martin et al. (1983). The Ws (Fig. 11) show a mild gradient, from a S3 tem- plate at the nucleus, to a bluer one, S4 at 5′′. The continuum ratios show a gradient from a S3–S4 template at the nucleus c(cid:13) 0000 RAS, MNRAS 000, 000–000 to a bluer one, S5–S6 at 5′′. Koski (1978) and Kay (1994) estimations of FC continuum contribution to the nuclear spectrum are not confirmed by our data, which do not show any detectable dilution in the Ws. This galaxy is another case in which the continuum ratios outside the nucleus have values bluer than we would expect from the Ws analysis. IC 1816: This galaxy was identified as a Seyfert 1 by Fairall (1988), but the lack of variability lead Winkler (1992) to question this classification. Our spectrum does not show broad emission lines, which lead us to classify it as a Seyfert 2. The Ws of this galaxy (Fig. 12) have values typical of S3 templates in the inner 4′′, with an apparent dilution in Mg caused by contamination by [NI] emission. The Ws decrease to values of a S5 template at 8′′ from the nucleus. The con- tinuum ratios have values corresponding to templates bluer than the Ws, S5 at the nucleus, changing to S6 at 8′′. ESO 417-G6: Our spectrum was obtained with the slit oriented along the extended emission (NW-SE direction, ac- cording to Mulchaey et al. 1996). The Ws show no gradient (Fig. 13), having values typical of a S3 template. The appar- ent dilution in the Mg band is due to contamination by [NI] emission. The continuum ratios show a gradient, changing from S3 at the nucleus, to a bluer S5 template at ≈ 5′′ from the nucleus. Mrk 607: According to De Robertis & Osterbrock (1986a) the nuclear spectrum of this galaxy shows high- excitation lines, like [FeVII] and [FeX], but only narrow per- mitted lines. Kay (1994) obtained a 10% FC contribution to 22 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt the nuclear flux, and estimates a 4.3% continuum polariza- tion in the 3200–6300 A range. The Ws and continuum ratios (Fig. 14) indicate similar templates, S2–S3 at the nucleus, S5 at 6′′, suggesting a weak starforming ring, and S3 outwards. Our spectrum of this edge-on galaxy was obtained with the slit positioned along the major axis, which may explain the large variations in the Ws in such small scales. Considering these variations, we cannot evaluate whether the nuclear Ws are diluted or not. NGC 1358: The nuclear FC contributes with 17% of the flux and is polarized by 1.7% in the 3200–6300 A range, according to Kay (1994). The Ws of NGC 1358 (Fig. 15) show no systematic gradient and have values typical of a S1 template, the same result obtained by Storchi-Bergmann & Pastoriza (1989). The only exception is Ca K, which reaches values larger than S1 at r = −6′′. The continuum ratios show a gradient from values typical of a S1 template at the nucleus to bluer values typical of a S5 template outwards. This galaxy is another case in which the continuum ratios outside the nucleus have values bluer than the ones expected from the analysis of the Ws. NGC 1386: The Ws are typical of a S3 template at the nucleus (Fig. 16), S4–S5 at 10′′ from the nucleus, due to the presence of spiral arms (Storchi-Bergmann et al. 1996a), and S3 outwards. The continuum ratios indicate a S1 template at the nucleus, with a gradient to values typical of S5–S6 at 10′′ and outwards. CGCG 420-015: This galaxy was classified as Seyfert 2 by de Grijp et al. (1992). The W values are indicative of a S3 template at the nucleus, with G-band and Mg showing a small gradient to S2 outwards, consistent with a small dilu- tion in the nucleus (Fig. 17). The continuum ratios indicate a S1 template at the nucleus, redder than the value predicted from the Ws, with a gradient to S3–S4 at 4′′ and farther out. Care must be taken when analyzing the Ws and continuum ratios of this galaxy, because it was observed with the slit positioned almost perpendicular to the parallatic angle, in- troducing large differential refraction effects. ESO 362-G8: The Ws are similar to a S5–S6 template at the nucleus, with a gradient to S4–S5 at 4′′ and farther out (Fig. 18). The continuum ratios are similar to a S5 template at the nucleus, changing to bluer values, typical of S6–S7 templates outwards. The Ws of this galaxy indicate bluer templates in the nuclear region, which is due to the presence of young stars, revealed by strong HI absorption features. This galaxy also has continuum ratios outside the nucleus bluer than the values predicted from the Ws. Mrk 1210: Tran, Miller & Kay (1992) found polar- ized broad Hα and Hβ components in the spectrum of this galaxy. Later Tran (1995a,b,c) confirmed this result and de- termined that the FC contributes to 25% of the nuclear flux at 5500 A, while Kay (1994) estimates a 64% FC contribu- tion at 4400 A. Nevertheless, the Ws show almost no gra- dient indicative of such a dilution (Fig. 19). The Ws cor- respond to S4 and S5 templates. The continuum ratios are similar to S5–S6 at the nucleus, with a gradient to S6 out- wards, which is bluer than the template estimated from the Ws. There is an apparent dilution of the Mg and Na features in the nuclear spectrum due to emission line contamination by [NI] and HeI respectively. by Phillips, Charles & Baldwin (1983). According to Pogge (1989a), the [OIII] image is symmetrical but more extended than the stars profile. The W profiles (Fig. 20) do not show any evidence of a stellar population gradient, with values similar to S2–S3 templates. The continuum ratios show a gradient from a S2–S3 template at the nucleus, to S4–S5 at 5′′ and outwards, bluer than the values indicated by the Ws. IRAS 11215-2806: This galaxy was classified by de Grijp et al. (1992) as a Seyfert 2. The Ca K and G-band Ws (Fig. 21) present a gradient from a S4 template at the nucleus to S2 in the outer regions, indicating dilution by a blue continuum. Mg presents almost no gradient, with values typical of S3–S4 templates. The continuum ratios present a gradient in the opposite direction, from S4 at the nucleus, to S5 outwards. MCG-05-27-013: This galaxy was classified as Seyfert 2 by Terlevich et al. (1991). The Ws of Ca K are typical of a S3 template at the nucleus, changing to S1 outwards, which suggests dilution by a FC (Fig. 22). The G-band has a smaller gradient, with values similar to a S2 template at the nucleus and S1 outwards, while for Mg the values are similar to a S2 template, with no gradient. The continuum ratios have values larger than S1 at the nucleus, due to strong red- dening, changing to values similar to S4 in the outer regions, bluer than the values predicted by the analysis of the Ws. Fairall 316: This is a Seyfert 2 galaxy discovered by Fairall (1981). The Ws show no gradient (Fig. 23), with Ca K having values typical of S2, while G-band and Mg have values typical of S1. The continuum ratios have values of S1 template at the nucleus, decreasing outwards to values bluer than those predicted by the Ws, similar to a S3 template. NGC 5135: NGC5135 was described by Phillips, Charles & Baldwin (1983) as having a composite nucleus with characteristics of both Seyfert 2 and Starburst. Thuan (1984) obtained IUE spectra of this galaxy, confirming the dual nature of the nucleus, which presents both emission lines typical of Seyfert 2s and absorption lines typical of Starbursts, the latter component contributing with 25% of the total UV emission. The Ws (Fig. 24) show a gradient from values typical of a S7 template at the nucleus to S6–S5 at 7′′. The continuum ratios also show a gradient from S6 at the nucleus to S5 at 7′′. This gradient is due to the presence of young stars in the nuclear region, as revealed by strong HI absorption features. NGC 5643: This galaxy was classified by Phillips et al. (1983) as a low luminosity Seyfert 2. Schmitt, Storchi- Bergmann & Baldwin (1994) presented [OIII] and Hα im- ages, as well as optical spectra of NGC5643. The images show the high-excitation gas to be extended along the E-W direction by ≈ 20′′. The stellar population is moderately old in the central region, showing evidence of absorption lines diluted by a blue continuum, supposed to be scattered nu- clear light. The Ws of this galaxy present a mild gradient to the E of the nucleus (i.e., towards negative r in Fig. 25), suggesting the presence of scattered light up to at least 4′′ from the nu- cleus. In the outer regions, the Ws correspond to templates S3–S4, while the continuum ratios correspond to S3 to the E of the nucleus and S1 to the W. Bonatto et al. (1989) ob- tained that the nuclear stellar population of this galaxy can be represented by a S3 template, similar our result. NGC 3081: This galaxy was classified as a Seyfert 2 NGC 6300: Storchi-Bergmann & Pastoriza (1989) ob- c(cid:13) 0000 RAS, MNRAS 000, 000–000 served the nuclear stellar population of this galaxy to be old, with a small contribution from intermediate age stars. The Ws do not show a clear gradient (Fig. 26), with Ca K varying between S2 and S3 templates, while G-band and Mg vary between S4 and S5. The continuum ratios decrease from values much larger than those of a S1 template at the nucleus to values typical of a S6 template at 10′′ and farther out, bluer than the values predicted from the Ws in this region. NGC 6890: Storchi-Bergmann, Bica & Pastoriza (1990) studied the nuclear optical spectrum of this galaxy, finding that the stellar population is old and that Hα may have a broad component. The Ws show a gradient in Ca K and G band, which change from S4 at the nucleus to S3 in the outer regions, but not in Mg, which has values typical of a S3 template, with no gradient (Fig. 27). The continuum ratios have values larger than S1 at the nucleus, decreas- ing to S2 in the outer regions. These results are similar to the ones obtained by Storchi-Bergmann, Bica & Pastoriza (1990). NGC 7130: Like NGC 5135, this galaxy has a Seyfert 2 nucleus surrounded by a Starburst (Phillips et al. 1983). Thuan (1984) presents IUE spectra which show both high- excitation emission lines typical of Seyfert 2s and absorption lines typical of Starburst. The starburst component domi- nates the UV emission, with ≈ 75% of the flux. Shields & Filippenko (1990) showed that the narrow line region have two kinematical components and emission line ratios which vary from those of AGN in the nucleus to those typical of HII regions outwards. Both Ws and continuum ratios show a marked gradient (Fig. 28). The Ws change from values similar to S7 at the nucleus to S6 outwards, while the continuum ratios change from S6–S7 at the nucleus to S5–S6 outwards. This gradient is due to the presence of a circumnuclear starforming region. NGC 7582: Morris et al. (1985) found circumnuclear Hα emission in this Seyfert 2 galaxy, suggesting a ring of ro- tating HII regions in the galaxy plane. The Ws of this galaxy show a dilution, changing from S7 at the nucleus to S3–S4 in the outer regions (Fig. 29). The continuum ratios are similar to S3 at the nucleus and NE (r < 0), but decrease to values similar to S5–S6 towards r > 0. The continuum ratios to the SW are bluer than what we would expect from the Ws results. The dilution measured in this object is probably due to its inner star-forming ring. 4.1.3 Radio Galaxies 3C 33: This is a FR II radio source. Koski (1978) esti- mated that the FC contributes to 19% of the flux at 5000 A, observed through a 2.7′′×4′′ aperture. Antonucci (1984) obtained spectropolarimetric data, which shows wavelength independent polarization, but no broad polarized emission lines. The Ws of this radio-galaxy (Fig. 30) have values simi- lar to an E3 or E6 template, though W(MgI) is more similar to an E1 template. The continuum is redder at the nucleus, with values of an E2 template, decreasing to values typical of an E3 or E6 template outwards. The FC contribution es- timated by Koski (1978) would produce a dilution of the Ws in the nuclear region. Our data do not support this result, as the W-profiles are essentially flat. c(cid:13) 0000 RAS, MNRAS 000, 000–000 The Stellar Content of Active Galaxies 23 Pks 0349-27: The Ws (Fig. 31) do not show a gradient and have values typical of an E1–E2 template, whereas the continuum ratios show a gradient from an E2 template at the nucleus to a bluer, E3–E4 template outwards. Pictor A: This is a FR II Broad Line Radio Galaxy with a strong double lobed radio source oriented along the E- W direction (Christiansen et al. 1977). Halpern & Eracleous (1994) detected double-peaked Balmer lines, not present in previous spectra. The Ws and continuum ratios (Fig. 32) show a strong dilution by the FC at the nucleus. Their val- ues correspond to a S4 template outwards, except for the G-band, where the values are similar to those of a S1 tem- plate in the outer regions. The remarks about the uncertain continuum and W measurements in the nucleus of MCG-02- 33-034 also apply to Pictor A. Pks 0634-20: This is a FR II radio source. Simpson, Ward & Wilson (1995) did not detect a broad Paα line and show that the K-L color, [OIII] and soft X-ray fluxes are consistent with the spectral energy distribution of a quasar absorbed by AV ≈ 30 mag. The Ws (Fig. 33) are consistent with an E2 stellar population, without a gradient. The con- tinuum ratios indicate bluer templates, E3 at the nucleus, decreasing to values typical of E6 outwards. Pks 0745-19: According to Baum & Heckman (1989), this narrow line radio galaxy does not have a well defined FR class. The Ws indicate an E8 template at the nucleus, with a gradient to values typical of E4 outwards (Fig. 34), suggesting dilution by a FC. The continuum ratios indicate a redder, E2 stellar population. However, it should be no- ticed that these continuum ratios were not corrected by fore- ground reddening, because this galaxy lies too close to the Galactic plane (b < 4◦). 4.1.4 LINERs and Normal Galaxies NGC 1326: Storchi-Bergmann et al. (1996a) obtained nar- row band [OIII] and Hα images, which reveal a circum- nuclear ring at r ≈ 6′′. As for NGC 1097 and other galaxies containing star-forming rings, the Ws and color profiles of this galaxy (Fig. 35) confirm the presence of a ring. The radial behavior of the Ws and continuum ratios do not ev- idence the presence of a FC in the nucleus. The Ws have values typical of a S3 template at the nucleus, S6 at the ring and S2–S3 outwards. The continuum ratios indicate a S1 template at the nucleus (redder than the one indicated by the Ws), S6 at the ring, and S3–S4 outwards. NGC 1598: According to Phillips et al. (1984), the nu- clear spectrum of this galaxy is similar to that of LINERs, with evidence of young stars at the nucleus, which is also surrounded by a ring of HII regions. The Ws and contin- uum ratios (Fig. 37) indicate similar templates: S7 at the nucleus, in agreement with the presence of young stars, and S5 from 4 to 6′′ reaching values typical of a S6 template outwards. NGC 2935: This is a normal galaxy with a nuclear ring of star formation (Buta & Crocker 1992). The Ws of Ca K and G-band are similar to S4 at the nucleus, S5 at the ring (5′′) and S2 outwards, while Mg is similar to S2 at the nucleus, S3 at the ring and S2 outwards (Fig. 39). The continuum ratios are similar to S3 at the nucleus, S5 at the ring and S3 outwards. NGC 4303: This is a Sersic-Pastoriza galaxy, classified 24 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt as a LINER by Huchra, Wyatt & Davis (1982). Filippenko & Sargent (1985) observations show that HII regions are prominent in the nucleus and circumnuclear region. They also found a substantially broader component in each line and propose we are seeing a faint Seyfert 2 or LINER nucleus hidden by HII regions. The Ws are similar to a S5 template at the nucleus, decreasing to values similar to a S6 template at the ring (5′′; Fig. 40). From 7′′ to 15′′ from the nucleus the Ws are similar to a S4 template, decreasing to S6–S7 in the outer regions. The continuum ratios behave like the Ws, with the exception of the nucleus and ring, where they indicate redder templates, S4 and S5 respectively. In the other regions Ws and continuum ratios indicate similar templates. Our results are in good agreement with those of Bonatto et al. (1989). IC 4889: Phillips et al. (1986) found Hα, [NII] and [OII] emission in this normal elliptical galaxy. The Ws show no significant gradient, with values typical of an E1 template, while the continuum ratios are E1 at the nucleus and de- crease to values similar to those of an E3 template in the outer regions (Fig. 41). NGC 5248: Both Ws and continuum ratios show the presence of a ring at r = 10′′ from the nucleus (Fig. 42). The values of Ca K and G-band are typical of a S5 template at the nucleus, S6 at the ring, S4 at 15′′ from the nucleus, decreasing to values typical of S6–S7 templates outwards. Mg and continuum ratio values indicate redder templates, S4 at the nucleus, S5 at the ring, S2–S3 at 15′′, decreasing to S5–S6 outwards. Our result for the nuclear stellar population is similar to the one obtained by Bonatto et al. (1989). NGC 6221: This southern spiral galaxy was classi- fied as a Seyfert 2 by V´eron, V´eron & Zuiderwijk (1981) and Pence & Blackman (1984), based on the detection of a faint broad Hβ component, an [OIII] line broader than Hβ and to the proximity to a hard X-ray source (Marshall et al. 1979, Wood et al. 1984). However, Schmitt & Storchi- Bergmann (1995) recently proposed that the identification of NGC6621 with the X-ray source is probably wrong and that it would rather be related to the high-excitation Seyfert 2 galaxy ESO138-G01. Pence & Blackman (1984) data also show that the nucleus is surrounded by a Starburst. Our W profiles (Fig. 43) show a gradient from a S7 tem- plate at the nucleus to S5 at 5′′ and farther out, in agreement with the presence of a nuclear starburst, confirmed by the gradient also observed in the continuum ratios, which cor- respond to S6 at the nucleus and to S5 at 5′′ and farther out. 5 DISCUSSION The present set of high quality long-slit spectra contains a wealth of information on the properties of AGN and of the stellar population of their host galaxies. In this section we discuss some of the global results revealed by our analysis and their possible implications. 5.1 Stellar Populations The values of the continuum colors and Ws of the stellar lines found in our analysis cover a wide range, as can be seen in Figs. 3–44. Even in inner regions of the bulge not affected by circumnuclear star-formation or by a non-stellar FC, the galaxies exhibit substantial differences in the Ws and colors, reflecting a variety of stellar population properties. This was confirmed by our characterization of the spectra in terms of Bica's (1988) templates, which showed that, despite a predominance of S2 and S3 templates, every spectral type from S1 to S7 appears in the representation of the spectra in the inner regions of our sample galaxies. This observed variety of stellar population properties illustrates the inadequacy of using a single starlight tem- plate to evaluate and remove the stellar component from AGN spectra. The importance of an accurate evaluation of the starlight component cannot be underestimated. A 'tem- plate mismatch' affects not only the derived emission line fluxes, but also, and more strongly, the determination of the strength and shape of a residual FC. We have seen, for instance, that many of the nuclei in the present sample show no evidence for dilution of the stellar lines in compar- ison with off-nuclear spectra of the same galaxy. However, some dilution would most probably be measured if a differ- ent galaxy was used as a starlight template. The cases of Mrk 348, 573, 1210 and 3C33 are illustrative in this respect. Whereas Koski (1978), Kay (1994) and Tran (1995a) ob- tained large contributions of a FC comparing their spectra with a normal galaxy template (§4.1), we find little evidence for dilution of the absorption lines comparing the nuclear Ws with those measured a few arcseconds outside the nucleus. This brings up the question of which is the best way to estimate the stellar component in an AGN spectrum. By far the most commonly adopted procedure is to use the spec- trum of a normal galaxy as a starlight template. Given the variety of stellar populations found in AGN, if a template galaxy is to be used at all then it has to be chosen among a library contemplating a large variety of stellar population characteristics. An alternative technique, implicitly used in this paper, is to use an off-nucleus spectrum of the object as the starlight template. This is arguably a more robust method, as the best representation of the stellar popula- tion in a given galaxy is the galaxy itself! The extrapolation to zero radius, however, involves the assumption that the stellar population in the bulge of the host galaxy does not change dramatically in the inner few arcseconds. A compar- ison between these two methods of starlight evaluation will be presented in a future communication. 5.2 Dilution factors Figs. 3–44 demonstrate that several galaxies show signs of dilution of the stellar lines in the nucleus. In this section we compare the nuclear Ws to those a few arcseconds outside the nucleus in order to estimate the fraction of the contin- uum associated with a featureless spectral component. In order to obtain Ws representative of the neighbour- hood of the nucleus we have co-added typically four extrac- tions, two in each side of the nucleus, located at distances ranging from 2 to 12′′ (see Table 4). Extractions adjacent to the nucleus were avoided whenever possible, since, due to seeing, they are often contaminated by the nuclear spec- trum. The typical distance of the off-nuclear extractions was 4′′, corresponding to physical radii between ∼ 0.5 and 2 kpc for the distances of the galaxies in the sample. After constructing these off-nuclear templates and mea- c(cid:13) 0000 RAS, MNRAS 000, 000–000 suring their Ws, the FC fraction of the total continuum spec- trum at the wavelength of an absorption line was estimated by fFC(λ) = Woff−nuc − Wnuc Woff−nuc (1) where Wnuc and Woff−nuc are the Ws in the nucleus and in the off-nuclear template respectively. Table 4 presents the results of this analysis. Galaxies not listed in Table 4 show no signs of dilution, and even for some objects in the table (e.g., NGC 1097, Mrk 348) the evidence for dilution is only marginal-by 'marginal' we mean within 2σ of a spurious dilution caused by the errors in the W measurements. We recall that dilution of the Ws by less than 10% cannot be reliably detected with this method due to the errors in the Ws. We concentrated this analysis in the Ca K, G-band and Mg lines. However, as indicated in the Table, the Mg line is very often contaminated by [Fe VII] and/or [NI] in the nuclear spectrum, producing an apparent dilution. In such cases the fFC(5176A) fractions have to be taken as upper limits of the actual dilution. The line which is less affected by such spurious effects is Ca K. Empty slots in Table 4 occur whenever an increase in the W, instead of dilution, was observed. Of the broad-lined objects in our sample, only NGC 526a does not exhibit a systematic dilution of the Ws in the nucleus. The other possible exception is NGC 1097, which does show some dilution in the Ca K line, but at a marginal level. The highest values of fFC occur for MCG-02-33-034 and Pictor A, where it reaches nearly 80% at the wave- length of Ca K. However, as discussed in §3, these values are not as reliable as for other Seyfert 1s due to the un- certain positioning of the continuum for these two objects. Indeed, their nuclear spectra are so complex that the G-band and Mg lines are completely filled by emission, preventing a determination of fFC at the corresponding wavelengths. Perhaps the most intriguing aspect of Table 4 is the low degree of dilution detected in Seyfert 2s. Of the 20 Seyfert 2s in our sample, 13 are listed in Table 4, and only 9 of these (IC 1816, ESO 362-G8, NGC 5135, 5643, 6890, 7130, 7582, MCG -05-27-013 and IRAS 11215-2806) have dilution con- vincingly detected. As reviewed in §4.1, NGC 5135, 7130, 7582 and ESO 362-G8 are composite nuclei, showing fea- tures of both Seyfert 2s and Starbursts. In these cases, as for the LINER NGC 1598 (Fig. 37), the dilution is almost certainly dominated by the young stellar component, not a genuine AGN continuum. This lowers the count to 5 objects with detected dilution out of 16 'pure' Seyfert 2s. Marginal indications of dilution were found in Mrk 348, Mrk 1210, NGC 1358 and CGCG 420-015. In the other objects, the FC, if present at all, contributes 10% or less in the optical. This result is further discussed below. 5.2.1 Implications for FC2 The low levels or lack of dilution in Seyfert 2s found in the present analysis is an interesting result in light of the current debate over the existence and origin of a FC in Seyfert 2s and its implications for unified models. It has been shown that the FC in Seyfert 2s cannot be simply the continuum of a hidden AGN scattered into our line of sight, otherwise broad emission lines should also be seen, in contradiction c(cid:13) 0000 RAS, MNRAS 000, 000–000 The Stellar Content of Active Galaxies 25 with the very definition of type 2 Seyferts (Cid Fernandes & Terlevich 1992, 1995, Heckman et al. 1995). Furthermore, the low levels of polarization (Miller & Goodrich 1990), and the observation of reflected broad lines substantially more polarized than the FC (Tran 1995c) are also in conflict with the simple obscuration/reflection picture. These facts lead to the idea that a second FC component (dubbed FC2 by Miller 1994) must be present in Seyfert 2s. The nature of this component remains unknown, though both a circumnuclear starburst (Cid Fernandes & Terlevich 1995, Heckman et al. 1995) and free-free emission from the scattering region (Tran 1995c) have been suggested. Our observations seem to indicate that a FC (and conse- quently FC2) is not an universal phenomenon in Seyfert 2s, and that previous determinations of the FC strength have been overestimated. A smaller FC contribution would alle- viate, if not completely solve, all problems listed above. A good way to further investigate this issue would be to obtain spectropolarimetry for the Seyfert 2s in the present sample. We expect galaxies for which no dilution of the Ws was detected to show little or no polarization. IC 1816, NGC 5643, NGC 6890, MCG -05-27-013 and IRAS 11215-2806, on the other hand, should exhibit large polarizations and Seyfert 1 features in the polarized spectra. In the cases of composite nuclei, spectropolarimetry could help disentan- gling the starburst and Seyfert 2 components and possibly reveal a hidden AGN. As a final word of caution, we note that while through- out this paper we have attributed the dilution of absorp- tion lines in the nucleus of active galaxies to a FC, dilution can also be caused by young stars, as shown to occur in circumnuclear star forming rings (e.g., Figs. 3, 36 and 38) and in the nuclei of the Starburst galaxies NGC 6221 and 7130 (Figs. 43 and 28). It is a known problem in optical spectroscopy of AGN that it is often difficult to distinguish between a young stellar population and a FC (e.g., Miller & Goodrich 1990, Cid Fernandes & Terlevich 1992, Kay 1994). UV spectroscopy offers the most direct way of removing this ambiguity (e.g., Leitherer, Robert & Heckman 1995, Heck- man et al. 1995). 5.2.2 Metallicity gradients In measuring the dilution of absorption lines with respect to their strengths ∼ 4′′ (or ∼ 1 kpc) outside the nucleus we are making the assumption that the stellar population in the bulge of the host galaxy does not change appreciably from r ∼ 4′′ to r = 0. This approximation might not be valid if the galaxy presents a strong metallicity (Z) gradient, lead- ing to an increase of the Ws towards the nucleus (e.g., D´ıaz 1992). If such a gradient is present, then a flat W profile as the one found in many of our galaxies could reflect the com- pensating effects of a FC diluting the enhanced metal lines in the nucleus§. In this section we investigate this possibil- ity by means of an approximate treatment of the potential effects of a metallicity gradient. To estimate we have used ∆ log Z/∆ log r = 0.2, a typical value obtained for r ∼ 3′′ in elliptical galaxies (Davies, Sadler & Peletier 1993 Z-gradient the > § We thank Dr. R. Terlevich for pointing this out to us. 26 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt DILUTION FACTORS Object Type r [′′] fFC(λ) [%] 3930 A 4301 A 5176 A NGC 1097 NGC 1598c NGC 7213 ESO 362-G18 NGC 6814 MCG-02-33-034 NGC 6860 Pictor A Mrk 348 Mrk 1210 IC 1816 ESO 362-G8c NGC 1358 NGC 5643 NGC 6890 CGCG 420-015 IRAS 11215-2806 MCG-05-27-013 NGC 5135c NGC 7130c NGC 7582c NGC 6221c PKS 0745-19 Lin-1 Lin Lin-1 1 1 1 1 BLRG 2 2 2 2 2 2 2 2 2 2 2-SB 2-SB 2-SB SB NLRG 4.0 4.0 3.6–7.2 7.0–12.0 5.0–10.8 4.0–7.0 4.0–7.0 3.7–6.5 1.9–3.7 4.0–8.0 4.0 4.0–7.0 4.0–6.0 3.6–5.4 1.8–3.6 2.0–4.0 4.0–7.0 4.0–7.0 3.6–7.2 3.6–7.2 4.0–8.0 3.6–5.4 1.9 6±5 73±4 31±4 46±6 47±4 76±5 51±4 77±3 1±5 11±6 13±5 25±6 0±5 19±4 17±5 4±5 24±4 10±5 37±8 47±5 70±4 52±4 55±13 - 39±6 33±6b 59±5b 53±6b 100b 64±6b 100b 9±6 12±7 15±6 15±7 1±7 33±6 39±7 15±6 34±5 12±7 51±7 62±6 60±6 73±4 3±20 - 39±6 - 22±6a 22±5a 100a 50±5 100a 9±7a 26±7a 30±5a 11±5 - 33±4a - 5±7a 30±5a - 36±8a 35±5a 57±5a 49±6a 15±11a Table 4. Column 3 lists the range in distances from the nucleus used to compute Woff−nuc. Columns 4–6 list the FC fractions corresponding to the Ca K, G-band and Mg lines respectively. a contamination by [FeVII] and/or [NI]. b contamination by broad Hγ. c dilution due to young stars in the nucleus. and references therein). Assuming that the nuclear stellar population sampled in our 2′′ × 2′′ aperture is character- ized by a mean radius of 0.5′′, and considering a typical off-nuclear extraction distance of 4′′, this corresponds to an increase of ∼ 0.2 dex in Z from r = 4′′ to the nucleus, cor- responding to the inner ∼ 1 kpc of the galaxy. This value is also typical of Z-gradients mapped from the radial vari- ations of the O abundances in HII regions in galactic disks (e.g., Vila-Costas & Edmunds 1992 and references therein). To convert this gradient to a W-gradient we used the results of Jablonka, Alloin & Bica (1992). We adopted their calibra- tion for an old, globular cluster population, corresponding to the steepest W-Z relation, thus resulting in the largest possible W-variations for a fixed Z-gradient. This exercise resulted in the following variations of the Ws due to a Z-gradient: ∆W(CaK) = 1.2 A, ∆W(G- band) = 0.8 A and ∆W(Mg) = 0.5 A. Given that the assumptions above were all made in the sense of maximiz- ing the effects of a Z-gradient, these variations should be regarded as upper limits. These are small variations, being of the same order of the errors in W (§3.1). The net varia- tions with respect to the typical values of W found in our off-nuclear extractions (∼ 13 A for Ca K and ∼ 8 A for both G-band and Mg) would correspond to changes of, at most, 10% in W(Ca K), 8% in W(G-band) and 6% in W(Mg). These results, while certainly not closing the question, seem to indicate that Z-gradients do not constitute a seri- ous caveat to the interpretation of W-profiles in AGN. Only galaxies with a FC weaker than 6–10% could have a dilution of the metal lines masked by the Z-gradient. Whilst we do not think that a compensation of the effects of a FC and a Z-gradient provides a viable interpretation for the ubiqui- tous absence of dilution in our sample galaxies, particularly in Seyferts 2, further investigations on the strength and de- tectability of Z-gradients in active galaxies would clearly be wellcome, as this issue also bears on the question of whether an off-nuclear extraction provides a better starlight tem- plate for the nuclear stellar population than a normal galaxy (§5.1). 5.3 Color gradients The continuum ratios in the sample galaxies exhibit an inter- esting radial behavior. As seen in Figs. 3–44, the F5870/F4020 and F4630/F3780 ratios very often peak in the nucleus, indi- cating a redder continuum. This seems to be an extremely ubiquitous behavior. The two kinds of objects which do not follow this rule are those containing nuclear starbursts, namely, NGC 1598, 6221, 5135, 7130, and broad-lined ob- jects, the most extreme examples being MCG -02-33-034 and Pictor A. In particular, all Seyfert 2s, with the exception of NGC 5643 which shows a complex color profile, have nuclei redder than the extranuclear regions. This effect is not an artifact of our pseudo continuum determination, as we ver- ified that the color-profiles are nearly identical if instead of the pseudo-continuum one defines line-free bands close to the pivot points to compute mean fluxes and flux ratios. This is a somewhat surprising result, given that the c(cid:13) 0000 RAS, MNRAS 000, 000–000 general prejudice is that AGN are bluer than normal galax- ies, and as such they should also be bluer than their own bulges. On the other hand, if taken as evidence of redden- ing, the color profiles would fit in to the currently favored view that the nuclear regions of Seyfert 2s are rich in dust. Some support to this interpretation comes from the behavior of the Na line (partially produced in interstellar medium), whose W profiles in some galaxies, notably ESO 417-G6, NGC 1386, ESO 362-G8, NGC 6300 and NGC 7582, peaks in the nucleus, mimicking the color profiles. Examination of the radial behavior of the Hα/Hβ emission line ratio would provide a test of this interpretation. Another intriguing result was found from the compar- ison of the stellar population templates inferred from the Ws with those inferred from the continuum ratios for the Seyfert 2s. With few exceptions, the off-nuclear colors indi- cate a spectral template bluer than that indicated by the absorption lines. At present we cannot give a clear expla- nation for this apparent contradiction, but we speculate on two possible interpretations. (1) The effect is due to a strong contribution from metal rich (thus strong lined) intermedi- ate age (thus blue) stars, a population not well represented by any of templates S1–S7. This would be a genuine stellar population effect. (2) The blue continuum is due to a scat- tered AGN component while the absorption lines originate in a metal rich population. This would result in an extended dilution of the stellar features, whose Ws would therefore be underestimated, leading to the incorrect attribution of a spectral template less metallic than the actual population. Detailed modeling will be required to test these possibilities. 6 SUMMARY We have reported the results of a high signal-to-noise ra- tio long slit spectroscopy study of 39 active galaxies and 3 normal galaxies. The run of stellar absorption lines and continuum colors with distance from the nucleus was in- vestigated using a consistent methodology, aiming a global characterization of the stellar populations and the detection of a nuclear FC through the dilution of conspicuous absorp- tion lines with respect to off-nuclear extractions. Our results can be summarized as follows. (1) Radial variations of the stellar populations were de- tected in most cases. Star-forming rings, in particular, were found to leave clear imprints in the equivalent widths and color profiles. (2) The stellar populations in the inner arcseconds of active galaxies are varied, as inferred from the wide spread of equivalent widths and colors measured in this work. This fact alone raises serious doubts as to whether it is appropri- ate to use a single starlight template to evaluate and remove the stellar component from AGN spectra. (3) The equivalent width profiles were used to measure the dilution of the absorption lines in the nucleus with re- spect to off-nuclear spectra. Dilution by a nuclear FC was detected in most broad lined objects in the sample. (4) Galaxies undergoing star-formation in their nuclei, including composite starburst + Seyfert 2 objects, also show dilution of the nuclear absorption lines. The dilution in these cases is due to young stars, not to a FC. These galaxies also c(cid:13) 0000 RAS, MNRAS 000, 000–000 The Stellar Content of Active Galaxies 27 show a clear color gradient, their nuclei being bluer than off-nuclear regions. (5) About 50% of the Seyfert 2s in the sample show no signs of dilution of their absorption lines in the nucleus, indicating that a FC, if present, contributes less than 10% of the total flux in the optical range. Possible consequences to the (controversial) nature of the FC in Seyfert 2s were discussed. (6) All but one of the Seyfert 2s present a color gradient in the sense that the spectrum gets redder as we approach the nucleus, indicative of the presence of dust. (7) There is an apparent discrepancy between the spec- tral templates assigned from the equivalent widths of the absorption lines and from the continuum colors, an effect observed in the off-nuclear spectra of most Seyfert 2s in the present sample. Tentative interpretations involving either a stellar population effect or an extended blue continuum were briefly outlined. Acknowledgments: We thank Dr. E. Bica for use- ful discussions and remarks on an early version of this manuscript. HRS acknowledges the hospitality of the De- partment of Physics at Florian´opolis. Support from FAPEU- UFSC is also duly acknowledged. RCF work was partially supported by CNPq under grant 300867/95-6. This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, CALTECH, under contract with the National Aeronautics and Space Administration. REFERENCES Antonucci, R. R. J. 1984, ApJ, 278, 499 Antonucci, R. R. J. 1993, ARA&A, 31, 473 Aretxaga, I., Terlevich, R. 1994, MNRAS, 269, 462 Baum, S. A., Heckman, T., Brindle, A., van Breugel, W. & Miley, G. 1988, ApJS, 68, 643 Baum, S. A. & Heckman, T. 1989, ApJ, 336, 681 Bica, E. 1988, A&A, 195, 76 Bica, E. & Alloin, D. 1986, A&A, 166, 83 Bica, E., Pastoriza, M. G., Maia, M., Da Silva, L. A. L. & Dottori, H. 1991, AJ, 102, 1702 Binette, L., Wilson, A. S. & Storchi-Bergmann, T. 1996, A&A, 312, 365 Bonatto, C. J., Bica, E. & Alloin 1989, A&A, 226, 23 Buta, R. 1986, ApJS, 61, 631 Buta, R. 1987, ApJS, 64, 383 Buta, R., Crocker D. A. 1992 AJ, 103, 1804 Christiansen, W. N., Frater, R. H., Watkinson, A., O'Sullivan, J. D. & Lockhart, I. A. 1977, MNRAS, 181, 183 Cid Fernandes, R. 1996, Vistas in Astronomy, 40, 143 Cid Fernandes, R. & Terlevich, R. 1992, in "Relationships be- tween Starburst and Active galaxies", Filippenko, A. (ed.), (ASP Conference Series), 241 Cid Fernandes, R. & Terlevich, R. 1995, MNRAS, 272, 423 Colbert, E. J. M., Baum, S. A., Gallimore, J. F., O'Dea, C. & Christensen, J. A. 1996, ApJ, 467, 551 de Grijp, M. H. K., Keel, W. C., Miley, G. K., Goudfrooij, P. & Lub, J. 1992, A&AS, 96, 389 Davies R. L., Sadler E. M., Peletier R. F. 1993, MNRAS, 262, 650 De Robertis, M. M. & Osterbrock, D. E. 1986a, ApJ, 301, 98 De Robertis, M. M. & Osterbrock, D. E. 1986b, ApJ, 301, 727 D´ıaz, A. I. 1985, Ph.D. thesis, University of Sussex 28 R. Cid Fernandes, T. Storchi-Bergmann & H. R. Schmitt D´ıaz, A. I. 1992, in "Relationships between Starburst and Active galaxies", Filippenko, A. (ed.), (ASP Conference Series), 181 Fairall, A. P. 1988, MNRAS, 233, 691 Fairall, A. P. 1981, MNRAS, 196, 417 Ferland, G. J. & Osterbrock, D. E. 1986, ApJ, 300, 658 Filippenko, A. V. & Halpern, J. P. 1984, ApJ, 285, 458 Filippenko, A. V. & Sargent, W. L. W. 1985, ApJS, 57, 503 Filippenko, A. V. & Sargent, W. L. W. 1988, ApJ, 324, 134 Garcia-Vargas, M. L., D´ıaz, A. I., Terlevich, R. & Terlevich, E. Schmitt, H. R., Storchi-Bergmann, T. & Baldwin, J. A. 1994, ApJ, 423, 237 Sekiguchi, K. & Menzies, J. W. 1990, MNRAS, 245, 66 S´ersic, J. L. & Pastoriza, M. 1965, PASP, 77, 287 (SP65) S´ersic, J. L. & Pastoriza, M. 1967, PASP, 79, 152 (SP67) Shields, J. C. & Filippenko, A. V. 1990, AJ, 100, 1034 Stauffer J. R., ApJ, 262, 66 Simpson, C., Ward, M. J. & Wilson, A. S. 1995, ApJ, 454, 683 Storchi-Bergmann, T., Baldwin, J. A. & Wilson, A. S. 1993, ApJ, 1990, Ap&SS, 171, 65 410, L11 Goodrich, R. W. & Veilleux, S. & Hill, G. J. 1994, ApJ, 422, 521 Halpern, J. P. & Eracleous, M. 1994, ApJ, 433, L17 Halpern, J. P & Filippenko, A. V. 1984, ApJ, 285, 475 Haniff, C. A., Wilson, A. S. & Ward, M. J. 1988, ApJ, 334, 104 Harnett, J. I. 1987, MNRAS, 227, 887 Heckman, T., Krolik, J., Meurer, G., Calzetti, D., Kinney, A, Koratkar, A., Leitherer, C., Robert, C. & Wilson, A. S. 1995, ApJ, 452, 549 Ho, L. C., Filippenko, A. V. & Sargent, W. L. W. 1993, ApJ, 417, 63 Huchra, J. P., Wyatt, W. F. & Davis, M. 1982, AJ, 87, 1628 Jablonka P., Alloin D., Bica E. 1990, A&A, 235, 22 Kay, L. E. 1994, ApJ, 430, 196 Keel, W. C. 1983, ApJ, 269, 466 Kinney, A. L., Bohlin, R. C., Calzetti, D., Panagia, N. & Wyse, R. F. G. 1993, ApJS, 86, 5 Storchi-Bergmann, T., Bica, E. & Pastoriza, M. G. 1990, MN- RAS, 245, 749 Storchi-Bergmann, T. & Bonatto, C. J. 1991, MNRAS, 250, 138 Storchi-Bergmann, T., Eracleous M., Halpern J., Wilson A., Fil- ippenko A. 1995a, ApJ, 443, 617 Storchi-Bergmann, T., Kinney, A. L. & Challis, P. 1995b, ApJS, 98, 103 Storchi-Bergmann, T. & Pastoriza, M. G. 1989, ApJ, 347, 195 Storchi-Bergmann, T., Rodriguez-Ardila, A., Schmitt, H. R., Wil- son, A. S. & Baldwin, J. A. 1996a, ApJ, 472, 83 Storchi-Bergmann, T., Wilson, A. S. & Baldwin, J. A., 1996b, ApJ, 460, 252 Storchi-Bergmann, T., Wilson, A. S., Mulchaey, J. S. & Binettec, L. 1996, A&A, 312, 357 Storchi-Bergmann T., Bica E., Kinney A., Bonatto C. 1997, MN- RAS, submitted Kinney, A. L., Antonucci, R. R. J., Ward, M. J., Whittle, M. & Tadhunter, C. N., Fosbury, R. A. E. & Quinn, P. J. 1989, MNRAS, Wilson, A. S. 1991, ApJ, 377, 100 Koski, A. T. 1978, ApJ, 223, 56 Kruper, J. S., Urry, C. M., Canizares, C. R. 1990, ApJS, 74, 347 Kukula, M. J., Pedlar, A., Baum, S. A. & O'Dea, C. 1995, MN- RAS, 26, 1262 L´ıpari, S., Tsvetanov, Z. & Macchetto, F. 1993, ApJ, 405, 186 Malkan, M. A. & Filippenko, A. V. 1983, ApJ, 275, 477 Marshall, F. E., Boldt, C. A., Holt, S. S., Mushotzky, R. F., Pravdo, S. H., Rotschild, R. E., Serlemitsos, P. J. 1979, ApJS, 40, 657 Martin, P. G., Thompson, I. B., Maza, J. & Angel, J. R. P. 1983, ApJ, 266, 470 Miller, J. S. & Goodrich, R. W. 1990, ApJ, 355, 456 Morris, S., Ward, M., Whittle, M., Wilson, A. S. & Taylor, K. 1985, MNRAS, 216, 193 Mouri, H., Taniguchi, Y., Kawara, K. & Nishida, M. 1989, ApJ, 346, L73 Mulchaey, J. S., Wilson, A. S. & Tsvetanov, Z. 1996, ApJS, 102, 309 Osterbrock, D. E. & De Robertis, M. M. 1985, PASP, 97, 1129 Pence, W. D. & Blackman, C. P. 1984, MNRAS, 207, 9 Phillips, M. M. 1979, ApJ, 227, L121 Phillips, M. M., Charles, P. A. & Baldwin, J. A. 1983, ApJ, 266, 485 Phillips, M. M., Jenkins, C. R., Dopita, M. A., Sadler, E. M. & Binette, L. 1986, AJ, 91, 1062 Phillips, M. M., Pagel, B. E. J., Edmunds, M. G. & D´ıaz, A. I. 1984, MNRAS, 210, 701 Pogge, R. W. 1989a, ApJ, 345, 730 Pogge, R. W. 1989b, ApJS, 71, 433 Pogge, R. W. & De Robertis, M. M. 1993, ApJ, 404, 563 Ruiz, M., Rieke, G. H. & Schmitt, G. D. 1994, ApJ, 423, 608 Saikia, D. J., Pedlar, A., Unger, S. W. & Axon, D. J. 1994, MN- RAS, 270, 46 Schmidt, A. A., Bica, E. & Alloin, D. 1990, MNRAS, 243, 620 Schmitt, H. R. 1994, Msc. Thesis, Universidade Federal do Rio Grande do Sul, Brazil Schmitt, H. R., Bica, E. & Pastoriza, M. G. 1996, MNRAS, 278, 965 Schmitt, H. R. & Kinney, A. L. 1996, ApJ, 463, 497 Schmitt, H. R. & Storchi-Bergmann, T. 1995, MNRAS, 276, 592 240, 225 Terlevich, R., Melnick, J., Mesagosa, J., Moles, M. & Copetti, M. V. F. 1991, A&AS, 91, 285 Thuan, T. X. 1984, ApJ, 281, 126 Tran, H. D. 1995a, ApJ, 440, 565 Tran, H. D. 1995b, ApJ, 440, 578 Tran, H. D. 1995c, ApJ, 440, 597 Tran, H. D., Miller, J. S. & Kay, L. E. 1992, ApJ, 397, 452 Ulvestad, J. S. & Wilson, A. S. 1984a, ApJ, 278, 544 Ulvestad, J. S. & Wilson, A. S. 1984b, ApJ, 285, 439 Ulvestad, J. S. & Wilson, A. S. 1989, ApJ, 343, 659 Unger, S. W., Pedlar, A., Neff, S. G. & de Bruyn, A. G. 1984, MNRAS, 209, 15p V´eron-Cetty, M.-P., V´eron, P. 1986, A&AS, 66, 335 V´eron, M. P., V´eron, P. & Zuiderwijk, E. J. 1981, A&A, 98, 34 Weaver, K. A., Wilson, A. S. & Baldwin, J. A. 1991, ApJ, 366, 50 Winge C., Peterson B. M., Horne K., Pogge R. W., Pastoriza M., Storchi-Bergmann T. 1995, ApJ, 445, 680 Winkler, H. 1992, MNRAS, 257, 677 Winkler, H., Glass, I. S., van Wyk, F., Marang, F., Spencer Jones, J. H., Buckley, D. A. H. & Sekiguchi, K. 1992, MNRAS, 257, 659 Wood, K. S., Meekins, J. F., Yentis, D. J., Smathers, H. W., McNutt, D. P., Bleach, R. D., Byran, E. T., Friedman, H., Meidav, M. 1984, ApJS, 56, 507 Yee, H. K. C. 1980, ApJ, 241, 894 c(cid:13) 0000 RAS, MNRAS 000, 000–000 This figure "FIG1.jpg" is available in "jpg"(cid:10) format from: http://arxiv.org/ps/astro-ph/9801309v1 This figure "FIG_ZOOMab.jpg" is available in "jpg"(cid:10) format from: http://arxiv.org/ps/astro-ph/9801309v1 This figure "FIG_ZOOMcd.jpg" is available in "jpg"(cid:10) format from: http://arxiv.org/ps/astro-ph/9801309v1
astro-ph/0605548
2
0605
2006-10-24T03:51:55
Imprint of Gravitational Lensing by Population III Stars in Gamma Ray Burst Light Curves
[ "astro-ph" ]
We propose a novel method to extract the imprint of gravitational lensing by Pop III stars in the light curves of Gamma Ray Bursts (GRBs). Significant portions of GRBs can originate in hypernovae of Pop III stars and be gravitationally lensed by foreground Pop III stars or their remnants. If the lens mass is on the order of $10^2-10^3M_\odot$ and the lens redshift is greater than 10, the time delay between two lensed images of a GRB is $\approx 1$s and the image separation is $\approx 10 \mu$as. Although it is difficult to resolve the two lensed images spatially with current facilities, the light curves of two images are superimposed with a delay of $\approx 1$ s. GRB light curves usually exhibit noticeable variability, where each spike is less than 1s. If a GRB is lensed, all spikes are superimposed with the same time delay. Hence, if the autocorrelation of light curve with changing time interval is calculated, it should show the resonance at the time delay of lensed images. Applying this autocorrelation method to GRB light curves which are archived as the {\it BATSE} catalogue, we demonstrate that more than half light curves can show the recognizable resonance, if they are lensed. Furthermore, in 1821 GRBs we actually find one candidate of GRB lensed by a Pop III star, which may be located at redshift 20-200. The present method is quite straightforward and therefore provides an effective tool to search for Pop III stars at redshift greater than 10. Using this method, we may find more candidates of GRBs lensed by Pop III stars in the data by the {\it Swift} satellite.
astro-ph
astro-ph
Draft version September 10, 2018 Preprint typeset using LATEX style emulateapj v. 14/09/00 6 0 0 2 t c O 4 2 2 v 8 4 5 5 0 6 0 / h p - o r t s a : v i X r a IMPRINT OF GRAVITATIONAL LENSING BY POPULATION III STARS IN GAMMA RAY BURST LIGHT CURVES Y. Hirose1, M. Umemura1, A. Yonehara2,3, and J. Sato1 [email protected] Draft version September 10, 2018 ABSTRACT We propose a novel method to extract the imprint of gravitational lensing by Pop III stars in the light curves of Gamma Ray Bursts (GRBs). Significant portions of GRBs can originate in hypernovae of Pop III stars and be gravitationally lensed by foreground Pop III stars or their remnants. If the lens mass is on the order of 102 − 103M⊙ and the lens redshift is greater than 10, the time delay between two lensed images of a GRB is ≈ 1s and the image separation is ≈ 10 µas. Although it is difficult to resolve the two lensed images spatially with current facilities, the light curves of two images are superimposed with a delay of ≈ 1 s. GRB light curves usually exhibit noticeable variability, where each spike is less than 1s. If a GRB is lensed, all spikes are superimposed with the same time delay. Hence, if the autocorrelation of light curve with changing time interval is calculated, it should show the resonance at the time delay of lensed images. Applying this autocorrelation method to GRB light curves which are archived as the BATSE catalogue, we demonstrate that more than half of the light curves can show the recognizable resonance, if they are lensed. Furthermore, in 1821 GRBs we actually find one candidate of GRB lensed by a Pop III star, which may be located at redshift 20−200. The present method is quite straightforward and therefore provides an effective tool to search for Pop III stars at redshift greater than 10. Using this method, we may find more candidates of GRBs lensed by Pop III stars in the data by the Swift satellite. Subject headings: gamma-ray burst -- -gravitational lensing -- PopIII stars 1. introduction The recent observation of the cosmic microwave background by Wilkinson Microwave Anisotropy Probe (WMAP) suggests that the reionization of the universe took place at redshifts of 8 . z . 14 (Spergel et al. 2003; Kogut et al. 2003; Page et al. 2006; Spergel, D., et al. 2006). This result implies that first generation stars (Pop III stars) were possibly born at z & 10 if UV photons emit- ted from Pop III stars are responsible for cosmic reioniza- tion. However, there is no direct evidence that Pop III stars actually formed at z > 10. Obviously, it is impos- sible with current or near future facilities to detect the emission from a Pop III star at such high redshifts (e.g., Mizusawa et al. 2004). But, gamma ray bursts (GRBs) can be detected even at z > 10, if they arise there. GRBs are the only currently available tool for probing first gen- eration objects in the universe. If one uses the data of absolute magnitude for GRBs with known redshifts, one can expect that more than half GRBs are detectable if they occur at z > 10 (Lamb & Reichart 2000). Roughly 3000 GRBs have been detected to date, but redshifts have been measured only for 30 GRBs (Bloom et al. 2003), among which the most distant is GRB 000131 at z = 4.5 (Andersen et al. 2000). But, if empirical relations be- tween the spectral properties and the absolute magnitude are used, the GRBs detected to date may include events at z > 10 (Lloyd-Ronning et al. 2002; Yonetoku et al. 2004; Murakami et al. 2005). In addition, recently a new GRB satellite, Swift, has been launched (Gehrels et al. 2004). Swift is now accumulating more data of GRBs at a high rate. Recently, GRB 050904 detected by Swift, in terms of metal absorption lines and Lyman break, turns out to have occurred at z = 6.295 (Kawai et al. 2006). The discovery of the association between GRB 030329 and SN 2003dh has demonstrated that at least a portion of long bursts in GRBs are caused by collapse of massive stars (Kawabata et al. 2003; Price et al. 2003; Uemura et al. 2003). On the other hand, Pop III stars are ex- pected to form in a top-heavy fashion with the peak at 100−103M⊙ in the initial mass function (IMF) (e.g., Naka- mura & Umemura 2001 and references therein). Also, the theoretical study by Heger & Woosley (2002) suggests that Pop III stars between 100M⊙ and 140M⊙ may end their lives as GRBs accompanied by the core collapse into black holes. Heger et al. (2003) estimate, assuming the IMF by Nakamura & Umemura (2001), that 5% of Pop III stars can result in GRBs. In the context of cold dark matter cosmology, more than 10-30% of GRBs are expected to occur at z & 10, assuming that the redshift distributions of GRBs trace the cosmic star formation history (Bromm & Loeb 2002). Thus, observed GRBs highly probably con- tain GRB originating from Pop III stars at z & 10. The firm methods to measure redshifts are the detec- tion of absorption and/or emission lines of host galaxies of GRBs (e.g., Metzger et al. 1997), or the Lyα absorption edge in afterglow (Andersen et al. 2000). However, these methods cannot be applied for all GRBs, but have been successful to determine redshifts only for 30 GRBs (Bloom et al. 2003). Instead, some empirical laws have been ap- plied to much more GRBs. They include a variability- 1 Center for Computational Sciences, University of Tsukuba, Ibaraki 305-8577, Japan 2 Department of Physics, University of Tokyo, Hongo, Bunkyo, Tokyo 113-0033, Japan 3 Astronomisches Rechen-Institut, Zentrum fur Astronomie, Universitat Heidelberg, Monchhofstrasse 12-14, 69120 Heidelberg, Germany 1 2 Hirose et al. luminosity relation (Fenimore & Ramirez-Ruiz 2000), a lag-luminosity relation (Norris et al. 2000), Ep-luminosity relation (Amati et al. 2002), and the spectral peak energy- to-luminosity relation (Yonetoku et al. 2004). Apply- ing these relations to GRBs, the redshift distributions of GRBs are derived (Fenimore & Ramirez-Ruiz 2000; Nor- ris et al. 2000; Schaefer et al 2001; Lloyd-Ronning et al. 2002; Yonetoku et al. 2004). Some analyses conclude that a portion of GRBs are located at z & 10. However, it is still controversial whether such an indirect technique is correct or not. In this paper, we propose a novel method to constrain the redshifts of GRBs that may originate from Pop III stars at z & 10. In the present method, the effects of the gravitational lensing by Pop III stars are consid- ered. The lensing of GRBs is considered for the first time by Paczy´nski (1986, 1987), who proposed the pos- sibility that a soft gamma-ray repeater is produced by gravitational lensing of a single burst at cosmological dis- tance. Also, Loeb & Perna (1998) first discussed the mi- crolensing effect of GRB afterglows, and Garnavich; Loeb & Stanek (2000) found the candidate microlensed after- glow (GRB 000301C). The rates of such events are further discussed from theoretical points of view (Koopmans & Wambsganss 2001; Wyithe & Turner 2002; Baltz & Hui 2005). Blaes & Webster (1992) argue the method to de- tect cosmological clumped dark matter by using the prob- ability of detectable GRB lensing. Nemiroff et al. (1993) and Marani et al. (1999) search for the compact dark matter candidate using actual GRBs data obtained by the Burst and Transient Source Experiment (BATSE) satel- lite on the Compton Gamma Ray Observatory satellite. They focus on large mass lenses up to 106M⊙, which cause the delay time-scale of several tens-100 s. On the other hand, Williams & Wijers (1997) investigate the influence on GRB light curve of the millisecond gravitational lens- ing caused by each star in a lensing galaxy. In addition, Nemiroff & Marani (1998) argue that it is possible to place constraints on the cosmic density of dark matter, baryons, stars, and so on, by microlensing by stellar mass objects. In the present method, we focus on the gravitational lens- ing by Pop III stars. If the mass of Pop III stars is on the order of 102 − 103M⊙ and the redshift is greater than 10, the time delay between two lensed images of a GRB is ≈ 1 s. Quite advantageously, this time delay is longer than the time resolution (64 ms) of GRB light curves and shorter than the duration of GRB events, which is several tens to 100 sec for long bursts. Thus, we can see the super- imposed light curves of two lensed images. The present method seeks for the imprint of gravitational lensing by Pop III Stars in GRB light curves. We attempt to extract the imprint of lensing by calculating the autocorrelation of light curves. In this paper, we assume a standard ΛCDM cosmo- logical parameter: H0 = 70 km s−1 Mpc−1, ΩM = 0.3, ΩΛ = 0.7, and Ωb = 0.04. The paper is organized as fol- lows: In §2, the formalism of gravitational lensing and the estimation of time delay between two images are provided. In §3, the method to find the evidence of lensing by Pop III stars is proposed. Also, we demonstrate the potentiality of the present method for artificially lensed GRBs, and de- scribe how to determine the redshifts of lensed GRBs. In §4, we apply this method to 1821 GRB data obtained by BATSE, and find a candidate of lensed GRB. §5 is devoted to the conclusions. 2. gravitational lensing We consider a GRB lensed by a foreground Pop III star. Here, we presuppose the lens model of a point mass. The Einstein ring radius gives a typical scale of gravitational lensing, which is expressed as θE ≡(cid:18) 4GML c2 DLS DOSDOL(cid:19)1/2 , (1) where G is the gravity constant, c is the speed of light, ML is the mass of a lens object, and DLS, DOS, and DOL are respectively angular diameter distances between the lens and the source, the observer and the source, and the observer and the lens. A point mass lens produces two images with angular directions of θ = β 2  1 ±s1 + 4(cid:18) θE  β (cid:19)2  , (2) where β is the angle of lens from a line-of-sight to the source. We hereafter express the image with θ > θE by image 1, and that with θ < θE by image 2. The brightness of the image 1 and the image 2 are respectively magnified by A1,2 = 1 4 [(1 + 4f −2)1/2 + (1 + 4f −2)−1/2 ± 2], (3) where f = β/θE. Thus, the image 1 is brighter than the original one, while the image 2 is fainter. In the case of a lens of Pop III star, the Einstein radius is estimated as θE ≃ 10(cid:18) ML 103M⊙(cid:19)1/2 D 4 × 104Mpc!−1/2 µas, (4) where D ≡ DOSDOL/DLS. Obviously, this angular sepa- ration is impossible to resolve by current facilities. Hence, we can just observe the superposition of two images. However, the light curves of two images are superim- posed with a time delay caused by the gravitational lens- ing, as shown in Figure 1. The arrival time of signals for a lensed image is expressed as DOSDOL (1 + zL) (cid:20) 1 2 (θ − β)2 − Ψ(θ)(cid:21) , (5) t(θ) = c DLS where zL is the redshift of lens object, and Ψ is so called lens potential. For the point mass lens model, Ψ is ex- pressed as Ψ(θ) = DLS 4GML DOSDOL c2 lnθ/θC, (6) where θC is constant (Narayan & Bartelmann 1997). Then, the time delay between two images is given by ∆t(zL, ML, f ) = t(θ2) − t(θ1) ∝ ML(1 + zL). (7) It should be noted that ∆t is determined solely by the mass and redshift of lens, regardless of the source redshift. In other words, the time delay places a constraint just on the lens, not on the source. However, if the lens redshift (zL) is determined, it gives the minimum value of the source redshift (zS) since zS must be higher than zL. Fig. 2 illustrates the relation between the time delay ∆t and the magnification ratio between image 1 and image 2, Imprint of Gravitational Lensing 3 assuming the lens redshift of 50. This figure shows that the lens with & 104M⊙ yields the time delay longer than the standard GRB duration, if the typical delay time-scale is assessed by f ≈ 1. (Note that, as shown later, if f be- comes larger than unity, the ratio of magnification becomes smaller and therefore the contribution of image 2 becomes difficult to extract. Also, if f becomes smaller than unity, the probability of lensing goes down. ) On the other hand, the lens with < 10M⊙ leads to ∆t shorter than the time resolution of light curves, and therefore the information of delay is buried. The mass scale of 102 − 103M⊙ expected for Pop III stars gives 10−1 s . ∆t . 1 s, which is longer than the time resolution and shorter than the GRB du- ration. Hence, this mass range appears to be suitable for extracting the time delay information. However, the actual GRB light curves generally exhibit variabilities with time-scale shorter than ∆t. Thus, it is not straightforward to extract the time delay information. To demonstrate this difficulty, we show the light curve of GRB 930214 and the artificially lensed light curve in Fig- ure 3, where ML = 103M⊙, zL = 50, and f = 0.5 are assumed. The time delay is ∆t = 1.0 s in this case. This figure clearly shows that if we observe only the superim- posed lensed light curve, it seems impossible to recognize by appearance that this light curve is lensed. Hence, we invoke a new technique to discriminate a lensed GRB from unlensed one. 3. autocorrelation method 3.1. Theory We pay attention to the fact that all spikes in a light curve are individually lensed. Then, many pairs with time separation of ∆t appear in the light curve, as schemati- cally shown in Figure 4 (b). To detect those pairs, we em- ploy the autocorrelation method (e.g., Geiger & Schneider 1996). The autocorrelation, C(δt), is defined as C(δt) = Pi I(ti + δt)I(ti) Pi I(ti)2 , (8) where I(ti) is the number of photons contained in a i-th bin in the GRB light curve. If there are pairs with ∆t, the autocorrelation (8) is expected to show the resonance "bump" around ∆t, as shown in Figure 4 (c). Then, we can evaluate the time delay by the existence of this bump. 3.2. Robustness The autocorrelation method is simple and well defined, but the issue we should check is its applicability for the actual GRB light curves. To test the robustness of this method, we produce artificially lensed light curves for GRBs in BATSE archived data, and calculate the auto- correlation. We use 1821 light curves in the BATSE cata- logue with the time resolution of 64 ms 4. Unless otherwise specified, we adopt the data of T90, where T90 is defined by the duration such that the cumulative photon counts increase from 5% to 95% of the total GRB photon counts (Kouveliotou et. al. 1993, Koshut et. al. 1996). Then, the summation in equation (8) is taken in the range of T90 − δt. But, if the data in T90 start with the bins whose time resolution is worse than 1024 ms, we neglect those low resolution bins. 4 http : //cossc.gsfc.nasa.gov/batse/BATSE Ctlg/duration.html In Figure 5, the resultant autocorrelation is shown for 10 GRB light curves. In each panel, a thin solid line repre- sents the autocorrelation for the original light curve, while a thick solid line is the autocorrelation for the artificially lensed light curve, where ML = 103M⊙, zL = 50, and f = 0.5 are assumed, the same as Figure 3, and there- fore the time delay of lensed images is ∆t = 1 s. We can see that there is no bump in C(δt) for the original light curve, whereas a bump emerges around δt = 1 s in C(δt) for the artificially lensed light curve. Note that C(δt) for the artificially lensed light curve is stronger than C(δt) for all of the original light curve, owing to the am- plification by gravitational lensing. C(δt) for the artifi- cially lensed light curve is fit by the polynomial of 8th degree. With using the best fit polynomial F (δt), we define the dispersion, σ, of the autocorrelation curve by j=1[C(δtj) − F (δtj)]2/n, where n is the number of bins. The levels of ±3σ are shown by dashed lines. The zoomed view around the bump of C(δt) for the artificially lensed light curve is also shown in each small panel. For these GRBs, bumps exceeding 3σ appear if lensed, cor- responding to the time delay between two lensed images, ∆t. σ2 =Pn 3.3. Dependence on f In fact, not all GRB light curves exhibit bumps in C(δt) when lensed. The fraction of GRBs which show bumps exceeding 3σ in C(δt) depends on the value of f = β/θE. Also, the time delay ∆t and the magnification A1,2 de- pend on f . To demonstrate this, we show the dependence on f of the autocorrelation for the artificially lensed light curve of GRB 930214, in Figure 6. It is clear that if f is larger, the bump is suppressed. This is because the con- tribution of image 2 becomes smaller with increasing f . In this GRB, the bump over 3σ disappears at f > 1.0. The value of f at which the bump disappears differs in each GRB. Using 220 GRB data, which are used in Fen- imore & Ramirez-Ruiz (2000), we obtain the fraction of GRBs which show bumps over 3σ in C(δt) as a function of f . In this analysis, ∆t = 1.0 s is assumed. The resul- tant fraction is shown in Figure 7 (a). For f = 0.5, about a half of GRBs show bumps exceeding 3σ in C(δt). In contrast, the cross-section of the gravitational lensing is proportional to f 2, which is also shown in Figure 7 (a). We evaluate the probability density of GRBs exhibiting bumps over 3σ by multiplying the fraction for which a bump appears by f 2. The normalized probability density against f is shown in Figure 7 (b). As a result, the proba- bility is peaked around f = 1 and the standard deviation corresponds to ∆f ≈ 0.25. It is noted that this probabil- ity density is found to be hardly dependent on the value of ∆t. 3.4. Optical depth Here, we estimate the optical depth of gravitational lens- ing by Pop III stars. If we assume that the fraction α of baryonic matter composes Pop III stars at z ≥ zIII, then the optical depth is given by τ (zS) =Z zS zIII nL(zL)σL cdt dzL dzL, (9) 4 Hirose et al. where σL is the cross-section given as σL = π(DOLθE)2 and nL(zL) is the number density of the lens objects given as nL(zL) = α (1 + zL)3. (10) 3H 2 0 8πG Ωb ML Then, the optical depth (9) is calculated as τ (zS) = 3 5 αΩb(cid:26)(cid:20) (1 + zS)5/2 + 1 −(cid:20) (1 + zIII)5/2 + 1 (1 + zS)5/2 − 1(cid:21) ln(1 + zS) (1 + zIII)5/2 − 1(cid:21) ln(1 + zIII)(cid:27) (11) in the Einstein-de Sitter universe (Turner, Ostriker, & Gott 1984; Turner & Umemura 1997). Here, we assume α = 0.1 and zIII = 10. The resultant optical depth of Pop III star lensing is shown in Fig. 8. Since τ (zS) is the probability that a source is located inside the Einstein ring radius (f ≤ 1), it is not the probability of bump de- tection. The probability of bump appearance, p(f ), is a decreasing function of f , as shown in Fig. 7 (a). Since the optical depth that f is in the range of [f, f +df ] is given by dτ (zS) = τ (zS)df 2 = τ (zS)2f df , the probability of bump detection is given by P (zS) =Z ∞ 0 2f p(f )τ (zS)df . (12) The resultant bump detection probability is also shown in Fig. 8. From this figure, the probability turns out to be ≈ 0.001 for zS = 20 − 40. If we take into account that more than 10-30% of GRBs occur at z & 10 (Bromm & Loeb 2002), the expectation number of bump detection for lensed GRBs is assessed to be one in a few thousand GRBs. 4. a candidate for grb lensed at z ≈ 60 4.1. Data analysis As shown above, the autocorrelation of intrinsic light curves exhibits no bumps for almost all GRBs. But, a few in 1000 GRBs might show bumps in C(δt) even for intrin- sic light curves. Hence, we calculate the autocorrelation of all GRB light curves available in BATSE catalogue, which amount to 1821 GRBs. As a result, we have found one candidate, GRB 940919 (BATSE trigger number 3174), in which a 3σ bump in C(δt) appears. The light curve and the autocorrelation of this GRB is shown in Figure 9. As seen in panel (b), a bump exceeding 3σ appears at ∆t = 0.96 s. 4.2. Statistical significance To check the statistical significance of a bump in C(δt) of GRB 940919, we make a test with mock light curves. Here, we generate mock light curves using a smoothed cor- relation function that dose not show any bump, and inves- tigate whether bumps appear in correlation functions just from pure statistical fluctuations. From Wiener-Kihntchine theorem, the power spectrum of light curves are given by I(ω)2 = F [Pi I(ti)I(ti + δt)] , where F denotes the Fourier transformation and ω = 2π/δt. We can generate mock light curves by the inverse (13) Fourier transformation of I(ω). Here, in order to add fluc- tuation to I(ω), we take random Gaussian distributions, where I(ω) is the standard deviation and the phase is random in the range of [0, 2π]. Then, the mock light curve is given by (14) I(t) = F −1[I(ω)] =Pω I(ω) cos(−φ − ωt), where I(ω) is a random sample in the Gaussian distribu- tion, and φ is the random phase shift from 0 to 2π. We produce 2000 mock light curves using a correlation func- tion, and recalculate the autocorrelation C(δt) by equation (8). As a result, we have found that no bump higher than 3σ appears in the C(δt) of 2000 mock light curves. A part of the recalculated C(δt) are shown in Figure 10. Thus, it is unlikely that a bump in the correlation arises as a result of pure statistical fluctuations. 4.3. Light curve decomposition As a further test for the lensing of GRB 940919 light curve, we attempt to decompose the light curve, assuming that it is the superposition of two lensed light curves with ∆t = 0.96 s, and analyze the decomposed light curves. The decomposition is made by the following recurrence formula; Itot(t) = I1(t) + I2(t), I2(t) = A2 A1 I1(t − ∆t), (15) (16) where Itot(t) is the observed intensity, and I1(t) and I2(t) are intensities for image 1 and 2, respectively. If these two equations are combined, I1(t) can be expressed by I1(t) = N Xj=0(cid:18)− A2 A1(cid:19)j Itot(t − j∆t). (17) The summation is taken in the range of t− j∆t ≧ 0, where t = 0 is the starting point of T90. Then, we can derive also I2(t) by equation (15). In Figure 11, the decomposed light curves are shown in the case of f = 1 which is the most probable case as shown in §3.3. The application of this decomposition method for the finite amount of data does not guarantee that the light curve is successfully decom- posed into two lensed light curves. Hence, to check the validity of this decomposition method, we calculate the cross-correlation of two decomposed light curves by Cc(δt) = , (18) Pi I1(ti)I2(ti + δt) pPi I1(ti)2pPi I2(ti + δt)2 where δt is the time shift. The result is shown in Figure 12. As clearly shown in this figure, the cross-correlation is peaked when δt accords with ∆t = 0.96 s. Also, each decomposed light curve shows no bump higher than 3σ in the autocorrelation for a reasonable range of f . Hence, we can conclude that the light curve of GRB 940919 is suc- cessfully decomposed and is likely to be the superposition of two lensed light curves. Imprint of Gravitational Lensing 5 4.4. Redshift estimation Here, we constrain the redshift of the lens object. As shown in equation (7), ∆t just determines ML(1 + zL), ex- cept for f . If the probability density against f (Fig. 7 (b) ) is applied, we can derive a suitable range for ML(1 + zL). The dark gray region in Figure 13 represents the suitable range for ∆t = 0.96 s. On the other hand, the hatched region shows the mass range of Pop III stars obtained by Nakamura & Umemura (2001). Combining these two re- gions, the allowed redshift of the lens object is at least zL ≈ 10. If f = 1 is adopted, the redshift ranges from zL ≈ 20 to zL ≈ 200, where the most probable one is zL ≈ 60. Nonetheless, there still exists another possibility that the lens object is as massive as ∼ 104M⊙ located at zL ≃0, as seen in Fig. 13. Loeb (1993) and Umemura et al. (1993) suggested that relic massive black holes are candidates for such an object. Sasaki & Umemura (1996) place a con- straint on ΩBH from the UV background intensity and the Gunn-Peterson effect in the context of a cold dark mat- ter cosmology. They find that the black hole mass density might be as low as ΩBH/Ωb . 10−3. Therefore, the ex- pected number of bump detection for massive black holes is by two orders of magnitude lower than that for Pop III stars. 5. conclusions To place constraints on the redshifts of GRBs which originate from Pop III stars at z > 10, we have proposed a novel method based on the gravitational lensing effects. If the lens is Pop III stars with 102 − 103M⊙ at z > 10, the time delay between two lensed images of a GRB is ≈ 1 s. This time delay is longer than the time resolution (64 ms) of GRB light curves and shorter than the duration of GRB events. Therefore, if a GRB is lensed, we observe the superposition of two lensed light curves. We have con- sidered the autocorrelation method to extract the imprint of gravitational lensing by Pop III stars in the GRB light curves. Using BATSE data, we have derived the probabil- ity of the resonance bump in the autocorrelation function, which is an indicator for the gravitational lensing. Apply- ing this autocorrelation method to GRB light curves in the BATSE catalogue, we have demonstrated that more than half of the light curves can show resonance bumps, if they are lensed. Furthermore, in 1821 GRB light curves, we have found one candidate of GRB lensed by a Pop III star at z ≈ 60. The present method is quite straightforward and therefore provides an effective tool to search for Pop III stars at redshift greater than 10. Although the num- ber of GRBs with available data is 1821 in this paper, the Swift satellite is now accumulating more GRB data. If the present method is applied for those data, more candidates for GRBs lensed at z > 10 may be found in the future. These can provide a firm evidence of massive Pop III stars born at high redshifts. We thank T. Murakami and D. Yonetoku for helpful in- formation and fruitful discussion, and S.Mao for his valu- able comments. Numerical simulations were performed with facilities at the Center for Computational Sciences, University of Tsukuba. One of the author (AY) acknowl- edges to the Japan Society for the Promotion of Science 09514, Inoue Foundation for Science, and JSPS Postdoc- toral Fellowships for Research Abroad. This work was supported in part by Grants-in-Aid for Scientific Research from MEXT 16002003 (MU). REFERENCES Amati, L., et al. 2002, A&A, 390, 81 Andersen, M. I., et al. 2000, A&A, 364, L54 Baltz, E. A., & Hui, L. 2005, ApJ, 618, 403 Blaes, O. M., & Webster, R. L. 1992, ApJ, 391, L63 Bloom, J. S., Frail, D. A., & Kulkarni, S. R. 2003, ApJ, 594, 674 Bromm, V., & Loeb, A. 2002 ApJ, 575, 111 Fenimore, E. E., & Ramirez-Ruiz, E. 2000, astro-ph/0004176 Garnavich, P. M., Loeb, A., & Stanek, K. Z. 2000, ApJ, 544, L11 Gehrels, N., et al. 2004, ApJ, 611, 1005 Geiger, B., & Schneider, P. 1996, MNRAS, 282, 530 Heger, A., & Woosley, S. E. 2002, ApJ, 567, 532 Heger, A., et al. 2003, ApJ, 591, 288 Kawabata, K. S., et al. 2003, ApJ, 593, L19 Kawai, N. et al. 2006, Nature, 440, 184 Kogut, A., et al. 2003, ApJS, 148, 161 Koopmans, L. V. E., & Wambsganss, J. 2001, MNRAS, 325, 1317 Koshut, T. M., Paciesas, W. S., Kouveliotou, C., van Paradijs, J., Pendleton, G. N., Fishman, G.J., & Meegan, C. A. 1996, ApJ, 463, 570 Kouveliotou, C., et al. 1993, ApJ, 413, L101 Lamb, D. Q., & Reichart, D. E. 2000, ApJ, 536, 1 Lloyd-Ronning, N. M, Fryer, C. L., & Ramirez-Ruiz, E. 2002, ApJ, 574, 554 Loeb, A. 1993, ApJ, 403, 542 Loeb, A., & Perna, R. 1998, ApJ, 495, 597 Marni G. F., Nemiroff, R. J., Norris, J. P., Hurley, K., & Bonnell, J. T. 1999, ApJ, 512, L13 Metzger, M., et al. 1997, Nature, 387, 879 Mizusawa, H., Nishi, R., & Omukai, K. 2004, PASJ, 56, 487 Murakami, T., Yonetoku, D., Umemura, M., Matsubayashi, T. Yamazaki, R. 2005, ApJ, 625, L13 Nakamura, F., & Umemura, M., 2001, ApJ, 548, 19 Narayan, R., & Bartelmann, M. 1999, in Formation of Structure in the Universe, ed. A. Dekel & J. P. Ostriker (Cambridge: Cambridge Univ. Press), 360 Nemiroff, R. J., et al. 1993, ApJ, 414, 36 Nemiroff, R. J., & Marani, G.F. 1998, ApJ, 494, L173 Norris, J. P., Marani, G. F., & Bonnell, J. T. 2000, ApJ, 534, 248 Paczy´nski, B. 1986, ApJ, 308, L43 Paczy´nski, B. 1987, ApJ, 317, 51 Page, L. et al. 2006, astro-ph/0603450 Price, P. A., et al 2003, Nature, 423, 844 Sasaki, S., & Umemura, M. 1996, ApJ, 462, 104 Schaefer, B. E., Deng, M., & Band D. L. 2001, ApJ, 563, L123 Spergel, D., et al. 2003, ApJS, 148, 175 Spergel, D., et al. 2006, astro-ph/0603449 Turner, E. L., Ostriker, J. P., & Gott, J. R. 1984, ApJ, 284, 1 Turner, E. L., & Umemura, M. 1997, ApJ, 483, 603 Uemura, M., et al. 2003, Nature, 423, 843 Umemura, M., Loeb, A., & Turner, E. L. 1993, ApJ, 419, 459 Yonetoku, D., Murakami, T., Nakamura, T., Yamazaki, R., Inoue, A., K., & Ioka, K. 2004, ApJ, 609, 935 Williams, L. L. R., & Wijers, R. A. M. J. 1997, MNRAS, 286, L11 Wyithe, J. S. B., & Turner, E. L. 2002, ApJ, 575, 650 6 Hirose et al. Fig. 1. -- Schematic diagram of lensed GRB light curve. The dotted line represents the original light curve, while the dashed and dot-dashed lines represent the light curves of image 1 and image 2, respectively. The superimposed light curve of lensed images is shown by the solid line. The intensities are normalized by the maximum intensity of original light curve. Fig. 2. -- Relation between the time delay and the magnification ratio (A2/A1) between two images. The redshift of lens object is fixed to 50. The curves show lens masses of 10-105M⊙ with an increment of 1 order of magnitude. In the right vertical axis, f = β/θE, is also shown corresponding to the magnification ratio. The time resolution of BATSE data, ∆t = 64 ms, is represented by the leftmost gray region. Imprint of Gravitational Lensing 7 Fig. 3. -- The original light curve of GRB 930214 obtained by BATSE and artificially lensed light curves are presented. The lens is assumed to be a Pop III star with ML = 103M⊙ at zL = 50, while the source is located at zS = 51. Lensed image 1, image 2, and the total light curves are shown. The time delay between two images is ∆t = 1.0 s under the assumption of f = 0.5. Fig. 4. -- (a) Schematic figure of a spiky GRB light curve. (b) The light curve as the superposition of two lensed light curves with the time delay of ∆t. (c) Autocorrelation function calculated for light curve (b). 8 Hirose et al. (a) GRB 921118 (b) GRB 930214 (c) GRB 930405 (d) GRB 930704 (e) GRB 931013B ) t δ ( C ) t δ ( C 1 0.999 0.998 0.997 0.996 0.995 0.994 1 0.999 0.998 0.997 0.996 0.995 0.994 0.993 1 0.9995 0.999 0.9985 ) t δ ( C 0.998 0.9975 0.997 0.9965 0.996 ) t δ ( C ) t δ ( C 1 0.998 0.996 0.994 0.992 0.99 1 0.999 0.998 0.997 0.996 0.995 0.994 0.993 0 0.5 1 1.5 δt [s] 2 2.5 3 ) t δ ( C 1 0.9995 0.999 0.9985 0.998 0.9975 0.997 0.9965 0.996 ) t δ ( C ) t δ ( C 1 0.999 0.998 0.997 0.996 0.995 0.994 1 0.999 0.998 0.997 0.996 0.995 0.994 0.993 0.992 1 0.999 0.998 ) t δ ( C 0.997 0.996 0.995 1 0.9995 0.999 0.9985 ) t δ ( C 0.998 0.9975 0.997 0.9965 0.996 0.9955 0.995 0 (f) GRB 940312 (g) GRB 940722 (h) GRB 950808 (i) GRB 951124 (j) GRB 960414 0.5 1 1.5 δt [s] 2 2.5 3 Fig. 5. -- Autocorrelation against δt for 10 GRB light curves. In each panel, a thin curve shows the autocorrelation for the original light curve, while a thick curve is that for the artificially lensed light curve. Here, ML = 103M⊙, zL = 50, and f = 0.5 are assumed and therefore the time delay of lensed images is ∆t = 1 s. A dotted curve is the best fitting for the autocorrelation for lensed light curve, and two dashed curves show ±3σ level from the best fitting. In each panel, a zoomed view around a correlation bump is also shown. Imprint of Gravitational Lensing 9 ) t δ ( C ) t δ ( C 1.001 1 0.999 0.998 0.997 0.996 0.995 0.994 0.993 0.992 1.001 1 0.999 0.998 0.997 0.996 0.995 0.994 0.993 0.992 (a) f=0.1 0 0.5 1 1.5 δt [s] 2 2.5 3 (c) f=1.0 0 0.5 1 1.5 δt [s] 2 2.5 3 ) t δ ( C ) t δ ( C 1.001 1 0.999 0.998 0.997 0.996 0.995 0.994 0.993 0.992 1.001 1 0.999 0.998 0.997 0.996 0.995 0.994 0.993 0.992 (b) f=0.5 0 0.5 1 1.5 δt [s] 2 2.5 3 (d) f=2.0 0 0.5 1 1.5 δt [s] 2 2.5 3 Fig. 6. -- Dependence on f = β/θE of the autocorrelation for the artificially lensed light curve of GRB 930214, which is the same GRB as panel (b) in Fig. 5. The time delay is ∆t = 0.1 s. The meanings of curves are the same as Fig. 5. It is seen that the bump is weaker for larger f . 10 Hirose et al. 100 (a) ] % [ y t i l i b a b o r p 80 60 40 20 0 0 1 (b) y t i s n e d y t i l i b a b o r p 0.8 0.6 0.4 0.2 0 0 probability f2 0.5 1 f 1.5 2 2.5 f2 8 6 4 2 0 0.5 1 f 1.5 2 2.5 Fig. 7. -- (a) A dot-dashed curve shows the probability of exhibiting bumps exceeding 3σ. Here, ∆t = 1.0 s is assumed. A dashed line is f 2, which is proportional to the cross-section of gravitational lens. (b) The probability density of exhibiting bumps against f . The gray scales represent 1σ, 2σ, and 3σ around f = 1 (from dark to light). Imprint of Gravitational Lensing 11 0.006 0.005 0.004 0.003 0.002 0.001 y t i l i b a b o r p / t h p e d l a c i t p o 0 0 optical depth probability 20 40 zs 60 80 100 Fig. 8. -- Probability of bump detection for lensed GRBs, which is shown by a solid line. A dashed line is the optical depth of gravitational lensing, when a GRB as a source object is located at zS ≥ 10 and 10% of Ωb at zL ≥ 10 contributes to gravitational lensing as a lens object. (a) 740 720 700 680 660 640 620 600 580 560 s m 4 6 / r e b m u n n o t o h p T1 GRB 940919 (b) 1 ) t δ ( C 0.998 0.996 0.994 0.992 0.99 0.988 0.986 540 -5 0 5 10 t[s] 15 20 25 30 0.984 0 0.5 1 2 2.5 3 1.5 δt[s] Fig. 9. -- (a) The intrinsic light curve of GRB 940919. (b) Autocorrelation against δt. A 3σ bump appears at ∆t = 0.96 s. 12 Hirose et al. 1.04 1.02 1 0.98 ) t δ ( C 0.96 0.94 0.92 0 0.5 1 1.5 δt [s] 2 2.5 3 Fig. 10. -- Autocorrelation for mock light curves. A thick curve is the autocorrelation for the original light curve (GRB 000421). Other thin dashed curves are a part of autocorrelations for 2000 mock light curves based on the same GRB. Image 1 640 620 600 580 560 540 520 500 480 s m 4 6 / r e b m u n t n o o h p -5 0 5 10 15 20 25 30 t [s] Image 2 95 90 85 80 75 70 s m 4 6 / r e b m u n t n o o h p -5 0 5 10 15 20 25 30 t [s] Fig. 11. -- The decomposed light curves for GRB 940919, assuming that the observed light curve is the superposition of two lensed light curves with ∆t = 0.96 s. Imprint of Gravitational Lensing 13 1.0005 1 0.9995 0.999 ) t δ ( c C 0.9985 0.998 0.9975 0.997 0.9965 0 0.5 1 1.5 δt [s] 2 2.5 3 Fig. 12. -- The cross-correlation of decomposed light curves shown in Fig. 11. A dotted line is the best fit curve, and short-dashed curves are ±3σ levels. The cross-correlation shows a peak well above 3σ, when δt is the same as ∆t = 0.96 s. / M L M 106 105 104 103 102 10 1 1 10 1+zL 102 103 Fig. 13. -- The dark gray region shows the suitable range of ML(1 + zL) for GRB 940919. Here, 3σ range of f (see Fig. 7b) is adopted. The central thick solid line corresponds to the case of f = 1, which is most probable. The hatched region is the mass range of Pop III stars by Nakamura & Umemura (2001).
astro-ph/0103319
1
0103
2001-03-20T17:20:30
Gamma Ray Bursts statistical properties and limitations on the physical model
[ "astro-ph" ]
The present common view about GRB origin is related to cosmology. There are two evidences in favour of this interpretation. The first is connected with statistics, the second is based on measurements of the redshifts in the GRB optical afterglows. Red shifts in optical afterglows had been observed only in long GRB. Statistical errors, and possibility of galactic origin of short GRB is discussed; their connection with Soft Gamma Repeaters (SGR) is analyzed.
astro-ph
astro-ph
Gamma Ray Bursts statistical properties and limitations on the physical model G.S.Bisnovatyi-Kogan October 31, 2018 Abstract The present common view about GRB origin is related to cosmology. There are two evidences in favour of this interpretation. The first is connected with statistics, the second is based on measurements of the redshifts in the GRB optical afterglows. Red shifts in optical afterglows had been observed only in long GRB. Statistical errors, and possibility of galactic origin of short GRB is discussed; their connection with Soft Gamma Repeaters (SGR) is analyzed. 1 Introduction Cosmological origin of GRB had been first suggested in [Prilutsky and Usov(1975)] soon after their discovery. The present model of cosmological GRB based on production of gamma quanta from neutrino collisions ν + ¯ν → e+ + e− was first considered in [Berezinsky and Prilutsky(1987)]. The efficiency of transforma- ∼ 6 × 1053 ergs into gamma quanta was estimated tion of the neutrino flux Wν ¯ν as ∼ 6 × 10−6, giving a pulse Wγ ∼ 3 × 1048 ergs. It could explain the cos- mological GRB only at rather narrow pulse beam. In the giant GRB 990123 the isotropic energy production is very large [Kulkarni et al.(1999)] in gamma ≈ 2.3 × 1054 ergs, and in optics Wopt ∼ 1051 ergs. Simultaneous strong Wγ beaming in gamma and optical bands is rather unplausible. Strong beaming would modify the observed smooth optical light curve in presence of a source rotation. Some problems in GRB interpretation and modelling are discussed. 2 Statistics and restrictions to the model BATSE data start to deviate from the uniform distribution with 3/2 slope at rather large fluences, for which KONUS data are well defined. Analysis of KONUS data had been done in [Higdon and Schmidt(1990)]. Taking into ac- count selection effects, the resulting value V /Vmax = 0.45 ± 0.03 was obtained. KONUS data had been obtained in conditions of constant background. Similar analysis [Schmidt(1999)] of BATSE data, obtained in conditions of substantially 1 variable background, gave resulting V /Vmax = 0.334 ± 0.008. These two results seems to be in contradiction, because KONUS sensitivity was only 3 times less than that of BATSE, where deviations from the uniform distribution V /Vmax = 0.5 in BATSE data are still rather large [Fishman and Meegan(1995)]. In presence of a threshold deviations of V /Vmax from its uniform Euclidean value 0.5 may be connected with the errors in determination of the burst peak luminosity or total fluence [Bisnovatyi-Kogan(1997)]. Such errors may be con- nected with spectral differences, variable sensitivity of detectors for bursts com- ing from different directions, variable background. All these reasons lead to underestimation of the burst luminosity, and decrease the slope of the curve log N (log S). There is no angular correlation between GRB and sample of any other objects in the universe. From the energy conservation law it follows W < M c2, where M is a mass of the source. A proper account of physical laws put much stronger restrictions to the energy output. Calculations of ns-ns collisions gave energy output in (X, γ) region not exceeding 1050 ergs [Ruffert and Janka(1998)], and similar re- sults characterize ns-bh collision [Ruffert and Janka(1999)]. Magnetorotational explosion does not give larger energy output in (X, γ) region, transforming about 5% of the rotational energy into a kinetic one [Ardeljan et al.(2000)]. The problems with vaguely defined "hypernova" model had been discussed in [Blinnikov and Postnov(1998)]. The largest γ-ray production efficiency, close to 100%M c2 may be expected, if GRB originate from matter-antimatter star collisions. That arises a problem of antistar creation in the early universe. Simultaneous γ, X and optical observations in GRB, accompanied by spec- tral and polarization experiments are very important. Search of hard X-ray lines and of annihilation 0.511 keV, line declared by KONUS, remain to be a puzzle which should be solved. Cosmological GRB explosion in a dense molecular cloud would lead to a specific optical light curve [Bisnovatyi-Kogan and Timokhin(1997)], which discovery would reveal conditions in the region of cosmological GRB ex- plosion. Study of hard γ afterglow, similar to the one observed by EGRET [Schneid et al.(1992)] is expected in a near future. 3 Short GRB and SGR All afterglows had been measured only for long bursts. It cannot be excluded that short bursts could have another, may be Galactic origin. There is also a possibility, that short bursts are connected with a giant bursts observed in 3 soft gamma repeaters (from 4 known). At larger distances only giant bursts, appeared as short GRB could be observed. If we accept the present inter- pretation of SRG, as galactic and LMC sources at distances 10-50 kpc, than only giant bursts should be visible in the nearest galaxies as weak (about few 10−7) short GRB. Taking into account that Andromeda is ∼ 4 times more massive than our Galaxy [Vorontsov-Velyaminov(1972)], we should expect [Bisnovatyi-Kogan(1999)] to see about 10 short GRB in its direction during the 2 observation time, while no one was yet observed. Another large galaxy in the local group of galaxies Maffei IC 1805 is also more massive than the Galaxy, and short GRB from it are also expected. Presently SRG are interpreted as young neutron stars with very strong mag- netic field - "magnetars" [Duncan and Thompson(1992)]. The estimation of the distance and, consequently, the luminosity is based on SGR identification with supernovae remnants (SNR), leading to large energy losses. This interpretation has several theoretical objections [Bisnovatyi-Kogan(1999)]. 1. Hard gamma pulsars observed in 3 SGR have luminosities strongly ex- ceeding the critical Eddington luminosity. At such luminosity a strong mass loss should smear out the pulses. 2. Rotational energy losses estimated from the period increase rate are much smaller than the observed gamma and X-ray luminosity even in a quies- cent state. In the magnetar model the energy comes from the annihilation of magnetic field. Such annihilation should be accompanied by creation of ener- getic electrons and radio-emission. The radio-emission of SGR is very weak, its discovery is very difficult, and still not firmly established. 3. Giant bursts observed in 3 SGR at present interpretation are accompanies by a huge energy production, part of which should go into particle acceleration and kinetic energy outbursts. It should influence the near-by SNR, and produce a visible changes in radio and optics, similar to those produced by pulsar glitches in the Crab nebula, when much smaller amount of energy is released. No such changes had been reported up to now, probably because they have not been present there. Another interpretation of SGR, free of these contradictions needs a smaller distance to SGR, what is possible if the connection with SNR would not be confirmed. Note, that all SGR are situated at the very edge, or even outside of the SNR envelope, requiring very high 1000-3000 km/s speed of the neutron star. Refusing this connection and suggesting ∼ 10 − 30 times smaller distance to SGR would remove the upper objections. The even smaller distances are less probable, because most SGR are situated in the galactic disk, and so should be situated at distances larger than this disk thickness. Existence of one SGR outside the galactic disk direction could indicate to its big age during which it could leave the galactic disk. Discovery of big population of neutron stars in the globular clusters and in the galactic bulge, as recycled pulsars, indicate to existence of neutron stars in the whole volume of the Galaxy. 4 Acknowledgments This work was partially supported by RFBR grant 99-02-18180 and INTAS-ESA grant 120. The authors are grateful to Prof. J.C.Wheeler and the Organizing Committee for support and hospitality, and to B.Schaefer for discussion. 3 References [Ardeljan et al.(2000)] Ardeljan, N. V., Bisnovatyi-Kogan, G. S., Moiseenko, S. G., A&A, 355, 1181-1190 (2000). [Berezinsky and Prilutsky(1987)] Berezinsky, V. S., Prilutsky, O. F., A&A, 175, 309-311 (1987). [Bisnovatyi-Kogan(1997)] Bisnovatyi-Kogan, G. S., A&A, 324, 573-577 (1997). [Bisnovatyi-Kogan(1999)] Bisnovatyi-Kogan, G. S., Astro-ph/9911275 ; Proc. Workshop Vulcano 1999. [Bisnovatyi-Kogan and Timokhin(1997)] Bisnovatyi-Kogan, G. S., Timokhin, A. N.Astron. Zh., 74, 483-496 (1997). [Blinnikov and Postnov(1998)] Blinnikov, S. I., Postnov, K. A., Month. Not. R.A.S., 293, L29-L32 (1998). [Duncan and Thompson(1992)] Duncan, R. C., Thompson, C., ApJ Lett., 392, L9-L13 (1992). [Fishman and Meegan(1995)] Fishman, G. J., Meegan, C. A., Ann. Rev. A&A, 33, 415-458 (1995). [Higdon and Schmidt(1990)] Higdon, J. C., Schmidt, M., ApJ, 355, 13-17 (1990). [Kulkarni et al.(1999)] Kulkarni, S. R. et al., Nature, 398, 389-399 (1999). [Prilutsky and Usov(1975)] Prilutsky, O. F., Usov, V. V., Astrophys. Sp. Sci., 34, 387-393 (1974). [Ruffert and Janka(1998)] Ruffert, M., Janka, H.-T., A&A, 338, 535-555 (1998). [Ruffert and Janka(1999)] Ruffert, M., Janka, H.-T., A&A, 344, 573-606 (1999). [Schmidt(1999)] Schmidt, M., ApJ Lett., 523, L117-L120 (1999). [Schneid et al.(1992)] Schneid, E. J. et al., A&A, 255, L13-L16 (1992). [Vorontsov-Velyaminov(1972)] Vorontsov-Velyaminov, B. A., Extragalactic As- tronomy, Nauka. Moscow (1972). 4
astro-ph/0401596
1
0401
2004-01-28T16:43:37
QSO + galaxy association and discrepant redshifts in NEQ3
[ "astro-ph" ]
Spectroscopy and deep imaging of the group NEQ3 are presented. This system is formed by three compact objects with relative separations of ~2.6 and ~2.8 arcsec, and a lenticular galaxy at ~17 arcsec from the geometric centre of the group. A diffuse filament is located on a line joining the three compact objects and the main galaxy. Analysis of these observations confirms the redshift previously known for three of the objects (z=0.1239 for the main galaxy, and z=0.1935 and 0.1939 for two of the compact objects). We have also determined the previously unknown redshift of the third compact object as z=0.2229. Using the relative strength and width of the main spectral lines, we have classified the compact objects as two HII galaxies and one QSO (the object at z=0.1935). With cross-correlation techniques, we have tentatively estimated the redshift of the filament as z=0.19 (although a weaker component also appears at z=0.12) so that it is probably associated with the halo of the two compact objects at this redshift. The two objects at redshift ~0.19 represent possibly one of the more clear examples of starburst (and perhaps QSO activity) driven by interaction. These, and the relation between these two objects and the other two at (~0.12 and ~0.22) make the nature of this system intriguing, being difficult to explain the whole association on conventional scenarios.
astro-ph
astro-ph
QSO+Galaxy association and discrepant redshifts in NEQ3 C. M. Guti´errez Instituto de Astrof´ısica de Canarias, E-38205, La Laguna, Tenerife, Spain [email protected] and M. L´opez-Corredoira Astronomisches Institut der Universitat Basel. Venusstrasse 7. CH-4102 Binningen, Switzerland ABSTRACT Spectroscopy and deep imaging of the group NEQ3 are presented. This sys- tem is formed by three compact objects with relative separations of ∼2.6 and ∼2.8 arcsec, and a lenticular galaxy at ∼17 arcsec from the geometric centre of the group. A diffuse filament is located on a line joining the three compact objects and the main galaxy. Analysis of these observations confirms the redshift previously known for three of the objects (z = 0.1239 for the main galaxy, and z = 0.1935 and 0.1939 for two of the compact objects). We have also determined the previously unknown redshift of the third compact object as z = 0.2229. Us- ing the relative strength and width of the main spectral lines we have classified the compact objects as two HII galaxies and one QSO (the object at z = 0.1935). With cross-correlation techniques, we have tentatively estimated the redshift of the filament as z = 0.19 (although a weaker component also appears at z = 0.12) so that it is probably associated with the halo of the two compact objects at this redshift. The two objects at redshift ∼ 0.19 represent possibly one of the more clear examples of starburst (and perhaps QSO activity) driven by interaction. These, and the relation between these two objects and the other two at (∼ 0.12 and ∼ 0.22) make the nature of this system intriguing, being difficult to explain the whole association on conventional scenarios. Subject headings: galaxies: photometry, individual (NEQ3), cosmology: anoma- lous redshifts -- 2 -- 1. Introduction In the last three years we have started a programme to study several apparently inter- acting systems with different redshifts. These constitute the so-called problem of anomalous redshifts (e.g. Burbidge 1996 and references therein). The first system that we studied in detail is formed by NGC 7603, its companion galaxy NGC 7603B, the filament apparently connecting both galaxies, and two objects within it (L´opez-Corredoira & Guti´errez 2002). In that article we showed that the two objects in the filament have redshifts of 0.24 and 0.39 respectively. In L´opez-Corredoira & Guti´errez (2003) we have presented subsequent observations on these objects and detected others at high redshift in the same field. Sim- ple probability computations indicate the unlikelihood that this configuration is merely a projection effect. We continued our study of several other systems in different observing campaigns con- ducted during 2002. Here, we present our observations and analysis of the system NEQ3 (the designation was proposed by Sulentic & Arp 1976 and means Non Equilibrium System Number 3). Although a rather intriguing system, the only other study of it, surprisingly, has been conducted by Arp (1977). The system is centred at RA(J2000) = 12h 09m 51.6s, Dec.(J2000) = +39◦ 31′ 29′′ at 10.8 arcmin from the exhaustively studied large active galaxy NGC 4151. Other interesting associations were found by Arp in the region around this galaxy. The system comprises three compact objects with relative separations of 2.6 and 2.8 arcsec with respect to the central object, all of them lying along the minor axis of an apparent lenticular galaxy at ∼17 arcsec. Arp measured a redshift of 0.12 for the main galaxy and 0.19 for two of the compact objects. Arp described the objects as compact with emission lines. A filament is situated along the line connecting the main galaxy and the three compact objects. All the above facts attracted our attention and made this system one of our top priorities. The goals of our study were to ascertain: i) the redshift of the third compact object, ii) the nature of these objects, iii) the redshift of the mentioned filament, iv) whether there are other objects with anomalous redshifts in the near field, v) whether there is evidence for distortions in the profile of the lenticular galaxy, vi) how likely it is that this optical association is merely a projection effect, and vii) what physical mechanisms could originate a system like this. In this article we present the first results of this study. A more detailed analysis covering the photometry and velocity radial profile of the main galaxy and spectra of other objects in the field is in progress and will be presented in a future paper. -- 3 -- 2. Observations The observations presented here comprise broad band imaging and long-slit spectroscopy with intermediate resolution. These observations were taken at the NOT1 (on 2002 December 2) and WHT2 (on 2002 December 28 -- 29). We follow a standard reduction method explained in L´opez-Corredoira & Guti´errez (2003). Table 1 presents a summary of the observations. We observed the field in the Sloan r′ filter. We used the NOT with the ALFOSC instrument, which has a field of view ∼6 arcmin with a spatial sampling of 0.188 arcsecs. The r′ band was relatively calibrated using an image taken with the IAC 80 telescope 3 and corrected to the SDSS r′ filter through the relations given in Smith et al. (2002). The spectra were taken with the red arm of the ISIS instrument and the R158R grism at the WHT. This configuration gives a sampling of 1.63 A and a resolution, measured from the width of the spectral arc lines, of ∼8 A for a slit width of 1.2 arcsec. The spectra have a useful range between ∼3900 and ∼9600 A. Figure 1 presents a false-colour scale of our image in the r′ band. Superimposed are the positions in which we positioned the slits. These positions were chosen to cross the three compact objects, the main galaxy, the filaments within them and several other promising objects in the field. For clarity, the objects analysed here have been arbitrary labelled from 1 to 4. 3. Analysis Figure 2 shows the system in detail. Objects 1, 2 and 3 appear rather compact, although two small distortions (see also Figure 1) in object 2, seem to point in the direction of the two close companions. Unfortunately our ability to detect small-scale structure is severely restricted by poor seeing; for example, the seeing (1.4 arcsec) in the r′ band corresponds to ∼4 kpc at a cosmological redshift of z = 0.19. We see how the system is surrounded by a diffuse emission. We have delineated this emission down to isophote 26.8 mag arcsec−2. We also discern a filament (previously noted by Arp) along the line of the minor axis of object 1The Nordic Optical Telescope (NOT) is jointly operated on the island of La Palma by Denmark, Finland, Iceland, Norway, and Sweden, in Spain's Roque de los Muchachos Observatory of the Instituto de Astrof´ısica de Canarias (IAC). 2The William Herschel Telescope (WHT) is operated by the Isaac Newton Group and the IAC in Spain's Roque de los Muchachos Observatory. 3The IAC-80 telescope is located at Spain's Teide Observatory (Tenerife) and is operated by the IAC. -- 4 -- 4. It seems also that the filament decreases in brightness from object 1 -- 2 to object 4, which looks rather symmetric with a morphological shape typical of a lenticular (S0/Sa) galaxy, its spectrum presenting only absorption lines characteristic of an old population (Ca H, Ca K, MgI (λ5180), NaI(λ5892)). Using them we have estimated a redshift of 0.1239 ± 0.0005 for this galaxy. The only signs of distortion are the filament in the direction to objects 1 -- 3, and an extended emission feature to the north that seems to enclose a weak object (r′ = 21.8 mag) situated at ∼ 13 arcsec. We have determined the magnitudes of the four objects using the Sextractor software (Bertin & Arnouts 1996). The magnitudes of objects 1 -- 4 in the r′ band are respectively 19.8, 19.6, 20.2 and 17.3. As a consequence of the proximity of the three objects and the existence of diffuse emission around them, the uncertainty in the quoted magnitudes should be ∼0.1 -- 0.2 mag. The spectra of objects 1, 2 and 3 are dominated by emission lines whose identification allow us to determine their redshift unambiguously. The main features are the lines of the hydrogen Balmer series, and the lines of OII and OIII. Other minor features such as SI, NI, NII and HeI are also detected. Figure 3, shows the spectra (corrected of redshift) around the lines Hβ and Hα. Table 2 presents the equivalent widths of each of these lines. The positions of all of these lines is consistent within the uncertainty in our spectral resolution. The resulting redshifts are 0.1935 ± 0.0002, 0.1939 ± 0.0005 and 0.2229 ± 0.0002 for objects 1, 2 and 3 respectively. Tentatively we have identified some possible absoption features in the spectra of the three objects which could be associated to the host or normal underlying stellar population. Object 1 has a typical broad band line spectrum, while objects 2 and 3 have only narrow emission lines. They have strong star formation (EW(Hα + NII) = 100 and 70 mA respectively; Carter et al. 2001). We have classified the emission line objects using the diagnostics considered by Ver´on, Gon¸calves, & Ver´on-Cetty (1997) based on the line ratios of OIII(λ5007)/Hβ and NII(λ6584)/Hα. Because of the spectral proximity of the lines involved in each ratio, the effects of reddening are minimized and the unaccuracies on the determination of the continuum largely cancel out. According to these criteria, object 1 is a QSO/Seyfert 1 galaxy, while objects 2 and 3 are HII galaxies. The spectrum of the filament is very noisy and does not have any obvious feature. How- ever, using the subroutine f xcor of IRAF4 with a field bright star as template, we tentatively identify a maximum in the cross-correlation function which corresponds approximately to a redshift of 0.19. A secondary maxima appears also at z = 0.12, but we do not exclude 4IRAF is the Image Reduction and Analysis Facility, written and supported by the IRAF programming group at the National Optical Astronomy Observatories (NOAO) in Tucson, Arizona. possibly contaminnation from the main galaxy. -- 5 -- 4. Discusion As in other cases of anomalous redshift, the optical configuration of the system formed by NEQ3 (objects 1 to 3), the close lenticular galaxy (object 4) and the additional features such as the filament apparently connecting them, and the diffuse emission in the northern direction seem to be clear indications of proximity and interaction. As discussed in L´opez-Corredoira & Guti´errez (2003), examples like this, in which galaxies interact through filaments and show distortions in the halos, are relative common. An interpretation that explains the configuration as equivalent to other systems in interaction would be clearly preferred over one in which the configuration is purely a projection effect. Objects 1 and 2 have redshifts of z = 0.1935 and 0.1939 respectively and are separated by 2.8 arcsec. Considering the redshift as indicative of cosmological distance, the above numbers are translated into a difference of ∼100 ± 200 km s−1 in velocity and a separation of ∼8 kpc. Bahcall et al. (1997), in a study with the Hubble Space Telescope, found that most of the detected QSO hosts have a galaxy at less than 25 kpc. This companion tends to be an elliptical galaxy for radio-loud QSOs, and an elliptical or spiral galaxy for radio- quiet QSOs. Canalizo & Stockton (2001) have studied the environment of nine objects with intermediate colors between ultra luminous infrared galaxies and QSOs and have found close companion in some of them and morphological evidence of interaction in eight of them. In general the companion show a mixture of young and old stellar population but are not star forming galaxies. In that sample one of the system which perhaps resemble more NEQ3 is IRAS00275-2859 which shows a secondary nucleus (or a giant HII region) at ∼ 2.7 arcsec from the QSO and a tail with several clumps extending 12 arcsec. Differently from the cases studied by these authors, in NEQ3 the two objects (the QSO and the HII galaxy) at redshift ∼ 0.19 have similar magnitudes in the r band. The association of objects 1 and 2 is particularly interesting because it is one of the cases with a smaller angular separation, the galaxy is undergoing an intense burst of star formation, and there is a diffuse filament possibly associated with this pair. All of these facts seem to indicate a strong interaction between the QSO and the HII galaxy. This interaction could be the responsible of the intense star formation in object 2 and perhaps also of the QSO activity (Stockton 1982). In a conventional scenario the role of objects 3 and 4 is unclear. For instance, could the difference in redshift between objects 1 and 2, and object 3 (∼0.03) be produced by a -- 6 -- difference in peculiar velocity? The difference would be 9000 km s−1. As far as we know, interactions between galaxies with such differences in velocity have not been observed and it would be difficult to explain in the framework of models of galaxy formation. In favour of the physical relation between the pair of objects 1 -- 2 and object 3 is the main halo which seems surround the three objects uniformily and the slight elongation (see Figures 1 and 2) of object 2 through objects 1 and 3. We have discarded the possibility of gravitational lensing produced by object 4, because the combination of amplification and separation between the sources and the lens would require much more massive lensing structures such as the cores of galaxy clusters. This amplification by galaxy clusters has been used by Ellis et al. (2001) to detect high redshift galaxies. Finally, is it just a merely projection effect the proximity of object 4? Is this proximity related with the presence of the possible filament extending outwards in the direction of the three compact objects? From the galaxy counts by Metcalfe et al. (1991) and assuming Poission statistics, we have made a simple probability calculation. Based on that, Table 3 presents an estimation of the integrated number of galaxies with magnitudes up the one of objects 2, 3 and 4. We start the computation with the pair of objects at z ∼ 0.19, and compute the probability (P1) to find an object as bright as 3 (which density is N3 in a circle of ∼ 3 arcsec of radius (area A1). Assuming Poisson statistics (and onsidering that the density of each kind of objects in the area considered is very low) we have P1 = N3 × A1 ∼ 4 × 10−3. The probability to have object 4 at the end of the filament (assuming roughly an area for this filament A2 ∼ 15 × 2.5 = 37.5 arcsec−2) is P2 = N4 × A2 ∼ 5 × 10−4. Then the total probability is: P = P1 ∗ P2/2 ∼ 10−6. This low probability makes unlikely an explanation based in merely a projection effect of the four objects. 5. Summary • New observations of a system of anomalous redshift are presented. These comprises, broad band, and intermediate resolution spectroscopy of four close objects. • The redshifts of the main galaxy (0.1239 ± 0.0005) and the two previously known compact objects (0.1935 ± 0.0002 and 0.1939 ± 0.0002) have been confirmed with better accuracy. • We have delineated the diffuse emission apparently connecting the main galaxy and the three compact objects up to the isophote 26.8 mag arcsec−2 in the SDSS r′ filter. • The redshift of the third compact object has been estimated as 0.2229 ± 0.0002. -- 7 -- • We have shown that the filament probably has a redshift of z = 0.19 and seems to be mainly associated with the compact objects. • From the ratio of several emission lines, we have classified the three emission line objects as a QSO (object 1) and HII galaxies (objects 2 and 3) • The probability of the whole aggregation being fortuitous has been computed as ∼10−6. We want to thank Fernanda Art´ıguez Carro, who kindly made the observations at the IAC 80 that allow a relative calibration of the image taken in the r′ filter with the NOT telescope. -- 8 -- REFERENCES Arp, H. 1977, ApJ, 218, 70 Bahcall, J. N., Kirhakos, S., Saxe, D. H., & Schneider, D. P. 1997, ApJ, 479, 642 Bertin, E., & Arnouts, S. 1996, A&A, 117, 393 Burbidge, G. 1996, A&A, 309, 9 Carter, B. J., Fabricant, D. G., Geller, M. J., Kurtz, M. J., & McLean, B. 2001, ApJ, 559, 606 Canalizo, G., & Stockton, A. 2001, ApJ, 555, 719 Ellis, R., Santos, M. R., Kneib, J. P., & Kuijken, K. 2001, ApJ, 560, L119 L´opez-Corredoira, M., & Guti´errez, C. M. 2003, A&A (submitted) L´opez-Corredoira, M., & Guti´erez, C. M. 2002, A&A, 390, L15 Metcalfe, N., Shanks, T., Fong, R., & Jones, L. R. 1991, MNRAS, 249, 498 Smith, J. A. et al. 2002, AJ, 123, 2121 Stockton, A. 1982, ApJ, 257, 335 Sulentic, J. W. & Arp, H. 1976, first year report on NSF grant Ver´on, P., Goncalves, A. C., & Ver´on-Cetty, M.-P. 1997, A&A, 319, 52 This preprint was prepared with the AAS LATEX macros v5.2. -- 9 -- TABLES Table 1. Observations Imaging Filter u′ g ′ r′ Exp. time (s) Seeing (arcsec) 2 × 1800 1 × 900 5 × 1800 Spectroscopy 2.7 2.7 1.4 Position Exp. time (s) 1 2 3 4 × 1800 1 × 1800 1 × 1800 Slit width 3.0 1.2 1.2 -- 10 -- Table 2. Emission lines Line EW (A) Object 1 Object 2 Object 3 OII (λ3727) NeIII (λ3869) Hγ Hβ OIII (λ4959) OIII (λ5007) NI(λ5198, 5200) HeI(λ5876) OI (λ6300) NII (λ6548) Hα NII (λ6584) SII (λ6717) SII (λ6731) 26 17 6 7 41 117 1 6 2 10 4 6 7 28 11 3 14 2 8 2 2 4 8 66 28 16 13 37 8 2 5 5 52 17 10 3 -- 11 -- Table 3. Density of objects Object r (mag) N (degree−2) 2 3 4 19.6 20.2 17.3 1500 2500 200 -- 12 -- FIGURE CAPTIONS 1.A false-colour scale in the r′ band of the region around the system NEQ3 (only a ∼ 40′′ × 40′′ section has been plotted). The plot also show the position of the slits (see main text for the details). The objects studied in this article are labelled as 1 -- 4. 2. A grey-scale image in the r′ band of the three compact objects, the main galaxy and the filament apparently connecting them. The contours correspond to 25.6, 26.0 and 26.8 mag arcsec−2. North is to the right and east to the top. 3. Main spectral features (corrected of redshift) of the three emission line objects analysed in this paper. -- 13 -- Spectrograph slit positions 1 2 3 4 Objects Pos. 2 E filament (z=0.19) (z=0.1239) 4 10" Pos. 1 Increasing flux QSO 1 (z=0.1935) 2 (z=0.1939) 3 (z=0.2229) N Pos. 3 -- 14 -- -- 15 --
astro-ph/0106245
2
0106
2001-07-25T14:01:01
CMB constraints on spatial variations of the vacuum energy density
[ "astro-ph", "gr-qc", "hep-ph", "hep-th" ]
In a recent article, a simple `spherical bubble' toy model for a spatially varying vacuum energy density was introduced, and type Ia supernovae data was used to constrain it. Here we generalize the model to allow for the fact that we may not necessarily be at the centre of a region with a given set of cosmological parameters, and discuss the constraints on these models coming from Cosmic Microwave Background Radiation data. We find tight constraints on possible spatial variations of the vacuum energy density for any significant deviations from the centre of the bubble and we comment on the relevance of our results.
astro-ph
astro-ph
CMB constraints on spatial variations of the vacuum energy density P. P. Avelino1,2∗ A. Canavezes1† J. P. M. de Carvalho1,3‡ and C. J. A. P. Martins1,4,5§ 1 Centro de Astrof´ısica, Universidade do Porto, Rua das Estrelas s/n, 4150-762 Porto, Portugal 2 Dep. de F´ısica da Faculdade de Ciencias da Univ. do Porto, Rua do Campo Alegre 687, 4169-007 Porto, Portugal 3 Dep. de Matem´atica Aplicada da Faculdade de Ciencias da Univ. do Porto, Rua das Taipas 135, 4050 Porto, Portugal 4 Department of Applied Mathematics and Theoretical Physics, Centre for Mathematical Sciences, University of Cambridge Wilberforce Road, Cambridge CB3 0WA, U.K. 5 Institut d'Astrophysique de Paris, 98 bis Boulevard Arago, 75014 Paris, France (June 13th, 2001) In a recent article, a simple 'spherical bubble' toy model for a spatially varying vacuum energy density was introduced, and type Ia supernovae data was used to constrain it. Here we generalize the model to allow for the fact that we may not necessarily be at the centre of a region with a given set of cosmological parameters, and discuss the constraints on these models coming from Cosmic Microwave Background Radiation data. We find tight constraints on possible spatial variations of the vacuum energy density for any significant deviations from the centre of the bubble and we comment on the relevance of our results. I. INTRODUCTION Recent Cosmic Microwave Background (CMB) [1] and type Ia supernovae [2] data have provided some reasonably strong evidence for an accelerating local universe, implying that most of the 'missing mass' of the universe should be in a non-clustered form, such as a cosmological constant or quintessence. Yet it is well known that these measurements are mostly local, so one should be especially careful in the way they are used. In particular, it is almost always assumed that the equation of state of the dark matter is constant for low redshifts. However, this is by no means well justified, and a redshift dependence would arise if there are space and/or non-trivial time variations of the cosmological parameters. Two 'natural' ways in which this would happen are quintessence models with a non-trivial equation of state, and models where a late-time phase transition [3 -- 6] produces wall-like defects which separate regions with different values of the cosmological parameters, notably the matter and vacuum energy densities (and hence also the Hubble constant). Note that even though the motivations for the two types of models are quite different, their observational consequences can be fairly similar, and distinguishing between the two may be a non-trivial task. In a recent paper [6], three of the present authors have introduced a very simple 'toy model' that aims to mimic this type of cosmological scenario with a minimal number of free parameters. It assumes that we live inside a spherical bubble with a given set of cosmological parameters, which is surrounded by a region where these parameters are different. In terms of redshift, there will be a single discrete jump on in these parameters as one goes through the wall. A perhaps more realistic, but also definitely more complicated toy model (in the sense of having more free parameters) would be one with various different 'shells', and hence several jumps in the cosmological parameters at various redshifts. Ultimately, one could get to the continuous limit where one has the cosmological parameters varying as continuous functions of redshift in a non-trivial way. Something along these lines is discussed in [7]. In our previous work [6], we have used the recent measurements of the luminosity versus redshift relation using Type Ia supernovae, which have now been observed out to z ∼ 1.7 [8], to constrain the simplest models of this type. ∗Electronic address: pedro @ astro.up.pt †Electronic address: alex @ astro.up.pt ‡Electronic address: mauricio @ astro.up.pt §Electronic address: C.J.A.P.Martins @ damtp.cam.ac.uk 1 It was found that presently available observations are only constraining at very low redshifts z ∼< 0.5 (as was to be expected), but it was also independently confirmed that the high red-shift supernovae data does prefer a relatively large positive cosmological constant. Here we will discuss the constraints on this type of models coming from CMB data, while also allowing for a further effect. Indeed, we allow for the possibility that we are not at the centre of the inner domain or, in other words, that there is a further anisotropic component in the CMB. Obviously the observed near-isotropy of the CMB will impose quite strong constraints on our position, but it is still important to determine how much freedom is allowed if one considers that part of the observed anisotropy comes from such a displacement. In the next section we briefly review our toy model and discuss the evolution of cosmological perturbations within its framework. In Sect. III we show, mainly as an amusing interlude, how one could design a 'mimic model' of this kind which would reproduce the CMB spectrum predicted by standard paradigms such as inflation. Our main results are presented and discussed in Sect. IV, and we conclude in Sect. V. II. COSMOLOGICAL PERTURBATIONS IN A SPHERICAL BUBBLE MODEL We shall again consider a simplified model in which the universe consists of a spherically symmetric region (domain) with a given set of cosmological parameters, which is surrounded by another region where the values of the cosmological parameters (in particular, that of the vacuum energy density) are different. We shall refer to these two values of the vacuum energy density as V− and V+ respectively. Note that the cosmological parameters will in general be different in both regions, but their variations are not totally independent -- they are such as to make the universe flat both inside and outside the domain (this point is discussed in detail in [6]). As in our previous work [6], we assume that the red-shift of the domain wall, as measured by an observer at the centre of the inner spherical domain, is z∗. However, we now allow for the possibility that we are not at the centre of the domain. Specifically, we assume that we live a red-shift z∆ ≤ z∗ away from this 'central' observer. Again, we are assuming that the thin region separating the two domains considered (domain wall) does not generate relevant CMB fluctuations. This happens if the potential of the field is small enough at the origin [6]. We shall also assume that the domain walls are frozen in comoving coordinates, which again will be a good approximation provided that friction is important [9]. We shall parametrize the vacuum energy density by ΩΛ ≡ V− ρc , where ρc is the critical density, and we define ∆ΩΛ as ∆ΩΛ(r) = δρΛ(r) ρc = ρΛ(r) − V− ρc , (1) (2) where ρΛ(r) is the vacuum energy density at the point in question. Hence, this can have two possible values: 0 if we are inside the inner region, and (V+ − V−)/ρc in the outer domain. In the conformal-Newtonian gauge, the line-element for a flat Friedmann-Robertson-Walker background and scalar metric perturbations can be written as ds2 = a2(η)(cid:2)(1 + 2Φ)c2dη2 − (1 − 2Φ)(dr2 + r2dθ2 + r2 sin2 θdφ2)(cid:3) , (3) assuming that the anisotropic stresses are small. Here, Φ is the metric perturbation, c is the speed of light in vacuum, a is the scale factor, η is the conformal time, and r, θ and φ are spatial coordinates. Given that the vacuum energy becomes dominant only for recent epochs we shall be concerned with the evolution of perturbations only in the matter-dominated era, thus neglecting the contribution of the radiation component. The evolution of the scale factor a is governed by the Friedmann equation1, which can be written as H2 = Ω0 ma−1 + Ω0 Λa2 . (4) 1A dot denotes a derivative with respect to conformal time, H = a/a, the index '0' means that the quantities are to be evaluated at the present time and we have taken a0 = 1 and H0 = 1 (so that the conformal time is measured in units of H−1 0 ). 2 Note that the background matter and vacuum energy densities at an arbitrary epoch can be written as Ωm = Ω0 m Ω0 m + Ω0 Λa3 and ΩΛ = 1 − Ωm. The linear evolution equation of the scalar perturbations is given by (see for example [10]) Φ + 3H Φ + [2 H + H2]Φ = 4πGa2δp = − 3 2 a2∆Ω0 Λ, where δp is the pressure perturbation and 2 H = −Ω0 ma−1 + 2Ω0 Λa2. (5) (6) (7) Given that the source term in the outer domain (∝ a2∆Ω0 following initial conditions for eq. (6): Λ) is only important near the present time, we assume the Φ(0) = 0, Φ(0) = 0, (8) both in the inner and the outer regions. We note that since the source term is absent in the inner region, the metric perturbation is always zero there. The novel feature of this type of model is that in the presence of the scalar metric perturbations defined by eqn. (6) there is a shift in the CMB temperature given by ∆T T = −2Φ+. (9) where Φ+ is the value of Φ in the outer region at the time when the light crossed the domain wall (recall that here we are assuming that Φ = 0 in the inner region). This temperature shift is due to the relative velocity between comoving observers on either side of the domain wall. In order to compute the spectrum of CMB fluctuations generated in this model as a function of its parameters (given the values of Ω0 Λ) we need to calculate the time when the CMB radiation that reaches us today crossed the domain wall. This, of course, depends on the direction we are looking at if we're not at the centre of the inner bubble. m and Ω0 We will parametrize this direction by the angle θ = n · ncl where the unit vector n defines an arbitrary direction on the sky and ncl ≡ r/r where r is the vector that points from the observer on earth to the closest point on the domain wall. Using the Friedmann equation it is straightforward to compute the comoving distances d(z∗) and d(z∆) corresponding to the red-shifts z∗ and z∆, d(z) = cZ z 0 dz′ [Ω0 m(1 + z′)3 + Ω0 Λ] 1/2 . We find it useful to parametrize our position within the bubble by ǫ ≡ d(z∆) d(z∗) , (10) (11) that is, the ratio of our comoving distance to the bubble centre and the comoving bubble radius. On the other hand, the comoving distance to the domain wall as seen by us as a function of the angle θ is given by and the conformal time when the CMB radiation that reaches us today crossed the domain wall at that particular direction is then simply given by D(θ) = p(d(z∗) cos θ − d(z∆))2 + (d(z∗) sin θ)2, (12) η(θ) = η0 − D(θ) c . (13) The value of Φ+(θ) ≡ Φ(η) can now be calculated using eqn. (6). Before moving on to discuss the results of this analysis in Sect. IV, we make a brief detour to discuss an interesting property of this type of models. 3 Let us assume for the sake of simplicity that Ωm = 1 and ΩΛ = 0 such that eq. (6) becomes III. A MIMIC MODEL Φ + 6 η Φ = η−6 ∂(η6 Φ) ∂η = − 3 8 η2∆Ω0 Λ , (14) where we have made use of the fact that with our conventions one has a = η2/η2 equation with the boundary conditions 0 and η0 = 2H−1 0 = 2. Solving this in the outer region we obtain Φ(0) = 0, Φ(0) = 0, Φ(η) = − 1 96 η4∆Ω0 Λ = − 1 6 (1 + z)−2∆Ω0 Λ. (15) (16) Let us consider a domain wall which is at a red-shift z from us. Consider small spatial fluctuations in the red-shift of the domain wall parametrized by ∆z(θ, φ) where θ and φ are angular coordinates. These fluctuations will induce CMB perturbations with ∆T T = −2∆Φ+ = − 2 3 ∆z(1 + z)−3∆Ω0 Λ. (17) where we have subtracted the multipole of order 0 of the CMB fluctuations. This has been done for a flat matter dominated background but it is obvious to see that it can be done more generally. This demonstrates that it would be possible in general to generate any spectrum of CMB fluctuations by an appropriate design of the domain wall. It would, of course, be extremely unlikely that a domain wall would have the precise shape necessary to produce the CMB spectrum predicted in popular models for the origin of CMB and large scale structure (e.g. inflation). However, it would not be so implausible that this model could have some kind of impact on the CMB fluctuations. Irrespective of the likelihood of this model, we think it serves to highlight an important point, namely that a good test to the predictions of current models for structure formation consists in studying to what extent these could be produced by other models (even if they might seem unlikely in the light of our current theoretical prejudices). IV. RESULTS AND DISCUSSION Our results can be conveniently summarized by figs. 1 and 2. We have assumed that the present local universe Λ = 0.7, and calculated the maximum value allowed by Λ ) as a function of the parameters ǫ and z∗. Figs. is characterized by cosmological densities Ω0 currently existing observations for ∆Ω0 1 and 2 show the result of this calculation for the dipole and quadrupole measured by COBE [11]. Λ in the outer domain (∆Ωmax m = 0.3 and Ω0 As expected, there are no direct CMB constraints, to first order, if we happen to be at the centre of the domain, and they become progressively tighter as we move away from this position. We note that if the observer is right at the centre of the spherical domain, a further (second-order) contribution would come from the integrated Sachs-Wolfe effect. Being second order, this would be subdominant everywhere except extremely close to the centre (ǫ ∼ 0), and in this limit the supernovae constraints discussed in [6] are currently much stronger. For this reason we have ignored this effect in the present analysis, although it should be considered in the future when more precise data becomes available. Also as expected, there are no constraints in the limits of an infinitely small (z∗ → 0) or infinitely large (z∗ → ∞) inner region. This means that for any given ǫ there will be some intermediate value of the redshift for which the the allowed variation in the vacuum energy density is minimal. In fact it turns out that, at least to a first approximation, this particular redshift is independent of our position within the bubble (which is measured by ǫ). For the currently favoured cosmological model, with Ω0 Λ = 0.7, this redshift is found to be z ∼ 2.2. We also point out that, the quadrupole constraints (Fig 2) are, in general, much stronger than the dipole constraints (Fig 1). This is due to the much larger amplitude of the observed dipole CMB anisotropy compared with the quadrupole. m = 0.3 and Ω0 Several models attempting to reconstruct the matter distribution from the observed velocity field have been pro- posed. However, even the best ones are left with an unexplained relatively high residual velocity of ∼ 200km s−1, too large to be an unknown motion of our galaxy with respect to the Local Group reference frame (see for example 4 FIG. 1. The maximum variation in the vacuum energy density with respect to the present local value (parametrized by ∆Ωmax Λ ) allowed by the COBE dipole, as a function of our position within the spherical bubble. This position is parametrized by the redshift, z∗, of the spherical wall as measured by an observer at its centre, and by the ratio, ǫ, of our comoving distance to the bubble centre and the comoving bubble radius. FIG. 2. The maximum variation in the vacuum energy density with respect to the present local value (parametrized by ∆Ωmax Λ ) allowed by the COBE quadrupole , as a function of our position within the spherical bubble. This position is parametrized by the redshift, z∗, of the spherical wall as measured by an observer at its centre, and by the ratio, ǫ, of our comoving distance to the bubble centre and the comoving bubble radius. 5 [12,13]). The possibility that this discrepancy might be explained by a model such as the 'bubble' model presented in this paper is, however, unfavoured by our results. The relatively large values of ∆Ωmax allowed by the COBE dipole make this possibility rather unlikely. Λ In any case, our analysis confirms the intuitive expectation that the CMB constraints on anisotropies of this kind are extremely tight, or in other words that if such a model is a realistic approximation to the evolution of the universe, then we must be very close to the centre of the domain. Still another way of saying it is that, if non-trivial variations of the cosmological parameters as a function of the redshift did occur in the past, they are more likely to result from scenarios where the redshift variations are the result of time variations which maintain spatial homogeneity or isotropy (such as quintessence models, for example), since in scenarios where there are also spatial variations we are forced by current observations to occupy a fairly privileged (and arguably unlikely) position. V. CONCLUSIONS In an earlier work we have introduced a very simple 'spherical bubble' toy model for a spatially varying vacuum energy density, and type Ia supernovae data was used to constrain it. Here we have generalized this model, in order to account for the possibility that we may not be at the centre of a region with a given set of cosmological parameters. We have then discussed the constraints on this model coming from CMBR data. As expected, any significant deviations from isotropy are tightly constrained. Ultimately, our toy model aims to be a simple mimic for non-trivial variations of cosmological parameters. As we have pointed out before, these variations as one probes higher redshifts could result either from genuine time-variations (an example of which would be quintessence models with a non-trivial equation of state) or from spatial variations (an example of which would be models where late-time phase transitions produce a domain-like structure). The present work, together with [6] quantitatively confirms the intuitive expectation that models with spatial vari- ations are much more constrained, in the sense that the near-isotropy of the CMBR imposes quite strong restrictions on our position within such a domain, whereas no such constraints exist in the case of genuine time variations. Ob- viously, our toy model is rather simplified, and a more detailed analysis is required to constrain specific models. We shall return to this interesting topic in a future publication. ACKNOWLEDGMENTS C. M. is grateful for the hospitality at IAP and DARC (Observatoire de Paris-Meudon), where this work was completed. A. C. and C. M. are funded by FCT (Portugal) under "Programa PRAXIS XXI" (grant nos. PRAX- ISXXI/BPD/18826/98 and FMRH/BPD/1600/2000 respectively). We thank "Centro de Astrof´ısica da Universidade do Porto" (CAUP) for the facilities provided. [1] C.B. Netterfield et al., astro-ph/0104460 (2001); C. Pryke et al., astro-ph/0104490 (2001); R. Stompor et al., astro-ph/0105062 (2001). [2] A. G. Riess et al., Astron. J. 116, 1009 (1998); P. M. Garnavich et al., Ap. J. Lett. 493, L53 (1998); S. Perlmutter et al., Ap. J. 517, 465 (1999). [3] P. P. Avelino, J. P. M de Carvalho and C. J. A. P. Martins, Phys. Lett. B501, 257 (2001). [4] P.P. Avelino and C.J.A.P. Martins, Phys. Rev. D62 (2000), 103510 (2000). [5] P.P. Avelino, J.P.M. de Carvalho, C.J.A.P. Martins and J.C.R.E. Oliveira, astro-ph/0004227 (to appear in Phys. Lett. B). [6] P.P. Avelino, J P.M. de Carvalho and C.J.A.P. Martins, astro-ph/0103075 (to appear in Phys. Rev. D). [7] Y. Wang and P.M. Garnavich, astro-ph/0101040 (2001); M. Tegmark, astro-ph/0101354 (2001). [8] A.G. Riess et al., astro-ph/0104455 (2001). [9] C.J.A.P. Martins and E.P.S. Shellard, hep-ph/0003298 (2000). [10] V.F. Mukhanov, H.A. Feldman and R.H. Brandenberger, Phys. Rep. 215, 203 (1992). [11] C.L. Bennett et al, Ap. J. Lett. 464,1 (1996). 6 [12] J.L. Tonry et al., astro-ph/9907062 (1999). [13] M. Rowan-Robinson et al., MNRAS, in press (see also astro-ph/9912223). 7
astro-ph/0010198
1
0010
2000-10-10T18:37:41
Galaxy Modelling - II. Multi-Wavelength Faint Counts from a Semi-Analytic Model of Galaxy Formation
[ "astro-ph" ]
(Abridged) This paper predicts self-consistent faint galaxy counts from the UV to the submm wavelength range. The STARDUST spectral energy distributions described in Devriendt et al. (1999) are embedded within the explicit cosmological framework of a simple semi-analytic model of galaxy formation and evolution. We build a class of models which capture the luminosity budget of the universe through faint galaxy counts and redshift distributions in the whole wavelength range spanned by our spectra. In contrast with a rather stable behaviour in the optical and even in the far-IR, the submm counts are dramatically sensitive to variations in the cosmological parameters and changes in the star formation history. Faint submm counts are more easily accommodated within an open universe with a low value of $\Omega_0$, or a flat universe with a non-zero cosmological constant. This study illustrates the implementation of multi-wavelength spectra into a semi-analytic model. In spite of its simplicity, it already provides fair fits of the current data of faint counts, and a physically motivated way of interpolating and extrapolating these data to other wavelengths and fainter flux levels.
astro-ph
astro-ph
A&A manuscript no. (will be inserted by hand later) Your thesaurus codes are: 11(11.05.2; 13.21.1; 13.09.1; 13.18.1) ASTRONOMY AND ASTROPHYSICS Galaxy Modelling -- II. Multi -- Wavelength Faint Counts from a Semi -- Analytic Model of Galaxy Formation Julien E. G. Devriendt1,2 and Bruno Guiderdoni1 1 Institut d'Astrophysique de Paris, 98 bis Boulevard Arago, F -- 75014 Paris, France 2 Nuclear & Astrophysics Laboratory, Keble Road, Oxford OX1 3RH, United Kingdom Received ?? / Accepted ?? self -- consistent Abstract. This paper predicts faint galaxy counts from the UV to the submm wavelength range. The stardust spectral energy distributions de- scribed in Devriendt et al. (1999) (Paper I) are embedded within the explicit cosmological framework of a simple semi -- analytic model of galaxy formation and evolution. We begin with a description of the non -- dissipative and dis- sipative collapses of primordial perturbations, and plug in standard recipes for star formation, stellar evolution and feedback. We also model the absorption of starlight by dust and its re -- processing in the IR and submm. We then build a class of models which capture the luminosity bud- get of the universe through faint galaxy counts and red- shift distributions in the whole wavelength range spanned by our spectra. In contrast with a rather stable behaviour in the optical and even in the far -- IR, the submm counts are dramatically sensitive to variations in the cosmologi- cal parameters and changes in the star formation history. Faint submm counts are more easily accommodated within an open universe with a low value of Ω0, or a flat universe with a non -- zero cosmological constant. We confirm the suggestion of Guiderdoni et al. (1998) that matching the current multi -- wavelength data requires a population of heavily -- extinguished, massive galaxies with large star for- mation rates (∼ 500 M⊙ yr−1) at intermediate and high redshift (z ≥ 1.5). Such a population of objects probably is the consequence of an increase of interaction and merg- ing activity at high redshift, but a realistic quantitative description can only be obtained through more detailed modelling of such processes. This study illustrates the implementation of multi-wavelength spectra into a semi -- analytic model. In spite of its simplicity, it already pro- vides fair fits of the current data of faint counts, and a physically motivated way of interpolating and extrapolat- ing these data to other wavelengths and fainter flux levels. Key words: cosmology: structure formation -- cosmology: galaxy counts Send offprint requests to: [email protected] 1. Introduction In the dense gas clouds that harbour starbursts, the ultra- violet (UV) light of young stars is absorbed by dust grains which, in turn, release their thermal energy at infrared (IR) and submillimetre (submm) wavelengths. Thus, un- derstanding the star formation history of galaxies clearly requires a correct assessment of the UV to submm lumi- nosity budget. The most straightforward and simple obser- vational probe of such a luminosity budget is the analysis of the faint galaxy counts obtained at various wavelengths. In this purview, this paper proposes self -- consistent theo- retical predictions of faint galaxy counts at optical, IR and submm wavelengths that can be directly compared with the current host of data, and used to prepare observational strategies with forthcoming instruments. In the local universe, only 30 % of the bolometric lu- minosity is released in the IR/submm wavelength range (Soifer and Neugebauer, 1991), and the effect of extinc- tion can thus be considered as a mere correction that does not change the main evolutionary trends elaborated from optical studies. However, there is now a growing amount of evidence that this fraction was much higher in the past. Indeed, the discovery of the Cosmic Infrared Background (CIRB) at a level ten times higher than the no -- evolution predictions based on the iras local IR luminosity func- tion, and twice as high as the Cosmic Optical Background obtained from optical counts, showed that dust extinc- tion and emission play a major role in defining the lu- minosity budget of high -- redshift galaxies (Puget et al. (1996), Guiderdoni et al. (1997), Schlegel et al. (1998), Fixsen et al. (1998), Hauser et al. (1998)). Since this ma- jor breakthrough, deep surveys with the iso satellite at 15 µm (Oliver et al. (1997), Aussel et al. (1999), Elbaz et al. (1999)) and 175 µm (Kawara et al. (1998), Puget et al. (1999)), and with the SCUBA instrument at 850 µm (Smail et al. (1997), Barger et al. (1998), Hughes et al. (1998), Eales et al. (1999), Barger et al. (1999a)) have begun to break the CIRB into its brightest contribu- tors. Although identification and spectroscopic follow -- up of the submm sources are not easy, the preliminary results of such studies seem to show that part of these sources 2 J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model are the high -- redshift counterparts of the local luminous and ultraluminous IR galaxies (LIRGs and ULIRGs) dis- covered by iras (Smail et al. (1998), Lilly et al. (1999), Barger et al. (1999b)). In parallel to this pioneering explo- ration of the "optically -- dark" and "infrared -- bright" side of the universe, a more careful examination of the Canada -- France Redshift Survey (CFRS) galaxies at z ∼ 1, and Lyman Break Galaxies at z ∼ 3 and 4 do show a signifi- cant amount of extinction (Flores et al. (1999), Steidel et al. (1999), Meurer et al. (1999)). The previous estimates of the UV fluxes, and consequently of the star formation rates, in these objects have to be respectively multiplied by factors 3 and 5 to take into account the effect of extinc- tion. Dust seems to be present at still higher redshifts. For instance, it is seen in a lensed galaxy at z = 4.92 (Soifer et al., 1998) and even in a Lyman α galaxy at z = 6.68 (Chen et al., 1999). The synthetic spectra of stellar populations in galax- ies are easily computed from spectrophotometric models of galaxy evolution. Unfortunately, most of these models neglect the influence of dust on the spectral appearance of galaxies. Guiderdoni & Rocca -- Volmerange (1987) pro- posed a first modelling of the effect of dust extinction. Later, Mazzei et al. (1992) basically used the same recipe for extinction, but they also computed dust emission to get spectral energy distributions (SEDs) from the UV to the far -- IR. Complete sets of synthetic spectra are now available from the grasil (Silva et al., 1998) and star- dust models (Devriendt et al. (1999), hereafter Paper I) of spectrophotometric evolution. These models share the same spirit, but they differ by a number of details. The grasil SEDs in the IR are computed from a more sophis- ticated model of transfer, that is more explicit physically, but involves several free parameters, whereas stardust SEDs in the IR are computed with a minimal number of free parameters, by weighing various dust components to reproduce the observed iras colour -- luminosity relations. These spectra can be used in phenomenological mod- els of faint galaxy counts, that extrapolate the evolution of the local galaxies backwards under the assumption of pure luminosity evolution. For instance, the predictions of faint galaxy counts at optical wavelengths by Guiderdoni & Rocca -- Volmerange (1990) used their optical spectra with extinction, whereas Franceschini et al. (1991), (1994) used the Mazzei et al. spectra with dust extinction and emission to produce the first set of counts at optical and FIR wave- lengths. However, semi -- analytic models of galaxy forma- tion (hereafter SAMs) are a much more powerful approach to describe the physical processes that rule galaxy forma- tion and evolution within an explicit cosmological con- text (White & Frenk (1991), Lacey & Silk (1991), Kauff- mann et al. (1993), Cole et al. (1994), Somerville & Pri- mack (2000)). White & Frenk (1991), Lacey et al. (1993), Kauffmann et al. (1994), and Cole et al. (1994) proposed predictions of faint galaxy counts at optical wavelengths (basically the B and K bands) from their models. How- ever, predictions at IR/submm wavelengths from a SAM were produced much later by Guiderdoni et al. ((1997), (1998), hereafter GHBM). But this first study did not give the corresponding predictions at optical wavelengths, and was restricted to the Ω0 = 1 standard Cold Dark Matter (CDM) model. In this paper, we implement the stardust spectra into a SAM to make predictions of faint galaxy counts and redshift distributions from the UV to the submm wavelength range, very much in the spirit of GHBM. We also extend the SAM to other cosmologies, and study the sensitivity of the results to the cosmological parameters and star formation history. Although our approach has a number of shortcomings which are due to the simplicity of our model, this paper primarily intends to show that (i) the implementation of stardust SEDs into SAMs is straightforward because of its small number of free pa- rameters; (ii) the model gives fits that are already very satisfactory in spite of the simplicity of the approach; and (iii) the outputs are a physically motivated tool to interpo- late or extrapolate the current observations of faint counts to other wavelengths and/or flux levels. This is needed to prepare the observational strategies with the forthcoming IR/submm satellites sirtf, first and planck, as well as with the Atacama Large Millimetre Array. In section 2, we briefly describe how we connect the stardust spectra with the various physical processes that are relevant to galaxy formation, within our SAM. We point out the differences with GHBM. Section 3 discusses the values of the free parameters that define the so -- called "quiescent" mode of star formation. Section 4 focuses on the sensitivity of the faint galaxy counts to a change in the cosmological parameters. Section 5 studies the sensi- tivity of the faint galaxy counts to galaxy evolution, and more specifically to the presence of a heavily -- extinguished "starburst" mode of star formation similar to the one in ULIRGs. Finally, a fiducial model is proposed. We discuss our results in Section 6. 2. The basics of our semi -- analytic model In the SAM ab initio approach, galaxies form from Gaus- sian random density fluctuations in the primordial matter distribution, dominated by CDM. Bound perturbations grow along with the expanding universe, until gravitation makes them turn around and (non -- dissipatively) collapse. As a result, they end up as virialized halos. Then the collisionally -- shocked baryonic gas cools down radiatively, and settles at the bottoms of the potential wells where it is rotationally -- supported. Stars form from the cold gas and evolve. At the end of their lifetimes, they inject energy, gas and heavy elements back into the interstellar medium. The chemical evolution is computed, and the recipes developed in GHBM and in Paper I give the amount of optical lu- minosity that is absorbed by dust and thermally released J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model 3 at IR/submm wavelengths. Finally, overall SEDs from the UV to the submm are computed with stardust. We refer the reader to GHBM for a detailed descrip- tion of how to compute the mass distribution of collapsed dark matter halos from the peaks formalism introduced by Bardeen et al. (1986), and Lacey and Silk (1991) in an Einstein -- de Sitter universe. We give in appendix A the quantities which enable us to extend this formalism to low matter -- density universes with or without a cosmological constant. As we follow closely the prescriptions in GHBM, we only mention in the following subsections the quanti- ties which differ from their work. 2.1. Gas We assume that a universal "baryonic fraction" ΩB/Ω0 of the pristine gas gets locked up within each dark matter halo, where it is collisionally ionised by the shocks occur- ring during virialization. Because it can cool radiatively, gas then sinks into the potential wells of the halos. The cooling time depends on the gas metallicity. Here, we de- cide to adopt the cooling function given by Sutherland and Dopita (1993) for one third of solar metallicity. This choice is motivated by the fact that it is the average value that is observed today in clusters, and probably, as argued by Renzini (1999), the average value of the low -- redshift universe as a whole. Under this assumption, the cooling time is underestimated in high -- redshift halos where the gas is more metal -- poor. However, these objects are also smaller and denser on an average, so that their cooling times are already very short. We assume that the gas stops falling into the dark matter potential wells when it reaches rotational equilib- rium, and forms rotating thin disks (see e.g. Dalcanton et al. (1997) and Mo et al. (1998)). Following these authors, we adopt for the thin disk an exponential surface density profile with scale length rd, and truncation radius rt, such as: Σ(r) =(cid:26) Σ(0) exp(− r 0 rd ) if r ≤ rt , if r > rt where rt is defined as the minimum value between the virial radius rvir, and fcrd. The free parameter fc defines the extent of the cold gas disk. We then relate the exponential scale length of the cold gas disk rd to the initial radius rvir, through conservation of specific angular momentum (Fall and Efstathiou, 1980). As shown by Mo et al. (1998), stability criteria yield: rd = 1 √2 λrvir ×(cid:18)1 − f 2 c exp(−fc) 2 [1 − (1 + fc) exp(−fc)](cid:19)−1 , (1) where λ ≡ JE1/2G−1M −5/2 ≃ 0.05 ± 0.03 is the well -- know dimensionless spin parameter. As this will be important later, we emphasise that the simple formalism used here does not allow us to form spheroids through mergers/interactions of galaxies. There- fore, in section 5, we will define a "starburst mode" which phenomenologically accounts for this process. 2.2. Stars The only time scale available in our gas disks is the dy- namical time scale tdyn ≡ 2πrd/Vc. Therefore, guided by observational data (Kennicutt (1998)), we assume that the complicated physical processes ruling star formation lead, at least in a disk galaxy, to a global star formation rate (SFR) with the simple law: SFR(t) = Mgas(t) βtdyn . (2) where Mgas(t) is the total mass of cold gas in the disk at time t. We introduce an efficiency factor β−1 as a sec- ond free parameter. The IMF is chosen to be Salpeter's, with slope x = 1.35 between masses md = 0.1 and mu = 120M⊙. The stardust spectrophotometric and chemical evo- lution model presented in Paper I is then used to compute metal enrichment of the gas as well as the UV to NIR spectra of the stellar populations produced with such star formation rates. Details on the stellar spectra, evolution- ary tracks and yields can be retrieved from this paper and references therein. 2.3. Feedback Along with producing metals, massive stars which, at the end of their lifetimes, explode in galaxies, eject hot gas and heavy elements into the interstellar and/or intergalactic medium. We focus here on the modelling of this "stellar feedback", which is inspired from Dekel and Silk (1986). The average binding energy of a mass of gas Mgas(t) distributed within a truncated exponential disk at time t, which is gravitationally -- dominated by its dark matter halo, is given by: c + V 2 Mgas(t)(cid:2)V 2 esc(rt)(cid:3) ,(3) 1 πr2 t Z rt 0 2πrMgas(r, t)Φ(r)dr ≃ 1 2 where Φ(r) is the gravitational potential of a singular isothermal sphere truncated at virial radius rvir, and: Vesc(r) = Vc(cid:20)2 (cid:18)1 − ln r rvir(cid:19)(cid:21)1/2 is its escape velocity at radius r. (4) As a result, the energy balance between the gravita- tional binding energy and the kinetic energy pumped by supernovae into the interstellar medium yields the frac- tion of stars F⋆ that formed before the triggering of the galactic wind (at time tW ): F⋆ = M⋆(tW ) M⋆(tW ) + Mgas(tW ) = 1 1 + (Vhot/Vc)2 , (5) 4 with: J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model Vhot ≡(cid:20) 2ηSN ESN ǫSN 1 + (Vesc(rt)/Vc)2(cid:21)1/2 , (6) where the energy available per supernova is ESN = 1051erg, and the number of supernovae per mass unit of the stars that just formed is ηSN = 7.5 × 10−3 for a Salpeter IMF. The SN heating efficiency ǫSN is a third free parameter. By taking the initial gas mass available for star formation to be the initial cold gas mass minus the gas mass lost in the galactic wind, one then approximate chemical evolution by the closed -- box model described in Paper I. Note that we have neglected any dynamical effect due to mass loss in the previous analysis. 2.4. Dust Part of the luminosity released by stars is absorbed by dust and re -- emitted in the IR/submm range. We now briefly outline how we compute the luminosity budget of our ob- jects within the SAM. We emphasise that this is a major improvement with respect to GHBM, as for the first time, stellar and dust emission are linked self -- consistently. As in Paper I, we proceed to derive the IR/submm dust spec- tra with three steps: (i) computation of the optical depth of the disks, (ii) computation of the amount of bolomet- ric energy absorbed by dust, and (iii) computation of the spectral energy distribution of dust emission. The first step is easily completed because we know the sizes of our objects from eq. 1 and the definition of the truncation radius rt, and we obtain the mass of gas Mgas(t) and metallicity Zg(t) as a function of time through our model of chemical evolution. We then use the scaling of the extinction curve with gas column den- sity and metallicity described in Guiderdoni & Rocca- Volmerange (1987) to compute the face -- on optical depth of our objects at any wavelength: τ z λ(t) =(cid:18) Aλ AV(cid:19)Z⊙(cid:18) Zg(t) Z⊙ (cid:19)s(cid:18) hNH(t)i 2.1 × 1021 at cm−2(cid:19) , (7) where the mean H column density (accounting for the presence of helium) reads: hNH(t)i = Mgas(t) 1.4µmpπr2 t . (8) The second step is more delicate because it involves choosing a "realistic" geometry distribution for the rela- tive distribution of stars and dust. We model galaxies as oblate ellipsoids where dust and stars are homogeneously mixed, and scattering is taken into account. As explained in Paper I, the model gives a decent fit of the sample of local spirals analysed by Andreani & Franceschini (1996). Finally, the third step involves an explicit modelling of the dust grain properties and sizes. We use the three -- component model described in D´esert et al. (1990) for the Fig. 1. Influence of the different cosmologies on the UV/near -- IR faint counts. Dots stand for ΛCDM, dashes for OCDM, and solid lines for SCDM. Data are from Hogg et al. (1997) (U band), Williams et al. (1996) (F300WAB, B & I bands), Arnouts et al. (1997) (B band), Bertin & Dennefeld (1997) (B band), Gardner et al. (1996) (B, I & K bands), Metcalfe et al. (1995) (B band), Weir et al. (1995) (B band), Smail et al. (1995) (I band), Le F`evre et al. (1995) (I band), Moustakas et al. (1997) (K band), and Djogorvski et al. (1995) (K band). Milky Way with polycyclic aromatic hydrocarbons, very small grains and big grains, and we allow a fraction of the big grain population to be in thermal equilibrium at a warmer temperature if our galaxies undergo a massive starburst. The weights of these four components are fixed in order to reproduce the relations of IR/submm colours with bolometric IR luminosity LIR that are observed lo- cally, as detailed in Paper I. Once the full (UV/submm) spectral energy distributions of individual objects are computed following such a method, we build populations of galaxies for which we derive galaxy counts and redshift distributions. We present these results in the following sec- tions. 3. The free parameters In addition to the cosmological parameters h, Ω0, λ0, ΩB and σ8h−1 , and to the choice of the IMF, which is assumed to be constant throughout a Hubble time, we basically have three astrophysical free parameters in the current version of our simple semi -- analytic model: the star for- mation efficiency β−1, the SN heating efficiency ǫSN , and J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model 5 ciency of SN explosions in a disk galaxy, because SN bub- bles can blow their energy out of the disk without altering the cold gas (see e.g. De Young & Heckman (1994), and Lobo & Guiderdoni (1999) for an examination of the is- sue within a SAM). Consequently, ǫSN could be very low. We adopt ǫSN = 0.03 in the following. Increasing ǫSN de- creases the normalisation and slope of the optical and IR counts, since star formation is quenched in galaxies with still higher masses, that form at still lower redshifts. Third, the average value of the disk truncation pa- rameter fc, which measures the gaseous disk extension, is around 6, from the sample of spiral galaxies with various morphological types observed by Bosma (1981), and used by GHMB. However, this number is probably uncertain by about a factor 2, and it can be adjusted within this range in order to match the UV/optical/near -- IR counts as well as iras counts, as far as this parameter fixes the amount of dust absorption in a disk galaxy. Increasing fc increases the normalisation of the optical counts and decreases that of the IR counts, since extinction decreases. As explained in Paper I, the set of stardust spectra depends on (i) the mass of baryons in the galaxy, (ii) the star formation timescale t∗, (iii) the age t of the stellar population, and (iv) the parameter called fH that links the gas mass fraction to the gas surface density, and is used in the computation of the face -- on optical depth. In the SAMs, these quantities are computed directly from the cold gas mass Mbar, the dynamical time tdyn, the collapse redshift zcoll and observed redshift z, and the disk expo- nential length rd, provided the values of the cosmological parameters are chosen, and the astrophysical parameters β, ǫSN and fc are fixed. Thus the implementation of the stardust spectral energy distributions within our SAM is very straightforward and does not bring new free pa- rameters. The above -- mentioned values of the free parameters de- fine the so -- called "quiescent mode" of star formation, sim- ilar to what is observed in local disks. As a reference point, we list, for the ΛCDM cosmology, the properties of a disk galaxy hosted by a halo of about 7.5 × 1011M⊙, which collapses at a redshift ∼ 3 (meaning that the age of the "Milky Way" -- class spiral galaxy that sits in this halo is about 11.4 Gyr) with a spin parameter ≃ 0.08. At red- shift 0, such a galaxy has turned about 88 % of its total 7.5 ×1010M⊙ of cold gas into stars. Its disk exponential scale length is about 3.5 kpc, yielding a gaseous disk ex- tending to 21 kpc with an average hydrogen column of 8 × 1020at cm−2 and a metallicity of 0.02. This, in turn implies a face on optical depth in the B band of 0.7 re- sulting in values of MB = −19.8 and LIR = 2 × 1010L⊙ for the face -- on absolute B magnitude and the bolometric IR luminosity (between 3 and 1000 µm) respectively. Although this discussion is only valid, strictly speak- ing, for a given cosmology, it is unlikely that different cosmological parameters will significantly affect physical parameters like star formation or feedback efficiency. We Fig. 2. Influence of the different cosmologies on the mid -- IR/FIR/submm counts. Data are from Elbaz et al. (1999) (15 µm), Kawara et al. (1998) and Puget et al. (1999) (175 µm), Smail et al. (1997), Eales et al. (1999) and Barger et al. (1999a) (850µm). Coding for the lines is the same as in Fig 1. the disk truncation parameter fc which is used to compute the gas column density and the face -- on optical depth. As a matter of fact, there is not much freedom in the choice of these parameters. First, the value of the star formation efficiency β−1 de- duced from Kennicutt's data (Kennicutt, 1998) is about β ≃ 50 for our definition, and is valid for galaxies ranging from quiescent objects to very active starbursts. We refer to GHBM for a discussion of how this prescription actually compares to the data, especially to the so -- called "Roberts times" in nearby disks, and just mention that the differ- ence between the value in this paper and the one used in GHBM (β = 100) stems from the different prescriptions used to compute disk sizes, which result in our disks being about twice as large as theirs. As mentioned by Kennicutt, there is a lot of scatter in the data (± 30 -- 50 %), which, along with plausible systematics in the calibration of the different star formation estimators, should make the value of β uncertain by at least 20 %. Increasing β decreases the normalisation and slope of the optical and IR counts, because star formation is lower, and takes place at lower redshifts. Second, recent numerical simulations (Thornton et al., 1998) suggest that the SN heating efficiency ǫSN is ≃ 0.09. However, there is much uncertainty on the actual effi- 6 J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model therefore consider our astrophysical parameters as inde- pendent of the cosmological model. In the next section, we keep the same values of the astrophysical parameters, and we study the predictions of the SAM for the sets of cosmological parameters that are displayed in table 1. Cosmological Mod- els Ω0 λ0 h ΩB σ8h−1 SCDM ΛCDM OCDM 1.0 0.3 0.3 0.0 0.7 0.0 0.5 0.7 0.7 0.015 h−2 0.58 0.015 h−2 1.0 0.015 h−2 1.0 Table 1. Parameters of the different cosmologies. relatively "stable" behaviour extends to the FIR range. In sharp contrast, the differences between the predictions of the various cosmological models are spectacular in the submm range : at 850 µm and for the flux level of 10 mJy, the model predicts ∼ 100 times more sources (or ∼ 10 times brighter sources) in the OCDM cosmology than in the SCDM. An interesting trend also comes out of these figures: with the quiescent model, any low matter -- density universe does a better job at matching the ISOPHOT counts at 175 µm and the SCUBA counts at 850 µm than a critical one. Our OCDM is even able to fit the submm counts "naturally", without any additional ingredient. The influence of cosmology on the faint counts is pro- duced by the complicated combination of several effects. For instance, for lower values of the density parameter Ω0, either with zero cosmological constant, or with zero curvature, we have the following changes: 1. For halos of a given mass, the collapse occurs earlier on an average (see Fig A.1); 2. The halo number density is lower (see Fig A.1); 3. With a fixed value of ΩB, the baryon fraction is higher, so that there is, on an average, more fuel for star for- mation per halo of a given mass; 4. Volume elements are larger; 5. Luminosity distances are larger; 6. The time versus redshift relationship changes, and the amount of evolution undergone by the sources at any redshift with respect to z = 0 is larger. These six effects act in different ways. If they are taken separately, points 3, 4, and 6 increase the slope of the counts whereas points 1, 2, and 5 decrease the slope. In addition to this, there is the effect of the k-correction. In the optical, and NIR, the k -- correction is positive, and can- cels out partly what is occurring at high redshift in such a way that the faint counts are weakly sensitive to cos- mology. At optical wavelengths, the net effect is that faint counts in low matter -- density universes are below those in the SCDM. This conclusion is opposite to the predic- tions of phenomenological models based on backward evo- lution of the local luminosity function, under the assump- tion of monolithic collapse and pure luminosity evolution (see e.g. (Guiderdoni and Rocca-Volmerange, 1990)). The latter models predict that the OCDM is over the SCDM. The origin of this discrepancy is that the phenomenologi- cal models do not take points 1, 2 and 3 into account. Fig. 3. Influence of the different cosmologies on multi -- wavelength redshift distributions of galaxies. Coding for the lines is the same as in Fig 1. Data are from Crampton et al. (1995) (Canada -- France Redshift Survey), and Ashby et al. (1996) (North Ecliptic Pole region). The predicted curves in the I band and at 60 µm have been renormalised to the total number of galaxies in the data. 4. Sensitivity of faint counts to cosmological parameters We emphasise that our purpose here is not to determine values of h, Ω0 or λ0, but rather to answer the following question: what is the net effect of changing the cosmo- logical parameters on faint galaxy counts from the UV to the submm. We explore the set of cosmological parame- ters displayed in table 1. All the astrophysical parameters are fixed at the "natural values" of the "quiescent" mode of star formation based on the local universe. On Figs. 1 and 2, we show multi -- wavelength counts obtained for the SCDM, ΛCDM, and OCDM cosmologies defined in table 1. From these figures, one clearly sees that, in agreement with Heyl et al. (1995) and Somerville & Primack (2000), the UV to NIR counts are relatively in- sensitive to changes in the cosmological parameters. This J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model 7 This weak sensitivity extends to the mid -- IR and FIR, but the behaviour of the submm counts is very different. In contrast with the optical, NIR, mid -- IR and FIR, the k -- correction is negative at submm wavelengths (see Fig 17 of Paper I for an illustration), and enhances the effects of cosmology and evolution at high redshift. The dominant effects are the earlier collapse (see Fig A.1) and larger volume available at high redshift in low matter -- density universes. Such an effect is particularly marked on the redshift distribution of the sources given in fig 3. These predic- tions are compared with the faintest redshift survey in the I band (the CFRS, Lilly et al. (1995), Crampton et al. (1995)), with IAB < 22.5, and the North Ecliptic Pole Region (NEPR) survey from iras at 60 µm, with 60 ≤ S60 ≤ 150 mJy (Ashby et al., 1996). The CFRS sur- vey is correctly reproduced by the quiescent model, what- ever the cosmology, though all quiescent models seem to overpredict low -- luminosity objects and produce a peak at too low a redshift with respect to the data (∼ 0.3 in- stead of ∼ 0.5). This is clearly due to the overproduc- tion of low -- luminosity objects in the luminosity function. There is not much sensitivity to cosmology. At 60 µm, the various cosmologies predict different redshift distribu- tions. The SDCM peaks at low redshift, without any high -- redshift tail, as anticipated by GHBM. The ΛCDM and the OCDM peak at higher redshift, with broader distri- butions. The OCDM already seems to overpredict high -- z galaxies in the NEPR at 60 µm. Finally, the sensitivity of the redshift distributions to cosmology is spectacular in the submm range. The wavelengths 175 µm and 850 µm and the flux cuts at 100 and 2 mJy respectively cor- respond to on -- going redshift surveys of the ISOPHOT and SCUBA sources. There are no firm results for these surveys because of identification problems (Downes et al. (1999), Smail et al. (1999)), and we prefer not to plot data. However, our quiescent models do not contain merg- ers and therefore do not account for the massive ULIRGs seen by ISOPHOT and SCUBA. This has to be taken into account in the model, though a low matter -- density universe lessens noticeably the importance of the contri- bution of ULIRGs to the cosmic FIR luminosity. We now try to assess the impact of such mechanisms on our results phenomenologically. 5. Sensitivity of faint counts to the star formation history Our simple SAM is not able to compute either the merg- ing history of halos, or of the galaxies they host. However, we know that locally there is a tight correlation between major mergers on one side, LIRGs and ULIRGs on the other: at least 95 % of them are currently undergoing ma- jor mergers (see for instance Sanders & Mirabel (1996)). It also seems fairly safe to assume that ISOPHOT and SCUBA sources are the high-redshift counterparts of such mergers. As a matter of fact, one could sum up the qual- itative information from currently available datasets as follows. First, the objects seen by SCUBA have to be either very massive, or very efficient to extract energy from the gas, simply because their bolometric luminos- ity is larger than 1012L⊙. Second, they have to be highly extinguished because most of this luminosity is emitted in the IR/submm. Third, for such numerous bright sources not to have been detected in the IRAS NEPR redshift survey at 60 µm, they have to be located in majority at redshifts greater than about ∼ 1.5, which seems to be the case for some of the SCUBA sources (Barger et al., 1999b). Fig. 4. IR counts for the SCDM "quiescent" model (solid lines), the "burst" model (dots), and a still more efficient model to extract luminosity from the gas (dashes). The luminosities (resp. number densities) of ULIRGs are mul- tiplied (resp. divided) by a factor 2 in the efficient model as compared to the "burst" model. In light of these observational facts, and as in GBHM, we define an ad-hoc "starburst" model, simply by push- ing the limits of our quiescent models (SCDM, OCDM, or ΛCDM), still powering the sources with star formation according to a Salpeter IMF. This consists merely in trans- forming a fraction of high -- redshift quiescent objects into ULIRGs, while keeping all the parameters of the model fixed. The obvious interest of such an exercise is to as- sess whether one is able to reproduce the SCUBA source counts, along with preserving the quality of the fits of the optical counts used to calibrate the quiescent model, in the various cosmologies. We hereafter focus on the SCDM 8 J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model cosmology, for which the "quiescent" mode of star forma- tion is unable to reproduce the submm counts. rate averaged over this period is ∼ 540 M⊙ yr−1. The starburst galaxy has a typical column density of about 7× 1022at cm−2, and a metallicity of 0.03, yielding a face -- on optical depth in the B band of 128. Its absolute B mag- nitude and bolometric IR luminosity (between 3 and 1000 µm) reach MB = −19.58 and 2 × 1012L⊙ respectively. Fig. 5. Multi -- wavelength redshift distributions for the SCDM "quiescent" model (solid lines), the "burst" model (dots), and a still more efficient model to extract luminos- ity from the gas (dashes), as in Fig 4. In order to build such an ad-hoc model, we use the reasonable recipe that follows: 1. A fraction of objects with halo masses larger than 1012M⊙ goes through a ULIRG phase when their host halos collapse; their SFRs and optical depths are typi- cally two orders of magnitude higher than those of the z = 0 Milky Way (e.g. Rigopoulou et al. (1996)). We tune our β and fc parameters to obtain such proper- ties for the heavily -- extinguished burst mode of star formation. Typically, we take β = 0.5 and fc = 0.5 for the starbursts. 2. These ULIRGs are mainly located at z > 1.5, which we enforce by requiring that their fraction evolves pro- portionally to the squared density, i.e. as (1 + zcoll)6. 3. Their number density at redshift 0 is consistent with the iras luminosity function of (Soifer and Neuge- bauer, 1991). As a result of this phenomenological recipe, a typi- cal halo of mass 1012M⊙, with reduced spin parameter λ ≃ 0.04, that collapses at redshift ≃ 3, hosts by red- shift ≃ 2.8 (180 Myr after the starburst was triggered) a ULIRG of size 1 kpc that has consumed 98 % of its 1011M⊙ of cold gas initially present. The star formation Fig. 6. UV/near-IR counts for the fiducial model (solid line). Of course, such a model is quite drastic, but once again, it should be considered as the necessary extension of the quiescent models to produce the correct amount of FIR/submm luminosity. The interesting result is that such a SCDM model in which all massive objects that form at redshifts higher than 1.5 are ULIRGs produces almost enough IR/submm luminosity to match the ISOPHOT and SCUBA counts, as can be seen in Fig 4. This is also the typical luminosity one can extract from star forma- tion with a Salpeter IMF without ruining the UV/IR cal- ibration of the counts. For instance, decreasing the mass above which the ULIRG phenomenon occurs by an order of magnitude strongly decreases the optical counts. We also have to examine the possibility that a more efficient mechanism powers these sources, for instance a top -- heavy IMF, with all the energy available through stel- lar nucleosynthesis being reprocessed in the IR/submm. The main features of such a model have been discussed in GHBM who take this solution to accommodate submm counts easily in an SCDM cosmology. We refer the reader to that paper for details. To test this possibility, we simply take our burst model and multiply the luminosity output J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model 9 Fig. 7. Mid -- IR/submm counts for the fiducial model (solid line). Fig. 8. Multi -- wavelength redshift distributions for the fiducial model (solid line). of each ULIRG in the infrared per unit mass, LIR/M , by a factor 2, while lowering the number of ULIRGs in the model by 2. This is to say, we trade the number of sources for more luminosity per source. Fig 4 and Fig 5 show that the influence of such a redistribution on the counts is weak. Of course, any combination of luminosity and number density of ULIRGs is possible. In light of the previous work, and bearing in mind that we want to describe multi -- wavelength galaxy counts, we can define a "best guess" model within a given cosmologi- cal model. We hereafter retain the ΛCDM model as a typ- ical example, since the optical and submm counts with the ΛCDM model and the quiescent mode of star formation only are intermediate between the SCDM and OCDM. We take h = 0.7, Ω0 = 0.3, λ0 = 0.7, ΩB = 0.015h−2, and σ8h−1 = 1. In terms of our astrophysical parameters, we keep the standard value ǫSN = 0.03, and we take β = 50, fc = 6 for the quiescent galaxies, and β = 0.5, fc = 0.5 for the starbursts. One ULIRG dwells in each halo that is more massive than 1012M⊙ and collapses before red- shift 1.5. This ULIRG population evolves as the density squared at lower z, so that at redshift 0, its number density is about 10−7 Mpc−3, corresponding to only one ULIRG for 2500 halos ≥ 1012M⊙. The predictions for the faint counts are given in Figs 6, and 7. The model provides a good fit of the faint counts at optical wavelengths (though the bright counts are slightly overestimated). The quality of the fit nicely compares with other faint counts obtained from SAMs (e.g Kauffmann et al. (1994)). The ISOCAM 15 µm data and iras 60 µm data are also fairly repro- duced, though the observed slope of the 15 µm counts seem to be slightly steeper than the model. The fit of the submm counts is also very satisfactory. The redshift distributions are given in Fig 8. The CFRS predictions now peak almost at the correct red- shift. The NEPR predictions still exhibit a high -- redshift tail as in GHBM, in contrast with the data, but the level is much lower than in GHBM. We recall that the NEPR sample is polluted by a supercluster in the first redshift bin. Moreover, a recent follow -- up of this sample with ISO- CAM at 15 µm seems to show that some of the sources are multiple and that the optical identifications might be ambiguous in these cases (Aussel et al., 2000). The rel- ative levels of the two peaks in the redshift distribution at 175 µm are sensitive to the flux cut -- off. Most of the sources in the redshift distribution for the SCUBA deep surveys at 850 µm are predicted to be at z > 1, but the comparison with data is still difficult because of identifica- tion uncertainties (see e.g. Barger et al. (1999b) corrected after Smail et al. (1999)). Finally, Fig 9 shows the Cosmic Background obtained by integrating the faint counts, and compare the predic- tions with current data in the optical, IR and submm. Whereas introducing ULIRGs in an ad -- hoc way into our simple models suffices to reproduce the Cosmic IR Background and the submm counts at 850 µm, it falls marginally short of getting the required diffuse back- ground flux at 140 and 240 µm, though it reproduces the 10 J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model We have not explored a large set of statistical prop- erties in this paper, but have only illustrated the ability of this approach to reproduce the optical/IR/submm lu- minosity budget by producing predictions of faint galaxy counts at UV, visible, NIR, mid -- IR, FIR, and submm wavelengths. As in GHBM, we have defined a quiescent mode of star formation which corresponds to the "natu- ral" values of these astrophysical parameters as they are suggested by local observations of disks (for β and fc), or by numerical simulations (for ǫSN ). We have then studied the influence of the cosmological parameters Ω0 and Λ on the faint counts. In SAMs, the cosmological parameters influence the counts at various stages of the computation, through halo collapse as well as through the relationship of the cosmic times, luminosity distances and volume elements to redshifts. These quanti- ties intervene in the computation of the counts in different ways, and their effect is dimmed or enhanced by the k -- corrections. The net result on the optical and NIR counts, with the influence of a positive k -- correction, is a rather weak sensitivity to the values of the cosmological param- eters. In contrast, because of the negative k -- correction that enhances what happens at high redshift, the submm counts show a strong sensitivity to cosmology. We know that some of the SCUBA sources detected at 850 µm are the high -- redshift counterparts of local LIRGs and ULIRGs. These objects harbour heavily -- extinguished starbursts due to gas inflows triggered by in- teraction/merging. Our SAM is unable to address this pro- cess, and, as in GHBM, we chose to implement a heavily -- extinguished starburst mode of star formation, by increas- ing the number fraction of massive objects that undergo this stage, as the squared density. Of course, the LIRGs and ULIRGs can also be powered by a top -- heavy IMF. We have shown the strong sensitivity of the submm counts to the details of the ULIRG scenario, because of the negative k -- correction, contrasting it with the weak influence on the optical counts. We have also produced redshift distributions at vari- ous wavelengths and flux cuts. Here again, the sensitivity of the results to cosmology and evolution is dramatic at submm wavelengths. We got a fair fit of the CFRS red- shift distribution in the I band, and the redshift distri- bution predicted for the NEPR survey looks much bet- ter than in GHBM. At submm wavelengths, we predict a double -- peaked distribution with nearby (mostly qui- escent) sources and distant (mostly starburst) sources. The relative weight of the two broad peaks is sensitive to cosmology and evolution. Sufficient statistics in an ob- servational sample could in principle help disentangling these effects. Unfortunately, the identification process of the submm sources is difficult, either in ISOPHOT or in SCUBA samples, and we might have to wait for multi -- wavelength observations with forthcoming satellites such as sirtf and first to solve this issue. In this context, our predicted counts are also a useful ingredient to anal- Fig. 9. Diffuse Background light for the fiducial model (solid line). ISOPHOT counts brighter than 100 mJy at 175 µm. These galaxies contribute only 10 % of the background. So this discrepancy may be due only to the fact that the 175 µm counts below 100 mJy are much steeper than our predic- tions. The model is too low by a factor of 2 with respect to the points corrected for warm galactic dust by Lagache et al. (1999), which are themselves a factor of 1.5 below the points without such a correction by Hauser et al. (1998). The difficulty to fit the points might indicate that this correction is still underestimated. Finally, one should also be aware that a contribution of intergalactic dust (with a grey extinction curve) to the background light is also possible (Aguirre and Haiman, 1999). Adding these extra components might help reconcile models and observations. We conclude from these figures that this fiducial model gives a satisfactory estimate of the luminosity budget of galaxies, and allows us to interpolate or extrapolate the observed faint counts to other wavelengths and fainter flux levels. 6. Discussion and conclusions In this paper, we have proposed a first implementation of the set of stardust synthetic spectra into a SAM of galaxy formation. Although our model is quite simple, and cannot properly handle the merging history trees of halos and galaxies, we have shown that the implementation of stardust is quite straightforward. This implementation can be easily achieved in more sophisticated SAMs. J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model 11 yse current data, and to prepare observational strategies with these satellites. In this purview, we have proposed a new "fiducial model" in a ΛCDM cosmology which can take the place of the model "E" in GHBM, and which is available upon request. Merging -- triggered starbursts and subsequent bulge formation are key -- processes in the paradigm of hierar- chical galaxy formation. It turns out that these pro- cesses are particularly apparent at IR/submm wave- lengths, and almost invisible at optical wavelengths. So multi -- wavelength observations are required to constrain the history of galaxy formation in the quiescent and star- burst modes. The complete merging history of galaxies has to be followed in detail, especially during the short periods of interaction and merging which produce IR lu- minous sources. A hybrid method of galaxy formation us- ing high -- resolution N -- body simulations to plant semi -- analytic galaxies is being developed and will be used to quantify the importance of merging processes (Hatton et al. (2000) and following papers). In this context, the reader is invited to notice the following point: any change in the luminosities and number densities of the ULIRGs results into spectacular changes in the submm counts, because of their super -- Euclidean regime. This sort of "instabili- ties" in the behaviour of the faint counts is going to be very useful to constrain the luminosity and duration of the starbursts in more refined models. Throughout this paper, we have assumed that dust heating is powered by star formation. The other possible engine is the presence of heavily -- extinguished AGNs lo- cated at the centers of ULIRGs. In local samples, heating is dominated by starbursts, except in the most luminous galaxies (Genzel et al. (1998), Lutz et al. (1998)). So far, we do not know how this situation evolves with redshift. The redshift distribution of the SCUBA sources seems to peak around redshift 2.5 where the observed redshift dis- tribution of quasar activity may also peak (Pei, 1995). Per- haps a sophisticated combination of AGNs and starbursts with a top -- heavy IMF is responsible for the evolution seen by SCUBA, and we signal that attempts to include both types of ingredients are already discussed in the literature (e.g. Blain et al. (1999)). However, properly disentangling all these components definitely requires a more sophis- ticated model than the one presented here. We therefore defer this issue to a later study, but remark that AGNs can be implemented self -- consistently within the framework of the hybrid method (Kauffmann and Haehnelt, 2000), so that one could hope to set quantitative limits on their respective contributions to the submm fluxes. The situation is indeed as complex from the theoret- ical as well as the observational points of view. That is why this simple modelling should be considered as an ex- ploratory step to probe the complex issues involved in multi -- wavelength counts, and to design a more satisfying approach of the problem. Acknowledgements. We are pleased to thank David Elbaz for communicating pertinent comments on an early version of this paper as well as for providing us with his data in electronic form. We also acknowledge fruitful discussion with Herv´e Aus- sel, Fran¸cois R. Bouchet, Guilaine Lagache, and Jean -- Loup Puget. References Aguirre, A. and Haiman, Z., 1999, ApJ submitted Andreani, P. and Franceschini, A., 1996, MNRAS 283, 85 Arnouts, S., De Lapparent, V., Mathez, G., Mazure, A., Mellier, Y., Bertin, E., and Kruszewski, A., 1997, A&AS 124, 163 Ashby, M. L. N., Hacking, P. B., Houck, J. R., Soifer, B. T., and Weisstein, E. W., 1996, ApJ 456, 428+ Aussel, H., Cesarsky, C. J., Elbaz, D., and Starck, J. L., 1999, A&A 342, 313 Aussel, H., Coia, D., Mazzei, P., De Zotti, G., and Frances- chini, A., 2000, A&AS 141, 257 Bardeen, J. M., Bond, J. R., Kaiser, N., and Szalay, A. S., 1986, ApJ 304, 15 Barger, A. J., Cowie, L. L., and Sanders, D. B., 1999a, ApJ 518, L5 Barger, A. J., Cowie, L. L., Sanders, D. B., Fulton, E., Taniguchi, Y., Sato, Y., Kawara, K., and Okuda, H., 1998, Nature 394, 248 Barger, A. J., Cowie, L. L., Smail, I., Ivison, R. J., Blain, A. W., and Kneib, J. P., 1999b, AJ 117, 2656 Bertin, E. and Dennefeld, M., 1997, A&A 317, 43 Blain, A. W., Jameson, A., Smail, I., Longair, M. S., Kneib, J.-P., and Ivison, R. J., 1999, MNRAS Bosma, A., 1981, AJ 86, 1825 Chen, H. W., Lanzetta, K. M., and Pascarelle, S., 1999, Nature 398, 586 Cole, S., Aragon-Salamanca, A., Frenk, C. S., Navarro, J. F., and Zepf, S. E., 1994, MNRAS 271, 781+ Crampton, D., Le F`evre, O., Lilly, S. J., and Hammer, F., 1995, ApJ 455, 96+ Dalcanton, J. J., Spergel, D. N., and Summers, F. J., 1997, ApJ 482, 659+ de Young, D. S. and Heckman, T. M., 1994, ApJ 431, 598 Dekel, A. and Silk, J., 1986, ApJ 303, 39 D´esert, F. X., Boulanger, F., and Puget, J. L., 1990, A&A 237, 215 Devriendt, J. E. G., 1999, Ph.D. thesis, Universit´e de Paris XI. Devriendt, J. E. G., Guiderdoni, B., and Sadat, R., 1999, A&A 350, 381 Djorgovski, S., Soifer, B. T., Pahre, M. A., Larkin, J. E., Smith, J. D., Neugebauer, G., Smail, I., Matthews, K., Hogg, D. W., Blandford, R. D., Cohen, J., Harrison, W., and Nelson, J., 1995, ApJ 438, L13 Downes, D., Neri, R., Greve, A., Guilloteau, S., Casoli, F., 12 J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model Hughes, D., Lutz, D., Menten, K. M., Wilner, D. J., Andreani, P., Bertoldi, F., Carilli, C. L., Dunlop, J., Genzel, R., Gueth, F., Ivison, R. J., Mann, R. G., Mel- lier, Y., Oliver, S., Peacock, J., Rigopoulou, D., Rowan- Robinson, M., Schilke, P., Serjeant, S., Tacconi, L. J., and Wright, M., 1999, A&A 347, 809 Eales, S., Lilly, S., Gear, W., Dunne, L., Bond, J. R., Hammer, F., Le F`evre, O., and Crampton, D., 1999, ApJ 515, 518 Elbaz, D., Cesarsky, C. J., Fadda, D., Aussel, H., D´esert, F. X., Franceschini, A., Flores, H., Harwit, M., Puget, J. L., Starck, J. L., Clements, D. L., Danese, L., Koo, D. C., and Mandolesi, R., 1999, A&A 351, L37 Fall, S. M. and Efstathiou, G., 1980, MNRAS 193, 189 Fixsen, D. J., Dwek, E., Mather, J. C., L., B. C., and Shafer, R. A., 1998, ApJ , in press Flores, H., Hammer, F., Thuan, T. X., C´esarsky, C., Desert, F. X., Omont, A., Lilly, S. J., Eales, S., Cramp- ton, D., and Le F`evre, O., 1999, ApJ 517, 148 Franceschini, A., De Zotti, G., Toffolatti, L., Mazzei, P., and Danese, L., 1991, A&AS 89, 285 ture 394, 241+ Kauffmann, G., Guiderdoni, B., and White, S. D. M., 1994, MNRAS 267, 981+ Kauffmann, G. and Haehnelt, M., 2000, MNRAS 311, 576 Kauffmann, G., White, S. D. M., and Guiderdoni, B., 1993, MNRAS 264, 201+ Kawara, K., Sato, Y., Matsuhara, H., Taniguchi, Y., Okuda, H., Sofue, Y., Matsumoto, T., Wakamatsu, K., Karoji, H., Okamura, S., Chambers, K. C., Cowie, L. L., Joseph, R. D., and Sanders, D. B., 1998, A&A 336, L9 Kennicutt, R. C., 1998, ApJ 498, 541 Lacey, C., Guiderdoni, B., Rocca-Volmerange, B., and Silk, J., 1993, ApJ 402, 15 Lacey, C. and Silk, J., 1991, ApJ 381, 14 Lagache, G., Abergel, A., Boulanger, F., D´esert, F. X., and Puget, J. L., 1999, A&A 344, 322 Lilly, S. J., Eales, S. A., Gear, W. K. P., Hammer, F., Le F`evre, O., Crampton, D., Bond, J. R., and Dunne, L., 1999, ApJ 518, 641 Franceschini, A., Mazzei, P., De Zotti, G., and Danese, L., Lilly, S. J., Tresse, L., Hammer, F., Crampton, D., and Le 1994, ApJ 427, 140 F`evre, O., 1995, ApJ 455, 108+ Gardner, J. P., Sharples, R. M., Carrasco, B. E., and Frenk, C. S., 1996, MNRAS 282, L1 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., Tac- coni, L., and Thatte, N., 1998, ApJ 498, 579+ Guiderdoni, B., Bouchet, F. R., Puget, J. L., Lagache, G., Lobo, C. and Guiderdoni, B., 1999, A&A 345, 712 Lutz, D., Spoon, H. W. W., Rigopoulou, D., Moorwood, A. F. M., and Genzel, R., 1998, ApJ 505, L103 Mazzei, P., Xu, C., and de Zotti, G., 1992, A&A 256, 45 Metcalfe, N., Shanks, T., Fong, R., Gardner, J., and Roche, N., 1996, in IAU Symposia, Vol. 171, pp 225+ Meurer, G. R., Heckman, T. M., and Calzetti, D., 1999, and Hivon, E., 1997, Nature 390, 257+ ApJ 521, 64 Guiderdoni, B., Hivon, E., Bouchet, F. R., and Maffei, B., Mo, H. J., Mao, S., and White, S. D. M., 1998, MNRAS 1998, MNRAS 295, 877 295, 319 Guiderdoni, B. and Rocca-Volmerange, B., 1987, A&A Moustakas, L. A., Davis, M., Graham, J. R., Silk, J., Pe- 186, 1 Guiderdoni, B. and Rocca-Volmerange, B., 1990, A&A 227, 362 Hatton, S. J., Ninin, S., Devriendt, J. E. G., Bouchet, F. R., Guiderdoni, B., Stohr, F., and Vibert, D., 2000, MNRAS in prep Hauser, M. G., Arendt, R. G., Kelsall, T., Dwek, E., Ode- gard, N., Weiland, J. L., Freudenreich, H. T., Reach, W. T., Silverberg, R. F., Moseley, S. H., Pei, Y. C., Lubin, P., Mather, J. C., Shafer, R. A., Smoot, G. F., Weiss, R., Wilkinson, D. T., and Wright, E. L., 1998, ApJ 508, 25 Heath, D. J., 1977, MNRAS 179, 351 Heyl, J. S., Cole, S., Frenk, C. S., and Navarro, J. F., 1995, MNRAS 274, 755 Hogg, D. W., Pahre, M. A., McCarthy, J. K., Cohen, J. G., Blandford, R., Smail, I., and Soifer, B. T., 1997, MN- RAS 288, 404 Hughes, D. H., Serjeant, S., Dunlop, J., Rowan -- Robinson, M., Blain, A., Mann, R. G., Ivison, R., Peacock, J., Ef- stathiou, A., Gear, W., Oliver, S., Lawrence, A., Lon- gair, M., Goldschmidt, P., and Jenness, T., 1998, Na- terson, B. A., and Yoshii, Y., 1997, ApJ 475, 445+ Oliver, S. J., Goldschmidt, P., Franceschini, A., Ser- jeant, S. B. G., Efstathiou, A., Verma, A., Gruppioni, C., Eaton, N., Mann, R. G., Mobasher, B., Pearson, C. P., Rowan-Robinson, M., Sumner, T. J., Danese, L., Elbaz, D., Egami, E., Kontizas, M., Lawrence, A., MCMahon, R., Norgaard-Nielsen, H. U., Perez- Fournon, I., and Gonzalez-Serrano, J. I., 1997, MN- RAS 289, 471 Peebles, P. J. E., 1980, The large-scale structure of the universe, Princeton University Press Pei, Y. C., 1995, ApJ 438, 623 Puget, J. L., Abergel, A., Bernard, J. P., Boulanger, F., Burton, W. B., Desert, F. X., and Hartmann, D., 1996, A&A 308, L5 Puget, J. L., Lagache, G., Clements, D. L., Reach, W. T., Aussel, H., Bouchet, F. R., Cesarsky, C., D´esert, F. X., Dole, H., Elbaz, D., Franceschini, A., Guiderdoni, B., and Moorwood, A. F. M., 1999, A&A 345, 29 Renzini, A., 1999, in J. Walsh and M. Rosa (eds.), Chem- ical Evolution from Zero to High Redshift, Springer -- Verlag J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model 13 Richstone, D., Loeb, A., and Turner, E. L., 1992, ApJ 393, 477 Rigopoulou, D., Lawrence, A., and Rowan-Robinson, M., 1996, MNRAS 278, 1049 Sanders, D. B. and Mirabel, I. F., 1996, ARA&A 34, 749+ Schlegel, D. J., Finkbeiner, D. P., and Davis, M., 1998, ApJ 500, 525+ Silva, L., Granato, G. L., Bressan, A., and Danese, L., 1998, ApJ 509, 103 Smail, I., Hogg, D. W., Yan, L., and Cohen, J. G., 1995, ApJ 449, L105 Smail, I., Ivison, R. J., and Blain, A. W., 1997, ApJ 490, L5 Smail, I., Ivison, R. J., Blain, A. W., and Kneib, J. P., 1998, ApJ 507, L21 Smail, I., Ivison, R. J., Kneib, J.-P., Cowie, L. L., Blain, A. W., Barger, A. J., Owen, F. N., and Morrison, G., 1999, MNRAS 308, 1061 Soifer, B. T. and Neugebauer, G., 1991, AJ 101, 354 Soifer, B. T., Neugebauer, G., Franx, M., Matthews, K., and Illingworth, G. D., 1998, ApJ 501, L171 Somerville, R. and Primack, J., 2000, MNRAS in press Steidel, C. C., Adelberger, K. L., Giavalisco, M., Dickin- son, M., and Pettini, M., 1999, ApJ 519, 1 Sugiyama, N., 1995, ApJS 100, 281+ Sutherland, R. S. and Dopita, M. A., 1993, ApJS 88, 253 Thornton, K., Gaudlitz, M., Janka, H. T., and Steinmetz, M., 1998, ApJ 500, 95+ Weir, N., Djorgovski, S., and Fayyad, U. M., 1995, AJ 110, 1+ White, S. D. M. and Frenk, C. S., 1991, ApJ 379, 52 Williams, R. E., Blacker, B., Dickinson, M., Dixon, W. V. D., Ferguson, H. C., Fruchter, A. S., Giavalisco, M., Gilliland, R. L., Heyer, I., Katsanis, R., Levay, Z., Lu- cas, R. A., McElroy, D. B., Petro, L., Postman, M., Adorf, H.-M., and Hook, R., 1996, AJ 112, 1335+ Zel'dovich, Y. B., 1965, Adv. Astr. Ap. 3, 241+ where Ω0 is the current matter density (in critical density units), ΩB is the baryon density, and h = H0/(100 km/s/Mpc) is the reduced Hubble constant. In the linear regime, the equation of motion is solved for the expansion factor, a, and the solutions for the growth of the density contrast δ (see e.g. Peebles (1980)) are derived. There are two such solutions (Heath, 1977), which form a complete set (Zel'dovich, 1965) and read: Dd[z] = H0(cid:8)Ω0(1 + z)3 + (1 − Ω0 − λ0)(1 + z)2 + λ0(cid:9) Dg[z] = Dd[z] a2 dx, (A.4) 1 2 (A.3) 0Z ∞ z (1 + x) D3 d[x] where the subscripts d and g respectively stand for the decaying and growing modes, and λ0 = Λ/(3H 2 0 ) is the re- duced cosmological constant. If one further assumes that the initial peculiar velocity of the perturbation is zero (i.e. that the perturbation simply moves along with the ex- panding universe), the density contrast δ[z] in the linear regime grows as: δ[z] = (cid:26) 3 2 Ω0(1 + zi) + 1 − Ω0 − λ0(cid:27) H 2 a2 0 0 Dg[z] δi, (A.5) where the subscript i stands for the initial quantities. In the non -- linear regime, one considers an isolated spherical perturbation of radius ri, at time ti, which has a uniform overdensity δi ≡ (ρp[ti] − ρb[ti])/ρb[ti] with re- spect to the background (δi ≪ 1), and encloses a mass M = 4/3πr3 i ρb[ti](1 + δi). We assume that the perturba- tion is bound, and that its peculiar velocity is nil. The equation of motion is integrated to obtain the time tm at which the perturbation reaches its maximum expansion radius rm: tm = ti + 1 H0Z rm/ri 1 x1/2 Dp[x]1/2 dx, where Dp[x] = λ0 x3 + {(1 − Ω0 − λ0)(1 + zi)2 − Ω0δi(1 + zi)3} x + Ω0(1 + δi)(1 + zi)3, (A.6) (A.7) (A.8) Appendix A: Dark matter halos in any cosmology We suppose that perturbations of the matter density field when the universe becomes matter dominated are com- pletely characterised by their power spectrum P (k): P (k) ∝ knT 2(k), where T (k) is the transfer function (see fit given in Ap- pendix G of Bardeen et al. (1986)). We further assume a post-inflation Harrison -- Zel'dovich power spectrum (n = 1) for these perturbations, and take the shape parameter Γ used in the computation of T (k) to be (Sugiyama, 1995): (A.1) and rm/ri is the first real root > 1 of the cubic equation (c.f. Richstone et al. (1992)): Dp[x] = 0. (A.9) After it has reached this maximum radius at time tm, the perturbation, by symmetry, collapses on a time scale tcoll = 2tm − ti. Thus, with the previous equations, the critical density contrast δ0 linearly extrapolated till today (eq. A.5) is explicitly related to the collapse redshift zcoll of the perturbation for any cosmology, just by computing the redshift to which the collapse time of a perturbation with overdensity δi corresponds in the unperturbed uni- verse (provided tcoll is smaller than the age of the uni- verse): Γ = Ω0h exp"−ΩB(1 + ph/0.5 Ω0 )# , (A.2) tcoll ≡Z ∞ zcoll 1 (1 + x)Dd[x] dx. (A.10) 14 J. Devriendt & B. Guiderdoni: II. Multi -- Wavelength Counts from a Semi -- Analytic Model 0 0 0 -1 -1 -1 -2 -2 -2 -3 -3 -3 -4 -4 -4 -5 -5 -5 0 0 0 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 Fig. A.1. Evolution of the formation rate of dark mat- ter halos (masses indicated on the figure) for three differ- ent cosmologies: SCDM (solid line), OCDM (dashes), and ΛCDM (dots). The parameters used for these models are given in table 1. If the resulting virialized perturbation can be approxi- mated by a singular isothermal sphere truncated at virial radius rvir, then rvir/rm is the solution of the following cubic equation (see Devriendt (1999)): 0 r3 m 4λ0H 2 3GM x3 − ( 12 5 + 0 r3 m 6λ0H 2 5GM )x + 1 = 0. (A.11) Finally, the velocity of a test particle moving on a circular orbit in the isothermal sphere reads: Vc = √2σ =r GM rvir − 2λ0 9 H 2 0 r2 vir. (A.12) Implementing these results into the peaks formalism described in GHBM enables one to derive formation rates for dark matter halos as a function of redshift. Fig A.1 illustrates this halo formation rate for three typical cos- mologies gathered in table 1.
astro-ph/0206151
2
0206
2002-06-11T21:15:26
HETE-2 Localization and Observations of the Short, Hard Gamma-Ray Burst GRB020531
[ "astro-ph" ]
The {\it HETE-2} (hereafter \HETE) French Gamma Telescope (FREGATE) and the Wide-field X-ray Monitor (WXM) instruments detected a short ($t_{50} = 360$ msec in the FREGATE 85-300 keV energy band), hard gamma-ray burst (GRB) that occurred at 1578.72 SOD (00:26:18.72 UT) on 31 May 2002. The WXM flight localization software produced a valid location in spacecraft (relative) coordinates. However, since no on-board real-time star camera aspect was available, an absolute localization could not be disseminated. A preliminary localization was reported as a GCN Position Notice at 01:54:22 UT, 88 min after the burst. Further ground analysis produced a refined localization, which can be expressed as a 90% confidence rectangle that is 67 arcminutes in RA and 43 arcminutes in Dec (90% confidence region), centered at RA = +15$^{\rm h}$ 14$^{\rm m}$ 45$^{\rm s}$, Dec = -19$^\circ$ 21\arcmin 35\arcsec (J2000). An IPN localization of the burst was disseminated 18 hours after the GRB (Hurley et al. 2002b). A refined IPN localization was disseminated $\approx$ 5 days after the burst. This hexagonal-shaped localization error region is centered on RA = 15$^{\rm h}$ 15$^{\rm m}$ 03.57$^{\rm s}$, -19$^\circ$ 24\arcmin 51.00\arcsec (J2000), and has an area of $\approx$ 22 square arcminutes (99.7% confidence region). The prompt localization of this short, hard GRB by \HETE and the anti-Sun pointing of the \HETE instruments, coupled with the refinement of the localization by the IPN, has made possible rapid follow-up observations of the burst at radio, optical, and X-ray wavelengths.
astro-ph
astro-ph
DRAFT VERSION NOVEMBER 20, 2018 Preprint typeset using LATEX style emulateapj v. 14/09/00 HETE-2 LOCALIZATION AND OBSERVATIONS OF THE SHORT, HARD GAMMA-RAY BURST GRB020531 D. Q. LAMB,1 G. R. RICKER,2 J.-L. ATTEIA,3 K. HURLEY,4 N. KAWAI,5,6 Y. SHIRASAKI,5,9 T. SAKAMOTO,5,6,10 T. TAMAGAWA,6 C. GRAZIANI,1 J.-F. OLIVE,3 A. YOSHIDA,5,7 M. MATSUOKA,8 K. TORII,6 E. E. FENIMORE,10 M. GALASSI,10 T. TAVENNER,10 T. Q. DONAGHY,1 M. BOER,3 J.-P. DEZALAY,3 R. VANDERSPEK,2 G. CREW,2 J. DOTY,2 G. MONNELLY,2 J. VILLASENOR,2 N. BUTLER,2 J. G. JERNIGAN,4 A. LEVINE,2 F. MARTEL,2 E. MORGAN,2 G. PRIGOZHIN,2 S. E. WOOSLEY,11 T. CLINE,12 I MITROFANOV,13 D. ANFIMOV,13 A. KOZYREV,13 M. LITVAK,13 A. SANIN,13 W. BOYNTON,14 C. FELLOWS,14 K. HARSHMAN,14 C. SHINOHARA,14 R. STARR,12 J. BRAGA,15 R. MANCHANDA,16 G. PIZZICHINI,17 K. TAKAGISHI,18 AND M. YAMAUCHI18 Draft version November 20, 2018 ABSTRACT The HETE-2 (hereafter HETE ) French Gamma Telescope (FREGATE) and the Wide-field X-ray Monitor (WXM) instruments detected a short (t50 = 360 msec in the FREGATE 85-300 keV energy band), hard gamma- ray burst (GRB) that occurred at 1578.72 SOD (00:26:18.72 UT) on 31 May 2002. The WXM flight localization software produced a valid location in spacecraft (relative) coordinates. However, since no on-board real-time star camera aspect was available, an absolute localization could not be disseminated. A preliminary localization was reported as a GCN Position Notice at 01:54:22 UT, 88 min after the burst. Further ground analysis produced a refined localization, which can be expressed as a 90% confidence rectangle that is 67 arcminutes in RA and 43 arcminutes in Dec (90% confidence region), centered at RA = +15h 14m 45s, Dec = -19◦ 21′35′′(J2000). An IPN localization of the burst was disseminated 18 hours after the GRB (Hurley et al. 2002b). A refined IPN localization was disseminated ≈ 5 days after the burst. This hexagonal-shaped localization error region is centered on RA = 15h 15m 03.57s, -19◦ 24′51.00′′(J2000), and has an area of ≈ 22 square arcminutes (99.7% confidence region). The prompt localization of this short, hard GRB by HETE and the anti-Sun pointing of the HETE instruments, coupled with the refinement of the localization by the IPN, has made possible rapid follow-up observations of the burst at radio, optical, and X-ray wavelengths. The time history of GRB020531 at high (> 30 keV) energies consists of a short, intense spike followed by a much less intense secondary peak. Its time history is thus similar to that seen in many short, hard bursts. Analysis of the FREGATE and WXM time histories gives durations for the burst of t50 = 1.36 s in the WXM 2 - 25 keV energy range, and 1.10 s, 0.86 s, 0.62 s, and 0.36 s in the FREGATE 6- 13, 14-30, 31-84, and 85-400 keV energy bands. The duration of the burst thus increases with decreasing energy, which is similar to the behavior of long GRBs. The photon number flux, photon energy flux, and energy fluence of the burst in the 50-300 keV energy band in 1.25 seconds are 3.0 ph cm- 2 s- 1, 6.4 × 10- 7 erg cm- 2 s- 1, and 8.0 × 10- 7 erg cm- 2, respectively. The spectrum of the burst evolves from hard to soft, which is also similar to long GRBs. These similarities to the properties of long GRBs, and other similarities previously known, suggest that short, hard GRBs are closely related to long GRBs. Subject headings: gamma rays: bursts (GRB020531) 1 Department of Astronomy and Astrophysics, University of Chicago, 5640 South Ellis Avenue, Chicago, IL 60637. 2 Center for Space Research, Massachusetts Institute of Technology, 70 Vassar Street, Cambridge, MA, 02139. 3 Centre d'Etude Spatiale des Rayonnements, CNRS/UPS, B.P.4346, 31028 Toulouse Cedex 4, France. 4 Space Sciences Laboratory, University of California at Berkeley, 601 Campbell Hall, Berkeley, CA, 94720. 5 Department of Physics, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8551, Japan. 6 RIKEN (Institute of Physical and Chemical Research), 2-1 Hirosawa, Wako, Saitama 351-0198, Japan. 7 Department of Physics, Aoyama Gakuin University, Chitosedai 6-16-1 Setagaya-ku, Tokyo 157-8572, Japan. 8 Tsukuba Space Center, National Space Development Agency of Japan, Tsukuba, Ibaraki, 305-8505, Japan. 9 National Astronomical Observatory, Osawa 2-21-1, Mitaka, Tokyo 181-8588 Japan. 10 Los Alamos National Laboratory, P.O. Box 1663, Los Alamos, NM, 87545. 11 Department of Astronomy and Astrophysics, University of California at Santa Cruz, 477 Clark Kerr Hall, Santa Cruz, CA 95064. 12 NASA Goddard Space Flight Center, Greenbelt, MD, 20771. 13 Space Research Institute, Profsojuznaya Str. 84/32, 117810, Moscow, Russia. 14 Department of Planetary Sciences, Lunar and Planetary Laboratory, Tucson, AZ 85721-0092. 15 Instituto Nacional de Pesquisas Espaciais, Avenida Dos Astronautas 1758, São José dos Campos 12227-010, Brazil. 16 Department of Astaronomy and Astrophysics, Tata Institute of Fundamental Research, Homi Bhabha Road, Mumbai, 400 005, India. 17 Consiglio Nazionale delle Ricerche (IASF), via Piero Gobetti, 101-40129 Bologna, Italy. 18 Faculty of engineeering, Miyazaki University, Gakuen Kibanadai Nishi, Miyazaki 889-2192, Japan. 1 2 Lamb et al. 1. INTRODUCTION It has been known for nearly a decade that gamma-ray bursts (GRBs) appear to fall into two classes: short (≈ 0.2 sec), harder bursts, which comprise 20-25% of all bursts; and long (≈ 20 sec), softer bursts, which comprise 75-80% of the total (Hur- ley et al. 1992; Lamb, Graziani, and Smith 1993; Kouveliotou et al. 1993). Norris, Scargle & Bonnell 2000). The spectra of the two classes of bursts differ: the spectra of the bursts become softer as the bursts become longer (Dezalay et al. 1992; Kouve- liotou et al. 1993; Dezalay et al. 1996). There is also evidence that the brightness distributions of the two classes of bursts dif- fer (Graziani and Lamb 1994; Belli 1997; Tavani 1998); how- ever, the difference in the brightness distributions can be ex- plained by the difference in their durations, which causes the sampling distance for short bursts to be smaller than for the long bursts (Graziani and Lamb 1994). In addition the V /Vmax val- ues (Schmidt 2001) and the angular distributions (Kouveliotou et al. 1993) of the two classes appear to be identical. Thanks to the rapid dissemination of accurate GRB localiza- tions by BeppoSAX (Costa et al. 1997), much has been learned in the past five years about GRBs. This has included the dis- coveries that GRBs have X-ray (Costa et al. 1997), optical (van Paradijs et al. 1997), and radio (Frail et al. 1997) afterglows. Redshifts and host galaxies are now known for more than two dozen GRBs (see, e.g., Lamb 2002). However, all of these dis- coveries relate to long GRBs. In contrast, to date nothing is known about the distance to or the nature of the short GRBs, despite extensive efforts. The Burst and Transient Source Experiment (BATSE) on the Compton Gamma-Ray Observatorylocalized localized numer- ous short GRBs in near-real time. Although many had large error boxes, one (trigger 6788) was localized to an error cir- cle of ∼ 30 square degrees, which was searched for an optical counterpart within 12 s to a magnitude of 14.98 (Kehoe et al. 2001). The results were negative. The Third Interplanetary Net- work (IPN) derived localizations for four short GRBs (000607, 001025B, 001204, and 010119) with delays of 15 -- 65 hours. But in three of these cases, the opportunity for follow-up ob- servations was compromised either by the burst being close to the Sun (000607, 65◦) or close to the Galactic plane (001025B, b ≈ 4◦; 010119, b ≈ 5◦). Only one burst (001204) was opti- mally placed on the sky for follow-up observations. However, in this case the delay (65 hours) in deriving a localization for the burst hampered follow-up efforts. Despite an accepted Bep- poSAX ToO program, no X-ray follow-up observations were possible because of Sun-angle or other operational constraints, except in the case of 001204. However, the delay in deriving the localization of this burst made the success of any X-ray follow- up observation unlikely, and therefore none was carried out. In this Letter we report the detection and prompt localization of a short, hard GRB by HETE-2 (hereafter HETE ) (Ricker et al. 2002a,b). On 31 May 2002 at 1578.73 SOD (05:15:50.56) UT on 31 May 2002 UT, the HETE-2 French Gamma Telescope (FREGATE) instrument (Atteia et al. 2002) and the Wide-field X-ray Monitor (WXM) instrument (Kawai et al. 2002) detected a bright, short (duration ∼ 300 msec in the FREGATE 30-400 keV energy band), hard GRB. The prompt localization of this short, hard GRB by HETE and the anti-Sun pointing of the HETE instruments has made possible rapid follow-up observa- tions of it at radio, optical, and X-ray wavelengths. We also de- scribe the properties of GRB020531 derived from observations of it using the FREGATE and WXM instruments on HETE , which provided unprecedented spectral and temporal coverage of this short, hard GRB. 2. OBSERVATIONS 2.1. Localization The HETE FREGATE and WXM instruments detected a short (duration ∼ 300 msec in the FREGATE 30-400 keV energy band), hard gamma-ray burst (GRB) that occurred at 1578.72 SOD (00:26:18.72 UT) on 31 May 2002. The fluence of the burst was not large enough for it to be detected by the HETE Soft X-Ray camera (SXC) (Monnelly et al. 2002). The WXM flight localization software produced a valid location in spacecraft (relative) coordinates. However, since no on-board real-time star camera aspect was available, an absolute localiza- tion could not be disseminated. A preliminary ground-analysis localization was reported as a GCN Position Notice at 01:54:22 UT, 88 minutes after the burst. Further ground analysis pro- duced a refined localization, which can be expressed as a 90% confidence rectangle that is 67 arcminutes in RA and 43 ar- cminutes in Dec, centered at RA = +15h 14m 45s, Dec = -19◦ 21′35′′(J2000) (see Figure 1). The refined localizaton was re- ported as a GCN Position Notice at 03:57:14 UT, 211 minutes after the burst. By restricting the search to the crossing window defined by the HETE position, a weak GRB was identified in the Ulysses data. An initial IPN localization of the burst was derived by triangulation using HETE FREGATE, Ulysses GRB and Mars Odyssey HEND data, and a 46 square arcminute error box was disseminated 18 hours after the GRB (Hurley et al. 2002b). This localization was further refined approximately five days after the burst. The resulting hexagonal-shaped local- ization error region is centered on RA = 15h 15m 03.57s, -19◦ 24′51.00′′(J2000), and has an area of ≈ 22 square arcminutes (99.7% confidence region) (see Figure 1) (Hurley et al. 2002c). We have used the imaging capabilities of the WXM to greatly improve the S/N of GRB020531 as seen by the WXM. Using the WXM photon time- and energy-tagged data (TAG data), we selected only WXM photons (1) during the portion of the burst that maximized the S/N of the burst time history, using Spiffy Trigger (Graziani 2002); (2) on the seven wires illuminated by the GRB but not by the bright Galactic X-ray source Sco X-1, which was in the field-of-view (FOV) of the WXM at the time; (3) on those portions of these seven wires that were illuminated by the burst; and (4) the bins on those portions of the wires that were illuminated by the burst, using the refined IPN localiza- tion of the burst (Hurley et al. 2002b) and the mask pattern of the coded aperture of the WXM. These "cuts" on the photon TAG data increased the S/N of the burst in the WXM data by nearly a factor of two. While these "cuts" on the WXM photon TAG data do not allow an improved localization of the GRB by the WXM, their success provides independent evidence that the refined IPN error region for the burst is correct. 2.2. Temporal Properties Figure 2 shows the time history of GRB020531 in the WXM 2-25 keV energy band and in various FREGATE energy bands. Table lists the t50 and t90 durations of the burst in the WXM 2-25 kev energy band and in various FREGATE energy bands. Figure 3 shows the burst time history in the entire WXM 2-25 keV energy band, and in four other WXM energy bands, utiliz- ing the four "cuts" on the WXM photon TAG data that increase HETE-2 Observations of GRB020531 3 the S/N of the burst in the WXM data, and that are described in §2.1 section. The t50 and t90 durations of GRB020531 in the 85-400 keV energy band were 0.36 s and 0.74 s; thus GRB020531 is a short, hard GRB. Like many short bursts (see, e.g., 010119; Hurley et al. 2002), the time history of GRB020531 consists of a short, intense spike followed by a less intense and softer secondary peak. At low energies, the burst consists of two peaks, each lasting < 1 sec and separated by about 2 seconds (see Figure 3). Thus, at low energies, the primary and secondary peaks are comparable in duration and in intensity. Figure 4 shows the t50 and t90 durations of GRB020531 in various energy bands. The duration of the burst increases with decreasing energy. A χ2 fit to the t50 values, assuming that all of the values have the same relative uncertainty, yields logt50 = 0.42 - 0.38 log(E/1keV); a similar fit to the t90 values yields logt90 = 1.42 - 0.62 log(E/1keV). This behavior is simi- lar to that seen in long GRBs (Fenimore et al. 1994). 2.3. Spectrum Table 3 gives the S/N of the detection of the burst, the peak photon flux, the peak energy flux, and the energy fluence of the burst in the WXM and in various FREGATE energy band. Table 4 gives the best-fit parameters for the spectrum of GRB020531 in the WXM and in FREGATE. Figure 5 shows the expected and observed photon counts in FREGATE energy loss bins from 6 -- 400 keV (upper panel) and the residuals (lower panel) for the best fit power-law spectrum given in Table 4. Figure 6 shows the corresponding photon counts in the WXM energy loss bins from 2-25 keV for the four wires in the X- detector and the two wires in the Y-detector (Kawai et al. 2002) used in the WXM spectral fit. The spectra derived independently in the WXM and in FRE- GATE are consistent with one another. The results show that the observed spectrum of GRB020531 is fully consistent with a single power-law spectrum with slope α = - 1.2 ± 0.06. The power-law spectrum of GRB020531 extends from 2 -- 400 keV, and shows no evidence of a cutoff at high energies to the highest energies (≈ 400 keV) observed by FREGATE. We have explored the spectral evolution of the burst in the WXM 2 -- 25 keV energy band. Using Spiffy Trigger (Graziani 2002), we found two time intervals during the burst that max- imized the S/N of the time history. These two time intervals start 0.00 sec and 1.12 sec after the trigger time for the burst, and last ∆t1 = 0.80 sec and ∆t2 = 0.72 sec. We have used the four "cuts" on the WXM photon TAG data that are described in §2.1, and that increase the S/N of the burst in the WXM data, in order to determine the spectrum of the burst in the WXM 2 -- 25 keV energy band in these two time intervals. Table 4 shows the resulting best-fit power-law parameters for these two time intervals. Comparing the value of the power-law index (α = 1.26 ±0.06) for the spectrum during the first peak, as deter- mined from the FREGATE data, and the value of the power-law index (α = 2.186+1.38 - 0.95) for the spectrum during the second peak, as determined from the WXM data, provides evidence at the ≈ 90% confidence level that the spectrum of the secondary peak is softer than the spectrum of the primary peak. Such spectral softening is similar to that seen in long GRBs (see, e.g., Band et al. 1993). 3. DISCUSSION HETE has detected and localized GRB020531, a short, hard burst. The prompt localization of the burst by HETE and the anti-Sun pointing of the HETE instruments, coupled with the later precise localization of the burst by the IPN, has allowed rapid follow-up of the GRB not only by small aperture, large FOV robotic telescopes (e.g., Park et al. 2002, Boer et al. 2002) but also by large aperture, modest FOV telescopes (e.g., Fox et al. 2002, Lamb et al. 2002, West et al. 2002, Miceli et al. 2002, Dullighan et al. 2002). Figure 7 shows that these observations have placed much more severe upper limits on any optical af- terglow of a short, hard GRB than ever before. However, these constraints do not rule out the existence of optical afterglows of short, hard GRBs that are similar to the optical afterglows of long GRBs (see Hurley et al. 2002, Figure 3). The time history of GRB020531 at high (> 30 keV) energies consists of a short, intense spike followed by a much less in- tense secondary peak. Its time history is thus similar to that seen in many short, hard bursts. The time history of GRB020531 at low (< 25 keV) energies consists of two peaks of similar dura- tion (≈ 0.80 sec) and intensity, the first of which corresponds in time to the short, intense spike seen at high energies. The spectrum of the short, intense spike is well described by a single power-law spectrum with index α = - 1.2 ± 0.06 from 6 -- 400 keV. Such a steep spectrum is quite unusual (Paciesas et al. 2001) but not unprecedented (see, e.g., Hurley et al. 2002; Figure 2 [GRB001204]) for short, hard GRBs. The steepness of the spectrum of GRB020531 may explain in part the fact that the WXM on HETE was able to detect and localize this par- ticular short, hard burst, whereas the Wide-Field Cameras on BeppoSAX were ultimately unsuccessful in dectecting or local- izing any short, hard GRBs, despite great efforts (Gandolfi et al. 2002). The spectrum of the secondary peak, which is comparable in duration and in intensity in the 2 -- 25 keV energy band to the primary peak, is also well described by a single power-law spectrum, but the spectrum is softer than the spectrum of the primary peak at the 90% confidence level. Two qualitatively different models have been suggested to explain GRBs: merging compact objects and the collapse of rotating massive stars. It is difficult to make short bursts in the collapsar model (MacFadyen & Woosley 1999) and there have been suggestions (e.g., Fryer, Woosley, & Hartmann 1999) that short hard bursts are the observational counterpart of merging neutron-star and neutron-star black-hole pairs. While collap- sars and merging compact objects both get their energy from a hyperaccreting black hole, the dimension and mass of the disk is smaller in the latter, hence the time scale is shorter. If these mergers go on far from the galaxies where the compact objects are made, their afterglows might be faint. The spectral and temporal behavior measured by HETE for GRB020531 could indicate a central engine that remains on with a declining power after the principal burst (reflecting a de- clining accretion rate?) or may be a consequence of the compe- tition between internal and external shocks in making the GRB. The short time scale is certainly more consistent with the merg- ing compact object hypothesis, but also possibly consistent with the supranova model (Vietri & Stella 1998, 1999) since the lat- ter produces a similar compact disk and black hole. This could give a short burst should such bursts prove to be associated with massive stars. The properties of GRB020531 as measured by HETE have different characteristics at different energies: more complex time structure at lower energies, increasing duration with de- 4 Lamb et al. creasing energy, a power-law spectrum over the 2-400 keV en- ergy range but spectral softening with time). These properties of GRB020531 as measured by HETE are similar to those of long bursts, and when taken together with the previously know properties of short, hard GRBs described in the Introduction (similar brightness distributions, V /Vmax values, angular distri- butions) suggest that short, hard GRBs are closely related to long GRBs. 4. CONCLUSIONS HETE has detected and localized a short, hard GRB, estab- lishing its capability to do so -- and demonstrating that the de- tection and localization of short, hard GRBs in the hard x-ray energy band is possible. This has important implications for Swift and for other future GRB missions. The prompt, precise localization of GRB020531 by HETE and the IPN have allowed rapid follow-up observations, which have placed much more severe limits on the brightness of any radio and optical afterglows from short, hard GRBs. The complement of soft x-ray, hard x-ray, and gamma-ray instruments (SXC, WXM, and FREGATE) on HETE provides unprecedented temporal and spectral coverage of short, hard GRBs. With the currently projected long orbital lifetime (> 10 yrs) and excellent health of the HETE spacecraft and in- struments, the results described for GRB020531 in this Letter demonstrate that HETE can continue to provide an unprece- dented opportunity to study short, hard GRBs, and possibly to determine the distance to and the nature of these bursts. ACKNOWLEDGMENTS The HETE mission is supported in the US by NASA con- tract NASW-4690; in Japan, in part by the Ministry of Edu- cation, Culture, Sports, Science, and Technology Grant-in-Aid 13440063; and in France, by CNES contract 793-01-8479. KH is grateful for Ulysses support under Contract JPL 958056, for HETE support under Contract MIT-SC-R-293291, and for Mars Odyssey support under the NASA LTSA program. G. Pizzichini acknowledges support by the Italian Space Agency. REFERENCES Atteia, J-L, et al. 2002, In-flight Performance and First Results from the Kawai, N., et al. 2002, In-Orbit Performance of the WXM Instrument on FREGATE Instrument on HETE, in WH20017. bibitem[Band1993] Band, D. et al. 1993, ApJ, 413, 281 Belli, B. 1997, in Proc. 25th Int. Cosmic-Ray Conf. (Durban), 41 Boer, M. Klotz, A., Atteia, J.-L., Pollas, C. & Pinna, H. 2002, GCN Circular HETE, in WH2001∗∗. Kehoe, R. et al. 2001, Ap.J. 554, L159 Kouveliotou, C. et al. 1993, ApJ, 413, L101 Lamb, D. Q. 2002, Gamma-Ray Bursts as a Probe of Cosmology, in Graziani, C., Lamb, D. Q. 1994, in AIP Conf. Proc. 307, Gamma-Ray Bursts, ed. G. J. Fishman, J. J. Brainerd, and K. Hurley (New York: AIP), 227 Hurley, K. 1992, in AIP Conf. Proc. 265, Gamma-Ray Bursts, ed. W. Paciesas & G. Fishman (New York: AIP), 3 Hurley, K. 2002a, ApJ, 567, 447 Hurley, K., et al. 2002b, GCN Circ. 1402 Hurley, K., et al. 2002c, GCN Circ. 1407 and Science Overview, in WH2001∗∗. Ruffert, M. & Janka, H. 1999, A&A, 344, 573 Schmidt, M. 2001, ApJ, 559, L79 Tavani, M. 1998, ApJ, 497, L21 Vietri, M., & Stella, L. 1998, ApJL, 507, L45 Vietri, M., & Stella, L. 1999, ApJL, 527, L43 West, D. 2002, GCn Circular 1406 1408 Butler, N., Dullighan, A., Ford, P., Monnelly, G., Ricker, G., Vanderspek, R., Hurley, K. & Lamb, D. 2002, GCN Circular 1415 Costa, E. et al. 1997, IAU Ciruclar No. 6576 Dezalay, J.-P. et al. 1992, in AIP Conf. Proc. 265, Gamma-Ray Bursts, ed. W. Paciesas & G. Fishman (New York: AIP), 304 Dezalay, J.-P. et al. 1996, ApJ, 471, L27 Dulligan, Monnelly, G., Butler, N., Vanderspek, R., Ford, P. & Ricker, G. 2002, GCN Circular 1411 bibitem[Fenimore1994] Fenimore, E. E. 1994, ApJ, 547, 315 Fox, D. & Bloom, J. S. 2002, GCN Circular 1400 Frail, D. et al. 1997, Nature 389, 261 Frail, D. A. & Berger, e. 2002, GCN Circular 1418 Fryer, C. L., Woosley, S. E., Hartmann, D. H. 1999, ApJ, 526, 152 Gandolfi, G. et al. 2002, BeppoSAX Results on Short Gamma-Ray Bursts, in WH2001∗∗. WH2001∗∗. Lamb, D. Q., Graziani, C. & Smith, I. A. 1993, ApJ, 413, L11 Lamb, D. Q. et al. 2002, GCN Circular 1403 MacFadyen, A. & Woosley, S. 1999, ApJ, 524, 262 Miceli, A., Lamb, D. Q., Zucker, D., Covey, K., Dembicky, J. & Hastings, N. C. 2002, GCN Circular 1416 Monnelly, G. et al. 2002, HETE Soft X-Ray Camera Imaging: Calibration, Performance, and Sensitivity, in WH2001∗∗. Paciesas, W. S., Preece, R. D., Briggs, M S., and Malozzi, R. S. 2001, in Gamma-Ray Bursts in the Afterglow Era, (Rome, Italy, 17-20 October 2000), ESO Astrophysics Symposia, Springer (Berlin ), p. 13 van Paradijs, J. et al. 1997, Nature 386, 686 Park, H. S., Williams, G. G., Lindsay, K. 2002, GCN Circular 1404 Ricker, G. R., et al. 2002, GCN Circ. 1399 Ricker, G.R., et al. 2002, High Energy Transient Explorer (HETE): Mission 7 WH2001 = Gamma-Ray Burst and Afterglow Astronomy 2001: A Workshop Celebrating the First Year of the HETE Mission, Woods Hole, MA, November 2001, to be published in the AIP Conference Proceedings (AIP Press: New York). HETE-2 Observations of GRB020531 5 TABLE 1 GRB020531 ERROR BOX COORDINATES Source HETE WXM IPN αJ2000.0 15 14 45 15 17 00 15 17 00 15 12 30 15 12 30 δJ2000.0 -19 21 35 -19 43 00 -19 00 00 -19 00 00 -19 43 00 15 15 03.57 15 14 53.98 15 15 14.46 15 15 17.07 15 15 12.51 15 14 53.75 15 14 49.67 -19 24 51.00 -19 24 18.15 -19 21 38.39 -19 21 35.32 -19 25 32.46 -19 27 58.57 -19 28 03.27 Comment center corner corner corner corner center corner corner corner corner corner corner Note. -- Units of right ascension are hours, minutes, and seconds; units of declination are degrees, arcminutes, and arcseconds. TABLE 2 TEMPORAL PROPERTIES OF GRB020531. Instrument Energy Band (keV) t50 (s) t90 (s) HETE WXM 2 - 25 1.36 4.56 HETE FREGATE 6-13 14-30 31-84 85-400 1.10 0.86 0.62 0.36 5.58 3.46 4.52 0.74 ENERGY EMISSION PROPERTIES OF GRB020531. TABLE 3 Energy Band (keV) Photon Flux (ph cm- 2 s- 1) 2 - 25 8 - 400 32 - 400 25 - 100 50 - 300 2.9 8.4 4.4 3.0 3.0 Energy Flux (erg cm- 2 s- 1) (2.7 ± 0.8) × 10- 8 10.1 × 10- 6 9.1 × 10- 7 2.6 × 10- 7 6.4 × 10- 7 Energy Fluence (erg cm- 2) - 1.4) × 10- 8 (4.9+1.5 12.7 × 10- 7 11.4 × 10- 7 3.1 × 10- 7 8.0 × 10- 7 Note. -- All WXM parameters are for a joint fit to six wires (XA0, XA1, XA2, XB0, YB0, and YB1). 6 Lamb et al. PARAMETERS FOR BEST-FIT POWER-LAW SPECTRUM OF GRB020531. TABLE 4 Instrument WXM ∆t (s) Energy Band (keV) 0.00-1.84 0.00-0.80 1.12-1.84 2 - 25 " " FREGATE (-0.145)-(+1.31) 6-400 Scale Factor Power-Law Index 3.936+0.99 - 0.93 -- -- (8.05 ± 0.56) × 10- 2 α 1.57+0.48 - 0.44 1.15+0.80 - 0.54 3.06+2.55 - 1.27 1.26 ± 0.06 Note. -- The scale factor for the WXM fit is normalized at 1 keV, while that for the FREGATE fit is normalized at 30 keV. The WXM parameters are for a joint fit to six wires (XA0, XA1, XA2, XB0, YB0, and YB1). HETE-2 Observations of GRB020531 7 FIG. 1. -- The final rectangular HETE WXM error box for GRB020531 (dashed line). The rectangle completely encloses the Chicago 90% confidence region error circle and RIKEN 90% confidence region error rectangle (which utilyzed the same data). Note that the WXM flight location lies well inside the confidence region. Also shown is the hexagonal-shaped refined IPN error box for GRB020531 (thin solid line), determined by triangulation using the HETE FREGATE, Ulysses GRB, and Mars Odyssey HEND data for the burst. 8 Lamb et al. FIG. 2. -- FREGATE time histories of GRB020531, binned in 80 msec bins. Top to bottom: the 6-13 keV, 13-30 keV, 30-85 keV, and 85-300 keV bands. HETE-2 Observations of GRB020531 9 FIG. 3. -- Time history of GRB020531 in the WXM 2-25 keV, 2-8 keV, 8-15 keV, and 15-25 keV energy bands, binned in 80 msec bins. 10 Lamb et al. FIG. 4. -- Duration of GRB020531 versus energy. The energy bins have been chosen so that each contains approximately the same number of photons. The durations t50 and t90 increase with decreasing energy as E- 0.38 and E- 0.62, respectively. This behavior is similar to that seen in long GRBs. HETE-2 Observations of GRB020531 11 FIG. 5. -- Upper panels: Observed counts (crosses) compared to predicted counts (histogram) for the best-fit spectrum in the FREGATE energy band 6-400 keV during the first 0.00-1.25 sec of the burst. Lower panels: residuals. 12 Lamb et al. FIG. 6. -- Observed counts compared to predicted counts for the best-fit spectrum in the WXM energy band 2-25 keV for the first 1.84 seconds of the burst, and the residuals. A specific detector response matrix is calculated for each anode wire. Top panels: Observed counts (crosses) compared to predicted counts (histograms) for wires XA0, XA1, XA2, and XB0 (upper panel); residuals (lower panel). Bottom panels: Observed counts (crosses) compared to predicted counts (historams) for wires YB0, and YB1 (upper panel); residuals (lower panel). HETE-2 Observations of GRB020531 13 FIG. 7. -- Limiting magnitudes versus time for any optical afterglow from follow-up observations of GRB020531. The limiting magnitudes shown include both limits derived using small aperture, large FOV robotic telescopes (Park et al. 2002, Boer et al. 2002, West), a small aperture, modest FOV telescope (West et al. 2002), and large aperture, small FOV telescopes (Lamb et al. 2002, West et al. 2002, Miceli et al. 2002, Dullighan et al. 2002).
0710.1805
1
0710
2007-10-09T15:34:59
The dependence of the viscosity-parameter on the disk scale height profile
[ "astro-ph" ]
It is shown that the height scale for accretion disks is a constant whenever hydrostatic equilibrium and sub-sonic turbulence regime hold in the disk. In order to have a variable height scale, processes that do contribute with an extra term to the continuity equation are needed. This makes the viscosity parameter much greater in the outer region and much smaller in the inner region. Under these circumstances, turbulence is a presumable source of viscosity in the disk.
astro-ph
astro-ph
The dependence of the viscosity- parameter on the disk scale height profile Cesar Meirelles Filho & Nelson Leister Instituto de Astronomia, Geof´ısica e de Ciencias Atmosf´ericas Universidade de Sao Paulo R. do Matao, 1226, 05508-090 Sao Paulo, SP, Brasil [email protected] ABSTRACT It is shown that the height scale for accretion disks is a constant whenever hydrostatic equilibrium and sub-sonic turbulence regime hold in the disk. In order to have a variable height scale, processes that do contribute with an extra term to the continuity equation are needed. This makes the viscosity parameter much greater in the outer region and much smaller in the inner region. Under these circumstances, turbulence is a presumable source of viscosity in the disk. Subject headings: radiative transfer accretion disks - hydrodynamics, turbulence, dwarf novae, 1. Introduction In the last 35 years, accretion disks have become one of the most intense research ar- eas in theoretical astrophysics. The reason for that cannot be explained just by invoking their ubiquity in a variety of different astrophysical environment. At that point, one should recognize the role played by the Shakura & Sunyaev (1973) α standard model, which made appeal to a very intuitive, simple and neat physics. This model has, as its main assumptions, geometrical thinness, large optical thickness, hydrostatic equilibrium in z, perpendicular to the plane of the disk, and the viscosity parametrization in terms of an unknown α parameter. It was meant to apply under very specific conditions, such as those occurring in the outer parts of accretion disks in Bynary Systems, around very Young Stellar Objects, and in flows associated with the central engine for Active Galactic Nuclei (Papaloizou & Lin 1995). Soon, people realized that the approach used to treat viscosity was general and could be used under quite different conditions. Somehow this brought some ease to the long standing question -- 2 -- of viscosity in accretion disks, encouraging people to tackle related problems, even without knowing what the source of viscosity really is. So successful and profuse the applications were that, today, one speaks of, at least, four models of accretion disks: the α-standard model, suc- cessfully applied to Cataclysmic Variables, Transient X-Ray Sources, accretion disks around Active Galactic Nuclei, accretion disks in Young Stellar Objects (YSO) (Smak 1999; Lasota 2001; Menou et al. 2000; King et al. 2007; Begelman 1985; Lin 1989); the ADAF model, where the acronym stands for Advection Dominated Accretion Flows, used to explain X and γ-ray emission from underluminous X-Ray Binaries and Active Galactic Nuclei (Perna et al. 2000; Narayan & Yi 1995; Becker & Le 2003; Narayan et al. 2002); the CDAF model, Con- vection Dominated Accretion Flows, applied to underfed and radiatively inneficient hard X-Ray Binaries and Active Galactic Nuclei (Abramowicz et al. 2002; Narayan et al. 2000; Igumenshchev 2002; Igumenshchev et al. 2003; Narayan et al. 2002; Quataert & Gruzinov 2000; Ball et al. 2001); the Shapiro et al. (1976) two-temperature accretion disk model ap- plied to Cygnus X-1. Despite a remarkable difference among these models and systems to which they apply, the α-parametrization works quite well. One of the reasons for that would be the weak dependence of the disk properties on α (King et al. 2007). Besides, it seems that the spread on the values of α to fit all these systems is not large, i.e., fairly similar values of α result in reasonable agreement with observations (King et al. 2007). There are, however, some points that require more detailed treatment, and these are related to the disk scale height. By this we mean not only its value but, above all, how it behaves along the disk . These questions are of fundamental importance when one is concerned with charac- teristic time scale lengths, turbulence regimes, and criteria to choose among different energy transport models. To make this point more clear, we shall focus on the α-standard model, but our criticism applies to them all. The assumed constancy of the viscosity parameter is very decisive on the α-standard model, leading to a disk scale height that goes like r 8 , and yields the same behavior along r for both the effective temperature and the temper- 4 . As a matter of fact, it leads to a ature at the mid-plane (z = 0) of the disk, T ≈ r− constant optical depth along r. This result, apparently consistent, is made possible only by an unsound interpretation of a formal solution to the continuity equation. Using the same assumptions of the α-standard model, but the constancy of the viscosity parameter, and withdrawing the flaw by taking the correct solution to the continuity equation, the results change quite significantly. Now, α will go as r11.25, with a huge variation along the disk. If constancy of α is indeed required, one will have to look for some process that add a term in the continuity equation in such a way as to give the correct disk scale height. It is our claim in this letter that the questions we have just mentioned cast some doubts about the way the disk scale height is obtained and its implications in the determination of α. The results we have obtained in this letter do modify previous results in the area, and do modify our current knowledge about fundamental issues in accretion disk theory. Besides, they are 9 3 very important as far as full consistency is required and have not been considered so far. -- 3 -- 2. The solution to the continuity equation and the radial height scale profile In this section we are going to show how the radial scale height is related to the mass continuity equation. We also show that, in order to have a disk scale height varying with radial distance, one needs to look for processes that add an extra term to the continuity equation. Since the disk is assumed to have azymuthal symmetry, being under hydrostatic equilibrium in z-direction, the continuity equation reads, 1 r ∂ ∂ r (r ρ Vr ) = 0 , (1) with r, ρ, Vr being, respectively, radial distance, mass density and radial velocity. It should be said that the above equation neglects mass transport due to turbulence, which is equivalent to assuming sub sonic turbulence regime. Now, we set f (r , z ) = r ρ Vr and expand f in powers of z to obtain f (r , z ) = ∞ Xn=0 z2 n (2 n)!(cid:18) ∂2 n ∂ z2 n f (r , z )(cid:19)z=0 , (2) where account was taken of reflexion symmetry over z. Inserting f (r , z ) given by eq.(2) into eq.(1) gives 1 r Setting z = 0, we must have z2 n (2 n)! ∞ Xn=0 ∂ ∂ r(cid:18) ∂2 n ∂ z2 n f (r , z )(cid:19) = 0 . or ∂ ∂ r f (r 0) = 0 , f (r , 0 ) ≡ C0 , where C0 is constant. For z 6= 0, we recall that ∂ ∂ r(cid:18) ∂2 n ∂ z2 n f (r , z )(cid:19)z=0 =(cid:18) ∂2 n ∂ z2 n ∂ ∂ r f (r , z )(cid:19)z=0 , which, by eq.(1), should give Then ∂2 n ∂ z2 n ∂ ∂ r f (r , z ) = 0 . (cid:18) ∂2 n ∂ z2 n f (r , z )(cid:19)z=0 ≡ constant ≡ C2 n , (3) (4) (5) (6) (7) (8) and the most general solution to eq.(1) should be written -- 4 -- f (r , x ) = C0 C2 n C0 (2 n) ! x2n , ∞ Xn=0 where x = z ℓ , and C0 = (r ρ Vr )z=0. Finally, we may write for the accretion rate, and since this implies M = 4 π ℓ C0 C2 n C0 (2 n + 1) ! , ∞ Xn=0 ∂ ∂ r M = 0 , ∂ ∂ r ℓ = 0 , (9) (10) (11) (12) because the C2 n are all constants. We, then, must conclude that, under hydrostatic equilib- rium together with sub sonic turbulence, the height scale of the disk is not allowed to vary along the radial distance r. 3. The dependence of the viscosity parameter on the disk height scale In this section, we are going to highlight some consequences of the conclusions we have drawn in the last section. In order to proceed, let us recall some results very familiar from the accretion disks theory. Let us start from the hydrostatic equilibrium equation, ∂ ∂ z P = −ρ Ω2 z , which serves to define the disk height scale ℓ, i.e., P = ρ Ω2 ℓ2 , (13) (14) with P being the pressure, and Ω the Keplerian angular velocity. The angular momentum conservation equation may be written M Ω r = 4 π ∂ from which we obtain ρ = ∂ r(cid:0)α P ℓ2 r(cid:1) , 2 π α Ω ℓ3 S , M (15) (16) α being the viscosity parameter, and -- 5 -- S = 1 −(cid:16) r1 r (cid:17) 1 2 (17) takes into account null boundary condition for the torque at r = r1, the inner radius of the disk. Assuming the disk to be cooled by radiative transport in z direction, we may write, in the diffusion approximation, Fz = −c ∂ ∂ τ Pr , (18) with c, Fz, τ , Pr being, respectively, velocity of light, radiative flux in z direction, optical depth, and radiation pressure. Replacing differentials by finite differences, and recalling the definition of effective temperature, i.e., eq.(18) may be written as σ Tef f 4 = 3 4 π M Ω2 , 3 4 π σ M Ω2 τ = T 4 , (19) (20) with σ being the Stefan-Boltzmann constant, and T the temperature at the mid plane of the disk. In the outer parts of the disk, the opacity is mainly given by the free-free opacity.So, using a Rosseland mean opacity, averaged over z, eq.(20) will be rewritten as T 4 = 2.62 × 1026 M Ω2 ρ2 T −3.5 ℓ . Finally, for a gas pressure dominated disk, we obtain from eq.(21), α2 = 1.33 × 10−51 17 M 3 M34 x22.5 y20 , (21) (22) M17, M34, x, y are, respectively, accretion rate in units of 1017 g s−1, mass of the where central object in units of 1034 g, the radial distance in units of the inner radius, the disk scale height in units of the inner radius. The inner radius r1 is assumed to be 3 Rg, Rg being the gravitational radius. According to King et al. (2007), if constraints on α are required, one should make resource to observations of systems subject to temporal behavior, such as dwarf nova outbursts (Warner 2003; Cannizzo 2001), outbursts of X-ray transients (Lasota 2001), protostellar accretion disks (Hartmann et al. 1998), FU Orionis outbursts (Lodato & Clarke 2004), or optical variability in Active Galactic Nuclei (Starling et al. 2004). If we take, for instance, the viscous time-scale, t = −Z r1 r r ν dr , (23) where ν is the kinematic viscosity, we obtain for the standard model after little algebra, using eq.(22), -- 6 -- ℓ r = y x = 1.63 × 10−3 17 M 0.225 M34 0.325 x−0.319 d t0.25 x0.125 , (24) (25) (26) xd being the disk size in our units. Now, inserting this expression into eq.(22), it yields αs = 207.53 M34 1.375 M −0.375 17 x1.575 d t−1.25 , or, to make a comparison with Smak (1999), t = 71.39 M34 1.1 M −0.3 17 αs −0.8xd 1.25 . It should be remarked that, contrary to Smak (1999) paper, we use the disk scale height dependence on T , the temperature at z = 0, not Tef f , which we believe to be the correct procedure. Essentially, this makes our scale height greater by a factor of τ 0.125, τ being the optical depth. Our results differ from Smak (1999) by minor discrepancies in the exponents, but by a numerical factor of about two orders of magnitude greater. Finally, using our formulation, we obtain ℓ r = y x = 2.1 × 10−3 M17 M34!0.1875 t0.125 x−1 , and for α0, the value of α directly related to the viscous time, α0 = 0.1 xd −1.575 αs , (27) (28) with αs given by eq.(25). Expression (28) is the value of α at x = 1. The value of α anywhere in the disk is α = x11.25 α0 . (29) From eq.(25) and eq.(28) we see that the viscosity parameter related to the viscous time will be much smaller when we employ the correct solution to the continuity equation. At 9.675 greater than the value obtained with the the outer radius of the disk it will be 0.1 xd standard model. 4. Analysis and conclusions A rapid inspection on eq.(22) reveals a very strong dependence of α on the radial distance, due to the assumptions of sub sonic turbulence and hydrostatic equilibrium. The -- 7 -- α parametrization of the viscosity yields hardly credible results. In a disk of xd = 100, α varies by a factor of 3 × 1022 as compared to its value at x = 1. Assuming equality between the length scale of the turbulence and the height scale of the disk, since the disk is assumed to be thin, we have for y = 0.01, sub sonic turbulence only for x ≤ 3.04; and for y = 0.1, only for x ≤ 23.56. The results are highly dependent on the extent and thinness of the disk. The thicker the disk, the more sub sonic it will be. It should be stressed that the conclusions we have drawn are based on the analysis of a solution obtained under conditions that, under the usual procedure, would result in the α standard model, which gives a disk scale height varying with r 8 , a necessary condition needed to have constant α. However, the constancy of ℓ along r is not an assumption, but it stems as a consequence from a rigorous solution to the mass continuity equation. In other words, the assumption of constant α is not compatible with the solution to the mass continuity equation. To make things compatible, we should have 9 (30) 1 r ∂ ∂ r (r ρ Vr) + L(cid:16)ρ , ~V(cid:17) = 0 , ∂ r , ∂ where L = L(cid:0) ∂ ∂ z (cid:1) is an operator applied to ρ and ~V . If we insist in hydrostatic equi- librium, L describes turbulent mass transport. It is beyond our aim, in this letter, to go any further on this matter but, if a disk height scale dependence on r is essential to have physically meaningful results, a urgent search for physical processes that do contribute with an extra term to the mass transport equation is indeed required. In that respect, it is very unlikely to discard turbulent mass transport in the disk as one of the reasons to have the disk scale height varying along the radial distance, and, if so, turbulence is a presumable source of viscosity in the disk. The points we have raised in this letter deserve attention by themselves but in no way exhaust the subject. The value of the disk height scale and the way it behaves along r are decisive to know, among other things, what is the relevant energy transport mechanism in a given region and how it varies along the disk. It is our intention to to consider these issues, in a more detailed way, in a future article. The authors acknowledge support from FAPESP . Abramowicz, M.A., Igumenshchev,I.V., Quataert, E., & Narayan, R. 2002, ApJ, 565, 1101 REFERENCES Shakura, N.I., & Sunyaev, R.A. 1973, A&A, 24, 337 Ball, G.H., Narayan, R., & Quataert, E.2001, ApJ, 552, 221 -- 8 -- Becker, P., & Le, T. 2003, ApJ, 588,408 Begelman, M.C. 1985, in Astrophysics of Active Galaxies and Quasi Stellar Objects, Mill Valley, CA, University Science Books, 411, 452 Cannizzo, J.K., 2001, ApJ, 556, 847 Hartmann, L., Calvet, N., Gullbring, E., & D'Alessio, P. 1998, ApJ, 495, 385 Igumenshchev, I.V. 2002, ApJ,577, L31 Igumenshchev, I.V., Narayan, R., & Abramowicz, M.A. 2003, ApJ, 592, 1042 King, A.R., Pringle, J.E., & Livio, M. 2007, MNRAS, 376, 1740 Lasota, J.P. 2001, New Astronomic Review ,45 , 449 Lin, D.N.C. 1989, in Dynamics of Astrophysical Discs, ed. Sellwood, J.A., Cambridge Uni- versity Press, 27, 42 Lodato, C., & Clarke, C.J., 2004, MNRAS, 353, 841 Menou, K., Hameury, J.M., & Lasota, J.P. 2000, MNRAS, 314, 498 Narayan, R., & Yi, I. 1995, ApJ, 452, 710 Narayan, R., Igumenshchev, I.V., &Abramowicz, M.A. 2000, ApJ, 599, 798 Narayan, R., Quataert, E., Igumenshchev, I.V., & Abramowicz, M.A. 2002, ApJ, 577, 295 Papaloizou, J.C.B., & Lin, D.N.C. 1995, ARA&A, 33, 505 Perna, R., Raymond, J., & Narayan, R. 2000, ApJ, 541, 898 Quataert, E., & Gruzinov, A. 2000, ApJ, 539, 809 Shapiro, S., Lightman, A.P., & Eardley, D.M. 1976, ApJ, 204, 187 Smak, J. 1999, Acta Astronomica, 49, 391 Starling, R.L.C., Siemiginowska, A., Uttley, P., & Soria, R. 2004, MNRAS, 347, 67 Warner, B. 3003, Cataclysmic Variable Stars, Cambridge Univ. Press., Cambridge Yuan, F., Quataert, E., & Narayan, R. 2003, ApJ, 598, 301 This preprint was prepared with the AAS LATEX macros v5.2.
astro-ph/0406214
2
0406
2004-07-30T04:43:49
SNEWS: The SuperNova Early Warning System
[ "astro-ph" ]
This paper provides a technical description of the SuperNova Early Warning System (SNEWS), an international network of experiments with the goal of providing an early warning of a galactic supernova.
astro-ph
astro-ph
SNEWS: The SuperNova Early Warning System Pietro Antonioli‡, Richard Tresch Fienberg§, Fabrice Fleurotk, Yoshiyuki Fukuda¶, Walter Fulgione+ , Alec Habig*, Jaret Heise♯, Arthur B McDonalda , Corrinne Millsb,f , Toshio Nambac , Leif J Robinson§, Kate Scholbergd †, Michael Schwendenerk, Roger W Sinnott§, Blake Staceyd , Yoichiro Suzukic , R´eda Tafiroutkg , Carlo Vigorito+ , Brett Virene , Clarence Virtuek, and Antonino Zichichi‡ ‡ University of Bologna and INFN Bologna, Via Irnerio 46, 40126 Bologna, Italy § Sky & Telescope, 49 Bay State Rd, Cambridge, MA 02138-1200, USA k Department of Physics and Astronomy, Laurentian University, Sudbury, Ontario P3E 2C6 Canada ¶ Department of Physics, Miyagi University of Education, Sendai, Miyagi 980-0845, Japan + IFSI-CNR Torino, University of Torino and INFN Torino, Via P.Giuria 1, 10125 Torino, Italy * Department of Physics, University of Minnesota, Duluth, MN 55812-2496, USA ♯ Los Alamos National Laboratory, Los Alamos, NM 87545 a Department of Physics, Queen’s University, Kingston, Ontario K7L 3N6, Canada b Department of Physics, Boston University, Boston, MA 02215, USA c Institute for Cosmic Ray Research, University of Tokyo, Kashiwa, Chiba 277-8582, Japan d Department of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA e Brookhaven National Laboratory, P. O. Box 5000, Upton, NY 11973-5000 f Current address: Department of Physics, University of California, Santa Barbara, Santa Barbara, CA 93106 USA g Current address: Department of Physics, University of Toronto, Toronto, Ontario M5S 1A7, Canada Abstract. This paper provides a technical description of the SuperNova Early Warning System (SNEWS), an international network of experiments with the goal of providing an early warning of a galactic supernova. † To whom correspondence should be addressed E-mail: [email protected] SNEWS: The SuperNova Early Warning System 2 1. Introduction The famous supernova SN1987A in the Large Magellanic Cloud (LMC) brought the field of supernova neutrino astrophysics to life. Two water Cherenkov detectors, Kamiokande II and IMB, detected 20 events between them [1, 2, 3, 4]; two scintillator detectors, Baksan and LSD [5, 6] also reported observations. The sparse SN1987A neutrino data were sufficient to confirm the baseline model of gravitational collapse causing type II SNe and to put limits on neutrino properties (such as a ¯νe mass limit of around 20 eV.) To make distinctions between different theoretical models of core collapse and supernova explosions and to extract more information about neutrino properties, we await the more copious neutrino signal which the new generation of large neutrino experiments will detect from the next such event in our Galaxy. When the core of a massive star at the end of its life collapses, less than 1% of the gravitational binding energy of the neutron star will be released in the forms of optically visible radiation and the kinetic energy of the expanding remnant. The remainder of the binding energy is radiated in neutrinos, of which ∼1% will be electron neutrinos from an initial “neutronization” burst and the remaining 99% will be neutrinos from the later cooling reactions, roughly equally distributed among flavors. Average neutrino energies are expected to be about 13-14 MeV for νe , 14-16 MeV for ¯νe , and 20-21 MeV for all other flavors. The neutrinos are emitted over a total timescale of tens of seconds, with about half emitted during the first 1-2 seconds. Reference [7] summarizes the expected features of a core collapse neutrino signal; more recent simulation work can be found in e.g. [8, 9]. A core-collapse supernova in our Galaxy will bring a wealth of scientific information. The neutrino signal will provide information about the properties of neutrinos themselves and astrophysicists will learn about the nature of the core collapse. One unique feature of the neutrino signal is that it is prompt — neutrinos emerge on a timescale of tens of seconds, while the first electromagnetic signal may be hours or days after the stellar collapse. Therefore, neutrino observation can provide an early alert that could allow astronomers a chance to make unprecedented observations of the very early turn-on of the supernova light curve; even observations of SNe as young as a few days are rare for extra-galactic supernovae. The environment immediately surrounding the progenitor star is probed by the initial stages of the supernova. For example, any effects of a close binary companion upon the blast would occur very soon after shock breakout. UV and soft x-ray flashes are predicted at very early times. Finally, there may be entirely unexpected effects—no supernova has ever been observed very soon after its birth. Although the neutrino signal will be plentiful in practically all galactic core collapses, it is possible that there will be little or no optical fireworks (the supernova “fizzles”); the nature of any observable remnant would then be very interesting. This paper focuses on the prompt alert which is possible using the neutrino signal. We will describe the technical aspects of the system. Section 2 gives an overview of SNEWS, and Section 3 briefly covers the expected SNEWS: The SuperNova Early Warning System 3 signal in current detectors. Section 4 discusses some issues associated with SNEWS. Section 5 introduces the individual experiments’ monitors. Section 6 covers SNEWS implementation and defines the coincidence conditions and alert scheme. Section 7 describes the results of the “high-rate” system test performed in 2001. Section 8 describes the alert to the astronomical community. Section 9 gives future directions. The final section summarizes. 2. SNEWS Overview The SNEWS (SuperNova Early Warning System) collaboration is an international group of experimenters from several supernova neutrino-sensitive experiments. The primary goal of SNEWS is to provide the astronomical community with a prompt alert for a galactic supernova. An additional goal is to optimize global sensitivity to supernova neutrino physics, by such cooperative work as downtime coordination. The idea of a blind central coincidence computer receiving signals from several experiments has been around for some time (e.g. [10].) In addition to the basic early warning advantages of a neutrino detector, there are several benefits from a system involving neutrino signals from two or more different detectors. First, if the supernova is distant and only weak signals are recorded, a coincidence between signals from different detectors effectively increases the sensitivity by allowing reductions in alarm thresholds and allowing one to impose a minimum of (possibly model-dependent) expectations on the form of the signal. Second, even if a highly sensitive detector such as Super- K is online, requiring a coincidence among several detectors effectively reduces the “non-Poissonian” background present for any given detector and enormously increases the confidence in an alert.‡ Background alarms at widely separated laboratories are highly unlikely to be correlated. Without the additional confidence from coincident neutrino observations, it would be very difficult for any individual detector to provide an automated alert to astronomers. Finally, using signals from more than one detector, there is some possibility for determining the direction of the source when a single detector alone can provide no information (see reference [11].) Unfortunately triangulation is in practice quite difficult to do promptly, and cannot point as well as individual detectors. An important question for SNEWS is: how often is a galactic supernova likely to occur? Estimates vary widely, but are typically in the range of about one per 30 years (e.g. [12].) This is frequent enough to have a reasonable hope of observing one during the next five or ten years, but rare enough to mean that we must take special care not to miss anything when one occurs. The charter member experiments of SNEWS are Super-Kamiokande (Super-K) in Japan, the Sudbury Neutrino Observatory (SNO) in Canada and the Large Volume ‡ “Non-Poissonian” refers to background alarms whose rate cannot be well predicted according to a constant-background-rate Poisson distribution. Detector effects such as flashing phototubes and electronics problems fall under this category. Rates may also be locally Poissonian, just non-stationary. SNEWS: The SuperNova Early Warning System 4 Detector type scintillator water Cherenkov heavy water Material C,H H20 D20 long string water Cherenkov liquid argon high Z/neutron radio-chemical H2O Ar Pb, Fe 37Cl, 127 I, 71Ga Energy Time Point Flavor ¯νe n y y y y y ¯νe all n y NC: n νe , ¯νe y y CC: y n y n ¯νe νe y y y all n y y n n n νe Table 1. Supernova neutrino detector types and their primary capabilities. Detector (LVD) in Italy§. Representatives from AMANDA, IceCube, KamLAND, Borexino, Mini-BooNE, Icarus, OMNIS, and LIGO participate in the SNEWS Working Group, and we hope will eventually join the active coincidence. There is currently a single coincidence server, hosted by Brookhaven National Laboratory. We expect that additional machines will be deployed in the future. The BNL computer continuously runs a coincidence server process, which waits for alarm datagrams from the experiments’ clients, and provides an alert if there is a coincidence within a specified time window (10 seconds for normal running.) We have implemented a scheme of “GOLD” and “SILVER” alerts: GOLD alerts are intended for automated dissemination to the community; SILVER alerts will be disseminated among the experimenters, and require human checking. As of this writing, no inter-experiment coincidence, real or accidental, has ever occurred (except in high rate test mode), nor has any core collapse event been detected within the lifetimes of the currently active experiments. 3. The Supernova Signal and Current Detectors There are several classes of detectors capable of observing neutrinos from gravitational collapse. Most supernova neutrino detectors are designed primarily for other purposes, for proton decay searches, solar and atmospheric neutrino physics, accelerator e.g. neutrino oscillation studies, and high energy neutrino source searches. Table 1 gives a brief overview of the supernova neutrino detector types. More detailed information about supernova detection capabilities can be found in reference [14]. To summarize briefly: scintillator and water Cherenkov detectors are sensitive primarily to ¯νe ; those with neutral current capabilities (heavy water, high Z/neutron, and also water Cherenkov and scintillator to some extent) are sensitive to all flavors. Water Cherenkov and heavy water detectors have significant pointing capabilities. All except radiochemical can see neutrinos in real-time. All have energy resolution except long string water Cherenkov and radiochemical. § MACRO[13] was another charter member, and was involved with SNEWS until it turned off in 2000. SNEWS: The SuperNova Early Warning System 5 Table 2 lists specific supernova neutrino detectors and their capabilities [15, 16, 17, 18, 19, 20, 21, 22, 23]. For a summary of supernova neutrino capabilities of future detectors, please see [24, 25]. k ¶ Detector Super-K SNO LVD AMANDA Baksan Mini-BooNE KamLAND Borexino Icarus OMNIS LANNDD UNO/Hyper-K Location Japan Canada Type # of events @8.5 kpc 7000 300 450 200 Mass (kton) 32 H2O Ch. 1.4 H2O, 1 D2O scint. Italy 1 long string Meff ∼0.4/pmt Antarctica 50 scint. 0.33 Russia 200 0.7 scint. USA 300 Japan 1 scint. 100 scint. 0.3 Italy 200 Italy 2.4 liquid argon 2000 USA 2 high Z (Pb) 6000 liquid argon 70 USA USA/Japan >100,000 600-1000 H2O Ch. Status running running running running running running running 200x 200x proposed proposed proposed Table 2. Specific supernova neutrino detectors. The expected numbers of events are approximate, and refer to yields in the dominant channels. 4. The Three “P”’s In order to make the best use of a neutrino burst supernova alert, the astronomical community needs the “three P’s”: “prompt”, “pointing” and “positive”. We comment on each of these below. k Note that the currently running Super-K II (after reconstruction in 2002) has nearly the same supernova sensitivity as Super-K I; a slight increase in energy threshold due to loss of phototubes will cause only a few percent loss of total signal events. ¶ Gravitational wave detectors deserve some note here. Large interferometer experiments such as LIGO, Virgo, GEO, TAMA and ACIGA[26] as well as cryogenic antennas belonging to the IGEC collaboration[27] may have the capability of detecting gravitational wave signals from asymmetric supernova explosions (although the details of a stellar collapse gravitational wave signal are not yet well understood.) When these detectors reach maturity over the next several years, they will become an important part of a stellar collapse network, and combined neutrino and gravitational wave data will be an extremely valuable source of information for testing supernova models. The gravitational wave signal may be more prompt even than the neutrino signal, and in fact, may provide a t = 0 for a neutrino time of flight mass measurement (see e.g. [28].) The scientific potential from combined gravitational wave and neutrino signals from stellar collapse is exciting and largely unexplored territory. SNEWS: The SuperNova Early Warning System 6 4.1. “Prompt” The alert must be as prompt as possible to catch the early stages of shock breakout, which occurs within hours (or less) of core collapse. We estimate an alert dissemination time of five minutes or less for an automated (GOLD) alert. A SILVER alert involving human-checked alarms would take longer, optimistically 20 minutes or so, but perhaps longer. 4.2. “Pointing” Clearly, the more accurately we can point to a core collapse event using neutrino information, the more likely it will be that early light turn-on will be observed by astronomers. Even for the case when no directional information is available (e.g. for a single scintillator detector online) it is still useful for astronomers to know that a gravitational collapse event has occurred. However any pointing information at all is extremely valuable. The question of pointing to the supernova using the neutrino data has been examined in detail in reference [11]. There are two ways of pointing with neutrinos: first, individual detectors can make use of asymmetric reactions for which the products “remember” the direction of the incoming neutrino. Second, the timing of the neutrino signals in several detectors can be used to do triangulation. Reference [11] estimates roughly 5◦ pointing accuracy for Super-K and 20◦ pointing accuracy for SNO, given a galactic center core collapse. Triangulation is less promising, and presents practical difficulties: it requires immediate and complete exchange of event-by-event information, which is difficult in practice, and we do not plan to attempt it promptly. We do not anticipate that SNEWS will disseminate pointing information as part of the initial alert message in the short term (although this may change); this information will come from the individual experiments, and may not be available immediately. Each experiment establishes its own protocol for making estimated pointing information available. 4.3. “Positive”: There must be no false supernova alerts to the astronomical community. A single experiment cannot realistically decrease the false alert rate to zero, since there will always be some residual rate of false alerts from Poissonian and non-Poissonian sources. However, by requiring an inter-experiment coincidence, the false alert rate can be decreased to nearly zero: this is the great strength of SNEWS. We have chosen the nominal acceptable average false alert rate to be one per century. The following section is devoted to the question of ensuring a false alert rate which is sufficiently low. 4.3.1. False Alerts The fundamental motivation for the SNEWS coincidence is the reduction of false alerts. We categorize the possibilities for false alerts below: (i) Accidental Coincidences SNEWS: The SuperNova Early Warning System 7 Individual experiment rate 1/week 10 12 ) r y ( l a 10 11 v r e t n 10 10 I 10 9 10 8 10 7 10 6 10 5 10 4 10 3 10 2 10 2 3 4 5 6 7 8 9 Number of active experiments Figure 1. Average interval between accidental alerts for an n-fold coincidence of N experiments, for a 10 second coincidence window and a uniform individual experiment background alarm rate of one per week. Accidental (random) coincidences imply that there was no actual association with an astronomical event and that the coincidence occurred by chance. The rough expected rate of accidental coincidences can be calculated by assuming equal, constant, uncorrelated alarm rates for each experiment. Figure 1 shows the average interval between accidental alerts for an n-fold coincidence of N experiments, for a 10 second coincidence window and an individual experiment background alarm rate of one per week. This plot shows that an individual experiment alarm rate of one per week is acceptable only if four or fewer experiments are online, or if a three-fold coincidence if required; otherwise a lower individual experiment rate is required. Based on these considerations, the requirement for an experiment to participate in SNEWS is an average alarm rate of no more than 1 per week. We may adjust the criteria defined in this paper if more than four experiments are running. In reality, individual alarm rates are not strictly Poissonian and change with time. However, they may be Poissonian on shorter time scales, and so long as instantaneous individual rates do not exceed a certain value, the accidental coincidence rate can be made as small as desired. In Section 6.6 we detail how we deal with potentially changing alarm rates. SNEWS: The SuperNova Early Warning System 8 (ii) Non-astrophysical Correlated Bursts The possibility of some correlation between bursts seen in the individual detectors, which is not astrophysical in origin, exists to the extent that some credible coupling can be shown to exist between detectors. For participating detectors that may be physically close to one another there are a large number of possible couplings from a shared local environment (electrical noise, ambient pressure, seismic, etc..) For participating detectors that are very well separated from one another one has to invoke more fanciful and substantially less probable couplings such as solar activity, solar flares, or widescale upper atmospheric electrical disturbances. The most credible coupling for separated detectors may well be the seismic one, but even that is not really plausible. (iii) Malicious Actions A fake alert sent to the astronomical community by hackers breaking in to our machines is a remote possibility, but one which we view seriously. Breaching several machines at the individual experiments and creating false alarms at the client datagram level would have the same end effect, but would require more knowledge of the detailed and widely different operation of several detectors, so we feel this is much less of a concern. To prevent malicious actions, we take a serious approach to the security of the server and the connections to it. We require that the server be housed at a national laboratory with designated personnel to take responsibility for the security and maintenance of the machine. 4.4. Privacy Another “P” (relevant to experimenters more than to astronomers) is “Privacy”. To satisfy inter-experiment privacy needs (and in addition to help ensure secure data transmission), we have set up a formal set of rules for data sharing and have structured the collaboration around these rules. The “SNEWS subgroup” is a working group of a few people per experiment, designated by our Advisory Board (spokespeople of the active experiments.) Only subgroup members have access to the alarm data from all experiments. Subgroup members agree not to propagate information without explicit approval from the Advisory Board. 5. Description of Individual Experiments’ Supernova Triggers In this section we will describe briefly the online supernova monitoring systems of Super-K, SNO and LVD, which provide alarm input to the SNEWS coincidence. The supernova capabilities of the detectors are well known and details can be found elsewhere [14]; other details of the triggers, monitoring systems and analyses are also described elsewhere [29, 30, 31]. Although each of these three experiments takes a somewhat different approach to real-time monitoring, every one is sensitive to a galactic supernova and can provide an alarm on a timescale of minutes. SNEWS: The SuperNova Early Warning System 9 A note on SNEWS terminology: an “alarm” refers to a supernova neutrino burst candidate detected by an individual experiment, according to conditions defined for that experiment. An “alert” refers to a coincidence between alarms, and the conditions which define an alert are described in section 6. This section describes the individual experiment alarm conditions only. The detailed neutrino event information comprising the alarm bursts is not sent to the coincidence server. 5.1. Super-K Supernova Online Monitor The Super-K supernova alarm system involves software that does a prompt pre-analysis before full reconstruction. Roughly two-minute chunks of data called “subruns” are sent from the event builder via “express-line” to a dedicated supernova burst monitor machine, skipping the usual steps required before data are sent to the offline processes. The low energy trigger events are then searched for clusters in several time windows (0.5 seconds, 2 seconds, 10 seconds.) If the “pre-multiplicity” thresholds are exceeded for any of these time windows, then the “pre-candidate” is passed to a second monitor program for further analysis. Standard noise reduction algorithms similar to those applied for solar neutrino analysis [32] are applied, and the search is performed again, this time on events passing the cuts and with a higher energy threshold. Full energy reconstruction is not done (to save time), but vertices and directions are reconstructed. If any cluster passes a second pass multiplicity threshold, the multiplicity N in a 20 second time in addition, the mean distance between event positions in the window is counted; C PN −1 i=1 PN cluster, Rmean = 1 j=i+1 ~ri − ~rj , where C is the number of pairs, is calculated. Background supernova alarms arise from muon spallation events and “flashing” tubes. Both of these types of fake clusters have event vertex distributions which are highly non- uniform, and which will yield small Rmean values. Candidate clusters with sufficiently high N and R are considered to be supernova candidates. Because reconstruction of thousands of events in a real supernova burst (∼ 5000 events) could require an hour or more to fully analyze, pre-alarms are generated after 100 events if a candidate is found (roughly a 1 minute timescale.) If an alarm burst candidate is found, a datagram is sent to the SNEWS server, and shiftworkers are alerted. Detailed information about the candidate is made available to shiftworkers (present 24 hours a day onsite.) The shiftworker checks for the existence of spallation muons, examines reconstructed vertices and their goodness, and also checks the exploded view of the PMT hit pattern. A preliminary estimate of the supernova direction from elastic scattering is available at this point. If a good supernova candidate is identified, an offline process will re-analyze to provide full, precise reconstruction within a few hours. SNEWS: The SuperNova Early Warning System 10 5.2. SNO Supernova Online Monitor At SNO, custom readout electronics collect the PMT data underground and pass that information to an event builder. Built events are then sent to an event dispatcher process running on a surface computer, which is used for online monitoring. A fully detailed description will be found elsewhere [30]. Summarized here is the basic machinery of the trigger, which consists of three distinct levels which are fast and completely automated: • Level 1: This is the burst monitor which looks in the datastream for a certain number of events above a certain energy threshold within a certain time window. At present the multiplicity threshold is set to 30 events above approximately 4 MeV in a 2 second time window which provides a good sensitivity to a galactic supernova. Dynamic thresholds are used whenever calibration sources are introduced in the detector. Bursts satisfying the multiplicity criteria are written to a data file and then transferred to an analysis machine. • Level 2: At this level events are calibrated and analyzed on an event-by-event basis. The main task of this second-level trigger is to identify events with anomalous time and charge as well as events with geometric signatures of particular detector pathologies in order to cut them from the burst data set. For example, events with low charge to number of hit ratio usually indicate electrical pickup. A set of data cleaning cuts are applied which are meant to reject known instrumental background with a very high efficiency. Cherenkov events pass those cuts with very little sacrifice. • Level 3: If more than 35% of the events composing a burst survive the data cleaning cuts, an alert is sent to the SNEWS server and a dialout computer contacts the members of the SNO supernova trigger group. In the meantime a more in-depth analysis is performed to extract fitted vertices and direction cosines. The relative event fractions occurring in the D2O/H2O volumes are extracted and a search algorithm uses the events’ direction cosines to find the electron scattering (ES) events, which are expected to best convey information about the direction of the possible supernova. The Level 2 analysis produces a set of histograms which are mainly useful for quick burst diagnostic by the operator and any interested party. Besides hit and time distributions, crate/slot/channel occupancies are provided, which are expected to be flat for a supernova signal. The Level 3 analysis produces a set of histograms using fitted vertices in both the light and heavy water volumes as well as angular distribution of the events’ fitted directions. Each burst is catalogued and automatically archived on the SNO private WWW server. 5.3. LVD Supernova Online Monitor In the LVD experiment the scintillator counting rate is continuously monitored by a DAQ task, which examines all data collected in real time. A simple and fast muon rejection SNEWS: The SuperNova Early Warning System 11 algorithm makes a pre-selection of ν -candidate signals, registered by the experiment with a 12.5 ns time precision. This first selection level does not apply cuts on pulse energy and topological distribution. A separate on-line monitor task looks for burst candidates from the reduced data stream. The search algorithm is based on a pure statistical analysis of the time sequence of events including some additional cuts. The code processes the sequence in order to extract significant clusters of pulses having an expected frequency, induced by the accidental background, lower than a predefined threshold. At this level, pulse energy is required to be in the 7-100 MeV range in order to avoid fluctuation effects at the edge of energy threshold and problems due to electronic noise, as well as to reject single counter muon signals. After these cuts the background pulse frequency is found to be very stable and corresponds, for the full LVD configuration, to fB = 0.2 Hz. The resulting ν -pulse candidate time sequence, collected inside a 1000- pulse deep circular buffer, is processed by the alarm module of the monitoring code. Buffered events are processed in fixed time windows ∆T = 20 s originating at the start run time. For each asynchronous window the number Nν of contained single pulses is obtained. Then the Poissonian probability PC to have k ≥ Nν events in the cluster is ·e−λ λk , where λ = fB ∗ 20 is the mean calculated according to: PC (k ≥ Nν , λ) = P∞ k=Nν k ! expected number of pulses due to background rate. To optimize sensitivity, the online frequency fB is evaluated each time a new pulse is inserted into the buffer. The alarm threshold probability is obtained from the above expression by fixing a global alarm frequency fA . The predictive capability of the selection algorithm has been checked with real and simulated data as a function of the required global alarm frequency (fA = 1/hour, ...1/year.) Finally, to reduce the number of false alarms, for each selected candidate a topological check is applied. For a real supernova burst candidate a uniform distribution of pulses between involved counters is expected. If not, counters with abnormal high counting are excluded and the resulting cluster is re-analyzed. Surviving clusters are considered to be candidate alarms, and corresponding datagrams are sent to the SNEWS servers; all related information is saved for further analysis. Online event buffering and processing gives less than 2 minutes delay between the burst time and the alarm notification. The LVD shiftworker and experts within the collaboration are notified. 6. SNEWS Coincidence Implementation This section describes the hardware setup and software developed for SNEWS. 6.1. The Coincidence Server There is currently a single SNEWS workstation running Linux at Brookhaven National Laboratory, which serves as the coincidence server. Previously, we had had servers at LNGS and Kamioka, but moved to BNL in fall of 2003, for ease of security and SNEWS: The SuperNova Early Warning System 12 maintenance with the resources available there. The software has capabilities for dealing with multiple servers, and more may be added in the future. A second, identical machine is kept running and in synch with the primary server, so that immediate failover is possible in case of a problem with the primary server. The coincidence server remains behind the BNL firewall. Only very limited access to SNEWS subgroup members and the BNL system administrator is permitted. In addition, the server is housed in a physically secure location. If additional servers are added to the network, they will be sub ject to similar security requirements. 6.2. Coincidence Software The SNEWS software involves client and server programs which implement a simple datagram exchange via socket, employing TCP/IP protocol, and encrypted via OpenSSL. The code is designed to be easily portable to diverse operating systems. The client software is provided to the individual experiments in the form of a library of subroutines that may be called by an experiment’s supernova watch software to initiate a datagram transfer. The package also provides standalone tools for testing. The server software runs in a standalone mode, and most of the time simply waits to receive datagrams from the clients. It maintains two queues: a normal queue and a “high-rate” queue, for test alarms. When an alarm datagram is received, it is placed on a queue according to its flag (see Section 6.5.) One month’s worth of alarms are stored in the queue. Received alarms are written to disk, and are read in from disk if the server is stopped and restarted. Every time an alarm is received, the last 24 hours’ worth of alarms on the queue is searched for a coincidence. See Section 6.6 for detailed coincidence conditions. When a client initiates a connection, the server employs several layers of checks to validate the origin of the datagram. Only the IP addresses of the client machines of the involved experiments are allowed to submit packets. In addition the client and server exchange certificates which have been verified by the SNEWS Certificate Authority, and the server rejects the connection if any check fails. 6.3. SNEWS Shifts SNEWS subgroup members share shiftwork on a regular cycle. Shift duties include a check twice daily to ensure that the server is running, that network connectivity is up, and that communication capability is in good order. Individual experiment alarm rates are monitored by SNEWS subgroup members, so that any long term increase of rate over the 1/week limit may be addressed. 6.4. SNEWS Operational Modes We have established a well-defined operational mode for SNEWS, which we expect to develop in a series of managed transitions between operational modes. For instance, SNEWS: The SuperNova Early Warning System 13 new experiments or new coincidence servers will be added or removed. Each operational mode is identified by a number and the date when it came into effect, and will specify in detail the participants, the coincidence conditions, the alert classifications, and the procedures for action in case of different alarm conditions. The following sections outline the conditions for the operational mode we anticipate for the near future. 6.5. SNEWS Packet Types and Flags Each participating experiment may generate and send to the server different types of alarm datagrams. The alarm datagrams include a packet type, and a level flag. The packet type can be PING, ALARM, or RETRACTION. The level flag can be TEST, GOOD, POSSIBLE, RETRACTED or OVERRIDE. Datagrams having packet field values which do not belong to any of these categories are discarded by the server. Packet Types • PING: Ping packets are used for test purposes only and cause nothing more than a message printed to the coincidence server log. • ALARM: Alarm packets contain information about individual experiment alarms; what the server does with them depends on the level flag. • RETRACTION: Retraction packets contain information about previously sent alarms to be retracted from the server’s alarm queues. Level Flags • TEST: This flag indicates a datagram packet intended for test use as well as for any high-rate test mode. • POSSIBLE : This flag indicates an alarm generated during scheduled operations (i.e. maintenance, calibration, tests, etc.) or other known anomalous conditions. It is up to each experimental collaboration to set this flag inside the packet when appropriate. • GOOD: This flag indicates an alarm generated during normal detector conditions. • RETRACTED: This flag is set for retraction packets (note that this information is redundant—all packets of RETRACTION type will be retracted regardless of level flag.) • OVERRIDE: This flag indicates an alarm that has been confirmed as good. 6.6. Coincidence Definition The general coincidence definition implemented in the coincidence code may generate either of two types of alert: GOLD or SILVER. A GOLD alert is generated if al l of following conditions (1 through 4) are met: (i) There is a 2 or more -fold coincidence (by UT time stamp) within 10 seconds, involving at least two different experiments. (The time window refers to the maximum separation of any of the alarms in the coincidence.) SNEWS: The SuperNova Early Warning System 14 (ii) At least two of the experiments involved are at physically separated laboratories. This condition is automatically satisfied for the current operational mode. (iii) Two or more of the alarms in the coincidence are flagged as GOOD. It is the responsibility of each participant experiment to flag the alarm sent to the SNEWS server(s) appropriately. The specific criteria for GOOD/POSSIBLE alarms are locally defined by each experiment. (iv) For at least two of the experiments involved in the coincidence, the rate of good alarms for several past time intervals {Ti} = {10 minutes, 1 hour, 10 hours, 1 day, 3 days, 1 week, 1 month} preceding the first alarm of the coincidence candidate, must be consistent with the λmax =1/week requirement.+ We define the precise condition as follows: if an experiment sent {ni} alarms in each of the last intervals {Ti} before the first event of the coincidence, then the Poisson probabilities Pi for ni or more alarms in Ti , n=ni (λmaxTi )ne−λmax Ti /n!, Pi = P∞ for each interval Ti , must each be greater than Pthr = 0.5%. This corresponds to the condition that each {ni} must be less than {1, 2, 2, 3, 4, 5, 11} for the preceding intervals {Ti} for an alarm to be GOLD. When the first criterion is satisfied, but at least one of the other criteria is not satisfied, the generated alert is flagged as SILVER. In this case the alert has to be checked by the individual experiment collaborations before any public announcement. No alert will be sent to the community by SNEWS until (and if ) there is an upgrade to GOLD. 6.7. Rate-dependent GOLD Coincidence Suppression The last criterion—demotion to SILVER based on past rate history—deserves some additional discussion. The purpose of this criterion is to protect against short term rate increases from one or more experiments. The suppression is effective: see Figure 2. However it comes at a slight cost: if one assumes a constant Poisson background rate of 1 per week for all three experiments, criterion 4 will result in demotion of about 4% of true GOLD alerts to SILVER, just due to Poisson fluctuations in the previous time windows. However, the protection against unexpected increases in background rate is probably worth this small loss (note that typically individual experiment alarm rates will be less than 1 per week anyway.) One might also worry that long term rate increases might cause increased demotion of true GOLD to SILVER. We have evaluated the overall average rate increase from any single experiment that would result in 90% of true coincidences being demoted. Figure 3 shows the effect of changing Pthr . The value of Pthr chosen was 0.5%, which gives fairly + These intervals represent real time, not live time, since full live time information will not be available to the coincidence server. SNEWS: The SuperNova Early Warning System 15 Maximum allowed rate Unsuppressed coincidence rate Suppressed coincidence rate ) y r u t 2.25 n e c r e p s t r e l 1.75 a ( e t a r e c n e d 1.25 i c n i o c l a t n e 0.75 d i c c A 1.5 2 1 0.5 0.25 0 1 2 3 4 5 6 Rate increase factor for single experiment Figure 2. The green dashed line shows the maximum allowed accidental coincidence rate (1 per century.) The red line shows the expected accidental coincidence rate of 2 out of 3 experiments (for a 10 second coincidence window), under the assumption that one of the three experiences an increase in rate by a factor f (and the others maintain a 1 per week rate), as a function of rate increase factor f . The blue dotted line shows the expected coincidence rate after the past rate history suppression has been applied. low true GOLD suppression (4%); and at this threshold any overall single experiment rate increase of more than a factor of 4 will result in demotion of 90% of coincidences.∗ We feel this GOLD and SILVER scheme strikes the right balance between danger of losing true coincidences due to too-stringent criteria and danger of issuing false alerts to astronomers. 6.8. Demotion and Promotion Although we hope to avoid ever being in the situation where retraction of a GOLD alert is necessary, any experiment may reflag from GOOD (or POSSIBLE) to RETRACTED ∗ Note that long term average alarm rates will be monitored by shiftworkers, and subgroup members wil l be notified if rates of their experiments exceed the nominal 1 per week limit, so any such rate increase will be temporary. Also note that if alarm rate increases have clearly been corrected, by subgroup agreement on a case-by-case basis individual alarms may be retracted after the fact, so as not to decrease true GOLD coincidence efficiency. SNEWS: The SuperNova Early Warning System 16 Gold demotion probability threshold optimization 0.6 0.5 0.4 0.3 0.2 0.1 0 5 4 3 2 1 0 -4 10 10 -3 -2 10 Probability threshold in single interval -4 10 10 -3 -2 10 Probability threshold in single interval Figure 3. Top plot: total probability of demoting a true GOLD coincidence as a function of Pthr , assuming three experiments with 1/week alarm rate. Bottom plot: factor by which average rate of any single experiment would have to increase in order for 90% of coincidences to be demoted. The Pthr chosen is 0.005. its own alarm after data checking. The server will then automatically reevaluate and reissue the alert based on alarms in the past day of its memory: the result may be still GOLD, demotion to SILVER, or no alert at all. For the latter case, the SNEWS subgroup is notified, and a RETRACTED alert will be issued to the same mailing list as for GOLD and posted on the public web page. Experiments may also send OVERRIDE packets: a GOLD alert may also be generated if condition 1 is satisfied and at least one alarm in the coincidence is OVERRIDE and at least one is GOOD, regardless of whether the other conditions are satisfied. This case allows an override of past high-rate history demotion (or other conditions that could tag an alarm as POSSIBLE) for a human-checked alert. Figure 4 summarizes the sequence of events and GOLD vs. SILVER decisions. 7. High Rate Test Results During an approximately two-month period in April-June of 2001, Super-K, LVD and SNO subgroup members performed a “high rate test” of the coincidence software. The SNEWS: The SuperNova Early Warning System 17 Figure 4. Flowchart summarizing the sequence of events and decisions that determine whether an alert is GOLD or SILVER. purpose was two-fold: first, to check the robustness of the software and work out any remaining bugs; and second, to increase confidence in our understanding of the expected coincidence rates. The test was highly successful. The idea of the high rate test was to lower the thresholds of the experiments’ supernova monitors such that coincidence alerts increased to a non-negligible rate, due to the Poissonian nature of the data. Each experiment set its supernova monitor burst search parameters to yield a test alarm rate somewhere in the range of 10-100/day. In addition, in order to increase artificially the coincidence rate, the coincidence window was increased from its standard value of 10 seconds to 400 seconds. The individual experiment alarm and coincidence rates were somewhat non- stationary, which was not unexpected. The results were analyzed via a “time- shift” method (see below) to show that alarms were uncorrelated, and that recorded coincidences were consistent with expected rates. The alarms received as a function of time, and coincidences as a function of time, are shown in Figures 5 and 6. The numbers of individual alarms, and numbers of 2- and 3-fold coincidences are shown in Table 3. The rates are roughly constant over most periods, although there is clearly some “burstiness”. SNEWS: The SuperNova Early Warning System 18 Alarm times Super-K 10 0 April 20 SNO 10 0 April 20 LVD 10 0 April May June July May June July May June July Figure 5. Number of alarms received from the individual experiments, plotted in 11 hour bins. Experiments can be “dead” to SNEWS for many reasons: actual detector deadtime, online supernova monitor problems, network problems, or coincidence server problems. To estimate the deadtime, we used the data themselves and counted improbably long gaps as deadtime. This method automatically takes into account dead time from all causes. After removing the dead intervals, we calculated the overlapping live periods for each pair of detectors, as well as the 3-experiment overlap time. Note that there are some dead intervals common to all detectors due to network trouble; in particular the period from May 22-25 represents a problem with network availability to the test server at the Kamioka site (note that we expect very high uptime at the current BNL server site.) Based on these known alarm rates and live periods, we then calculate expected accidental coincidence rates. For the purpose of comparison with expected coincidence rates, we have calculated the “raw” number of coincidences from the individual experiment alarms arriving at the server: the number of raw coincidences is defined as the number of times the individual experiment alarms are separated by a maximum test time window of τ =400 s. Note that in the real case, according to the rate history-based demotion algorithm SNEWS: The SuperNova Early Warning System 19 Coincidence times Super-K+SNO May June July May June July Super-K+LVD SNO+LVD May June July Super-K+SNO+LVD May June July 2 1 0 April 4 2 0 April 4 2 0 April 1.5 1 0.5 0 April Figure 6. 2- and 3-fold coincidences, plotted in 11 hour bins. “Unique” coincidences only are shown. described above, the coincidence server will suppress redundant alerts. If there is more than one alarm from a given experiment within the time window, the server will send only one GOLD alert, corresponding to the coincidence of the first alarm from each experiment (assuming all conditions are satisfied) – whereas multiple “raw” coincidences would be counted for this cluster. Subsequent coincidences would be demoted to SILVER. The number of “unique” coincidences is the number of coincidences with different first alarm times. The final column of Table 3 shows both “raw” and “unique” (in parentheses) numbers of coincidences. Figure 6 shows “unique” coincidences. The coincidence server output was checked to verify consistency with the calculated raw coincidences. For stationary, uncorrelated Poisson point processes, the rate of N -fold coincidences between N detectors is given by obs ! N λcoinc = N τ N −1 Yi=1 T N where τ is the coincidence window (the maximum separation of events for a coincidence), Tobs is the total common observing time, and Ni is the number of events Ni , (1) SNEWS: The SuperNova Early Warning System 20 observed by the ith detector. For example, for a 2-fold coincidence between detectors i and j , the expected number of coincidences is 2NiNj τ /Tobs ij . The uncertainties on the expectated rate values are calculated by propagating the uncertainties on the live time. However, equation 1 is strictly valid only for stationary processes, and this assumption is clearly violated in our case (see Figure 5.) Therefore we take a different approach to calculate expected coincidence rates: to predict more generally the number of accidental coincidences from these non-stationary alarm sequences, we apply a “time- shift” method [33, 34, 35]: for any pair of detectors, we shift all of one experiment’s alarm time values by an offset ∆t, and determine the number of coincidences nc for that time offset value. This procedure is repeated for many values of ∆t; the mean and standard deviation of the distribution of nc values then gives both the expected number of observed coincidences and its expected spread, which we then compare with the observed number of raw coincidences. Similarly, we time-shift one of the three experiment’s alarm time series by ∆t to determine the expected 3-fold coincidence rate. The plot of nc versus time offset value should be flat, and show no spike at zero (or any other) offset, if there are no correlations between the different experiments’ alarm times. The results of this analysis for 2-fold coincidences are shown in Figure 7 (a similar plot, although with lower statistics, results for 3-fold coincidences.) We use time shifts ranging from -150 to 150 hours at 1000 second intervals. Live time is taken into account in the time-shifted sample by shifting the offset experiment’s live period by the same offset and then re-evaluating the overlap time. The mean and RMS values of the resulting shifted coincidence rates are used to determine the expected number of coincidences for each combination in Table 3. Table 3 shows the expected and observed numbers of events. The expected numbers of coincidences from equation 1 do not exactly match the expected numbers from the time-shift method, even considering live time estimate uncertainty. Presumably this is due to the somewhat non-stationary nature of the alarm sequence. The number of observed coincidences do match the time-shift expectations well within the expected spread. In addition, the time shift plots show no evidence of correlations between experiments, as expected. Although these somewhat non-stationary data, taken at lowered threshold, do not necessarily imply that rates will also be non-stationary when thresholds are raised and running conditions are normal, one can never be completely sure that individual experiment rates will not increase unexpectedly. This is the motivation for the rate- dependent GOLD suppression scheme of section 6.7. The coincidence server now has capability for continuous high rate testing, using tagged TEST alarms in parallel with normal alarms. SNEWS: The SuperNova Early Warning System 21 Coincidences/day versus time offset 2.5 2 1.5 1 0.5 0 4 3.5 3 2.5 2 1.5 3 2 1 -150 -100 -50 0 50 100 150 SK/SNO -150 -100 -50 0 50 100 150 SK/LVD -150 -100 -50 0 50 100 150 SNO/LVD Hours Figure 7. Rate of 2-fold coincidences for each experiment pair, as a function of time offset in hours. The rate was determined using overlap live time after the time offset. 8. The Alert to the Astronomical Community At the supernova early alert workshop of 1998, the conclusion from the astronomer working group [36] was that “the message will spread itself ” and that SNEWS will need to do no more than send out emails to as many astronomers as possible. SNEWS maintains a mailing list of interested parties, including both professionals and amateurs, to be alerted in the case of a coincidence. In an ideal case, the coincidence network provides the astronomical community with an event time and an error box on the sky at which interested observers could point their instruments. In a realistic case, the size of the error box is dependent on the location of the supernova and the experiments which are online, and may be very large (and at this time will not be available in the initial alert message.) However, members of the mailing list with wide-angle viewing capability (satellites, small telescopes) should be able to pinpoint an optical event quickly. Although an unknown fraction of galactic supernovae will be obscured by dust, many will be visible to amateurs with modest equipment. Regardless of the quality of neutrino pointing available, however, the advance warning alone gives observers of all kinds valuable time to get to their observatories and prepare to gather data as soon as an accurate position is determined. SNEWS: The SuperNova Early Warning System 22 Experiment Combination Common SK/SNO/LVD live time alarms (days) 24.1+1.1 SK/SNO −0.5 44.6+1.1 SK/LVD −0.9 27.7+0.7 SNO/LVD −0.6 SK/SNO/LVD 19.6+1.1 −0.6 334/187/- 576/-/1025 -/189/646 276/144/431 Ncoinc expected (eqn) 24.1+1.0 −2.2 122.6+4.9 −5.8 40.8+1.6 −2.0 2.9+0.3 −0.5 Ncoinc expected (shift) 24.9 ± 7.0 133.8 ± 13.7 46.4 ± 9.2 4.2 ± 2.9 Ncoinc observed raw (unique) 30 (17) 149 (112) 52 (41) 4 (4) Table 3. Alarm and coincidence summary. The first column indicates the 2 or 3-fold experiment combination. The second column gives the overlap live time for that combination with estimated uncertainties. The next column gives the numbers of alarms from the experiments which are within the overlap live time. The fourth column gives the number of expected coincidences according to equation 1; the error reported is the systematic error only, from uncertainty in the live time estimate. The fifth column gives the expectation and its RMS based on the time shift method. The final column gives the observed number of coincidences during the test; the number outside parentheses indicates the “raw” number of coincidences defined as the number of alarms with maximum time separation of 400 seconds (to be compared to the predictions); the number in parentheses is the number of “unique” coincidences tagged by the coincidence server (see text.) A Target of Opportunity proposal for the Hubble Space Telescope, “Observing the Next Nearby Supernova”, aiming to take advantage of early supernova light based on an early warning, was approved [37] for Cycle 13 and was operational for Cycles 8 through 12. 8.1. Amateur Astronomers The large pool of skilled and well-equipped amateur astronomers is also prepared to help locate a nearby supernova. The editors of Sky & Telescope magazine have set up a clearinghouse for amateur observers in search for first light (and a precise optical position as early as possible) [38], via their AstroAlert service [39]. This was started by former editor-in-chief Leif Robinson, and has the continued support of current editor- in-chief Rick Fienberg. In collaboration with the American Association of Variable Star Observers, they have developed a set of criteria for evaluating amateur responses to an alert, so that a reliable precise position can be disseminated as early as possible. For instance: there must be at least two consistent reports, demonstrated lack of motion, lack of identification with known asteroid and variable star databases, variability consistent with supernova light curves and, if the information is available, a spectrum consistent with known supernova types. On February 14 2003, Sky & Telescope performed a test for amateurs. A transient target (the asteroid Vesta at a near-stationary point in its retrograde loop) was selected, which at the time was about magnitude 6.7. Sky & Telescope issued an alert (very carefully tagged as a test) to their mailing list, with a given 13-degree uncertainty SNEWS: The SuperNova Early Warning System 23 radius. They received 83 responses via the web response form, and more by email. The responses were of world-wide distribution, and although many observers experienced poor conditions, six were successful in identifying the target. From this experience, they have suggested refinements to optimize amateur astronomer strategy. A second test is planned soon, and should be a regular occurrence. 8.2. SNEWS Alerts We maintain two alert mailing lists which will be sent to automatically by the SNEWS coincidence software in the case of an alert. The first is the GOLD alert list, which includes all astronomers who have signed up, including Sky & Telescope and the HST astronomers, and is to be an automated alert. The second mailing list will be for SILVER alerts, and is to be sent to neutrino experimenters only. These alerts will be checked out by shiftworkers at their respective experiments before an alert is issued; each experiment is responsible for making sure the SILVER alert messages reach shiftworkers. Each experimental collaboration defines its own protocol for acting on a SNEWS SILVER or GOLD alert. For both SILVER and GOLD cases, a message containing the following information: • UTC time of the coincidence, • all detectors involved in the coincidence, and • the types of alarms (GOOD, POSSIBLE) for each experiment involved in the coincidence will be automatically sent by the server to the SNEWS subgroup members. The information may also be posted to a restricted SNEWS subgroup page for SILVER, and a public page for GOLD. To allow the confirmation of a SNEWS alert as really coming from SNEWS, any alerts will be public key signed using the SNEWS key. This key has the ID# 68DF93F7, and is available on the network of public PGP keyservers such as http://pgp.mit.edu/ Note that there is no restriction on individual experiments making any announcement based on individual observation in the case of absence of a SNEWS alert, SILVER or GOLD, or preceding or following any SNEWS alert message. Any individual experiment may publicly announce a supposed supernova signal following a dispatched SILVER alert which has not yet been upgraded to GOLD. In this case the information that a previous SILVER alert from the SNEWS server(s) has been received should be cited. 9. Status and Future Prospects At the time of this writing, SILVER alerts only between Super-K and LVD are activated. We are working towards having the operational mode described in this paper to be SNEWS: The SuperNova Early Warning System 24 activated in the very short term, comprising automated GOOD alarms from Super- K and LVD, but automated POSSIBLE alarms only from SNO, such that SNO will participate in a GOLD alert only if at least two other experiments’ GOOD alarms are present. We also expect SNEWS to incorporate more galactic-supernova-sensitive neutrino detectors over the next few years. In addition, we may expand the network of servers with additional secure sites. 10. Summary In summary, several supernova neutrino detectors are now online. If a stellar core collapse occurs in our Galaxy, these detectors will record signals from which a wealth of physical and astrophysical information can be mined. An early alert of a gravitational collapse occurrence is essential to give astronomers the best chance possible of observing the physically interesting and previously poorly observed early turn-on of the supernova light curve. A coincidence of several neutrino experiments is a very powerful technique for reducing “non-Poissonian” false alarms to the astronomical community, in order to allow a prompt alarm. We have implemented such a system, currently incorporating several running detectors: LVD, SNO and Super- K. We expect to expand the network in the near future, and move to a more automated mode in the near future. Acknowledgments We would like to thank the members of the Super-K, SNO and LVD experimental collaborations, without whose hard work the SNEWS pro ject would not be possible. We thank Sam Aronson and Tom Schlagel of Brookhaven National Laboratory for providing a secure home for the server. We are grateful to John Bahcall, Barry Barish, John Beacom, Janet Conrad, Ed Kearns, Janet Mattei, Jim Stone, and Larry Sulak for advice and support; others from the early days of SNEWS are Adam Burrows, Robert Kirschner, and Mark Vagins. We also thank the members of the SNEWS working group from other experiments. SNEWS is funded by National Science Foundation awards PHY-0303196 and PHY-0302166. References [1] Hirata K S et al. 1987 Phys. Rev. Lett. 58 1490 [2] Bionta R M et al. 1987 Phys. Rev. Lett. 58 1494 [3] Bratton C B et al. 1988 Phys. Rev. D 37 3361 [4] Hirata K S et al. 1988 Phys. Rev. D 38 448 [5] Alekseev E N et al. 1987 JETP. Lett. 45 589 [6] Aglietta M et al. 1987 Europhys. Lett. 3 1315 [7] Burrows A et al. 1992 Phys. Rev. D 45 3361 [8] Thompson T et al. 2003 Astrophys. J. 592 434 SNEWS: The SuperNova Early Warning System 25 [9] Liebendorfer M et al. Preprint astro-ph/0207036 [10] Cline D B 1990 Proceedings of the Supernova Watch Workshop (Santa Monica) [11] Beacom J and Vogel P 1999 Phys. Rev. D 60 033007 (Beacom J and Vogel P 1999 Preprint astro-ph/9811350) [12] Tammann G A et al. 1994 Ap. J. Supp. 92 487 [13] Ahlen S et al. 1992 Astropart. Phys. 1 11 Ambrosio M et al. 1998 Astropart. Phys. 8 123-133 [14] Scholberg K 2000 Nucl. Phys. Proc. Suppl. 91 331 (Scholberg K 2000 Preprint hep-ex/0008044) [15] Aglietta M et al. 1992 Nuov. Cim. A 105, 1793 [16] Fukuda S et al. 2003 Nucl. Instrum. Meth. A 501 418 [17] Boger J et al. 2000 Nucl. Instrum. Meth. A 449 172 (Boger J et al. 2000 Preprint nucl-ex/9910016) [18] Jacobsen J E, Halzen F and Zas E 1994 Phys. Rev. D 49 1758 Jacobsen J E, Halzen F and Zas E 1996 Phys. Rev. D 53 7359 [19] Bueno A, Gil-Botella I and Rubbia A 2003 Preprint hep-ph/0307222 [20] Sharp M et al. 2002 Phys. Rev. D66 013012 (Sharp M et al. 2002 Preprint hep-ph/0205035) [21] Cadonati L et al. 2002 Astropart. Phys. 16 361 (Cadonati L et al. 2002 Preprint hep-ph/0012082) [22] R. N. Boyd et al. 2003 Nucl. Phys. A 718 222 [23] C. K. Jung 2000 Preprint hep-ex/0005046 [24] Homestake Collaboration 2003 NUSEL Science Book Preprint nucl-ex/0308018 [25] Scholberg K 2003 http://mocha.phys.washington.edu/∼int talk/WorkShops/Neutrino02/ Working Groups/People/Scholberg K/scholberg thurs solarstellarwg.pdf [26] Hughes S A et al. 2001 Preprint astro-ph/0110349 [27] Astone P et al. 2003 Phys. Rev. D 68 022001 (Astone P et al. 2003 Preprint astro-ph/0302482) [28] Arnaud N et al. 2000 Phys. Rev. D 65 033010 (Arnaud N al. 2000 Preprint hep-ph/0109027) [29] Fukuda Y 2003 Proc. of Origin of Matter and Evolution of Galaxies 2000 (World Scientific) 209 Super-K collaboration, in preparation [30] Virtue C 2001 Nucl. Phys. Proc. Suppl. 100 326 (Virtue C 2001 Preprint astro-ph/0103324) SNO collaboration, in preparation [31] Fulgione W 1999 Nucl. Phys. Proc. Suppl. 70 469 [32] Fukuda S et al. 2001 Phys. Rev. Lett. 86 5651 (Fukuda S et al. 2001 Preprint hep-ex/0103032) [33] Amaldi E et al. 1989 Astron. Astrophys. 216 325 [34] Prodi G A et al. 2000 Int. J. Mod. Phys. D 9 237 (Prodi G A et al. 2000 Preprint astro-ph/0003106) [35] Allen Z A et al. 2000 Phys. Rev. Lett. 85 5046 (Allen Z A et al. 2000 Preprint astro-ph/0007308) [36] http://hep.bu.edu/∼schol/workshop/workshop summary.html [37] Bahcall J (P.I.) 2003 Observing the next nearby supernova HST proposal http://presto.stsci.edu/public/propinfo.html [38] Robinson L J August 1999 Sky & Telescope 30 [39] http://SkyandTelescope.com/observing/proamcollab/astroalert/ 10264
0711.0779
2
0711
2008-01-10T20:22:02
Nonlinear Evolution of Anisotropic Cosmological Power
[ "astro-ph" ]
There has been growing interest in the possibility of testing more precisely the assumption of statistical isotropy of primordial density perturbations. If it is to be tested with galaxy surveys at distance scales <~ 10 Mpc, then nonlinear evolution of anisotropic power must be understood. To this end, we calculate the angular dependence of the power spectrum to third order in perturbation theory for a primordial power spectrum with a quadrupole dependence on the wavevector direction. Our results suggest that primordial power anisotropies will be suppressed by <~ 7% in the quasilinear regime. We also show that the skewness in the statistically anisotropic theory differs by no more than 1% from that in the isotropic theory.
astro-ph
astro-ph
Nonlinear Evolution of Anisotropic Cosmological Power Shin'ichiro Ando and Marc Kamionkowski California Institute of Technology, Mail Code 130-33, Pasadena, CA 91125 (Dated: November 5, 2007; revised January 8, 2008; accepted January 10, 2008) There has been growing interest in the possibility of testing more precisely the assumption of statistical isotropy of primordial density perturbations. If it is to be tested with galaxy surveys at distance scales < ∼ 10 Mpc, then nonlinear evolution of anisotropic power must be understood. To this end, we calculate the angular dependence of the power spectrum to third order in perturbation theory for a primordial power spectrum with a quadrupole dependence on the wavevector direction. Our results suggest that primordial power anisotropies will be suppressed by < ∼ 7% in the quasilinear regime. We also show that the skewness in the statistically anisotropic theory differs by no more than 1% from that in the isotropic theory. PACS numbers: 98.80.-k It is commonly assumed that primordial density per- turbations are statistically isotropic, and statistical isotropy is a prediction of most structure-formation theo- ries. The notion of statistical isotropy can be quantified, though, by considering models in which it is broken, and a growing literature has discussed physical models that produce primordial perturbations that are not statisti- cally isotropic [1]. The manifestations of departures from statistical isotropy can take on many forms, but one simple ex- ample [2] predicts a primordial power spectrum P (k) with a quadrupole dependence on the direction k of the wavevector k, P (k) = A(k) [1 + g∗P2(µk)] , (1) where P2(x) = (3x2 − 1)/2 is the second Legendre poly- nomial, µk = k · e, and e is a preferred direction.1 Refer- ence [3] constructed a CMB minimum-variance estimator for the power-anisotropy amplitude g∗ and showed that the Planck satellite will be sensitive to a value of g∗ as small as ∼0.02, a number which can be estimated roughly by ∼10 N −1/2 , where Npix is the number of statistically independent pixels on the sky detected by Planck. pix Similar arguments suggest that the sensitivity of a galaxy survey, like the Sloan Digital Sky Survey [4] or Two-Degree Field [5], should have a comparable sensitiv- ity. One issue that will arise, though, in testing statisti- cal isotropy of primordial perturbations is the quasilinear evolution of the power spectrum. The root-mean-square density-perturbation amplitude becomes of order unity at distance scales ∼10 h−1 Mpc, and so quasilinear evo- lution of the density field must be taken into account if the mass distribution measured on these scales in the Universe today is to be used to infer the primordial mass distribution. 1 Note that our g∗ is 3/2 times that in Ref. [2]. In this paper, we study the nonlinear evolution of den- sity perturbations to see how nonlinearity affects statisti- cal isotropy. Does nonlinear evolution amplify departures from statistical isotropy? Or possibly suppress them? To take the first steps to address this question, we calculate the power spectrum, to third order in perturbation the- ory, under the assumption that primordial perturbations have the form given in Eq. (1). The bottom line is that quasilinear evolution suppresses departures from statisti- cal isotropy, but by only < ∼ 7% compared with the linear theory. Thus, galaxy surveys in the linear/quasilinear regime will still be useful diagnostics for departures from statistical isotropy. We also calculate skewness, finding that the quadrupole power anisotropy changes it by no more than 1%. To proceed, we use third-order perturbation theory to determine whether a primordial power anisotropy is amplified or suppressed in the quasilinear regime. The power spectrum P (k, z) is defined by the ensemble aver- age of the two-point correlation of the Fourier transform δ(k, z) of the density perturbation through hδ(k1, z)δ(k2, z)i = (2π)3δD(k1 + k2)P (k1, z), (2) where δD is the Dirac delta function. The density per- turbations can be expanded in terms of the linear-theory density-perturbation amplitude, which has a redshift de- pendence proportional to the linear growth factor D(z), δ(k, z) = ∞ Xn=1 δn(k)Dn(z). (3) In the linear regime, each Fourier mode grows at the same rate, and so the linear-theory power spectrum Plin(k), corresponding to the linear density perturbation δ1(k), evolves in such a way that the quadrupole dependence of the primordial power spectrum [Eq. (1)] is preserved. However, the nonlinear power spectrum will have differ- ent anisotropy structure, and we shall investigate it fol- lowing prescription developed in Refs. [6, 7, 8]. Reference [6] showed that solution of the nonrelativis- tic fluid equations (i.e., continuity, Euler, and Poisson equations) allows one to write the nth-order density- perturbation amplitude δn in terms of the linear pertur- bation δ1 through where the sum on n is restricted to even positive inte- gers. To the next-to-leading order, the expansion can be written, 2 δn(k) = Z dq1 (2π)3 · · ·Z dqn × (2π)3Fn(q1, · · · , qn)δ1(q1) · · · δ1(qn). (2π)3 δD(q1 + · · · + qn − k) (4) For any n, the function Fn can be obtained by recursive relations given in Ref. [6]. In particular, the expressions for F2 and F3 that are directly relevant for our purpose are explicitly given by Eqs. (A2) and (A3) of Ref. [6].2 As all odd moments of δ1(k) vanish, the power spectrum to third order is given as (2π)3δD(k1 + k2)P (2)(k1, z) = D2(z)hδ1(k1)δ1(k2)i + D4(z) [hδ2(k1)δ2(k2)i +hδ1(k1)δ3(k2)i + hδ3(k1)δ1(k2)i] . (5) The second term, which evolves as D4(z), is the next- to-leading term evaluated to the third order of density perturbation δ3, while the first term is the linear part. We further define a quantity Pmn as hδm(k1)δn(k2)i = (2π)3δD(k1 + k2)Pmn(k1), (6) and with this definition, Eq. (5) is rewritten as P (2)(k, z) = D2(z)P11(k)+D4(z)[P22(k)+2P13(k)]. (7) We then need to write P22 and P13 in terms of the linear spectrum Plin = P11. This procedure is straight- forward, and the results are very similar to those given in Ref. [8]: P22(k) = 2Z dq (2π)3 hF (s) 2 (q, k − q)i2 (8) 3 (q, −q, k)Plin(q), (9) P13(k) = 3Plin(k)Z × Plin(q)Plin(k − q), dq (2π)3 F (s) where the symmetrized function F (s) n is obtained by sum- ming n! permutations of Fn over its n arguments and di- viding by n!. The only difference from the expressions in Ref. [8] is that the linear power spectrum now depends on both magnitude k and direction k of wavevector k. In order to illuminate the anisotropy structure, we ex- pand Eqs. (8) and (9) with Legendre polynomials. The most general power spectrum today (D = 1) can be ex- panded in Legendre polynomials Pn(x) as P (k) = Plin(k, µk) + ∞ Xm,n=0 gm ∗ Pn(µk)Bmn(k), (10) 2 Strictly speaking, F2 in Ref. [6] is for an Einstein-de Sitter Uni- verse, but Ref. [9] shows that its form in a ΛCDM model with Ωm ≃ 0.3 differs by less than 1%; the same should be true for F3. P (2)(k) = Alin(k) + B00(k) + g2 ∗B20(k) +(cid:8)g∗[Alin(k) + B12(k)] + g2 ∗B24(k)P4(k), + g2 ∗B22(k)(cid:9) P2(k) (11) where Alin, defined by Eq. (1), is the isotropic linear power spectrum. Expressions for the expansion coeffi- cients Bmn(k) are given at the end of the paper. We then rewrite Eq. (11) as P (2)(k) = A(2)(k)h1 + g(2) 4 (k)P4(µk)i , +g(2) 2 (k)P2(µk) g(2) 2 (k) = g∗c1(k) + g2 g(2) 4 (k) = g2 ∗c3(k), ∗c2(k), (12) (13) (14) with the following definitions of A(2), c1, c2, and c3: A(2)(k) = Alin(k) + B00(k) + g2 ∗B20(k), (15) A(2)(k)ci(k) =   Alin(k) + B12(k) [i = 1] [i = 2] [i = 3] B22(k) B24(k) . (16) We note that the primordial quadrupole anisotropy also affects the isotropic part of the next-to-leading order power spectrum, A(2)(k). While this is interesting, that correction is expected to be very small, being suppressed by g2 ∗. Thus, hereafter we neglect this term; keeping it only gives correction of the order of g3 2 and g(2) 4 , respectively. For the linear power spectrum Alin(k), we use the fit- ting formula for the transfer function given in Ref. [10] with current values for the relevant cosmological param- eters [11]. ∗ to g(2) ∗ and g4 If g∗ ≪ 1, we may neglect all the terms propor- tional to g2 ∗, and thus c1(k) is the only quantity relevant for anisotropy structure. As linear theory simply gives c1(k) = 1, the quantity c1(k) − 1 represents enhancement of the quadrupole anisotropy due to quasilinear evolu- tion. In Fig. 1, we plot c1(k) − 1. This figure shows that at large scales k < ∼ 10−2 Mpc−1, the nonlinear effect is subdominant and thus anisotropic structure is the same as that for the linear theory: c1(k) ≈ 1. The quadrupole anisotropy then decreases at smaller scales and it reaches minimum (c1 − 1 ≃ −0.07) in the quasilinear regime, k ∼ 0.1 Mpc−1. Third-order perturbation theory becomes less accurate for even higher wavenumbers, k > ∼ 0.1 Mpc−1, at z = 0. However, it remains very accurate at such (comoving) scales in the high-redshift Universe, z ≥ 1 [12]. For gen- eral redshifts z, the anisotropic power spectrum is given by Eqs. (12) -- (16) with replacements Alin → D2(z)Alin 3 rameters. Strictly speaking, the validity of third-order perturbation theory breaks down when the terms higher order in ∆2(k) become important; in this case, one will need to go to fourth-order or even higher-order correc- tions to obtain the evolved power spectrum. We now compute the skewness S3 ≡ (cid:10)δ3(cid:11) /(cid:10)δ2(cid:11)2 from the anisotropic primordial power spectrum, to second or- der in perturbation theory. The standard result, for an Einstein-de Sitter Universe, is S3 = 34/7 [13, 14], and the result for a ΛCDM Universe differs by less than 1%. The three-point correlation function at zero lag is a Fourier transform of hδ(k1, z)δ(k2, z)δ(k3, z)i, and thus the lead- ing contribution comes from D4(z)hδ1(k1)δ1(k2)δ2(k3)i, etc. Rewriting δ2 in terms of the linear fluctuation δ1 and the integration kernel F2, and using the definition of linear power spectrum, we obtain, hδ3(z)i = 6D4(z)Z dk1 (2π)3 Z dk2 (2π)3 F (s) 2 (k1, k2) × Plin(k1)Plin(k2). (20) After integrating over the directions of wavevectors, k1 and k2, we find hδ3(z)i = D4(z)(cid:18) 34 7 + ∗ 8g2 175(cid:19)(cid:20)Z dkk2 2π2 Alin(k)(cid:21)2 . (21) On the other hand, the variance hδ2i can be computed in a similar manner, and to the same order, we obtain hδ2(z)i = D2(z)Z dk (2π)3 Plin(k) = D2(z)Z dkk2 2π2 Alin(k). (22) Therefore, we find the skewness to be S3 = hδ3(z)i hδ2(z)i2 = 34 7 + ∗ 8g2 175 . (23) The requirement that the power spectrum be positive definite restricts the value 0 < g∗ < 1, and so the largest possible deviation from the isotropic value of 34/7 is less than 1%. Thus, the skewness will not be an effective discriminant for anisotropic power. To conclude, we used third-order perturbation the- ory to calculate the quasilinear evolution of a primordial power spectrum. Our results show that nonlinear evo- lution suppresses the primordial anisotropy. The sup- pression is ∼7% at k = 0.1 Mpc−1 and z = 0 compared with the linear theory. This must be taken into account when interpreting the result of searches in the quasilin- ear regime for primordial anisotropy, and it indicates that nonlinear scales are still valuable as tests for the isotropy of primordial power, since the suppression is weak. We also found that the standard prediction for the skewness is little changed if the primordial power spectrum has a quadrupole anisotropy. Our perturbative results seem to indicate that a quadrupole anisotropy is enhanced at smaller scales, but FIG. 1: Coefficients of anisotropic terms in the third-order power spectrum, c1(k) − 1, c2(k), and c3(k). The definitions are given in Eqs. (12) -- (14). Solid curves are the result of nu- merical integration, while dotted curves are the fitting func- tions (17) -- (19). and Bmn → D4(z)Bmn. The enhancement of anisotropy is then given by c1(k, z) − 1, which is well approxi- mated by [c1(k)−1]D2(z) in the quasilinear regime where Alin > Bmn. It is −0.03 at k = 0.1 Mpc−1 and z = 1; the suppression is weaker. One can also see amplified anisotropy for k > ∼ 0.3 Mpc−1 at redshifts when the third- order approach is still valid at such scales. The corrections on the order of g2 ∗ are represented by c2(k) and c3(k), both of which are also shown in Fig. 1. Besides it is suppressed by additional g∗, c2(k) is intrin- sically smaller than c1 − 1, thus giving only a minor cor- rection. While possessing different anisotropy structure, the higher multipole term proportional to c3(k) would also be small in the quasilinear regime. The results of our computations can be approximated by c1(k) = 1 − 0.463∆2(k) + 0.886∆4(k) − 0.407∆6(k),(17) c2(k) = −0.133∆2(k) + 0.195∆4(k) − 0.079∆6(k),(18) c3(k) = −0.163∆2(k) + 1.062∆4(k) − 1.082∆6(k) + 0.373∆8(k), (19) in terms of the linear-theory density-perturbation ampli- tude at wavenumber k, ∆2(k) = (k3/2π2)Alin(k). These fitting functions are plotted as dotted curves in Fig. 1. They were obtained explicitly for the current best-fit cos- mological parameters [11], but when written in terms of ∆2(k), should also be accurate for other cosmological pa- this enhancement cannot be trusted at low redshifts, as it occurs at scales where our perturbative approach breaks down. It seems counterintuitive to think that growth of perturbations in the highly nonlinear regime can act to enhance the primordial anisotropy, but it will require an N-body simulation to be sure. Likewise, we imagine that departures from statistical isotropy that are higher order in angle (octupole, etc.) will also be suppressed by quasi- linear evolution, but we leave that calculation for future work. Before closing, we provide explicit expressions for the quantities in Eq. (10). For the numerical evaluation of Eqs. (8) and (9), we shall choose k along the z-direction, k = (0, 0, 1), and the preferred direction e in the x-z plane, e = (p1 − µ2 k, 0, µk). We then use the spheri- cal coordinate for q = q(p1 − µ2 cos φ,p1 − µ2 sin φ, µ), and dq = q2dqdµdφ. With this notation, Bmn in Eq. (10) are given as Bmn = Smn + 2Tmn, (24) Smn = Tmn = 0 k4 dµ dqZ 1 [7kµ + q(3 − 10µ2)]2 (k2 + q2 − 2kqµ)3 Imn π2 Z ∞ × Alin(q)Alin(cid:16)pk2 + q2 − 2kqµ(cid:17) , −1 π2 k2Alin(k) Z ∞ dqAlin(q)(cid:20)Jmn + Kmn(q2 − k2)3 ln(cid:18) k + q k − q(cid:19)(cid:21), 0 (25) (26) where Smn and Tmn correspond to P22 and P13, respec- tively. Explicit forms of Imn, Jmn, and Kmn are K00 = K12 = K20 = = 7q2 + 2k2 672k5q3 , 6k6 + 41k4q2 + 76k2q4 + 21q6 5376k7q5 7 18 K24 K22 = 7 10 6k6 + 25k4q2 + 20k2q4 + 21q6 26880k7q5 4 (35) (36) (37) , , and all the other components vanish. Note that T00 is the same as Eq. (19) of Ref. [8]. MK acknowledges the hospitality of the Aspen Cen- ter for Physics, where part of this work was completed. This work was supported by the Sherman Fairchild Foundation (SA), DoE DE-FG03-92-ER40701, NASA NNG05GF69G, and the Gordon and Betty Moore Foun- dation (MK). [1] C. Gordon, W. Hu, D. Huterer and T. Crawford, Phys. Rev. D 72, 103002 (2005); J. G. Cresswell, A. R. Liddle, P. Mukherjee and A. Riazuelo, Phys. Rev. D 73, 041302 (2006); S. H. S. Alexander, hep-th/0601034; A. Berera, R. V. Buniy and T. W. Kephart, JCAP 0410, 016 (2004); R. V. Buniy, A. Berera and T. W. Kephart, Phys. Rev. D 73, 063529 (2006); G. V. Chibisov and Yu. V. Shtanov, Int. J. Mod. Phys. A 5, 2625 (1990); G. V. Chibisov and Yu. V. Shtanov, Sov. Phys. JETP 69, 17 (1989) [Zh. Eksp. Teor. Fiz. 96, 32 (1989)]; A. E. Gumrukcuoglu, C. R. Contaldi and M. Peloso, JCAP 0711, 005 (2007); C. Armendariz-Picon, JCAP 0709, 014 (2007); J. F. Donoghue, K. Dutta and A. Ross, astro-ph/0703455; R. A. Battye and A. Moss, Phys. Rev. D 74, 041301 (2006); C. G. Bohmer and D. F. Mota, arXiv:0710.2003 [astro-ph]; T. S. Pereira, C. Pitrou and J. P. Uzan, JCAP 0709, 006 (2007). [2] L. Ackerman, S. M. Carroll and M. B. Wise, Phys. Rev. D 75, 083502 (2007). [3] A. R. Pullen and M. Kamionkowski, Phys. Rev. D 76, 103529 (2007). [4] http://www.sdss.org [5] http://www.aao.gov.au/2df [6] M. H. Goroff, B. Grinstein, S. J. Rey and M. B. Wise, Astrophys. J. 311, 6 (1986). [7] N. Makino, M. Sasaki and Y. Suto, Phys. Rev. D 46, 585 (1992). [8] B. Jain and E. Bertschinger, Astrophys. J. 431, 495 (1994). I00 = I12 = I20 = I22 = I24 = J00 = J12 = k2 + q2 − 2kqµ 392 , 2q2(3µ2 − 1) + (k2 − 2kqµ)(3µ2 + 1) 2q2 − 4kqµ + k2(3µ2 − 1) 784 3920 , (k2 + q2)(3µ2 − 1) − kqµ(3µ2 + 1) 2744 (27) , (28) (29) (30) , 9 1 54880 (cid:2)4k2(3µ2 − 1) + q2(35µ4 − 30µ2 + 3) −8kqµ(5µ2 − 3)(cid:3) , k2 q2 1008 (cid:18)6 k4(cid:19) , q2 − 79 + 50 k2 − 21 k2 40320 (cid:18)90 q2 − 5006 + 1732 q2 k2 q4 1 (31) (32) k4 q4 + 375 k6(cid:19) , q6 q4 k4 − 315 −300 J20 = 7 10 J22 = 7 18 J24 = 1 201600 (cid:18)90 k4 q4 + 135 k6(cid:19) , q6 q4 k4 − 315 +540 k2 q2 − 1846 − 268 q2 k2 (33) [9] M. Kamionkowski and A. Buchalter, Astrophys. J. 514, 7 (1999). [10] D. J. Eisenstein and W. Hu, Astrophys. J. 511, 5 (1999). [11] D. N. Spergel et al., Astrophys. J. Suppl. 170, 377 (2007). [12] D. Jeong and E. Komatsu, Astrophys. J. 651, 619 (2006). [13] P. J. E. Peebles, The Large-Scale Structure of the Uni- verse (Princeton Univ. Press, Princeton, 1980). (34) [14] J. N. Fry, Astrophys. J. 279, 499 (1984).
astro-ph/0603549
1
0603
2006-03-20T23:49:47
Linear radio structures in selected Seyfert and LINER galaxies
[ "astro-ph" ]
High resolution MERLIN 5 GHz observations (0.04") of 7 Seyfert galaxies, selected as the ones previously showing evidence of collimated ejection, have been compared with high resolution archive HST data. The radio maps reveal rich structures in all the galaxies. NGC 2639 and TXFS 2226-184 have multiple knot parsec-scale extended structures, Mrk 1034, Mrk 1210, NGC 4922C and NGC 5506 reveal one-sided jets, while IC 1481 exhibits a jet-like extension. The close correlation between the radio-emitting relativistic plasma and the ionized gas in the inner regions of these galaxies allows us to study in detail the physics close to the center of low luminosity AGN.
astro-ph
astro-ph
Linear radio structures in selected Seyfert and LINER galaxies E. Xanthopoulos∗,†, A. H. Thean ∗∗, A. Pedlar ∗∗ and A. M. S. Richards ∗∗ ∗University of California Davis, Department of Physics, Davis, CA 95616 †IGPP/Lawrence Livermore National Laboratory, Livermore, CA 94550 ∗∗University of Manchester, Jodrell Bank Observatory, Macclesfield, Cheshire SK11 9DL, England Abstract. High resolution MERLIN 5 GHz observations (0 ′′. 04) of 7 Seyfert galaxies, selected as the ones previously showing evidence of collimated ejection, have been compared with high resolution archive HST data. The radio maps reveal rich structures in all the galaxies. NGC 2639 and TXFS 2226-184 have multiple knot parsec-scale extended structures, Mrk 1034, Mrk 1210, NGC 4922C and NGC 5506 reveal one-sided jets, while IC 1481 exhibits a jet-like extension. The close correlation between the radio-emitting relativistic plasma and the ionized gas in the inner regions of these galaxies allows us to study in detail the physics close to the center of low luminosity AGN. Keywords: Seyfert and LINER galaxies, Radio observations, Jets PACS: 98.54.Cm, 98.38.Fs INTRODUCTION High resolution radio observations have revealed double and triple radio sources in many Seyfert nuclei, suggesting that the collimated ejection of material is occurring in these objects in the same way as the radio jets in quasars and radio galaxies. But very few PLot file version 1 created 25-MAY-2004 16:17:20 CONT: MKN1034 IPOL 4994.000 MHZ MKN1034.J2000.1 32 11 49.05 PLot file version 2 created 11-MAY-2004 14:33:45 CONT: NONE JHKs cub 3.000 mrk1034.2mass.1 6 ) 0 0 0 2 J ( I N O T A N L C E D I 49.00 48.95 48.90 48.85 48.80 48.75 48.70 02 23 21.975 21.970 21.965 21.960 21.955 21.950 RIGHT ASCENSION (J2000) Cont peak flux = 2.2187E-03 JY/BEAM Levs = 1.978E-04 * (1, 2, 3, 4, 5, 6, 7, 8, 10, 16, 32, 64) 32 12 15 00 ) 0 0 0 2 J ( I N O T A N L C E D I 11 45 30 21.945 21.940 02 23 24 23 22 21 RIGHT ASCENSION (J2000) 20 Cont peak flux = 2.0765E+02 Levs = 2.145E+00 * (1, 2, 4, 8, 16, 32, 64, 128, 256) (Left:) 5 GHz MERLIN map of Mrk 1034. Contour FIGURE 1. levels are at 1.978e- 4×(1,2,3,4,5,6,7,8,10,16,32,64) Jy/beam. The peak flux is 2.219 mJy/beam and the rms noise level is∼98 m Jy/beam. (Right:) Ks (2.17 m m) Two Micron All Sky Survey (2MASS) image of Mrk 1034. Seyfert galaxies have been found to have radio structures that can be described as jets (Kukula et al. 1993; Ghosh et al. 1994). The angular resolution of MERLIN at 5 GHz is equivalent to that of the Hubble Space Telescope (HST), making these radio images ideal for comparison with the structure of the narrow-line region (NLR) and extended narrow line region (ENLR) and perfect cases to study in detail the individual small scale "jets" as revealed in e.g. Capetti et al. (1999). The purpose of the present study is to: a) Investigate the variety of collimated ejection in low luminosity AGN and increase the small number of multiple component "jets" known, b) Determine the alignment of the collimated ejection on scales of a few parsecs and compare this alignment with estimates made from extended NLR studies from HST and other optical observations, c) Involve hydrodynamical simulations and multi-wavelength data in order to constrain the bow shock and similar models and get a more complete picture of this class of AGN and, d) Compare the 5 GHz continuum observations with 22 GHz MERLIN observations of the same galaxies, since the coincidence of the position of the two emissions may mark the location of the obscured central engine. MAPS AND RESULTS Mrk 1034: The MERLIN 5 GHz map (Fig. 1) of Ark 81 - one of the Seyfert 1 galaxies of the pair Ark 80, Ark 81 that are interconnected- shows an E-W elongated radio structure (20 mas) with a 10 mas jet extension to the E consistent with a highly collimated radio jet in that direction. Mrk 1034 has not been observed yet with HST. However, the inner contours of a 2MASS near-IR image of Mrk 1034 (Fig. 1) follow the same elongation along the E-W direction. Mrk 1210: The near-UV emission revealed in a high resolution (0 ′′. 027) ACS/HRC HST (F330W) image of Mrk 1210 follows closely the ∼15 mas one-sided jet unveiled in the MERLIN 5 GHz map (Fig. 2) of this amorphous Seyfert 2 galaxy. PLot file version 1 created 03-JUN-2004 18:18:47 CONT: MKN1210 IPOL 4994.000 MHZ MKN1210.J2000.1 05 06 50.05 ) 0 0 0 2 J ( I N O T A N L C E D I 50.00 49.95 49.90 49.85 49.80 49.75 49.70 49.65 49.60 08 04 05.870 05.865 05.860 05.855 RIGHT ASCENSION (J2000) 05.850 05.845 Cont peak flux = 2.2838E-02 JY/BEAM Levs = 4.899E-04 * (1, 2, 4, 6, 8, 10, 12, 16, 20, 32, 40, 45) PLot file version 10 created 21-JUN-2004 18:45:40 GREY: MKN1210 IPOL 4994.000 MHZ MRK1210.J2000.1 CONT: NONE IPOL 4994.000 MHZ MRK1210O.LGEOM.2 ) 0 0 0 2 J ( I N O T A N L C E D I 05 06 50.2 50.1 50.0 49.9 49.8 49.7 49.6 49.5 05.87 05.86 05.85 08 04 05.88 RIGHT ASCENSION (J2000) Grey scale flux range= 0.800 7.000 MilliJY/BEAM Cont peak flux = 6.3979E+00 ELECTRON Levs = 8.411E-03 * (1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 700) 7 6 5 4 3 2 1 05.84 05.83 (Left:) 5 GHz MERLIN map of Mrk 1210. Contour FIGURE 2. levels are at 4.899e- 4×(1,2,4,6,8,10,12,16,20,32,40,45) Jy/beam. The peak flux is 22.83 mJy/beam and the rms noise level is ∼ 163 m Jy/beam. (Right:) HST ACS/HRC (F330W) contour map of Mrk 1210 overlaid on the MERLIN 5 GHz greyscale image. NGC2639: The MERLIN 5 GHz map (Fig. 3) of this well-known LINER megamaser galaxy reveals a bright core and symmetrical E-W "wings" with rich structure (multiple knots), that is highly correlated with the UV star formation morphology of the HST ACS/HRC image (Fig. 3). PLot file version 2 created 27-MAY-2004 15:43:09 CONT: NGC2639 IPOL 4994.000 MHZ NGC2639.J2000.1 PLot file version 3 created 22-JUN-2004 15:36:42 GREY: NGC2639 IPOL 4994.000 MHZ NGC2639.J2000.1 CONT: NONE IPOL 4994.000 MHZ NGC2639ACS.LGEOM.1 ) 0 0 0 2 J ( I N O T A N L C E D I 50 12 20.6 20.4 20.2 20.0 19.8 19.6 19.4 19.2 08 43 38.15 38.10 38.05 RIGHT ASCENSION (J2000) Cont peak flux = 9.3511E-02 JY/BEAM Levs = 1.297E-04 * (1, 2, 4, 8, 16, 32, 64, 128, 256, 512) 50 12 21.0 20.5 20.0 19.5 19.0 ) 0 0 0 2 J ( I N O T A N L C E D I 38.00 38.10 08 43 38.15 38.00 RIGHT ASCENSION (J2000) Grey scale flux range= 80.0 230.0 MicroJY/BEAM Cont peak flux = 9.8145E-02 ELECTRON Levs = 2.594E-02 * (1, 1.500, 2, 3, 4) 38.05 220 200 180 160 140 120 100 80 37.95 (Left:) 5 GHz MERLIN map of NGC 2639. Contour FIGURE 3. levels are at 1.297e- 4×(1,2,4,8,16,32,64,128,256,512) Jy/beam. The peak flux is 93.5 mJy/beam and the rms noise level is ∼48 m Jy/beam. (Right:) HST ACS/HRC (F330W) contour map of NGC 2639 overlaid on the MERLIN 5 GHz greyscale image. PLot file version 1 created 27-AUG-2004 12:48:08 BOTH: NGC4922A NGC4922AB.WFPFIN.1 29 19 05 17.0 17.5 18.0 RIGHT ASCENSION (J2000) Grey scale flux range= 25.00 40.00 Cont peak flux = 1.9780E+03 Levs = 1.547E+00 * (-1, 1, 2, 4, 8, 16, 25, 32, 40, 64, 128, 256, 512) FIGURE 4. (Left:) 5 GHz MERLIN map of NGC 4922C (PGC 044896). Contour levels are at 3.803e- 4×(1,2,4,6,8,9,10,16,32,64,128,256,512,1024) Jy/beam. The peak flux is 3.97 mJy/beam and the rms noise level is ∼126 m Jy/beam. (Right:) HST WFPC2 (F606W) image of the triple system NGC 4922. NGC 4922: The MERLIN 5 GHz map (Fig. 4) detects the middle component C, PGC 044896 (FIRST J130125.2+291849 radio source), from the three (3) galaxy system PLot file version 1 created 03-JUN-2004 15:20:59 CONT: NGC4922A IPOL 4994.000 MHZ NGC4922B.J2000.1 29 18 49.65 ) 0 0 0 2 J ( I N O T A N L C E D I 49.60 49.55 49.50 49.45 49.40 13 01 25.270 25.265 25.260 25.255 RIGHT ASCENSION (J2000) 25.250 Cont peak flux = 3.9780E-03 JY/BEAM Levs = 3.803E-04 * (1, 2, 3, 4, 6, 8, 9, 10, 16, 32, 64, 128, 256, 512, 1024) B C A 40 38 36 34 32 30 28 26 ) 0 0 0 2 J ( I N O T A N L C E D I 10 15 20 25 30 35 40 45 50 55 13 01 16.5 NGC 4922 (NGC 4922A/B are both Seyfert 2) as shown in HST WFPC2 (F606W) im- age (Fig. 4). The radio structure is elongated in the NE-SW direction, in good agreement with the orientation of the interacting galaxies. NGC 5506: The MERLIN 5 GHz map shows a NW radio loop/jet (∼5 mas in extent) that follows closely the rich near-UV emission (Fig. 5). PLot file version 2 created 03-JUN-2004 19:39:39 CONT: NGC5506 IPOL 4994.000 MHZ NGC5506.J2000.1 -03 12 27.2 -03 12 26.8 PLot file version 5 created 29-JUN-2004 21:14:47 GREY: NGC5506 IPOL 4994.000 MHZ NGC5506.J2000.1 CONT: NGC5506 IPOL 4994.000 MHZ NGC5506FOC.LGEOM.1 ) 0 0 0 2 J ( I N O T A N L C E D I 27.3 27.4 27.5 27.6 27.7 27.8 27.9 28.0 28.1 28.2 14 13 14.91 14.90 14.89 14.88 14.87 14.86 14.85 RIGHT ASCENSION (J2000) Cont peak flux = 5.7292E-02 JY/BEAM Levs = 2.185E-03 * (1, 2, 4, 6, 8, 10, 12, 16, 20, 24) ) 0 0 0 2 J ( I N O T A N L C E D I 27.0 27.2 27.4 27.6 27.8 14 13 14.92 14.90 14.88 14.86 RIGHT ASCENSION (J2000) 14.84 14.82 Grey scale flux range= 2.00 20.00 MilliJY/BEAM Cont peak flux = 1.5766E+02 Levs = 5.000E+00 * (-1, 1, 2, 4, 8, 16, 32, 64, 128, 256, 512) 20 15 10 5 (Left:) 5 GHz MERLIN map of NGC 5506. Contour FIGURE 5. levels are at 2.185e- 3×(1,2,4,6,10,12,16,20,24) Jy/beam. The peak flux is 57.29 mJy/beam and the rms noise level is ∼ 546 m Jy/beam. (Right:) HST ACS/HRC (F330W) contour map of NGC 5506 overlaid on the MERLIN 5 GHz greyscale image. PLot file version 3 created 27-FEB-2005 16:25:02 GREY: TXFS2226 TXFS2226HA.LGEOM.1 CONT: TXFS2226 TXFS2226.J2000.1 -18 10 45.4 ) 0 0 0 2 J ( I N O T A N L C E D I 45.6 45.8 46.0 46.2 46.4 46.6 46.8 47.0 22 29 12.58 12.56 12.54 12.52 12.50 RIGHT ASCENSION (J2000) Grey scale flux range= 0.000 6.000 Cont peak flux = 1.4255E-02 JY/BEAM Levs = 2.500E-04 * (-1, 1, 2, 4, 8, 16, 32, 64, 128, 512) 6 5 4 3 2 1 0 12.48 FIGURE 6. (Left:) The average red/green continuum HST WFPC2 image of the host galaxy of TXFS 2226-184. (Right:) 5 GHz MERLIN map of TXFS 2226-184 overlaid on the HST WFPC2 Ha image. Contour levels are at 1.750e-4×(1,2,4,8,16,32,64,128,256,512) Jy/beam. The peak flux is 14.25 mJy/beam and the rms noise level is ∼ 65 m Jy/beam. TXFS 2226-184: The multiple knot radio structure (Fig. 6) coincides with the nucleus of the host galaxy and then is aligned with the Ha emission on the larger scale. This alignment reflects an interaction between the radio jet and the interstellar medium. The Ha and radio structures are perpendicular to the galaxy major axis (Fig. 6). IC 1481: The radio map of this LINER water maser galaxy and a possible site of ongoing merger, shows elongated structure (Fig. 7) which agrees with VLBI observations that have shown that water maser emission is associated with thin, parsec-scale molecular disks located near the spout of radio continuum jets. PLot file version 1 created 28-FEB-2005 16:42:47 CONT: IC1481 IPOL 4994.000 MHZ IC1481.ICLN1.1 05 54 22.05 PLot file version 2 created 11-MAY-2005 19:05:12 GREY: IC1481 IC1481.J2000.1 CONT: U54V0905 IC1481.LGEOM.2 05 54 24.5 ) 0 0 0 2 J ( I N O T A N L C E D I 22.00 21.95 21.90 21.85 21.80 21.75 21.70 21.65 21.60 23 19 25.090 25.085 25.080 25.075 25.070 RIGHT ASCENSION (J2000) Cont peak flux = 1.0535E-03 JY/BEAM Levs = 1.024E-04 * (-1, 1, 2, 4, 6, 8, 10) ) 0 0 0 2 J ( I N O T A N L C E D I 24.0 23.5 23.0 22.5 23 19 25.12 25.10 25.08 25.06 25.04 25.02 RIGHT ASCENSION (J2000) Grey scale flux range= 0.000 1.000 MilliJY/BEAM Cont peak flux = 1.7759E+02 Levs = 1.136E+00 * (-1, 1, 2, 4, 8, 16, 32, 64, 128, 256, 512) 0.8 0.6 0.4 0.2 0.0 25.00 24.98 (Left:) 5 GHz MERLIN map of FIGURE 7. levels are at 1.750e- 4×(1,2,4,8,16,32,64,128,256,512) Jy/beam. The peak flux is 14.25 mJy/beam and the rms noise level is ∼ 65 m Jy/beam. (Middle:) The red continuum HST WFPC2 image of the host galaxy. (Right:) The Ha HST WFPC2 contour map overlaid on 5 GHz MERLIN greyscale image. IC 1481. Contour DISCUSSION It is not widely appreciated that Seyferts, in common with radio galaxies, may show evidence for collimated ejection and/or radio jets. The conical or biconical morphology of the UV continuum in Seyferts has been found to be aligned with the radio axis, and we can postulate that the radio jets define the axis of the UV cone and also the central engine. In this area, high resolution images obtained by MERLIN at 5 GHz have provided the clearest evidence for linear radio jets in Seyfert galaxies that have been previously poorly resolved in VLA images. For this reason we observed with MERLIN at 5 GHz and 0 ′′. 04 resolution, 7 Seyfert galaxies that were found to show hints of such structures in previous lower resolution radio observations. The higher resolution observations allow us to identify the orientation and geometry of the nuclear disk and further examine, for the ones that was possible, the relation of the nuclear radio structure to the inner and large scale morphology, emission and dust lanes seen in optical images. This way we are able to build up a more complete picture of the physics of the low luminosity AGN. REFERENCES 1. Capetti, A., Axon, D. J., Macchetto, F., Marconi, A., Winge, C., Astrophys. J, 516, 187–194 (1999). 2. Ghosh, T., Schilizzi, R. T., Miley, G. K., de Bruyn, A. G., Kukula, M. J., Pedlar, A., Graham, D., Saikia, D. J., in "Multi-wavelength continuum emission of AGN", edited by Courvoisier T. J.-L., Blecha, A., Proceedings of IAU Symp. 159, Kluwer, Dordrecht, 1994, p. 426. 3. Kukula, M. J., Ghosh, T., Pedlar, A., Schilizzi, R. T., Miley, G. K., de Bruyn, A. G., Saikia, D. J., Month. Not. Roy. Astron. Soc., 264, 893–899 (1993).
astro-ph/0510848
1
0510
2005-10-31T13:39:22
An Intermediate-band imaging survey for high-redshift Lyman Alpha Emitters: The Mahoroba-11
[ "astro-ph" ]
We present results of our intermediate-band optical imaging survey for high-$z$ Ly$\alpha$ emitters (LAEs) using the prime focus camera, Suprime-Cam, on the 8.2m Subaru Telescope. In our survey, we use eleven filters; four broad-band filters ($B$, $R_{\rm c}$, $i^\prime$, and $z^\prime$) and seven intermediate-band filters covering from 500 nm to 720 nm; we call this imaging program as the Mahoroba-11. The seven intermediate-band filters are selected from the IA filter series that is the Suprime-Cam intermediate-band filter system whose spectral resolution is $R = 23$. Our survey has been made in a $34^\prime \times 27^\prime$ sky area in the Subaru XMM Newton Deep Survey field. We have found 409 IA-excess objects that provide us a large photometric sample of strong emission-line objects. Applying the photometric redshift method to this sample, we obtained a new sample of 198 LAE candidates at $3 < z < 5$. We found that there is no evidence for evolution of the number density and the star formation rate density for LAEs with $\log L({\rm Ly}\alpha) ({\rm erg s^{-1}}) > 42.67$ between $z \sim 3$ and 5.
astro-ph
astro-ph
An Intermediate-band imaging survey for high-redshift Lyman Alpha Emitters: The Mahoroba-11 Sanae F. Yamada,1 Shunji S. Sasaki,1 Ryoko Sumiya,1 Kazuyoshi Umeda,1 Yasuhiro Shioya,1 Masaru Ajiki1 , Tohru Nagao,2,3 Takashi Murayama,1 and Yoshiaki Taniguchi1 [email protected] [email protected] 1Astronomical Institute, Graduate School of Science, Tohoku University, Aramaki, Aoba, Sendai 980-8578 2National Astronomical Observatory of Japan, 2-21-1 Osawa, Mitaka, Tokyo 181-8588 3INAF -- Osservatorio Astrofisico di Arcetri, Largo Enrico Fermi 5, 50125 Firenze, Italy (Received 2005 January 0; accepted 0 0) Abstract We present results of our intermediate-band optical imaging survey for high-z Lyα emitters (LAEs) using the prime focus camera, Suprime-Cam, on the 8.2m Subaru In our survey, we use eleven filters; four broad-band filters (B, Rc, i′, Telescope. and z′) and seven intermediate-band filters covering from 500 nm to 720 nm; we call this imaging program as the Mahoroba-11. The seven intermediate-band filters are selected from the IA filter series that is the Suprime-Cam intermediate-band filter system whose spectral resolution is R = 23. Our survey has been made in a 34′ × 27′ sky area in the Subaru XMM Newton Deep Survey field. We have found 409 IA- excess objects that provide us a large photometric sample of strong emission-line objects. Applying the photometric redshift method to this sample, we obtained a new sample of 198 LAE candidates at 3 < z < 5. We found that there is no evidence for evolution of the number density and the star formation rate density for LAEs with log L(Lyα)(erg s−1) > 42.67 between z ∼ 3 and 5. Key words: cosmology: observations -- early universe -- galaxies: formation -- galaxies: evolution 1 1. Introduction Formation and evolution of galaxies have been intensively discussed in this decade. This progress has been supported by the Hubble Space Telescope (HST) and ground-based 8-10 m class telescopes such as the W. M. Keck telescopes, VLT, the Gemini Telescopes, and the Subaru Telescope. In particular, the Hubble Deep Fields (Williams et al. 1996, 2000) and the Hubble Ultra Deep Field (Beckwith et al. 2003) make it possible to search for very faint galaxies beyond z = 5 (e.g., Lanzetta et al. 1996; Weymann et al. 1998; Bunker et al. 2004: Yan, Windhorst 2004). Early star formation in galaxies could occur beyond z = 5 (e.g., see for reviews, Taniguchi et al. 2003b, Spinrad 2003). Yet, the peak of star formation in the universe appears to occur at z ∼ 1 (Madau et al. 1996; Steidel et al. 1999; Ouchi et al. 2004; Giavalisco et al. 2004). Although this observational trend has been widely accepted, we have not yet fully understood which physical processes play crucially important roles during the course of evolution of galaxies. Therefore, in order to understand the whole history of galaxies in the universe from high redshift to the present day, it is important to carry out systematic searches for galaxies at any redshift. It has been also claimed that the basic dynamical properties of galaxies could be deter- mined at 1 < z < 3; i.e., the origin of the Hubble type of galaxies could be established at this redshift interval (e.g., Kajisawa, Yamada 2001). In addition, early star formation in galaxies as well as early growth of supermassive black holes in their nuclei should be clarified unambigu- ously in terms of the hierarchical structure formation paradigm (Madau et al. 1996; Barger et al. 2001; Kauffmann, Haehnelt 2000). In order to explore this issue, it is absolutely necessary to investigate observational properties of galaxies beyond z = 1. Indeed, many observational programs have been devoted to this problem by finding high-z galaxies. Searches for such high-z galaxies have been carried out mainly by the following two methods. One is the optical broad-band color selection technique (e.g., Steidel, Hamilton 1992; Steidel et al. 1999; Madau et al. 1996; Lanzetta et al. 1996; Giavalisco et al. 2004 and references therein). The other method is deep searches for strong Lyα emitters by using an optical narrow-band filter (Hu et al. 1996; Cowie, Hu 1998; Kudritzki et al. 2000; Rhoads et al. 2000; Ajiki et al. 2003; Kodaira et al. 2003; Taniguchi et al. 2005). We also note that the slitless grism spectroscopy is another method to search for the high-z galaxies (Kurk et al. 2004; Martin & Sawicki 2004; Pirzkal et al. 2004). The first method provides a relatively large sample of high-z Lyman break galax- ies (LBGs) and their follow-up spectroscopy tells us global properties of high-z galaxies. However, such studies are biased to investigations of relatively bright galaxies because a certain magnitude-limited sample is used in such a study. On the other hand, the second method provides us a sample of strong emission-line objects, most of which are missed in a magnitude- limited sample, although it is not clear whether or not they are typical faint normal galaxies. 2 However, because of the narrowness of the filter bandpass, such a study can only probe strong emission-line objects in a small volume of the universe (see for a review, Taniguchi et al. 2003b). In order to overcome the demerit of searches with a narrow-band filter, searches with multi narrow-band filters have been made in some cases (Rhoads et al. 2000; Shimasaku et al. 2004). Even in their pioneering surveys, only a couple of filters were used and thus their survey volumes were still small. More recently, new surveys with a combination between typical broad-band filters and intermediate- and narrow-band filters have been conducted. The most successful survey made so far is the COMBO-17 survey in which 17 filters are used (Wolf et al. 2003a; see also Bell et al. 2004); 5 broad-band filters and 12 medium-band ones. Indeed, this survey has been used to make a systematic search for strong emission-line objects such as active galactic nuclei between z ∼ 1 and z ∼ 5 (Wolf et al. 2003b). In order to search for strong Lyα emitters systematically, we have conducted a new deep optical imaging survey with 11 filters using the prime focus camera, Suprime-Cam (Miyazaki et al. 2002), on the 8.2m Subaru Telescope (Kaifu et al. 2000; Iye et al. 2004). In this survey, we use a set of seven intermediate-band (IA) filters covering wavelengths from ≈ 527nm to 709nm, corresponding to the Lyα redshift from z = 3.4 and z = 4.8 (Hayashino et al. 2000; Taniguchi 2004; Fujita et al. 2003; Taniguchi et al. 2003b; Ajiki et al. 2004). This new data set allows us to investigate the cosmic star formation history systematically from z = 4.8 to z = 3.3 for the first time. In this paper, we present our results and then discuss the nature of strong Lyα emitters at high redshift. Throughout this paper, we adopt a flat universe with Ωm = 0.3, ΩΛ = 0.7, and H0 = 70 km s−1Mpc−1, and we use the AB magnitude system (e.g. Oke 1974). 2. Observations and data reduction 2.1. Observations Deep and wide-field B-, R-, i′-, and z′-band imaging data of a southward 30′ × 24′ area in the Subaru/XMM-Newton Deep Survey Field (the SXDS field1) centered at α(J2000) = 2h18m00s and δ(J2000) = −5◦12′00′′ were obtained using the Suprime-Cam (Miyazaki et al. 2002) on the 8.2m Subaru Telescope during a period between 2000 August, and 2002 January by the SXDS team. The Suprime-Cam consists of ten CCD chips, each of which has 2048 × 4096 pixels, providing a field of view of 27′ × 34′. In the SXDS field, X-ray deep survey by the XMM-Newton satellite as well as multi- broad-band imaging observations are on-going over 1.3 square degree area. Therefore, the field is one of the most suitable target fields for our IA filter survey. Carefully examining the entire SXDS field, we selected the southward field of SXDS field which do not contain any bright stars. Note that a sky area of 13.′7 ×13.′7 in the central part of the SXDS field was already 1 See http://www.naoj.org/Science/SubaruProject/SDS/. 3 observed by Fujita et al. (2003) using an IA filter, IA574. However, our observing field does not overlap with it (see figure 2). In addition to these broad-band image data, intermediate-band images were obtained during an open-use observing program, S02B-163 (PI = K. Kodaira), using the follow- ing IA filters: IA527, IA574, IA598, IA624, IA651, IA679, and IA709, on Suprime- Cam / Subaru Telescope during a period between 2002 October and 2002 November. The center wavelength and F W HM of these IA filters are summarized in table 1; see also http://www.awa.tohoku.ac.jp/~tamura/astro/filter.html , Hayashino et al. (2000, 2003), and Taniguchi (2001). The transmission curves of the filters used in our observations are shown in figure 1. A journal of our all observations is given in table 1. All of the observations were made under photometric conditions, and the seeing size was ≃ 1.′′0. 2.2. Data Reduction The individual CCD data were reduced and combined using IRAF and the mosaic-CCD data reduction software developed by Yagi et al. (2002). The following spectrophotometric standard stars were observed to calibrate the IA filter imaging data; LDS749B, LTT9491, G93- 48, and Feige110 [references here Oke (1974), Oke (1990), Landolt (1992), and Stone (1996)]. The combined images for the individual bands were aligned and smoothed with Gaussian kernels to match their seeing sizes. The final images cover a contiguous 944 arcmin2 area (34′.71 × 27′.20) with a PSF FWHM of 1.′′04 for the broad-band and IA-band data. The final IA image stacked with the seven IA images is shown in figure 3. After masking out areas which are affected by the starlight, we obtained the final image whose effective sky coverage is 526 arcmin2. Note that the data reduction was made by the team of S02B-163 program in which the authors in this paper were included as collaborators. 3. Results 3.1. Source Detection and Photometry For photometry and source detection of our observational data, we use SExtractor version 2.1.7 (Bertin & Arnouts 1996). As for the source detection in all images, we use the limiting magnitudes for a 3σ detection with a 2′′ diameter aperture : B = 28.2, Rc = 27.4, i′ = 27.0, z′ = 25.8, IA527 = 26.8, IA574 = 26.5, IA598 = 26.5, IA624 = 26.7, IA651 = 26.7, IA679 = 26.8, and IA709 = 26.6. In the above source detection, we have detected ∼ 70000 sources down to IA = 27 in each IA image. As shown in the left panel of figure 4, we can detect objects to ∼ 27 mag with above condition. In order to examine the completeness of our IA images, we show results of the number counts in the IA imaging as a function of AB magnitude in figure 4. These results show that our IA imaging appears complete down to 26.5 -- 27 mag for each IA image. In order to examine further how accurately we select emission-line objects in our IA 4 images, we have made simulations using IRAF ARTDATA (e.g., Kajisawa et al. 2000; Fujita et al. 2003). For this purpose, we generate two sets of 300 model galaxies for each magnitude interval, ∆m = 0.2 mag. Model galaxies in the first set obey the de Vaucouleurs r1/4 law light distribution while those in the second set obey the exponential law. Their sky positions, half- light radius (from 1 to 7 kpc), and ellipticities are randomly determined. We put these galaxies into the CCD data together with Poisson noises. After smoothing model-galaxy images to match to the seeing size of our observation, we try to detect them using SExtractor with the same procedure as that we used in our reduction. In the right panel of figure 4, the detectability is shown as a function of IA magnitude for each IA filter. These results also show that our IA imaging appears complete down to 26.5 -- 27 mag for each IA image. 3.2. Selection of IA-Excess Objects We describe our procedures to select IA-excess objects from our data set. For this purpose, we need a continuum image for each IA image. Since the central wavelength of each IA filter does not generally match to that of a certain broad band filter, we have to generate a custom continuum image for each IA filter. In previous studies (e.g., Steidel et al. 2000; Fujita et al. 2003), the broad band image whose effective wavelength is close to that of the concerned narrow-band or intermediate-band one was used as a continuum image. However, if we follow this way, continuum break objects such as LBGs could also be selected as a strong IA-excess object. In order to reject such objects, we have adopted the image of the broad filter band (D[IA]) whose central wavelength is longer than that of the concerned IA filter as a continuum image: D[IA] is Rc-band for IA527, IA574, and IA598, and i′-band for IA624, IA651, IA679, and IA709. Hereafter, we refer these continuum D[IA], or D[IAnnn] where nnn = 527, 574, 598, 624, 651, 679, or 709. In figure 5, we show the color - magnitude diagram between D[IA] − IA and IA for objects in our IA catalogs. Taking the scatter in the D[IA] − IA color into account, we define strong IA-excess objects with the following criterion, D[IA] − IA ≥ mag(EWrest = 20A ), (1) where mag(EWrest = 20A ) is the magnitude corresponding to EWrest = 20A determined as, D[IA] − IA = −2.5log FWHM (IA) EWobs + F W HM(IA) , (2) where F W HM(IA) is the F W HM of IA filter (see table 1). mag(EWrest=20A ) are 0.332, 0.325, 0.314, 0.320, 0.311, 0.311, and 0.339 for IA527, IA574, IA598, IA624, IA651, IA679, and IA709, respectively. These criteria are shown by horizon lines in figure 5. We also add the criteria for error, D[IA] − IA > 3σ(D[IA] − IA) and IA < 3σ(IA) (3) (see curved lines in figure 5). Then, we select 74, 47, 42, 48, 60, 77, and 61 strong IA- 5 excess objects in the IA527, IA574, IA598, IA624, IA651, IA679, and IA709 filter images, respectively. 3.3. Selection of Lyman α Emitter Candidates Because of the large observed equivalent width, it seems that most IA-excess objects may be LAEs at high redshift. However, there may be still a possibility that some of them are strong emission-line objects at lower redshifts; e.g., Hα emitters at z ∼ 0.00 to 0.10, [O III] λ 5007 emitters at z ∼ 0.03 to 0.45, and [O II] λ 3727 emitters at z ∼ 0.38 to 0.94. Here, we try to select LAEs from IA-excess objects. Although a color - color diagram is usually used to separata the LAEs from low-z emission-line galaxies (e.g., Taniguchi et al. 2005), it seems difficult to separate LAEs properly especially for blue IA bands. We therefore adopt the photometric redshift (SED fitting) technique using all 11 bands photometric data. Using this method, we can select LAEs with enough accuracy for all IA bands. The detailed discussion on the accuracy of the photometric redshift is shown in appendix 1. For this purpose, we generate model galaxy SEDs using the population synthesis model, GALAXEV, developed by Bruzual & Charlot (2003). The SEDs of local galaxies are well reproduced by models whose star-formation rate declines exponentially (the τ model) : i.e., SF R(t) ∝ exp(−t/τ ), where t is the age of galaxy and τ is the time scale of star formation. In this study, we use τ = 1 Gyr models with Salpeter's initial mass function (the power index of x = 1.35 and the stellar mass range of 0.1 ≤ m/M⊙ ≤ 100) to derive various SED types. We note that the SED templates derived by Coleman et al. (1980) for elliptical galaxies, Sbc, Scd, Irr, and SB, correspond to those using t =8, 4, 3, 2, and 1 Gyr, respectively. We calculate SEDs with ages of t=0.1, 0.5, 1, 2, 3, 4, and 8 Gyr. When we use only optical broad band photometry for estimates of the photometric redshift, we need not seriously take account of the contribution of emission lines to the broad band flux. However, in our case, we cannot neglect the contribution of some strong emission lines because our photometric data contain intermediate-band data. In order to include possible emission-line fluxes into our models, we calculate the number of ionizing photons, NLyc, from the SED which is calculated from the above population synthesis model. Then we can estimate Hβ luminosity, L(Hβ) using the following formula (Leitherer & Heckman 1995), L(Hβ) = 4.76 × 10−13NLyc erg s−1. (4) Other strong emission-line luminosities, such as for [O ii], [O iii], Hα, and so on, are estimated by typical line ratios relative to Hβ (PEGASE: Fioc & Rocca-Volmerange 1997). In addition to the contribution of strong emission lines to SEDs, we also take the following two effects into account in SED templates. One is the absorption by interstellar medium in a galaxy itself. For this correction, we use the starburst reddening curve of Calzetti et al. (2000), Fi(λ) = Fo(λ)100.4Es(B−V )k′(λ), (5) 6 where Fi(λ) is the intrinsic stellar continuum, Fo(λ) is the observed stellar continuum, and k′(λ) = A′(λ)/Es(B − V ) is the starburst reddening curve. The expression of k′(λ) is k′(λ) = 2.659(−1.857 + 1.040/λ) + R′ V for 0.63µm ≤ λ ≤ 2.20µm; = 2.659(−2.156 − 1.509/λ − 0.198/λ2 + 0.011/λ3 + R′ V ) for 0.12µm ≤ λ < 0.63µm, (6) V = A′(V )/Es(B − V ) = 4.05. And another effect is the absorption from intergalactic where R′ gas. Following the method in Madau et al. (1996), we describe the observed mean specific flux of a source at redshift zem as hf (νobs)i = (1 + zem)L(νem) 4πd2 L he−τ i, (7) where νobs = νem/(1 + zem), dL is the luminosity distance corresponding to zem and he−τ i is the average transmission. Then, we try to fit the observed SEDs of IA-excess objects using SED templates gener- ated by the above procedures. In this fitting, we calculate the standardized "relative likelihood" between zph=0 and zph=5.0. We adopt a redshift of the primary peak of likelihood as the most probable photometric redshift only when the peak value is higher by a factor of two than that of the secondary peak. Otherwise, we judge that we cannot obtain any reliable photometric redshift. In this way, we have obtained photometric redshifts for 64, 40, 20, 26, 16, 17, and 15 LAEs found in IA527, IA574, IA598, IA624, IA651, IA679, and IA709 selected catalogs, respectively. We summarize the number of LAEs in table 2. The photometric catalogs of LAEs are shown in table 3. We show the spatial distributions of LAEs in figure 6. The areas shown by grey color are the masked region. Although it is interesting to investigate clustering properties of these objects, a significant part of the sky area is masked on each IA image and thus it seems difficult to obtain a firm result on the clustering properties. Therefore, we do not discuss this issue further in this paper. 3.4. Equivalent Widths We show the distributions of observed equivalent widths, EWobs, for the high-z LAE sample found in the previous section for those found in each IA filter in figure 7. We note that a LAE found in IA574 catalog have very large EWobs; 6840.6 A2. These data are not shown in figure 7. We find that LAEs with smaller EWobs tend to be more populous. We also show the histograms for the total sample in the lower-right panel of figure 7. Next, we show the distributions of rest-frame equivalent widths for the same samples in figure 8 where EW0(Lyα) 2 We comment on this large EWobs. The Rc-band flux of this object is smaller than 2σ. Using the 2σ flux, we derive the lower limit of EWobs = 1288 A. 7 = EWobs(Lyα)/(1 + z). 3.5. Lyman α luminosity We estimate the Lyα luminosity for each LAE found in our survey using the following relation, L(Lyα) = 4πd2 Lf (Lyα) erg s−1, (8) where dL is the luminosity distance, and f (Lyα) is the Lyα flux estimated using D[IA] and IA magnitude below, f (Lyα) = {fν(IA) − fν(D[IA])}∆ν erg s−1 cm−2, (9) fν(IA) = 10−0.4(IA+48.6) where 10−0.4(D[IA]+48.6) erg s−1 cm−2 Hz−1, and ∆ν is the bandwidth in unit of Hz. culation, we assume all the LAEs exist at the redshift shown in table 2 for each IA-band. fν(D[IA]) erg s−1 cm−2 Hz−1, = In this cal- 4. Discussion 4.1. Number Densities of Lymanα Emitters Since we have found LAEs at z ≃ 3.3 -- 4.8, we can investigate the number density evo- lution of LAEs as a function of redshift. The faintest log L(Lyα) is different for different IA band: 42.32 for IA527, 42.51 for IA574, 42.56 for IA598, 42.56 for IA624, 42.61 for IA651, 42.63 for IA679, and 42.67 for IA709, where L(Lyα) is in units of erg s−1. In order to compare with samples which is obtained by the same selection limit, we use the LAEs which is brighter than the IA limit, logL(Lyα) = 42.67. The number of LAEs brighter than logL(Lyα) = 42.67 is 10 (IA527), 15 (IA574), 10 (IA598), 15 (IA624), 12 (IA651), 15 (IA679) and 15 (IA709), respec- tively, where L(Lyα) is again in units of erg s−1. Our effective survey area is 526 arcmin2. Then the volume covered by each filter is 3.36×105 (IA527), 3.63×105 (IA574), 3.86×105 (IA598), 3.85×105 (IA624), 4.05×105 (IA651), 4.12×105 (IA679), and 3.78×105 (IA709) Mpc3, respec- tively. Therefore, we obtain the number density of LAEs for each filter as follows; n(LAE) ≃ 3.0 × 10−5 Mpc−1 at z ≃ 3.34, 4.1 × 10−5 Mpc−1 at z ≃ 3.72, 2.6 × 10−5 Mpc−1 at z ≃ 3.93, 3.9 × 10−5 Mpc−1 at z ≃ 4.12, 3.0 × 10−5 Mpc−1 at z ≃ 4.35, 3.6 × 10−5 Mpc−1 at z ≃ 4.58, and 3.9 × 10−5 Mpc−1 at z ≃ 4.82. These results are shown in figure 9. Taking the estimated error into account, we may conclude that the LAE number density is constant at 3.3 < z < 4.8. Let us compare our results with published results. Fujita et al. (2003) obtained n(LAE) ≃ 6.4 × 10−5 at z ≃ 3.7, being smaller by a factor of two than our value at z ≃ 3.7. This may be attributed to their survey is shallower than ours. Cowie & Hu (1998) made deep imaging survey for LAEs at z ∼ 3.4 in the Hubble Deep Field North and SSA22 and then found n(LAE) ∼ 1 × 10−3 Mpc−1. Kudritzki et al. (2000) obtained n(LAE) ∼ 1 × 10−3 Mpc−1 for a blank field at z ≃ 3.1. Steidel et al. (2000) also obtained a large value for the SSA22a field; n(LAE) ∼ 4 × 10−3 Mpc−1 at z ≃ 3.1. Although the field studied by Steidel et al. (2000) is a so-called 8 over density region, the LAE number density appear to depend on the survey depth. In this respect, since our LAE survey is a homogeneous one for LAEs between z ∼ 3.3 - 4.8, the results shown in figure 9 can be used to investigate the evolution of LAE number density for the first time. Therefore, the constant number density of LAEs is one of important results in this study. 4.2. Lyα Luminosity Distribution In order to estimate contribution of LAEs to the cosmic star formation rate density at high redshift, we need a reliable Lyα luminosity function based on a large sample of LAEs. Since we have found 198 reliable candidates of LAEs at z ≃ 3.3 -- 4.8, our sample is useful for this purpose. In figure 10, we show the distribution of Lyα luminosities. We note that in this figure each bin width is log L(Lyα) = 0.2 where L(Lyα) is in units of erg s−1. Our derived Lyα luminosities range from 1042.3 to 1043.2 erg s−1. These luminosities are higher than typical ones found in previous LAE surveys with use of a narrowband filter. The reason for this is that our IA filters have wider FWHMs than the narrowband filters used in the previous LAE surveys and thus our survey tend to find LAEs with a larger Lyα equivalent width. Therefore, we may underestimate the density at low luminosity range (log L(Lyα) < 42.5) by the limit of L(Lyα) at high-redshift. . 4.3. UV Luminosity Distribution Since the LAE candidates are selected by the estimation of photometric redshift, their UV magnitudes are bright enough to be detected on our R or i′ image. Therefore, these candidates would be detected as LBG candidates. In this respect, our LAE candidates can be regarded as subsamples of LBGs. Therefore, it is interesting to compare UV luminosities of our LAEs with those of LBGs at similar redshifts. For this purpose, we estimate UV luminosities of our LAEs using broad band photometric data. We adopt Rc-band or i′-band as UV continuum; R for the LAEs found in IA527, IA574, and IA598 catalogs, while i′ for those in IA624, IA651, IA679, and IA709 catalogs. The results for LAEs of each IA catalog are shown in figure 11. In the lower-right panel of figure 11, we also show the results for all LAEs in our study. In this panel, we show the results of z ≃ 5.7 LAE survey made by Hu et al. (2004) (filled squares). Our results appear to be quite similar to their results although our redshift range (3.3 < z < 4.8) is smaller than the redshift at the survey field of Hu et al. This suggests that the UV luminosity of LAEs does not show strong evolution from z ≃ 5.7 to z ≃ 3.3 although we need more data to confirm this. Then we compare our results with those obtained for LBGs at z ∼ 3 -- 4 (Steidel et al. 1999). As shown in figure 11, the number density of LBGs is systematically higher than that of LAEs in any UV luminosity (by a factor ∼ 5). This may in part explain why the SFRD derived from LAEs is significantly smaller than that derived from LBGs. This will be discussed later 9 (see Section 4.4.3). 4.4. Star Formation Rate 4.4.1. Star Formation Rate Based on Lyα Luminosity We estimate the SFR for the emitters at each seven IA filters. We adopted the formula from Kennicutt (1998), SF R = 7.9 × 10−42L(Hα) M⊙ yr−1 (10) where L(Hα) is in units of erg s−1, and from the case B recombination theory (Brocklehurst 1971), L(Lyα) = 8.7L(Hα), (11) where L(Lyα) is also in units of erg s−1, we can obtain the conversion relation between the SFR and the Lyα luminosity, SF R(Lyα) = 9.1 × 10−43L(Lyα) M⊙ yr−1 . (12) Using this relation, we estimate the SFR for all LAEs. The results are given in figure 12. The obtained SFR ranges between 1.8 M⊙ yr−1 and 17.0 M⊙ yr−1 with a median SFR of 4.1 M⊙ yr−1. These results are consistent with previous surveys for high-z LAEs (e.g., Cowie & Hu 1998; Keel et al. 1999; Kudritzki et al. 2000; Fujita et al. 2003; Ajiki et al. 2003; Cuby et al. 2003; Taniguchi et al. 2005). 4.4.2. Star Formation Rate Based on UV Luminosity We can also estimate SFR from rest-frame UV continuum and then we examine whether or not the SFR derived from the Lyα luminosity is consistent with that derived from the UV continuum luminosity for our sample. As described in section 4.3, we can estimate the rest- frame UV luminosity for all LAEs. Then we obtain the SFR based on UV luminosity using the following relation, SF R(UV) = 1.4 × 10−28Lν M⊙/yr, (13) where Lν is in units of erg s−1 Hz−1 (Kennicutt 1998). The use of continuum magnitudes to estimate the SFR avoids the extremely complex problems of both the Lyα escape process and the uncertainties in the correction of the Lyα fluxes for intergalactic scattering which are present in the determination of the Lyα luminosity function. 4.4.3. Comparison between SF R(Lyα) with SF R(UV) Now we compare SF R(Lyα) with SF R(UV). The results are shown in figure 12 for LAEs found in each IA catalog. In the final panel of this figure, we also show results for all LAEs. We find that the two kinds of SFRs appears to be consistent within a factor of two for most of our LAEs, in particular for LAEs at redshift between z ≃ 3.3 (IA527) and z ≃ 3.9 (IA598). 10 However, for most LAEs at redshift between z ≃ 4.1 and z ≃ 4.8 (i.e., z > 4), there appears a tendency that SF R(Lyα) is systematically smaller by a factor of ≈ 1.3 than SF R(UV). Such tendency is often found LAEs beyond z = 5 (e.g., Hu et al. 2004; Ajiki et al. 2003; Taniguchi et al. 2005 and references therein). In order to see this tendency more clearly, we show the average SF R(Lyα)/SF R(UV) ratio, hlog SF R(Lyα)/SF R(UV )i, as a function of redshift in figure 13; the averages logSF R(Lyα)/SF R(UV ) are 0.03, 0.09, 0.92, −0.10, −0.04, −0.07, and −0.11 for IA527, IA574, IA598, IA624, IA651, IA679, and IA709, respectively. Although these averages seem to slightly decrease with increasing redshift as shown in figure 13, it can be said that they are almost constant, i.e., hlogSF R(Lyα)/SF R(UV )i ≃ 1. We summarize the observational properties of LAEs written above in table 4. 4.4.4. Cosmic Evolution of the Star Formation Rate Density The cosmic star formation history is one of the important key issues related to the formation and evolution of galaxies. In particular, the cosmic star formation history at high redshift has been mainly investigated by using galaxy samples based on the optical broad-band color selection; i.e., the Lyman break method (e.g., Madau et al. 1996; Steidel et al. 1999). More recently, LAEs have been also used to investigate the cosmic star formation history (e.g., Ajiki et al. 2003; Taniguchi et al. 2005; see for a review Taniguchi et al. 2003b). The use of LAE samples as well as LBG ones is important because such strong LAEs could contribute to the cosmic star formation rate significantly. However, the majority of them may be missed in broad-band photometric samples because they are often too faint to be contained in a broad- band magnitude-limited sample. One concern with surveys with a narrowband filter is that the redshift coverage is inevitably small. Therefore, information on the SFR is quite limited for a certain, discrete redshift. Since the LAEs found in this study are located from z ≃ 4.8 to z ≃ 3.3, they allow us to investigate the cosmic star formation history viewed from LAEs systematically in the concerned redshift range. This is indeed one of main purposes of our IA filter survey. We obtain the star formation rate density (SFRD) for each LAE sample at z ∼ 3.3 -- 4.8. The results are shown in the upper panel of figure 14 with previous results. Here, we use SFR calculated from Lyα emission. The SFRDs shown in figure 14 are the simple sum of SFRs divided by the survey volume for all LAE candidates (open circles in the both panels) and LAE candidates of log L(Lyα > 42.67) (filled cirles in the lower panel). The errors of the SFRD which arise from the photometric error is 8 % at most. Since we have not corrected for extinction for Lyα emission and we have not made summing up by using a Lyα luminosity function, these SFRDs are lower limit. Our results show that SFRD for all LAE candidates basically decreases with increasing redshift. The decreasing trend may be caused from our selection limit. We compare the SFRD for LAEs with log L(Lyα) > 42.67 (the filled circles in the lower panel of figure 14). The SFRD 11 derived for LAEs with log L(Lyα) > 42.67 is nearly constant at z = 3.3 -- 4.8. We can detect the LAE candidates from logL(Lyα)≃ 42.4 to logL(Lyα)≃ 43.5 (see figure 10). However since we observe only the bright-end of the number densities, we may miss a number of low-luminosity LAEs. The data points with an upward arrow in figure 14 show the results of LAE searches. Since Lyα emission from high-redshift objects is absorbed either by the intergalactic medium or by the gas in the system itself or both, the blueward flux of Lyα emission should be under- estimated. Therefore it could be desired that the SFRD using Lyα method must be corrected for this effect. 5. Summary We have presented seven optical intermediate-band and multicolor observations of the Subaru / XMM-Newton Deep Field obtained with the Suprime-Cam on the 8.2 m Subaru telescope. The intermediate-band image covered a sky area of 526 arcmin2 in the Subaru/XMM- Newton Deep Field (Ouchi et al. 2001). Our survey volume amounts to 2.9×106Mpc3 when we adopt a flat universe with Ωmatter = 0.3, ΩΛ= 0.7, and H0 = 70 km s−1 Mpc−1. We summarize the major conclusions of this study as follows: (1) In our survey we have found 409 IA-excess objects whose rest-frame Lyα emission- line equivalent widths are greater than 20 A. Applying the photometric redshift technique, we obtain a sample of 198 Lyα emitter candidates. (2) From this LAE sample, we find that the number density of LAEs at z ∼ 3.3 -- 4.8 is n(LAE) ∼ 10−4 Mpc−3 on average. (3) In order to investigate host galaxy properties of our LAEs, we investigate the rest- frame UV luminosity distributions of our LAE sample. The LAEs found in our study are fainter than those found for LAEs at z ≃ 5.7 (Hu et al. 2004). We also find that the number density of LBGs is systematically higher by a factor of ≈ 5 than LAEs in any UV luminosity. (4) Using the observed Lyα luminosity, we estimate the SFR of our LAEs ranging from 1.8 to 17.0 M⊙ yr−1 with a median of 4.1 M⊙ yr−1. We also estimate the SFR us- ing the rest-frame UV luminosity. Comparing these two SFRs, we find that the average SF R(Lyα)/SF R(UV) ratios range from 0.5 to 2. This ratio is nearly constant between z ∼ 3 and 5. (5) Finally, we investigate the cosmic star formation history based on the star formation rate density (SFRD). We obtain SF RD ∼ 10−3.6 M⊙ yr−1 Mpc−3 on average. Although we find a systematic decrease of SF RD with increasing redshift for all LAEs found in our survey, the SFRDs derived by using only bright LAEs (log L(Lyα) > 42.6 appear to be nearly constant between z = 3.3 and 4.8. We would like to thank the Subaru Telescope staff and the SXDS team for their invalu- 12 able assistance. We would also like to thank K. Kodaira, S. Okamura, T. Yamada, K. Ohta, K. Shimasaku, and M. Ouchi for many useful suggestions, comments, and encouragement during the course of this study. We also thank T. Hayashino and H. Tamura for their kind techni- cal help for construction of the IA filter system. This work was financially supported in part by the Ministry of Education, Culture, Sports, Science and Technology (Nos. 10044052, and 10304013) and JSPS (No. 15340059). MA, SSS, and TN are JSPS fellows. IRAF 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. 13 Appendix 1. Accuracy of our selection of LAEs using photometric redshifts Since we select LAEs from IA-excess objects applying the photometric redshift method, it is important to examine if there is no significant selection effect as a function of redshift. In order to demonstrate it, we perform computer simulations in the following way. By comparing differences between the true (model) and estimeated (photometric) redshifts, we can estimate the accuracy of our classification. First, we make the simulated data from the galaxy SEDs taking account of the pho- tometric errors. The galaxy SEDs are generated by using the population synthesis model, GALAXEV (Bruzual & Charlot 2003). To make various kinds of SEDs with strong emission line, we calculate the emission-line fluxes as a function of EW0(Hα), EW0([OIII]), EW0([OII]), and EW0(Lyα) instead of being proportional to the ionizing photon production rate. We adopt three cases of EW0 for each line: EW0 = 100, 200, and 400 A for low-z emission line galaxies, EW0 = 65, 130, and 260 A for LAEs. Second, we select emission-line galaxies from simu- lated catalogs using the same criteria written in section 3.2. Third, we apply the photometric redshift method to the simulated data and compare the photometric redshift (zphot) with the model redshifts (zmodel). Comparisons between zmodel and zphot are shown in figure 15. All the model LAEs are selected as LAEs using our photometric redshift method. Although some of the model low-z emission line galaxies are selected as LAEs, the fraction of misclassification is less than 10 % for the all IA-filter bands. We, therefore, conclude that there is no significant selection effect as a function of redshift. 14 Table 1. The journal of observation Band λc (A ) B 4364 Total F W HM Obs. Date Total Integ. Time Mlim(AB) (A ) 1008 2000 Nov 24,25 2002 Jan 13 (s) (3σ, 2′′φ) 18000 28.2 Rc 6410 1576 2000 Aug 1 2000 Nov 21-24 2001 Nov 17 2002 Jan 13 2000 Nov 25 2002 Jan 13 Total i′ 7589 1535 Total z′ 9024 1409 2001 Oct 15,19,20 IA527 5272 IA574 5743 IA598 6000 IA624 6226 IA651 6502 IA679 6788 IA709 7082 242 271 294 299 322 336 318 2002 Nov 6 2002 Nov 6 2002 Nov 6 2002 Oct 30,31 2002 Oct 30,31 2002 Oct 30,31 2002 Oct 30,31 12000 27.4 13200 5700 5280 7200 6720 10560 9600 10560 11520 27.0 25.8 26.8 26.5 26.5 26.7 26.7 26.8 26.6 15 Table 2. The numbers of LAE (EWrest > 20 A) candidates. Filter z(Lyα) EWobs D[IA] − IA IA-excess objects† LAE candidates‡ (EWrest = 20A) (mag) IA527 3.336 IA574 3.723 IA598 3.934 IA624 4.120 IA651 4.347 IA679 4.582 IA709 4.824 86.711 94.457 98.684 102.401 106.941 111.645 116.480 0.332 0.325 0.314 0.320 0.311 0.311 0.339 † The number of IA-excess objects (see section 3.2). ‡ The number of LAEs (see section 3.3). 74 47 42 48 60 77 61 64 40 20 26 16 17 15 16 Table 3. Catalog of LAE candidates ID ∆RA ∆DEC B R i′ z′ IA527 IA574 IA598 IA624 IA651 IA679 IA709 (arcmin) (arcmin) IA527 7048 7988 9442 9620 11523 11538 13069 15427 15823 17533 31.12 17.98 4.69 3.87 32.95 29.70 15.33 14.95 13.32 7.25 19286 16.25 3.68 32.53 32.98 17.27 13.88 23.54 27.31 10.55 28.10 27691 28584 28996 29083 37812 38785 41668 44636 47969 49443 51844 27.84 54027 55020 55100 55848 57223 57844 59074 59405 60224 60455 60664 8.21 22.84 25.17 20.90 13.32 30.39 24.21 24.81 12.82 16.47 12.93 5.92 12.04 > 28.6 2.10 2.26 2.65 2.62 3.08 3.08 3.46 3.99 4.09 4.47 4.90 24.9 27.6 26.0 27.5 26.7 27.2 26.8 24.5 26.8 25.3 24.6 24.7 24.1 24.7 24.5 27.1 > 26.2 25.5 > 26.9 > 26.9 25.3 25.3 25.0 25.4 25.4 24.5 27.0 25.5 24.6 26.7 25.4 24.6 27.0 25.4 24.6 26.6 25.4 27.4 > 27.4 > 26.2 26.2 > 26.9 > 26.9 > 27.1 27.0 > 27.2 > 27.0 26.2 26.6 26.1 26.4 > 26.2 26.5 > 26.2 26.3 > 26.2 25.4 25.5 25.5 26.3 26.4 26.5 > 26.9 26.2 26.3 26.0 27.0 26.1 26.4 26.2 26.8 > 27.2 26.1 26.3 26.7 26.8 26.2 28.4 > 27.8 > 27.4 > 26.2 26.1 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 > 27.0 27.1 27.1 27.4 26.7 26.5 26.8 27.0 > 26.2 26.3 > 26.2 26.6 > 26.2 25.8 25.6 25.7 26.2 26.8 26.7 26.4 26.5 27.0 26.2 26.8 26.9 26.7 26.9 26.3 > 27.1 27.0 > 27.2 26.6 26.7 26.6 6.87 > 28.6 > 27.8 > 27.4 > 26.2 25.8 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 > 27.0 7.12 7.22 7.19 9.34 9.45 10.14 10.97 26.1 26.1 27.7 26.8 26.6 27.6 26.3 11.70 26.3 12.52 13.04 13.23 13.25 13.43 13.75 13.89 27.6 27.0 27.9 26.9 27.1 28.5 27.0 25.8 25.7 26.7 25.9 26.0 26.8 25.9 25.5 27.1 26.9 25.8 27.3 26.7 26.4 25.9 25.7 25.9 25.3 25.3 24.9 25.8 25.8 26.6 > 26.2 25.9 > 26.9 25.8 25.6 26.0 > 26.2 25.3 25.4 26.5 26.5 26.0 25.9 26.8 25.8 26.1 25.8 25.6 26.6 25.6 26.8 25.6 25.5 26.6 25.7 26.3 26.0 25.8 26.9 26.0 26.3 25.8 25.6 26.7 25.9 26.1 26.9 > 26.2 25.8 > 26.9 26.8 > 27.1 > 27.1 27.0 > 27.0 25.7 25.4 26.0 25.3 25.4 24.3 25.7 25.5 26.4 25.5 26.0 25.9 26.4 25.3 25.9 25.6 26.0 25.5 27.1 > 26.2 26.0 > 26.9 > 26.9 > 27.1 26.5 > 27.2 > 27.0 27.1 > 26.2 25.6 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 25.6 25.5 25.2 26.1 25.7 25.7 25.7 25.6 26.6 25.7 27.4 > 26.2 26.0 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 > 27.0 26.5 26.1 26.3 > 26.2 25.8 24.7 26.3 > 26.9 26.9 > 27.1 > 27.2 26.1 > 26.9 26.9 26.4 26.6 26.9 26.9 27.4 > 27.4 > 26.2 26.0 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 > 27.0 26.0 26.2 > 26.2 25.4 > 26.9 26.6 26.3 26.1 26.3 26.7 14.14 > 28.6 27.6 > 27.4 > 26.2 26.2 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 > 27.0 14.24 14.45 14.52 14.52 25.7 27.1 26.4 27.4 25.4 26.4 25.7 26.7 25.4 25.7 26.5 > 26.2 25.6 25.9 26.9 > 26.2 24.7 25.5 25.2 25.9 25.4 25.3 26.4 > 26.9 25.7 25.9 26.8 > 26.9 25.5 26.2 25.6 26.9 25.3 26.6 25.6 26.1 25.5 26.5 25.8 25.4 26.5 25.8 27.1 > 27.0 17 Table 3. (Continued.) ID ∆RA ∆DEC B R i′ z′ IA527 IA574 IA598 IA624 IA651 IA679 IA709 (arcmin) (arcmin) 63293 69085 70551 71679 72094 72647 13.47 10.84 27.75 23.43 15.20 9.68 74055 19.68 74299 75353 77460 78162 80388 81380 82145 82756 83157 86220 87141 87553 88785 89485 89896 90874 94639 94941 95272 96625 99496 100145 101147 9.17 18.23 23.33 9.64 13.88 33.07 2.78 22.50 25.82 11.33 4.41 11.80 30.56 24.51 30.36 12.26 33.17 24.45 3.67 102766 10.22 6580 23.34 10969 11379 9.65 3.01 15.21 27.3 16.51 16.89 17.15 17.23 17.36 17.67 17.72 17.99 18.48 18.67 19.15 19.44 19.63 19.73 19.86 20.53 20.77 20.85 21.13 27.5 26.7 26.7 26.5 27.4 27.8 27.6 27.6 27.0 25.6 27.9 26.8 27.6 27.6 26.5 27.2 26.7 27.1 27.2 21.30 27.0 22.98 23.64 23.81 24.04 24.38 2.02 3.06 3.15 26.5 27.4 27.5 27.8 27.2 26.8 27.0 28.5 26.5 27.4 26.0 25.9 25.8 26.9 27.4 27.6 26.9 26.1 24.8 25.8 26.6 27.0 25.5 26.5 25.8 26.4 26.1 26.2 26.7 27.0 26.7 25.3 25.4 26.7 26.4 > 26.2 25.8 26.8 26.4 26.4 27.0 > 26.2 26.2 > 26.9 26.7 > 27.1 26.0 25.9 25.8 25.7 25.7 25.8 27.1 > 26.2 25.2 25.0 25.4 25.6 25.9 26.2 26.1 25.9 26.0 25.9 26.4 > 26.9 26.1 25.6 25.8 26.7 26.5 26.7 25.8 25.9 26.0 26.3 26.5 26.8 > 27.0 25.9 25.8 25.9 27.3 > 26.2 25.9 > 26.9 > 26.9 > 27.1 > 27.1 27.0 > 27.0 26.5 > 26.2 26.1 > 26.9 > 26.9 26.6 > 27.1 > 27.2 > 27.0 27.1 > 26.2 25.7 > 26.9 26.8 > 27.1 27.2 > 27.0 26.6 > 27.2 26.9 26.1 24.8 26.1 24.8 25.4 24.3 25.9 24.9 26.4 24.9 26.0 24.8 26.0 25.9 25.8 26.7 26.1 24.8 26.8 25.6 26.0 24.8 25.6 26.0 25.1 26.2 25.9 27.1 > 26.2 27.4 > 27.4 26.0 > 26.9 > 26.9 26.6 > 27.1 > 27.2 25.3 25.8 26.0 26.7 25.9 26.8 25.8 25.6 25.8 26.6 > 27.1 > 27.2 > 27.0 27.3 > 26.2 25.5 > 26.9 > 26.9 > 27.1 > 27.1 25.6 25.8 26.5 > 26.2 25.6 25.4 24.5 25.8 24.9 25.9 26.4 26.0 26.0 26.7 25.9 26.7 > 26.2 25.2 > 26.9 > 26.9 26.1 26.1 26.2 > 26.2 25.5 25.5 26.0 26.3 26.1 26.1 25.7 26.4 26.1 26.1 26.2 26.1 27.0 25.6 26.6 25.4 25.4 26.7 > 27.2 > 27.0 25.7 26.5 26.1 26.2 26.0 26.2 26.4 26.4 25.8 26.7 26.1 26.0 2.73 21.38 > 28.6 > 27.8 > 27.4 > 26.2 26.1 > 26.9 > 26.9 > 27.1 27.0 > 27.2 > 27.0 28.10 10.37 21.69 22.51 26.8 28.0 26.4 27.4 26.3 > 26.2 25.4 > 26.9 26.3 26.6 26.2 26.6 26.3 27.2 > 26.2 25.8 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 > 27.0 9.14 22.58 28.4 > 27.8 > 27.4 > 26.2 26.1 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 > 27.0 22.69 > 28.6 27.6 > 27.4 > 26.2 26.1 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 > 27.0 26.2 26.2 > 26.2 25.5 > 26.6 26.3 26.1 26.4 26.4 27.2 > 27.4 > 26.2 25.7 > 26.9 > 26.9 26.7 > 27.1 > 27.2 26.8 > 26.2 25.6 > 26.9 26.7 > 27.1 26.4 27.1 26.1 26.7 26.6 26.4 25.8 26.0 > 26.9 > 26.9 > 27.1 > 27.1 26.9 > 27.0 26.7 > 26.2 25.7 > 26.9 26.4 > 27.1 > 27.1 > 27.2 26.7 IA574 25.4 25.5 25.1 25.8 26.1 26.0 26.4 > 26.2 > 27.2 24.9 24.5 25.7 26.2 25.4 25.4 25.6 26.7 > 27.1 25.5 25.6 26.9 25.5 25.4 25.4 25.4 27.1 > 27.0 18 Table 3. (Continued.) ID ∆RA ∆DEC B R i′ z′ IA527 IA574 IA598 IA624 IA651 IA679 IA709 (arcmin) (arcmin) 3.51 28.4 27.5 > 27.4 > 26.2 > 27.2 26.0 > 26.9 26.9 > 27.1 > 27.2 > 27.0 27.3 > 26.2 > 27.2 25.9 > 26.9 > 27.1 > 27.1 > 27.2 5.62 > 28.6 5.85 6.57 9.03 27.5 27.9 27.2 9.52 > 28.6 11.64 11.96 13.02 13.53 13.66 13.68 14.04 14.03 27.9 27.7 27.6 27.2 27.5 28.2 28.5 27.6 14.36 > 28.6 14.78 16.41 16.47 16.71 17.10 27.7 27.1 27.5 27.8 28.1 17.21 > 28.6 17.71 18.09 18.37 18.54 18.69 18.89 18.96 19.32 26.8 28.4 26.9 27.1 26.7 28.2 27.3 27.5 19.65 28.3 13034 21251 22223 25092 34695 36987 45242 46603 51420 53464 53930 31.65 14.75 11.00 16.05 19.17 15.02 28.40 32.09 28.60 29.48 3.36 54185 18.45 5.22 21.14 11.44 31.72 27.52 15.68 17.41 22.09 26.74 20.31 26.20 8.82 8.84 21.00 25.58 11.48 28.22 28.68 3.03 4.29 24.16 22.42 55740 55786 57174 58770 65463 65839 66933 68446 69065 71041 72707 73613 74326 75127 75930 76201 77116 78959 79754 82975 85436 90651 91865 27.5 26.5 26.5 25.7 27.3 26.6 26.3 25.9 25.7 25.5 26.4 26.4 26.3 26.9 26.1 25.8 25.7 26.7 26.2 27.8 25.4 26.4 > 26.2 26.7 25.6 > 26.9 26.7 > 26.2 > 27.2 25.5 25.8 26.6 27.1 > 26.2 > 27.2 25.7 24.7 25.8 26.4 25.7 26.6 26.5 > 26.2 26.7 25.7 > 26.9 26.3 26.5 25.8 26.5 26.8 25.7 26.8 26.8 25.7 26.9 26.3 26.5 25.7 26.4 > 27.1 27.1 > 27.0 26.7 26.6 26.4 25.5 > 27.2 25.7 26.0 25.8 25.5 26.1 25.6 > 26.2 26.4 > 26.2 > 27.2 26.3 26.0 > 27.2 26.1 > 26.2 26.6 24.4 25.2 25.0 24.8 25.6 25.6 25.5 26.8 26.4 25.9 25.8 25.6 26.5 26.7 26.7 25.8 25.6 25.1 26.0 25.4 26.0 > 27.1 26.5 26.4 25.8 25.7 25.5 26.4 26.5 26.9 26.7 > 27.0 26.0 26.1 25.7 25.8 26.6 26.8 26.3 25.7 25.9 25.7 25.8 26.3 26.7 26.5 27.0 > 26.2 > 27.2 25.8 > 26.9 > 27.1 26.8 > 27.2 > 27.0 25.9 > 26.2 25.7 25.9 26.5 25.6 25.8 26.1 26.3 26.9 26.8 27.1 26.3 > 26.2 > 27.2 25.3 25.2 25.2 25.8 25.5 26.3 25.8 25.0 26.1 26.5 27.3 > 26.2 > 27.2 26.0 > 26.9 25.2 25.2 26.2 24.7 25.4 25.9 25.7 25.7 26.5 26.4 26.8 25.5 26.4 25.7 25.9 26.5 26.1 26.3 25.6 25.9 26.5 26.3 26.2 25.8 25.8 26.8 26.3 27.0 > 27.2 > 27.0 25.3 25.3 25.4 27.7 > 27.4 > 26.2 > 27.2 26.0 > 26.9 > 27.1 > 27.1 > 27.2 > 27.0 25.4 25.6 25.5 27.1 25.9 26.2 27.0 25.5 25.6 25.5 25.7 26.1 25.9 26.4 26.2 26.5 24.5 25.2 24.9 25.5 25.8 25.5 25.4 25.8 25.5 27.1 > 26.2 > 27.2 25.5 > 26.9 > 27.1 25.9 > 26.2 26.2 26.2 > 26.2 > 27.2 25.0 25.4 26.0 26.4 26.8 > 26.2 > 27.2 25.2 > 26.9 25.8 26.2 27.0 25.4 25.7 25.5 26.5 25.9 26.0 25.5 25.9 25.7 25.6 25.6 25.5 26.9 > 27.0 26.1 26.3 25.8 26.7 26.8 26.8 > 27.2 19.82 > 28.6 > 27.8 > 27.4 > 26.2 > 27.2 25.9 > 26.9 > 27.1 > 27.1 > 27.2 > 27.0 20.60 28.5 > 27.8 > 27.4 > 26.2 > 27.2 26.0 > 26.9 26.8 26.6 > 27.2 > 27.0 21.39 28.1 27.4 27.0 > 26.2 > 27.2 25.9 > 26.9 > 27.1 > 27.1 27.1 > 27.0 22.51 > 28.6 > 27.8 > 27.4 > 26.2 9.85 22.81 27.4 25.7 25.5 25.3 19 27.0 27.1 25.8 > 26.9 > 27.1 > 27.1 > 27.2 25.0 25.7 25.7 25.8 25.8 26.7 25.5 Table 3. (Continued.) ID ∆RA ∆DEC B R i′ z′ IA527 IA574 IA598 IA624 IA651 IA679 IA709 (arcmin) (arcmin) 96523 99338 32.25 14.02 23.86 > 28.6 24.48 28.4 26.6 27.1 26.7 > 26.2 > 27.2 25.6 26.2 > 27.1 27.2 > 26.2 > 27.2 25.8 > 26.9 > 27.1 26.3 26.5 27.0 > 27.0 27.0 > 27.0 10519 8.44 12390 15.71 2.93 3.39 3.65 7.15 8.70 9.77 13.66 13.66 14.04 14.76 15.13 15.46 16.46 27.4 27.6 28.0 28.3 28.2 27.5 28.2 27.5 28.4 28.0 28.4 27.4 27.6 7.22 21.87 23.33 16.71 33.03 3.36 5.22 18.42 15.24 31.10 15.68 8.87 16.79 > 28.6 22.09 22.99 27.30 9.98 11.33 32.35 3.23 2.62 33.00 15.97 31.80 18.15 27.74 19.97 19.27 22.64 20.48 > 28.6 20.98 21.70 22.25 22.83 23.61 27.5 27.6 28.2 27.6 28.5 4.91 > 28.6 5.00 > 28.6 5.10 > 28.6 14.13 > 28.6 14.78 > 28.6 15.06 > 28.6 15.45 28.3 15.59 > 28.6 16.40 > 28.6 16.91 > 28.6 25.9 25.3 25.7 26.6 26.3 25.6 26.3 25.5 26.3 25.6 25.5 25.8 26.4 26.7 26.2 26.5 26.0 24.7 26.3 26.2 25.7 25.5 25.4 25.9 26.7 26.1 26.6 26.2 13299 27435 33977 37912 53728 53761 55275 58192 59826 60845 64675 65983 80344 81987 85081 87226 89279 92776 18239 18481 18929 56248 58954 60304 61944 62565 65736 68104 26.0 26.3 25.9 26.5 25.7 25.7 IA598 26.0 > 26.2 25.5 25.4 26.0 > 26.2 26.4 26.7 26.5 26.4 > 26.2 > 27.2 25.7 > 27.2 26.4 25.7 26.4 25.6 25.8 26.4 27.1 > 26.9 25.6 > 26.2 26.4 26.3 25.7 26.0 > 27.2 25.9 26.1 24.8 25.6 26.2 25.2 24.9 25.1 26.2 25.3 25.9 25.4 > 27.1 25.6 25.1 25.5 25.0 25.4 25.1 26.4 25.8 26.9 25.6 26.6 25.7 25.9 25.5 25.9 26.7 25.9 25.4 26.0 26.7 26.5 > 27.2 25.7 26.8 25.6 26.4 25.7 25.6 26.3 25.8 26.8 26.0 26.5 25.5 25.8 26.8 26.2 25.8 26.3 25.8 26.7 26.1 27.7 > 27.4 > 26.2 > 27.2 > 26.9 26.0 > 27.1 > 27.1 > 27.2 > 27.0 25.7 26.1 25.9 25.9 26.5 26.7 26.7 25.3 26.9 > 26.2 27.1 > 26.9 27.0 > 26.2 > 27.2 > 26.9 26.3 > 26.2 > 27.2 26.3 25.8 > 27.2 26.5 25.8 26.1 > 26.2 > 27.2 > 26.9 25.1 25.0 25.5 25.5 25.5 25.0 25.8 26.7 26.9 26.4 25.6 > 27.1 25.2 24.1 26.6 25.1 25.7 25.9 26.6 27.1 26.2 26.4 26.4 25.0 25.9 25.9 27.1 25.9 25.9 26.8 26.8 > 27.0 26.6 26.5 26.6 24.8 24.3 24.4 26.3 27.1 > 27.4 > 26.2 > 27.2 IA651 25.5 26.4 25.9 > 27.1 > 27.1 > 27.2 25.8 25.7 26.5 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 26.5 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 25.5 > 27.2 > 26.9 > 26.9 > 25.3 25.7 > 27.2 > 26.9 25.6 > 26.2 26.5 26.0 > 26.2 > 27.2 26.2 26.8 26.3 26.6 26.6 27.1 > 26.2 > 27.2 > 26.9 > 26.9 26.2 > 26.2 > 27.2 26.8 26.8 25.5 25.8 25.5 26.3 26.3 26.7 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 26.5 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 20 25.3 25.4 25.3 25.1 24.7 25.4 25.9 25.4 25.8 25.6 26.8 26.7 25.9 25.7 25.5 25.9 27.2 26.1 26.9 26.3 26.1 26.3 26.1 24.8 26.7 26.4 27.0 26.4 25.8 25.4 26.1 27.0 26.2 26.5 26.6 ID ∆RA ∆DEC B R i′ z′ IA527 IA574 IA598 IA624 IA651 IA679 IA709 Table 3. (Continued.) (arcmin) (arcmin) 70056 87781 4.12 9.50 17.39 28.4 21.64 28.4 95072 33.02 23.39 > 28.6 96758 97673 98943 9065 11843 16415 18878 19202 24769 33768 46786 47730 52102 57794 60173 62116 64479 65576 74655 89098 9.34 23.84 > 28.6 30.68 14.27 24.03 28.2 24.33 > 28.6 11.03 12.88 12.58 19.37 16.85 18.54 2.80 11.24 19.04 18.29 18.81 24.90 18.75 13.46 13.78 23.99 32.65 3.02 > 28.6 3.78 > 28.6 5.04 > 28.6 5.73 > 28.6 5.83 > 28.6 7.37 > 28.6 9.83 > 28.6 13.33 > 28.6 13.55 > 28.6 14.62 > 28.6 16.05 28.4 16.68 > 28.6 17.17 > 28.6 17.75 > 28.6 18.07 > 28.6 20.40 > 28.6 24.33 > 28.6 26.0 25.7 26.8 26.2 26.1 25.6 26.8 25.1 26.1 26.9 26.7 25.5 26.1 26.3 26.2 26.2 26.6 25.9 26.1 26.3 25.6 27.2 26.2 26.1 25.7 25.5 > 27.2 > 26.9 > 26.9 > 27.1 25.7 27.0 26.6 > 26.9 > 27.1 25.4 25.2 26.3 25.7 26.7 25.8 27.3 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 25.5 > 27.2 > 27.0 26.1 26.0 26.6 > 26.9 > 26.9 > 27.1 26.2 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 25.6 > 26.2 > 27.2 26.5 26.4 26.7 25.3 25.4 25.2 IA679 26.7 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 24.8 24.6 26.3 25.6 26.3 26.0 25.0 26.3 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 26.7 25.7 25.4 25.8 24.1 25.3 26.0 26.5 25.7 26.6 25.1 26.5 26.9 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 26.7 25.8 > 27.0 26.7 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 25.6 25.9 25.5 > 27.2 > 26.9 26.4 > 27.1 26.1 > 27.2 > 26.9 > 26.9 > 27.1 26.9 26.6 26.0 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 25.8 26.3 25.7 > 27.2 > 26.9 > 26.9 > 26.5 > 27.1 26.1 > 27.2 26.7 26.8 27.0 26.0 26.8 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 25.6 25.9 26.2 25.2 25.7 > 27.2 26.7 > 26.9 > 27.1 26.8 26.1 > 27.2 > 26.9 > 26.6 > 27.1 > 27.1 25.7 > 27.2 > 26.9 > 26.6 > 26.3 25.2 > 27.2 26.6 26.2 26.3 26.4 26.4 27.0 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 26.4 26.2 26.8 > 26.9 > 26.9 > 27.1 > 27.1 25.6 24.7 25.4 25.4 25.3 25.4 25.7 25.0 25.1 25.5 24.8 25.9 25.4 IA709 26.7 25.7 25.8 25.9 25.5 25.8 26.7 26.1 26.1 26.0 25.2 26.6 26.2 25.1 25.8 25.5 25.3 24.9 25.3 25.5 25.2 25.0 8190 13683 6.33 3.29 2.65 > 28.6 26.0 25.5 26.0 > 27.2 26.5 > 26.9 > 27.1 26.2 25.9 4.01 > 28.6 > 27.8 27.3 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 18701 13.61 5.25 > 28.6 31658 33815 34592 38261 51167 3.29 21.05 22.59 20.65 8.35 > 28.6 8.91 > 28.6 9.08 > 28.6 9.95 > 28.6 3.36 13.00 > 28.6 55557 13.08 14.03 > 28.6 27.2 27.2 26.2 26.7 26.8 26.9 26.0 26.2 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 26.0 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 25.4 25.3 > 27.2 > 26.9 > 26.9 26.8 > 27.1 > 27.2 26.0 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 26.2 25.8 25.5 26.0 > 27.2 > 26.9 > 26.9 27.0 > 27.1 > 27.2 25.8 > 27.2 > 26.9 > 26.6 > 27.1 > 27.1 > 27.2 26.1 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 25.6 21 Table 3. (Continued.) ID ∆RA ∆DEC B R i′ z′ IA527 IA574 IA598 IA624 IA651 IA679 IA709 (arcmin) (arcmin) 59617 66451 72647 80850 83599 84928 4.65 9.95 25.39 29.77 32.00 11.99 14.98 > 28.6 16.61 > 28.6 18.09 > 28.6 20.09 > 28.6 20.73 > 28.6 21.07 > 28.6 25.9 26.5 27.0 26.6 27.2 26.8 25.4 25.7 25.8 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 25.9 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 27.3 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 26.5 26.2 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 27.2 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 26.6 > 26.2 > 27.2 > 26.9 > 26.9 > 27.1 > 27.1 > 27.2 24.4 25.2 25.8 25.5 25.7 25.2 22 References Ajiki, M., et al. 2002, ApJ, 576, L25 Ajiki, M., et al. 2003, AJ, 126, 2091 Ajiki, M., et al. 2004, PASJ, 56, 597 Barger, A. J., et al. 2001, AJ, 122, 2177 Becker, R. H., et al. 2001, AJ, 122, 2850 Beckwith, S. V. W. et al. 2003, BAAS, 202, 1705 Bell, E. F., et al. 2004, ApJ, 608, 752 Bertin, E., & Arnouts, A. 1996, A&AS, 117, 393 Brocklehurst, M. 1971, MNRAS, 153, 471 Bruzual, A. G., & Charlot, S. 2003, MNRAS, 344, 1000 Bunker, A. J., Stanway, E. R., Ellis, R. S., McMahon, R. G. 2004, MNRAS, 355, 374 Calzetti, D., Armus, L., Bohlin, R. C., Kinney, A. L, Koornneef, J., & Storchi-Bergmann, T. 2000, ApJ, 533, 2 Cohen, J. G., Hogg, D. W., Blandford, R., Cowie, L. L., Hu, E., Songalia, A., Shopbell, P., & Richberg, K. 2000, ApJ, 545, 32 Coleman, G. D., Wu, C.-C., & Weedman, D. W. 1980, ApJS, 43, 393 Connolly, A. J., Szalay, A. S., Dickinson, M., Subbarao, M. U., & Brunner, R. J. 1997, ApJ, 486, 11 Cowie, L. L., Songalia, A., Hu, E. M., & Cohen, J. G. 1996, AJ, 112, 839 Cowie, L. L., & Hu, E. M. 1998, AJ, 115,1319 Cuby, J.-G., Le Fevre, O., McCracken, H., Cuillandre, J. C., Magnier, E., Meneux, B. 2003, A&A, 405, 19 Dawson, S., Stern, D., Bunker, A. J., Spinrad, H., & Dey, A. 2001, AJ, 122, 598 Dey, A., Spinrad, H., Stern, D., Graham, J. R., & Chaffee, F. H. 1998, ApJ, 498, L93 Fan, X., Strauss, M. A., Schneider, D. P., Becker, R. H., White, R. L., Narayanan, V. K., Richards, G. T., Pentericci, L., et al. 2002, AAS, 20111414 Fioc, M., & Rocca-Volmerange, B. 1997, A&A, 326, 950 Fujita, S. S., et al. 2003, AJ, 125, 13 Giavalisco, M., Dickinson, M., Ferguson, H. C., Ravindranath, C., Kretchmer, C., Moustakas, L. A., Madau, P., Fall, S. M., et al. 2004, ApJ, 600, L103 Haiman, Z. & Spaans, M. 1999, ApJ, 518, 138 Hayashino, T., et al. 2000, SPIE, 4008, 397 Hayashino, T., et al. 2003, PNAOJ, 7, 33 Hu, E. M., McMahon, R. G., Egami, E. 1996, ApJ, 459, L53 Hu, E. M., Cowie, L. L., McMahon, R. G., Capak, P., Iwamuro, F., Kneib, J. -P., Maihara, T., & Motohara, K. 2002, ApJ, 568, L75 Hu, E. M., Cowie, L. L., Capak, P., McMahon, R. G., Hayashino, T., & Komiyama, Y. 2004, AJ, 127, 563 Iwata, I., Ohta, K., Tamura, N., Ando, M., Wada, S., Watanabe, C., Akiyama, M., & Aoki, K. 2003, PASJ, 55, 415 Iye, M., et al. 2004, PASJ, 56, 381 23 Kaifu, N., et al. 2000, PASJ, 52, 1 Kajisawa, M., Yamada, T. 2001, PASJ, 53, 833 Kajisawa, M., et al. 2000, PASJ, 52, 53 Kauffmann, G., Haehnelt, M. 2000, MNRAS, 311, 576 Keel, W. C., Cohen, S. H., Windhorst, R. A., & Waddington, I. 1999, AJ, 118, 2547 Kennicutt, R. 1998, ARA&A, 36, 189 Kodaira, K., et al. 2003, PASJ, 55, L17 Kudritzki, R. -P., et al. 2000, ApJ, 536, 19 Kurk, J. D., Cimatti, A., di Serego Alighieri, S., Vernet, J., Daddi, E., Ferrara, A., & Ciardi, B. 2004, A&A, 422, L13 Landolt, A. U. 1992, AJ, 104, 340 Lanzetta, K. M., Yahil, A., & Fern´andez-Soto, A. 1996, Nature, 381, 759 Leitherer, C. & Heckman, T. M. 1995, ApJS, 96, 9 Lilly, S. J., LeF`evre, O., Hammer, F., & Crampton, D. 1996, ApJ, 460, L1 Loeb, A., & Barkana, R. 2001, ARA&A, 39, 19 Lowenthal, J. D., Koo, D. C., Guzman, R., Gallego, J., Phillips, A. C., Faber, S. M., Vogt, N. P., Illingworth, G. D., & Gronwall, C. 1997, ApJ, 481, 637 Madau, P., Ferguson, H. C., Dickinson, M. E., Giavalisco, M., Steidel, C. C., & Fruchter, A. 1996, MNRAS, 283, 1388 Martin, C. L. & Sawicki, M. 2004, ApJ, 603, 414 Miyazaki et al. 2002, PASJ, 54, 833 Oke, J. B. 1974, ApJS, 27, 21 Oke, J. B. 1990, AJ, 99, 1621 Osterbrock, D. E. 1989, Astrophysics of Gaseous Nebulae and Active Galactic Nuclei. University Science Books, Sausalito, CA Ouchi, M., et al. 2001, ApJ, 558, L83 Ouchi, M., Shimasaku, K., Okamura, S., Furusawa, H., Kashikawa, N., Ota, K., Doi, M., Hamabe, M., et al. 2004, ApJ, 611, 660 Partridge, R. B., & Peebles, P. J. E. 1967, ApJ, 147, 868 Pettini, M., Shapley, A. E., Steidel, C. C., Cuby, J. -G., Dickinson, M., Moorwood, A. F. M., Adelberger, K. L., & Giavalisco, M. 2001, ApJ, 554, 981 Pirzkal, N., et al. 2004, ApJS, 154, 501 Pritchet, C. J. 1994, PASP, 106, 1052 Rhoads, J. E., Malhotra, S., Dey, A., Stern, D., Spinrad, H., Jannuzi, B. T. 2000, ApJ, 545, L85 Rhoads, J. E., Dey, A., Malhotra, S., Stern, D., Spinrad, H., Jannuizi, B. T., Dawson, S., Brown, M. J. I., & Landes, E. 2003, AJ, 125, 1006 Santos, M. R., Ellis, R. S., Kneib, J. -P., Richard, J., & Kuijken, K. 2004, ApJ, 606, 683 Shapley, A. E., Steidel, C., Adelberger, K. L., Dickinson, M., Giavalisco, M., & Pettini, M. 2001, ApJ, 562, 95 Shapley, A. E., Steidel, C. C., Pettini, M., & Adelberger, K. L. 2003, ApJ, 588, 65 Shimasaku, K., et al. 2004, ApJ, 605, L93 24 Spergel, D. N., Verde, L., Peiris, H. V., Komatsu, E., Nolta, M. R., Bennett, C. L., Halpern, M., Hinshaw, G., et al. 2003, ApJS, 148, 175 Spinrad, H. 2003, "Astrophysics Update" in press, Mason, J. (Ed.) Springer Praxis Books in Astrophysics and Astronomy Steidel, C. C., Hamilton, D. 1992, AJ, 104, 941 Steidel, C. C., Giavalisco, M., Dickinson, M., & Adelberger, K. L. 1996a, AJ, 112, 352 Steidel, C. C., Giavalisco, M., Pettini, M., Dickinson, M., & Adelberger, K. L. 1996b, ApJ, 462, L17 Steidel, C. C., Adelberger, K. L., Giavalisco, M., Dickinson, M., & Pettini, M. 1999, ApJ, 519, 1 Steidel, C. C., Adelberger, K. L., Shapley, A. E., Pettini, M., Dickinson, M., & Giavalisco, M. 2000, ApJ, 532, 170 Stone, R. P. S. 1996, ApJS, 107, 423 Taniguchi, Y. 2004, Studies of Galaxies in the Young Universe with New Generation Telescope, Proceedings of Japan-German Seminar, held in Sendai, Japan, July 24-28, 2001, Eds.: N. Arimoto and W. Duschl, p. 107 Taniguchi, Y. et al. 2003a, ApJ, 585, L97 Taniguchi, Y. et al. 2003b, JKAS, 36, 123 (erratum 36, 283) Taniguchi, Y., et al. 2005, PASJ, 57, 165 Thompson, D., & Djorgovski, S. 1995, AJ, 110, 3 Weymannm, R. J., et al. 1998, ApJ, 505, L95 Williams, R. E., et al. 1996, AJ, 112, 1335 Williams, R. E., et al. 2000, AJ, 120, 2735 Wolf, C., Meisenheimer, K., Rix, H.-W., Borch, A., Dye, S., Kleinheinrich, M. 2003a, A&A, 401, 73 Wolf, C., Wisotzki, L., Borch, A., Dye, S., Kleinheinrich, M., Meisenheimer, K. 2003b, A&A, 408, 499 Yagi , M., Kashikawa, N., Sekiguchi, M., Doi, M., Yasuda, N., Shimasaku, K., & Okamura, S. 2002, AJ, 123, 66 Yahata, N., Lanzetta, K. M., Turnshek, D. A., & Oke, J. B. 2000, ApJ, 538, 493 Yan, H., Windhorst, R. A. 2004, ApJ, 612, L93 25 Table 4. Properties of LAE candidates ID EWobs EWrest (A) (A) fLyα LLyα Lν(U V ) MU V SF RLyα SF RU V (erg s−1 cm−2) (erg s−1) (erg s−1 Hz−1) (mag) (M⊙yr−1) (M⊙yr−1) 7048 7988 9442 9620 11523 11538 13069 15427 15823 17533 19286 112.7 509.4 89.5 484.4 284.2 397.5 212.6 989.4 333.6 330.0 412.9 26.0 117.5 20.6 111.7 65.6 91.7 49.0 228.2 76.9 76.1 95.2 IA527 6.97E-17 6.95E+42 3.96E-17 3.95E+42 2.56E-17 2.56E+42 2.17E-17 2.17E+42 3.62E-17 3.61E+42 3.61E-17 3.60E+42 2.92E-17 2.91E+42 2.77E-17 2.76E+42 2.73E-17 2.72E+42 3.28E-17 3.27E+42 3.18E-17 3.17E+42 27691 1926.8 444.4 4.14E-17 4.13E+42 28584 28996 29083 37812 38785 41668 44636 47969 49443 51844 54027 55020 55100 55848 57223 57844 59074 59405 60224 60455 60664 112.8 254.9 292.7 174.3 182.3 336.0 159.5 520.4 432.4 536.8 181.6 558.4 316.0 984.6 632.3 175.1 689.7 204.7 319.9 154.8 259.9 26.0 58.8 67.5 40.2 42.0 77.5 36.8 120.0 99.7 123.8 41.9 128.8 72.9 227.1 145.8 40.4 159.1 47.2 73.8 35.7 59.9 2.21E-17 2.20E+42 5.38E-17 5.37E+42 2.33E-17 2.32E+42 2.94E-17 2.93E+42 2.75E-17 2.74E+42 2.52E-17 2.52E+42 2.64E-17 2.64E+42 1.27E-16 1.27E+43 2.36E-17 2.35E+42 3.78E-17 3.77E+42 3.42E-17 3.41E+42 2.53E-17 2.52E+42 2.53E-17 2.52E+42 1.04E-16 1.03E+43 2.78E-17 2.77E+42 2.75E-17 2.74E+42 2.36E-17 2.36E+42 5.80E-17 5.78E+42 3.52E-17 3.51E+42 3.17E-17 3.16E+42 2.20E-17 2.19E+42 26 1.32E+29 1.65E+28 6.10E+28 9.56E+27 2.71E+28 1.93E+28 2.92E+28 5.96E+27 1.74E+28 2.12E+28 1.64E+28 4.58E+27 4.17E+28 4.50E+28 1.70E+28 3.60E+28 3.21E+28 1.60E+28 3.53E+28 5.21E+28 1.16E+28 1.50E+28 4.02E+28 9.64E+27 1.71E+28 2.24E+28 9.36E+27 3.34E+28 7.30E+27 6.04E+28 2.34E+28 4.36E+28 1.80E+28 -21.2 -18.9 -20.4 -18.4 -19.5 -19.1 -19.6 -17.8 -19.0 -19.2 -18.9 -17.6 -20.0 -20.0 -19.0 -19.8 -19.7 -18.9 -19.8 -20.2 -18.6 -18.9 -19.9 -18.4 -19.0 -19.3 -18.3 -19.7 -18.1 -20.4 -19.3 -20.0 -19.1 6.3 3.6 2.3 2.0 3.3 3.3 2.6 2.5 2.5 3.0 2.9 3.8 2.0 4.9 2.1 2.7 2.5 2.3 2.4 11.6 2.1 3.4 3.1 2.3 2.3 9.4 2.5 2.5 2.1 5.3 3.2 2.9 2.0 18.4 2.3 8.5 1.3 3.8 2.7 4.1 0.8 2.4 3.0 2.3 0.6 5.8 6.3 2.4 5.0 4.5 2.2 4.9 7.3 1.6 2.1 5.6 1.4 2.4 3.1 1.3 4.7 1.0 8.5 3.3 6.1 2.5 Table 4. (Continued.) ID EWobs EWrest (A) (A) fLyα LLyα Lν(U V ) MU V SF RLyα SF RU V (erg s−1 cm−2) (erg s−1) (erg s−1 Hz−1) (mag) (M⊙yr−1) (M⊙yr−1) 63293 69085 70551 71679 72094 72647 74055 74299 75353 77460 78162 80388 81380 82145 82756 83157 86220 87141 87553 88785 89485 228.1 478.3 236.5 297.5 107.3 610.5 687.6 695.1 499.8 188.5 152.7 700.2 124.5 290.3 692.1 391.4 240.8 319.6 489.7 170.3 196.1 52.6 110.3 54.5 68.6 24.8 140.8 158.6 160.3 115.3 43.5 35.2 161.5 28.7 67.0 159.6 90.3 55.5 73.7 112.9 39.3 45.2 2.27E-17 2.27E+42 2.12E-17 2.11E+42 3.78E-17 3.77E+42 5.03E-17 5.02E+42 2.09E-17 2.08E+42 4.07E-17 4.05E+42 2.97E-17 2.96E+42 2.50E-17 2.50E+42 3.30E-17 3.29E+42 2.78E-17 2.78E+42 7.24E-17 7.23E+42 2.92E-17 2.91E+42 2.35E-17 2.34E+42 2.58E-17 2.58E+42 4.31E-17 4.30E+42 9.34E-17 9.32E+42 2.34E-17 2.33E+42 5.82E-17 5.80E+42 5.07E-17 5.06E+42 2.41E-17 2.40E+42 2.59E-17 2.58E+42 89896 1722.7 397.3 2.95E-17 2.95E+42 90874 94639 362.4 850.1 94941 1062.3 95272 96625 99496 100145 101147 102766 6580 10969 11379 665.4 227.3 667.9 426.9 379.6 341.3 124.7 381.4 420.0 83.6 196.1 245.0 153.4 52.4 154.0 98.4 87.5 78.7 26.4 80.7 88.9 3.97E-17 3.96E+42 3.57E-17 3.56E+42 2.69E-17 2.68E+42 2.44E-17 2.43E+42 2.90E-17 2.89E+42 3.53E-17 3.52E+42 3.34E-17 3.33E+42 2.28E-17 2.27E+42 2.79E-17 2.78E+42 IA574 2.99E-17 3.88E+42 8.44E-17 1.09E+43 2.86E-17 3.71E+42 27 2.12E+28 9.44E+27 3.41E+28 3.61E+28 4.15E+28 1.42E+28 9.21E+27 7.67E+27 1.41E+28 3.15E+28 1.01E+29 8.89E+27 4.02E+28 1.90E+28 1.33E+28 5.09E+28 2.07E+28 3.88E+28 2.21E+28 3.01E+28 2.82E+28 3.65E+27 2.33E+28 8.95E+27 5.39E+27 7.80E+27 2.72E+28 1.12E+28 1.67E+28 1.28E+28 1.74E+28 7.23E+28 6.68E+28 2.06E+28 -19.2 -18.3 -19.7 -19.8 -19.9 -18.8 -18.3 -18.1 -18.8 -19.6 -20.9 -18.3 -19.9 -19.1 -18.7 -20.2 -19.2 -19.9 -19.3 -19.6 -19.5 -17.3 -19.3 -18.3 -17.7 -18.1 -19.5 -18.5 -19.0 -18.7 -19.0 -20.6 -20.5 -19.2 2.1 1.9 3.4 4.6 1.9 3.7 2.7 2.3 3.0 2.5 6.6 2.6 2.1 2.3 3.9 8.5 2.1 5.3 4.6 2.2 2.4 2.7 3.6 3.2 2.4 2.2 2.6 3.2 3.0 2.1 2.5 3.0 1.3 4.8 5.0 5.8 2.0 1.3 1.1 2.0 4.4 14.2 1.2 5.6 2.7 1.9 7.1 2.9 5.4 3.1 4.2 3.9 0.5 3.3 1.3 0.8 1.1 3.8 1.6 2.3 1.8 2.4 3.5 10.0 3.4 10.1 9.4 2.9 Table 4. (Continued.) fLyα LLyα Lν(U V ) MU V SF RLyα SF RU V (erg s−1 cm−2) (erg s−1) (erg s−1 Hz−1) (mag) (M⊙yr−1) (M⊙yr−1) ID EWobs EWrest (A) (A) 13034 21251 22223 25092 34695 36987 45242 796.8 895.4 352.1 311.1 390.0 805.4 388.8 168.7 189.6 74.6 65.9 82.6 170.5 82.3 2.57E-17 3.33E+42 2.93E-17 3.80E+42 2.81E-17 3.64E+42 2.59E-17 3.36E+42 6.87E-17 8.90E+42 3.08E-17 4.00E+42 2.84E-17 3.69E+42 9.74E+27 9.88E+27 2.41E+28 2.51E+28 5.31E+28 1.16E+28 2.21E+28 3.00E+28 4.43E+28 5.29E+28 6.02E+28 2.65E+28 2.83E+28 2.93E+28 1.72E+28 3.48E+28 4.84E+28 5.23E+28 2.05E+28 3.32E+28 6.14E+27 7.14E+28 8.05E+27 6.90E+28 5.50E+28 6.05E+28 1.43E+28 4.19E+28 3.32E+28 1.57E+28 1.59E+27 7.80E+27 1.12E+28 7.91E+27 5.29E+28 -18.4 -18.4 -19.4 -19.4 -20.2 -18.6 -19.3 -19.6 -20.0 -20.2 -20.4 -19.5 -19.5 -19.6 -19.0 -19.8 -20.1 -20.2 -19.2 -19.7 -17.9 -20.5 -18.2 -20.5 -20.3 -20.4 -18.8 -20.0 -19.7 -18.9 -16.4 -18.1 -18.5 -18.1 -20.2 3.0 3.5 3.3 3.1 8.1 3.6 3.4 14.6 4.3 4.7 6.2 3.2 3.2 3.3 3.4 4.3 3.5 3.1 3.1 3.4 3.5 6.1 3.4 9.3 2.9 4.8 5.0 6.0 3.8 7.3 4.3 3.3 3.4 4.0 5.3 1.4 1.4 3.4 3.5 7.4 1.6 3.1 4.2 6.2 7.4 8.4 3.7 4.0 4.1 2.4 4.9 6.8 7.3 2.9 4.6 0.9 10.0 1.1 9.7 7.7 8.5 2.0 5.9 4.7 2.2 0.2 1.1 1.6 1.1 7.4 46603 1243.6 263.3 1.24E-16 1.61E+43 51420 53464 53930 54185 55740 55786 57174 58770 65463 65839 66933 68446 247.0 225.6 265.3 307.9 285.8 286.1 505.4 316.0 184.0 152.8 383.0 258.9 52.3 47.8 56.2 65.2 60.5 60.6 3.62E-17 4.70E+42 3.96E-17 5.13E+42 5.29E-17 6.86E+42 2.71E-17 3.51E+42 2.68E-17 3.48E+42 2.78E-17 3.60E+42 107.0 2.88E-17 3.73E+42 66.9 39.0 32.3 81.1 54.8 3.65E-17 4.73E+42 2.95E-17 3.82E+42 2.65E-17 3.44E+42 2.61E-17 3.38E+42 2.85E-17 3.69E+42 69065 1450.1 307.0 2.95E-17 3.82E+42 71041 219.1 46.4 5.18E-17 6.72E+42 72707 1092.7 231.4 2.91E-17 3.78E+42 73613 74326 75127 75930 76201 77116 343.1 135.7 204.0 893.4 364.5 291.9 72.6 28.7 43.2 189.2 77.2 61.8 7.85E-17 1.02E+43 2.47E-17 3.20E+42 4.09E-17 5.30E+42 4.23E-17 5.49E+42 5.06E-17 6.56E+42 3.21E-17 4.17E+42 78959 1189.9 251.9 6.17E-17 8.00E+42 79754 6840.6 1448.4 3.61E-17 4.69E+42 82975 1087.8 85436 769.0 90651 1303.0 91865 255.4 230.3 162.8 275.9 54.1 2.81E-17 3.64E+42 2.86E-17 3.71E+42 3.41E-17 4.42E+42 4.47E-17 5.80E+42 28 Table 4. (Continued.) ID EWobs EWrest (A) (A) fLyα LLyα Lν(U V ) MU V SF RLyα SF RU V (erg s−1 cm−2) (erg s−1) (erg s−1 Hz−1) (mag) (M⊙yr−1) (M⊙yr−1) 96523 99338 426.4 637.8 90.3 135.1 3.23E-17 4.20E+42 2.92E-17 3.79E+42 2.29E+28 1.38E+28 -19.3 -18.8 10519 12390 13299 27435 33977 37912 53728 53761 55275 58192 262.0 110.9 214.0 547.0 265.3 181.9 302.9 169.6 388.4 175.6 53.1 22.5 43.4 110.9 53.8 36.9 61.4 34.4 78.7 35.6 IA598 3.34E-17 4.94E+42 2.61E-17 3.86E+42 3.27E-17 4.83E+42 3.92E-17 5.80E+42 2.49E-17 3.68E+42 3.17E-17 4.69E+42 2.73E-17 4.03E+42 3.11E-17 4.61E+42 3.42E-17 5.05E+42 2.97E-17 4.39E+42 59826 1162.9 235.7 2.93E-17 4.33E+42 60845 64675 65983 80344 81987 85081 87226 89279 92776 149.3 299.1 383.9 570.6 261.4 381.6 321.7 224.3 578.8 30.3 60.6 77.8 115.7 53.0 77.3 65.2 45.5 2.84E-17 4.20E+42 4.34E-17 6.42E+42 3.20E-17 4.73E+42 3.61E-17 5.34E+42 2.70E-17 3.99E+42 2.85E-17 4.21E+42 3.88E-17 5.74E+42 8.82E-17 1.30E+43 117.3 2.48E-17 3.66E+42 IA624 15802 1150.5 224.7 3.76E-17 6.21E+42 16965 18537 18866 26804 30834 31226 283.3 196.6 398.9 230.9 666.1 125.6 55.3 38.4 77.9 45.1 130.1 24.5 3.03E-17 5.00E+42 2.54E-17 4.19E+42 2.32E-17 3.82E+42 2.48E-17 4.10E+42 2.72E-17 4.49E+42 2.84E-17 4.69E+42 33170 1607.5 314.0 3.84E-17 6.34E+42 38991 40098 128.9 532.4 25.2 104.0 2.88E-17 4.75E+42 4.31E-17 7.11E+42 29 4.58E+28 8.46E+28 5.50E+28 2.58E+28 3.37E+28 6.27E+28 3.24E+28 6.60E+28 3.16E+28 6.07E+28 9.06E+27 6.84E+28 5.22E+28 3.00E+28 2.28E+28 3.71E+28 2.69E+28 4.34E+28 1.41E+29 1.54E+28 1.36E+28 4.45E+28 5.38E+28 2.42E+28 4.48E+28 1.70E+28 9.43E+28 9.96E+27 9.29E+28 3.37E+28 -20.1 -20.7 -20.3 -19.4 -19.7 -20.4 -19.7 -20.4 -19.7 -20.4 -18.3 -20.5 -20.2 -19.6 -19.3 -19.8 -19.5 -20.0 -21.3 -18.9 -18.7 -20.0 -20.2 -19.4 -20.0 -19.0 -20.8 -18.4 -20.8 -19.7 3.8 3.5 4.5 3.5 4.4 5.3 3.4 4.3 3.7 4.2 4.6 4.0 3.9 3.8 5.8 4.3 4.9 3.6 3.8 5.2 11.9 3.3 5.7 4.5 3.8 3.5 3.7 4.1 4.3 5.8 4.3 6.5 3.2 1.9 6.4 11.8 7.7 3.6 4.7 8.8 4.5 9.2 4.4 8.5 1.3 9.6 7.3 4.2 3.2 5.2 3.8 6.1 19.8 2.2 1.9 6.2 7.5 3.4 6.3 2.4 13.2 1.4 13.0 4.7 Table 4. (Continued.) ID EWobs EWrest (A) (A) fLyα LLyα Lν(U V ) MU V SF RLyα SF RU V (erg s−1 cm−2) (erg s−1) (erg s−1 Hz−1) (mag) (M⊙yr−1) (M⊙yr−1) 52800 53168 56258 57053 59547 60562 61041 61778 62001 63836 68957 69429 83307 489.3 465.8 670.8 355.1 178.6 207.4 228.5 459.3 220.5 492.5 116.2 195.6 922.2 93077 1171.9 99353 99689 553.8 571.6 18239 18481 18929 56248 58954 60304 61944 62565 65736 68104 70056 87781 656.0 597.6 181.8 241.2 459.4 267.7 657.8 340.9 465.6 410.6 254.0 173.1 95.6 91.0 131.0 69.4 34.9 40.5 44.6 89.7 43.1 96.2 22.7 38.2 180.1 228.9 108.2 111.6 122.7 111.8 34.0 45.1 85.9 50.1 2.30E-17 3.80E+42 2.28E-17 3.76E+42 3.31E-17 5.45E+42 2.55E-17 4.21E+42 2.73E-17 4.50E+42 2.20E-17 3.64E+42 2.81E-17 4.64E+42 2.39E-17 3.95E+42 3.51E-17 5.79E+42 6.35E-17 1.05E+43 2.34E-17 3.86E+42 2.86E-17 4.71E+42 5.19E-17 8.56E+42 3.99E-17 6.58E+42 4.03E-17 6.66E+42 3.58E-17 5.91E+42 IA651 4.16E-17 7.79E+42 3.90E-17 7.30E+42 2.19E-17 4.10E+42 3.30E-17 6.18E+42 6.68E-17 1.25E+43 2.70E-17 5.06E+42 123.0 2.48E-17 4.64E+42 63.8 87.1 76.8 47.5 32.4 2.82E-17 5.29E+42 2.41E-17 4.51E+42 2.65E-17 4.96E+42 2.48E-17 4.64E+42 2.44E-17 4.57E+42 95072 1327.7 248.3 4.26E-17 7.98E+42 96758 97673 98943 312.4 369.1 160.0 58.4 69.0 29.9 2.97E-17 5.56E+42 3.06E-17 5.73E+42 2.32E-17 4.35E+42 IA679 30 1.96E+28 2.04E+28 2.05E+28 2.99E+28 6.36E+28 4.43E+28 5.12E+28 2.17E+28 6.63E+28 5.37E+28 8.38E+28 6.08E+28 2.34E+28 1.42E+28 3.03E+28 2.61E+28 3.13E+28 3.22E+28 5.94E+28 6.75E+28 7.18E+28 4.98E+28 1.86E+28 4.09E+28 2.55E+28 3.18E+28 4.81E+28 6.96E+28 1.58E+28 4.69E+28 4.09E+28 7.17E+28 -19.1 -19.2 -19.2 -19.6 -20.4 -20.0 -20.2 -19.2 -20.5 -20.2 -20.7 -20.4 -19.3 -18.8 -19.6 -19.4 -19.6 -19.7 -20.3 -20.5 -20.6 -20.1 -19.1 -19.9 -19.4 -19.7 -20.1 -20.5 -18.9 -20.1 -19.9 -20.5 3.5 3.4 5.0 3.8 4.1 3.3 4.2 3.6 5.3 9.5 3.5 4.3 7.8 6.0 6.1 5.4 7.1 6.6 3.7 5.6 2.7 2.9 2.9 4.2 8.9 6.2 7.2 3.0 9.3 7.5 11.7 8.5 3.3 2.0 4.2 3.7 4.4 4.5 8.3 9.4 11.4 10.1 4.6 4.2 4.8 4.1 4.5 4.2 4.2 7.3 5.1 5.2 4.0 7.0 2.6 5.7 3.6 4.5 6.7 9.7 2.2 6.6 5.7 10.0 Table 4. (Continued.) ID EWobs EWrest (A) (A) fLyα LLyα Lν(U V ) MU V SF RLyα SF RU V (erg s−1 cm−2) (erg s−1) (erg s−1 Hz−1) (mag) (M⊙yr−1) (M⊙yr−1) 9065 11843 16415 18878 19202 24769 33768 46786 47730 52102 57794 60173 62116 64479 65576 74655 89098 8190 13683 18701 31658 33815 34592 38261 51167 55557 59617 66451 72647 80850 83599 84928 476.5 268.6 492.9 615.4 606.7 398.4 198.0 265.8 174.4 458.6 583.6 265.6 327.7 273.6 137.0 624.1 469.6 154.1 978.3 321.4 267.5 177.6 272.0 309.8 225.2 176.5 460.3 171.7 907.7 286.9 946.8 809.3 85.4 48.1 88.3 110.3 108.7 71.4 35.5 47.6 31.2 82.2 2.29E-17 4.87E+42 7.76E-17 1.65E+43 3.56E-17 7.56E+42 2.55E-17 5.41E+42 2.97E-17 6.30E+42 5.62E-17 1.19E+43 2.12E-17 4.50E+42 2.48E-17 5.27E+42 2.01E-17 4.27E+42 3.19E-17 6.77E+42 104.5 2.60E-17 5.51E+42 47.6 58.7 49.0 24.6 111.8 84.1 26.5 168.0 55.2 45.9 30.5 46.7 53.2 38.7 30.3 79.0 29.5 155.9 49.3 162.6 139.0 3.59E-17 7.61E+42 3.45E-17 7.31E+42 2.20E-17 4.67E+42 2.74E-17 5.80E+42 2.25E-17 4.77E+42 3.14E-17 6.66E+42 IA709 2.02E-17 4.83E+42 2.59E-17 6.20E+42 2.21E-17 5.30E+42 2.40E-17 5.75E+42 2.78E-17 6.66E+42 2.37E-17 5.68E+42 2.18E-17 5.21E+42 2.28E-17 5.45E+42 2.45E-17 5.88E+42 6.84E-17 1.64E+43 1.96E-17 4.68E+42 2.43E-17 5.83E+42 2.02E-17 4.84E+42 2.81E-17 6.73E+42 4.08E-17 9.76E+42 31 2.81E+28 1.69E+29 4.22E+28 2.42E+28 2.86E+28 8.24E+28 6.25E+28 5.46E+28 6.74E+28 4.06E+28 2.60E+28 7.88E+28 6.14E+28 4.70E+28 1.17E+29 2.10E+28 3.90E+28 9.00E+28 1.82E+28 4.73E+28 6.17E+28 1.08E+29 6.00E+28 4.83E+28 6.94E+28 9.56E+28 1.02E+29 7.83E+28 1.84E+28 4.84E+28 2.04E+28 3.46E+28 -19.5 -21.5 -20.0 -19.4 -19.6 -20.7 -20.4 -20.2 -20.5 -19.9 -19.4 -20.6 -20.4 -20.1 -21.1 -19.2 -19.9 -20.8 -19.1 -20.1 -20.4 -21.0 -20.4 -20.1 -20.5 -20.9 -20.9 -20.6 -19.1 -20.1 -19.2 -19.8 4.4 15.0 6.9 4.9 5.7 10.9 4.1 4.8 3.9 6.2 5.0 6.9 6.7 4.3 5.3 4.3 6.1 4.4 5.6 4.8 5.2 6.1 5.2 4.7 5.0 5.3 14.9 4.3 5.3 4.4 6.1 8.9 3.9 23.6 5.9 3.4 4.0 11.5 8.8 7.6 9.4 5.7 3.6 11.0 8.6 6.6 16.3 2.9 5.5 12.6 2.5 6.6 8.6 15.1 8.4 6.8 9.7 13.4 14.3 11.0 2.6 6.8 2.9 4.8 1 IA527 574 624 598 679 709 651 i i n o s s m s n a r T 0.5 0 5000 1 6000 7000 B RC i' z' 0.5 0 6000 4000 10000 Wavelength (angstrom) 8000 Fig. 1. Transmission curves of the filters used in this study. Lower panel shows the four broad band filters and upper panel shows a close up for the seven IA filters. 32 Fig. 2. The entire filed of view of the SXDS field. Our MAHOROBA field (34.′71 × 27.′20) is a sky area in the southern part of the SXDS field while that of Fujita et al. (2003a) is located in the central part(13.′7 × 13.′7). 33 Fig. 3. The stacked image obtained with the seven IA filters. 34 Fig. 4. Left : Number counts per arcmin2 in 0.2 mag bin. Right : The detection completeness of artificial galaxies using the simulation. Filled squares show the results from the data sets using the exponential law. Open squares show that of the de Vaucouleurs law. 35 Fig. 5. Color-magnitude diagrams for objects with IA <27 at each IA filter. Red dotted horizontal line corresponds to the criterion of EWrest = 20A. Red dashed line corresponds to 3σ of D[IA]. Red solid curve corresponds to 3σ of D[IA] − IA. Blue open circles show LAE candidates. 36 Fig. 6. The spatial distributions of LAE candidates. The gray shadow show the masked area. The three large area in these figure are masked for the starlight contamination. 37 Fig. 7. EWobs distribution of LAE candidates found in each IA-excess catalog. Vertical line in each panel corresponds to mag(EWrest = 20A). In the panel for IA574, one large EWobs objects is not shown; its equivalent width is EWobs = 6840 A. 38 Fig. 8. EW0 distribution of LAE candidates in each IA-excess catalog. Vertical line in each panel corresponds to mag(EWrest = 20A). In the panel for IA574, one large EWobs objects is not shown; its equivalent width is EWobs = 1448 A. 39 ) 3 − c p M ( ) 7 6 . 2 4 > A y L g o l ( n g o l −3 −4 −5 −6 3 4 z 5 Fig. 9. Number densities of LAE candidates are shown as a function of redshift. The horizon error bar simply shows the redshift coverage of each IA filter. 40 Fig. 10. Luminosity distributions of LAE candidates in each IA-excess catalog. The lower right panel shows the results for all LAE candidates. The vertical lines show the limit of L(Lyα) at each filter. 41 Fig. 11. Luminosity distributions of the absolute U V magnitude for LAE candidates; 1 σ errors are also shown. Filled squares in the lower-right panel show the results for z ∼ 5.7 LAEs (Hu et al. 2004). Solid line in the same panel show the UV luminosity function for LBGs at z ∼ 4 (Ouchi et al. 2004). 42 Fig. 12. Comparison between SF R(Lyα) and SF R(UV) for LAEs in each IA-excess catalog. The following relations are also shown; 1) SF R(Lyα) = SF R(UV) (dashed line), 2) SF R(Lyα) = 2×SF R(UV) (dotted line), 3) SF R(Lyα) = 0.5 × SF R(UV) (dash-dotted line), and 4) the fitting result (solid line). The lower-right panel shows the results for all LAEs.43 1 0 > ) V U R F S ( / ) A y L ( R F S g o < l −1 3 4 z 5 Fig. 13. The average ratio of SF R(Lyα) to SF R(UV) is shown As a function of redshift. Our results are shown by filled squares. 44 Fig. 14. Upper panel shows the cosmic evolution of the star formation rate density as a function of redshift. Our results are shown by large open circles (for all LAEs). Results of previous LAE surveys are also shown; The data sources are Taniguchi et al. (2005, asterisk), Ouchi et al. (2004, double diamond), Fujita et al. (2003a, double triangle), Cowie & Hu (1998, double circle), Kudritzski et al. (2000, double square). All the data above show the lower limit of SFRD. We therefore show these data with the up-arrow. The data using Lyman break method are shown by the points with error of SF RD. Those based on LBGs are also shown; Steidel et al. (2000, open inverse triangles), Madau et al. (1998, filled triangles), Connolly et al. (1997, open triangles), Lilly et al. (1996, open circles), Iwata et al. (2003, filled inverse triangle), Giavalisco et al. (2003, open diamonds). The other data sources of SFRD are Gallego et al. (1996, plus), Tresse & Maddox (1998, cross), Fujita et al. (2003b, open square), Treyer et al. (1998, filled square). Lower panel shows the cosmic evolution of the star formation rate density as a function of redshift. We compare the lower limit of SFRD derived from our data. The large open circles show the sum of SFRD for all LAEs and the large filled circles show the sum of SFRD for LAEs with log(Lyα) > 42.67. 45 Fig. 15. Comparison between zmodel and zphot for simulated catalogs selected as emission-line galaxies. 46
0804.4139
4
0804
2009-01-14T13:37:40
The spectroscopic orbit and the geometry of R Aqr
[ "astro-ph" ]
R Aqr is one of the closest symbiotic binaries and the only D-type system with radial velocity data suitable for orbital parameters estimation. The aims of our study are to derive a reliable spectroscopic orbit of the Mira component, and to establish connections between the orbital motion and other phenomena shown by R Aqr. We reanalize velocity data recently published by McIntosh & Rustan complemented by additional velocities. We find an eccentric orbit (e=0.25) with a period 43.6 yr. This solution is in agreement with a resolved VLA observation of this system. We demonstrate that the last 1974-1981 increase of extinction towards the Mira occured during its superior spectroscopic conjunction, and can be due to obscuration by a neutral material in the accreting stream. We also show that jet ejection is not connected with the orbital position.
astro-ph
astro-ph
Astronomy&Astrophysicsmanuscript no. raqrfinal November 6, 2018 c(cid:13) ESO 2018 9 0 0 2 n a J 4 1 ] h p - o r t s a [ 4 v 9 3 1 4 . 4 0 8 0 : v i X r a The spectroscopic orbit and the geometry of R Aqr M. Gromadzki1 and J. Mikołajewska1 Nicolaus Copernicus Astronomical Center, Bartycka 18, 00-716 Warszawa, Poland e-mail: marg,[email protected] Received October ??, ????; accepted ???? ??, ???? ABSTRACT Context. R Aqr is one of the closest symbiotic binaries and the only D-type system with radial velocity data suitable for orbital parameter estimation. Aims.The aims of our study are to derive a reliable spectroscopic orbit of the Mira component, and to establish connections between the orbital motion and other phenomena exhibited by R Aqr. Methods. We reanalyze and revise the velocity data compiled by McIntosh & Rustan complemented by additional velocities. Results.We find an eccentric orbit (e = 0.25) with a period 43.6 yr. This solution is in agreement with a resolved VLA observation of this system. We demonstrate that the last increase in extinction towards the Mira variable in 1974 -- 1981 occurred during its superior, spectroscopic conjunction, and can be due to obscuration by a neutral material in the accreting stream. We also show that jet ejection is not connected with the orbital position. Key words. stars: binaries: symbiotic -- stars:binaries:spectroscopic -- stars: individual: R Aqr 1. Introduction R Aqr is a symbiotic binary surrounded by an hourglass neb- ula (Solf & Ulrich 1985), and both its cool and hot components are unique. The cool component is a Mira variable with a pulsa- tion period of 387d with SiO and H2O maser emission, which is rare, with only three among 48 symbiotics with Miras exhibit- ing this emission (Whitelock 2003). The hot component shows sporadic, jet ejections (e.g. Kellogg et al. 2007 and Nichols et al. 2007), and Nichols et al. (2007) also detected a 1734 s peri- odic oscillation in Chandra X-ray observations, which suggested that the hot companion is a magnetic white dwarf. Willson et al. (1981) interpreted a brightness reduction in pulsation maxima as the eclipsing of the Mira by a companion surrounded by an extended gas cloud. This interpretation was supported by two facts. First, during the last hypothetical eclipse (1974 -- 1980), a minimum was observed in the near-IR light curve similar to that corresponding to dust obscuration in other symbiotic Miras (e.g., Whitelock et al. 1983). Secondly, a spectrum obtained during the first such event (1886 -- 1894) exhibits only faint emission lines of hydrogen and no trace of M type spectral features (Townley et al. 1928). In the historical, visual, light curve, three phases of reduced maxima are present spaced at 44 yr intervals and this value was adopted by Willson et al. (1981) as the orbital period. Additionally, the O − C diagram showed variation with a period of ∼22 yr. Wallerstein (1986) used a number of spectral absorp- tion lines associated with the Mira in an attempt to determine the orbital elements of R Aqr. His results were consistent with the 44 yr period, but uncertain. Hinkle et al. (1989) combined these archival data with radial velocities (RV) determined from near- IR CO and Ti I lines, and obtained a 44 yr period and a highly eccentric, e = 0.6, orbital solution. Send offprint requests to: M. Gromadzki McIntosh & Rustan (2007) collected radial velocity data of SiO maser emission of R Aqr (Lepine et al. 1978; Cohen & Ghigo 1980; Spencer et al. 1981; Lane 1982; Hall et al. 1990; Cho et al. 1996; Boboltz 1997; Hollis et al. 2000; Alcolea et al. 1999; Pardo et al. 2004; McIntosh & Rustan 2007) com- plemented by previous radial velocities derived from optical absorption lines (Merrill 1950; Jacobsen & Wallerstein 1975; Wallerstein 1986) and near-IR CO and Ti I lines (Hinkle et al. 1989). All of these features are associated with the cool compo- nent or material in its close neighborhood and reflect its motion around the mass center. In astrophysical masers, emission orig- inating in many maser spots moves around the Mira variable. Spectra of masers exhibit a series of emission lines with vari- ous velocities associated with different maser spots. McIntosh (2006) studied changes in the velocity centroid (VC) of SiO maser emission from 76 single evolved giants, and showed that the mean difference in the VCs measured at different epochs was only 0.065 ± 2.00 km s−1, and that the VC of the emission dis- tribution can be used to derive the velocity of the star. Based on this result, McIntosh & Rustan (2007) adopted the VC in their work for all kinds of velocity data analysis. McIntosh & Rustan (2007) also estimated orbital elements using a program developed by Gudehus (2001). They inferred an eccentric orbit, e = 0.52, with a period 34.6 yr. Their orbital period was significantly shorter than any previous estimates as well as the period derived in our present study. Their other were of reasonable accuracy if we interpret their system mass, equal to 0.043 M , as the mass function, and their semi-major axis of the system, of 3.7 AU, as the semi-major axis of the orbit of the Mira component. We also note that not all points listed in their Table 1 were plotted on their figures showing VC versus orbital phase, namely, they did not measurement from Alcolea et al. (1999). In the text, the authors wrote that points from Hollis et al. (2000) are listed in their Table 1, although they were not ⊙ 2 M. Gromadzki and J. Mikołajewska: The spectroscopic orbit and the geometry of R Aqr displayed there. These data points were plotted on the figures showing VC versus Julian days, but they were missing on the figure of VC versus orbital phase. The authors also did not com- pare the orbit solution with the resolved image of the binary (Hollis et al. 1997), nor include radial velocity measurements be- fore 1946 (Merrill 1935, 1950). These points were also ignored in all previous papers about the orbit estimation from radial ve- locities (Wallerstein 1986, Hinkle et al. 1989). The authors also neglected radial velocity measurements of the SiO maser by Zuckerman et al. (1979), Hollis et al. (1986), Hall et al. (1987), Allen et al. (1989), Heske (1989), Jewell et al. (1991), Patel et al. (1992), Gray et al. (1995), Schwarz et al. (1995), Hollis et al. (2000, 2001), Imai et al. (2001), Cotton et al. (2004, 2006), and Kang et al. (2006). We also note that the observations stud- ied by Pardo et al. (2004) included the observations analyzed by Alcolea et al. (1999), which included the data used by Martinez et al. (1988). These three data sets are thus not entirely different, and the later studies were based simply on data collected over longer time. This paper contains an analysis of the revised radial velocity data listed in Table 1 of McIntosh & Rustan (2007) and addi- tional velocity data (listed here in Table 1), as well as discussion of the connections between the orbital motion and other phe- nomena shown by R Aqr. 2. Orbit determination The data from Table 1 of McIntosh & Rustan (2007) were re- vised and supplemented by additional radial velocity measure- ments from Merrill (1935, 1950), Zuckerman (1979), Martinez et al. (1988), Jewell et al. (1991), Schwarz et al. (1995), Hollis et al. (2000, 2001), Cotton et al. (2004, 2006), and Kang et al. (2006). The blue-violet absorption lines in Mira variables show con- stant shift of between −5 km s−1 and −10 km s−1 with respect to near-IR velocities (Hinkle et al. 1984; and references therein). In our study, a shift equal to −6 km s−1 was adopted. The blue- violet absorption lines do not show any variation with pulsation period and do not need any additional correction. Absorption lines from the red part of the optical spectrum exhibit signifi- cant variation but there is no way of correcting for this effect in a single measurement. For this reason, we did not use three data points from Jacobsen & Wallerstein (1975), which had been previously used for the spectroscopic orbit determinations by Wallerstein (1986), Hinkle et al. (1989), and McIntosh & Rustan (2007). In the near-IR, the radial velocity (RV) variations of R Aqr were dominated by stellar pulsations (∆V ∼ 40 km s−1). To re- move changes due to the pulsations, Hinkle et al. (1989) com- plete a linear fit to the infrared velocity curve between phase 0.1 and 0.8, and then adopted as the stellar velocities the resid- uals remaining after subtracting the velocity-curve fit from each measurement plus the center-of-mass velocity of the Mira of −29.1 km s−1. The value of the center-of-mass velocity was es- timated from the CO ∆ν = 3 velocity at phase 0.36, following the results of Hinkle, Scharlach & Hall (1984). Unfortunately, Hinkle et al. (1989) used an old ephemeris of Campbell (1955). Since R Aqr displays significant variation in its pulsation period (see Fig. 1), we repeated these corrections with the ephemeris JD(max) = 2 442 404.2(±0.4) + 388.1(±0.1)× E, derived for the 1975 -- 1989 period covering Hinkle et al. (1989) observations. To calculate the orbital solution all data points were averaged in 387-day bins corresponding to the pulsation period of the Mira component. The velocity changes due to the orbital motion are negligible during such an interval. The data obtained from differ- ent methods were binned separately. Since different maser tran- sitions are formed at different distances from the stellar surface, they were also binned separately. We excluded single measure- ments for various maser transitions (Allen et al. 1989; Patel et al. 1992; Gray et al. 1995; Imai et al. 2001) obtained for epochs covered by more extensive data sets. In Table 1, we list the radial velocity data used to derive the orbital elements of R Aqr and then converted to local standard of rest (LS R) velocities, which are also plotted in Fig. 1. We also decided to exclude from our analysis the SiO maser data obtained during the period JD2 446 000 -- 47 500 (data from Hollis et al. 1986, Heske 1989, Alcolea et al. 1999, and Hall et al. 1990) because the maser emission showed at that time signif- icantly different behaviour e.g. a few fainter emission, irregular contours than in later epochs, as well as the masers in isolated Miras (see Pardo et al. 2004). The SiO masers in R Aqr display a ring-like morphology ∼ 31 mas (∼ 6.2[d/200 pc] AU) in diam- eter (Boboltz et al. 1998). So, these masers lie very close to the stellar surface, namely at a distance of only ∼ 1.5 radii from the Mira (see Sect. 3) in a region dominated by stellar pulsations and permeated by circumstellar shocks. In addition, the SiO masers can be seriously affected by tidal interactions during the perias- tron passage when the size of the SiO maser region became com- parable to the Roche lobe radius of the Mira. In fact, according to our orbital solution (see below) the Alcolea et al. (1999) ob- servations were obtained around periastron. The deep minimum in both VCs and flux of maser emission that appeared around JD2 446 700 -- 47 300, coincided with the jet formation (Nichols et al. 2007) and a rise in the UV continuum (Meier & Kafatos 1995). We suspect that the SiO masers may have been partly dis- rupted by rising UV radiation during the jet formation, and tidal interaction during the periastron passage and then gradually re- build. To identify the spectroscopic orbit, we first estimated a pe- riod of RV variations of 43.6 yr using Ortfit (based on the method of Schwarzenberg-Czerny 1996), and then calculated the orbital parameters: T0, the time of periastron passage, γ, the baricentral velocity, K, the semi-amplitude, ω, the longitude of periastron, and e, the eccentricity with fixed period. We applied Bertiau's program (1967) based on the Lehmenn-Filh´es method. As a re- sult, we obtained an eccentric, e = 0.25±0.07, orbit with reason- able errors (Table 2). Figure 2 displays the data and the best-fit velocity curve versus orbital phase. Figure 1 also compares the RV variations with the O − C data for the Mira pulsations based visual data collected by am- ateur observers from AAVSO. The O -- C is a plot showing the difference between the observed (O) times of maximum light and the values calculated according to an adopted ephemeris (C) as a function of time. A description of the O -- C method can be found in a review by Zhou (2003). To derive a new pulsation ephemeris, we first estimated the time of each maximum in the light curve by fitting a third-order Fourier polynomial. Then a linear ephemeris: JD(max) = 2 416 070 ± 4 + 387.30 ± 0.07 × E was fitted to the obtained times of maxima. The resulting O-C diagram shows a parabolic shape that corresponds to an increase in the period rate of 0.013±0.002 day per cycle. After subtracting this parabolic trend, a sinusoidal variation with a period of 22.5 years and amplitude of about 8 days (Fig. 1) is visible. M. Gromadzki and J. Mikołajewska: The spectroscopic orbit and the geometry of R Aqr 3 1920 1920 1920 1920 1920 1940 1940 1940 1940 1940 1960 1960 1960 1960 1960 1980 1980 1980 1980 1980 2000 2000 2000 2000 2000 20 20 20 20 20 10 10 10 10 10 0 0 0 0 0 -10 -10 -10 -10 -10 -20 -20 -20 -20 -20 -30 -30 -30 -30 -30 ] ] ] ] ] s s s s s y y y y y a a a a a d d d d d [ [ [ [ [ - - - - - C C C C C O O O O O ] ] ] ] ] 1 1 1 1 1 - - - - - s s s s s m m m m m k k k k k [ [ [ [ [ V V V V V R R R R R -22 -22 -22 -22 -22 -26 -26 -26 -26 -26 -30 -30 -30 -30 -30 -34 -34 -34 -34 -34 20000 20000 20000 20000 20000 30000 30000 30000 30000 30000 40000 40000 40000 40000 40000 JD-2400000 JD-2400000 JD-2400000 JD-2400000 JD-2400000 50000 50000 50000 50000 50000 Fig. 1. The O − C diagram (top) and RV variations (bottom) for R Aqr. Observations from different sources are marked like in Table 1. The solid curve represents the fitted orbital elements (Table 2). The dotted curve represents the orbital model proposed by Hollis et al. (1997). The dashed curve represents solution, with modified ω = 87 which better fits the VLA observations. Bars show times of hypothetic eclipse, short arrows represent moments of jet ejection (Nichols et al. 2007), whilst long arrows show periastron passage. Table 2. Orbital elements for R Aqr. ties in the orbital parameters (see Table 2). The mass function sets a lower limit to the hot companion mass. Element Porb (day) γ (km s−1) K (km s−1) e ω (deg) T0 Σ(O − C)2 σ (km s−1) Value 15 943±471 -24.9±0.2 4.0±0.4 0.25±0.07 106±19 2 444 019±728 40 1.11 3. Discussion Our orbital solution yields the semi-major axis of the Mira of KP 2π and the mass function of ag sin i = √1 − e2 = 5.68+0.85 −0.83 AU, f (M) = 1 2πG · PK3(1 − e2)3/2 = (Mh sin i)3 (Mh + Mg)2 = 0.096+0.042 −0.032M ⊙, where Mg, and Mh represent masses of the giant and the hot component, respectively, and the errors are set by the uncertain- ⊙ , and Mh = 0.5 -- 1.4 M For further discussion, we assume that the symbiotic binary R Aqr consist a Mira variable and a white dwarf (WD) compan- ion with an orbit inclined at i = 70◦ (∼ 72◦ according to Solf & Urlich 1985). Assuming plausible components masses, Mg = 1 -- 1.5 M , for the Mira and the WD, respec- tively, the total mass of system Mtot = Mh + Mg should be within the range 1.5 -- 3 M . The component masses are also limited by the orbital solution. Therefore, if we consider f (M) = 0.064 -- 0.138 M , ⊙ Mh = 0.57 -- 1.02 M , and the mass ratio q = Mg/Mh = 1.2 -- 2.1 (Fig. 3). These values are in good agreement with the commonly adopted model of symbiotic binaries. The semi-major axis of the system, a, is 14.2 -- 16.8 AU, which corresponds to 71 -- 84 mas on the sky for a distance of d = 200 pc (see below). , we derive Mtot = 1.6 -- 2.5 M ⊙ ⊙ ⊙ and Mg = 1 -- 1.5 M ⊙ ⊙ Our orbital solution can be tested and refined by using re- solved observations of R Aqr (Hege et al. 1991; Hollis et al. 1997). In particular, the binary components of R Aqr were re- solved by VLA observations (Hollis et al. 1997) obtained on 20 Nov 1996 (JD=2 450 407). The measured separation between the VLA image in SiO maser v = 1, J = 1 − 0 transition and the continuum emission in a 50 MHz bandwidth at 43.165 GHz as- sociated with an Hii region surrounding the hot companion was 4 M. Gromadzki and J. Mikołajewska: The spectroscopic orbit and the geometry of R Aqr Table 1. Radial velocities of R Aqr used for orbital elements determination. JD 2422255 2423039 2429581 2430336 2432235.5 2433019.3 2441237 2442939 2443527.5 2444078 2444356 2444509 2444748 2445535 2445574.5 2445862 2445890.6 2446378.5 2447247 2447338 2447634 2447870 2448240 2448283 2448696 2448998.4 2449435 2449439 2449804 2450068.8 2450407 2450948 2451390.4 2451785.8 2451892 2453253 2453784.3 2454112 RV [km s−1] -21.6 -22.0 -30.7 -30.2 -28.5 -29.0 -21.7 -22.0 -26.6 -25.9 -27.0 -25.6 -30.1 -27.7 -27.6 -30.1 -28.0 -27.8 -28.5 -28.9 -28.5 -27.1 -27.0 -28.3 -27.4 -26.4 -27.0 -26.3 -27.0 -26.8 -24.0 -24.0 -25.6 -23.2 -24.3 -23.0 -22.7 -22.8 Spectral range Symbol Visual Visual Visual Visual Visual Visual Visual Near-IR Visual Visual Near-IR Visual Near-IR Near-IR v = 1, J = 1 − 0 v = 1, J = 1 − 0 v = 1, J = 1 − 0 v = 1, J = 1 − 0 Near-IR v = 1, J = 1 − 0 v = 1, J = 1 − 0 v = 1, J = 1 − 0 v = 1, J = 2 − 1 v = 1, J = 1 − 0 v = 1, J = 1 − 0 v = 1, J = 2 − 1 v = 1, J = 1 − 0 v = 1, J = 2 − 1 v = 1, J = 1 − 0 v = 1, J = 1 − 0 v = 1, J = 1 − 0 v = 1, J = 1 − 0 v = 1, J = 2 − 1 v = 1, J = 2 − 1 v = 1, J = 1 − 0 v = 1, J = 1 − 0 v = 1, J = 1 − 0 v = 1, J = 1 − 0 G# G# H# H# H# H# (cid:8) > + (cid:4) × (cid:7) (cid:7) > N (cid:7) > > > △ △ H  H H  H  H ♦ 2 2 D C ⊕ 4 ▽ ▽ References Merrill 1935 Merrill 1935 Merrill 1950 Merrill 1950 Merrill 1950 Merrill 1950 Jacobsen & Wallerstein 1975 Hinkle et al. 1989 Lepine et al. 1978, Zuckerman 1979 Cohen & Ghigo 1980, Spencer et al. 1981, Lane 1982 Cho et al. 1996, Jewell et al. 1991 Lane 1982 Wallerstein 1986 Wallerstein 1986 Hinkle et al. 1989 Wallerstein 1986 Hinkle et al. 1989 Hinkle et al. 1989 Martinez et al. 1988 Hinkle et al. 1989 Martinez et al. 1988 Pardo et al. 2004 Schwarz et al. 1995 Pardo et al. 2004 Pardo et al. 2004 Schwarz et al. 1995 Pardo et al. 2004 Schwarz et al. 1995 Pardo et al. 2004 Boboltz et al. 1997 Hollis et al. 2000 Hollis et al. 2000 Cotton et al. 2006 McIntosh & Rustan 2007 McIntosh & Rustan 2007 Kang et al. 2006, Hollis et al. 2000 Kang et al. 2006 Hollis et al. 2001, Cotton et al. 2004, McIntosh & Rustan 2007 ρ = 55 ± 2 mas (11[d/200 pc] AU), and the position angle was θ = 18◦ ± 2◦. This can be compared with the projected com- ponent separation on the sky plane calculated from our orbital elements. The orbital elements (Table 2) combined with the or- bital inclination of i = 70◦ (Solf & Urlich 1985) inferred the projected separation of ρ = 0.44a ∼ 31 -- 37 [d/200 pc]−1 mas, and the difference between the apparent binary position angle and the position angle of the line of nodes (the binary orienta- tion on sky) of θ − Ω = −109◦. Our predicted component separation on the sky is consis- tent with the VLA observations only if the distance is somewhat lower than 200 pc or we underestimate the ρ value due to uncer- tainties in our orbital elements. The predicted value of ρ should in fact increase, changing ω (see also Fig. 4), and/or increasing e, and/or decreasing T0. For example, ρ should increase to 0.65a (46 -- 55 [d/200 pc]−1 mas) just by adopting a lower value of ω = 87, i.e. lower only by an amount approximately equal to the standard-deviation uncer- tainty of ω. The distance to R Aqr is also relatively uncertain, and pub- lished estimates based on various methods are in the range 180 -- 260 pc (e.g. 180 pc by Solf & Ulrich 1985; 181 pc by Lepine et al. 1978; 240 pc using the most recent period-luminosity relation from Whitelock et al. 2008; and 260 pc by Baade 1943, 1944). The Hipparcos parallax of R Aqr is 5.07±3.15 mas (Perryman et al. 1993), which corresponds to a distance of 197+323 −75 pc. Our determination of the parameters of the spectroscopic orbit is thus in a good agreement (within ≤ 1-σ errors in the orbital parame- ters) with those inferred from the VLA resolved observations. The value of θ = 18◦, measured with the VLA, implies that Ω = 127◦. Thus, the position of the binary orbit on the sky repro- duces well the general picture of the nebula around R Aqr, i.e. the orbital plane is perpendicular to the jets (see Fig. 7 in Solf & Urlich 1985). The orbital parameters derived for the speckle observations (Hege et al. 1991) completed on 16 Oct 1983 are inconsistent with our orbit in any case, because the measured separation, ρ = 124 ± 2 mas which corresponding to 25[d/200 pc] AU, is too large for any orbital phase. Hege et al. (1991) probably did not detect the Mira companion but rather an Hα emission region in the SW counterjet. Our orbital solution predicts the times of spectroscopic con- junctions at Tconj I = 2 450 859 (Feb 1998), Tconj II = 2 443 658 (May 1978) for the inferior ( the Mira is in front of the hot component), and the superior (the Mira behind) conjunction, re- spectively. The superior conjuction coincides with the last dust obscuration/eclipse (see also Fig. 4). Although the conjunction times could differ by up to 1 -- 2 years due to uncertainties in the M. Gromadzki and J. Mikołajewska: The spectroscopic orbit and the geometry of R Aqr 5 ] ] ] 1 1 1 - - - s s s m m m k k k [ [ [ s s s e e e r r r ] ] ] ] 1 1 1 1 - - - - s s s s m m m m k k k k [ [ [ [ V V V V R R R R 2 2 2 -2 -2 -2 -22 -22 -22 -22 -26 -26 -26 -26 -30 -30 -30 -30 0.0 0.0 0.0 0.0 0.5 0.5 0.5 0.5 1.0 1.0 1.0 1.0 Orbital phase Orbital phase Orbital phase Orbital phase 1.5 1.5 1.5 1.5 Fig. 2. Bottom: The RV data and fitted orbital curve (Table 2) against orbital phase. Symbols are the same as in Fig 1. The solid curve represent our orbital solution. The dashed curve represents solution, with modified ω = 87 which better fits the VLA obser- vations. Top: Difference between the RV data and our orbital fit (solid line in the lower pannel). ] ] ] O• O• O• M M M [ [ [ h h h M M M 1.2 1.2 1.2 1.0 1.0 1.0 0.8 0.8 0.8 0.6 0.6 0.6 0.4 0.4 0.4 0.8 0.8 0.8 0 . 1 3 8 0 . 1 3 8 0 . 1 3 8 0 . 0 9 6 0 . 0 9 6 0 . 0 9 6 0 . 0 6 4 0 . 0 6 4 0 . 0 6 4 1.0 1.0 1.0 1.2 1.2 1.2 1.4 1.4 1.4 1.6 1.6 1.6 1.8 1.8 1.8 Mg [MO• ] Mg [MO• ] Mg [MO• ] Fig. 3. The permitted component masses constrained by the ob- served mass function as well as its maximum and minimum val- ues (solid lines labeled with the values of f (M)), and assuming i = 70◦. Dotted and dashed lines show the solution for i = 75◦, and i = 90◦, respectively. orbital parameters, especially T0, e, and ω, the obscuration lasted about 6 years, and the coincidence between these two events is obvious. ⊙ Adopting plausible binary parameters, i.e. the mass ratio q ∼ 1.6, Mtot = 2 M , and a = 15 AU (see above), we esti- mate the minimum component separation to be a(1−e) ∼ 11AU, and the Roche lobe radius for the Mira, to be RL ∼ 4 AU. The near-IR interferometry of R Aqr gave the diameter of the Mira expressed as a diameter of uniform disk, ΘUD = 14.06 -- 20.8 mas in K, 17.7 mas in J, and 17.7 -- 19.07 mas in H, in intermedi- ate and minimum pulsation phase, respectively (van Belle et al. 1996; Millan-Gabet et al. 2005; Ragland et al. 2006). These val- ues correspond to the average radius of the Mira component, Rg ∼ 2 AU. Even during periastron passage, the Mira variable therefore remains relatively far from filling the Roche lobe. Many detached interacting binaries, including the symbiotic binaries as well as many X-ray binaries show evidence for much higher mass transfer rate than predicted by models of spherically symmetric winds. This phenomenon can be accounted for by wind focused towards the compact component. Podsiadlowski & Mohamed (2007) proposed a model for o Cet binary sys- tem that can explain this focusing. In their model, a slow wind from Mira fills its Roche lobe and then the matter falls -- via the L1 Lagrangian point -- into an accreting stream onto an accre- tion disk around the companion. This model is even more rele- vant to R Aqr where the component separation is significantly smaller than that in o Cet. In particular, it could explain the in- crease in extinction towards the Mira during the last "eclipse" in 1974-1981 caused by a neutral material in the accreting stream. According to our orbital solution, the streaming material can ob- scure the Mira at that time, and in addition, the proximity of periastron (see Fig. 4) could give rise to enhanced mass transfer. A similar enhanced, wind-obscuration scenario for R Aqr was discussed by Mikołajewska & Kenyon (1992), although in that case the source of the increased mass loss was proposed to be ) of the a helium flash above a massive core (Mcore ∼ 1.36M ⊙ Mira. The orbital configuration of R Aqr could also explain the ob- servations of the OH and H2O masers that originate in circum- stellar regions more distant from the Mira variable (e.g. H2O masers zone is 10 -- 20 AU) than the SiO maser emission region. The OH and H2O maser lines were not detected in 1979 and 1984 (Cohen & Ghigo 1980; Norris et al. 1984) when the hot companion was in front of Mira. Ivison et al. (1994) reported detection of weak OH and H2O maser emission in 1993, when the Mira was in front of the hot companion and the masers were shielded from its radiation. They claimed, however, that the OH maser detection was only tentative, and that the velocity mea- surement was very unusual (an expansion velocity of 10 km s−1 is rather large for this system), and only the red peak should have been detected, and additional spectra obtained using the VLA during 1994 June did not show the OH maser line (Ivison et al. 1998). The H2O line appeared more promising, although weak for a Mira variable with relatively strong SiO masers. It was also clearly detected during 3 epochs in 1993 -- 1995 (Ivison et al. 1998). Ivison et al. (1998) proposed three possible mech- anisms that could inhibit the OH and H2O maser formation in their natural environment, involving either the orbital motion of the companion or its UV radiation and fast wind, or line obscu- ration at low frequencies by optically thick, ionized gas. The apparent correlation between the orbital position and the O−C for the Mira pulsations is extremely interesting. In particu- lar, minima in the O−C diagram occur around the periastron (see Fig 1). This may be due to some distortion of the Mira caused by tidal interaction during the periastron passage, for example if the Mira became elongated towards the hot component, the pulsation may need more time for propagation in this direction. According to our orbital model, the main maximum of O − C appears close to the inferior conjunction, when the Mira eclipses the hot companion, whereas the secondary, lower maximum ap- pears in the middle of the two conjunctions, when both compo- nents are well separated and we can see the elongated side. Finally, the orbital solution proposed is inconsistent with the 17 yr orbital period for R Aqr proposed by Nichols et al. (2007). M. Gromadzki and J. Mikołajewska: The spectroscopic orbit and the geometry of R Aqr 6 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 E2 E2 E2 E2 E2 E2 E2 E2 S S S S S S J1 E1 E1 E1 E1 E1 E1 E1 E1 E1 J2 To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer To observer M1 M1 M1 V V V V V V V V Mira as obscuration by the accretion stream. We also showed that the jet ejection (Nichols et al. 2007) is not connected with the orbital position. Finally, we note that September 2012 should be a perfect time to complete resolved imaging of R Aqr, since the Mira component will then be at minimum brightness and the separation between the components on sky will be close to its maximum possible value. Acknowledgements. This study made use of the American Association of Variable Star Observers (AAVSO) International Database contributed by ob- servers worldwide. We are grateful to the anonymous referee for valuable com- ments. We also thank Radek Smolec for comments to the first version of this paper and Wojtek Pych for providing the Ortfit software. This work was partly supported by the Polish Research grant N203 395534. M2 M2 M2 References Alcolea, J., Pardo, J.R., Bujarrabal, V., et al. 1999, A&AS, 139, 461 Allen, D.A., Hall, P.J., Norris, R.P., et al. 1989, MNRAS, 236, 363 Baade, W. 1943, Carnegie Inst. Washington Yearb., 42, 17 Baade, W. 1944, Carnegie Inst. Washington Yearb., 43, 12 Bertiau, F.C. 1967, IAUS, 30, 227 Boboltz, D. 1997, Ph.D. thesis, Virginia Polytechnic Univ. Boboltz, D., Diamond, P.J., & Kemball, A.J. 1998, ASPC 144, 2438 Campbell L. 1955, Studies of LongPeriod Variables, AAVSO Pub. Cho, S.-H., Kaifu, N., & Ukita, N. 1996, A&AS, 115, 117 Cohen, N. L., & Ghigo, F. D. 1980, AJ, 85, 451 Cotton, W.D., Mennesson, B., Diamond, P.J., et al. 2004, A&A, 414, 275 Cotton, W.D., Vlemmings, W., Mennesson, B., et al. 2006, A&A, 456, 339 Gray, M.D., Ivison, R.J., Yates, J.A., et al. 1995, MNRAS, 277, 67 Gudehus, D. 2001, BAAS, 33, 850 Hall, P.J., Robin, M., Wark, R.M., & Wright, A.E. 1987, Proc. Astron. Soc. Aus., 7, 50 Hall, P. J., Allen, D. A., Troup, E. R., Wark, R. M., & Wright, A. E. 1990, MNRAS, 243, 480 Hege, E. K., Allen, C. K., & Cocke, W. J. 1991, ApJ, 381, 543 Heske, A. 1989, A&A, 208, 77 Hinkle, K., Scharlach, W., & Hall, D. N. 1984, ApJS, 56, 1 Hinkle, K., Wilson, T., Scharlach, W., & Fekel, F. 1989, AJ, 98, 1820 Hollis, J. M., Michalitsianos, A. G., Kafatos, M., Wright, M. C. H., & Welch, W. J. 1986, ApJ, 309, L53 Hollis J. M., Pedelty, J. A., & Lyon, R. G. 1997, ApJ, 482, L85 Hollis, J. M., Pedelty, J. A., Forster, J., et al. 2000, ApJ, 543, L81 Hollis, J., Boboltz, D., Pedelty, J., White, S., & Forster, J. 2001, ApJ, 559, L37 Ivison, R. J., Seaquist, E. R., & Hall, P. J. 1994, MNRAS, 269, 218 Imai, H., Miyoshi, M., Ukita, N., et al. 2001, PASJ, 53, 259 Ivison, R. J., Yates, J. A., & Hall, P.J. 1998, MNRAS, 295, 813 Jacobsen, T. S., & Wallerstein, G. 1975, PASP, 87, 269 Jewell, P.R., Snyder, L.E., Walmsley, C.M., Wilson, T.L., & Gensheimer, P.D.1991, A&A, 242, 211 Kang, J., Cho, S.-H., Kim, H.-G., et al. 2006, ApJSS, 165, 360 Kellogg, E., Anderson, C., Korreck, K., et al. 2007, ApJ, 664, 1079 Lane, A. 1982, Ph.D. thesis, Univ. Massachusetts Lepine, J., LeSqueren, A., & Scalise, E. 1978, ApJ, 225, 869 Martinez, A., Bujarrabal, V., & Alcolea, J. 1988, A&AS, 74, 273 McIntosh, G. C. 2006, AJ, 132, 1046 McIntosh, G. C., & Rustan, G. 2007, AJ, 134, 55 Meier, S. R., & Kafatos, M., 1995, ApJ, 451, 359 Merrill, P. W. 1935, ApJ, 81, 312 Merrill, P. W. 1950, ApJ, 112, 514 Mikołajewska, J., & Kenyon, S. J. 1992, MNRAS, 256, 177 Mikołajewska, J. 1999, in "Optical and Infrared Spectroscopy of Circumstellar Matter", ASP Conference Series Vol. 188, eds. E. Guenther & B. Stecklum, p.291 Millan-Gabet, R., Pedretti, E., Monnier, J.D., et al. 2005, ApJ, 620, 961 Nichols, J. S., DePasquale, J., Kellogg, E., et al. 2007, ApJ, 660, 651 Norris, R. P., Haynes, R. F., Wright, A. E., & Allen, D. A. 1984, Proc. Astron. Soc. Aust., 5, 562 Pardo, J. R., Alcolea, J., Bujarrabal, V., et al. 2004, A&A, 424, 145 Patel, Nimesh A., Joseph, Antony, & Ganesan, R. 1992, J. Astrophys. Astron., 13, 241 Perryman, M.A.C., Lindegren, L., Kovalevsky, J., at al. 1997, A&A, 323, 49 Posiadlowski, Ph., & Mohamed, S. 2007, in "Evolution and chemistry of symbi- otic star, binary post-AGB and related objects", eds. J Mikołajewska, & R. Szczerba, Baltic Astronomy, 16, 26 Ragland, S., Traub, W. A., Berger, J.-P., et al. 2006, ApJ, 652, 650 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 -1.5 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Fig. 4. The orbit of the Mira (+) and hot (×) component in the R Aqr binary system in steps of ∆φ = 0.1. In this representa- tion, the stars move anti-clockwise. The dotted circle represents the Mira boundary at φ = 0 (periastron passage). The solid dot marks the mass center. Axes are in units of the semi-major axis a. The positions of Mira at the beginning and end of eclipse is marked by E1, and E2, respectively whereas the main and secondary maxima in O − C diagram are denoted by M1 and M2, respectively. J1 and J2 mark the two jet ejection episodes. Open squares connected with dotted line and marked by V show the component position during the VLA observation (Hollis et al. 1997). Open dots connected with dotted line and marked by S show the component position during the speckle observation (Hege et al. 1991). They noticed that radio observations obtained in 1987 and 2004 exhibited non-thermal spectra, which are usually interpreted to be a signature of a newly emitted jet (in stellar sources older jets are generally thermal), suggested that the jet formation is cou- pled with periastron. In our opinion, such a short period appears unrealistic not only because it is not reflected by the RV variation but also because in that case there would not be sufficient room for a Mira with a dust envelope. The 17-yr period implies a semi- major axis of ∼ 8 AUa whereas the minimum component sepa- ration for D-type symbiotic binaries is about 10 -- 15 AU, which for typical masses (see above) indicates orbital periods longer than 20 yrs (Mikołajewska 1999). It appears that the jet ejection in R Aqr connot be connected with the orbital motion. 4. Conclusions Based on published radial velocities of its Mira component, we have derived new orbital parameters for the symbiotic binary R Aqr. We have found in particular, the orbital period of 43.6 yr, and showed that the mass function is consistent with the pres- ence of a typical, 1 -- 1.5 M , Mira variable accompanied by a 0.6 -- 1 M white dwarf. We also showed that our spectroscopic orbit is consistent with the VLA astrometry (Hollis et al. 1997) and inconsistent with the speckle interferometery (Hege et al. 1991). Our orbital model allow us to interpret the "eclipses" of ⊙ ⊙ M. Gromadzki and J. Mikołajewska: The spectroscopic orbit and the geometry of R Aqr 7 Schwarzenberg-Czerny, A. 1996, ApJ, 460, L107 Solf, J., & Ulrich, H. 1985, A&A, 148, 274 Spencer, J., Winnberg, A., Olnon, F., et al. 1981, AJ, 86, 392 Schwarz, H.E., Nyman, L.-A., Seaquist, E.R., & Ivison, R.J. 1995, A&A, 303, 833 Townley, S. D., Cannon, A. J., & Campbell, L. 1928, Ann. Harv. Coll. Obs., 79, 161 van Belle, G. T., Dyck, H. M., Benson, J. A., & Lacasse, M. G. 1996, AJ, 112, 2147 Wallerstein, G. 1986, PASP, 98, 118 Whitelock, P.A. 2003, in "Symbiotic Stars Probing Stellar Evolution", ASP Conference Series Vol. 303, eds. R.L.M. Corradi, J. Mikołajewska and T.J. Mahoney, p.41 Whitelock, P.A., Feast, M. W., Catchpole, R. M., Carter, B. S., & Roberts, G. 1983, MNRAS, 203, 351 Whitelock, P.A., Feast, M.W., & van Leeuwen, F. 2008, MNRAS, 386, 313/arXiv:0801.4465 Willson, L. A., Garnavich, P., & Mattei, J. A. 1981, Inf. Bull. Variable Stars, 1961, 1 Zhou, A. 1999, Publ. Beijing Astron. Obs., 33, 17/arXiv:astro-ph/0304066v1 Zuckerman, B. 1979, ApJ, 230, 442
astro-ph/0610299
1
0610
2006-10-10T22:08:16
Transient X-ray sources in the field of the Unidentified Gamma-Ray Source TeV J2032+4130 in Cygnus
[ "astro-ph" ]
We present an analysis of Chandra ACIS observations of the field of TeV J2032+4130, the first unidentified TeV source, detected serendipitously by HEGRA. This deep (48.7 ksec) observation of the field follows up on an earlier 5 ksec Chandra director's discretionary observation. Of the numerous point-like X-ray sources in the field, the brightest are shown to be a mixture of early and late-type stars. We find that several of the X-ray sources are transients, exhibiting rapid increases in count rates by factors 3-10, and similar in nature to the one, hard absorbed transient source located in the earlier Chandra observation of the field. None of these transient sources are likely to correspond to the TeV source. Instead, we identify a region of diffuse X-ray emission within the error circle of the TeV source and consider its plausible association.
astro-ph
astro-ph
Astrophysics and Space Science manuscript No. (will be inserted by the editor) 6 0 0 2 t c O 0 1 1 v 9 9 2 0 1 6 0 / h p - o r t s a : v i X r a R. Mukherjee · E. V. Gotthelf · J. P. Halpern Transient X-ray sources in the field of the Unidentified Gamma-Ray Source TeV J2032+4130 in Cygnus Received: date / Accepted: date Abstract We present an analysis of Chandra ACIS ob- servations of the field of TeV J2032+4130, the first uniden- tified TeV source, detected serendipitously by HEGRA. This deep (48.7 ksec) observation of the field follows up on an earlier 5 ksec Chandra director's discretionary ob- servation. Of the numerous point-like X-ray sources in the field, the brightest are shown to be a mixture of early and late-type stars. We find that several of the X- ray sources are transients, exhibiting rapid increases in count rates by factors 3 -- 10, and similar in nature to the one, hard absorbed transient source located in the ear- lier Chandra observation of the field. None of these tran- sient sources are likely to correspond to the TeV source. Instead, we identify a region of diffuse X-ray emission within the error circle of the TeV source and consider its plausible association. Keywords gamma-rays: individual (3EG J2033+4118, TeV J2032+4130) · gamma-rays: observations · X-rays: stars 1 Introduction TeV J2032+4130 was the first unidentified gamma-ray source detected at TeV energies. The source was discov- ered serendipitously in the direction of the Cygnus OB2 stellar association region by the HEGRA CT-System at La Palma [1,2] in observations originally devoted to Cygnus X -- 3. The HEGRA observations, carried out be- tween 1999 and 2002, found TeV J2032+4130 to be a steady gamma-ray source, with the integrated flux mea- sured above 1 TeV at ≈ 5% of that of the Crab Nebula. The best fit HEGRA position for the source is 20h31m57s± R. Mukherjee Department of Physics & Astronomy, Barnard College, Columbia University, New York, NY 10027. E-mail: [email protected] E. V. Gotthelf & J. P. Halpern Columbia Astrophysics Laboratory, Columbia University, New York, NY 10027 6.s2stat ± 1.s0sys, +41◦29′56.8′′ ± 1.′1stat ± 1.′0sys (J2000) [3]. The source was reported to be extended, with a Gaussian 1σ radius of ∼ 6.′2 (±1.′2stat ± 0.′9sys). TeV J2032+4130 was also detected in the Whipple archival data taken in 1989 and 1990 [4], with some indication that the source may be variable on the time scale of sev- eral years. It is interesting to note that the error circle of TeV J2032+4130 overlaps the edge of the 95% confidence error ellipse of an EGRET source, 3EG J2033+4118. However, it is not clear if they are associated. The field of TeV J2032+4130 was initially observed by Chandra during a short 5 ksec exposure [5]. In an at- tempt to understand the possible origins of the source, we performed a multiwavelength study of the region, car- rying out optical identifications and spectroscopic classi- fications of the bright X-ray sources in the Chandra ACIS image and (archival) ROSAT PSPC data [6]. The X-ray sources detected were found to be a mix of early- and late- type stars, and there was no compelling counter- part to the gamma-ray source. However, in our study of the Chandra ACIS field, we did find an unusual new, hard absorbed source that was both transient and rapidly vari- able. We reported on the detection of this source (Chan- dra Source 2 in [6]) as the brightest source in the Chandra field, located at 20h31m43.755s, +41◦35′55.17′′ (J2000), at a distance of 7′ from the centroid of the TeV emission. We detected a coincident reddened optical counterpart with the MDM Observatory 2.4m telescope, but with- out any emission or absorption features in its spectrum. Although the transient source was the brightest of the Chandra sources, it was noticeably absent from earlier ROSAT or Einstein images. At the time of our initial study of this field, we considered the possibility of this transient X-ray source being a candidate for a "proton" blazar [7], a radio-weak gamma-ray source that could be associated with TeV J2032+4130 [6]. However, without knowing the exact nature of this source, we were unable to consider it a compelling counterpart. Since the original study, a deep 50 ksec Chandra ob- servation of the TeV J2032+4130 field has been acquired. An analysis of this data recently summarized in [8] found 2 R. Mukherjee et al. ≈ 240 point-like X-ray sources in the Chandra field, but no obvious diffuse X-ray counterpart to TeV J2032+4130. We have reanalyzed this new Chandra exposure and find that at least seven of the brightest X-ray point sources are either flare stars or transients. The brightest Chan- dra source, "Source 2" from our earlier study is no longer detected. We are now convinced that Chandra Source 2 reported in [6] is a flare star, not associated with TeV J2032+4130. In this paper we (a) summarize the prop- erties of the several transient sources discovered in the Chandra field of TeV J2032+4130, (b) identify a diffuse candidate X-ray counterpart, and (c) review our conclu- sions about the possible nature of the gamma-ray source. 2 X-ray Observations On 2004 July 12, Chandra acquired a 48.7 ks obser- vation of the field of TeV J2032+4130 with the front- illuminated, imaging CCD array of the Advanced CCD Imaging Spectrometer (ACIS-I). ACIS is sensitive to pho- tons in the energy range 0.2 -- 10 keV with a spectral res- olution of ∆E/E ∼ 0.1 at 1 keV. Data reduction and analysis were performed using the standard analysis soft- ware packages, CIAO, FTOOLS, and XSPEC. Figure 1 shows the Chandra image of the region, with the position of TeV J2032+4130 marked. There are numerous point- like X-ray sources near the centroid of the TeV source: Butt et al. (2006) find 240 point-like X-ray sources in a recent study of the field [8]. We have marked the posi- tions of the brightest point sources, those having at least 100 photons and a signal-to-noise ratio 5σ or greater, in Fig. 1. The positions, count rates and hardness ratios of these sources are given in Table 1. A comparison of this field with the earlier 5 ksec Chandra exposure [6] shows that several of the point sources from the earlier Chandra observation are detected in this deeper exposure. How- ever, it is notable that the brightest X-ray source from the earlier observation is absent from the image shown in Fig. 1. The position of this transient source discovered by [6] is marked in the figure with a triangle. We find that several of the sources in the Chan- dra field have ordinary stellar counterparts. Many of the stars in the Cyg OB2 association are among the strongest stellar X-ray sources in the Galaxy. Likely optical identi- fications of the Chandra sources are given in Table 1. The magnitudes listed in the table are from the USNO-A2.0 and USNO-B1.0, where available, or from the MT91 [9] compilation of stars in Cyg OB2, or from the optical images obtained by us during our earlier study of this field [6]. Two of the sources have no optical counterparts. Both were detected in the earlier Chandra observation, and have no optical counterpart to a limiting magni- tude greater than 23 [6]. As in our earlier analysis, we find these to be the hardest sources in the image. They are likely to be active galaxies, highly absorbed by the image of Fig. 1 Chandra ACIS-I the field of TeV J2032+4130. The positions of the marked sources are given in Table 1. The small square marks the centroid of TeV J2032+4130, and the circle is the estimated Gaussian 1σ ex- tent of the TeV emission [3]. The triangle marks the brightest Chandra source in an earlier 5 ksec observation of the region [6], noticeably absent from this image. Galactic ISM, and are unlikely to be nearby, old neutron stars. 3 Transient X-ray Sources in the Field of TeV J2032+4130 Seven of the sources in Table 1 were not detected in the earlier ROSAT [10] or Chandra [6] observations of this field. Thus, they may be described as transient sources. Figure 2 shows the lightcurve of the brightest of these sources (# 4), constructed from the 48.7 ks Chandra ob- servation. The aperture of the source and background regions are indicated in the figure. The background is seen to be sufficiently stable and has little effect on the source light curves. The figure shows that the count rate rose by more than a factor of 10 in the final 15 ks of the observation, after remaining faint for the first 35 ks. We see a similar behavior in the case of the other tran- sient sources. We believe that these are flare stars, which commonly exhibit X-ray flaring activity (e.g. [11]). Transient X-ray sources in the field of TeV J2032+4130 3 Table 1 Chandra Sources in the Field of TeV J2032+4130: Likely Counterparts Optical Positiona R.A. Decl. Name Sp. Type B mag R mag . . . ID X-ray Positiona Ctsb HRc R.A. Decl. 20 31 56.50 +41 37 22.00 20 32 46.23 +41 36 16.03 20 31 51.87 +41 31 18.91 20 32 12.78 +41 29 50.94 20 31 23.58 +41 29 49.29 20 32 11.60 +41 29 01.41 20 32 25.78 +41 28 42.28 20 32 13.84 +41 27 11.66 20 32 27.63 +41 26 21.76 20 32 38.72 +41 25 14.75 20 32 11.32 +41 24 52.02 20 31 37.32 +41 23 37.19 20 31 51.30 +41 23 23.44 20 32 33.84 +41 23 04.46 20 32 37.85 +41 22 08.79 20 32 22.42 +41 18 18.97 20 32 40.66 +41 14 28.96 20 32 31.85 +41 14 12.15 20 33 10.83 +41 15 12.56 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 808 0.56 445 −0.22 206 0.76 843 0.15 274 −0.21 113 −0.05 160 −0.18 290 −0.75 179 −0.60 440 −0.31 301 −0.08 232 −0.51 722 −0.78 623 0.17 931 0.10 28127 −0.55 16558 −0.26 9440 −0.13 14479 −0.28 . . . . . . . . . . . . . . . . . . . . . 20 32 46.240 +41 36 16.0 MT91 321 . . . . . . 20 32 12.763 +41 29 51.24 20 31 23.573 +41 29 49.45 20 32 11.600 +41 29 01.48 20 32 25.731 +41 28 42.89 20 32 13.836 +41 27 12.33 Cyg OB2 4 20 32 27.663 +41 26 22.44 MT91 258 20 32 38.580 +41 25 13.6 MT91 299 20 32 11.303 +41 24 52.69 20 31 37.267 +41 23 36.01 MT91 115 20 31 51.319 +41 23 23.79 MT91 152 20 32 33.862 +41 23 04.27 20 32 37.820 +41 22 08.98 20 32 22.425 +41 18 18.96 Cyg OB2 5 20 32 40.959 +41 14 29.29 Cyg OB2 12 B5Iab: 20 32 31.556 +41 14 08.48 MT91 267 20 33 10.736 +41 15 08.22 Cyg OB2 9 G6 V G3 V . . . . . . . . . O7e ... O5Iab:e O7 III((f)) O7.5V 11.23 > 23.2 10.28 > 23.7 19.2 15.1 15.5 17.3 10.2 10.4 17.0 13.1 13.1 16.7 18.5 8.1 ... 11.8 ... 11.42 12.03 13.90 13.40 10.64 14.41 15.06 12.61 (a) Units of right ascension are hours, minutes, and seconds. Units of declination are degrees, arcminutes, and arcseconds. (b) Total counts in a 12′′ radius aperture. The total included background is estimated as 1 -- 3 counts. (c) Hardness ratio (HR) is defined as: S(0.5−2keV)−S(2−10keV) S(0.5−2keV)+S(2−10keV) , where S is the source counts in a given energy band. In comparison, the only transient source in our ear- lier Chandra field was also highly variable during the brief 5 ks observation. It remained faint for the first 3.5 ks, but increased its count rate by ten-fold in the final 1.5 ks. This source is not detected in the deep Chandra observation of the field. Flare stars are generally dim, red (class-M) dwarfs that are seen to exhibit unusually violent activity in op- tical and/or X-ray bands, and sometimes in the radio Fig. 2 Lightcurves (top) and local background (bottom) of Chandra source 4 (Table 1). The aperture sizes for the source extraction and background regions are indicated in the figure. A time binning of 200 s was used. The dots below the light curve are the arrival times of the individual photons. The background is demonstrated to be stable and has negligible effect on the source light curve. and ultraviolet bands. Flare stars are not known to be gamma-ray emitters. It is unlikely that any of the tran- sient X-ray sources are point source counterparts to TeV J2032+4130. One of the suggestions for the origin of an extended region of TeV emission, as in the case of TeV J2032+4130, is inverse Compton scattering from a jet-driven termina- tion shock from Cyg X-3 or an as yet undetected micro- quasar [1]. In [6] we were motivated to further study the one transient source in the field in order to investigate if it could be such a jet source. Based on the new Chandra observation and the detection of several similar transient sources, it is clear that the one transient source in [6] is not a microquasar, or responsible for the TeV emission in any way. 4 Diffuse X-ray Emission in the Field of TeV J2032+4130 We carried out an analysis to search for diffuse, extended X-ray emission in the TeV source region. Figure 3 shows an image of the diffuse emission only, made by locat- ing and cutting out the point sources, and smoothing the resulting image with a Gaussian kernel of sigma 14′′. This image is exposure and vignetting corrected in the broad energy band of 0.3 to 8 keV. The extended emis- sion centered on 20h32m13.4s, +41◦27′10.4′′ (J2000) is detected at a significance of 6.1σ, and has an extent of roughly ∼ 1.6′ diameter, with a few features extending further. It is possibly associated with Cyg OB2 #4. For 4 R. Mukherjee et al. ation at d = 1740 pc [9], the corresponding luminosity is ∼ 3 × 1031 erg s−1. TeV J2032+4130 appears to be an extended source, unlikely to have a point source counterpart at other wave- lengths. It seems to be related to the massive Cyg OB2 association and the massive stars in the region. Aharo- nian et al. (2002) [1] discuss two possible origins of the extended TeV emission from the source. The emission could be hadronic in origin, arising from the accelera- tion of hadrons in shocked OB star winds and interaction with local, dense gas cloud, and subsequent π0 decay. Or, the TeV emission could be inverse Compton scattering in a jet-driven termination shock from Cyg X-3 or an as yet undetected microquasar. In a recent study, Butt et al. (2006) [8] find that a surface density plot of the point-like X-ray sources in the Chandra field shows an excess consistent with the size and position of TeV J2032+4130. One proposal made by these authors is that the TeV source is a composite of several point sources, and it is possible that several of point X-ray sources are responsible for the TeV emission. The fact that we detect hard X-ray emission within the error circle of TeV J2032+4130 is quite interesting. Together with the TeV observations, it points to the fact that high energy particles are being accelerated in the stellar winds associated with the massive stars in the re- gion. It is not obvious, however, that the diffuse emission is related to the TeV source. We need deeper observa- tions of the region in order to derive an X-ray spectrum of the diffuse emission. It would also be important to see if future observations (with better angular resolution) of TeV J2032+4130 with VERITAS or MAGIC indicate any spatial correlation between the gamma-ray and X- ray emissions. Further observations at TeV energies with ground-based atmospheric Cherenkov telescopes as well as space-based experiments like GLAST are needed to help us resolve the nature of this source. Acknowledgements This publication makes use of data obtained from HEASARC at Goddard Space Flight Center and the SIMBAD astronomical database. R. M. acknowledges support from NSF grant PHY-0244809. References 1. Aharonian, F., et al. 2002a, A&A, 393, L37. 2. Rowell, G. & Horns, D. 2002, in The Gamma-Ray Uni- verse, ed. A. Goldwurm, D. Neumann, & J. Tran Thanh Van (Hanoi: Gioi Publ.), 385. 3. Aharonian, F., et al. 2005, A&A, 431, 197. 4. Lang, M. J., et al. 2004, A&A, 423, 415. 5. Butt, Y., et al. 2003, ApJ, 597, 494. 6. Mukherjee, R., Halpern, J. P., Gotthelf, E. V. et al. 2003, ApJ, 589, 487. 7. Mannheim, K. 1993, A&A, 269, 67. 8. Butt, Y., et al. 2006, ApJ, 643, 238. 9. Massey, P., & Thompson, A. B. 1991, AJ, 101, 1408. 10. Waldron, W. L., et al. 1998, ApJS, 118, 217. 11. Haisch, B. M., et al. 1983, ApJ, 267, 280. Fig. 3 Chandra ACIS-I image of the field of TeV J2032+4130 showing diffuse emission only in the 0.3-8 keV band. The image has been exposure and vignetting corrected. The circle is the estimated Gaussian 1σ extent of the TeV emission [3]. the circular aperture of 1.6′ diameter, the diffuse X-ray flux is 8 × 10−14 erg cm−2 s−1 in the 0.5 -- 10 keV band, assuming a power-law spectral model with photon index of 1.5, typical of non-thermal spectra and total Galac- tic NH = 1.5 × 1022 cm−2. By comparing the exposure and vignetting corrected images of the region in the soft and hard energy bands of 0.3 -- 2.0 keV and 2.0 -- 8.0 keV, respectively, we find no significant softening of the spec- trum in the high energy band. The corresponding hard- ness ratio is −0.48. 5 Discussion and Conclusions Based on the deep Chandra observations of the TeV field, we summarize our principal findings as follows. We find several new transient sources in the Chandra field of TeV J2032+4130. These are similar in nature to the one transient source found in the earlier 5 ks Chan- dra observation of the field. We are convinced that these transient sources are flare stars, unlikely to be associated with the TeV source. Mukherjee et al. (2003) [6] considered the candidacy of the one transient source detected in the field of TeV J2032+4130 for either a "proton blazar" or a jet source responsible for TeV emission via inverse Compton scat- tering. Based on the new data, we are convinced that this is not the case. We find no convincing point source counterpart to TeV J2032+4130 in the X-ray band. We find significant hard diffuse X-ray emission within the error circle of TeV J2032+4130. If the source of the diffuse emission is embedded in the Cygnus OB2 associ-
0707.1652
1
0707
2007-07-11T17:16:51
The Offline Software Framework of the Pierre Auger Observatory
[ "astro-ph" ]
The Pierre Auger Observatory is designed to unveil the nature and the origins of the highest energy cosmic rays. The large and geographically dispersed collaboration of physicists and the wide-ranging collection of simulation and reconstruction tasks pose some special challenges for the offline analysis software. We have designed and implemented a general purpose framework which allows collaborators to contribute algorithms and sequencing instructions to build up the variety of applications they require. The framework includes machinery to manage these user codes, to organize the abundance of user-contributed configuration files, to facilitate multi-format file handling, and to provide access to event and time-dependent detector information which can reside in various data sources. A number of utilities are also provided, including a novel geometry package which allows manipulation of abstract geometrical objects independent of coordinate system choice. The framework is implemented in C++, and takes advantage of object oriented design and common open source tools, while keeping the user side simple enough for C++ novices to learn in a reasonable time. The distribution system incorporates unit and acceptance testing in order to support rapid development of both the core framework and contributed user code.
astro-ph
astro-ph
The Offline Software Framework of the Pierre Auger Observatory S. Argir`o a, S.L.C. Barroso b,1, J. Gonzalez c,2, L. Nellen d,3, T. Paul c,2,∗, T.A. Porter e,4, L. Prado Jr. f ,1 , M. Roth g,5, R. Ulrich g , and D. Veberic h,6 a INFN and University of Torino, Via P. Giuria 1, I-10125 Torino, Italy bCentro Brasileiro de Pesquisas F´ısicas, Rua Dr. Xavier Sigaud, 150, Rio de Janeiro-RJ, CEP 22290-180, Brazil cNortheastern University, 360 Huntington Ave., Boston, MA, 02115 USA dDepartamento de F´ısica de Altas Energ´ıas, Instituto de Ciencias Nucleares, Universidad Nacional Autonoma de M´exico, M´exico D.F., C.P 04510 eLouisiana State University, Baton Rouge, LA, 70803, USA and Santa Cruz Institute for Particle Physics, University of California, Santa Cruz, CA, 95046, USA f Instituto de F´ısica Gleb Wataghin, Universidade Estadual de Campinas UNICAMP, Campinas-SP, CP 6165 CEP 13083-970, Brazil gKarlsruhe Institute of Technology KIT, University and Forschungszentrum Karlsruhe, POB 3640, D-76021 Karlsruhe, Germany hUniversity of Nova Gorica, Vipavska 13, PO Box 301, SI-5001, Nova Gorica, Slovenia Abstract The Pierre Auger Observatory is designed to unveil the nature and the origins of the highest energy cosmic rays. The large and geographically dispersed collaboration of physicists and the wide-ranging collection of simulation and reconstruction tasks pose some special challenges for the offline analysis software. We have designed and implemented a general purpose framework which allows collaborators to contribute algorithms and sequencing instructions to build up the variety of applications they require. The framework includes machinery to manage these user codes, to organize the abundance of user-contributed configuration files, to facilitate multi-format file handling, and to provide access to event and time-dependent detector information which can reside in various data sources. A number of utilities are also provided, in- cluding a novel geometry package which allows manipulation of abstract geometrical ob jects independent of coordinate system choice. The framework is implemented in C++, and takes advantage of ob ject oriented design and common open source tools, while keeping the user side simple enough for C++ novices to learn in a reasonable Preprint submitted to Elsevier Science 23 December 2013 time. The distribution system incorporates unit and acceptance testing in order to support rapid development of both the core framework and contributed user code. Key words: offline software, framework, ob ject oriented, simulation, cosmic rays PACS: 07.05.Bx, 07.05.Kf, 07.05.Tp, 29.85.+c 1 Introduction The offline software framework of the Pierre Auger Observatory (1) provides an infrastructure to support a variety of distinct computational tasks necessary to analyze data gathered by the observatory. The observatory itself is designed to measure the extensive air showers produced by the highest energy cosmic rays (> 1019 eV) with the goal of discovering their origins and composition. Two different techniques are used to detect air showers. Firstly, a collection of telescopes is used to sense the fluorescence light produced by excited atmo- spheric nitrogen as the cascade of particles develops and deposits energy in the atmosphere. This method can be used only when the sky is moonless and dark, and thus has roughly a 15% duty cycle. Secondly, an array of detectors on the ground is used to sample particle densities and arrival times as the air shower impinges upon the Earth’s surface. Each surface detector consists of a tank containing 12 tons of purified water instrumented with photomultiplier tubes to detect the Cherenkov light produced by passing particles. The surface detector has a 100% duty cycle. A subsample of air showers detected by both instruments, dubbed hybrid events, are very precisely measured and provide an invaluable energy calibration tool. In order to provide full sky coverage, the baseline design of the observatory calls for two sites, one in the southern hemisphere and one in the north. The southern site is located in Mendoza, Argentina, and construction there is nearing completion, at which time the observatory will comprise 24 fluorescence telescopes overlooking 1600 surface ∗ corresponding author: [email protected] 1 The work of S.L.C.B. and L.P.Jr was supported by the Brazilian funding agencies CNPq and FAPESP. 2 The work of T.P. and J.G. was supported in part by the US National Science Foundation. 3 The work of L.N. was supported by CONACyT, DGAPA-UNAM (PAPIIT pro- gramme), and CIC-UNAM. 4 The work of T.A.P. was supported in part by the US Department of Energy. 5 The work of M.R. was supported by the HHNG-128 grant of the German Helmholtz association. 6 The work of D.V. was supported by the Slovenian research agency ARRS. 2 detectors spaced 1.5 km apart on a hexagonal grid. The state of Colorado in the USA has been selected as the location for the northern site. The requirements of this pro ject place rather strong demands on the software framework underlying data analysis. Most importantly, the framework must be flexible and robust enough to support the collaborative effort of a large number of physicists developing a variety of applications over the pro jected 20 year lifetime of the experiment. Specifically, the software must support simulation and reconstruction of events using surface, fluorescence and hybrid methods, as well as simulation of calibration techniques and other ancillary tasks such as data preprocessing. Further, as the experimental run will be long, it is essential that the software be extensible to accommodate future up- grades to the observatory instrumentation. The offline framework must also handle a number of data formats in order to deal with event and monitoring information from a variety of instruments, as well as the output of air shower simulation codes. Additionally, it is desirable that all physics code be “ex- posed” in the sense that any collaboration member must be able to replace existing algorithms with his or her own in a straightforward manner. Finally, while the underlying framework itself may exploit the full power of C++ and ob ject-oriented design, the portions of the code directly used by physicists should not assume a particularly detailed knowledge of these topics. The offline framework was designed with these principles in mind. Implemen- tation began in 2002, and the first physics results based upon this code were presented at the 29th International Cosmic Ray Conference (2). 2 Overview The offline framework comprises three principal parts: a collection of pro- cessing modules which can be assembled and sequenced through instructions provided in an XML file, an event data model through which modules can relay data to one another and which accumulates all simulation and recon- struction information, and a detector description which provides a gateway to data describing the configuration and performance of the observatory as well as atmospheric conditions as a function of time. These ingredients are depicted in Fig. 1. This approach of pipelining processing modules which communicate through an event serves to separate data from the algorithms which operate on these data. Though this approach is not particularly characteristic of ob ject oriented design, it was used nonetheless, as it better satisfies the requirements of physi- cists whose primary ob jective is collaborative development and refinement of algorithms. 3 Fig. 1. General structure of the offline framework. Simulation and reconstruction tasks are broken down into modules. Each module is able to read information from the detector description and/or the event, process the information, and write the results back into the event. These components are complemented by a set of foundation classes and util- ities for error logging, physics and mathematical manipulation, as well as a unique package supporting abstract manipulation of geometrical ob jects. Each of these aspects of the framework is described in more detail below. 3 User Modules and Run Control Experience has shown that most tasks of interest of the Pierre Auger Col- laboration can be factorized into sequences of self contained processing steps. Physicists prepare such processing algorithms in so-called modules, which they register with the framework via a macro. This modular design allows collabo- rators to exchange code easily, compare algorithms and build up a wide variety of applications by combining modules in various sequences. Modules inherit a common interface which declares the methods that carry out processing. Specifically, module authors must implement a Run method which is called once per event, as well as Init and Finish methods to be called at the beginning and end of a processing job. Authors invoke a macro in the module class declaration which registers a factory function used by the framework to instantiate the module when requested. The registry mechanism provides a fall-back hook that handles requests for modules not known to the registry. Dynamical loading using this fall-back mechanism is currently under development. Modules themselves are not instrumented with a means to place requirements on versions of other modules or on module execution order; instead, the configuration machinery described in section 4 is used to set such requirements. For most applications, run-time control over module sequences is afforded through a run control ler which invokes the various processing steps within the modules according to a set of externally provided instructions. We have constructed a simple XML-based (3) language for specifying these sequencing 4 instructions. The decision to employ XML for this task was based on several considerations. Firstly, sequencing via interpreted XML files allows one to set up or modify the behavior of a run without compiling, and so offers the con- venience of an interpreted language. In addition, XML syntax is simple and relatively familiar to the community, and hence imposes a minimal learning curve compared to most scripting languages. Furthermore, XML sequencing instructions can be straightforwardly logged using the same mechanisms em- ployed for other types of configuration logging, as described in some detail in section 4. Finally, XML proves sufficient to support our commonest applica- tions, which do not require especially intricate sequencing control. In fact the ma jority of applications require only two fundamental instructions. First, a module element instructs the run controller to invoke the processing method of the module whose name appears between the begin and end tags. Second, a loop tag is used to command looping over modules, including arbitrarily deep loop nests. This tag may be decorated with attributes indicating the number of times to loop, and whether or not the contents of the event should be pushed onto a stack before the modules contained in the loop begin acting upon it. It is appropriate to push the event to a stack if, for example, one wishes to repeat a random process starting from the same input. On the other hand, one would disable the push to stack in order to implement an iterative procedure distributed over several modules. Fig. 2 shows a simple example of the structure of a sequencing file. <sequenceFile> <loop numTimes="unbounded"> <module> SimulatedShowerReader </module> <loop numTimes="10" pushToStack="yes"> </module> <module> EventGenerator </module> <module> TankSimulator <module> TriggerSimulator </module> </module> <module> EventExporter </loop> </loop> </sequenceFile> Fig. 2. Simplified example in which an XML file sets a sequence of modules to conduct a simulation of the surface array. <loop> and <module> tags are interpreted by the run controller, which invokes the modules in the proper sequence. In this example, simulated showers are read from a file, and each shower is thrown onto the array in 10 random position by an EventGenerator. Subsequent modules simulate the response of the surface detectors and trigger, and export the simulated data to file. The pushToStack="yes" attribute instructs the Run Controller to store the event when entering the loop, and restore it to that state when returning back to the beginning of the loop. Note that XML naturally accommodates common sequencing requirements such as nested loops. Modules can signal the run controller via return codes of the Init, Run and Finish methods, and command it to break a loop or to skip all subsequent 5 modules up to the next loop tag. 4 Configuration Parameters, cuts and configuration instructions used by modules or by the framework itself are stored in XML files. A globally accessible central con- figurator points modules and framework components to the location of their configuration data, and creates parsers to assist in reading information from these locations. The locations of configuration data are specified in a so-called bootstrap file, and may comprise local filenames, URIs (4) or XPath (5) ad- dresses. The name of the bootstrap file is passed to the application at run time via the command line. The configuration mechanism can also concatenate all configuration data ac- cessed during a run and write it in a log file. This log file includes a preamble with a format identical to that of a bootstrap file, with XPath addresses spec- ifying the locations of all the configuration data in the file. In this way, a log file can subsequently be read, as though it were a bootstrap file, in order to reproduce a run with an identical configuration. This configuration logging mechanism may also be used to record the versions of modules and external libraries which are used during a run. External pack- age versions are determined at build time by our GNU autotools-based build machinery (6), which searches the local system for required packages and ver- ifies that versions of different packages are compatible with one another. The build system generates a script which can later be interrogated by the logging mechanism to record package versions employed for a particular run. Module versions are declared in the code by module authors, and are accessible to the configuration logging mechanism through the module interface. To check configuration files for errors, the W3C XML Schema (7) standard validation is employed throughout. This approach is used not only to vali- date internal framework configuration, but also to check configuration files of modules prepared by framework users. This has proved to be successful in saving coding time for developers and users alike, and facilitates much more detailed error checking than most users are likely to implement on their own. Further details on how configuration files are parsed and validated are given in section 6.1. The configuration machinery is also able to verify configuration file contents against a set of default files by employing MD5 digests (8). The default config- uration files are prepared by the framework developers and the analysis teams, and reference digests are computed from these files at build time. At run time, 6 the digest for each configuration file is recomputed and compared to the refer- ence value. Depending on run-time options, discrepant digests can either force program termination, or can simply log a warning. This machinery provides a means for those managing data analyses to quickly verify that configurations in use are the ones which have been recommended for the task at hand. 5 Data Access The offline framework provides two parallel hierarchies for accessing data: the event for reading and writing information that changes per event, and the read-only detector description for retrieving static or relatively slowly varying information such as detector geometry, calibration constants, and atmospheric conditions. 5.1 Event The Event data model contains all raw, calibrated, reconstructed and Monte Carlo data and acts as the principal backbone for communication between modules. The overall structure comprises a collection of classes organized fol- lowing the hierarchy normally associated with the observatory instruments, with further subdivisions for accessing such information as Monte Carlo truth, reconstructed quantities, calibration information and raw data. A non-exhaustive illustration of this hierarchy is given in Fig. 3. User modules access the event through a reference to the top of the hierarchy which is passed to the module interface by the run controller. Since the event constitutes the inter-module communication backbone, refer- ence semantics are used throughout to access data structures in the event, and constructors are kept private to prevent accidental copying of event compo- nents. For example, to retrieve the ob ject representing the first photomultiplier tube (PMT) ob ject of station number 157 belonging to the surface detector (S) portion of the event, one could write simply, PMT& pmt1 = theEvent.GetSEvent().GetStation(157).GetPMT(1); where theEvent is a reference to the top of the event hierarchy. The event is built up dynamically as needed, and is instrumented with a simple protocol allowing modules to interrogate the event at any point to discover its current constituents. This protocol provides the means for a given module to determine whether the input data required to carry out the desired processing is available. As an example, consider the case of the Monte Carlo 7 Fig. 3. Hierarchy of the event interface. The top level Event encapsulates ob jects rep- resenting Fluorescence and Surface events (FEvent and SEvent respectively), as well as reconstructed and simulated shower data (ShowerRecData and ShowerSimData respectively). These components are further subdivided into ob jects representing simulated, reconstructed and triggering data at the level of individual telescopes, tanks and photomultiplier tubes. truth belonging to a PMT ob ject called thePMT. Attempting to access a non- existent subcomponent thePMT raises an exception: PMT& iDontExistYet = thePMT.GetSimData();// exception Checking for the existence of the desired data, creating an event subcompo- nent, and retrieveing a handle to the data therein would be carried out as follows: if (!thePMT.HasSimData()) thePMT.MakeSimData(); PMT& thePMTSim = thePMT.GetSimData(); // check for SimData // create SimData // success The structure of the event interface cannot be modified by modules. While this restriction limits the flexibility available to module authors, it does facilitate straightforward module interchangeability, which, as discussed in section 9, is of primary importance in our case. In practice, when users find the event structure does not accommodate their needs, they may implement ad hoc inter-module communication as temporary solution, and propose the required event interface changes to the framework developers. Though this approach does require periodic developer intervention, it has not proved to be overly problematic for our pro ject. 8 It is worth noting that the use of an event data model as an algorithm commu- nication backbone is not an uncommon approach, and is employed by other high energy physics and astrophysics experiments (10; 11; 12). In our case, however, the data access methods are somewhat less generalized than the techniques employed by larger experiments (see for example (11)). Our feeling is that the cost incurred in terms of flexibility is reasonably offset by user-side simplicity. The event representation in memory, or transient event, is decoupled from the representation on file, or persistent event. When a request is made to write event contents to file, the data are transferred from the transient event through a so-called file interface to the persistent event, which is instrumented with machinery for serialization. Conversely, when data are requested from file, a file interface transfers the data from the persistent event to the appropriate part of the event interface. Users can transfer all or part of the event from memory to a file at any stage in the processing, and reload the event and continue processing from that point onward. Various file formats are handled using the file interface mechanism, including raw event and monitoring formats as well as the different formats employed by the AIRES (13), CORSIKA (14), CONEX (15) and SENECA (16) air shower simulation packages. Fig. 4 contains a diagram of this event input/output mechanism. Event serialization is implemented using the ROOT (17) toolkit, though the decoupling of the transient and persistent events is intended to allow for rel- atively straightforward changes of serialization machinery if this should ever prove to be advantageous. When it becomes necessary to modify the event structure, backward compatibility is provided via the ROOT schema evolu- tion mechanism. We note that similar strategies for event persistency and schema evolution have been adopted by other large experiments (18). 5.2 Detector Description The detector description provides a unified interface from which module au- thors can retrieve non-event data including the detector configuration and per- formance at a particular time as well as atmospheric conditions. Like the event interface, the detector interface is organized following the hierarchy normally associated with the observatory instruments, and provides a set of simple-to- use access functions to extract data. Data requests are passed by this interface to a registry of so-called managers, each of which is capable of extracting a particular sort of information from a particular data source. Data retrieved from a manager are cached in the interface for future use. In this approach, the user deals with a single interface even though the data sought may reside in any number of different sources. Generally we choose to store static detector 9 Fig. 4. Event input/output. The section labeled “Event Interface” portrays a subset of the hierarchy depicted in Fig. 3. Data are transferred between this transient event and persistent ob jects through a common file interface. Different file implementa- tions are able to read and/or write in different formats, including those used by the data acquisition systems (DAS formats), formats used by other simulation packages, as well as a “native” format (ROOTEventFile) which accommodates all raw data, reconstructed quantities, and Monte Carlo truth. information in XML files, and time-varying monitoring and calibration data in MySQL (19) databases. However, as the pro ject evolves it sometimes hap- pens that access to detector data in some other format is required, perhaps as a stop-gap measure. The manager mechanism allows one to quickly provide simple interfaces in such cases, keeping the complexity of accessing multiple formats hidden from the user. The structure of the detector description ma- chinery is illustrated in Fig. 5. Note that it is possible to implement more than one manager for a particular sort of data. In this way, a special manager can override data from a general manager. For example, a user can decide to use a database for the ma jority of the description of the detector, but override some data by writing them in an XML file which is interpreted by the special manager. The specification of which data sources are accessed by the manager registry and in what order they are queried is detailed in a configuration file. The configuration of the manager registry is transparent to the user code. The detector description is also equipped to support a set of plug-in functions, called models which can be used for additional processing of data retrieved through the detector interface. These are used primarily to interpret atmo- spheric monitoring data. As an example, users can invoke a model designed to evaluate attenuation of light due to aerosols between two points in the atmo- sphere. This model interrogates the detector interface to find the atmospheric conditions at the specified time, and computes the attenuation. Models can also be placed under command of a super-model which can attempt various methods of computing the desired result, depending on what data are available for the specified time. 10 Fig. 5. Machinery of the detector description. The user interface (left) comprises a hierarchy of ob jects describing the various components of the observatory. These ob jects relay requests for data to registry of managers (right) which handle multiple data sources and formats. The manager mechanism has also proved convenient for generating detec- tor data for use by specialized Monte Carlo simulations. For instance, when studying reconstruction techniques, it is sometimes useful to include hypo- thetical surface detector stations in a simulation run, where the positions of the hypothetical stations are defined in the reference frame of the shower be- ing simulated. A manager has been prepared which uses the simulated shower geometry to pro ject a user-defined collection of hypothetical stations onto the ground, overlaying them on the actual station positions. Both sorts of station information are accessible transparently via the same interface. 6 Utilities The offline framework is built on a collection of utilities, including an XML parser, an error logger, various mathematics and physics services including a novel geometry package, testing utilities and a set of foundation classes to represent ob jects such as signal traces, tabulated functions and particles. In this section, we describe the parsing and geometry packages in more detail. 6.1 XML Parsing and Validation As noted previously, XML is employed to store configuration data for frame- work components and user contributed modules. Our XML reading utility, named simply reader, is built upon the Xerces (20) validating parser, and is designed to provide ease-of-use at the expense of somewhat reduced flexibil- ity compared to the Xerces APIs. The reader utility also provides additional features such as units handling for dimensional quantities, expression evalu- 11 ation, and casting of data in XML files to atomic types, Standard Template Library (21) containers, and data types specific to the Auger Observatory software. Navigation through hierarchical data is supported by a handful of methods for locating children and siblings of a given XML element. Data casting is provided through overloaded access functions. For instance, a configuration file can contain a dimensional quantity with units specified by an expression in the tag attribute: <g unit="meter/second^2"> 9.8 </g> Upon request, the reader casts the data between the <g> tags to the data type used in the access function, evaluates the meter/second^2 expression in the attribute, and then uses the result to convert the 9.8 into pre-defined internal units. Validation rules are specified using the XML Schema standard (7), which is well supported by Xerces and which has proved to be more palatable to our user base than the older DTD (22) standard. The built-in Schema types have been extended with a collection of data types commonly used by the collaboration, including lists, three-vectors, and tabulated functions as well quantities which require an associated unit. For example, the line of XML shown above can be validated using the Schema fragment, <xs:element name="g" type="auger:doubleWithUnit"/> where the prefix xs denotes the standard Schema namespace, and the auger prefix indicates the namespace containing our extensions of the standard types. Here the doubleWithUnit type specification guarantees that exactly one dou- ble precision number appears in the element, and that a unit attribute is present in the corresponding tag. One can build up rather involved types in a straightforward manner. For in- stance, it is useful to have the ability to define functions using either tabulated values: <EnergyDistribution> <x unit="MeV"> 100 330 1000 </x> <y> 1 0.95 0.5 </y> </EnergyDistribution> or using a parametrization with limits: <EnergyDistribution> <PDF> 1/x </PDF> <min unit="MeV"> 100 </min> <max unit="GeV"> 1 </max> 12 </EnergyDistribution> The corresponding Schema rules require the <EnergyDistribution> to be specified by a group of XML elements: <xs:element name="EnergyDistribution"> <xs:complexType> <xs:group ref="distributionFunction"/> </xs:complexType> </xs:element> where this group in turn is defined as a choice between two sequences of XML elements dictating the two ways to specify the distribution function mentioned above: <xs:group name="distributionFunction"> <xs:choice> <xs:sequence> <xs:element name="x" type="auger:listOfDoublesWithUnits"/> <xs:element name="y" type="auger:listOfDoubles"/> </xs:sequence> <xs:sequence> <xs:element name="PDF" type="xs:string"/> <xs:element name="min" type="auger:doubleWithUnit"/> <xs:element name="max" type="auger:doubleWithUnit"/> </xs:sequence> </xs:choice> </xs:group> The combination of XML and XML Schema validation enables us to support quite detailed configurations and robust checking, both in the internal frame- work configuration and in user-provided physics modules. To give a specific example, we consider the case of a so-called “particle injector” module which is used to randomly draw particles from various distributions and pass them to downstream detector response simulation modules. In configuring this module we wish to support a number of options in different combinations, including: placing particles at specific locations relative to a detector, or distributing them over cylindrical or spherical surfaces around the detector; selecting dif- ferent particle types with differing probabilities; setting discrete energies or drawing energies from a distribution; setting discrete zenith and azimuthal angles or drawing them from distributions; and describing distributions either analytically through an expression, or using a tabulated function. The hier- archical model employed by XML allows one to notate all these options in a human-readable form. Further, XML Schema validation facilitates detailed policing of the corresponding configuration file, so that for instance one can require a distribution to be described either analytically or in tabular form, 13 as outlined in the example above. Note that the general effectiveness of XML and XML Schema in software for large experiments has been noted by other authors (9). 6.2 Geometry As discussed previously, the Pierre Auger Observatory comprises many instru- ments spread over a large area and, in the case of the fluorescence telescopes, oriented in different directions. Consequently there is no naturally preferred coordinate system for the observatory; indeed each detector component has its own natural system, as do components of the event such as the air shower itself. Furthermore, since the detector spans more than 50 km from side to side, the curvature of the earth cannot generally be neglected. In such a cir- cumstance, keeping track of all the required transformations when performing geometrical computations is tedious and error prone. This problem is alleviated in the offline geometry package by providing ab- stract geometrical ob jects such as points and vectors. Operations on these ob jects can then be written in an abstract way, independent of any partic- ular coordinate system. Internally, the ob jects store components and track the coordinate system used. There is no need for pre-defined coordinate sys- tem conventions, or coordinate system conversions at module boundaries. The transformation of the internal representation occurs automatically. Despite the lack of a single natural coordinate system for the observatory, there are several important coordinate systems available. A registry mecha- nism provides access to a selection of global coordinate systems. Coordinate systems related to a particular component of the detector, like a telescope, or systems which depend on the event being processed, such as a shower coordi- nate system, are available through access functions belonging to the relevant classes of the detector or event structures. Coordinate systems are defined relative to other coordinate systems. Ulti- mately, a single root coordinate system is required. It must be fixed by con- vention, and in our case we choose an origin at the center of the Earth. Other base coordinate systems and a caching mechanism help to avoid the con- struction of potentially long chains of transformations when going from one coordinate system to another. The following is a simple example of how the geometry and units packages are used together: Point pos(x*km, y*km, z*km, posCoordSys); 14 Vector dist(deltaX, deltaY, deltaZ, otherCoordSys); Point newPos = pos + dist; cout << "X = " << newPos.GetX(outCoordSys)/m << " meters"; The variables x, y, and z are provided by some external source, in the units indicated (km), whereas deltaX, deltaY, and deltaZ are results from a previ- ous calculation, already in the internal units. Coordinate systems are required whenever components are used explicitly. Units are used on input and output of data and when exchanging information with external packages. The surveying of the detector utilizes Universal Transverse Mercator (UTM) coordinates with the WGS84 ellipsoid. These coordinates are convenient for navigation. They involve, however, a non-linear, conformal transformation. The geometry package provides a UTMPoint class for dealing with positions given in UTM, in particular for the conversion to and from Cartesian co- ordinates. This class also handles the geodetic conventions, which define the latitude based on the local vertical (see Fig. 6), as opposed to the angle 90◦ − θ, where θ is the usual zenith angle in spherical coordinates. Fig. 6. The geodetic latitude λ is defined as the angle between the local vertical and a plane parallel to the equatorial plane. For an elliptical shape, it is not just the complement of the zenith angle θ in the definition of spherical coordinates. The high degree of abstraction makes use of the geometry package quite easy. Uncontrolled, repeated coordinate transformations, though, can be a problem both for execution speed and for numerical precision. To control this behavior, it is possible to force the internal representation of an ob ject to use a particular coordinate system. The geometry package guarantees that no transformations take place in operation where all ob jects are represented in the same coordinate system. This provides a handle for experts to control when transformations take place. 15 7 Build System and Quality Control To help ensure code maintainability and stability in the face of a large number of contributors and a decades-long experimental run, unit and acceptance testing are integrated into the offline framework build and distribution system. This sort of quality assurance mechanism is crucial for any software which must continue to develop over a timescale of many years. Our build system is based on the GNU autotools (6), which provide hooks for integrating tests with the build and distribution system. A substantial collection of unit tests has been developed, each of which is designed to com- prehensively exercise a single framework component. We have employed the CppUnit (23) testing framework as an aid in implementing these unit tests. In addition to such low-level tests, a collection of higher-level acceptance tests has been developed which is used to verify that complete applications con- tinue to function properly during ongoing development. Such acceptance tests typically run full physics applications before and after each code change and notify developers in case of any unexpected differences in results. As a distributed cross-platform pro ject, the Auger Offline software must be regularly compiled and checked on numerous platforms. To automate this process, we have employed the tools provided by the BuildBot pro ject (24). The BuildBot is a Python-based system in which a master daemon is informed each time the code repository has been significantly altered. The master then triggers a collection of build slaves running on various platforms to download the latest code, build it, run the unit and acceptance tests, and inform the appropriate developers in case problems are detected. This has proved to be a very effective system, and provides rapid feedback to developers in case of problems. 8 External packages The choice of external packages upon which to build the offline framework was dictated not only by package features and the requirement of being open- source, but also by our best assessment of prospects for longevity. At the same time, we attempted to avoid locking the design to any single-provider solution. To help achieve this, the functionality of external libraries is often provided to the client code through wrappers or fa¸cades, as in the case of XML parsing described in sections 3 and 6.1, or through a bridge, as in the case of the detector description described in section 5.2. The collection of external libraries currently employed includes ROOT (17) for serialization, Xerces (20) for XML parsing and validation, CLHEP (26) for expression evaluation and 16 geometry foundations, Boost (25) for its many valuable C++ extensions, and optionally Geant4 (27) for detailed detector simulations. 9 Examples In this section we demonstrate the application of the offline software to typical simulation and reconstruction user tasks. We consider specifically the case of hybrid simulation and reconstruction, which involves combining the sequences for surface and fluorescence detector simulation and reconstruction. The dis- cussion is meant to illustrate some of the advantages of modularization at the level of algorithms, as well as the simple XML-based sequencing control. 9.1 Hybrid Detector Simulation Simulation of events observed by the hybrid detector typically involves cre- ation of a shower using a Monte Carlo generator (13; 14; 15; 16), the simulation of the response and triggering of the surface array to the particles arriving at the ground, and the simulation of the telescope response and triggering to the profile of fluorescence light emitted along the shower track. Finally, event building and export to various data formats can be performed. Detector simulations can be broken down into a sequence of steps, each of which is generally encapsulated within a separate module. For example, simu- lation of the surface detector typically begins with a module which places the simulated shower impact point somewhere within the surface array configura- tion. This is followed by a module which uses this information to determine which particles enter into which water tanks. Subsequent modules then sim- ulate the particle energy loss and Cherenkov light emission in each tank, the response of the phototubes and tank electronics, and the local tank trigger. A final module simulates the response of the central trigger, which considers information from multiple detector components when determining whether to record the event. Simulation of a fluorescence event involves modules for plac- ing the simulated shower at some distance from one of the eyes, simulating the fluorescence and Cherenkov light emitted by the shower as it develops, and finally simulating the response of the fluorescence telescopes, electronics and triggering. In Fig. 7 we show a typical module sequence for hybrid detector simulation. Each <module> element designates an individual simulation step. The essential elements for the surface and fluorescence detector simulation are contained within the innermost loop of the module sequence, while the outer loop allows for processing of all Monte Carlo events in a file or collection of files. 17 <sequenceFile> <!-- Loop over all Monte Carlo showers on file --> <loop numTimes="unbounded"> <!-- Read in a Monte Carlo shower --> <module> EventFileReader </module> <!-- use each shower 10 times --> <loop numTimes="10" pushToStack="yes"> <!-- Position the shower in random spot on simulated array --> </module> <module> EventGenerator <!-- Simulate the surface detector response --> <module> ShowerRegenerator </module> </module> <module> G4TankSimulator </module> <module> PhototubeSimulator </module> <module> SdElectronicsSimulator <module> TankTriggerSimulator </module> <!-- Simulate the fluorescence detector response --> <module> ShowerLightSimulator </module> <module> LightAtDiaphragmSimulator </module> </module> <module> TelescopeSimulator </module> <module> FdBackgroundSimulator <module> FdElectronicsSimulator </module> </module> <module> FdTriggerSimulator <!-- Simulate the trigger, build and export event --> <module> CentralTriggerSimulator </module> </module> <module> EventFileExporter </loop> </loop> </sequenceFile> Fig. 7. Example of a hybrid detector simulation module sequence. The surface detector simulation up to the tank triggering step is done first, then the fluorescence simulation up to the local eye triggering is performed. The central triggering, event building and exporting are performed last. It does not matter whether surface detector or fluorescence detector simulation is performed first, though both must be completed before the central triggering and event building steps occur. Modularization allows one to easily substitute alternative algorithms to per- form a particular step of the simulation sequence. For instance, the detailed 18 tank response simulation can be replaced with a simplified, fast simulation by simply replacing the relevant <module> element. Such modularization of algorithms also allows collaborators to propose different approaches to a par- ticular aspect of the simulation process, and to compare results running under identical conditions. 9.2 Hybrid Event Reconstruction The hybrid event reconstruction module sequence is indicated in Fig. 8. This sequence begins with calibration of the fluorescence and surface detectors, a procedure which transforms real or simulated raw data into physical quantities. Afterwards a so-called pulse finding algorithm is used to further process the traces recorded by the fluorescence telescopes. Next, a series of geometrical reconstruction modules are employed. First the plane containing the shower axis and the eye which detected it is determined. A complete geometrical fit within this plane is performed, taking into account both the timing of the shower image as it traverses the telescope pixels as well as the timing and impact point on the surface detectors. A calculation of the light flux reaching the telescope aperture is then carried out. The last step is the profile reconstruction, which converts the fluorescence light profile recorded by the telescopes to a determination of the energy deposit at a given atmospheric depth along the shower axis. The outcome of these steps is depicted in Fig. 9. ... <module> FdCalibrator <module> SdCalibrator <module> FdPulseFinder </module> </module> </module> <module> FdSDPFinder <module> HybridGeometryFinder <module> FdApertureLight </module> </module> </module> <module> FdProfileReconstructor </module> ... Fig. 8. Hybrid detector reconstruction module sequence. This sequence fragment can be appended to the one shown in Fig. 7 to reconstruct a simulated shower. 10 Ongoing developments While the framework described in this note is actively used for analysis, there are a few substantial enhancements in preparation. 19 Fig. 9. Result of the hybrid reconstruction module sequence. The underlying event was created by simulations, and saved in the Auger Observatory raw data format. The figure shows the detector including the grid of 1600 surface detectors and the three (of four) fluorescence detectors which triggered on the event. The colors (from blue to red) indicate the evolution of time. The three camera images show the image of the shower recorded on the telescope pixels, with the signal intensity on each pixel indicated by color. Finally, the plot on the lower left shows the light profile arriving at the Los Leones telescope, indicating contributions from different light sources. First, we are developing an interactive visualization package which is fully in- tegrated into the framework and which will provide not only graphical display of reconstructed event properties and Monte Carlo truth, but also interac- tive control over configuration and reconstruction procedures. This package will complement existing visualization tools which we use to browse processed events. Second, Python (28) bindings for the framework are in preparation. Once 20 complete, all of the framework public interfaces will be exposed via Python, allowing users to prepare rapid prototypes of analysis and visualization tasks. Python-based module sequencing will also be supported, allowing more intri- cate run control than is currently afforded through our XML-base sequenc- ing system for cases when this may be desired. Although these bindings will provide convenience for testing ideas and developing algorithms, the exist- ing module sequencing system and XML-based run control will continue to be used for production runs, particularly as the logging features of this machinery is desirable for batch processing. Third, the user module system described in section 3 is being upgraded to support dynamical loading of modules. This will allow for easier use of modules with the interactive visualization system mentioned above, and support easier module distribution and shorter development cycles. Finally, the event persistency machinery discussed in section 5.1 is undergoing revision. Though the approach we have implemented has been successful in decoupling the in-memory event from the representation on disk, the design does impose a maintenance burden since any modification of the structure must be implemented both in the transient and persistent events. For the future we envisage a system employing a meta-description of the event which will be used to automatically generate the transient and persistent events as well as the Python bindings mentioned above. 11 Conclusions We have implemented a software framework for analysis of data gathered by the Pierre Auger Observatory. This software provides machinery to facilitate collaborative development of algorithms to address various analysis tasks as well as tools to assist in the configuration and bookkeeping needed for produc- tion runs of simulated and real data. The framework is sufficiently configurable to adapt to a diverse set of applications, while the user side remains simple enough for C++ non-experts to learn in a reasonable time. The modular de- sign allows straightforward swapping of algorithms for quick comparisons of different approaches to a problem. The interfaces to detector and event in- formation free the users from having to deal individually with multiple data formats and data sources. This software, while still undergoing vigorous de- velopment and improvement, has been used in production of the first physics results from the observatory. 21 Acknowledgment The authors would like to thank the various funding agencies which made this work possible, as well as the fearless early adopters of the offline framework. References [1] J. Abraham et al. [Pierre Auger Collaboration], “Properties and perfor- mance of the prototype instrument for the Pierre Auger Observatory,” Nucl. Instrum. Meth. A 523, 50 (2004). [2] S. Argir`o et al. [Pierre Auger Collaboration], “The offline software frame- work of the Pierre Auger Observatory,” FERMILAB-CONF-05-311-E-TD Presented at 29th International Cosmic Ray Conference (ICRC 2005), Pune, India, 3-11 Aug 2005. [3] http://www.w3.org/XML/ ; Elliotte Rusty Harold, W. Scott Means, “XML in a Nutshell”, O’Reilly, 2004, ISBN: 0-596-00764-7. [4] http://tools.ietf.org/html/rfc3986/ . [5] http://www.w3.org/TR/xpath . [6] http://www.gnu.org/software/autoconf ; http://www.gnu.org/software/automake ; http://www.gnu.org/software/libtool ; Gary V. Vaughn, Ben Ellison, Tom Tromey, Ian Lance Taylorm “GNU Autoconf, Automake, and Libtool” Sams, 2000, ISBN 1-57870-190-2. [7] http://www.w3.org/XML/Schema/ ; Eric van der Vlist, “XML Schema”, O’Reilly, 2002, ISBN 0-596-00252-1. [8] R. Rivest, “RFC 1321: The MD5 message-digest http://www.faqs.org/rfcs/ . [9] See for example S. Patton, “Concrete uses of XML in software development and data analysis”, International Conference on Computing in High-Energy Physics and Nuclear Physics (CHEP 2003), La Jolla, California, USA 24-28 March 2003. http://www-conf.slac.stanford.edu/chep03/ . [10] J. Kowalkowski, H. Greenlee, Q. Li, S. Protopopescu, G. Watts, V. White, J. Yu, “D0 offline reconstruction and analysis control framework”, Inter- national Conference on Computing in High-Energy Physics and Nuclear Physics (CHEP 2000), Padova, Italy, 7-11 Feb 2000. http://chep2000.pd.infn.it/ . [11] C.D. Jones, M. Paterno, J. Kowalkowski, L. Sexton-Kennedy, W. Tanen- baum, “The new CMS event data model and framework”, International Conference on Computing in High-Energy Physics and Nuclear Physics (CHEP2006), Mumbai, India, 13-17 Feb 2006.; algorithm”, 22 C.D. Jones, “Access to non-event data for CMS”, ibid; See also http://cmsdoc.cern.ch/cms/cpt/Software/html/General/. [12] T. DeYoung, “Icetray: A software framework for IceCube”, International Conference on Computing in High-Energy Physics and Nuclear Physics (CHEP2004), Interlaken, Switzerland, 27 September - 1 October 2004. http://www.chep2004.org . [13] S. Sciutto, AIRES User’s Manual and Reference Guide, http://www.fisica.unlp.edu.ar/auger/aires ; S.J. Sciutto, “AIRES: A system for air shower simulations (version 2.2.0),” arXiv:astro-ph/9911331. [14] D. Heck, J. Knapp, J.N. Capdevielle, G. Schatz, T. Thuow, Report FZKA 6019 (1998). [15] T. Bergmann et al., “One-dimensional hybrid approach to extensive air shower simulation”, Astropart. Phys. 26, 420 (2007). [16] H.J. Drescher, G. Farrar, M. Bleicher, M. Reiter, S. Soff, H. Stoecker, “A fast hybrid approach to air shower simulations and applications”, Phys.Rev. D67, 116001 (2003). [17] http://root.cern.ch/. [18] E. Moyse, F. Akesson, “Event data model in ATLAS”, International Conference on Computing in High-Energy Physics and Nuclear Physics (CHEP2006), Mumbai, India, 13-17 Feb 2006. http://www.tifr.res.in/∼chep06/ . [19] http://dev.mysql.com ; See for example Jon Stephens, Chad Russell, “Beginning MySQL Database Design and Optimization: From Novice to Professional” Apress, 2004, ISBN 1-59059-332-4. [20] http://xml.apache.org/. [21] see for example N. Josuttis, “The C++ Standard Library”, Addison- Wesley, 1999, ISBN 0-201-37926-0. [22] http://www.w3.org/TR/REC-xml/ . [23] http://cppunit.sourceforge.net/doc/1.8.0/ . [24] http://buildbot.sourceforge.net/ . [25] http://www.boost.org/ ; See for example Bjorn Karlsson, “Beyond the C++ Standard Library: An Introduction to Boost”, Addison-Wesley, 2005, ISBN 0-3211-3354-4. [26] http://proj-clhep.web.cern.ch/proj-clhep/. [27] http://geant4.cern.ch/; S. Agostinelli et al., Nucl. Instrum. Meth. A 506, 250 (2003); J. Allison et al., IEEE Transactions on Nuclear Science, 53, 270 (2006). [28] http://www.python.org/ ; Mark Lutz, O’Reilly, 2001, ISBN: 0-596-00085-5. “Programming Python”, 23
astro-ph/0601674
1
0601
2006-01-30T13:13:49
Probing the Sagittarius stream with blue horizontal branch stars
[ "astro-ph" ]
We present 2-degree field spectroscopic observations of a sample of 96 A-type stars selected from the Sloan Digital Sky Survey Data Release 3 (SDSS DR3). Our aim is to identify blue horizontal branch (BHB) stars in order to measure the kinematic properties of the tidal tails of the Sagittarius dwarf spheroidal galaxy. We confine our attention to the 44 classifiable stars with spectra of signal-to-noise ratio >15 per angstrom. Classification produces a sample of 29 BHB stars at distances 5-47 kpc from the Sun. We split our sample into three bins based on their distance. We find 10 of the 12 stars at 14-25 kpc appear to have coherent, smoothly varying radial velocities which are plausibly associated with old debris in the Sagittarius tidal stream. Further observations along the orbit and at greater distances are required to trace the full extent of this structure on the sky. Three of our BHB stars in the direction of the globular cluster Palomar (Pal) 5 appear to be in an overdensity but are in the foreground of Pal 5. More observations are required around this overdensity to establish any relation to Pal 5 and/or the Sgr stream. We emphasize observations of BHB stars have unlimited potential for providing accurate velocity and distance information in old distant halo streams and globular clusters alike. The next generation multi-object spectrographs provide an excellent opportunity to accurately trace the full extent of such structures.
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 1 -- ?? (2002) Printed 15 July 2018 (MN LaTEX style file v2.2) Probing the Sagittarius stream with blue horizontal branch stars L. Clewley⋆ & Matt J. Jarvis† Astrophysics, Department of Physics, Keble Road, Oxford, OX1 3RH, UK Released 2002 Xxxxx XX ABSTRACT We present 2-degree field spectroscopic observations of a sample of 96 A -- type stars selected from the Sloan Digital Sky Survey Data Release 3 (SDSS DR3). Our aim is to identify blue horizontal branch (BHB) stars in order to measure the kinematic properties of the tidal tails of the Sagittarius dwarf spheroidal galaxy. We confine our attention to the 44 classifiable stars with spectra of signal-to-noise ratio > 15A−1. Classification produces a sample of 29 BHB stars at distances 5 − 47 kpc from the Sun. We split our sample into three bins based on their distance. We find 10 of the 12 stars at 14 − 25 kpc appear to have coherent, smoothly varying radial velocities which are plausibly associated with old debris in the Sagittarius tidal stream. Further obser- vations along the orbit and at greater distances are required to trace the full extent of this structure on the sky. Three of our BHB stars in the direction of the globular cluster Palomar (Pal) 5 appear to be in an overdensity but are in the foreground of Pal 5. More observations are required around this overdensity to establish any relation to Pal 5 and/or the Sgr stream. We emphasize observations of BHB stars have unlimited potential for providing accurate velocity and distance information in old distant halo streams and globular clusters alike. The next generation multi-object spectrographs provide an excellent opportunity to accurately trace the full extent of such structures. Key words: branch -- Galaxy: structure -- Galaxy: stream galaxies: individual (Sagittarius) -- Galaxy: halo -- stars: horizontal 1 INTRODUCTION There have been numerous searches for streams of mate- rial responsible for the build-up of stellar halos in galaxies. In the Milky Way the best studied is the Sagittarius (Sgr) dwarf spheroidal galaxy (Ibata, Gilmore & Irwin 1994) and its stellar stream (e.g. Majewski et al. 2003). An extended stream of stars has also been uncovered in the halo of the Andromeda galaxy (M31), revealing that it too is cannibaliz- ing a small companion (e.g. Lewis et al. 2004). Such streams yield crucial information on the merging and accretion his- tory of galaxy halos and have been used to constrain the mass of the Milky Way halo (e.g. Johnston et al. 1999) and the mass of the halo in M31 (Ibata et al. 2004). There are numerous studies reporting stellar streams as- sociated with the tidal debris of the Sgr system either trail- ing or leading it along its orbit. The stars involved in the tidal disruptions can be used as test particles which can lead to an understanding of the shape (Helmi 2004) and strength of the Milky Way potential. A number of studies have sought ⋆ [email protected] † [email protected] to trace Sgr and its stream with a variety of stellar popula- tions. For example, Totten & Irwin (1998) found evidence for such streams using intrinsically rare, but very luminous, car- bon stars. The calibration of the carbon star distance scale remains controversial and is further complicated by variabil- ity, dust and metallicity effects. Numerous RR Lyrae stars have also been found to be associated with the Sgr stream (Ivezi´c et al. 2000; Vivas et al. 2001; Vivas et al. 2005). The Quasar Equatorial Survey Team (QUEST) survey (Vivas at al. 2004) found 85 RR Lyrae variables covering about 36◦ in right ascension, of which 16 were identified spectroscopi- cally (Vivas et al. 2005). An all-sky view of the Sgr stream has been investigated using M giants selected from the Two Micron All-Sky Survey (2MASS) database (Majewski et al. 2003, 2004). There is strength in diversity. Heterogeneous stellar populations not only provide an important check on the distance and velocity measurements of any putative stream but also probe different formation epochs. Blue Horizontal Branch (BHB) stars provide yet another stellar population ideal for exploring Halo structure (Sirko et al. 2004; Brown et al., 2004; Clewley et al., 2005). Newberg at al. (2002) find a distant sample of BHB stars that, they suggest, reside in 2 Clewley & Jarvis the trailing arm of the Sgr stream. Further, Monaco et al. (2003) suggest that a sample of BHB stars discovered in the core of the Sgr dwarf spheroidal galaxy are the counterpart of the stars observed by Newberg et al. (2002). Both BHB stars and RR Lyraes are excellent standard candles, enabling us to determine their distances to 5-10%; by contrast the distances to K and M giants are accurate to 20% - 25% (e.g. Dohm-Palmer et al. 2001; Majewski et al. 2003). BHB and RR Lyrae stars are also old and metal poor, so they are ideal for tracing old parts (> 5 Gyr) of tidal streams. The age of the Sgr M giants is controversial. Ma- jewski et al. 2003 considered these stars to be younger than 5 Gyr with a significant fraction of them only 2−3 Gyrs old (Majewski et al. 2003). This poses two problems for their use as tracers. First, they may not be useful halo test particles as such dynamically young tracers do not place very stringent constraints on the halo (Helmi 2004). Second, comparisons between the evolution timescales of the M giants and the dynamical timescale required for the stars to populate the stream suggest that the stars exist in a stream that took more than their age to form. This inconsistency has been addressed by Bellazzini et al. (2005) who finds the M gi- ants were considerably older (> 5 Gyr) than was originally thought and may even probe similar epochs as RR Lyraes or BHB stars. This paper is the first in a series that discusses the use of BHB stars to directly trace the Sgr stream. BHB stars are A -- type giants that are easily identified in the Galactic halo as they lie blueward of the main -- sequence turnoff (e.g. Yanny et al. 2000). Assembling clean samples of remote BHB stars has been stymied by the existence of a contaminating population of high -- surface -- gravity A -- type stars, the blue stragglers, that are around two magnitudes fainter. Previ- ous analyses required high signal-to-noise ratio (S/N ) spec- troscopy to reliably separate these populations (e.g. Kin- man, Suntzeff & Kraft 1994), making identification of BHB stars in the distant halo unfeasible. However, recently Clew- ley et al. (2002, 2004) developed two classification methods that now enable us to overcome the difficulties of cleanly separating BHB stars from blue stragglers. In this paper we use these classification techniques, and the Sloan Digital Sky Survey Data Release 3 (SDSS DR3) photometry, to isolate a sample of BHB stars along the Sgr stream. In this new survey we target stars along the equato- rial strip in the region 180◦ <RA< 249◦, which are sensitive to the different models that distinguish between halo flatness using the Sgr stream (e.g. Mart´ınez-Delgado et al. 2004). In Section 2 of this paper we describe the selection of the BHB candidates from the SDSS data set, and outline the prescription for transforming the SDSS g, r magnitudes of A -- type stars to B, V magnitudes detailed in Clewley et al. (2005). Section 3 provides a summary of the 2dF spectro- scopic observations and data reduction procedures. In Sec- tion 4 we classify these stars into categories BHB and blue straggler. Section 5 presents the results of the classification procedure and provides a summary table of distances and radial velocities of the stars classified as BHB. Three BHB stars in the direction of the globular cluster Pal 5 appear to be in an overdensity but in the foreground of Pal 5; we compare our measurements with those in the literature. In Section 6 we investigate the evidence that the stars reside in a stream and are associated with the Sgr tidal debris. Figure 1. Upper: A plot of RA against g for all the BHB star candidates in the SDSS satisfying our selection criteria (detailed in the text). Selection was confined to the equatorial strip and the range 180◦ <RA< 250◦ . There are 610 stars satisfying the selection criteria. Lower: A plot of RA against g for the 96 stars observed. The coordinates of the eight pointings are provided in Table 1. Filled circles are stars that are classified BHB. The large filled circles are candidate stream members. Finally, in Section 7 we provide a summary of the main con- clusions of the paper. In this paper we use the coordinate Rgal to denote Galactocentric distances and the coordinate R⊙ to denote heliocentric distances. 2 SELECTING THE BHB CANDIDATES 2.1 Colour selection Our aim was to use the SDSS photometric data to select a sample of candidate faint BHB stars, with minimal contam- ination by quasi-stellar objects (QSOs) and F -- type stars. We selected the candidate BHB stars using the SDSS point spread function (PSF) ugr magnitudes of stellar objects. The SDSS is 95% complete for point sources to (u, g, r, i, z) = (22.0, 22.2, 22.2, 21.3, 20.5) and is saturated at about 14 magnitudes in g, r and i and about 12 mag. in u. Conse- quently, we select stars with 14.3 < g < 19.5. The fainter limit was added in retrospect and is a consequence of the 2dF instrument capability and the observing conditions cou- pled with our strict signal-to-noise (S/N) requirements for classification, which we outline below. All the SDSS magni- tudes discussed in this paper have been corrected for Galac- tic extinction, using the map of Schlegel, Finkbeiner & Davis (1998). We limited ourselves to the equatorial stripe in the SDSS data set which is observable from the southern hemi- sphere, with 180◦ < RA < 249◦ (J2000). In selecting candidate distant BHB stars from the SDSS, we used the results of Yanny et al. (2000), who stud- ied the spatial distribution of a sample of A -- type stars se- lected from the SDSS using the (reddening -- corrected) colour Probing the Sagittarius stream with blue horizontal branch stars 3 No. l b 01 02 03 04 05 06 07 08 277.27 329.12 345.38 355.34 1.42 6.59 9.64 15.00 60.38 60.28 55.23 49.97 45.57 40.90 37.66 31.00 RA J 2000 12 02 13 42 14 22 14 54 15 18 15 42 15 58 16 30 Table 1. Galactic and equatorial coordinates of the centres of the eight fields observed in the survey, ordered by Right Ascension. The declination of all eight fields are centered along the equator. selection box −0.3 < g − r < 0.0, 0.8 < u − g < 1.4. On this basis, we adopted the colour cuts defined by Yanny et al. (2000), and limited candidate selection to objects with colour error σ(g − r) < 0.07. There are 610 objects in this sample. While these colour cuts should be nearly optimal in terms of the fraction of candidates that are A -- type stars, we still expect substantial contamination by blue stragglers. We use two methods to classify stars into categories BHB and blue straggler. As one method makes use of (B − V )0 colours we need to convert the extinction corrected g −r colours to B − V . In Clewley et al. (2005) we described a re- investigation of this conversion and compared it to Fukugita et al. (1996) and Smith (2002). We derived a cubic relation that provides an improved fit. The transformation is given by B −V = 0.764(g−r)−0.170(g−r)2 +0.715(g−r)3 +0.218.(1) We stress that this relation is specifically for A -- type stars, and is not expected to be reliable for other types of star. We chose 8 pointings roughly equidistant in RA along the stream. The coordinates of the centres of the pointings are provided in Table 1. The final selection includes 96 A -- type stars which are listed in Table 2. In this table Column 1 is our running number, and column 2 lists the coordinates. Successive columns provide the dereddened SDSS g magni- tude, and the dereddened u − g and g − r colours. The last column provides the dereddened B − V colour, calculated using the transformation shown in Equation 1. Looking ahead, our classification methods were devel- oped specifically for objects with strong Balmer lines, de- fined by EW Hγ > 13A, and with a signal-to-noise (S/N) in the continuum > 15A−1. Of the 96 selected candidates we are able to adequately classify 44 stars, of these 29 were clas- sified BHB. For the 52 stars that we are unable to classify the majority (37 stars) were because the S/N was insuffi- cient. In addition, there were 4 QSOs and 11 stars with EW Hγ < 13A. Figure 1 shows the distribution in RA and g magnitude of the selections. In Figure 1(lower) the 96 colour selected stars are shown as open circles with solid circles representing the 29 stars that are classified BHB. The horizontal stripes represent the pointings. Figure 2. CaII K line (3933 A) EW(A) for the sample plotted against (B − V )0. The curves represent lines of constant metallic- ity for [Fe/H] = -1.0, -2.0 and -3.0 taken from Wilhelm et al. (1999). The straight line represents a best fit to stars in the Pleiades and Coma clusters assumed to be of solar metallicity. The vertical line at (B − V )0 = 0.05 is the limit for which metal- licities can be determined. The 29 stars classified as BHB stars are marked by filled circles, the 15 stars classified blue straggler are marked by open circles. 3 SPECTROSCOPIC OBSERVATIONS Medium resolution spectra were obtained for the 96 candi- dates in eight pointings spread over four nights from 11- 14 May 2005 using the two-degree field (2dF; Lewis et al. 2002) multi-object spectrograph on the 3.9-m Anglo- Australian Telescope (AAT). The instrument was equipped with a 20482 Tek CCD, with a projected scale of 0.2′′pixel−1. We used the R1200B grating, giving a dispersion of 1.1 A pixel−1 and a spectral coverage of 3800 -- 4900 A, which includes the relevant lines Hδ, Hγ, and Ca II K λ3933A. With the 1′′fibres the FWHM resolution, measured by fit- ting Gaussian profiles to strong unblended arc lines, was 2.6 A which is sufficient for the line -- fitting procedure. Four BHB radial velocity standard stars in the globular cluster M5 were also observed nightly. Table 3 summarizes relevant information for the standards. This table provides a list of the identification, RA and Dec., V magnitude, (B − V )0 colour, and the heliocentric radial velocity, V⊙. Successive columns (6) to (11) in Table 3 contain averages of Hδ and Hγ lines measured from a S´ersic function discussed in more detail in §4. Total integration times between 7200 and 12000 seconds were chosen (split into 2400 second exposures) using the 2dF exposure-time calculator, based on the seeing, transparency, and lunar phase. This was done in order to achieve a mini- mum continuum S/N ratio of 15 A , which was required to classify the stars (Clewley et al. 2002). In the event, obser- vations of 37 of the 96 targets failed to achieve the required S/N , and therefore these targets cannot be reliably clas- sified. The failures were primarily in the cases where the seeing was poor and the targets were faint. All targets were observed near culmination, with a mean airmass of 1.2±0.1. −1 4 Clewley & Jarvis No. (1) 01 02 03 04 05 06 07 08 09 Identification (J2000) (2) g (3) (u − g)0 (4) (g − r)0 (5) (B − V )0 (6) J120341.440+001400.96 J155953.064+002931.70 J163544.242+001048.25 J143156.510+001251.44 J151955.913+001617.05 J120339.027+000522.44 J163518.461-001113.78 J125902.040+001331.21 J134332.838+001243.41 18.703 ± 0.020 19.018 ± 0.015 17.095 ± 0.010 15.668 ± 0.022 15.061 ± 0.015 14.421 ± 0.046 15.936 ± 0.008 18.259 ± 0.022 18.940 ± 0.018 1.092 ± 0.037 1.078 ± 0.054 1.132 ± 0.023 1.105 ± 0.030 1.204 ± 0.017 1.338 ± 0.060 1.051 ± 0.020 1.236 ± 0.039 1.184 ± 0.048 -0.024 ± 0.026 -0.094 ± 0.021 -0.192 ± 0.014 -0.117 ± 0.026 -0.054 ± 0.020 -0.096 ± 0.054 -0.224 ± 0.012 -0.212 ± 0.033 -0.093 ± 0.025 0.200 ± 0.0200 0.144 ± 0.0169 0.060 ± 0.0126 0.125 ± 0.0214 0.176 ± 0.0157 0.143 ± 0.0431 0.031 ± 0.0113 0.042 ± 0.0301 0.145 ± 0.0201 Table 2. Photometric data for the 96 target BHB candidates taken from the SDSS. Column (1) is our running number, column (2) lists the coordinates. Columns (3) to (5) list the dereddened g magnitude, u − g, g − r colours. The last column provides the (B − V )0 colour. The table is presented in its entirety in the electronic edition of Monthly Notices. The image frames were reduced with the 2dF pipeline (2dfDR v3.31, Bailey et al. 2003). This package carries out bias subtraction, flat-fielding, tram-line mapping to the fi- bre locations on the CCD, fibre extraction, arc identification and wavelength calibration. For the tram-line mapping and wavelength calibration, fibre flat-field frames and CuAr arc observations were made before and after each 45 min ex- posure. The arcs were used to derive accurate positions of the spectra on the CCD and to calculate the time-varying dispersion solution due to flexure of the instrument. This approach was very successful. We found, from the scatter of the radial velocity standard stars (table 3), a rms drift in the zero point of 9 kms−1 over the entire data set. 4 SPECTROSCOPIC ANALYSIS AND CLASSIFICATION 4.1 Analysis The spectra were used, in combination with the photometry, to classify the stars, measure the metallicity and calculate the radial velocities. For these measurements we followed the procedures set out in Clewley et al. (2002, 2004) exactly, and we refer the reader to those papers for full details. Below we briefly summarize the procedures we undertake to measure the Balmer lines, metallicities and distances. For the Balmer lines we normalized each spectrum to the continuum, and fitted a S´ersic function, convolved with a Gaussian that has the FWHM of the instrumental resolu- tion. We have two classification procedures. First, the Scale width -- Shape method, plots the scale width b, and the shape index c. Second, the D0.15 -- Colour method plots D0.15, which is the line width at a depth 15% below the continuum against (B − V )0. With some exceptions (including the A metallic (Am) and peculiar (Ap) stars) the strength of the Ca II K line at constant temperature can be used as an indicator of the metallicity of A-type stars (e.g. Pier 1983, Beers et al. 1992, Kinman et al. 1994). The metallicities were determined from the Ca II K 3933A , line by minimum−χ2 fitting a Gaus- sian to the continuum divided spectrum over the wavelength range 3919 -- 3949 A. The metallicities were derived by plot- ting Ca II K line EW against (B − V )0 and interpolating between lines of constant metallicity (see Figure 2). The uncertainty is established from the uncertainties of the two quantities plotted, and an additional uncertainty of 0.3dex is added in quadrature. This systematic error was established by using our methods on high S/N stars with known metal- licities. As the lines in Figure 2 converge towards the bluer colours we consider this plot only reliable for colours redder than (B − V )0 > 0.05. No attempt has been made to remove the possible contribution of interstellar Ca II K absorption from the stellar K measurements. The absolute magnitudes of BHB stars, MV (BHB), depend on both metallicity and colour (i.e. temperature). Preston, Shectman & Beers (1991) provided an empirical Luminosity-Colour relation for BHB stars derived from a fit to globular cluster BHB stars. Recently, this empirical re- lation was re-investigated by Brown et al. (2005), who pro- vided a physical basis for this relationship based on theoret- ical modelling. In Clewley et al. (2004) we derived a relation for the absolute magnitude of BHBs in two steps. First, we used the results from the Clementini et al. (2003) to deter- mine the slope of the relation, MV (RR) = α + β[Fe/H]. We fixed the zero point of the relation using the measurement by Gould & Popowski (1998) of the absolute magnitude of RR Lyrae stars, MV (RR) = 0.77 ± 0.13 mag at [Fe/H] = -1.60, derived from statistical parallaxes of Hipparcos ob- servations. Combining these two measurements resulted in MV (RR) = 1.112 + 0.214[Fe/H]. Second we adopted the cu- bic expression determined by Preston et al. (1991) for the (B − V )0 colour dependence of the difference in absolute magnitudes between BHB and RR Lyrae stars. This pro- duced the final expression for the absolute magnitude of BHB stars: MV (BHB) = 1.552 + 0.214[Fe/H] − 4.423(B − V )0 +17.74(B − V )2 0 − 35.73(B − V )3 0. (2) Distances and associated errors are then determined using the apparent magnitudes V0, and the corresponding photo- metric and metallicity errors. To compute V , we used the relation V = g′−0.53(g′ −r′) (Fukugita et al. 1996), here dis- regarding the subtle differences between the different SDSS magnitudes (g, g′, g∗, etc.). The result produces distance er- rors of 6 − 10% for our confirmed BHB stars. We caution that the MV -- metallicity relation is still controversial and so distance measurements may be systematically in error. We Probing the Sagittarius stream with blue horizontal branch stars 5 ID (1) RA (J2000) Dec. V0 (B − V )0 (2) (3) (4) V⊙ [km s−1] (5) D0.15(γδ) [A] (6) M5-II-78 J151826.93 + 020717.78 14.95 0.12 42.2 ± 1.1 M5-IV-05 J151835.34 + 020227.94 15.15 0.15 56.9 ± 1.2 M5-III-69 J151830.43 + 020224.57 15.06 0.18 56.2 ± 1.5 30.322 ± 0.501 30.276 ± 0.459 31.675 ± 0.482 31.510 ± 0.500 28.654 ± 0.489 29.282 ± 0.465 b(γδ) [A] (7) 7.778 7.966 8.365 8.421 7.196 7.248 c(γδ) A B θ (8) (9) (10) (11) 0.799 0.829 0.836 0.851 0.775 0.754 0.196 0.184 0.192 0.197 0.189 0.179 0.021 0.021 0.020 0.022 0.020 0.018 1.5139 1.5131 1.5164 1.5171 1.5113 1.5131 M5-I-53 J151836.35 + 020744.60 15.06 0.06 52.2 ± 1.4 31.966 ± 0.454 8.372 0.827 0.182 0.019 1.5173 Table 3. Spectroscopic measurements of four M5 globular cluster BHB stars. The names are from Arp (1955) and Arp (1962), the photometry is from Cudworth (1979), and the radial velocities are from Peterson (1983). Columns (1) and (2) are the number and coordinates. Columns (3) and (4) list the dereddened V magnitude and B − V colours. The radial velocity, corrected to the heliocentric frame, is provided in column (5). Successive columns (6) to (11) contain averages of Hδ and Hγ line measurements measured from a S´ersic function. Column (6) is the line width at a depth 15% below the continuum and the parameters in (7) and (8) are the scale width (b), and the shape index (c). The errors in these latter two quantities are shown in (9) to (11), these are A and B, the semi-major and semi-minor axes of the error ellipse in the b − c plane, and θ the orientation of the semi-major axis, measured anti-clockwise from the b-axis. Here the error corresponds to the 68% confidence interval for each axis in isolation [see Clewley et al. (2002) for further details]. Figure 3. Classification of all 44 stars with unambiguous classifications shown in Table 4, using the (a) the Scale width -- Shape and (b) D0.15 -- Colour classification methods. The solid curves are the classification boundaries explained in the text. Filled circles are stars classified BHB in both plots, i.e. the stars below the classification boundary in each plot. Open circles are stars classified A/BS. Filled triangles are stars below the classification boundary in only one plot but are nonetheless classified BHB. In plot (a) there are 23 stars below the boundary and 21 above it. In plot (b) there are 29 stars below the boundary and 15 above it. A total of 29 stars are classified BHB. refer the interested reader to a recent review of this subject by Cacciari & Clementini (2003) and Alves (2004). The absolute magnitudes of blue stragglers have been less well studied. As we do not analyse them here we sim- ply adopt the relation, MV (BS) = 1.32 + 4.05(B − V )0 − 0.45[Fe/H], derived by Kinman et al. (1994). 4.2 Classification In Figure 3 we plot the two diagnostic diagrams for the 44 classifiable stars in the survey. Figure 3(a) shows the Scale width -- Shape method. The line -- profile quantities b and c, averaged for Hγ and Hδ are plotted. Figure 3(b) shows the D0.15−colour method. Plotted are values of D0.15 for Hγ and Hδ against (B − V )0 for all the stars. The solid lines show the classification boundaries, taken from Clewley et al. (2002, 2004), with high -- surface gravity stars (i.e. main -- sequence A stars or blue stragglers, hereafter A/BS) above the line, and low -- surface gravity stars (i.e. BHB stars) below the line. In both plots stars classified BHB are plotted as solid symbols and stars classified A/BS are plotted open. As we discuss below, the eight triangles are stars that are classified as BHB by one classification method and not the other but were ultimately classified as BHBs. Figure 3 shows that of the 44 candidates, 29 are classi- fied BHB by the D0.15−colour method, and 23 are classified BHB by the Scale width -- Shape method. There are 21 stars classified BHB by both methods. A total of ten stars are clas- 6 Clewley & Jarvis No. (1) 03 04 05 06 12 13 14 16 17 18 20 21 22 24 25 26 28 29 30 31 32 33 35 36 42 43 44 47 48 49 51 58 63 64 66 67 68 70 72 73 74 75 76 77 78 79 80 81 82 83 87 88 92 93 94 96 99 100 102 S/N [A]−1 (2) D0.15 (γδ) [A] (3) 39.96 34.59 45.28 32.42 23.40 45.55 33.77 21.95 15.22 12.51 25.18 28.40 15.09 70.09 27.50 30.97 44.21 36.83 50.47 24.74 36.99 17.23 30.54 70.46 34.81 17.93 19.96 39.99 32.27 71.92 10.34 16.50 12.25 27.34 40.67 34.66 27.66 32.80 15.00 15.53 12.07 15.93 20.07 11.21 54.50 16.84 13.43 48.89 22.99 26.32 12.54 39.76 95.40 32.73 15.04 31.72 36.32 94.98 50.83 31.90 ± 0.43 40.47 ± 0.45 24.90 ± 0.33 26.57 ± 0.41 30.43 ± 0.74 37.64 ± 0.40 38.67 ± 0.47 32.93 ± 0.69 37.35 ± 1.10 25.45 ± 1.28 34.61 ± 0.65 36.53 ± 0.53 26.39 ± 1.16 28.30 ± 0.23 34.55 ± 0.60 41.41 ± 0.51 32.60 ± 0.45 31.44 ± 0.47 31.73 ± 0.28 33.37 ± 0.59 31.38 ± 0.52 44.10 ± 1.05 29.52 ± 0.44 29.53 ± 0.21 33.70 ± 0.49 31.68 ± 1.08 28.90 ± 0.85 37.10 ± 0.43 29.68 ± 0.44 32.40 ± 0.25 35.38 ± 1.68 26.50 ± 0.89 37.52 ± 1.42 23.89 ± 0.60 25.95 ± 0.38 28.71 ± 0.44 41.95 ± 0.63 30.96 ± 0.43 25.67 ± 1.09 26.14 ± 0.99 25.67 ± 1.30 32.21 ± 0.98 33.22 ± 0.84 41.70 ± 1.64 31.09 ± 0.31 35.34 ± 1.10 29.85 ± 1.20 29.55 ± 0.34 31.67 ± 0.65 27.84 ± 0.62 26.92 ± 1.33 28.17 ± 0.41 27.89 ± 0.17 32.11 ± 0.43 29.84 ± 1.22 30.63 ± 0.56 30.68 ± 0.40 30.73 ± 0.17 33.12 ± 0.42 b(γδ) c(γδ) A B θ EW(CaIIK) [Fe/H] [A] (4) 9.44 10.09 6.18 6.36 8.33 10.04 10.14 8.94 11.23 5.99 8.07 8.54 6.39 8.94 8.37 11.19 9.79 7.41 9.09 9.61 9.72 11.98 8.53 8.57 7.97 7.96 6.88 8.90 8.58 9.95 8.46 6.51 9.23 5.22 5.81 7.22 11.21 8.69 6.27 8.14 5.53 8.83 7.79 11.80 8.75 8.73 6.57 8.59 8.92 7.16 5.67 6.29 8.56 8.80 8.65 7.12 8.74 9.62 9.67 (5) (6) (7) (8) [A] (9) (10) 1.00 0.76 0.74 0.73 0.88 0.85 0.82 0.87 1.01 0.67 0.70 0.72 0.72 1.16 0.74 0.86 1.05 0.72 0.96 0.97 1.12 0.88 0.98 0.97 0.72 0.78 0.72 0.73 0.98 1.09 0.72 0.71 0.72 0.62 0.67 0.78 0.85 0.93 0.68 1.13 0.58 0.88 0.71 0.93 0.93 0.76 0.61 0.98 0.92 0.79 0.61 0.66 1.09 0.88 0.98 0.70 0.95 1.13 0.99 0.16 0.19 0.13 0.16 0.28 0.16 0.19 0.27 0.44 0.47 0.26 0.21 0.45 0.08 0.24 0.20 0.17 0.18 0.11 0.23 0.18 0.41 0.17 0.08 0.19 0.44 0.33 0.17 0.17 0.10 0.69 0.35 0.55 0.30 0.14 0.18 0.25 0.17 0.45 0.41 0.50 0.38 0.33 0.72 0.12 0.46 0.48 0.12 0.26 0.23 0.51 0.15 0.06 0.17 0.44 0.22 0.15 0.06 0.15 0.025 0.013 0.015 0.016 0.036 0.015 0.016 0.031 0.053 0.050 0.019 0.015 0.048 0.019 0.019 0.017 0.029 0.015 0.015 0.033 0.035 0.035 0.027 0.013 0.014 0.039 0.030 0.012 0.027 0.018 0.050 0.038 0.046 0.049 0.013 0.018 0.021 0.023 0.045 0.089 0.036 0.044 0.024 0.066 0.017 0.035 0.040 0.021 0.034 0.028 0.034 0.012 0.013 0.020 0.074 0.017 0.023 0.012 0.023 1.526 1.529 1.503 1.508 1.517 1.524 1.524 1.518 1.531 1.509 1.522 1.525 1.509 1.529 1.524 1.530 1.527 1.521 1.515 1.516 1.533 1.533 1.512 1.510 1.522 1.517 1.513 1.525 1.509 1.530 1.520 1.504 1.526 1.505 1.510 1.508 1.529 1.514 1.502 1.494 1.511 1.521 1.523 1.520 1.519 1.523 1.516 1.519 1.516 1.513 1.515 1.515 1.516 1.514 1.515 1.516 1.513 1.526 1.528 0.992 ± 0.13 1.411 ± 0.13 2.726 ± 0.12 4.098 ± 0.13 1.343 ± 0.14 0.959 ± 0.12 1.385 ± 0.13 1.608 ± 0.15 1.374 ± 0.17 2.668 ± 0.18 1.286 ± 0.14 1.402 ± 0.14 1.861 ± 0.18 0.565 ± 0.11 1.790 ± 0.14 2.538 ± 0.13 1.494 ± 0.12 2.019 ± 0.13 0.575 ± 0.12 0.948 ± 0.14 1.104 ± 0.13 1.321 ± 0.16 0.657 ± 0.13 0.407 ± 0.11 4.973 ± 0.13 1.028 ± 0.16 1.515 ± 0.15 2.058 ± 0.13 0.681 ± 0.13 0.653 ± 0.11 1.447 ± 0.20 1.335 ± 0.16 0.506 ± 0.18 2.034 ± 0.14 1.619 ± 0.12 1.905 ± 0.13 1.462 ± 0.14 1.486 ± 0.13 2.263 ± 0.18 0.374 ± 0.19 2.659 ± 0.18 1.171 ± 0.17 2.308 ± 0.15 1.346 ± 0.19 1.879 ± 0.12 1.387 ± 0.16 2.778 ± 0.17 1.000 ± 0.12 1.393 ± 0.14 2.077 ± 0.14 1.618 ± 0.18 1.588 ± 0.13 0.595 ± 0.11 1.088 ± 0.13 0.403 ± 0.18 2.394 ± 0.13 0.700 ± 0.13 0.717 ± 0.11 0.986 ± 0.12 -1.42 ± 0.16 -1.49 ± 0.27 -0.99 ± 0.29 0.00 ± 0.51 -1.09 ± 0.37 -1.89 ± 0.11 ... ... ... -1.31 ± 0.28 -1.23 ± 0.31 -1.13 ± 0.26 -1.67 ± 0.24 -1.88 ± 0.21 -1.50 ± 0.24 -1.58 ± 0.07 -1.44 ± 0.25 -0.67 ± 0.35 -1.46 ± 0.13 -1.60 ± 0.11 ... ... ... -1.33 ± 0.36 -1.47 ± 0.14 -1.71 ± 0.12 ... ... ... ... ... ... 0.00 ± 0.00 -1.77 ± 0.10 -1.72 ± 0.11 -1.37 ± 0.12 ... ... ... -1.84 ± 0.11 -1.68 ± 0.23 -1.57 ± 0.25 -2.93 ± 0.08 -1.17 ± 0.43 -1.83 ± 0.20 -1.28 ± 0.13 -1.11 ± 0.17 -1.12 ± 0.34 -1.31 ± 0.25 -1.71 ± 0.44 -1.04 ± 0.28 -2.06 ± 0.15 -1.29 ± 0.12 -1.71 ± 0.23 -1.26 ± 0.14 -1.82 ± 0.20 -0.74 ± 0.34 -1.74 ± 0.11 ... ... ... -1.48 ± 0.12 -1.71 ± 0.22 -1.88 ± 0.10 -1.52 ± 0.06 ... ... ... ... ... ... -1.34 ± 0.12 -1.78 ± 0.20 -1.57 ± 0.16 -1.58 ± 0.14 V⊙ [km s−1] (11) R⊙ [kpc] (12) -210.19 ± 09.49 -256.56 ± 08.20 -41.56 ± 05.17 60.82 ± 10.62 -122.20 ± 13.04 94.01 ± 09.00 -74.77 ± 06.16 -85.53 ± 08.73 -46.94 ± 17.74 -38.52 ± 14.58 102.15 ± 09.83 73.66 ± 05.07 -1.47 ± 12.58 -45.74 ± 09.75 9.73 ± 06.20 21.22 ± 07.40 -150.49 ± 07.19 34.95 ± 04.89 84.78 ± 08.22 -31.73 ± 07.07 27.57 ± 08.63 -236.21 ± 12.42 -124.38 ± 09.08 -79.52 ± 10.22 18.81 ± 14.56 160.75 ± 15.77 104.07 ± 10.50 18.68 ± 06.50 191.31 ± 09.09 -179.80 ± 15.74 -9.55 ± 18.51 70.13 ± 14.54 -184.24 ± 21.74 38.66 ± 12.05 -16.48 ± 06.52 -214.94 ± 06.28 -218.00 ± 08.86 -95.59 ± 06.81 49.97 ± 17.98 -10.93 ± 17.91 100.23 ± 15.48 -11.71 ± 15.06 53.50 ± 07.38 -1.98 ± 16.19 64.57 ± 03.66 -17.16 ± 18.60 -13.19 ± 20.07 84.43 ± 07.77 40.56 ± 08.55 -25.56 ± 05.57 -66.15 ± 17.24 -114.59 ± 07.40 -48.73 ± 10.09 -147.60 ± 09.00 13.16 ± 22.97 -51.85 ± 06.27 -56.20 ± 06.98 -113.25 ± 08.80 -5.25 ± 08.00 17.16 ± 1.07 4.42 ± 0.58 7.18 ± 0.45 5.28 ± 0.33 16.56 ± 1.04 8.38 ± 1.13 9.09 ± 0.96 22.99 ± 1.44 7.61 ± 0.90 55.48 ± 3.47 6.83 ± 0.93 4.91 ± 0.77 47.12 ± 2.95 7.80 ± 0.49 6.91 ± 1.00 2.32 ± 0.30 5.98 ± 0.37 5.43 ± 0.89 6.99 ± 0.44 17.57 ± 1.10 8.14 ± 0.51 12.51 ± 1.76 9.94 ± 0.62 8.08 ± 0.51 2.85 ± 0.45 16.69 ± 1.04 34.63 ± 2.17 5.59 ± 0.84 9.62 ± 0.60 9.44 ± 0.59 8.12 ± 1.17 37.35 ± 2.34 12.19 ± 1.99 19.34 ± 1.21 7.47 ± 0.47 11.41 ± 0.71 12.89 ± 1.50 18.70 ± 1.17 24.22 ± 1.52 17.24 ± 1.08 50.10 ± 3.14 11.14 ± 0.70 11.27 ± 1.74 18.60 ± 2.63 10.06 ± 0.63 5.79 ± 0.87 39.86 ± 2.49 13.26 ± 0.83 11.75 ± 0.74 17.97 ± 1.12 50.53 ± 3.16 6.81 ± 1.14 5.27 ± 0.33 9.33 ± 0.58 19.33 ± 1.21 6.34 ± 1.02 19.24 ± 1.20 5.68 ± 0.36 17.35 ± 1.09 Class. (13) BHB A/BS BHB BHB BHB A/BS A/BS BHB A/BS BHB? A/BS A/BS BHB BHB? A/BS A/BS BHB? A/BS? BHB BHB BHB? A/BS BHB BHB A/BS BHB BHB A/BS BHB BHB? A/BS? BHB A/BS? BHB? BHB BHB A/BS BHB BHB BHB BHB? BHB A/BS A/BS? BHB A/BS BHB? BHB BHB BHB BHB? A/BS? BHB? BHB BHB A/BS BHB BHB BHB Table 4. Spectroscopic data for the BHB star candidates. The columns (3) to (8) and (11) are the same as in Table 3. In addition, columns (1) and (2) give the number of the star and the spectrum continuum S/N per A. Column (9) is EW of the CaII K line and column (10) is the measured metallicity for each star with its error. The heliocentric distance is shown in column (12). Finally, the classification of the star is provided in column (13). sified BHB by one or other of the methods. There is clearly close agreement between the two classification methods. For the ambiguous classifications we combine the information provided by all the parameters. The uncertainties on each parameter define the 2D probability distribution functions for any point. By integrating these functions below the clas- sification boundary we can compute a probability P (BHB) that any star is BHB. Of the nine stars with ambiguous clas- sification eight are classified BHB. We also classify the re- maining 15 stars with inadequate spectroscopic S/N , but for these the classifications are given as BHB? or A/BS? to indi- cate that they are not reliable. We note that the four radial velocity standards (Table 2), previously classified BHB from high -- resolution spectroscopy, are all unambiguously classi- fied BHB in both plots. 5 RESULTS Of the 96 selected candidates only 44 have suitable spec- troscopic S/N and Hγ EW for reliable classification. The classification of these 44 candidates results in 29 BHB stars. We have nevertheless followed the classification procedures for the remaining 52 objects, but for clarity have omitted them from Figures 2 and 3. For these objects the final clas- sifications are flagged as questionable. The results of the measurements for all the candidate BHB stars are provided in Table 4. Table 5 contains a summary of the kinematic proper- ties of the final sample of 29 reliably classified BHB stars. Listed are the identification, RA and Dec., heliocentric ve- locity and distance, Galactic coordinates l and b and the Galactocentric distance and radial velocity Rgal and Vgal respectively. To convert the heliocentric quantities to Galac- tocentric quantities, the heliocentric radial velocities are first corrected for solar motion by assuming a solar peculiar ve- locity of (U, V, W ) = (-9,12,7), where U is directed outward Probing the Sagittarius stream with blue horizontal branch stars 7 No. RA (J2000) [◦] (2) 248.9343414 229.9829712 180.9126129 239.8858643 229.8426208 229.5568085 210.7175751 229.4858856 210.7082367 229.4539948 239.3611908 248.5698090 239.3440247 229.0891724 217.5550232 239.2290192 229.0451965 210.3879852 217.5059052 239.1218567 248.2692719 248.2459106 210.3329315 229.0241089 210.0473328 217.1094971 229.0039825 248.0934448 248.0232239 (1) 03 05 06 12 16 22 30 31 35 36 43 44 48 58 66 67 70 72 73 75 78 81 82 83 93 94 99 100 102 Dec. [◦ ] (3) V⊙ [km s−1] (4) R⊙ [kpc] (5) l [◦ ] (6) b [◦ ] (7) 0.1800683 0.2714026 0.0895671 -0.1522919 -0.3234651 0.1586571 0.3365301 0.4894420 0.0929688 -0.1177904 -0.3381351 -0.1052531 0.3121248 0.1762543 0.1511529 -0.4766546 -0.1293136 0.3614573 0.3670563 0.0064421 -0.3074609 -0.3497654 -0.0795943 -0.1868408 0.0858782 0.4875866 -0.2001127 -0.2781225 -0.1998732 -210.19 ± 09.49 -41.56 ± 05.17 60.82 ± 10.62 -122.20 ± 13.04 -85.53 ± 08.73 -1.47 ± 12.58 84.78 ± 08.22 -31.73 ± 07.07 -124.38 ± 09.08 -79.52 ± 10.22 160.75 ± 15.77 104.07 ± 10.50 191.31 ± 09.09 70.13 ± 14.54 -16.48 ± 06.52 -214.94 ± 06.28 -95.59 ± 06.81 49.97 ± 17.98 -10.93 ± 17.91 -11.71 ± 15.06 64.57 ± 03.66 84.43 ± 07.77 40.56 ± 08.55 -25.56 ± 05.57 -147.60 ± 09.00 13.16 ± 22.97 -56.20 ± 06.98 -113.25 ± 08.80 -5.25 ± 08.00 17.16 ± 1.07 7.18 ± 0.45 5.28 ± 0.33 16.56 ± 1.04 22.99 ± 1.44 47.12 ± 2.95 6.99 ± 0.44 17.57 ± 1.10 9.94 ± 0.62 8.08 ± 0.51 16.69 ± 1.04 34.63 ± 2.17 9.62 ± 0.60 37.35 ± 2.34 7.47 ± 0.47 11.41 ± 0.71 18.70 ± 1.17 24.22 ± 1.52 17.24 ± 1.08 11.14 ± 0.70 10.06 ± 0.63 13.26 ± 0.83 11.75 ± 0.74 17.97 ± 1.12 9.33 ± 0.58 19.33 ± 1.21 19.24 ± 1.20 5.68 ± 0.36 17.35 ± 1.09 16.0580 2.1616 277.9714 9.7586 1.3890 1.6456 338.5038 1.9410 338.2509 1.2502 9.1860 15.5566 9.8454 1.2279 348.3371 8.9468 0.8546 337.9907 348.5063 9.3660 15.1744 15.1186 337.4755 0.7725 337.1651 348.1101 0.7391 15.0951 15.1287 29.8770 45.3792 60.6244 37.2616 45.1006 45.6312 58.1867 45.8996 57.9829 45.5293 37.5826 30.0370 37.9719 45.9976 54.0754 37.6097 45.8307 58.3777 54.2734 37.9773 30.1850 30.1823 58.0257 45.8088 58.3123 54.6178 45.8152 30.3495 30.4506 Rgal [kpc] (8) 11.39 5.91 9.28 11.42 18.25 41.92 7.62 13.31 9.19 6.22 11.56 28.30 6.08 32.28 6.07 7.16 14.33 21.50 12.11 6.28 5.56 11.66 8.43 12.12 6.88 16.20 14.82 4.53 11.58 Vgal [km s−1] (9) -143.55 -24.11 -45.17 -79.61 -70.26 14.49 50.33 -14.97 -159.59 -64.68 201.41 168.95 233.88 84.87 -33.16 -175.04 -81.89 14.72 -27.09 29.36 128.09 147.76 3.84 -12.09 -184.58 -3.71 -42.82 -50.08 57.99 Table 5. Summary of positional and kinematic information for the BHB stars. Listed are the: identification, RA and Dec., heliocentric velocity and distance, Galactic coordinates l and b, and the Galactocentric distance and radial velocity Rgal and Vgal respectively. has a Heliocentric radial velocity, V⊙ = −58.7 ± 0.20 km s−1 (Odenkirchen et al. 2002), a distance R⊙ =23.2 kpc and [Fe/H] = −1.41 (Harris 1996; Clement at al. 2001). Pal 5 is remarkable as observations from the SDSS show a pair of tidal tails, which extend around 4 kpc in opposite directions from the cluster and contain more stars than the cluster itself (Odenkirchen et al. 2003; Koch et al. 2004). We compare our measurements of BHB stars around Pal 5 to previous work in the literature. There are four stars (numbers 99,83,70,31; shown as filled circles within the boxes in Figure 4) that are plausibly associated with Pal 5, i.e. they are located either within the projected cen- tral region of the cluster or in one of the tidal tails. For these four stars we measure a mean heliocentric velocity of −52.3 ± 31.8 and a mean metallicity of −1.43 ± 0.27. These val- ues are entirely consistent, albeit with large errors, with the measures by Harris (1996) and Odenkirchen et al. (2003). However the mean distances (18.4 ± 0.7 kpc) are incon- sistent with those reported in Harris (1996). Three of the four stars (numbers 99,83,70) are located in the central po- sition (see Figure 4) at distances 19.2, 18.0, 18.7 respectively (mean = 18.6 ± 0.6). The remaining star that could plausi- bly be in the stream (number 31) is at 17.57 ± 1.1 kpc. We investigate this apparent distance discrepancy further. The QUEST survey of RR Lyrae stars (Vivas et al. 2004) found five previously observed RR Lyrae stars in Pal 5 in the V band; the data for these stars are presented in Table 6. The stars in this table are 0.24 magnitudes fainter than our BHB stars, suggesting they are more dis- tant. Assuming an absolute magnitude of the RR Lyraes of MV (RR) = 1.112 + 0.214[Fe/H] (as we discuss in §4.1), and an [Fe/H] = −1.4, then the mean distance of this sample is 21.2 ± 0.93 kpc. Two further Pal 5 HB stars are available online via the Figure 4. Plot of the BHB stars with similar RA and DEC. as the Pal 5 globular cluster. The two squares are regions of Pal 5 taken from Koch et al. (2004) denoting the centre of Pal 5 (229.0,-0.1) and the trailing stream (229.5,0.5). The filled circles are stars at a similar distance and are therefore most likely to be associated with the globular cluster. from the Galactic Centre, V is positive in the direction of Galactic rotation at the position of the Sun, and W is pos- itive toward the North Galactic Pole. We have assumed a circular speed of 220 km s−1 at the Galactocentric radius of the Sun (Rgal = 8.0 kpc). 5.1 Comparisons with previous observations of Palomar 5 Our survey passes over the globular cluster Palomar (Pal) 5 (RA= 229.022083◦ , Dec.=−0.111389◦ [J2000]). This system 8 Clewley & Jarvis ID R.A. (J2000) Dec. V B − V R⊙ Refs. 399 400 401 402 404 15 15 57.25 15 15 57.97 15 15 58.31 15 16 05.79 15 16 12.79 -00 06 52.8 -00 11 23.1 -00 05 47.3 -00 11 12.3 -00 10 02.7 17.44 17.33 17.58 17.46 17.38 0.03 0.01 0.03 0.04 0.03 21.16 20.11 22.57 21.36 20.58 (1), (2) (1) (1) (1) (1) Table 6. Summary of position, colour, magnitude and distance of RR Lyrae stars observed in the QUEST survey (Vivas et al. 2004). References: (1) Kinman & Rosino (1962); (2) Clement et al. (2001); (2) Wu et al. (2005) ID 31 70 83 99 V B − V 17.185 ± 0.014 17.238 ± 0.016 17.036 ± 0.016 17.329 ± 0.016 0.043 ± 0.0166 0.083 ± 0.0174 0.196 ± 0.0147 0.067 ± 0.0178 S41 S40 17.453 ± 0.003 17.519 ± 0.002 0.103 ± 0.0031 0.087 ± 0.0023 R⊙ 17.57 18.70 17.97 19.24 20.02 20.33 Table 7. Summary of the magnitude, colours and distances for our stars in the direction of Pal 5. The two stars at the bottom of the table are HB stars in the core of Pal 5. Canadian Astronomy Data Centre(CADC)1 which are also observed in the SDSS. Like the QUEST RR Lyrae stars these stars are more than 0.2 magnitudes fainter than the those observed in this paper. Table 7 summarises the colours, and V magnitude of these two stars along with the four stars in our study. If we again assume a metallicity of −1.4 then the stars s41 and s40 are at 20.0 and 20.3 kpc respectively. We note for completeness that if the distances are derived using the SDSS colours and magnitudes as in §4.1 then we find the stars s41 and s40 are 20.9 and 21.9 kpc respectively. Therefore, it seems unlikely that our BHB stars in the direction of Pal 5 are part of the core of this cluster. How- ever, the stars 70, 83 and 99 occupy a volume of space of 0.0004 cubic kpc, which is a space density of around 8000 per cubic kpc. Kinman et al. (1994; equation 12) find a mean density of BHB stars at Rgal = 14 kpc to be approximately 0.9 per cubic kpc. The three stars therefore appear to be in an overdensity perhaps indicating that Pal 5 is extensive not only in the plane of the sky but also in depth. Clearly more observations are required to fully understand this structure and its possible relation to Pal 5 and/or the Sgr stream. 6 DISCUSSION Table 5 summarises the main observational results of the paper; a sample of BHB stars with measured radial veloc- ities in the vicinity of the Sgr stream. We now investigate whether there is evidence that these stars actually reside in this stream. In Figure 5 we plot Heliocentric radial velocity and dis- tance versus RA for the 29 BHB stars. In Figure 5(upper) we split the sample into three groups of R⊙: stars at 5 < 1 The data is available at: http://cadcwww.dao.nrc.ca/astrocat/ in the Stetson Standard Fields R⊙ < 14 kpc (open circles); (ii) stars at 14 < R⊙ < 25 kpc (filled circles); and stars at R⊙ > 25 (open triangles). We overplot on this figure simulations of the tidal stream of Sgr for various halo flatness taken from Mart´ınez-Delgado et al. (2004). The curves on this plot are for values of the axis ratio of the density distribution, qd, which range from 0.1 to 1.0 (upper to lower curves). However, the axis ratio of the potential, qp, is the quantity that determines the satellite orbit and hence the shape of the potential. This parameter is a function of qd and Galactocentric distance (see Fig. 11 in Mart´ınez-Delgado et al. 2004) and ranges from 0.5 (oblate potentials) to 1.0 (spherical potentials). In Figure 5(lower) we plot V⊙ versus RA for the same three groups of BHB stars. The plot reveals a possible cor- relation between V⊙ and RA which suggests that 10 of the 12 stars in the region 14 < R⊙ < 25 kpc could be associ- ated with a stream. It is clear, however, that the velocity distribution of the stars is systematically lower than any of the Mart´ınez-Delgado et al. (2004) models presented here. There could be many reasons for this. One plausible reason is that the Mart´ınez-Delgado model considers only the last (∼ 5 Gyr) orbit. As BHB stars are old, any putative stream might feasibly belong to a later orbit that is not considered in this model. In the next section we investigate the evi- dence that this putative stream is plausibly associated with the Sgr tidal debris. 6.1 Comparisons with previous work As we briefly discuss in §1 there have been numerous de- tections of the Sgr tidal tail around the sky using tracers that map different epochs, i.e orbits in which the debris was stripped from the satellite. For example, Majewski et al. (2003, 2004) isolated M giants in the Two Micron All- Sky Survey (2MASS) in order to map the position and ve- locity distribution of tidal debris from Sgr around the en- tire Galaxy. In Law, Johnston & Majewski (2005; hereafter LJM05) these stars are compared with numerous simulations of the tidal debris of the Sgr satellite in differing Galactic potentials. We refer the reader to this paper for details of the parameters used in this model. The simulations are cre- ated using the relatively young M giants as the principal observable. We note that BHB stars and carbon stars are expected to trace older debris than M giants. We compare these models with our BHB stars and other stellar types that are available. Before we are able to compare our data with such mod- els we need to convert our coordinates to the system de- fined in Majewski et al. (2003). In this coordinate system the zero plane of the latitude coordinate B⊙ coincides with Probing the Sagittarius stream with blue horizontal branch stars 9 Figure 5. Upper: Plot of R⊙ versus RA for the BHB stars. The BHB stars are separated in distance so that stars are at: (i) 5 < R⊙ < 14 kpc (open circles); (ii) 14 < R⊙ < 25 kpc (filled circles); and R⊙ > 25 (open triangles). The curves are taken from Mart´ınez-Delgado et al. (2004) and describe the axis ratio of the density distribution, qd. The values of qd range from 1.0 (dotted line) to 0.1 (solid line) respectively. Lower: A plot of V⊙ versus RA for the three groups. The symbols and curves are the same as those on the upper plot. the best-fit great circle defined by the Sgr debris, as seen from the Sun; the longitudinal coordinate Λ⊙ is zero in the direction of the Sgr core and increases along the Sgr trailing stream. Our sample resides in the region 260◦ < Λ⊙ < 320◦ and −17◦ < B⊙ < 16◦. This corresponds to a section of the leading portion of the orbit which is particularly sensi- tive to the halo shape. Indeed, M stars appear to strongly favour models with prolate rather than oblate halos (Helmi 2004). In contrast, by fitting planes to the leading and trail- ing debris and measuring the orbital precession of the Sgr debris Johnston, Law & Majewski (2005) find oblate halos are favoured. In Figure 6 we compare our BHB data with the best-fit numerical simulations. Overplotted are various other trac- ers which we discuss below. The M giants are represented by open stars. Figure 6 illustrates the differences for the halo potential models. We plot distances and radial veloc- ities of the best-fit simulations of the Galactic halo poten- tial, which was created by LJM05 for oblate (q=0.9, upper plots), spherical (q=1.0, middle plots) and prolate (q=1.25, lower plots) potentials. The bold points (green) are old de- bris stripped four orbits ago whilst faint points (red) are from earlier orbits. The BHB stars are shown as filled circles, open circles and filled triangles as in Figure 5. The filled circles repre- sent our putative stream. The BHB stars that are shown as open circles do not appear to be associated with the models in either velocity or distance. However, the 10 of 12 filled circles are plausibly associated with the older debris both in velocity and distance in the LJM05 model. More data is required to confirm whether the three distant BHB stars (shown as triangles) are also associated with a more distant younger part of the Sagittarius stream. They do not show a remarkable metallicity which might suggest they are in an earlier orbit of Sgr. Carbon stars (filled squares) are from the catalogue pro- vided by Totten & Irwin (1998), confining ourselves to the same Λ⊙ ranges as the BHB data and to similar distances (improved distance estimates are added from Totten, Irwin & Whitelock 2000). Some of the carbon stars (blue squares) do appear to follow the M stars in velocity, however their distances do not appear to be associated at all. This dis- crepancy might be due to the uncertainty in distance mea- surements for individual carbon stars. It is also plausible that the carbon stars trace older debris than the M giants as they can have larger ages. This sample however does not show any correlation with the BHB stars. The crosses in Figure 6 are eight metal poor K- giants discovered by Kundu et al. (2002) to have coher- ent, smoothly varying distances and radial velocities. The Kundu et al. (2002) K-giant stars represent relatively old debris stripped from Sgr three pericentric passages ago. The velocities of this sample fit the models reasonably well par- ticularly for the prolate potential shown in Figure 6(lower). However, the distances remain uncertain in all three models. In summary, the M stars and carbon stars predomi- nantly map out the earlier tidal debris. The velocity data for both tracers appears to qualitatively fit the prolate models rather well (Figure 6, lower). However the distance infor- mation is less clear, particularly for the carbon stars. The 10 Clewley & Jarvis Figure 6. Distances and radial velocities of the best-fit simulations of the Galactic halo potential created by Law, Johnston & Majewski (2005) for oblate potentials (top plots), spherical potentials (middle plots) and prolate potentials (bottom plots). The bold dots (green) are old debris stripped four orbits ago whilst faint dots (red) are from earlier orbits. Overplotted are: BHB observations (as in Fig. 5, i.e. filled and open circles and filled triangles); filled squares are carbon stars selected from Totten & Irwin (1998); open stars are M giants from Majewski et al. (2004) and crosses (magenta) are K-giants from Kundu et al. (2002). Probing the Sagittarius stream with blue horizontal branch stars 11 velocity information of the metal poor K-giants again ap- pears to fit the prolate models well but again the distance information does not support this case. The BHB data are too scarce in this study to distinguish competing models of halo flatness despite being comparitively accurate standard candles. We emphasise that these observations are over a section of the leading part of the stream. Conclusive ev- idence for halo flatness will probably require multi-epoch observational data from leading and trailing debris. 7 SUMMARY We close with a summary of the main points in this paper. We have presented the first results of a pilot study to trace out the Sgr stream using BHB stars. Spectroscopy of the A-type stars, obtained with the 2dF, produced a sample of 44 stars with data of suitable quality for classification into the classes BHB and A/BS. The final sample (Table 4) comprises 29 stars classified as BHB. The heliocentric distances range from 5 to 47 kpc, with heliocentric radial velocities accurate to 10 km s−1, on average, and distance errors < 10%. We find 10 of 12 BHB stars at 14 − 25 kpc are plausibly tracing a stream. By comparing this data with models from LJM05, we find that these stars may be associated with the older debris of Sgr both in velocity and distance. Further observations along the entire Sgr orbit both along the trail- ing and leading stream and at greater distances are clearly required to trace the full extent of this structure on the sky. A BHB analogue of the 2MASS survey would comple- ment such surveys by probing the older tidal debris, which may be crucial in constraining the shape of the Galaxy halo (e.g. Helmi 2004). For the BHB stars which reside in the putative stream, we find three of our BHB stars in the di- rection of the globular cluster Palomar (Pal) 5 appear to be in a foreground overdensity. More observations around these stars are required to establish any link to Pal 5 and/or the Sgr stream itself. Despite these controversies BHB stars have the distinct advantage of being more reliable standard candles than M stars and carbon stars and so the distance estimates are more precise. We emphasise observations of BHB stars have unlimited potential at providing accurate velocity and distance information on distant halo streams. The next generation multi-object spectrographs provide an excellent opportunity to accurately trace the full extent of such structures. ACKNOWLEDGEMENTS We would like to thank the referee for a number of help- ful comments. We thank David Law and David Mart´ınez- Delgado for providing us with their models. The Sgr coordinate system conversions made use of code at: http://www.astro.virginia.edu/∼ srm4n/Sgr/code.html. We thank Caroline Van Breukelen for useful comments. The au- thors acknowledge PPARC for financial support. This paper uses observations made on the Anglo Australian Telescope at Siding Springs using the 2dF instrument. We made use of the SDSS online database. 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 Na- tional Science Foundation, the U.S. Department of Energy, the Japanese Monbukagakusho, and the Max Planck Soci- ety. REFERENCES Alves D. R., 2004, NewA Rev., 48, 659 Arp H. C., 1955, AJ, 60, 317 Arp H. C., 1962, AJ, 135, 311 Bailey J. et al. 2003, http://www.aao.gov.au/2df/manual.html Beers T. C., Preston G.W, Shectman S.A., Doinidis S.P., Griffin K.E., 1992, AJ, 103, 267 Bellazzini M., Correnti M., Ferraro F.R., Monaco L., Mon- tegriffo P., astro-ph/0512109 Brown W. R., Geller M. J., Kenyon S. J.,Beers T. C., Kurtz M. J., Roll J. B., 2004, AJ, 127, 1555 Brown W. R., Geller M. J., Kenyon S. J., Kurtz M. J., Allende P. C., Beers T. C., Wilhelm R., 2005, AJ, 130, 1097 Cacciari C., Clementini G., LNP, 635, 105 lement C.M., 2001, AJ, 122, 2587 Clementini G., Gratton R., Bragaglia A., Carretta E., Di Fabrizio L., Maio M., 2003, ApJ, 125, 1309 Clewley L., Warren S.J., Hewett P.C., Norris J.E., Peterson R.C., Evans N.W., 2002, MNRAS, 337, 87 Clewley L., Warren S.J., Hewett P.C., Norris J.E., Evans N.W., 2004, MNRAS, 352, 285 Clewley L., Warren S.J., Hewett P.C., Norris J.E., Evans N.W., 2005, MNRAS, 362, 349 Cudworth K.M., 1979, AJ, 84, 1866 Dohm-Palmer R. C. et al. 2001, ApJ, 555, L37 Fukugita M. et al., 1996, AJ, 111, 1748 Gould A., Popowski P., 1998, ApJ, 508, 844 Harris W.E. 1996, AJ, 112, 1487 Helmi A., 2004, ApJ, 610, 97 Ibata R. A., Gilmore G. and Irwin M. J., 1994, Nature, 370, 6486, 194 Ibata R., Chapman S., Ferguson A.M.N., Irwin M., Lewis G., McConnachie A., 2004, MNRAS, 351, 117 Irwin M., Hatzidimitriou D., 1995, MNRAS, 277, 1354 Ivezi´c, Z et al., 2000, AJ, 120, 963 Johnston K.V., Law D.R., Majewski S.R., 2005, ApJ, 619, 800 Johnston K.V., Zhao H., Spergel D.N., Hernquist L., 1999, ApJ, 512, L109 Kinman T.D., Rosino L., 1962, PASP, 74, 499 Kinman T.D., Suntzeff N.B., Kraft R.P., 1994, AJ, 108, 1722 Koch A. et al., 2004, AJ, 128, 2274 Kundu A. et al., 2002, ApJ, 576, L125 Law D.R., Johnston K.V., Majewski S.R., 2005, ApJ, 619, 807 Lewis I. J., Cannon R. D., Taylor K. et al., 2002, MNRAS, 333, 279 Majewski S.R. et al., 2003, ApJ, 599, 1082 Majewski S.R. et al., 2004, AJ, 128, 245 Mart´ınez-Delgado D. and G´omez-Flechoso M. ´A., Aparicio A., Carrera R, 2004,ApJ, 601, 242 12 Clewley & Jarvis Monaco L., Bellazzini M., Ferraro F. R., Pancino E., 2003, ApJ, 597, 25 Newberg, H. J., et al. 2002, ApJ, 569, 245 Odenkirchen M., Grebel E. K., Dehnen W., Rix H.W., Cud- worth K. M.,2002, AJ, 124, 1497 Odenkirchen M. et al., 2003, AJ, 126, 2385 Peterson R.C, 1983, ApJ, 275, 737 Pier J.R, 1983, ApJS, 53, 791 Preston G.W., Shectman S.A., Beers T.C., 1991, ApJ, 375, 121 Schlegel D.J., Finkbeiner, D.P., Davis, M., 1998, ApJ, 500, 525 Sirko E., Goodman J., Knapp G.R., Brinkmann J., Ivezi´c, Z., Knerr E.J., Schlegel D., Schneider D.P., York D.G., 2004, AJ, 127, 899 Smith, J. A. et al. 2002, AJ, 123, 2121 Totten, E.J., Irwin, M.J., 1998, MNRAS, 294, 1 Totten, E.J., Irwin, M.J., Whitelock, P.A., 2000, MNRAS, 314, 630 Vivas A.K. et al., 2001, ApJ, 554, L33 Vivas A.K. et al., 2004, AJ, 127, 1158 Vivas A.K., Zinn R., Gallart C., 2005, AJ, 129, 189 Wu C., Qiu Y. L., Deng J. S., Hu J. Y., Zhao Y. H., 2005, AJ, 130, 1640 Yanny B. et al., 2000, AJ, 540, 825
astro-ph/0101202
1
0101
2001-01-12T13:05:04
Radio Plasma as a Cosmological Probe
[ "astro-ph" ]
Plasma containing relativistic particles appears in various forms in the inter galactic medium (IGM): As radio plasma released by active radio galaxies, as fossil radio plasma from former radio galaxies - so called `radio ghosts', as `cluster radio relics' in some clusters of galaxies, and as `cluster radio halos'. The impact of the different forms of radio plasma on the IGM and their use as diagnostic tools are briefly discussed.
astro-ph
astro-ph
RADIO PLASMA AS A COSMOLOGICAL PROBE TORSTEN A. ENssLIN Max-Planck-Institut fur Astrophysik, Karl-Schwarzschild-Str.1, 85740 Garching, Germany Plasma containing relativistic particles appears in various forms in the inter galac- tic medium (IGM): As radio plasma released by active radio galaxies, as fossil radio plasma from former radio galaxies -- so called radio ghosts -- , as cluster radio relics in some clusters of galaxies, and as cluster radio halos. The impact of the different forms of radio plasma on the IGM and their use as diagnostic tools are briefly discussed. 1 Radio Galaxies Radio galaxies produce radio cocoons, which create large cavities1 in the IGM gas filled with magnetic fields and relativistic particles: electrons and positrons or protons. The electrons (and positrons if present) of several GeV reveal their presence by synchrotron emission at radio wavelength. From the observed radio emission the minimal energy density in the radio plasma can be deduced. Since the radio plasma has to have a higher (or equal) pressure than the environment, a direct probe of the ambient gas pressure is given2. Also the ram pressure effects of swept up IGM material during the expansion of the cocoon can be used to determine IGM gas densities3,4. The morphology of the radio galaxy can give information about relative motions of the galaxy and the IGM5, but also allows to detect IGM shock waves6. The amount of energy released by radio galaxies into the gaseous Universe during the whole history of the Universe is large. Rough estimates give a ratio of 0.1 ... 1 between the energy released by radio galaxies and by the gravitationally driven structure formation7,8. Radio galaxies are therefore one of the best candidates in order to explain the necessary non-gravitational heating required by the observed entropy-floor of clusters and groups of galaxies9,10. Therefore remnant radio plasma should be a ubiquitous phase of the IGM. 2 Radio Ghosts Radiative energy losses let the radio emitting electrons in radio cocoons become invisible to our instruments within cosmologically short times (108 year). After- wards, the cocoon of fossil radio plasma is named radio ghosts11. But the electrons confined by the magnetic fields should still stay relativistic, and might reveal their presence by inverse Compton effects on the cosmic microwave background (CMB): the relativistic Sunyaev-Zeldovich (rSZ) effect. Since a CMB photon scattered by a relativistic electron gains a large factor in energy, it is practically removed from the CMB spectrum. Thus the rSZ effect is mostly an absorption at CMB frequen- cies. The optical depth of the total electron population residing in radio ghosts is of the order of τ ∼ 10−7, but this number has large uncertainties12. Since the expansion of the radio cocoons into the IGM should at least release some pressure ensslin: submitted to World Scientific on December 17, 2018 1 work to the environmental gas, the corresponding thermal SZ effect should have a Comptonization parameter of the order13,12 of y ∼ 10−5...−6. But not only the relativistic electrons may reveal their presence by scattering of radiation. Also the magnetic fields of radio ghosts are able to deflect charged particles even at the highest energies observed in the cosmic ray spectrum. If the spatial distribution of radio ghosts is sufficiently unclustered compared to the clus- tering of galaxies, then the expected number densities of ghosts would be sufficient to explain the mysterious isotropy of the ultra high energy cosmic ray particles, even if the sources are as inhomogeneously distributed as the galaxies14. 3 Cluster Radio Relics Recently, radio ghosts were detected by X-ray deficits in cluster core regions1. But very likely, they already showed up as cluster radio relics 20 years ago. These extended radio sources in some clusters of galaxies are not specially connected to any optical galaxy15. Cluster radio relics are preferentially located in peripheral regions of clusters. They have usually steep radio spectra and exhibit often a high degree of linear polarization. In several cases, their locations can be associated with strong shock waves in merging clusters of galaxies16. In the cases of Abell 2256 and Abell 1367 temperature substructures of the hot ICM gas could be detected17,18, which support the presence of a shock wave at the location of cluster relics in these clusters. For Abell 75419,20, Abell 225621, Abell 366722 and also the Coma23 cluster numerical simulations of merger events were satisfactorily fitted to the X-ray data, which also supports the shock wave-relic connection. The mechanism producing the cluster radio relics is very likely adiabatic compression of radio ghosts in an environmental shock wave24. This strengthens the internal magnetic fields and shifts the electron population to higher energies. If the upper cooling cutoff of the electron spectrum can be shifted above the radio emitting energies, the ghost's radio emission is revived and the ghost appears as a cluster radio relic. Cluster radio relics therefore not only allow the study of shock waves, but also give a view into the fossil Universe. Their number should be much higher at lower frequencies, due to the distribution of frequency cutoffs in their radio spectra. 4 Cluster Radio Halos Cluster radio halos appear preferentially in the center of clusters of merging galaxies15,25. They have morphologies similar to the X-ray morphologies of their host clusters26. Radio polarization could not be reported in any case. Their large physical size (∼ Mpc) require that the radio emitting electrons are accelerated or injected throughout the cluster volume. The energy source of all non-thermal processes in clusters should be either the kinetic energy of matter falling onto clusters, or the outflows from galaxies. The latter can be divided in galactic winds, which are strongest for starburst galaxies, and ejection of radio plasma from an AGN. All these processes can produce shock waves and inject turbulence into the ICM, and therefore produce conditions where Fermi mechanisms accelerate particles. For a brief review see Ensslin (1999)27. ensslin: submitted to World Scientific on December 17, 2018 2 A promising injection mechanism of relativistic electrons is secondary particle production from hadronic interactions of relativistic protons with the background gas28,29,30,31: p + p → 2 N + π± π± → µ± + νµ/¯νµ → e± + νe/¯νe + νµ + ¯νµ The lifetime of relativistic protons in the ICM is of the order of the Hubble time, or larger32,33. Thus they are able to travel large distances from their sources before they release their energy. The production of electrons via charged pions has to be accompanied by gamma ray production via neutral pions 29,33,34,31: p + p → 2 N + πo πo → 2 γ Thus clusters with radio halos might have gamma-ray halos, which would be, if detected, a direct proof for a hadronic origin of radio halos. 5 Conclusions Radio galaxies: • probe the thermodynamical state of the gaseous Universe. • produce large amounts of fossil radio plasma. Radio ghosts, the invisible descendents of radio galaxies: • may produce an relativistic Sunyaev-Zeldovich effect. • scatter and possibly isotropize ultra high energy cosmic ray particles. Cluster radio relics: • trace shock waves of the large scale structure formation flows. • are likely revived radio ghosts. Cluster radio halos: • indicate recent cluster merger events. • trace non-thermal processes in the intracluster medium. References 1. A. C. Fabian, J. S. Sanders, S. Ettori, G. B. Taylor, S. W. Allen, C. S. Crawford, K. Iwasawa, R. M. Johnstone, and P. M. Ogle. MNRAS, 318, L65, 2000. 2. L. Feretti, G. C. Perola, and R. Fanti. A&A, 265, 9, 1992. 3. K.-H. Mack, , U. Klein, C. P. O'Dea, A. G. Willis, and L. Saripalli, A&A, 329, 431, 1998. ensslin: submitted to World Scientific on December 17, 2018 3 4. A. P. Schoenmakers, K.-H. Mack, A. G. de Bruyn, H. J. A. Rottgering,U. Klein, and H. van der Laan, A&A Supp., 146, 293, 2000. 5. J. O. Burns. Science, 280, 400, 1998. 6. T. A. Ensslin, P. Simon, P. L. Biermann, U. Klein, S. Kohle, P. P. Kronberg, and K.-H. Mack. ApJ Lett., in press, astro-ph/0012404 7. T. A. Ensslin, Y. Wang, B. B. Nath, and P. L. Biermann, A&A, 333, L47, 1998. 8. T. A. Ensslin, Cluster Mergers and their Connection to Radio Sources, 24th meeting of the IAU, Joint Discussion 10, Manchester, England., 10, E9, 2000. astro-ph/0011052 9. T. J. Ponman, D. B. Cannon, and J. F. Navarro. Nature, 397, 135, 1999. 10. K. K. S. Wu, A. C. Fabian, and P. E. J. Nulsen, MNRAS, 318, 889, 2000. 11. T. A. Ensslin. In Ringberg Workshop on 'Diffuse Thermal and Relativistic Plasma in Galaxy Clusters' eds.H. Bohringer, L. Feretti, P. Schuecker, MPE Report, 271, 275, 1999., astro-ph/9906212 12. T. A. Ensslin and C. R. Kaiser. A&A, 360, 417, 2000. 13. M. Yamada, N. Sugiyama, and J. Silk, ApJ 522, 66, 1999. 14. G. Medina-Tanco and T. A. Ensslin. Astroparticle Physics, in press, 2000. 15. L. Feretti and G. Giovannini. In IAU Symp. 175, Extragalactic Radio Sources, 333, 1996. 16. T. A. Ensslin, P. L. Biermann, U. Klein, and S. Kohle. A&A, 332, 395, 1998. 17. U. G. Briel and J. P. Henry. Nature, 372, 439, 1994. 18. R. H. Donnelly, M. Markevitch, W. Forman, C. Jones, L. P. David, E. Chura- zov, and M. Gilfanov. ApJ, 500, 138, 1998. 19. K. Roettiger, J. M. Stone, and R. F. Mushotzky. ApJ, 493, 62, 1998. 20. N. Kassim, T. E. Clarke, T. A. Ensslin, A. S. Cohen, and D. Neumann. ApJ Lett., submitted, 2001. 21. K. Roettiger, J. O. Burns, and J. Pinkney. ApJ, 453, 634, 1995. 22. K. Roettiger, J. O. Burns, and J. M. Stone. ApJ, 518, 603, 1999. 23. J. O. Burns, K. Roettiger, M. Ledlow, and A. Klypin. ApJ Lett., 427, L87, 1994. 24. T. A. Ensslin and Gopal-Krishna. A&A, in press, 2001. 25. G. Giovannini, M. Tordi, and L. Feretti. New Astronomy, 4, 141, 1999. 26. F. Govoni, T. A. Ensslin, L. Feretti, and G. Giovannini. A&A, submitted, 2000. 27. T. A. Ensslin. In IAU Symp. 199, 'The Universe at Low Radio Frequencies', 1999. astro-ph/0001433. 28. B. Dennison. ApJ Lett., 239, L93, 1980. 29. W. T. Vestrand. AJ, 87, 1266, 1982. 30. P. Blasi and S. Colafrancesco. Astroparticle Physics, 12, 169, 1999. 31. K. Dolag and T. A. Ensslin. A&A, 362, 151, 2000. 32. T. A. Ensslin, P. L. Biermann, P. P. Kronberg, and X.-P. Wu. ApJ, 477, 560, 1997. 33. V. S. Berezinsky, P. Blasi, and V. S. Ptuskin. ApJ, 487, 529, 1997. 34. S. Colafrancesco and P. Blasi. Astroparticle Physics, 9, 227, 1998. ensslin: submitted to World Scientific on December 17, 2018 4
astro-ph/0407576
1
0407
2004-07-28T13:13:56
A dynamical model for the CSO-MSO-FRII evolution: hints from hot spot properties
[ "astro-ph" ]
Compact Symmetric Objects (CSOs) are considered the young counterparts of large doubles according to advance speeds measured or inferred from spectral ageing. Here we present a simple power law model for the CSO/FRII evolution based on the study of sources with well defined hot-spots. The luminosity of the hot spots is estimated under minimum energy conditions. The advance of the source is considered to proceed in ram pressure equilibrium with the ambient medium. Finally, we also assume that the jets feeding the hot spots are relativistic and have a time dependent power. Comparison with observational data allows to interpret the CSO-FRII evolution in terms of decreasing jet power with time.
astro-ph
astro-ph
A dynamical model for the CSO-MSO-FRII evolution: hints from hot spot properties M. Perucho 1 and J. M.a Mart´ı 1 1 Departamento de Astronom´ıa y Astrof´ısica, Universidad de Valencia. Dr. Moliner 50, 46100 Burjassot (Valencia), Spain e-mail: [email protected], [email protected] Abstract Compact Symmetric Objects (CSOs) are considered the young counterparts of large doubles according to advance speeds measured or inferred from spectral ageing. Here we present a simple power law model for the CSO/FRII evolution based on the study of sources with well defined hot-spots. The luminosity of the hot spots is estimated under minimum energy conditions. The advance of the source is considered to proceed in ram pressure equilibrium with the ambient medium. Finally, we also assume that the jets feeding the hot spots are rela- tivistic and have a time dependent power. Comparison with observational data allows to interpret the CSO-FRII evolution in terms of decreasing jet power with time. Keywords: galaxies:active-galaxies:jets-galaxies:ISM-radio continuum:galaxies 1 Introduction Compact Symmetric Objects (CSOs) are thought to be early stages of powerful ex- tended radio sources as first suggested by Phillips & Mutel (1982) and later worked out by Readhead et al. (1996). This view has been underpined by recent measurements of hot spot advance speeds (Owsianik & Conway 1998, Owsianik et al. 1998, Taylor et al. 2000, Tschager et al. 2000, Polatidis et al. 2003) and spectral ageing measures (Murgia et al. 2003). Current evolutionary models (see Fanti & Fanti 2002) relate luminosity and expan- sion velocity of a source to jet power and external gas density. The energy accumulated in the lobes drives the source expansion. Ram pressure equilibrium with the ambient medium is assumed. The volume of the source is usually inferred from self-similarity arguments whereas radio power is computed from equipartition assumptions. In all the models, jet power was considered as constant. 1 Here we discuss a model, presented in Perucho & Mart´ı (2002a, Paper II), for the long term evolution of powerful radio sources, which is an extension to time dependent jet power of the model introduced in Perucho & Mart´ı (2002b, Paper I) for CSOs. In order to avoid conjectures about the volume growth of the source (e.g., self-similarity), we concentrate in the study of the hot spots for which properties like size or luminosity can be derived reliably. In our model we assume that the advance work of the hot- spot is directly connected to the jet power. The remaining assumptions of our model are standard. The luminosity of the hot spots is estimated under minimum energy conditions. The advance of the hot spot proceeds in ram pressure equilibrium with the ambient medium. Finally, we also assume that the jets feeding the hot spots are relativistic. In section 2, we give the main equations derived from the model. In sectio 3 we present the observational data we compared with our model, and section 4 is devoted to discussion and conclusions drawn from this work. 2 A dynamical model for the evolution of hot spots in powerful radio sources Our model relies on three basic parameters. The first (β) is the exponent for the growth of linear size of a hot spot with time, rhs ∝ tβ as it propagates through an external medium with density varying with linear size (LS) as ρext ∝ (LS)−δ, where δ is the second parameter. Using the hot-spot advance speed, we can relate linear size with time and describe the evolution of physical parameters in terms of distance to the source of the jets feeding them. Considering that hot-spots advance with non-relativistic speeds, ram pressure equilibrium leads to, vhs = s Fj Aj,hsρext . (1) where Fj is the jet thrust and Aj,hs, the cross-sectional area of the jet at the hot spot, assumed to be ∝ r2 hs. The final step is to consider that for a relativistic jet, jet thrust and power, Qj, are simply related by Fj ≈ Qj/c. If we now allow for a dependence of the jet power with time like Qj ∝ tε (where ε is the third parameter), combine all the dependencies and integrate, we get vhs ∝ (LS)δ/2tε/2−β → t ∝ (LS)(1−δ/2)/(ε/2+1−β). Substituting now in the expressions for the hot spot radius and speed, we obtain rhs ∝ (LS)β(1−δ/2)/(ε/2+1−β), vhs ∝ (LS)(δ/2+ε/2−β)/(ε/2+1−β). (2) (3) The next equation in our model comes from the source energy balance. We assume (P dV )hs,adv adjusts that the power consumed by the source in the hot spot advance, 2 Figure 1: Contours of β as function of sv and sr for the model discussed in the text. Boxes bound the expected values of β for CSO-MSO evolution (solid lines) and MSO- FRII evolution (dashed lines). to the evolution of the jet kinetic power, i.e., (P dV )hs,adv(∝ Phsr2 hsvhs) ∝ Qj ∝ tε. (4) Finally, under the assumption of minimum energy, the luminosity of the hot-spot (Lhs ∝ P 7/4 hs r3 hs) may be expressed in terms of LS Lhs ∝ (LS){7/4[−δ/2(ε−2β)+ε/2−δ/2−β]+3β(1−δ/2)}/(ε/2−β+1) . (5) Inverting eqs. (3) and (5), we derive expressions for the evolution parameters of our model in terms of exponents for the evolution of observable quantities. β = sr 1 − sv , δ = 12 7 sr − 4 7 sL + sv, ε = 2/7sr + 4/7sL + sv 1 − sv , (6) where sr, sv and sL stand for the values of the exponents of the observed hot spot radius, advance velocity and luminosity as functions of LS. Figure 1 shows the variation of β and Figure 2 that of δ and ε with the different exponents. Let us note that the previous expressions not only provide a system of algebraic equations to obtain values for the theoretical parameters in our model. They also prove that our model is self- consistent since the variations of the theoretical parameters induced by those derived from observations agree with physical expectations. Therefore, if we are able to obtain values for the exponents in eqs. (3) and (5) from observational data assuming that the corresponding LS − Lhs and LS − rhs plots track the evolution of individual sources, we can derive expected values for β, δ and ε from the observational fits. 3 Observables from hot spots In order to apply our model to the evolution of powerful radio sources from the CSO to the FRII phases, we have compiled a sample of sources with linear sizes between 3 Figure 2: Contours of δ and ǫ as function of sr and sL for the model discussed in the text and different slopes for the hot spot expansion (continuous contours: sv = 0; dotted contours: sv = −0.5; dashed: sv = −0.2. Boxes bound the expected values of δ and ǫ for CSO-MSO evolution (dotted lines) and MSO-FRII evolution (dashed lines). tens of pcs to hundreds of kpcs and well defined hot spots. The sample of CSO is the same as the one used in Paper I. Sources were selected from the GPS samples of Stanghellini et al. (1997), Snellen et al. (1998, 2000) and Peck & Taylor (2000). We have chosen the sources with double morphology already classified in the literature as CSOs and also those whose components can be safely interpreted as hot spots even though the central core has not been identified yet. The criteria we have followed are quite similar to those used by Peck & Taylor (2000), although see Paper I for details. Seven Medium size (1 - 10 kpc) symmetric objects ("doubles") have been taken from the CSS-3CR sample of Fanti et al. (1985). Finally, 40 sources from the sample of FRII-3CR radio galaxies of Hardcastle et al. (1998) have been considered. See Paper II for further details. All the subsamples in our combined sample have similar flux density cut-offs: 1 Jy at 5 GHz for the CSOs and 10 Jy at 178 MHz for the MSOs and FRIIs. The differences in redshift among the three subsamples (z ≤ 1 for the CSOs; 0.3 < z < 1.6 for the MSOs; z < 0.3 for the FRIIs) enhances the luminosity drop between MSOs and FRIIs. Figure 3 displays the hot spot radius and luminosity with respect to projected linear size on logarithmic scales, together with the best linear fit for both CSO-MSO and MSO-FRII subsamples. Hot-spot linear sizes and luminosities have been calculated as explained in the Appendix of Paper I. Table 1 compiles the slopes characterizing the power law fits and their errors, as well as the correlation coefficients. 4 Discussion Assuming an evolutionary interpretation of the plots in Fig. 3, the most remarkable aspects of those plots are the self-similar growth of hot-spot radius with linear size for the first ten kpc of evolution and flattening for large sources, in agreement with 4 Figure 3: Log-log plot for the hot spot radius and luminosity versus projected linear size for the sources in the combined sample. Continuous lines correspond to the best linear fits for both CSO-MSO and MSO-FRII subsamples. rhs sr r Lhs sL r Model β δ ǫ CSO-MSO 1.0 ± 0.3 0.93 0.24 ± 0.14 0.26 MSO-FRII 0.40 ± 0.11 0.51 −1.32 ± 0.15 −0.69 sv = 0 1.0 ± 0.3 1.6 ± 0.6 0.4 ± 0.2 sv = −0.5 sv = 0 0.7 ± 0.2 0.4 ± 0.2 1.1 ± 0.7 −0.05 ± 0.15 1.4 ± 0.5 −0.6 ± 0.2 sv = −0.2 0.3 ± 0.2 1.2 ± 0.5 −0.75 ± 0.15 Table 1: Best fits for radius (sr) and luminosity (sL), along with their errors and correlation coef- ficients (r). In the right part of the table we write the values of the evolution parameters (β, δ and ǫ) which result from the calculated best fits and two different possible slopes for advance speed (sv), along with their errors, which are directly taken from Figs. 1,2 for β, δ and ε. Jeyakumar & Saikia (2000), and the change of sign of the slope for radio-luminosity also at ten kpc. This fact is consistent with the transition between the ISM and the IGM, also suggested by the disappearance of IR aligned emission after the CSS stage (de Vries et al. 2003), which could imply significant changes in the evolution. Fits can be used to constrain the parameters β, δ, ε of the model. These values appear listed in Table 1. Hot spots undergo a secular deceleration (from 0.2c in the first kpc to a value ≈ ten times smaller in FRIIs) and there are no observational indications of any acceleration, hence constant and decreasing hot spots velocities have been considered for the CSO-MSO and MSO-FRII fits. The break in the fits shown in Fig. 3 at 1-10 kpc produce very different values for the parameters in the CSO-MSO and MSO-FRII phases. On the contrary, there is a complete consistency between the fits for the CSOs alone (see Paper I) and those for the CSO-MSO phase. The hot spot expansion rate, β, decreases from ≈ 1 in the CSO- MSO phase to ≈ 0.4 in the MSO-FRII phase. Jet power increases with time during the first phase (ε ≈ 0, 0.4 depending on the value chosen for sv) and decreases in the 5 long term (ε ≈ −0.6, −0.7). It would be interesting to relate the time evolution of the jet power with the physical processes responsible for the jet production (i.e., accretion rate, black hole spin). The density profile is flat (δ < 2), consistent for the CSO-MSO stage with that derived by Pihlstroem et al. (2003) (δ ≃ 1.3) for GPS-CSS sources from HI detections, and the transition between the two phases is smooth, although this can be a result of the fitting process that washes out any steep gradient between a flat (δ ≈ 0), small core and the intergalactic medium. We also note that the density gradient depends strongly on sv and that this parameter is poorly known. A value of δ = 2 in the CSO-MSO phase will produce accelerating hot spots (sv = 0.5) and a large increase of the jet power (ε = 2). Finally, fixing sL and sr and taking δ = 0, we get sv = −1.5, which is too small to be maintained over a long distance: starting with a hot spot speed of 0.2c at 50 pcs, the hot spot speed at 0.5 kpc would have decreased up to 6 10−3c, much smaller than the present accepted values for CSO advance speeds (Polatidis et al. 2003, Murgia et al. 2003). It would be interesting to have upper limits of hot spots advance speeds in CSOs in order to constrain the density profile and the jet power evolution. Regarding the problem of trapped sources, a suitable configuration of external medium density and jet power evolution may lead to a number of sources which, along with core-jets, may contribute to the excess of small sources in count statistics (Marecki et al. 2003, Drake et al. 2003). Acknowledgments We thank Prof. R.Fanti for his interest in our work. This research was supported by Spanish Ministerio de Ciencia y Tecnolog´ıa (grant AYA2001-3490-C02-01). M.P. thanks the University of Valencia for his fellowship within the V SEGLES program and also thanks LOC and SOC of the 3rd GPS-CSS Workshop for their kindness and hospitality. References Drake, C., et al., 2003, PASA, in press Fanti, C., Fanti, R., Parma, P., Schilizzi, R.T., & van Breugel, W.J.M. 1985, A&A, 143, 292 Fanti, C., & Fanti, R. 2002, in Issues in Unification of AGNs, eds. Maiolino, R., Marconi, A., Nagar, N. (ASP Conference Series) Hardcastle, M.J., Alexander, P., Pooley, G.G., & Riley, J.M. 1998, MNRAS, 296, 445 Jeyakumar, S., & Saikia, D.J. 2000, MNRAS, 311, 397 Marecki, A., et al., 2003, PASA, in press Murgia, M., et al., 2003, PASA, in press Owsianik, I., & Conway, J.E. 1998, A&A, 337, 69 Owsianik, I., Conway, J.E., & Polatidis, A.G. 1998, A&AL, 336, 37 Peck, A.B., & Taylor,G.B. 2000, ApJ, 534, 90 Perucho, M., & Mart´ı, J.M. 2002b, ApJ, 568, 639 (Paper I) Perucho, M., & Mart´ı, J.M. 2002a, in preparation (Paper II) Phillips, R.B., & Mutel, R.L. 1982, A&A, 106, 21 Pihlstroem, Y., et al., 2003, PASA, in press Polatidis, A.G., et al., 2003, PASA, in press 6 Readhead, A.C.S., Taylor, G.B., Xu, W., Pearson, T.J., Wilkinson, P.N., & Polatidis, A.G. 1996, ApJ, 460, 612 Snellen, I.A.G., Schilizzi, R.T., de Bruyn, A.G., Miley, G.K., Rengelink, R.B., Rotgering, H.J.A., Bremer, M.N. 1998, A&AS, 131, 435 Snellen, I.A.G., Schilizzi, R.T., & van Langevelde, H.J. 2000, MNRAS, 319, 429 Stanghellini, C., O'Dea, C.P., Baum, S.A., Dallacasa, D., Fanti, R., Fanti, C. 1997, A&A, 325, 943 Taylor, G.B., Marr, J.M., Pearson, T.J., & Readhead, A.C.S. 2000, ApJ, 541, 112 Tschager, W., Schilizzi, R.T., Rotgering, H.J.A., Snellen, I.A.G., Miley, G.K. 2000, A&A, 360, 887 de Vries, W., et al., 2003, PASA, in press 7
astro-ph/0002391
2
0002
2000-03-24T20:15:57
A New Dark Matter Model for Galaxies
[ "astro-ph" ]
In this paper a new theory of Dark Matter is proposed. Experimental analysis of several Galaxies show how the non-gravitational contribution to galactic Velocity Rotation Curves can be interpreted as that due to the Cosmological Constant $\Lambda$. The experimentally determined values for $\Lambda$ are found to be consistent with those expected from Cosmological Constraints. The Cosmological Constant is interpreted as leading to a constant energy density which in turn can be used to partly address the energy deficit problem (Dark Energy) of the Universe. The work presented here leads to the conclusion that the Cosmological Constant is negative and that the universe is de-accelerating. This is in clear contradiction to the Type Ia Supernovae results which support an accelerating universe.
astro-ph
astro-ph
RHCPP00-1T astro-ph/0002391 MARCH 2000 A New Dark Matter Model for Galaxies George V. Kraniotis ♠ 1and Steven B. Whitehouse♠ 2 ♠ Centre for Particle Physics, Royal Holloway College, University of London, Egham, Surrey, TW20-0EX, U.K ABSTRACT In this paper a new theory of Dark Matter is proposed. Experimental analysis of several Galaxies show how the non-gravitational contribution to galactic Velocity Rotation Curves can be interpreted as that due to the Cosmological Constant Λ. The experimentally determined values for Λ are found to be consistent with those expected from Cosmological Constraints. The Cosmological Constant is interpreted as leading to a constant energy density which in turn can be used to partly address the energy deficit problem (Dark Energy) of the Universe. The work presented here leads to the conclusion that the Cosmological Constant is negative and that the universe is de-accelerating. This is in clear contradiction to the Type Ia Supernovae results which support an accelerating universe. 0 0 0 2 r a M 4 2 2 v 1 9 3 2 0 0 0 / h p - o r t s a : v i X r a [email protected] 2 [email protected] 1 Dark Matter The problem of missing or Dark Matter, namely that there is insufficient material in the form of stars to hold galaxies and clusters together, has been known since the pioneering work of Bessel, Zwicky and most recently Rubin [1]. The existence of non-luminous Dark Matter was first inferred in 1984 by Fredrich Bessel from gravitational effects on positional measurements of Sirius and Procyon. In 1933, Zwicky concluded that the velocity dispersion in Rich Clusters of galaxies required 10 to 100 times more mass to keep them bound than could be accounted for by luminous galaxies themselves. Finally, Trimble [2] noted that the majority of galactic rotation curves, at large radii, remain flat or even rise well outside the radius of the luminous astronomical object. The missing Dark Matter has been traditionally explained in terms of Dark Matter Halo's [3], although none of the Dark Matter Halo models have been very successful in explaining the experimental data [4, 5]. This paper will describe the missing matter (Dark Energy) in terms of a Cosmo- logical Constant which leads to a constant energy density. The experimental determination of galactic velocity rotation curves (VRC) has been one of the most important approaches used to estimate the "local" mass (energy) density of the Universe. Several sets of data from VRC's will be analysed and the contribution due to the Cosmogical Constant determined. 2 Constaints on the value of the Cosmological Con- stant It is interesting to estimate the allowed range of values for the Cosmological Con- stant within the constraints of General Relativity and observational astronomy, (for a comprehensive review, see Bahcall et.al. [6]). Starting from a General Relativity point of view, the Friedman energy equation is given by: 1 = 8πGN 3 ρmatter H 2 − (1) kc2 R2H 2 + c2Λ 3H 2 , where the Hubble Constant is denoted by H, the curvature term by k and GN denotes 1 the Newton gravitational constant. Eq.(1) can be rewritten as 1 = Ωm + Ωk + ΩΛ (2) Here the relative contributions to the energy density of the universe are given by, the mass, curvature and Cosmological Constant. If we assume that the curvature contribution is small: 1 = ΩM atter + ΩΛ (3) In order to satisfy equation (3), it was surprising to discover that only a narrow range of values for the observed Cosmological Parameters were allowed. A "reasonable" set of parameters consistent with observation are: HO = 100Kms−1Mpc−1, ρmatter = 5 × 10−30gcm−3, ΩΛ Ωmatter = 4.3 (4) and ΩM atter + ΩΛ = 1.4 (here we assume a small value for the curvature ∼ 0.4. (For an authoritative review of the matter/energy sources of the universe, see Turner [7]). It was found that observational constraints placed upon the range of values for the cosmological parameters lead to a surprisingly narrow range of possible values for the Cosmological Constant, the range being given by: 10−56 < Λ < 5 × 10−55cm−2. (5) 3 Experimental Results It was shown [10], within the Weak Field Approximation, that the Cosmological Constant at large radii could be determined from galactic velocity rotation curves. This contribution is given by: v2 Λ(r) = v2 obs(r) − v2 mass(r), leading to, v2 Λ/r = c2Λr 3 , at large r 2 (6) (7) Galaxy NGC 2403 NGC 4258 NGC 5033 NGC 5055 NGC 2903 NGC 3198 Cosmological Constant Radius 20Kpc ΛN GC2403 = 3.6 × 10−55cm−2 50 Kpc ΛN GC4258 = 5.5 × 10−55cm−2 40 Kpc ΛN GC5033 = 1.0 × 10−55cm−2 50 Kpc ΛN GC5055 = 1.4 × 10−55cm−2 30 Kpc ΛN GC2903 = 3.8 × 10−55cm−2 50 Kpc ΛN GC3198 = 5.0 × 10−56cm−2 Table 1: Absolute values of the Cosmological Constant are shown above. The results obtained by this analysis are shown in Table 1. The experimental values obtained for the Cosmological Constant fall within the range determined from General Relativity and observational constraints. While the initial results are promising, a thorough and systematic analysis of galactic rotation curves needs to be undertaken in order to confirm the trend. Previous results [10] reported for the value of the Cosmological Constant were 100 to 1000 times the "allowed value". This systematic error arose for two main reasons: the first by not taking the gradient of the curves at sufficiently large radii and the second by the lack of access to "real" experimental data leading to crude data analysis. The results presented in this paper suffer from the second problem, i.e. all the gra- dients were obtained from the data in the published literature and not from raw experimental data i.e. M33 Corbelli & Salucci [4], NGC 3198 [8] and all others from [12]. However experience has taught us that a cursory look at rotation curves will deter- mine which galaxies are candidates for explanation by a Cosmological Constant and which are not. Galaxies where the velocity rotation curve remains flat or rises at large radii, are immediate candidates. NGC 3198 is a good example, whereas others such as M33 [4] has clearly not relaxed to the Cosmological background, even at many times the galactic radii. A full explanation for M33 has to be sought in a different direction. Finally, a simple calculation of the effective mass density due to the Cosmological Constant in NGC 3198, ρef f = − c2Λ 4πGN (8) leads to a value of 5.4 × 10−29gcm−3 which is comparable to the HI mass density [13] at the outer disk of NGC 3198 galaxy. This is further confirmation that the 3 Cosmological Constant effect can be seen at galactic scale lengths. 4 Accelerating or Decelerating Universe? Recently there has been great interest in the Type Ia Supernovae results of Perlmut- ter et al [11] which suggest that the universe is accelerating. In this section we will show that the Weak Field Approximation coupled with galactic velocity rotation curve data inevitably lead to a negative Cosmological Constant. The equation for the VRC is given [10] by 3 − v2 r = − Gm r2 + c2Λ 3 r (9) We note that Eq.(9) is only strictly true for small and large radii, however it will serve to illustrate our arguments. Using the Newtonian limit of Einstein field equations we derived equation (9). It is important to realize that the Cosmological Constant obeys the equation of state given by, PΛ = −c2ρΛ, (10) where PΛ is the pressure term due to Λ. Taking the Newtonial limit in the absence of matter, Tµν = 0, the differential equation for the static Newtonian potential becomes leading to, ∇2Φ = −c2Λ ρef f = ρΛ + 3PΛ c2 = −2ρΛ (11) (12) If we arbitrary set Φ = 0 at the origin, then in spherical coordinates (11) has the solution Φ = − c2Λ 6 r2 [14]. Thus, the Cosmological Constant leads to the following correction to the Newtonial potential Φ = − Gm r − c2Λ 6 r2 (13) At small galactic radii the velocity versus radius contribution is well known and fol- lows Newtonian physics. For large radii a negative Cosmological Constant gives a positive contribution to the VRC which is what is actually observed. On the other hand the effect of a positive Cosmological Constant would be to lower the rotation 3In Ref.[10] eq.(15), there was a typographical sign error for one of the terms and also the negative pressure effect associated with Λ was not fully appreciated. 4 curve below that due to matter alone. The above simple argument, based on observational astronomy, allows only a nega- tive Cosmological Constant as a possible explanation for the galactic velocity rotation curve data. This is in clear disagreement with the Type Ia supernovae results [11]. However, given the uncertainties in the determination of the deceleration parameter, q0, derived from supernovae data [11] the approach outlined above has certain merits worth consideration. In summary these are , the Cosmological Constant is determined from direct mea- surement unlike the Supernovae results, the experimentally determined value is the correct order of magnitude as that required from cosmological constraints, and finally a negative Cosmological Constant is consistent, and indeed a natural physical expla- nation , for the observed galactic velocity rotation curve data. Finally, observations of global clusters of stars constrain the age of the universe and consequently place an observational limit on a negative Cosmological Constant [16] of , Λ ≤ 2.2 × 10−56cm−2. (14) Note, the Cosmological Constant derived from global cluster constraints is in agree- ment with the experimentally determined value derived from galactic velocity rotation curve data. 4.1 Experimental Tests-Dark Matter Halo vs Cosmological Constant It would be of some interest if it was possible to experimentally distinguish between the contribution of Dark Matter Halo's and Dark Energy (Cosmological Constant) to galactic rotation curves. We know that Dark Matter predicts a variation of mass at large radii given by [17], MDM (r) ∝ r while for Dark Energy due to a Cosmological Constant, MΛ(r) ∝ r3[ρΛ + (3PΛ/c2)]. (15) (16) With these different types of predictive variations it should be possible to design experimental tests to distinguish between the two phenomena. 5 5 Quark Hadron Phase transition In this section which is of more speculative nature, working within the Extended Large Number Hypothesis, and using the experimentally determined Cosmological Constant, we will demonstrate how the energy density for the Quark - Hadron can be estimated. However, it is useful to put into context the significance of the Cosmological Con- stant for many seemingly disparate branches of Physics. The figure 1 below shows the Cosmological Constant at the epicentre of Physics. The diagram demonstrates a dichotomy whereby several branches of Physics need a non-zero Cosmological Constant in order to explain key physical phenomena, whilst in others a non-zero value presents a fundamental problem. General Relativity Λ=0 Λ Cosmology Inflation Λ=0 Supernova Accelerating Universe Λ=0 Standard Model Λ=0 Astronomy Missing Matter Λ=0 Particle Physics Quark-Hadron Λ=0 String Theory Λ=0 Quantum Field Theory Λ=0 Figure 1: Λ at the epicentre of Physics It is also noted here that while fundamental theories of Particle Physics such as the Standard Model, Quantum Field Theory and String Theory have many major predictive successes they all have problem with a high vacuum energy density. On the other hand while the Extended LNH is formulated from a naive theory [9] it appears to correctly predict the correct vacuum energy density and other cosmological parameters. The Extended LNH relates the value of the Cosmological Constant to 6 the effective mass given by: Λ = G2 N m6 h4 ef f = c6L4 s h6 m6 ef f (17) Matthews [18] pointed out that when using the Extended LNH to determine today's cosmological parameters, the mass of the proton originally suggested by Dirac should be replaced by the energy density of the last phase transition of the Universe : Quark - Hadron. Note that in equation (17) there are no free parameters, Ls is normalised to the gravitational constant and corresponds to the fundamental length of String Theory. Using equation (17) and the Cosmological Constant determined from NGC 5033, the effective mass is given by mEf f ective = 332MeV (18) We will associate this value with the Quark - Hadron phase transition energy. (Other experimentally determined Cosmological Constant data give mQH in the range 295 - 410 MeV). The experimentally determined value within the LNH predicts the correct order of magnitude for the phase transition. The above result poses the question that it might be possible to gain insights on the quantum mechanical origin of the Universe, as Dirac [19, 20, 21] suggested, from direct observation of the present day Universe. Finally, does the Cosmological Constant provide the key to the integration of the various Physics disciplines as Figure 1 suggests? 6 Discussion Analysis of the galactic rotation curves show that the missing Galactic Dark Matter can be explained in terms of a Cosmological Constant. This contribution can be considered a prime candidate for the "Dark Energy" which is smoothly distributed throughout space, and contributes approximately 70% to the mass/energy of the Universe [7]. However, in order to support this thesis for the Cosmological Constant, thorough and systematic analysis of galactic velocity rotation data needs to take place. 7 It was shown how, within the Weak Field Approximation, that VRC data inevitably lead to a negative value for the Cosmological Constant in direct disagreement with the type Ia Supernovae data. Nevertheless, given the uncertainties in determining the deceleration parameter q0 [15], from the redshift-magnitude Hubble diagram us- ing Type Ia supernovae as standard candles, we believe our approach is worth further consideration. The experimental values determined for the Cosmological Constant are shown to lie within an acceptable range. These values, used within the Extended Large Num- ber Hypothesis, predict values for the Quark-Hadron phase transition energy in the range 295-410 MeV. It would be remarkable, if proved correct, that the Cosmological Constant could be directly determined from the analysis of galactic velocity rotation curves. Equally remarkable, if proved correct, is the idea that astronomical observations can shed light on the last quantum mechanical phase transition of the Universe, namely the Quark-Hadron. 7 Acknowledgements We would like to thank Paolo Salucci for invaluable discussions on Cosmological and Astronomical aspects related to this work and Alexander Love for useful comments on the manuscript. We also thank J. Hargreaves and D. Bailin for suggestions and useful discussions. We also would like to thank Deja Whitehouse for proof reading this document. George Kraniotis was supported for this work by PPARC. References [1] V. Rubin, A. Waterman, J. Kenney, astro-ph/9904050 [2] Trimble, Ann. Rev. Astron. Astrophysics, 1981, 25, 425,72 [3] D. Sciama, Modern Cosmology and the Dark Matter Problem, Cambridge Uni- versity Press, 1995 [4] E. Corbelli & Paolo Salucci, accepted by Mon. Not. R. Astron. Soc., astro- ph/9909252 8 [5] S. van Albada & R.Sancisi, Phil. Trans. R. Soc. Lond., A320, (1986),P447 [6] N. Bahcall, J. Ostriker, S. Perlmutter, P. Steinhardt, astro-ph/9906463 [7] Turner, astro-ph/9904051 [8] T.S. Van Albada, J.N. Bahcall, K. Begeman and R. Sanscisi, Astrophysical Jour- nal,295(1985)305 [9] Ya.B .Zel'dovich, Usp. Fiz. Nauk 95, 209 (1968) [10] S. B. Whitehouse & G. V. Kraniotis, astro-ph/9911485 [11] S. Perlmutter et. al., astro-ph/9812133, December 1998. (Type 1a Supernovae) [12] S.M. Kent, Astr.J.93(1987)81 [13] P. Salucci, private communication; K.G. Begeman, Astron. Astrophys.223(1989)47 [14] H.C.Ohanian & R. Ruffini, Gravitation & Spacetime, W.W.Norton & Company, 1996 [15] A.G. Riess, A.V. Filippenko, W. Li, B. P. Schmidt, astro-ph/9907038, accepted by the Astronomical Journal; F. Hoyle, G. Burbidge and J. V. Narlikar, "A dif- ferent Approach to Cosmology", Cambridge University Press, Cambridge, 2000; "How will it end?",New Scientist, 17 July 1999,page 4 [16] S.M. Carrol, W.H. Press and E.L. Turner, Ann. Rev. Astron. Astro- phys.30(1992)499 [17] E. W. Kolb & M.S. Turner, "The Early Universe", Frontiers in Physics;v. 69, Addison-Wesley (1990) [18] R.Matthews, Astronomy & Geophysics, Vol 33, no. 6, (1998), 19-20 [19] P. Dirac, Proc. Roy. Soc. A165, 198 (1939) [20] P.Dirac, Proc. R. Soc. Lond., A 338, (1974), 439 [21] P.Dirac, Proc. R. Soc. Lond., A 365, (1978), 19 9
astro-ph/9802261
1
9802
1998-02-19T20:48:51
Dynamical friction in dwarf galaxies
[ "astro-ph" ]
We present a simplified analytic approach to the problem of the spiraling of a massive body orbiting within the dark halo of a dwarf galaxy. This dark halo is treated as the core region of a King distribution of dark matter particles, in consistency with the observational result of dwarf galaxies having solid body rotation curves. Thus we derive a simple formula which provides a reliable and general first order solution to the problem, totally analogous to the one corresponding to the dynamical friction problem in an isothermal halo. This analytic approach allows a clear handling and a transparent understanding of the physics and the scaling of the problem. A comparison with the isothermal case shows that in the core regions of a King sphere, dynamical friction proceeds at a different rate, and is sensitive to the total core radius. Thus, in principle, observable consequences may result. In order to illustrate the possible effects, we apply this formula to the spiraling of globular cluster orbits in dwarf galaxies, and show how present day globular cluster systems could in principle be used to derive better limits on the structure of dark halos around dwarf galaxies, when the observational situation improves. As a second application, we study the way a massive black hole population forming a fraction of these dark halos would gradually concentrate towards the centre, with the consequent deformation of an originally solid body rotation curve. This effect allows us to set limits on the fraction/mass of any massive black hole minority component of the dark halos of dwarf galaxies. In essence, we take advantage of the way the global matter distribution fixes the local distribution function for the dark matter particles, which in turn determines the dynamical friction problem.
astro-ph
astro-ph
M on.N ot.R .A stron.Soc.000,000{000 (1994) P rinted 26 January 2014 (M N plain TEX m acros v1.6) D ynam ical friction in dw arf galaxies X .H ernandez1;2 and G erard G ilm ore1 1 Institute of A stronom y, C am bridge U niversity, M adingley R oad, C am bridge C B 3 0H A 8 9 9 1 b e F 9 1 1 v 1 6 2 2 0 8 9 / h p - o r t s a : v i X r a A B ST R A C T W e presenta sim pli(cid:12)ed analytic approach to the problem ofthe spiraling ofa m assive body orbiting w ithin the dark halo of a dw arf galaxy.T his dark halo is treated as the core region of a K ing distribution of dark m atter particles, in consistency w ith the observationalresult ofdw arfgalaxies having solid body rotation curves.T hus w e derive a sim ple form ula w hich providesa reliable and general(cid:12)rstordersolution to the problem ,totally analogousto the one corresponding to the dynam icalfriction problem in an isotherm alhalo.T hisanalyticapproach allow sa clearhandling and a transparent understanding ofthe physics and the scaling ofthe problem .A com parison w ith the isotherm alcase show s that in the core regions of a K ing sphere,dynam icalfriction proceedsata di(cid:11)erentrate,and issensitive to the totalcore radius.T hus,in principle, observable consequencesm ay result.In orderto illustrate the possible e(cid:11)ects,w e apply this form ula to the spiraling of globular cluster orbits in dw arf galaxies, and show how present day globular cluster system s could in principle be used to derive better lim its on the structure of dark halos around dw arf galaxies,w hen the observational situation im proves.A s a second application,w e study the w ay a m assive black hole population form ing a fraction ofthese dark halosw ould gradually concentrate tow ards the centre,w ith the consequentdeform ation ofan originally solid body rotation curve. T his e(cid:11)ect allow s us to set lim its on the fraction/m ass of any m assive black hole m inority com ponentofthe dark halosofdw arfgalaxies.In essence,w e take advantage ofthe w ay the globalm atter distribution (cid:12)xes the localdistribution function for the dark m atter particles,w hich in turn determ ines the dynam icalfriction problem . K ey w ords:galaxies:com pact { kinem atics and dynam ics { structure { dark m atter 1 IN T R O D U C T IO N O ver the past few years it has becom e apparent that dw arf galactic system sdistinguish them selvesfrom the larger class of galaxies in m ore w ays than their total lum inosity. R o- tation velocity studies in dw arf spirals and velocity dis- persion studies in dw arf irregulars and dw arf spheroidals have revealed the presence ofa dark,dynam ically dom inant com ponent, m uch in excess of w hat a scaled dow n version of a large galaxy w ould show . T he dark m atter halo of a dw arf galaxy is m ore m assive than an extrapolation of a single m ass/lum inosity relationship (cid:12) tted to m assive galax- ies w ould suggest (eg,that of K orm endy 1986),but it also appears to have a distinct type ofstructure. O ne of the best-studied dark m atter distributions is that ofthe Sagittarius dSph galaxy (Ibata etal1997).Ibata etal (1997) analyzed their kinem atics of this galaxy to de- rive a (radial) period of the orbit of Sgr about the G alaxy ofless than (cid:24) 1G yr.T his lim it,together w ith lifetim e con- straints derived from the ages ofthe old stellar populations in Sgr,show it hassurvived for m ore than 10 orbits.N um er- icalm odels ofthe survivalofa dSph galaxy orbiting inside a larger G alactic halo,how ever,im ply com plete tidaldisrup- tion in a few orbits (V elazquez and W hite 1995; Johnston etal1995).T he solution ofthis paradox suggested by Ibata etalis that the dark m atter halo of Sgr,w hose existence is derived directly from theirkinem atics,m usthave an approx- im ately constant (H eaviside ) density pro(cid:12) le,out to a cuto(cid:11) at the tidalradius.T his m odelis consistent w ith allavail- able constraints.T he rotation curves of gas-rich dw arf spi- rals also typically show 'solid-body'linear behavior in their inner few kpc (eg,C asertano & van A lbada 1990 or B urkert 1995). O ther exam ples of galactic system s w here observa- tions suggest the presence ofa constant density structure in the centralregions are the rotation curves ofde B lok et al. (1996) for LSB 's and P ryor & K orm endy (1990) for dSph density pro(cid:12) les.A dditionally,there is the case ofD D O 154, a nearby dw arfgalaxy the rotation curve ofw hich has been extensively studied in detail. T his galaxy show s a core re- gion,a m axim um ,and then an alm ost K eplerian decline,in substantialdisagreem entw ith m odelsofdark halosobtained in cosm ological sim ulations,as analyzed by B urkert & Silk (1997).T hese observationalresults are in disagreem ent w ith the singular pro(cid:12) les obtained in self-consistent cosm ological sim ulations (e.g. N avarro et al. 1996) so that the detailed structureofdark m atterhalosispresently the objectofsom e debate e.g.B urkert(1997).From a theoreticalpointofview , in our paper I(H ernandez & G ilm ore 1997) w e use a sim ple baryonic infall m odel to infer the initial dark halo pro(cid:12) les of late type galaxies. U sing the recently observed rotation curves ofthese system s (de B lok et al.1996) and other ob- servationalrestrictions to calibrate the initialhalo pro(cid:12) les, 2 X .H ernandez and G .G ilm ore w e show that a K ing pro(cid:12) le can accurately reproduce the observed rotation curves of these system s, as w ell as those ofnorm allate type galaxies.In thatpaperw e also show that centrally divergent density pro(cid:12) les are di(cid:14) cult to reconcile w ith the rotation curves ofLSB galaxies.In as m uch as the problem is not settled, w e explore the case of a constant density halo for dw arfgalaxies,(as observations suggest) in term s of its possible im plications for dynam ical friction in those system s. T hatis,in general,w hereasm ore lum inousspiralgalax- ies show a typically (cid:13) at rotation curve out to several disk scale radii, indicating isotherm al halos w ith density falling radially asr(cid:0) 2,dw arfgalaxiesappearto have approxim ately constantdensity halosoutto the edge ofthe stellar distribu- tion.M oreover,w hile the centralregions oflum inous galax- iesare gravitationally dom inated by stars,so thatinferences about the inner structure ofthe dark m atter are necessarily m odel-dependent,the central regions of dw arf galaxies are gravitationally dom inated by dark m atter.In this paper w e consider the consequences of this relatively (cid:13) at m ass den- sity pro(cid:12) le for dynam icalevolution in the dw arf galaxy,to consider the possibility that this dom inance by dark m atter can be exploited to constrain the nature ofthatdark m atter, particularly by exploiting dynam icalfriction. D ynam icalfriction hasim portantconsequencesin shap- ing the orbitalevolution of m assive bodies w ithin dark ha- los,such as globular clusters and any hypotheticalm assive black holes.T his has been used to set som e interesting con- straints on the param eters ofthe m assive objects w hich or- bit in the dark halo system s of large galaxies, such as the allow ed m asses ofblack holes,required to be consistentw ith dynam icalfriction not having m ade their orbits decay over their lifetim es e.g. H ut & R ees (1992). Sim ilarly,if the or- biting objects are better know n,as in the case of globular clusters,the problem can be inverted,and used to set lim its on the dark halo structure and the evolution ofthe globular clustersystem (eg A guilar,H utand O striker1988).N othing w hich is not derivable from the com plete form of the rota- tion curve can be derived about the distribution function ofthe dark m atter itself.In the case ofdw arfsystem s how - ever,w hich rem ain dark m atter dom inated into their inner regions,w e can use dynam icalfriction constraints to inves- tigate the distribution function of the dark m atter in the core region, determ ined by the total m atter distribution, even though only a fraction of the rotation curve m ay be accessible to observations.It is therefore interesting to use dynam icalfriction constraints in the case of the dark m at- terdom inated dw arfgalaxies,to try to learn m ore aboutthe dark halos in these system s. A lthough the problem re(cid:13) ects the gravitationalinterac- tions ofthe m assive body w ith the totality ofhalo particles, and strictly should be treated through n-body sim ulations,a few w ellgrounded assum ptions can sim plify the problem to allow an analyticaltreatm ent.T his has been done (e.g.B in- ney & Trem aine 1987) in the case ofthe isotherm alhalo,to provide a robust,generalpurpose analyticalform ula w hich can give a reliable (cid:12) rstordersolution,w hich although notas accurate as an n-body code,has the advantage ofproviding physicalinsightinto the solution and the scaling ofthe prob- lem .D ynam icalfriction is m ost sensitive to the low velocity particles, w hich are the ones w hich interact m ore strongly w ith the body undergoing this friction. T he presence of a core in a dark halo m eans e(cid:11) ectively changing the ratio be- tw een low and high velocity halo particles,tow ardsm ore low velocity ones,as the high energy particles are m ostly found on extended orbits. H ence, it seem s reasonable to suspect that a distribution function corresponding to a system w ith a core w ill change the problem of orbital spiraling due to dynam ical friction w ith respect to the case of an isother- m aldark m atter halo.Since dSph galaxies do have globular clusters in som e cases (Sgr has four) one is interested to know iftheir continued existence provides any usefullim its on the dark m atter.In this paper w e develop a form ula to- tally analogous to the one ofB inney & Trem aine (1987) for the gravitationalfriction on a m assive body orbiting w ithin an isotherm al sphere,but appropriate to the dw arf galaxy dark m atter problem . In this paper the dynam icalfriction form ulation for the core regions ofdw arfgalaxies is derived and applied in con- nection w ith tw o aspects ofthe dynam ics ofthese system s. First,w e try to provide inform ation on the structure ofthese dark halos, from regions beyond those accessible through direct m easurem ents of stellar velocities, using the global inform ation contained in the local distribution function of the halo particles, through dynam ical friction e(cid:11) ects. T he orbital spiraling of a globular cluster w ill depend on the global dark halo structure, as it is this w hich determ ines the localdistribution function,w hich in turn determ ines the dynam icalfriction problem .W e can (cid:12) nd w hat range ofhalo structures is consistent w ith som e observed globular cluster system age and orbital distribution.Second,w e consider a m ore general case, w here som e part less than 100 percent of the dark halo is in the form of m assive com pact objects (black holes?), and derive som e lim its on the m ass of any such black holes by requiring that their orbits should not have decayed over the lifetim e ofthe system . In dw arfgalaxies,it can not be expected that the den- sity pro(cid:12) le inferred from the regions w here the stars can be m easured extendsinde(cid:12) nitely in a shallow density pro(cid:12) le.It seem s m ore naturalthat w e should be seeing only the core region ofa distribution ofdark m atter,the rest ofthe halo being em pty of stars, see for exam ple P ryor & K orm endy (1990) and Lake (1990). K ing-m odel spheres represent a self-consistent solution to both a B oltzm ann and a Poisson equation,and have been found to (cid:12) tw ellthe end productsof N -body violent relaxation sim ulations,so w e adopt here the plausible sim plifying assum ption that w e m ay treat galactic dark halos as K ing spheres. W e note that in our paper I w e (cid:12) nd K ing pro(cid:12) les to adequately reproduce the rotation curves ofLSB and dw arfgalaxies.In that paper w e perform a detailed study of baryon dissipation w ithin dark halos, calibrated using a variety ofobservationalrelations. In Section 2) w e present the derivation of the sim ple analytical form ula describing orbital spiraling due to dy- nam icalfriction in the core regions ofdark halos,w hich w e apply in section 3) to the tw o problem s m entioned above, in connection to dw arfgalaxies.In section 4) w e present the conclusions ofthis paper. 2 T H E O R E T IC A L A P P R O A C H Take a body A ofm ass M m oving on a stable circular orbit around the centre ofa spherically sym m etricalm ass distri- bution (cid:26)(r),m ade up of an equilibrium distribution of self gravitating particlesofm assm ,w here M > m .C onsiderone gravitationalencounter betw een the body A and one ofthe background particles.T hisproducesa sm alldeviation in A 's orbit producing a negative change (cid:0) (cid:1) V in the originalfor- w ard velocity,V ,and a positive change + (cid:1) V perpendicular to the originalforw ard velocity,in the plane ofthe interac- tion.C onsider now the totality ofthe background particles. In this case allthe + (cid:1) V deviations w illaverage out to zero, and only a net (cid:6) ((cid:0) (cid:1) V ) w illrem ain,in the direction ofthe original velocity. T his deceleration is w hat constitutes dy- nam icalfriction. Ifone now integrates over the e(cid:11) ects ofparticles inter- acting w ith im pact param eters from 0 to bm a x ,a m axim um im pactparam eterrelevantto the problem ,and assum esthat the background particles m ove isotropically,one obtains: dV dt = (cid:0) 16(cid:25) 2 ln(cid:3) G 2m M V 2 Z V 0 2 f(v)v dv; (1) for the deceleration parallelto V experienced by A as a re- sult ofthe collective interactions w ith the background par- ticles having a distribution function f(v), isotropic in ve- locity space. In equation (1) (cid:3) = bm a x V 2 T =(G M ), and VT is a typicalvelocity of the system (see B inney & Trem aine 1987).It has been assum ed that ln(cid:3) is a constant,although this is not strictly true,but usually (cid:3) > > 1,and ln(cid:3) does not vary appreciably for m ost applications.Sim ilarly,ln(cid:3) is not sensitive to the choice of VT ,taken in this w ork as the circular velocity ofthe halo ofbackground particles.N um er- ical experim ents have con(cid:12) rm ed the validity of (1), and of ln(cid:3) = cte:e.g.B ontekoe & van A lbada (1987) and Zaritsky & W hite (1988),see B inney & Trem aine (1987) for a m ore detailed discussion,and a derivation ofequation (1). A s explained in the introduction,in order to use equa- tion (1) in the case ofdw arfgalaxies,w e shallassum e their dark m atter halos to be w ellrepresented by the core regions ofa K ing distribution,approxim ated by a constant density region,i.e., (cid:26)(r < r0)= (cid:26)0 (2) D ynam icalfriction in dwarf G alaxies 3 =) I(r)= (cid:18) 1 4(cid:25) erf(X A )(cid:0) 2X A p (cid:25) (cid:0) X e 2 A (cid:0) A 4X 3 3 p (cid:25) (cid:19) 2 (cid:0) Y e (5) A tthispointw e have to introduce an assum ption about the orbit ofbody A ,so that w e can evaluate X A .Since the dynam ical friction drag rem oves energy from A in propor- tion to (dV )2, and angular m om entum only in proportion to dV ,as tim e progresses,A w illtend to settle into an orbit ofm axim um angular m om entum .T his process leads to the circularisation ofthe orbits ofbodies under the in(cid:13) uence of dynam ical friction. In this w ork, w e w ill assum e that the orbits ofspiraling bodies are alw ays circular,(see for exam - ple W ahde & D onner (1996), w ho study the m ore di(cid:14) cult problem ofthe in(cid:13) uenceofthe disk on the dynam icalfriction problem ,also underthe assum ption ofcircular orbitsfor the spiraling body) in accordance w ith the objective ofderiving a sim ple form ula forthe process.In thiscase,V = Vc,w here: 2 c (r)= V G M H (r) r = 4(cid:25) 3 2 G (cid:26)0r =) Vc(r)= V0(r=r0) 2(cid:27) (cid:17) X ,and M H (r)refers N ow Vc(r0)= V0,and X c = Vc= to the dark halo m ass internalto radius r,w ith M H (r0) = M H ,the totalhalo m ass w ithin the core radius. p N ow w e need ve,the escape velocity.From equation (2), Z r0 2 e (r)= v 1 2 G M H (r) dr + r r2 Z 1 r0 G M H r2 dr: T he second integral is an underestim ate,as (cid:26)(r > r0) 6= 0 (except in a tidally truncated dSph,such as Sgr),but since w e are interested only in the region r < r0, w hich corre- sponds to X < 1:23, this underestim ate introduces little error{ notice the X 3 and the exp((cid:0) Y 2) in the relevant term in I(r).N ow , and f(v)= n0 (2(cid:25)(cid:27) 2)3=2 (cid:0) 2 exp((cid:0) v =2(cid:27) 2)(cid:0) exp((cid:0) v 2 e =2(cid:27) (cid:1) 2) ; (3) (cid:18) (cid:19) 2 e (r)= v 1 2 4(cid:25) 3 G (cid:26)0 0 (cid:0) r2 r2 2 2 + r 0 In equation (2) (cid:26)0 is the density w ithin the core region, (cid:0) (cid:1)1 2 , and (cid:27) r0 the core radius, de(cid:12) ned as r0 = is the isotropic velocity dispersion of the halo particles. In equation (3), n0 is a particle num ber density, n0 = (cid:26)0=m , and ve = ve(r)isthe escape velocity ofthe halo.Introducing equation (3) into equation (1),one obtains: 9(cid:27)2=4(cid:25)G (cid:26) 0 (cid:18) (cid:19) 2 e (r)= V v 2 c (r) 3r2 0 r2 (cid:0) 1 =) Y 2 = X 2 (cid:18) (cid:19) 3r2 0 r2 (cid:0) 1 (6) dV dt = (cid:0) 16(cid:25) 2 ln(cid:3) G 2 (cid:26)0M I(r)V (cid:0) 2 ; w here, (4) Introducing R = r0=r,Vc(R ) = (31=2(cid:27)R ) and Y 2 = (9=2 (cid:0) X 2) : I(r)= Z V (cid:18) 1 (2(cid:25)(cid:27) 2)3=2 0 exp( (cid:0) v2 2(cid:27)2 )(cid:0) exp( e (cid:0) v2 2(cid:27)2 (cid:19) ) 2 v dv: p U sing the substitutions X = v=( p 2(cid:27)), Y = ve=( 2(cid:27)) and X A = V p ( 2(cid:27)) ,I(r) becom es: I(r)= Z X A 1 (cid:25)3=2 0 (cid:16) 2 X (cid:0) X e 2 (cid:0) Y (cid:0) e (cid:17) 2 dX : I(X )= (cid:18) 1 4(cid:25) erf(X )(cid:0) 2X p (cid:25) exp((cid:0) X 2)(cid:0) 4X 3 p 3 (cid:25) (cid:19) exp(X 2 (cid:0) 9=2) :(7) N ow to calculate the spiraling ofA ,w e use the assum p- tion that it alw ays orbits at Vc,therefore,jL j= rM Vc and d(L =M ) dt = Vc dr dt 4 X .H ernandez and G .G ilm ore A lso,the acceleration in equation (4) corresponds to a drag force M (dVc=dt) parallelto Vc,and therefore a torque d(L =M ) dt = r dVc dt =) dr dt = r dVc Vc dt (8) N ow de(cid:12) ne (cid:28),the characteristic tim e-scale w ith w hich the orbit ofA decays as: (cid:28) = r 2(dr=dt)r using equation (8), (cid:28) = Vc 2(dVc=dt)r and from equation (4), (cid:28) = V 3 c 32(cid:25)2ln(cid:3) G 2M (cid:26)0I(X ) Substituting Vc for (cid:26)0,V0 for Vc and X for r,w e get: (cid:28) = (cid:16) V0 r0 (cid:17) (cid:16) 2 (cid:17) 3=2 r3 0 3 24(cid:25)ln(cid:3) M G (cid:18) (cid:19) X 3 I(X ) (9) N ow ifF (X )= X 3=I(X ),w e see in Fig.1 that F (0)= 16:89, rem aining alm ost constant as X increases, reaching 19 for X = 0:5,and increasing slightly to reach 29 at X = 1, and increasing slightly faster beyond that point.T his show s that orbitaldecay in the core region ofa K ing sphere from R = 1 to around R = 0:8 proceeds faster than exponential, and afterthatitbecom es essentially exponential,w ith tim e- scales: (cid:16) (cid:28) = (20) V0 r0 (cid:17) (cid:16) 2 (cid:17) 3=2 r3 0 3 24(cid:25)ln(cid:3) M G ; w hich com es to: (cid:28)D F = (cid:17) (cid:16) V0 r0 r3 0 3M ln(cid:3) G yr; (10) w hich is the (cid:12) nalresult ofthis section,w here [V0]= km =s, [r0]= kpc and [M ]= 105M (cid:12) .T he m ain di(cid:11) erences betw een this result and that for the isotherm al case in B inney & Trem aine is the density pro(cid:12) le, and the fact that v=(cid:27) is a function ofr,rather than X (cid:17) 1.T his yields an exponential tim e-scale independent of the radius at w hich the particle orbits and w hich is only a function ofthe centraldensity of the halo,and the totalcore radius. F igure 1. T he function involved in equation (9), F (X ) and the m ore rigorously derived function F 0(X ) described in A ppendix B , for 0 < X < 1, the range corresponding to R (cid:20) 0:82. N otice the lack of any strong variations in F (X ) over this range,w hich has been used to approxim ate this function by a constant, close to the value it takes in the inner regions. T he m ore rigorously correct expression F 0(X ) is seen not to di(cid:11)er m uch from F (X ), show ing that the escape velocity approxim ation introduces little error. 0, and (cid:3) and the units are the sam e as in equation (10). T he m ain di(cid:11) erence betw een the tw o cases, is that in the isotherm al one the spiraling body reaches the centre in a (cid:12) nite am ount of tim e, w hereas in the core region case the orbitalspiraling only asym ptotically reachesthe centre,w ith a half-orbit tim e,(cid:28)D F . To com pare these tw o results, consider an observed body at a galactocentric distance ofrB ,orbiting at a veloc- ity vB .W e w ish to com pare the evolution ofits orbit in an isotherm alsphere,w hich is an adequate (cid:12) rst approxim ation to the outerpartsofa dark halo in atleasta large galaxy,to the evolution of its orbit in a constant density core,w hich is the topicalm odelfor the inner regions ofdw arfgalaxies. For the constant density core, V0 = vB (cid:17) (cid:16) r0 rB 2.1 C om parison w ith the isotherm al case A tthispoint,itisinteresting to com pare equation (10)w ith the corresponding expression describing the spiraling of a body ofm ass M m oving on a circular orbit around the cen- tre ofan isotherm aldensity distribution,characterized by a constant circular velocity Vc, tD F = i Vc 2:64r2 M ln(cid:3) G yr; (11) (B inney & Trem aine 1987), w here tD F is the total tim e it takes for the body to spiralfrom an initialradius ri to r = and therefore,from equations (10) and (11), (cid:28)D F tD F = (cid:16) 1 8 (cid:17) (cid:16) (cid:17) 3 r0 rB In practice, the inner regions of large galaxies are baryon- dom inated rather than dark-m atter dom inated,and even in dw arfs the baryon com ponent is not alw ays negligible, so that interactions w ith the baryonic com ponent, w hich in- crease as the radius decreases,w illsubstantially m odify the orbit at late tim es.To avoid this unnecessary com plication, w e take four half-orbit tim es as representative ofthe orbital decay in the core region case. T his (cid:12) ducial num ber of half lives is only used for this particular com parison w ith the isotherm al case, and is not used again in any of w hat fol- low s. A fter this tim e the body w ill have decayed into an orbit w ith a radius 1=16 ofthe initialradius.In this case, 4(cid:28)D F tD F = (cid:16) 1 2 (cid:17) (cid:16) (cid:17) 3 r0 rB W e see that for the sim plest assum ption ofrB = r0,dynam - icalfriction decay tim escales are tw ice as fast in the case of a constantdensity core than in the case ofan isotherm aldis- tribution.T hat is,the e(cid:11) ects can be relatively large,enough to be observable.For (r0=rB )= 21=3 the totalorbitaldecay tim escale is equalin both cases,and becom es progressively longer in the constant density case,as the true core radius becom es larger than the observed orbitalradius.W hereasin the isotherm alcase the observed radialdistance and orbital velocity of the body uniquely determ ine how long the dy- nam icalfriction process w illtake to drive the body to r = 0, in the core region case one requires the centraldensity ( a velocity radius pair w ithin the region of interest) and the halo core radius.In the isotherm alcase equation (11) yields a (cid:12) xed tD F for a body w ith an observed radialdistance and rotation velocity,w hile equation (10) further requires an as- sum ed core radius for the system ,w hich could som etim es be inferred from consistency requirem ents (see the case of the Sgr dw arfbelow ). Itis thisdependence ofthe orbitaldecay process on the total core radius w hich could be used to derive structural halo param eters from orbital structures. In the isotherm al case,only the starting radius determ ines the spiraling pro- cess: there is no sensitivity to the global param eters. P re- sented di(cid:11) erently,the only param eter ofthe isotherm alhalo is Vc.It should be noted that the com parison betw een the tw o tim e scales is highly sensitive to r0=rB ,and therefore, the isotherm alform ula is in generalnot a reliable estim ate ofthe dynam icalfriction problem in cases w here a constant density core is suspected.A dditionally,in som e applications the actualevolution ofthe orbit m ight be relevant,in w hich case the better (cid:12) tting of equation (10) or equation (11) m ight revealthe presence of a core region in the dark halo in question. Finally,itshould be noted thatneither ofequation (10) nor (11) can be applied to the case of the orbitaldecay of dw arf spheroidal galaxies in the halo of our galaxy. D irect application of equation (10) w ould assum e that only the halo m atter is present, so that the baryonic com ponent of our galaxy,w hich has an im portant dynam icalcontribution overany possibly interesting dark halo core region,w ould be ignored.T hisw ould exclude an im portantfraction ofthe ro- tation velocity from consideration, resulting in erroneously short tim es being predicted.E quation (11) assum es that all the m atterresponsible forthe (cid:13) atrotation curve contributes to the dynam ical friction drag, w hich w ould again be an overestim ate ofthe dynam icalfriction,as the disk m aterial does not interact w ith the dw arfgalaxy in the sam e w ay as the halo particles.A dditionally,one w ould require a consis- tent distribution function for the halo particles,in the pres- ence ofdynam ically im portant disk and bulge com ponents. For the above reasons,equation (10) is only relevant to the internal dynam ics of dw arf galaxies, or other clearly dark m atter dom inated system s,w here a uniform radialdensity is suspected from observations. Such cases do exist (Ibata D ynam icalfriction in dwarf G alaxies 5 etal 1997), and m ay even be the norm w ith LSB galaxies (paper I). 3 A P P L IC A T IO N S In this section w e use equation (10) in tw o interesting prob- lem s,to present a theoreticalfram ew ork w hich should allow us to derive im portant constraints on the structure of the dark halos of dw arf galaxies, and to set som e constraints on the fraction of these halos w hich could be m ade up of m assive com pact objects. 3.1 C ore radii determ ination in dw arf galaxies W e can use equation (10) to obtain inform ation on the size of the core region of dw arf galaxies by considering the ef- fects ofdynam icalfriction on the globular cluster system sof these galaxies.It is ofcourse notoriously di(cid:14) cult to deduce the properties of a hypothetical destroyed parent popula- tion from few ,or no,survivors.N onetheless,our aim here is to illustrate the sensitivity ofsuch analyses to assum ptions m ade concerning the spatialdensity distribution ofthe dark m atter,independent ofcurrent observationallim itations. For sim plicity here,to illustrate the scale of the e(cid:11) ect to zeroth order,w e characterize lum inous globular clusters as having uniform param eters,in particular m asses close to M = 105 M (cid:12) ( see H arris 1991 for a review on the subject). Taking M = 105 M (cid:12) , representative of a typical globular cluster,bm a x = 3kpc,in the range ofthe solid body rotation regions of dw arf galaxies, and VT = 40km =s, used only to calculate ln(cid:3) , w hich is only m arginally sensitive to these values,w e obtain: (cid:28)D F = (cid:17) (cid:16) V0 r0 r3 0 27:8M G yr: (12) Suppose that a dw arfgalaxy is observed,having a solid body rotation curve out to the last m easured point,at r = 2:5kpc,w ith Vc(2:5kpc) = 30km =s (typicalvalues for these system s, for exam ple C arignan & B eaulieu (1989) for the case of D D O 154 in w hich case the H I rotation curve w as m easured beyond the extentofthe stellar content).T he null hypothesis w ould be to assum e that the core region of this galaxy m easures2:5kpc,extending only asfarasthe rotation curve could be m easured, w ith V0 = 30km =s, but this is clearly only a low er lim it.Ifw e take r0 = 2:5kpc and V0 = 30km =s,for globular clusters ofM = 105 M (cid:12) ,equation (12) gives: (cid:28)D F = (30=2:5)(2:53 =27:8) = 6:74G yr: T histim escale issu(cid:14) ciently shortcom pared to the ages of dw arf galaxies in the Local G roup as to be potentially interesting. In general, as the dw arf galaxies correspond to higher contrast initial (cid:13) uctuations, they becam e bound structures,and perhapsinitiated starform ation,earlierthan norm al,larger galaxies. A dditionally,direct stellar popula- tion studies in these system s have yielded population ages ofabout the age ofthe universe (e.g.H odge 1989,Ibata etal 1997).In view ofthe above,12G yr seem s like a suitable age for these system s. T he application of equation (12) show s that if the core radius m easured only 2:5kpc, alm ost 2 or- bit half-tim es have elapsed for the globular cluster system , w hich should therefore show signi(cid:12) cant dynam ical friction 6 X .H ernandez and G .G ilm ore e(cid:11) ects.Speci(cid:12) cally,one expects a low abundance of globu- lar clusters at large radialdistances and to (cid:12) nd them only concentrated very close to the centre of the system ,w here dynam icalfriction w ould drive them . W ere such a distribution observed, could one in fact reliably infer that dynam ical friction m ight have been to blam e? Suppose instead,that the actualdark halo core re- gion is 3:5kpc in size.In this case,as the density ofthe dark m atter w ould notchange,(V0=r0)rem ains the sam e,and w e obtain, (cid:28)D F = 6:74(3:5=2:5)3 = 18G yr: T his last value is larger than the age ofthe universe,and w e w ould therefore expect to see no dynam icalfriction e(cid:11) ects in the globular cluster system ofthis galaxy. C learly,the presentspatialdistribution ofglobularclus- ters in dw arfgalaxies is a (one-w ay) test ofthe im portance ofdynam icalfriction,by investigating the incidence ofspa- tially extended cluster system s in galaxies w ith dark-m atter dom inated,linear,rotation curves.T his program is not ap- plicable at the present tim e,as w e lack inform ation on the state and nature of globular cluster system s around dw arf galaxies. In fact, due to observational di(cid:14) culties, together w ith the intrinsic rarity ofglobular cluster system s in dw arf galaxies,there are presently only a handfulofdetections of globularclustersaround dw arfgalaxies.Itisthrough the use of accum ulated deep H ST and w ide angle photom etry and redshift surveys (e.g. SD SS,2dF) that this problem m ight be explored. A sa speci(cid:12) c exam ple,w e can take the case ofthe Sagit- tarius dw arf, the recently discovered dSph galaxy, w hose dark m atter distribution, as derived from stellar kinem at- ics, is in good agreem ent w ith the core m odeldiscussed in this paper. T he Sgr dw arf has four globular clusters, and has been studied fairly w ellkinem atically (Ibata etal1997). From the observed velocity dispersion ofstarsin thisgalaxy, w e adopt V (1kpc) = 20km =s.A dditionally,the tidalradius for this galaxy at its current position is (cid:20) 1kpc (Ibata et.al. 1997). A pplying equation (12) w ith these num bers,w e ob- tain (cid:28)D F = 0:72G yr,for globular clusters of 105M (cid:12) .Since the globular cluster system of this galaxy has not decayed com pletely,as such a short half life w ould suggest,the Sgr dw arf necessarily had originally a dark halo core radius of m ore than itspresenttidalradius.T he dependenceof(cid:28)D F on the totalcore radius ofthe galaxy allow s to reconcile theory w ith observations, as a larger core radius (not constrained by any direct observation) w ould result in m ore extended values for (cid:28)D F . Q uantifying the original size is necessarily inexact. If, for exam ple,w e assum e that there have elapsed only 2 half lives for this globular cluster system , equation (12) yields an original core radius of 2:4kpc, w hich is larger than the observed 1kpc tidalradius.T his is consistent w ith the (cid:12) nd- ing that the stellar population w hich has been associated w ith this galaxy presently spreads over 3kpc,show ing signs oftidaldisruption by the G alaxy.T his is w hat w ould be ex- pected if the originalextent of this galaxy had been larger than its present tidalradius,as equation (12) suggests.Fur- ther, it being the only dSph w ith a globular cluster sys- tem also suggest the Sagittarius dw arf had a large original size. T hus, the existence of a globular cluster system , to- gether w ith the constant density dark m atter m ass distri- bution derived from stellar kinem atics, and the dynam ical friction analysis of this paper, requires that Sgr is signi(cid:12) - cantly tidally stripped. T hus,through requiring thatthe dynam icalfriction de- cay tim e, (cid:28)D F be consistent w ith the spatial extent of the globular cluster system over the age of the galaxy w e have used equation (12) to set som e constraints on the size ofthe originalcore region ofthissystem .Ifw e took the form ula for the isotherm alhalo,w e w ould obtain tD F = 8(cid:28)D F = 5.7 G yr, also signi(cid:12) cantly shorter than the age ofthis system .Since, in the isotherm al case, there is no further dependence be- yond the observed position and orbitalvelocity,there w ould be no w ay to reconcile theory w ith observations in this case. T his exam ple serves to illustrate the di(cid:11) erences in the dy- nam ical friction problem betw een the isotherm al and con- stant density core cases, as w ellas to provide independent con(cid:12) rm ation for the dynam ical calculations of Ibata et al (1997), in the sense that the dark halo of the Sgr dw arf is probably characterized by a constant density pro(cid:12) le. 3.2 T he case of m assive black holes in the dark halos of dw arf galaxies In this sub-section w e shall use equation (10) to set som e lim its on the fractional part of dw arf galactic halos w hich could bem ade up ofm assive black holes.In thereview on the subject ofbaryonic dark m atter by C arr (1994),it is show n thatthe only non-excluded baryonic dark m attercandidates forthe halosofgalaxies (like ourow n)are brow n dw arfstars and m assive black holes in the range 103 (cid:0) 107 M (cid:12) . G rav- itational m icrolensing searches (eg A lcock et al. 1996) are idealprobes at low m asses,butare less sensitive to the very rare and very long tim e-scale eventscaused by m assive black holes.W e can provide a lim it in the case that m assive black holes m ake up only a part ofthe dark halo,by investigating the survivaltim e ofsuch a system againstdynam icalfriction decay. W e calculate the tim e it w ould take for such m assive black holes(m aking up a fraction (cid:13) ofthe totaldark halo)to spiraltow ards the centre ofthe dark halos ofdw arfgalaxies, as a result of dynam icalfriction w ith another dark m atter com ponent, m ade up of sm all particles. To (cid:12) rst order, w e can expect that after 1 orbit half-life allthe black holes for- m erly partofthe uniform density distribution w ithin r0 w ill have form ed a new density distribution w ithin r0=2,leaving the rem aining fraction of dark m atter as it w as originally. C learly,thisscenario for the evolution ofthe halo isonly ap- plicable to low black-hole m ass fractions.Ifthe black holes m ake up a large fraction ofthe halo,dynam icalfriction w ill only redistribute energy betw een the tw o com ponents,m ak- ing the particle distribution expand, w ith the black holes segregating only m arginally tow ards the centralregions,at w hich pointdynam icalfriction w ould stop operating.In this w ay,this m ethod is com plem entary to the m icro-lensing ap- proach, w hich is not sensitive to the possibility of only a sm all fraction of the halo being m ade up of m assive black holes. It is easy to show that the fractional increase in the rotation velocity at a radius r0=2n after n half-orbit tim es have elapsed,and a fraction (cid:13) (for (cid:13) < < 1) of the original constant density halo has concentrated interior to r0=2n is: Vc n (r0=2 )= t= n (cid:28) (cid:0) (1 (cid:0) (cid:13))+ (cid:13)2 3n (cid:1) 1=2 n (r0=2 ): Vc t= 0 T hisincrem entw ould only appearinw ardsofr0=2n ,the rotation curve beyond being reduced.It is this distortion of a solid body rotation curve,to one w ith a centralenhance- m ent, w hich w ould appear anom alous relative to observed rotation curves,and so provides the observational applica- tion here.In the lim iting case,a galaxy w hich had collected allofits black holes in the center w ould show an inner K e- plerian rotation curve,rather than the observed solid body rotation.T hus,thiscase can beexcluded by inspection.E ven for a black hole fraction as low as 0.05,after only tw o half- orbit tim e-scales,at r0=4 the rotation curve w ould show an increm ent of a factor of 2, w hich could easily be detected. For larger black hole fractions,e.g.0:125,by only one half- orbit tim e-scale the rotation velocity at r0=2 w ould show an increase of a factor of 1.4, w hich could also be easily de- tectable.W e take the m axim um (cid:13) for w hich this approach should be valid as 1=8, at w hich fraction after 1 half or- bit tim e the black hole com ponent w ould concentrate inte- rior to ro=2, w ith an average density equal to that of the background halo density.H igherblack hole fractions are not applicable, as this sim ple approach w ould predict dynam i- calfriction w ould concentrate the black hole com ponent to densities higher than the background halo densities,w hich is not physical. W e de(cid:12) ne the criticalm ass above w hich black holes can be ruled out, at a given fraction, as M c((cid:13)), from equation (12) to obtain: D ynam icalfriction in dwarf G alaxies 7 friction proceeds in general at a di(cid:11) erent rate than in the isotherm alcase.O rbitaldecay in the constant density case is rapid at (cid:12) rst,and slow s dow n as tim e progresses, rather than starting slow ly and accelerating w ith tim e, as in the isotherm alcase. 2)T he dynam icalfriction decay tim e in a constantden- sity dark halo core,presented in equation (10),can be ap- plied to severalproblem sinvolving dynam icalfriction w ithin dw arfgalaxies.T he sim ple analytic nature and generality of this expression allow s a clear understanding of the physics and of the scaling properties of dynam ical friction in this case. 3) D eterm ination and analysis of the spatial distribu- tion of the globular cluster system s around dw arf galaxies can provide insight not only into the evolutionary history of these system s,butalso into the structure oftheirdark halos. In the speci(cid:12) c exam ple of the Sagittarius dw arf spheroidal galaxy,the existence and spatialdistribution ofits globular cluster system provides direct evidence for substantialtidal stripping during the lifetim e ofthe galaxy,asw ellasm aking an isotherm alhalo pro(cid:12) le seem unlikely. 4) T he observed sm oothness ofinner rotation curves in dw arfgalaxies,together w ith the analysis here,tightly con- strains any m inority contribution to their dark halo from individual, high-m ass, objects. M assive black holes w ith m asses ofm ore than 104M (cid:12) form ing a fraction ofless than about0.1 ofthedark halosofdw arfgalaxiescan beexcluded, as such a contribution to the m ass distribution w ould evolve by dynam icalfriction to generate observable distortions to the rotation curves. M c((cid:13))= V0r2 0 n 417(1 (cid:0) (cid:13)) 105 M (cid:12) (13) A C K N O W L E D G M E N T S w here n is the num ber of half-orbit tim e-scales w hich have elapsed since the form ation ofthe system . For a typical dw arf galaxy w e can take V0 = 30km =s and r0 = 2:5kpc,ifw e evaluate equation (13) for n = 1 and (cid:13) = 1=8 w e obtain M c = 0:5 (cid:2) 105M (cid:12) . For black holes of greater m ass,the distortion to the solid body rotation curve after one halflife had passed w ould becom e noticeable.T his show s that black holes form ing less than 1=8 the m ass of dw arf galactic halos, w ith individual m asses of m ore than 0:5 (cid:2) 105M (cid:12) can be excluded.Low ering the fraction on the halo w hich the black holes constitute, or requiring 2 half lives to have elapsed before the e(cid:11) ects are noticeable only introduces a factor of about tw o, m aking the lim it m ass M c = 1 (cid:2) 105 M (cid:12) . Taking now V0 = 15km =s, and r0 = 1:0kpc, representative of a dSph galaxy, w e obtain M c = 4(cid:2) 103M (cid:12) .A gain,the uncertaintiesin the otherparam eters in equation (13)introduce a factorof2 in thisnum ber.From this w e see that black holes form ing a fraction of less than 1=8 of the dark halos of dSph galaxies can be ruled out, for m asses greater than 1 (cid:2) 104M (cid:12) .R uling out such sm all fractions is not uninteresting,as it is here w here other m ore direct m ethods becom e insensitive. 4 C O N C L U SIO N S From the analysis introduced in section 2 and the applica- tions ofsection 3,w e can conclude the follow ing: T he w ork of X . H ernandez w as partly supported by a D G A PA -U N A M grant. R E F E R E N C E S A guilar L.,H ut P.,O striker J.,1988,A pJ,335,720 A lcock C .,et al.,1996,A pJ,461,67 B inney J.,Trem aine S.,1987,G alactic D ynam ics.P rinceton U ni- verstiy P ress de B lok W .J.G ., M cG augh s.s., van der H ulst J.M ., 1996, M N - R A S,283,18 B ontekoe T j.R .,van A lbada T .S.,1987,M N R A S,224,349 B urkert A .,1995,A pJ,447,25 B urkert A .,1997,astro-ph/9703057 B urkert A .,Silk J.,1997,P reprint,astro-ph/9707343 C arignan C .,B eaulieu S.,1989,A pJ,347,760 C arr B .,1994,A R A & A ,32,531 C asertano,S.,& van A lbada,T .,1990 in B aryonic D ark M atter, eds D .Lynden-B ell& G .G ilm ore (K luw er,D ordrecht) 159 H arris W .E .,1991,A R A & A ,29,543 H ernandez X .,G ilm ore G .,1997,M N R A S,in press H odge P.,1989,A R A & A ,27,139 H ut P.,R ees M .J.,1992,M N R A S,259,27 Ibata R .A .,W yse,R .F .G ,G ilm ore,G .,Irw in,M ,& Suntze(cid:11), N , 1997 A J 113,634 Johnston K .,SpergelD .,H ernquist L.,1995,A pJ,451,598 K orm endy J., 1986 in N early N orm al G alaxies ed S.M . Faber (Springer-V erlag,N Y ) p163 1) T he inw ard spiraling of a m assive body orbiting w ithin the core region of a dark halo due to dynam ical Lake G .,1990,M N R A S,244,701 P ryor C .,K orm endy J.,1990,A J,100,127 8 X .H ernandez and G .G ilm ore V elazquez H .,W hite S.,1995,M N R A S,275,L23 W ahde M .,D onner K .J.,1996,A & A ,312,431 Zaritsky D .,W hite S.D .M .,1988,M N R A S,235,289 0(X )= I (cid:18) 1 4(cid:25) erf(X )(cid:0) 2X (cid:25)1=2 2 (cid:0) X e (cid:0) 4X 3 3(cid:25)1=2 (cid:19) (cid:0) (45=2(cid:0) X ) e (B 3) and (cid:28) = (cid:16) V0 r0 (cid:17) (cid:16) 2 (cid:17) 3=2 r3 0 3 24(cid:25)ln(cid:3) M G 0(X ); F w here F 0(X )= X 3=I0(X ). (B 4) W e (cid:12) nd,0 < (F (X )(cid:0) F0(X )) < 1:65, for 0 < X < 1, w ith the di(cid:11) erence betw een the tw o expressions decreasing rapidly as X ! 0, and increasing m onotonically as X in- creases,asillustrated in Fig 1.T herefore,considering a grad- ualcut-o(cid:11) in the density distribution,even a quite abrupt exponentialone,increasestheescape velocity w ithin thecore region enough to reduce the third term in I(X )e(cid:11) ectively to zero.T his has the e(cid:11) ect oflow ering I(X ) tow ards its value at X = 0, driving the solution closer to the exponential case, w ith the sam e half-life. T his is w hat one m ight have expected,as the m atter distribution exterior to the orbiting body does not enter into the dynam ical friction problem , except in as m uch as it determ ines the global distribution function for the halo particles i.e. (cid:27). T his dependence has already been taken into account in the expression for r0 and the assum ption ofa globalK ing distribution,and hence one should see no further dependence ofthe solution on the m atter distribution exterior to the core region. T he only approxim ation enters in considering the density w ithin the core region of the K ing sphere as constant. In as m uch as the m atter distribution beyond this region does not a(cid:11) ect the dynam ical friction problem , the distribution function, circular velocity and density pro(cid:12) le of the halo m odel are selfconsistent. Solving equation (B 4)num erically,one getsan identical expansion to equation (A 2), w ith coe(cid:14) cients: a1 = 0:25, a2 = 0:075,a3 = 0:0091,etc.N otice that 12(cid:25)3=2a1 = 16:70, w hich is consistent w ith F 0(0) = 16:70 and not far from F (0)= 16:89. T hispaperhasbeen produced using the R oyalA stronom ical Society/B lackw ellScience TEX m acros. A P P E N D IX A : SO LV IN G F O R T H E D E T A IL E D O R B IT A L E V O L U T IO N From equation (4) and equation (8), dr dt = (cid:0) 16(cid:25) 2 ln(cid:3) G 2M (cid:26)0rI(X ) V 3 c substituting Vc for (cid:26)0,V0 for Vc and X and r0 for r,w e get: dX dt = (cid:0) A I(X ) ; X 2 w here: A = (2 (cid:2) 35)1=2 (cid:25)G ln(cid:3) M r3 0 ( r0 V0 )= 277:3 M r3 0 ( r0 V0 (A 1) (cid:0) 1 )G yr w ith [V0]= km =s,[r0]= kpc and [M ]= 105M (cid:12) . Solving E q(A 1) num erically,one obtains: 12(cid:25) 3=2(a1lnX + a2X 2 + a3X 3 + :::) = (cid:0) A t+ C (A 2) W here C is given by the initial conditions, X at t= 0, and a1 = 0:25281, a2 = 0:07811, a3 = 0:01079 and so on. N o- tice that 12(cid:25)3=2 a1 = 16:89, w hich w e already knew from F(X = 0)= 16.89. In thisw ay,equation (A 2)can be used to trace theexact tem poralevolution ofthe orbit ofany m assive body,w ithin the core region ofa K ing sphere. A P P E N D IX B : T H E E F F E C T O F T H E M A T T E R D IST R IB U T IO N B E Y O N D R 0 To explore the dependence of the solution on having ne- glected the m atter content beyond the core radius, in this subsection w e calculate theescape velocity,ve(r)considering an exponentialcut-o(cid:11) starting at the core radius,i.e., (cid:26) (cid:26)(r)= (cid:26)0 r < r0 (cid:26)0e(cid:0) (R (cid:0) 1) r (cid:21) r0 (cid:27) ; (B 1) r0 3 rdr + r 0 Z 1 r0 (cid:19) dr r r W hich im plies, (cid:18) Z 2 e (r)= v 1 2 4(cid:25) 3 Z G (cid:26)0 1 (cid:16) 2 + 4(cid:25)G (cid:26) 0r 0 1 w hich gives, (cid:0) (R (cid:0) 1) e (cid:0) (R (cid:0) 1) (cid:0) e dR (cid:17) 5 R 2 (cid:0) 2 R 2 (cid:0) (R (cid:0) 1) e (cid:0) (cid:18) 2 R (cid:19) 2 e (r)= v 1 2 4(cid:25) 3 G (cid:26)0 r2 0 (cid:0) r2 2 2 2 + r 0 + 6r 0 (cid:18) (cid:19) 2 =) v e (r)= V 2 c (r) 0 15r2 r2 (cid:0) 1 ; (B 2) w hich is the equivalent ofequation (6),and re(cid:13) ects a higher escape velocity.T hisleadsto Y 2 = (45=2(cid:0) X 2),w hich m akes the equivalent ofequation (7) becom e:
astro-ph/0309042
1
0309
2003-09-02T08:15:11
Simultaneous radio and X-ray observations of the low-mass X-ray binary GX 13+1
[ "astro-ph" ]
We present the results of two simultaneous X-ray/radio observations of the low-mass X-ray binary GX 13+1, performed in July/August 1999 with the Rossi X-ray Timing Explorer and the Very Large Array. In X-rays the source was observed in two distinct spectral states; a soft state, which had a corresponding 6 cm flux density of ~0.25 mJy, and a hard state, which was much brighter at 1.3-7.2 mJy. For the radio bright observation we measured a delay between changes in the X-ray spectral hardness and the radio brightness of ~40 minutes, similar to what has been found in the micro-quasar GRS 1915+105. We compare our results with those of GRS 1915+105 and the atoll/Z-type neutron star X-ray binaries. Although it has some properties that do not match with either atoll or Z sources, GX 13+1 seems more similar to the Z sources.
astro-ph
astro-ph
Astronomy & Astrophysics manuscript no. (DOI: will be inserted by hand later) October 31, 2018 Simultaneous radio and X-ray observations of the low-mass X-ray binary GX 13+1 Jeroen Homan1, Rudy Wijnands2, Michael P. Rupen3, Rob Fender4, Robert M. Hjellming5, Tiziana di Salvo4, and Michiel van der Klis4 1 INAF - Osservatorio Astronomico di Brera, Via E. Bianchi 46, I-23807 Merate, Italy 2 School of Physics and Astronomy, University of St Andrews, St Andrews, Fife KY16 9SS, Scotland, UK 3 National Radio Astronomy Observatory, Socorro, NM 87801 4 Astronomical Institute 'Anton Pannekoek', University of Amsterdam, Kruislaan 403, 1098 SJ, Amsterdam, The Netherlands 5 Deceased Received / Accepted Abstract. We present the results of two simultaneous X-ray/radio observations of the low-mass X-ray binary GX 13+1, performed in July/ August 1999 with the Rossi X-ray Timing Explorer and the Very Large Array. In X-rays the source was observed in two distinct spectral states; a soft state, which had a corresponding 6 cm flux density of ∼0.25 mJy, and a hard state, which was much brighter at 1.3 -- 7.2 mJy. For the radio bright observation we measured a delay between changes in the X-ray spectral hardness and the radio brightness of ∼40 minutes, similar to what has been found in the micro-quasar GRS 1915+105. We compare our results with those of GRS 1915+105 and the atoll/Z-type neutron star X-ray binaries. Although it has some properties that do not match with either atoll or Z sources, GX 13+1 seems more similar to the Z sources. Key words. Accretion, accretion disks - Stars: individual: GX 13+1 - Stars: neutron - ISM: jets and outflows - X-rays: stars - Radio continuum: stars 1. Introduction Based on their correlated spectral and variability prop- erties, the brightest persistent neutron star low-mass X- ray binaries (LMXBs) are often divided in two groups: the atoll and Z sources (Hasinger & van der Klis 1989; van der Klis 1995a), after the tracks they trace out in X-ray colour-colour diagrams (CDs). Although atoll and Z sources share some variability and spectral proper- ties (in the X-ray band), there are significant differ- ences between the two groups: atoll sources are less lu- minous, have harder X-ray spectra and show stronger rapid time variability than the Z sources (van der Klis 1995a). Atoll sources are also less luminous in the radio (Fender & Hendry 2000). Although some of the observa- tional differences can be accounted for by differences in the mass accretion rate, with the Z sources probably ac- creting near the Eddington rate and atoll sources at rates ranging from near Eddington to less than 10% of it, it is generally believed that additional differences, e.g. in the Send [email protected] requests offprint to: Jeroen Homan, e-mail: neutron star properties, are required to explain all obser- vational differences (Hasinger & van der Klis 1989). The nature of GX 13+1, one of the brightest neu- tron star LMXBs, is still ambiguous. Schulz et al. (1989) grouped it with the high luminosity sources (Schulz et al. 1989), which included the six sources that were later la- beled as Z sources. Hasinger & van der Klis (1989) classi- fied it as a bright atoll source, although they noted that of all atoll sources it showed properties which were clos- est to those seen in the Z sources, most notably its flar- ing branch-like appearance in the CD and the featureless power law noise in the power spectrum that is typical for that spectral state. Although the EXOSAT observations analyzed by Hasinger & van der Klis (1989) only showed the source in the so-called 'banana branch' state, the bimodal behavior of the source reported by Stella et al. (1985) strongly suggests that the source occasionally en- ters a different state. The latter seemed to be confirmed by the first observations of the source with the Rossi X-ray Timing Explorer (RXTE), in which a clear two branched structure was found in the CD (Homan et al. 1998). Although the pattern in the CD resembled both Z and atoll source tracks, the resemblance to Z sources was 2 Jeroen Homan et al.: Simultaneous radio and X-ray observations of the low-mass X-ray binary GX 13+1 strengthened by the discovery (in the same observations) of a 57 -- 69 Hz quasi-periodic oscillation (QPO), which had properties similar to that of the horizontal branch QPO in the Z sources. The CD of GX 13+1 in a recent paper by Muno et al. (2001), which for the first time displays a 'complete' pattern, shows a sharp vertex, similar to the normal branch/flaring branch vertex seen in the Z sources and quite unlike the rather smooth curves seen in atoll sources. More recently, Schnerr et al. (2003) analyzed a large set of RXTE data and conclude that many of the source's properties do not fit within the atoll/Z frame- work, although they favor the option of an atoll source. They suggest that part of its unusual behavior can be explained with the presence of a relativistic jet, that is almost pointing directly towards us. Observations in the infrared suggest the pres- ence of a K giant secondary (Garcia et al. 1992; Bandyopadhyay et al. 1999) and a possible orbital or precessional modulation with a period of ∼20 days (Bandyopadhyay et al. 2002); these are properties that are thought to be more typical for Z sources than for atoll sources. GX 13+1 also shares a common mean ra- dio luminosity (Grindlay & Seaquist 1986; Garcia et al. 1988) with the Z sources and black hole candidates (Fender & Hendry 2000), suggesting a similar origin for the quiescent radio emission from persistent black hole and Z source X-ray binaries, and GX 13+1. None of the (other) atoll sources is consistent with this relation. Although the source is variable in radio on time scales of less than an hour, no clear relation between X-ray and radio luminosi- ties was found (Garcia et al. 1988). In this paper we present the results of a coordinated X-ray/radio campaign, which had as its principal aim to investigate the possibility that GX 13+1 shows a similar X-ray state dependence of its radio emission as is found in the Z sources. In most of those sources the radio lumi- nosity decreases from the horizontal branch to the flaring branch. We find similar behavior in GX 13+1. 2. Observations and analysis 2.1. X-ray observations Our X-ray data were obtained simultaneously with the the Proportional Counter Array (PCA; Zhang et al. 1993; Jahoda et al. 1996) and the High Energy X- ray Timing Experiment (HEXTE; Gruber et al. 1996; Rothschild et al. 1998) onboard RXTE (Bradt et al. 1993). The observations were performed in 1999 between July 31 23:52 UTC and August 1 10:39 UTC (obs. 1), and on August 04 between 02:44 and 12:05 UTC (obs. 2). The total exposure times for the two observations were, respec- tively, ∼22.5 ks and ∼18.3 ks. The PCA and HEXTE data were obtained in several different modes; the spectral and timing properties of these modes are given in Table 1. For all modes we discarded data taken during Earth occulta- tions and passages through the South Atlantic Anomaly. The Standard 2 data were used to produce light curves, colour curves, a colour-colour diagram (CD), and a hardness-intensity diagram (HID), and to perform a spec- tral analysis. Only data from Proportional Counter Units (PCUs) 0 and 2 were used (all layers), since these were the only two that were active during all our observations. The data were background subtracted; dead time correc- tions (∼3 -- 3.5%) were only applied for the spectral anal- ysis. A soft colour was defined as the ratio of count rates in the 4.2 -- 7.5 keV and 2.5 -- 4.2 keV energy bands, and a hard colour as the ratio of count rates in the 10.0 -- 18.5 keV and 7.5 -- 10.0 keV energy bands (these four energy bands correspond, respectively, to Standard 2 channels (running from 1 to 129) 7 -- 14, 3 -- 6, 21 -- 40, and 15 -- 20). A CD and a HID were produced by, respectively, plotting hard colour versus soft colour and hard colour versus the 2.5 -- 18.5 keV count rate, for each 16 s data point. The soft and hard colour curves, which had an initial time resolution of 16 s, were rebinned to a time resolution of 1024 s, to allow a better study of their variations on long time scales. The PCA spectra were created using the stan- dard FTOOLS V5.2 routines. Systematic errors of 0.6% were added and response matrices were produced using PCARSP (V8.0). The details of the models used to fit the spectra of GX 13+1 are discussed in Section 3.1. HEXTE light curves were produced by running the FTOOLS V5.2 script hxtlcurv (which automat- ically corrects for background and deadtime) on the E 8us 256 DX1F mode data of cluster A. Additional rebin- ning was applied, resulting in 1024 s time bins, to achieve better statistics. The rapid X-ray time variability of the source was studied in terms of power spectra. For this analysis we performed fast fourier transforms (FFTs) of the high time resolution (1/8192 s) data . Power spectra were produced with frequencies of 1/1024 -- 512 Hz (2 -- 26.5 keV band) and 1/16 -- 2048 Hz (2 -- 26.5 keV, 2 -- 6.3 keV, and 6.3 -- 26.5 keV bands). No background or dead time corrections were ap- plied to the data prior to the FFTs; the effects of dead time were accounted for by our fitting method. The power spec- tra were selected on time, count rate and/or colour and subsequently rms normalized according to a procedure de- scribed in van der Klis (1995b). The resulting power spec- tra were fitted with a constant, to account for the dead- time modified Poisson level, a power law, a zero-centered Table 1. Data modes RXTE/HEXTE. for the RXTE/PCA and Mode Standard 1 Standard 2 SB 125us 0 13 1s SB 125us 14 17 1s E 125us 64M 18 1s E 8us 256 DX1F Time res. (s) 1/8 16 1/8192 1/8192 1/8192 1/131072 Energy range (keV) Energy channels 2 -- 60 2 -- 60 2 -- 5.9 5.9 -- 7.6 7.6 -- 60 10 -- 250 1 129 1 1 64 256 Jeroen Homan et al.: Simultaneous radio and X-ray observations of the low-mass X-ray binary GX 13+1 3 Fig. 1. The evolution of GX 13+1 during the observations on 1999 August 1 (left column) and 1999 August 4 (right column). a,b: PCA Count rate in the 2 -- 25 keV range (16 s bins). c,d: HEXTE 20 -- 100 keV count rate (rebinned to 1024 s). e,f: Soft color (rebinned to 1024 s). g,h: Hard color (rebinned to 1024 s). i,j: VLA 6 cm radio flux density (1800 -- 4800 s [i] and 900 s [j]). Errors on the flux density in panel i are of the order of 50 µJy. For the PCA count rate and colours only data from PCUs 0 and 2 were used. For the HEXTE count rate only Cluster A was used. Note that all quantities have the same ranges for obs. 1 and 2, except for the hard colour (to allow an easier comparison the with changes in the radio flux density). Lorentzian, to account for the deviations from the power law noise, and a Lorentzian for a weak QPO in the second observation. Errors on the fit parameters were determined using ∆χ2 = 1. Upper limits on QPOs were determined by fixing the frequency and FHWM to (a range of) values and using ∆χ2 = 2.71 (95% confidence). 2.2. Radio observations The radio data were obtained with the Very Large Array (VLA) radio observatory in its most extended (A) config- uration.. GX 13+1 was observed at 6 cm on 1999 August 1 between 00:33 and 09:31 UTC (obs. 1), and on 1999 August 4 between 00:22 and 08:51 UTC (obs. 2). During 4 Jeroen Homan et al.: Simultaneous radio and X-ray observations of the low-mass X-ray binary GX 13+1 Fig. 2. (a) Colour-colour diagram and (b) hardness- intensity diagram for the two RXTE/PCA observations of GX 13+1. Points of the first observations are depicted by the open circles, those of the second observation by crosses. Each data point represent a 16 s interval. Typical error bars are shown in the lower-right corners. the first observation 25 of the 27 VLA antennas were used; 26 were used during the second observation. Due to tech- nical problems ∼39% (obs. 1) and ∼8% (obs. 2) of the total observing time was lost. Observations were made simultaneously in both circular polarizations in each of two independent 50 MHz bands centered on 4885.1 and 4835.1 MHz. Flux densities, which are all Stokes I, were referenced to those of 1328+307 (3C 286), taken to be 7.462 and 7.510 Jy at 4885.1 and 4835.1 MHz respectively (Perley et al., priv. comm.). The overall flux density scale is probably good to at least ∼ 5%. To calibrate the com- plex antenna gains the standard calibration source 1817- 254 was observed at 30 minute (obs. 1) or 6 minute (obs. 2) intervals; the reason for the denser sampling of the cali- bration source in obs. 2 was that during obs. 1 a large part of the data was rendered useless due to a combination of bad weather conditions and too long intervals between calibration observations. The data were reduced and an- alyzed using the Astronomical Image Processing System (AIPS). For each day, images were first made using the en- tire data sets, ignoring any variability of GX 13+1. These images were used to find nearby confusing sources, which were then subtracted from the original uv-data. These uv- data were then split into 15-minute bins (or longer, in the case of weak signals) and imaged to give the flux density history of GX 13+1. This procedure allows the full synthe- sis mapping of confusing sources, while retaining the best possible sensitivity to fluctuations of GX 13+1. The flux densities are the average of the of maximum flux density and the integrated flux density in the region containing the source in the cleaned image. Note that because of cal- ibration and other "dead" time, the amount of data in the each bin is usually less than the sampling time. Errors on the flux density were conservatively calculated as the sum of the squares of (1) the difference of the maximum and the integrated flux density and (2) the measured off- source rms. Early in obs. 1 and during the beginning and end of obs. 2 the source wandered away from the image center, probably because the elevations of GX 13+1 and Fig. 3. The ratio of the mean raw RXTE count rate spec- trum of observation 2 to that of observation 1, revealing an overall increase and a relative hardening of the spectrum. the calibrator were rather different there - the effects of these problems can clearly be seen in the increase of the error bars at the end of the radio light cure in Fig. 1j. 3. Results 3.1. X-ray observations From Figs. 1 -- 3 it is clear that the X-ray properties of the source changed between the first to the second observa- tion. The second observation has a higher count rate, both in the 2 -- 25 keV (Fig. 1a and b) and 20 -- 100 keV bands (Fig. 1c and d) and shows more variability in the 2 -- 25 keV band. Also, the second observation tends to be spec- trally harder; while the PCA count rates increased only by a factor of ∼1.3, the HEXTE count rates increased by a factor of ∼3.2. The spectral difference is most clearly seen in the CD and HID shown in Fig. 2, where the two ob- servations show up as distinct patches, and Fig. 3, which shows the ratio of the spectra of obs. 2 and 1. Note that the small bridge between the two patches in the HID (Fig. 2b) does not represent a real connection between the two patches, but corresponds to the upper right part of patch traced out during obs. 1 in the CD. It is not clear from our observations how the source bridges the gap between the two patches in HID - we refer to Schnerr et al. (2003) for an observation that shows a transition between the two patches. Although type I X-ray bursts have been ob- served from GX 13+1 (Fleischman 1985; Matsuba et al. 1995), none were seen during our observations. 3.1.1. X-ray spectra X-ray spectral fits were performed with XSPEC (Arnaud 1996, V11.2) to the 3-25 keV PCA spectra using two dif- ferent models. The first model is based on the contin- uum model used by Ueda et al. (2001), for their ASCA 1 -- 10 keV spectra of GX 13+1, and Sidoli et al. (2002), for their 2 -- 10 keV XMM-Newton spectra. It consists of a black body (bbody in XSPEC), and a disk black body (diskbb). Galactic neutral hydrogen absorption (phabs) was fixed at NH = 2.9 1022 atoms cm−2 (Ueda et al. Jeroen Homan et al.: Simultaneous radio and X-ray observations of the low-mass X-ray binary GX 13+1 5 Table 2. Spectral fit results. Errors on the fit parame- ters are 68% confidence limits, upper limits represent 95% confidence limits. The 3 -- 10 and 10 -- 25 keV fluxes are un- absorbed. Parameter NH (atoms cm−2) kTW (keV) kTe (keV) τ RW (km) EF E (keV) FWHM (keV) Fe EW (eV) Eedge (keV) τedge F3−10 (ergs cm−2s−1) F10−25 (ergs cm−2s−1) χ2 red (dof) Obs. 1 Obs. 2 2.9 1022 (fixed) 2.9 1022 (fixed) 0.98±0.02 2.88±0.06 7.2±0.2 21.7±0.9 6.42 (fixed) 0.78±0.13 223 9.1±0.1 0.12±0.03 1.5 10−8 1.8 10−9 0.6 (40) 0.92±0.01 3.04±0.04 8.10±0.15 27.2±0.7 6.42 (fixed) 0.4+0.2 −0.4 <96 9.1 (fixed) <0.03 2.0 10−8 3.6 10−9 0.9 (41) Fit parameters of the comptt model - kTW : input soft photon (Wien) temperature - kTe: plasma temperature - τ : plasma optical depth - RW : effective Wien radius for the soft seed photons (in 't Zand et al. 1999), here derived for an assumed distance of 7 kpc 2001). A reasonable fit was found for obs. 2 (χ2 red = 1.52 - with an unphysically small disk radius), but not for obs. 1 (χ2 red = 11.3). Adding a line (gauss) around 6.4 keV and an absorption edge (edge) around 8 keV (Ueda et al. 2001) did not lead to acceptable fits. The second model we tried was a thermal Comptonization model (comptt, Titarchuk 1994; Hua & Titarchuk 1995), which has re- cently been applied successfully to the broad band con- tinuum of several bright neutron star LMXBs (see, e.g., di Salvo et al. 2000; Iaria et al. 2001). The NH was again fixed to NH = 2.9 1022 atoms cm−2. For the second ob- servation a reasonable fit was obtained (χ2 red=1.35), but the first observation showed large residuals between 5 and 10 keV (χ2 red=15.6). Adding a line around 6.4 keV and an edge around 9 keV greatly improved the fit (χ2 red=0.6). For the second observation the addition of the line and edge led to a small improvement (χ2 red=0.9), with the two components not being detected significantly. The best-fit results with the four-component model are given in Table 2. Some caution should be taken with interpreting the spectral results, as the low line energy (which tended to decrease when not fixed) and the high edge energy indicate different ionization stages of Fe. Also the large spectral changes within obs. 1 may have been partly responsible for the observed residuals. For both observations no black body component was needed, as the Comptonization com- ponent provided good fits at low energies. For a more model independent comparison of the spec- tra of the two observations we plot the ratio of observa- tions 2 and 1 in Fig. 3. While below 7 keV the ratio is rather constant, above that energy the spectrum of obs. 2 becomes increasingly hard. x 10 spectra (2 -- 26.5 keV) Fig. 4. Power two RXTE/PCA observations (lower: obs. 1, upper: obs. 2) in an νP (ν) representation. Note that the upper power spectrum has been multiplied by a factor of 10, to avoid overlapping of the power spectra. The Poisson level has in both cases been subtracted for plotting purposes. of the 3.1.2. X-ray variability From the PCA light curves in Fig. 1 it is clear that, at least on time scales of minutes to hours, the source was more variable during the second observation. The 2 -- 26.5 keV power spectra of the two observations are shown in Fig. 4; the results of the power spectral fits are given in Table 3. Both power spectra are relatively featureless and their continuum can be well fit by a combination of power law noise and a zero-centered Lorentzian. While the power law noise has an identical index in the two observations, the FHWM of the two zero-centered Lorentzians, and hence their maxima in a νPν plot, differ by more than two orders of magnitude, suggesting that they might not be related. We added a Lorentzian to the fit function of the second observation, for a QPO at 60 Hz - this QPO has a sta- tistical significance of 3.2σ. No similar QPO was found in the first observation. We also measured the total rms in two frequency intervals; 0.001 -- 1 Hz and 1 -- 100 Hz. In both frequency intervals the fractional variability is stronger in the second observations, with the difference being more pronounced at high frequencies. We searched for high fre- quency QPOs in the 1/16 -- 2048 Hz power spectra with the FWHM fixed to 150 Hz, but only upper limits could be determined. For obs. 1 they were: 2.1% (2 -- 26.5 keV), 2.8%(2 -- 6.3 keV), and 4.0% (6.3 -- 26.5 keV) and those for obs. 2: 2.2% (2 -- 26.5 keV) , 3.7% (2 -- 6.3 keV), and 3.6% (6.3 -- 26.5 keV). 6 Jeroen Homan et al.: Simultaneous radio and X-ray observations of the low-mass X-ray binary GX 13+1 Table 3. Power spectral fit results. As a model indepen- dent measure of the variability we also include the total fractional rms in the 0.001 -- 1 Hz and 1 -- 100 Hz ranges. FWHM stands for full-width-at-half-maximum. Parameter Power law rmsa (%) Power law index Lorentzian rmsb (%) Lorentzian FWHM (%) Lorentzian freq. (Hz) QPO rmsc (%) QPO FHWM (Hz) QPO frequency (Hz) rms [0.001 -- 1 Hz] (%) rms [1 -- 100 Hz] (%) χ2 red(d.o.f ) Obs. 1 5.15±0.12 1.09±0.01 2.99±0.01 0.45±0.04 0 (fixed) Obs. 2 7.71±0.11 1.09±0.01 3.4±0.6 100±30 0 (fixed) <1 2.1±0.4 (3.2σ)d 18 (fixed) 60 (fixed) 6.6±0.02 3.17±0.10 1.07 (209) 18±6 60±2 9.15±0.02 5.93±0.06 1.01 (210) a Integrated between 0.001 and 1 Hz b Integrated between 0 and +∞ Hz c Integrated between −∞ and +∞ Hz d Significance is calculated from the power, not the fractional r.m.s. 3.2. Radio observations During the first observation the source was only sometimes detected significantly, at a very low level and with low variability; the average flux density was ∼0.25 mJy. The radio flux density in the second observations was much higher, varying between 1.3 and 7.2 mJy. The flux density showed a slow increase for about 4.5 hours, followed by a gradual decrease. No significant variability was detected on time scales less than 15 minutes. An upper limit on the size of the source of 0.15 arcsecond was obtained. 3.3. X-ray/radio comparison Comparing panels i and j of Fig. 1 with panels a and b of the same figure shows that, apart from the fact that the second observation is brighter both in X-rays and radio, there is no apparent relation between X-ray count rate and radio flux, at least not on the time scales on which the latter varies. On the other hand, a comparison of the radio flux and hard X-ray colour of obs. 2 reveals similar broad peaks centered around ∼05:00 UTC. As is obvious from panels 1h and 1j there exists a significant lag between the peaks of the hard colour curve and the radio light curve. Fitting the broad peaks in the hard color (02:52 -- 06:02 UTC) and radio curves (03:53 -- 06:53 UTC) both with a Lorentzian gives a lag between the two peaks of 42±3 minutes. 4. Discussion We have observed the neutron star LMXB GX 13+1 simul- taneously in X-rays and radio. The source was found in two clearly distinct X-ray states; the spectrally hard state had associated radio fluxes that were between 4 and 18 time higher than the maximum we detected in the softer X-ray state. This dependence of the radio flux on X-ray state is similar to what is found for other bright neutron star LMXBs (the Z sources; see Hjellming & Han 1995, and references therein) and more recently also in the less- luminous atoll source 4U 1728-34 (Migliari et al. 2003). More specifically, our observations strongly suggest that the radio flux of the source is related to the X-ray spectral hardness, on time scales of both hours and days. 4.1. Outflow It is generally believed that the radio emission from neu- tron star and black hole X-ray binaries is produced by highly relativistic electrons that interact with magnetic fields to produce synchrotron radiation. High resolution radio observations of a handful of X-ray binaries show that these electrons reside in powerful, collimated out- flows, commonly referred to as jets. It is assumed that in the X-ray binaries for which jets have not (yet) been directly observed the radio emission originates in a similar outflow. To see whether this could also be the case for GX 13+1, we estimate the size of the emission region. If, for an assumed spherical region, this size is larger than the bi- nary separation, the highly relativistic synchrotron plasma cannot be contained in the system and the most plausible option would then be that we are dealing with an out- flow. Assuming a maximum brightness temperature TB of ≤ 1012 K (see e.g. Kellermann & Pauliny-Toth 1969) and a distance of 7 kpc (Bandyopadhyay et al. 1999) we derive, following Fender (2003), a minimum size for the emitting region of ∼15 R⊙ (1012 cm) at the peak of the radio flare in obs. 2. A size estimate based on the fastest part (1.5 hr) of the rise of the radio flare (≤ c∆t ∼ 1.6 · 1014 cm) is consistent with this value. Even for a system with equal masses for the primary and secondary and an orbital pe- riod of 20 days the emission region is at least comparable and probably larger than the size of the binary system (∼ 3 · 1012 cm), suggesting an outflow is also present in GX 13+1. 4.2. X-ray/radio connection Although the exact mechanism for producing jet outflows in X-ray binaries is still not clear, the energy needed to achieve the inferred observed high bulk velocities suggests that they form in the inner parts of the accretion flow, where also most of the X-rays are produced. Simultaneous X-ray and radio observations of GRS 1915+105 seem to confirm such a direct link between the outflow and in- ner accretion disk (Mirabel et al. 1998; Klein-Wolt et al. 2002). Clear patterns of X-ray/radio behavior are also ob- served in other black hole and Z source X-ray binaries; in general the spectrally hard X-ray states have a higher radio luminosity than the spectrally soft states. Our observations of GX 13+1 reveal a similar pattern, with the second observation being both more luminous in the radio and showing a harder energy spectrum (Figs. 1 -- Jeroen Homan et al.: Simultaneous radio and X-ray observations of the low-mass X-ray binary GX 13+1 7 Table 4. Observed mean and maximum radio flux den- sities and estimated distances for the six Z sources and GX 13+1. This table is reproduced from Tables 2 and 3 in (Fender & Hendry 2000), with updated values for GX 13+1. Source Sco X-1 GX 17+2 GX 349+2 Cyg X-2 GX 5-1 GX 340+0 GX 13+1 Meana (mJy) 10 ± 3 1.0 ± 0.3 0.6 ± 0.3 0.6 ± 0.2 1.3 ± 0.3 0.6 ± 0.3 1.8 ± 0.3 Maxb (mJy) 22 13.4 1.3 3.4 1.6 0.6 7.7 distance Refs (kpc) 2.0 ± 1.0 7.5 ± 2.3 5.0 ± 1.5 8.0 ± 2.4 9.2 ± 2.7 11.0 ± 3.3 1,2 1,3 4,5 1,6,7 1 8 7 ± 1 9 -- 13 a Mean cm radio flux density b Maximum radio flux density at 6 cm (1999) 3: Refs 1: Penninx (1989), 2: Bradshaw et al. (1991), 5: Penninx et al. (1990a), Christian & Swank (1993), 9: 7: Cowley et al. Grindlay & Seaquist (1988), 11: Berendsen et al. (2000), 12: Bandyopadhyay et al. (1999), 13: this work 8: Penninx et al. (1986), 10: Garcia et al. 6: Hjellming et al. (1988), 4: Cooke & Ponman (1997), (1979), 3). While the change between the soft and hard spectral states of GX 13+1 most likely occurred on a time scale of one or two days, we also find a relation between the X-ray and radio properties on a time scale of a few hours: a hardening of the X-ray spectrum was followed by an increase in the radio luminosity with a delay of ∼40 min- utes. This is the first time such a short-term X-ray/radio connection and delay have been found and measured in a neutron star LMXB. It is interesting to note that a similar delay has also been found on several occasions in the galactic black hole X-ray binary GRS 1915+105 (Klein-Wolt et al. 2002). In that case however, the de- lay (∼45 -- 60 minutes) was with respect to the start of the dips in the RXTE/PCA count rate. These dips co- incided with a strong spectral hardening. Also, the X- ray dips and radio flares in GRS 1915+105 were a rapid recurring phenomenon, whereas the radio flare in GX 13+1 was more likely a singular event. Klein-Wolt et al. (2002) explained the observed delay in GRS 1915+105 as the time it takes for the flow to become optically thin in the radio; this effect is clearly observed in Fig. 9 of Dhawan et al. (2000), which shows light curves of the di- rectly imaged jet of GRS 1915+105 at different radio wavelengths that peak later with increasing wavelength (see also Pooley & Fender 1997). Observations of Sco X-1 (Fomalont et al. 2001) suggest that such a delay can also result from the transfer time of the energy from the core to the radio lobes. to explain the X-ray variations, our second observation of GX 13+1 suggests that a smaller fraction of the inner disk is ejected in this case. The fact that clear dips are ob- served in GRS 1915+105 might be related to the different nature of the compact object, or to a much larger amount of matter being expelled. Following Fender (2003) we can estimate the mean power of the radio event in obs. 2 to be ∼ 1.8 · 1036 erg s−1. While this is only ∼1 percent of the 3 -- 25 keV X-ray luminosity (∼ 1.4 · 1038 erg s−1), the kinetic energy associated to the (possibly relativistic) out- flow may increase this number to a larger fraction of the total accretion energy. Although short term variations in the radio luminosity of GX 13+1 have been found before, they were only com- pared to the X-ray count rates (Garcia et al. 1988), and not to the X-ray spectral properties. No correlations be- tween radio luminosity and X-ray count rate were found, most likely because X-ray data were only available for the radio weak part of their data set. The time scale of shortest radio fluctuations observed by Garcia et al. (1988) (a few hours) is consistent with that of the flare in our second ob- servation. The 6 cm peak flux density they measured was 2.2 mJy, about a factor of ∼3.3 lower than our maximum flux density. Interestingly, when fitting our X-ray spectra with the same NH used by Garcia et al. (1988) and using their assumed distance of 7 kpc we find a similar difference for the 1 -- 20 keV luminosity. 4.3. Comparison with Z and atoll source observations Before discussing the nature of GX 13+1 in view of the atoll/Z source classification we compare its X-ray/radio behavior with that of sources from both classes. Although Schnerr et al. (2003) concluded that GX 13+1 is neither a Z nor an atoll source, one could, purely based on its ap- pearance in the CD, argue that the source was in the atoll source banana branch and island state, during obs. 1 and 2 respectively, or the Z source flaring branch and normal branch. Based on observing campaigns of the Z sources GX 17+2 (Penninx et al. 1988), Cyg X-2 (Hjellming et al. 1990a), Sco X-1 (Hjellming et al. 1990b) and GX 5-1 (Tan et al. 1992), Penninx (1989) suggested that all Z sources share a common radio, UV, and X-ray lumi- nosity on their normal branch. Subsequent detections of GX 349+2 (Cooke & Ponman 1991) and GX 340+0 (Penninx et al. 1993) at approximately the predicted ra- dio brightness seemed to confirm this idea. Penninx (1989) derived a normal branch luminosity of ∼ 1.6 · 1038 erg s−1 in the 1.5-15 keV band. In the same energy band we ob- tain (extrapolating our fit to lower energies) ∼ 1.7 · 1038 erg s−1, which is remarkably close the value of Penninx (1989). The ratio of (radio flux density change)/(X-ray flux change) is much larger in GX 13+1 than in GRS 1915+105. If we assume that the inner disk is ejected to form the outflow, as has been proposed for GRS 1915+105 The assumption of Penninx (1989) might be not com- pletely valid; using the distance estimates and average flux densities from Fender & Hendry (2000) (see Table 4) it seems that there is a considerable spread in the average 8 Jeroen Homan et al.: Simultaneous radio and X-ray observations of the low-mass X-ray binary GX 13+1 radio luminosity of the six Z sources. Moreover, none of them has a higher value than GX 13+1. However, this average radio luminosity depends strongly on the time a source spends in a radio bright state. Taking the maxi- mum radio flux densities (at 6 cm) reported in the liter- ature (see references in Table 4) should partly correct for this, which results in GX 17+2 being the most luminous with GX 13+1 being second. The overall behavior of the radio brightness along the track in the CD is also similar to that observed in several of the Z sources (Penninx et al. 1988; Hjellming et al. 1990a,b). The only atoll source for which a clear pattern in the X-ray/radio emission has been found is 4U 1728 -- 34 (Migliari et al. 2003). In that source the radio flux density was highest in the island state; it increased by a factor ∼6 from the hard part of the island state toward the transi- tion between the island state and, what seemed to be, the (softer) banana branch. After this transition it dropped sharply by a factor of ∼6. While the decrease of radio flux with spectral hardness (in the island state) is opposite to overall behavior in GX 13+1 and several Z sources (where we see an increase with spectral hardness along the track in the CD), the factor of ∼6 difference between the island state and banana branch is similar in sign and magnitude to the difference between our second and first observation. The average X-ray luminosity of 4U 1728 -- 34 (assuming a distance of 5.2 kpc (Galloway et al. 2002)) is more than a factor 10 lower than that of GX 13+1, whereas the 6 cm flux density is probably more than a factor 100 lower. 4.4. Z or atoll? The nature of GX 13+1 and its place in the Z/atoll clas- sification scheme have recently been extensively studied and discussed by Schnerr et al. (2003). They found that its motion through the HID is in the opposite sense to that in the CD, with the X-ray count rate increasing again when the source moves into the spectrally hard state, con- trary to most atoll and Z sources (see our Fig. 2 and also Wijnands et al. (1997) for similar behavior in the Z source Cyg X-2). Moreover, the strength of the very low frequency variability also changes in the opposite sense to that observed in Z and atoll sources. In an attempt to fit GX 13+1 within the Z/atoll scheme as an atoll source, Schnerr et al. (2003) tried to explain the source's peculiar behavior with the presence of a relativistic jet with an axis nearly aligned to our line of sight. The radio emission could then be boosted by more than an order of magnitude, putting the radio emission of GX 13+1 in accordance with measurement and upper limits of other atoll source; the unusual X-ray phenomena could be the consequence of a better view of the X-ray emitting regions associated with the base of the jet, pos- sibly assisted by Doppler boosting. As we showed above, the X-ray and radio luminosi- ties of GX 13+1 are actually in the range expected for Z sources, without requiring any unusual jet geometries, making a Z source nature more likely in our opinion (the atoll sources GX 9+1, GX 9+9 and GX 3+1 have also similar X-ray luminosities but much lower radio luminosi- ties). This is supported by our spectral fit parameters, which are quite similar to those found in the Z sources (see e.g. di Salvo et al. 2000; Di Salvo et al. 2001) and less like those in the atoll sources (see e.g. Di Salvo et al. 2000; Barret et al. 2000; Barret & Olive 2002), and by the fact that GX 13+1 rarely shows type I X-ray bursts. Based on the morphology in our CD GX 13+1 was probably ob- served in the normal branch and flaring branch. Although no clear indications for normal branch oscillations are found, the band limited noise measured by Schnerr et al. (2003) has properties similar to the normal branch oscilla- tions found in GX 5-1 (Jonker et al. 2002) and GX 340+0 (Jonker et al. 2000) close to the normal branch/flaring branch vertex, where it is very broad (Q < 1) and rather weak (≤2% rms). In GX 5-1 and GX 340+0 this broad bump evolves into a strong 6 Hz QPO as the spectrum hardens - this is not observed in GX 13+1. There are ad- ditional differences with the other Z sources, in particular with respect to the behavior of the low frequency variabil- ity. Finally, we note that although the source was more similar to the Z sources than to the atoll sources during our observations (at a luminosity three times as high as during the EXOSAT era), this does not exclude that at lower luminosities it behaves more atoll-like. 5. Summary Our simultaneous X-ray/radio observations of GX 13+1 revealed a strong dependence of the radio brightness on the X-ray state of the source. In the hard spectral state, which was at least 4 times brighter in the radio than the soft state, we also found a correlation between the X-ray and radio properties on a short time scale, with changes in radio having a delay of ∼40 minutes with respect to those in X-rays. We attribute this delay to the time it takes for the flow to become optically thin in the radio. The absence of strong dips in the X-ray light curve during the radio flare suggests that only a small amount of the matter is redirected from the inflow to the outflow. On the basis of a comparison with atoll and Z sources we conclude that the source is more similar to the Z sources, although several properties of GX 13+1 remain unexplained. Acknowledgements. JH thanks Roald Schnerr for comments on an earlier version of the manuscript and Jon Miller for his help with the spectral data reduction. JH also acknowledges support from Cofin-2000 grant MM02C71842. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. References Arnaud, K. A. 1996, in ASP Conf. Ser. 101: Astronomical Data Analysis Software and Systems V, Vol. 5, 17 Jeroen Homan et al.: Simultaneous radio and X-ray observations of the low-mass X-ray binary GX 13+1 9 Bandyopadhyay, R. M., Charles, P. A., Shahbaz, T., & Jonker, P. G., van der Klis, M., Homan, J., et al. 2002, Wagner, R. M. 2002, ApJ, 570, 793 MNRAS, 333, 665 Bandyopadhyay, R. M., Shahbaz, T., Charles, P. A., & Jonker, P. G., van der Klis, M., Wijnands, R., et al. 2000, Naylor, T. 1999, MNRAS, 306, 417 ApJ, 537, 374 Barret, D. & Olive, J. 2002, ApJ, 576, 391 Barret, D., Olive, J. F., Boirin, L., et al. 2000, ApJ, 533, Kellermann, K. I. & Pauliny-Toth, I. I. K. 1969, ApJ, 155, L71+ 329 Klein-Wolt, M., Fender, R. P., Pooley, G. G., et al. 2002, Berendsen, S. G. H., Fender, R., Kuulkers, E., Heise, J., MNRAS, 331, 745 & van der Klis, M. 2000, MNRAS, 318, 599 Matsuba, E., Dotani, T., Mitsuda, K., et al. 1995, PASJ, Bradshaw, C. F., Fomalont, E. B., & Geldzahler, B. J. 47, 575 1999, ApJ, 512, L121 Migliari, S., Fender, R., Rupen, M., et al. 2003, MNRAS, Bradt, H. V., Rothschild, R. E., & Swank, J. H. 1993, in press, astro-ph/0305221 A&AS, 97, 355 Mirabel, I. F., Dhawan, V., Chaty, S., et al. 1998, A&A, Christian, D. J. & Swank, J. H. 1997, ApJS, 109, 177 Cooke, B. A. & Ponman, T. J. 1991, A&A, 244, 358 Cowley, A. P., Crampton, D., & Hutchings, J. B. 1979, ApJ, 231, 539 Dhawan, V., Mirabel, I. F., & Rodr´ıguez, L. F. 2000, ApJ, 543, 373 330, L9 Muno, M. P., Remillard, R. A., & Chakrabarty, D. 2001, ApJ, submitted, astro-ph/0111370 Penninx, W. 1989, in Hunt J., Battrick B., eds, 23rd ESLAB Symp. on Two Topics in X-Ray Astronomy, ESA SP-296, Volume 1: X Ray Binaries,, 185 -- 196 Di Salvo, T., Iaria, R., Burderi, L., & Robba, N. R. 2000, Penninx, W., Lewin, W. H. G., Zijlstra, A. A., Mitsuda, ApJ, 542, 1034 K., & van Paradijs, J. 1988, Nature, 336, 146 Di Salvo, T., Robba, N. R., Iaria, R., et al. 2001, ApJ, Penninx, W., Zwarthoed, G. A. A., van Paradijs, J., et al. 554, 49 di Salvo, T., Stella, L., Robba, N. R., et al. 2000, ApJ, 544, L119 Fender, R. 2003, astro-ph/0303339 Fender, R. P. & Hendry, M. A. 2000, MNRAS, 317, 1 Fleischman, J. R. 1985, A&A, 153, 106 Fomalont, E. B., Geldzahler, B. J., & Bradshaw, C. F. 1993, A&A, 267, 92 Pooley, G. G. & Fender, R. P. 1997, MNRAS, 292, 925 Rothschild, R. E., Blanco, P. R., Gruber, D. E., et al. 1998, ApJ, 496, 538 Schnerr, R. S., Reerink, T., van der Klis, M., et al. 2003, A&A, 406, 221 Schulz, N. S., Hasinger, G., & Truemper, J. 1989, A&A, 2001, ApJ, 553, L27 225, 48 Galloway, D. K., Psaltis, D., Chakrabarty, D., & Muno, Sidoli, L., Parmar, A. N., Oosterbroek, T., & Lumb, D. M. P. 2002, astro-ph/0208464, 8464 2002, A&A Garcia, M. R., Grindlay, J. E., Bailyn, C. D., et al. 1992, Stella, L., White, N. E., & Taylor, B. G. 1985, in Recent AJ, 103, 1325 Results on Cataclysmic Variables, 125 Garcia, M. R., Grindlay, J. E., Molnar, L. A., et al. 1988, Tan, J., Lewin, W. H. G., Hjellming, R. M., et al. 1992, ApJ, 328, 552 ApJ, 385, 314 Grindlay, J. E. & Seaquist, E. R. 1986, ApJ, 310, 172 Gruber, D. E., Blanco, P. R., Heindl, W. A., et al. 1996, Titarchuk, L. 1994, ApJ, 434, 570 Ueda, Y., Asai, K., Yamaoka, K., Dotani, T., & Inoue, H. A&AS, 120, C641 2001, ApJ, 556, L87 Hasinger, G. & van der Klis, M. 1989, A&A, 225, 79 Hjellming, R. & Han, X. 1995, in X-ray binaries (Cambridge Astrophysics Series, Cambridge, MA: Cambridge University Press, -- c1995, edited by Lewin, Walter H.G.; Van Paradijs, Jan; Van den Heuvel, Edward P.J.), p. 308 Hjellming, R. M., Han, X. H., Cordova, F. A., & Hasinger, G. 1990a, A&A, 235, 147 Hjellming, R. M., Stewart, R. T., White, G. L., et al. 1990b, ApJ, 365, 681 Homan, J., van der Klis, M., Wijnands, R., Vaughan, B., & Kuulkers, E. 1998, ApJ, 499, L41 Hua, X. & Titarchuk, L. 1995, ApJ, 449, 188+ Iaria, R., Burderi, L., di Salvo, T., La Barbera, A., & van der Klis, M. 1995a, in X-ray binaries (Cambridge Astrophysics Series, Cambridge, MA: Cambridge University Press, -- c1995, edited by Lewin, Walter H.G.; Van Paradijs, Jan; Van den Heuvel, Edward P.J.), p. 252 van der Klis, M. 1995b, in Proceedings of the NATO Advanced Study Institute on the Lives of the Neutron Stars, held in Kemer, Turkey, August 29-September 12, 1993. Editors, M.A. Alpar, U. Kiziloglu, and J. van Paradijs; Publisher, Kluwer Academic, Dordrecht, The Netherlands, Boston, Massachusetts, p. 301 Wijnands, R. A. D., van der Klis, M., Kuulkers, E., Asai, K., & Hasinger, G. 1997, A&A, 323, 399 Zhang, W., Giles, A. B., Jahoda, K., et al. 1993, Robba, N. R. 2001, ApJ, 547, 412 Proc. SPIE, 2006, 324 in 't Zand, J. J. M., Verbunt, F., Strohmayer, T. E., et al. 1999, A&A, 345, 100 Jahoda, K., Swank, J. H., Giles, A. B., et al. 1996, Proc. SPIE, 2808, 59
astro-ph/0202440
1
0202
2002-02-23T18:06:06
Effects of boundary conditions on the dynamics of the solar convection zone
[ "astro-ph" ]
Recent analyses of the helioseismic data have produced evidence for a variety of interesting dynamical behaviour associated with torsional oscillations. What is not so far clear is whether these oscillations extend all the way to the bottom of the convection zone and, if so, whether the oscillatory behaviour at the top and the bottom of the convection zone is different. Attempts have been made to understand such modes of behaviour within the framework of nonlinear dynamo models which include the nonlinear action of the Lorentz force of the dynamo generated magnetic field on the solar angular velocity. One aspect of these models that remains uncertain is the nature of the boundary conditions on the magnetic field. Here by employing a range of physically plausible boundary conditions, we show that for near-critical and moderately supercritical dynamo regimes, the oscillations extend all the way down to the bottom of the convection zone. Thus, such penetration is an extremely robust feature of the models considered. We also find parameter ranges for which the supercritical models show spatiotemporal fragmentation for a range of choices of boundary conditions. Given their observational importance, we also make a comparative study of the amplitude of torsional oscillations as a function of the boundary conditions.
astro-ph
astro-ph
Tavakol et al.: Effects of boundary conditions on the dynamics of solar convection zone 1 Abstract. Recent analyses of the helioseismic data have produced evidence for a variety of interesting dynamical behaviour associated with torsional oscillations. What is not so far clear is whether these oscillations extend all the way to the bottom of the convection zone and, if so, whether the oscillatory behaviour at the top and the bot- tom of the convection zone is different. Attempts have been made to understand such modes of behaviour within the framework of nonlinear dynamo models which include the nonlinear action of the Lorentz force of the dynamo generated magnetic field on the solar angular velocity. One aspect of these models that remains uncertain is the na- ture of the boundary conditions on the magnetic field. Here by employing a range of physically plausible bound- ary conditions, we show that for near-critical and mod- erately supercritical dynamo regimes, the oscillations ex- tend all the way down to the bottom of the convection zone. Thus, such penetration is an extremely robust fea- ture of the models considered. We also find parameter ranges for which the supercritical models show spatiotem- poral fragmentation for a range of choices of boundary conditions. Given their observational importance, we also make a comparative study of the amplitude of torsional oscillations as a function of the boundary conditions. Key words: Sun: magnetic fields -- torsional oscillations -- activity -- spatiotemporal fragmentation 2 0 0 2 b e F 3 2 1 v 0 4 4 2 0 2 0 / h p - o r t s a : v i X r a A&A manuscript no. (will be inserted by hand later) Your thesaurus codes are: missing; you have not inserted them ASTRONOMY AND ASTROPHYSICS Effects of boundary conditions on the dynamics of the solar convection zone Reza Tavakol⋆ 1, Eurico Covas⋆⋆1, David Moss⋆⋆⋆2, and Andrew Tworkowski†3 1 Astronomy Unit, School of Mathematical Sciences, Queen Mary, University of London, Mile End Road, London E1 4NS, UK 2 Department of Mathematics, The University, Manchester M13 9PL, UK 3 Mathematics Research Centre, School of Mathematical Sciences, Queen Mary, University of London, Mile End Road, London E1 4NS, UK Received ; accepted 1. Introduction Recent analyses of the helioseismic data, both from the Michelson Doppler Imager (MDI) instrument on board the SOHO spacecraft (Howe et al. 2000a) and the Global Oscillation Network Group (GONG) project (Antia & Basu 2000), have provided strong evidence that the previ- ously observed solar torsional oscillations (e.g. Howard & LaBonte 1980; Snodgrass, Howard & Webster 1985; Koso- vichev & Schou 1997; Schou et al. 1998), with periods of about 11 years, penetrate into the convection zone (CZ) to depths of at least 10 percent in radius. These studies have also produced rather conflicting re- sults concerning the dynamical behaviour near the bottom of the convection zone. Thus Howe et al. (2000b) find ev- idence for the presence of torsional oscillations near the tachocline situated close to the bottom of the convection zone, but with a markedly shorter period of about 1.3 years, whereas Antia & Basu (2000) do not find such os- cillations. Given the uncertainties in the helioseismic data, what is not certain so far is (i) whether torsional oscilla- tions do extend all the way to the bottom of the CZ and (ii) whether there are different oscillatory modes of be- haviour at the top and the bottom of the CZ. Work is in progress by a number of groups to re- peat these analyses in order to answer these observational questions. In parallel, attempts have been made to ap- proach these questions theoretically by modelling varia- tions in the CZ within the framework of nonlinear dynamo models which include a nonlinear action of the azimuthal component of the Lorentz force of the dynamo generated magnetic field on the solar angular velocity (Covas et al. 2000a,b; Covas et al. 2001a,b, see also erratum in Covas et al. 2002). According to these results, for most ranges of dy- namo parameters, such as the dynamo and Prantdl num- Send offprint requests to: R. Tavakol ⋆ e-mail: [email protected] ⋆⋆ e-mail: [email protected] ⋆⋆⋆ e-mail: [email protected] † e-mail: [email protected] bers, the torsional oscillations extend all the way down to the bottom of the convection zone. In addition, spatiotem- poral fragmentation/bifurcation (STF) has been proposed as a dynamical mechanism to account for the possible ex- istence of multi-mode behaviour in different parts of the solar CZ (Covas et al. 2000b, 2001a,b, 2002). In all these studies the underlying zero order angular velocity was cho- sen to be consistent with the recent helioseismic data. As in much astrophysical modelling, an important source of uncertainty in these models is the nature of their boundary conditions. Given this uncertainty, and the fact that boundary conditions can alter qualitatively the behaviour of dynamical systems, it is important to see whether employing different boundary conditions can sig- nificantly change the dynamics in the CZ, and in particu- lar whether the two dynamical modes of behaviour men- tioned above are robust with respect to plausible changes in the boundary conditions. This is important for two rea- sons. Firstly, in the absence of precise knowledge about such boundary conditions, it is important that the dy- namical phenomena of interest predicted by such mod- els can survive reasonable changes in ill-known boundary conditions. Secondly, it may in principle be possible for qualitative changes found as the boundary conditions are altered to be used as a diagnostic tool to determine the range of physically reasonable boundary conditions in the solar context. Here, by considering a number of families of boundary conditions, we show that the penetration of torsional os- cillations to the bottom of the CZ is indeed robust with respect to a number of plausible changes to the bound- ary conditions. We also find spatiotemporal fragmenta- tion in these models with a variety of, but not all, choices of boundary conditions. Given the observational impor- tance of the amplitudes of the torsional oscillations, we also make a comparative study of their magnitudes as a function of the boundary conditions. Reza Tavakol et al.: Tavakol et al.: Effects of boundary conditions on the dynamics of solar convection zone 3 2. The model the overshoot region, we allowed a simple linear decrease from η = 1 at r = 0.8 to η = 0.5 in r < 0.7. 3. The choice of boundary conditions Boundary conditions on magnetic fields are often rather ill -- determined when modelling astrophysical systems. This is certainly true in the case of the Sun and solar-type stars. Given this uncertainty, we shall consider a number of physically motivated families of boundary conditions and investigate the consequences of each on the dynamics of the CZ. In particular we shall study whether they al- low penetration of torsional oscillations all the way to the bottom of the CZ as well as supporting spatiotemporal fragmentation. We note that at θ = 0 and π symmetry conditions imply A = B = 0. In this article we shall con- centrate on the changes to the outer boundary conditions only. 3.1. Boundary conditions at r = r0 The detailed physics is uncertain near the base of the computational region (r = r0). Given that the angular momentum flux out of a region with boundary S from the magnetic stresses is RS(BBr sin θ)dS, we set B = 0 on r = r0 in order to ensure zero angular momentum flux across the boundary and, correspondingly, stress-free conditions were used for v′. The condition ∂A/∂r = A/δ crudely models A falling to zero at distance δ below r = r0 (cf. Moss, Mestel & Tayler 1990; Tworkowski et al. 1998). We chose δ = 0.03, but the general nature of the results is insensitive to this choice. Taking δ > 0 is computation- ally helpful as it reduces somewhat the field gradients near r = r0, although it is not essential. 3.2. The outer boundary conditions At the outer boundary r = R, we shall, in view of the uncertainties regarding the outer boundary conditions, consider a number of different but physically reasonable choices. One of the common choices for the outer boundary conditions adopted in literature is the 'vacuum' bound- ary condition, in which the poloidal field within r = R is smoothly joined, by a matrix multiplication, to an exter- nal vacuum solution; the azimuthal field B = 0. Given the dynamic nature of the solar surface, the usual vacuum conditions can, to some extent at least, be regarded as a mathematically convenient idealization. Some aspects of this issue have recently been discussed at length by Kitchatinov, Mazur & Jardine (2000), who de- rive 'non-vacuum' boundary conditions on both B and BP . We also consider families of boundary conditions which deviate from the vacuum conditions and refer to ∂r2 + 2 r (1) (2) = µ0ρr sin θ ∂ ∂r + 1 r2 sin θ ( ∂ (∇ × B) × B .φ + νD2v′, = ∇ × (u × B + αB − η∇ × B). We shall assume that the gross features of the large scale solar magnetic field can be described by a mean field dy- namo model, with the standard equation ∂B ∂t Here u = v φ − 1 2 ∇η, the term proportional to ∇η repre- sents the effects of turbulent diamagnetism, and the ve- locity field is taken to be of the form v = v0 + v′, where v0 = Ω0r sin θ, Ω0 is a prescribed underlying rotation law and the component v′ satisfies ∂v′ ∂t where D2 is the operator ∂ 2 ∂θ ) − 1 sin θ ) and µ0 is the induction constant. The assumption of axisymmetry allows the field B to be split simply into toroidal and poloidal parts, B = BT +BP = B φ+∇×A φ, and Eq. (1) then yields two scalar equations for A and B. Nondimensionalizing in terms of the solar radius R and time R2/η0, where η0 is the maximum value of η, and putting Ω = Ω∗ Ω, α = α0 α, η = η0 η, B = B0 B and v′ = Ω∗Rv′, results in a system of equations for A, B and v′. The dynamo parameters are Rα = α0R/η0, Rω = Ω∗R2/η0, Pr = ν0/η0, and η = η/η0, where Ω∗ is the solar surface equatorial angular velocity. Here ν0 and η0 are the turbulent magnetic diffusivity and viscosity re- spectively and Pr is the turbulent Prandtl number. Our computational procedure is to adjust Rω so as to make the cycle period be near the solar cycle period of about 22 years for the marginal dynamo number, and then to allow Rα and, to some extent, Pr to vary. The density ρ is assumed to be uniform. ∂θ (sin θ ∂ Eqs. (1) and (2) were solved using the code described in Moss & Brooke (2000) (see also Covas et al. 2000b) to- gether with the boundary conditions given below, over the range r0 ≤ r ≤ 1, 0 ≤ θ ≤ π. We set r0 = 0.64, and with the solar CZ proper being thought to occupy the region r >∼ 0.7, the region r0 ≤ r <∼ 0.7 can be thought of as an overshoot region/tachocline. In the following simulations we used a mesh resolution of 61 × 101 points, uniformly distributed in radius and latitude respectively. In this investigation, we took Ω0 in 0.64 ≤ r ≤ 1 to be given by an interpolation on the MDI data obtained from 1996 to 1999 (Howe et al. 2000a). We set α = αr(r)f (θ), where f (θ) was chosen to be sin2 θ cos θ or sin4 θ cos θ. The angular structure of α is quite uncertain, and both these forms have been used in the literature (see e.g. Rudiger & Brandenburg 1995) and their choice here is simply to make the butterfly diagrams more realistic. We took αr = 1 in all or part of the CZ (see below for details), with cubic interpolation to zero at r = r0 and r = 1 in the cases where αr 6= 1 everywhere. Throughout we take αr ≥ 0 and Rα < 0. Also, in order to take some account of the likely decrease in the turbulent diffusion coefficient η in 4 Reza Tavakol et al.: Tavakol et al.: Effects of boundary conditions on the dynamics of solar convection zone these as 'open'. As a convenient and flexible general form for the boundary conditions at the surface, we write r dA dr dB dr + n1A = 0, r + n2B = 0 (3) dr = Bθ = 0 and dB where n1 and n2 are constants that parameterize the boundary conditions and, to some extent, their degree of openness. With n1 = 1, n2 = 0 the two conditions for A and B reduce to d(rA) dr = 0 respec- tively. The condition Bθ = 0 has been adopted previously by some investigators. The limit n2 → ∞ gives the of- ten used B = 0. As n1 increases, the penetration of the poloidal field through the surface decreases, and in the limit n1 → ∞ the boundary condition is A = 0, and all the poloidal field lines then close beneath the surface r = R, which is thus the limiting field line. Using the vacuum boundary condition for BP gives poloidal field lines that mostly make a modest angle with the radial direction, and so we can anticipate that, by taking small values of n1 and large values of n2, we will obtain solutions that resemble in some ways those found by using the vacuum boundary conditions mentioned above. (But note that Eq. (3) gives strictly local conditions on the field components, whereas the vacuum condition on the poloidal field is essentially nonlocal.) We note one further technical point. The angular mo- mentum flux through r = R is non-zero if both BP and B are non-zero there. Whilst the Sun certainly is losing an- gular momentum, we are not trying to model this process here, and so will only consider models in which the angular momentum 'drift' of the dynamo region is small enough that we can consider it to be a unchanging background for the dynamo calculations. Now in order to find the range of values of n1 and n2 such that the resulting 'partially open' boundary condi- tions are physically plausible, we need to ensure that the chosen boundary conditions result in appreciable poloidal flux penetrating the surface. Thus we calculated the aver- age over the dynamo cycle of the ratio of the flux of the poloidal field at the surface to the corresponding value within the CZ, given by Fs = [R sin(θ)A(R, θ)]Surface [r sin(θ)A(r, θ)]Inner , (4) as a function of n1 say. Here R is the model radius, the numerator is evaluated at the surface r = R, and the denominator is evaluated inside the dynamo region ('CZ') r0 ≤ r ≤ R. As n1 increases we would expect this ratio to decrease. We consider a boundary condition to be, in principle, viable if the ratio Fs is not too small compared with the corresponding value in the 'standard' vacuum case. 4. Results Using the above model, we studied the dynamics in the convection zone subject to three sets of boundary condi- tions; namely, the vacuum boundary condition and two Fig. 1. The radial (r -- t) contours of the angular veloc- ity residuals δΩ as a function of time for a cut at 25 degrees latitude, with vacuum boundary conditions and Rα = −5.5, Pr = 0.6, Rω = 60000. Note the fragmenta- tion at the bottom of the convection zone and the resulting difference in periods of oscillations at the top and at the bottom. Darker and lighter regions represent positive and negative deviations from the time averaged background rotation rate. families of open boundary conditions which we shall refer to as boundary conditions (1) and (2). Also in order to demonstrate that penetration of the torsional oscillations, as well as spatial fragmentation, can occur with various changes in other ingredients of the model, we have chosen examples with different forms for these ingredients. 4.1. Vacuum boundary conditions With this choice of boundary conditions, we found that for critical and moderately supercritical regimes, the tor- sional oscillations extend all the way down to the bottom of the CZ (see also Covas et al. 2000a, where the critical value of the dynamo number for the onset of dynamo was found to be Rα ≈ −3.16). In addition we found ranges of dynamo parameters for which supercritical models showed spatiotemporal fragmentation. As an example of such frag- mentation, we show in Fig. 1 the radial contours of the an- gular velocity residuals δΩ as a function of time for a cut at 25 degrees latitude. In this case we took f (θ) = sin4 θ cos θ, with αr = 1 throughout the computational region. The parameter values used were Rα = −5.5, Pr = 0.6 and Rω = 60000. We also show in Fig. 2 the magnetic butter- fly diagram. We note that this is the first time STF has been ob- tained with vacuum boundary conditions. This is despite the fact that the range of parameter values for which STF is present for this model is rather wide, as is the case with the boundary conditions considered below. What seems to occur here is that the onset of spatiotemporal fragmen- Reza Tavakol et al.: Tavakol et al.: Effects of boundary conditions on the dynamics of solar convection zone 5 0.25 0.20 0.15 0.10 0.05 Fs 0.00 0 25 50 100 125 150 75 n1 Fig. 3. Fs as a function of n1 for the open boundary conditions (1) given by equation (3), with n2 = 0. Fig. 4. The radial (r -- t) contours of the angular velocity residuals δΩ as a function of time for a cut at 25 degrees latitude, with the boundary conditions given by (3) with n1 = 25, Rα = −5.0, Pr = 0.7, Rω = 50000. Note the fragmentation at the bottom of the convection zone and the resulting difference in periods of oscillations at the top and at the bottom. Darker and lighter regions represent positive and negative deviations from the time averaged background rotation rate. values used are Rα = −5.0, Pr = 0.7, Rω = 50000 and n1 = 25. Fig. 5 shows the angular velocity residuals at R = 0.92 (i.e. the near-surface torsional oscillations), with latitude and time, for the same parameter values as in Fig. 4. The poloidal field lines and toroidal field contours for this case are presented in Figs. 6 and 7. In Fig. 7, the effect of the open boundary condition on B is seen to be relatively minor -- most of the toroidal field is concentrated near the tachocline, and far from the surface. The poloidal field (Fig. 6) is more uniformly distributed. Fig. 2. Butterfly diagram of the toroidal component of the magnetic field B at r0 = 0.90R, with the vacuum boundary conditions and the parameter values given by Rα = −5.5, Pr = 0.6, Rω = 60000. Dark and light shades correspond to positive and negative values respectively. tation is close to a bifurcation point, which disrupts the butterfly diagram and rather confuses the diagnosis of the situation. 4.2. Open boundary conditions (1) We now take boundary conditions at r = R to be given by Eq. (3), where n2 = 0 and n1 is varied. In order to find the range of values of n1 which can be considered as physically plausible, we calculated the ratio of the poloidal fluxes Fs given by (4) as a function of n1. The results are given in Fig. 3. To give an idea of how this ratio compares to that of models with vacuum boundary conditions, we also calculated Fs with vacuum boundary conditions with the same parameter values as for Fig. 1, and found that Fs ∼ 0.25. As can be seen from Fig. 3 this is comparable in magnitude to the values obtained with the above open boundary conditions for a wide range of values of n1 given by n1 <∼ 40. With these open boundary conditions, we again found that for slightly and moderately supercritical dynamo regimes, the torsional oscillations extend all the way down to the bottom of the CZ. There are also ranges of dy- namo parameters for which spatiotemporal fragmentation is found. When n1 is close to 1 there is appreciable an- gular momentum drift, and so we do not consider these solutions further here. As an example of a case with STF (and with negligible angular momentum drift over the in- tegration interval), we show in Fig. 4 the radial contours of the angular velocity residuals δΩ as a function of time for a cut at 25 degrees latitude. In this case f (θ) = sin4 θ cos θ and αr is given by αr = 1 for 0.7 ≤ r ≤ 0.8 with cubic interpolation to zero at r = r0 and r = 1. The parameter 6 Reza Tavakol et al.: Tavakol et al.: Effects of boundary conditions on the dynamics of solar convection zone Fig. 5. Angular velocity residuals at R = 0.92 with lati- tude and time, with boundary conditions given by Eq. (3), n1 = 25, Rα = −5.0, Pr = 0.7, Rω = 50000. A tempo- ral average has been subtracted to reveal the migrating banded zonal flows. Darker and lighter regions represent positive and negative deviations from the time averaged background rotation rate. Fig. 7. The toroidal field contours BP ; same parameters as for Fig. 5. Fig. 6. The poloidal field lines; same parameters as in Fig. 5. Fig. 8. The radial (r -- t) contours of the angular velocity residuals δΩ as a function of time for a cut at 20 degrees latitude, with the open boundary conditions (2) and Rα = −16, Pr = 0.22, Rω = 48000, n1 = 2 and n2 = 400. Note the fragmentation at the bottom of the convection zone and the resulting difference in periods of oscillations at the top and a region near the bottom of the CZ around r0 = 0.7. Darker and lighter regions represent positive and negative deviations from the time averaged background rotation rate. 4.3. Open boundary conditions (2) We now consider boundary conditions given by Eq. (3), with n1 of order one and 1 ≤ n2 ≤ 500 (so B ≈ 0 for large n2). We found that for n1 values close to one our model again has significant angular momentum drift, which in- creases as n1 decreases to 1. For larger values of n1 the drift is negligible, and the ratio Fs is again 'reasonable', being comparable with the vacuum case. We found that putting n1 = 2 gave satisfactory behaviour, and this is the case that we discuss in detail below. For all such cases we again found that for slightly and moderately supercritical dynamo numbers, the oscillations extend all the way down to the bottom of the CZ. In ad- dition we found ranges of dynamo parameters for which the supercritical model show spatiotemporal fragmenta- tion. An example, which has negligible angular momen- tum drift over the integration interval, is shown in Fig. 8. Here αr = 1 for 0.7 ≤ r ≤ 0.8 with cubic interpolation to zero at r = r0 and r = 1 and f (θ) = sin2 θ cos θ. The Reza Tavakol et al.: Tavakol et al.: Effects of boundary conditions on the dynamics of solar convection zone 7 parameter values used were Rα = −16.0, Pr = 0.22 and Rω = 48000, and boundary conditions were given by (3) with n1 = 2 and n2 = 400. 4.4. Amplitudes of oscillations as a function of boundary conditions Another important issue from an observational point of view is the way the amplitudes of the torsional oscillations vary as a function of model ingredients and parameters, as well as with depth in the CZ. To begin with, we verified that for a given boundary condition the amplitudes of the oscillations increase as the dynamo number Rα and the Prantdl number Pr are increased (see also Covas et al 2001a). For orientation, we recall that the observed surface amplitudes in the case of the Sun are latitude dependent and of order of one nHz (see e.g. Howe et al. (2000b)). We also made a comparative study of the amplitudes as a function of changes in the boundary conditions. Briefly we found that typically the solutions with vacuum bound- ary conditions have rather smaller amplitudes of oscilla- tions, especially near the surface. For example, for the model of Fig. 1 (which has a high Rα, spatiotemporal fragmentation and thus would be expected to have higher amplitudes), we found the mean averaged amplitudes to be 0.72, 0.19, 0.09 nHz at the depths r0 = 0.70, 0.88 and 0.95 respectively. For the models with open boundary conditions, we found the amplitudes to be on average higher than the vacuum case, specially near the surface. We have sum- marised in Fig. 9 our calculations of the amplitudes of oscillations for the models with open boundary conditions given by Eq. (3) and n2 = 0, for a range of n1 given by 1 < n1 < 150. Here the dynamo parameters were Rα = −5.0, Pr = 0.7, Rω = 50000 and f (θ) = sin4 θ cos θ, with αr = 1 for 0.7 ≤ r ≤ 0.8, with cubic interpolation to zero at r = r0 and r = 1. As can be seen, the amplitudes grow at all depths with increasing n1 and saturate around n1 ∼ 50. As an example, the models with n1 values around n1 = 25, (corresponding to Fig. 4, with STF), have ampli- tudes that are more than double those found above with vacuum boundary conditions, down to the level r0 = 0.88. 5. Discussion We have made a detailed study of the effects of the bound- ary conditions on the dynamics in the solar convection zone, by employing various forms of outer boundary con- ditions. In all the models considered here (as well as other re- sults not reported), we find that in near-critical and mod- erately supercritical dynamo regimes the torsional oscilla- tions extend all the way down to the bottom of the CZ. In this way our results, taken altogether, demonstrate that such penetration is extremely robust with respect to all 1.00 0.80 0.60 0.40 0.20 ) z H n ( ' v r=0.70Ro r=0.88Ro r=0.95Ro 0.00 0 25 50 100 125 150 75 n1 Fig. 9. The mean amplitudes of torsional oscillations, at radii r0 = 0.70, 0.88 and 0.95. in models with open bound- ary conditions given by Eq. (3) and n2 = 0, with n1 rang- ing from 3 to 150. the changes we have considered both to the boundary conditions, and the dynamo parameters such as the dy- namo and Prantdl numbers, in addition to variations in the model ingredients such as the α and η profiles and the rotation inversion. We deduce, that if our dynamo model (which is ba- sically a standard mean field dynamo) has any validity, then observers should expect to find that the solar tor- sional oscillations penetrate to the tachocline. However, given the significant uncertainties that still remain in he- lioseismic measurements, especially the limited temporal extent of the data available, this issue may not be resolv- able at present (see e.g. Vorontsov et al. 2002). In all cases we have found supercritical dynamo regimes with spatiotemporal fragmentation for a range of, but not all, dynamo parameters. This, together with our previous work, shows that fragmentation occurs with a variety of forms of α (and also that it is not confined to a particular inversion for the solar angular velocity). For still more supercritical dynamo regimes we find a series of spatiotemporal fragmentations, leading eventually to spa- tiotemporal chaos, i.e. disappearance of coherence in the dynamo regime. These results are of potential importance in interpret- ing the current observations, especially given their diffi- culty in resolving the dynamical regimes near the bottom of the convection zone. However given the variety of dy- namical behaviour possible theoretically near the bottom of the CZ, we cannot comment definitively on the reported 1 − 3 yr oscillations. Finally, given the observational importance of the am- plitudes of the torsional oscillations, we have made a com- parative study of their magnitudes as a function of the boundary conditions. We found that on average the am- 8 Reza Tavakol et al.: Tavakol et al.: Effects of boundary conditions on the dynamics of solar convection zone plitudes are smaller for the models with vacuum boundary conditions than for those with open boundary conditions. An important ingredient that our model omits and which seems bound to have an effect on the amplitudes of tor- sional oscillations throughout the CZ, is that of density stratification. We intend to return to this issue in a future publication. Acknowledgements. RT cle Physics PPA/G/S/1998/00576. EC and DM acknowledges hospitality of the Astronomy Unit, QM. Parti- and Astronomy Research Council Grant the UK benefited from References Antia H.M., Basu S., 2000, ApJ, 541, 442 Covas E., Tavakol R., Moss D. & Tworkowski A., 2000a, A&A, 360, L21 Covas E., Tavakol R. & Moss D., 2000b, A&A, 363, L13 Covas E., Tavakol R. & Moss D., 2001a, A&A, 371, 718 Covas E., Tavakol R., Vorontsov, S. & Moss, D., 2001b, A&A, 375, 260 Covas E., Moss, D., & Tavakol R., 2002, Erratum, A&A, to appear Howard R. & LaBonte B. J., 1980, ApJ Lett. 239, 33 Howe R., et al., 2000a, ApJ Lett., 533, 163 Howe R., et al., 2000b, Science, 287, 2456 Kitchatinov L.L., Mazur M. V. & Jardine M., 2000, A&A, 359, 531 Kosovichev A. G. & Schou J., 1997, ApJ, 482, 207 Moss D., Mestel L. & Tayler R.J., 1990, MNRAS, 245, 550 Moss D. & Brooke J., 2000, MNRAS, 315, 521 Rudiger R. & Brandenburg A., 1995, A&A, 296, 557 Schou J., Antia H. M., Basu S. et al., 1998, ApJ, 505, 390 Snodgrass H. B., Howard R. F. & Webster L. 1985, Sol. Phys., 95, 221 Tworkowski A., Tavakol R., Brooke J.M., Brandenburg A., Moss D. & Tuominen, I., 1998, MNRAS, 296, 287 Vorontsov, S., Tavakol, R., Covas, E. & Moss, D., (2002) 'Solar cycle variation of the solar internal rotation: heleioseismic inversion and dynamo modelling', To appear in 'Proceed- ings of Granada Workshop on 'The evolving Sun and its influence on planetary environments', ASP (Astronomical Society of the Pacific) Conference Series, A. Gimenez, E. Guinan and B. Montesinos (eds), astro-ph/0201422, also available at http://www.eurico.web.com
astro-ph/0701629
2
0701
2007-06-17T19:08:53
Pre-Merger Localization of Gravitational-Wave Standard Sirens With LISA I: Harmonic Mode Decomposition
[ "astro-ph", "gr-qc" ]
The continuous improvement in localization errors (sky position and distance) in real time as LISA observes the gradual inspiral of a supermassive black hole (SMBH) binary can be of great help in identifying any prompt electromagnetic counterpart associated with the merger. We develop a new method, based on a Fourier decomposition of the time-dependent, LISA-modulated gravitational-wave signal, to study this intricate problem. The method is faster than standard Monte Carlo simulations by orders of magnitude. By surveying the parameter space of potential LISA sources, we find that counterparts to SMBH binary mergers with total mass M~10^5-10^7 M_Sun and redshifts z<~3 can be localized to within the field of view of astronomical instruments (~deg^2) typically hours to weeks prior to coalescence. This will allow targeted searches for variable electromagnetic counterparts as the merger proceeds, as well as monitoring of the most energetic coalescence phase. A rich set of astrophysical and cosmological applications would emerge from the identification of electromagnetic counterparts to these gravitational-wave standard sirens.
astro-ph
astro-ph
Pre-Merger Localization of Gravitational-Wave Standard Sirens With LISA: Harmonic Mode Decomposition Bence Kocsis,1, 2 Zoltán Haiman,3 Kristen Menou,3 and Zsolt Frei1 1Institute of Physics, Eötvös University, Pázmány P. s. 1/A, 1117 Budapest, Hungary 2Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138 3Department of Astronomy, Columbia University, 550 West 120th Street, New York, NY 10027 7 0 0 2 n u J 7 1 2 v 9 2 6 1 0 7 0 / h p - o r t s a : v i X r a The continuous improvement in localization errors (sky position and distance) in real time as LISA observes the gradual inspiral of a supermassive black hole (SMBH) binary can be of great help in identifying any prompt electromagnetic counterpart associated with the merger. We develop a new method, based on a Fourier decom- position of the time-dependent, LISA-modulated gravitational-wave signal, to study this intricate problem. The method is faster than standard Monte Carlo simulations by orders of magnitude. By surveying the parameter space of potential LISA sources, we find that counterparts to SMBH binary mergers with total mass M ∼ 105 -- 107M⊙ and redshifts z < 3 can be localized to within the field of view of astronomical instruments (∼ deg2) ∼ typically hours to weeks prior to coalescence. This will allow a triggered search for variable electromagnetic counterparts as the merger proceeds, as well as monitoring of the most energetic coalescence phase. A rich set of astrophysical and cosmological applications would emerge from the identification of electromagnetic counterparts to these gravitational-wave standard sirens. PACS numbers: I. INTRODUCTION One of the key objectives of the planned, low-frequency gravitational-wave (GW) detector LISA (Laser Interferomet- ric Space Antenna) is the detection of supermassive black hole (SMBH) binary mergers at cosmological distances. The observation of these chirping GW sources would deepen our understanding of (i) general relativity, e.g. by offering unique tests of spacetime physics in the vicinity of SMBHs [1, 2, 3, 4, 5], (ii) cosmology, by providing additional con- straints on the luminosity distance -- redshift relation [6, 7, 8], (iii) large-scale structure, by indirectly constraining hierarchi- cal structure formation scenarios [9, 10, 11, 13], and (iv) black hole astrophysics, e.g. by allowing accurate determinations of Eddington ratios, and other attributes of black hole accretion, in systems with SMBH mass and spin known independently, from the GW measurements [14, 15, 16]. From a purely astronomical point of view, one of the most attractive features of the LISA mission design is the possi- bility to constrain the 3-dimensional location (i.e. sky posi- tion and distance) of GW inspiral sources to within a small enough volume that the identification of potential electromag- netic (EM) counterparts to SMBH merger events can be con- templated seriously. Indeed, the accuracy of such LISA lo- calizations at merger are encouraging, with an error volume δΩ× δz = 0.3 deg2×0.1 for SMBH masses m1 = m2 = 106M⊙ at z = 1, for instance [17]. In Ref. [15], we have shown that this accuracy may be sufficient to allow an unique identifica- tion of the bright quasar activity that may be associated with any such SMBH merger. Another possibility, examined here in detail, is to monitor the sky for EM counterparts in real time, as the SMBH inspiral proceeds. This is arguably one of the most efficient ways to identify reliably (prompt) EM counterparts to SMBH merger events, since the exact nature of such counterparts is a priori unknown. Using the GW inspiral signal accumulated up to some look -- back time, tf, preceding the final coalescence, one already has a partial knowledge of where the source of GWs is located on the sky. Since the sky position is deduced primar- ily from the detector's motion around the Sun, one anticipates that angular positioning uncertainties will not change too dra- matically during the last few days before merger, so that a targeted EM observation of the final stages of inspiral may be a feasible task. Here, we present an in-depth study of the potential for such pre-merger localizations with LISA, while we discuss various astrophysical concepts and observational strategies for EM counterpart identifications in a companion work [18]. The main purpose of the present analysis is thus to deter- mine the accuracy of SMBH inspiral localizations with LISA, as a function of look -- back time, tf, prior to merger. The LISA detector is not uniformly sensitive to sources with different sky positions and angular momentum orientations. Results will thus generally depend on the fiducial values of these an- gles. Our first objective is to calculate the time-dependence of distributions of localization errors, for randomly oriented sources, over a large range of values for the SMBH masses and source redshift. A second objective of our analysis is to estimate source parameter dependencies for these distribu- tions of localization errors, i.e. how the 3-dimensional (sky position and distance) localization error distributions depend on the fiducial sky position of GW sources. This is useful to understand which regions of the sky may be best suited for the identification of EM counterparts to SMBH merger events. To the best of our knowledge, this angle dependence has not been explored in detail before, not even in terms of final errors at ISCO (i.e. at tf = tisco, when using the complete inspiral data- stream, up to the innermost stable circular orbit, or ISCO). Parameter estimation uncertainties for LISA inspirals have been considered previously, under a variety of approximations [5, 7, 8, 13, 17, 19, 20, 21, 22]. These studies differ in the levels of approximation adopted for the GW waveform, using various orders of the post-Newtonian expansion. The LISA signal output for these waveforms are obtained through a lin- ear combination of the two GW polarizations, h+(t) and h×(t), with the beam pattern functions, F+ and F×. The beam pat- terns define the detector sensitivity for the two polarizations. They are determined by the angles describing the instanta- neous orientation of the LISA constellation relative to the GW polarizations. As the LISA detector constellation orbits the Sun, with a one year period, F+ and F× are slowly changing in time and this introduces an additional time dependence in the LISA signal. As first shown by Cutler [19], the source sky po- sition can be determined with LISA using this modulation. In the formalism given by Cutler [19], this modulation couples time and angular dependencies in a complicated way, making the estimation of localization errors numerically costly for a large set of SMBH binary random orientations and parame- ters. Using a different approach, Cornish & Rubbo [23] have de- rived the orbital modulation in a much simpler form, in which the angular parameter dependence and the time dependence can be decoupled. Here, starting directly from the original Cutler [19] expression, we give an independent derivation of the Cornish & Rubbo [23] formula and write it in an equiv- alent form, from which decoupling is more evident. We do this by expanding the LISA response function into a discrete Fourier sum of harmonics of the fundamental frequency of LISA's orbital motion, f⊕ = 1 yr- 1. Since LISA's orbit does not include high frequency features, we expect this sum to be quickly convergent. In fact, it is clear from the Cornish & Rubbo [23] result that the expansion terminates at 4 f⊕ and that there are no higher order harmonics due to the detec- tor's motion. The series coefficients in the expansion are in- dependent of time and only depend on the relative angles at ISCO. We then develop a Fisher matrix formalism in which parameter error distributions can be mapped independently of time, while the time dependence can be computed indepen- dently of the specific SMBH binary orbital elements. A Monte Carlo simulation for random binary orientations then becomes a simple linear combination, without any integral evaluations. This greatly reduces the numerical cost of estimating param- eter uncertainty distributions, even at fixed observation time (e.g. to map distributions of errors at ISCO). We use this nu- merical cost advantage 1. to map the distribution of localization errors for the full three dimensional grid of SMBH total mass (M = 105 -- 108M⊙), redshift (z = 0.1 -- 7) and arbitrary look -- back time (tf) before merger, 2. to study how source localization error distributions vary systematically with sky position, and 3. to discuss implications, in terms of advance warning times, for prompt electromagnetic counterpart searches with large field-of-view astronomical instruments. We call this new approach the harmonic mode decomposi- tion (HMD). The method verifies that the amplitude modula- tion, which is restricted to frequencies less than 4 f⊕ = 1.3× 10- 7 Hz, is indeed a very slow modulation when compared to the GW frequency of LISA SMBH inspirals (0.03 mHz -- 1 Hz). One plausibly expects that physical parameters which 2 determine the amplitude modulation (like the source sky po- sition and orbital inclination relative to the detector) can be estimated independently of the parameters which determine the GW frequency (like masses, orbital phase, time to ISCO). In the HMD method, the two sets of parameters are naturally separated and can be estimated independently. In particular, parameters related to the modulation can essentially be deter- mined on a background of GW-cycle averaged signal. In the present work, we compute LISA inspiral localization errors with the approximation that high frequency signal parameters have strictly no cross-correlations with parameters related to the slow orbital modulation. In addition to the numerical ad- vantages mentioned above, the HMD formalism offers a clear interpretation of the time evolution of uncertainties for the slow modulation parameters. This can be used to gain a better understanding of the general evolutionary properties of local- ization errors. The following questions, that we address in detail in our work, are particularly relevant. (i) Under what conditions do the localization uncertainties scale simply with the measured signal -- to -- noise ratio, and how do these uncertainties evolve during the final stages of inspiral? (ii) To what extent do the high and low frequency signal parameters decouple? (iii) What are the best determined combinations of the an- gular parameters? (iv) How and why does the shape of the 3D localization er- ror ellipsoid change during the final week(s) of obser- vation? In our analysis, we neglect the "Doppler phase" due to LISA's orbital motion, SMBH spin precession effects and any finite SMBH binary orbital eccentricities. These approxima- tions are advantageous for the resulting simplicity, but the use of the HMD method is not restricted to these approxi- mations. We also outline a generalized HMD method which remains numerically much more efficient than standard meth- ods. We leave a numerical implementation of this general HMD method to future work. It will be particularly interest- ing to determine how our approximate results for the evolution of LISA localization errors are modified when spin precession effects are included, since spin precession effects were shown to improve the final localization errors by factors of 3 -- 5 at ISCO [17, 22]. The remainder of this paper is organized as follows. In § II we define our conventions and the assumptions made in our analysis. In § III we expand the LISA GW signal in Fourier modes and obtain the conversion from actual physical parame- ters to corresponding Fourier amplitudes. In § IV, we incorpo- rate these results into a Fisher matrix formalism and derive the expressions necessary to estimate correlation errors for HMD signals. In § V, we quantify the computational advantages of the HMD method. In § VI, we present results from Monte Carlo computations of the time evolution of localization er- rors and discuss results in terms of advance warning times for prompt electromagnetic counterpart searches. In § VII, we 3 - 5/8 - 3/8 0.25 M η 6z , (1) = 2.7× 10- 4 Hz(cid:18) t day(cid:19) - 3/8 develop toy models to interpret the time-dependence of LISA localization errors and to answer questions (i) -- (iv) above. We summarize our results and conclude in § VIII. merger is (e.g. eq. 3.3 in ref. [25]) f0(Mz,t) = 53/8 8π t- 3/8M - 5/8 z , II. ASSUMPTIONS AND CONVENTIONS This section is divided into three parts. First, we list the definitions of physical quantities used in this paper, in partic- ular the variables describing a SMBH inspiral. Second, we give the equations which determine the LISA inspiral signal. Third, we state all the assumptions made in this work. A. Definitions In general, an SMBH inspiral is described by a total of 17 parameters. These include 2 redshifted mass parameters, (Mz, ηz), 6 parameters related to the BH spin vectors, pspin, the orbital eccentricity, e, the source luminosity distance, dL, 2 angles locating the source in the sky, (θN, φN), 2 angles that describe the relative orientation of the binary orbit, (θNL, φNL), a reference time, tmerger, and a reference phase at ISCO, φISCO, and the orbital phase, φorb. Throughout this work, we restrict ourselves to circular orbits by omitting the orbital eccentricity, e, and instead of the orbital phase, φorb, we use the look -- back time before merger, t, as our evolutionary time parameter. The LISA signal for a GW inspiral is determined by the above set of parameters and two additional angular parameters describ- ing the orientation of LISA, (Ξ, Φ). We elaborate on the defi- nitions of our mass and angular parameters below. 1. Mass Parameters For component masses m1 and m2, the total mass is M = m1 + m2, the reduced mass is µ = m1m2/M, the symmet- ric mass ratio is η = µ/M and the chirp mass is defined as M = Mη3/5 [24]. Throughout this work, we use geometrical units: G ≡ c ≡ 1. In this case, the mass can be expressed in units of time: 106M⊙ ≡ 4.95 sec. The measured GW wave- forms are insensitive to the cosmological parameters, if they are expressed in terms of the luminosity distance and the red- shifted mass parameters, e.g. mz = (1 + z)m (same for red- shifted chirp and reduced masses). 2. Time Parameters We write a generic look -- back time (or "observation time") before merger as t, and a generic redshifted GW frequency (or "observation frequency") as f [37]. We use the leading order (i.e. Newtonian) approximation for the frequency evolution. Therefore, the observed frequency at look -- back time t before or equivalently t0(Mz, f ) = 5(8π f )- 8/3M = 6.7 min(cid:18) f fc(cid:19) - 5/3 z - 8/3 - 5/3 - 1 0.25M η 6z , (2) where M6z is the redshifted total mass in units of 4× 106M⊙, η0.25 = η/0.25 is the symmetric mass ratio (η0.25 = 1 for equal component masses, § II C), fc = c/R⊕ = c/(1 AU) = 2.00 mHz is the inverse light-travel time across the radius of the LISA orbit, and the null index stands for the order of approximation. The inspiral phase extends until the innermost stable circular orbit (ISCO), at 6M, is reached - 1M- 1 z = 1.1 mHz× M- 1 6z , - 1Mz = 33 min× η - 1 0.25M6z. f ≤ fISCO = 6- 3/2π t ≥ tISCO = 5(3/2)4η (3) (4) where tISCO is the (observer-frame) look -- back time before merger corresponding to the ISCO, and fISCO is the (observer- frame) frequency at ISCO. In the present work, we fix the start of the observation (i.e. when the source first enters LISA's frequency band) at look -- back time ti, and examine how the value of an end-of- observation time, tf, prior to merger affects the precision on source localization. We restrict ourselves to pre-ISCO inspi- ral signals, corresponding to tf ≥ tISCO. Note that any instanta- neous look -- back time t associated with an observation lasting from look -- back times ti to tf must obey tISCO ≤ tf ≤ t ≤ ti in our notation. 3. Angular Parameters LISA is an equilateral triangle-shaped interferometer with an arm-length of 5 × 106 km, orbiting around the Sun. The constellation trails 20◦ behind the Earth and is tilted 60◦ rel- ative to the ecliptic. The detector plane precesses around the orbital axis with the same one-year period as the orbital period [26]. Following closely Refs. [19] and [17], including in nota- tion, we define two coordinate systems. The barycentric frame is tied to the ecliptic, with x, y lying in the ecliptic plane and z normal to it. The detector reference frame tied to the de- tector, with z′ normal to the detector plane, while x′, y′ are in the plane and co-rotating with the detector so that the arms are described by time-independent vectors. We refer to the barycentric frame with normal coordinates and to the detector frame with primed coordinates. The unit vectors defining the source location on the sky, N, and the SMBH binary orbital angular momentum, L, are described by polar angles (θN, φN) and (θL, φL) in the ecliptic frame, (θ′N, φ′N) and (θ′L, φ′L) in the detector frame: N(θN, φN) = zcos θN + xsin θN cos φN + ysin θN sin φN, L(θL, φL) = zcos θL + xsin θL cos φL + ysin θL sin φL. (5) (6) Since we assume no SMBH spins, orbital angular momen- tum is conserved and the (θN, φN, θL, φL) coordinates are time- independent properties of the sources. Let (Ξ, Φ) be the two angles specifying the orientation of the LISA system in the ecliptic: Φ describes its orbital phase during its motion around the Sun, while Ξ describes the ro- tation of the triangle around its geometrical center. If their values at merger are written Ξ0 and Φ0, then at an arbitrary look -- back time t: 4 The explicit definitions are given in Appendix B. Let us refer to the angles at the reference time t = tmerger = 0 as Ω(0). Although Φ ≡ Φ(t) and Ξ ≡ Ξ(t) are time-dependent, as given by (7,8) α is a time-independent combination, unlike the time-dependent γ ≡ γ(t). The angles at t = 0 are thus given by Ω(0) = (θN, φN, θNL, φNL, α, γ0). These angles have the interesting property that they pos- sess isotropic a priori distributions, like the original Ω(0) vari- ables, but the measured GW waveforms expressed in terms of these new variables are much simpler than when they are ex- pressed in terms of the original set eqs. (5 -- 12). Two additional quantities which are useful to describe the sensitivity of the detector in various directions are the antenna beam patterns [19]: Ξ(t) = Ξ0 - ω⊕t, Φ(t) = Φ0 - ω⊕t, (7) (8) F×,+(Ω) = 1 + cos2 θ′N 2 cos2φ′N cos2ψ′N ± cosθ′N sin 2φ′N sin 2ψ′N, (14) where ω⊕ ≡ 2π/ yr is the orbital angular velocity around the Sun. The time dependence of the detector normal vector z′ can be expressed as z - √3 2 1 2 x cos Φ - √3 2 z′ = y sin Φ. (9) The detector angles are given by cos θ′N = 1 2 cos θN - √3 φ′N = Ξ + tan- 1" √3 sin θN cos(Φ - φN), 2 2 cos θN + 1 2 sin θN sin(Φ - φN) sin θN sin(Φ - φN) (10) (11) # . where the sign ± is defined to be positive for F×, and negative for F+. Equation (14) and the transformation rules eqs. (5 -- 12) define the time evolution of the antenna beam patterns for a given set of final angles Ω(0) as the LISA system orbits around the Sun. Note that the LISA system is equivalent to two inde- pendent orthogonal-arm interferometers which are rotated by 45◦ relative to each other [19]. Both data-streams are given by the same equations (see eq. [21] below), modulo a change - π/4 of one of the angles for the second detector: φ′II (or equivalently αII = αI - π/4 using our time-independent an- gular variables). Thanks to this simple relationship between the two data-streams, it is possible to carry out all the calcu- lations for the first data-stream, and later include the second data-stream in the final expression by varying the fiducial an- gle α. N = φ′I N Let us also define ψ′, the polarization angle of the GW wave- form, as [17] tan ψ′ = L· z′ - ( L· N)(z′ · N) N· ( L× z′) . (12) Note that there are only 6 independent angular parameters (θN, φN, θL, φL, Ξ, Φ). Other detector specific quantities like θ′N, φ′N, θ′L, φ′L, and ψ′ can be expressed in terms of these 6 independent parameters using eqs. (5 -- 12). Let us introduce a new set of 6 independent angles, Ω = (θN, φN, θNL, φNL, α, γ), (13) with the following definitions: • θNL is the relative latitude of L and N (i.e. the inclina- tion of the binary orbit to the line of sight), • φNL is the relative longitude of L and N, • α ≡ Ξ - Φ + φN - • γ(t) ≡ Φ(t) - φN. 3π 4 , 4. Grouping the Parameters We group the most important parameters describing the in- spiral as follows: pslow ≡ {dL, Ω}, pfast ≡ {Mz, µz,tmerger, φISCO}, pspin ≡ {2 spin magnitudes,4 spin angles}. (15) (16) (17) This organization of parameters has fundamental importance in our formalism. As we show in § II B, the parameters pfast and pspin relate to the high frequency GW signal, while the pa- rameters pslow relate to the distinctly slow orbital modulation. B. LISA Inspiral Signal Waveform For a circular binary inspiral, the two polarizations of GW signal are well approximated by the restricted post-Newtonian expressions h+(t) = 2M5/3(π f )2/3 h×(t) = - 4M5/3(π f )2/3 dL dL (1 + cos2 θNL) cos φGW(t), cos θNL sin φGW(t). (18) (19) The GW phase φGW(t) ≡ φGW(pfast, pspin;t), which is twice the orbital phase , φ(t) = 2φorb(t), can be expanded into the series φGW(pfast, pspin;t) ≈φISCO + φ0(Mz;t) + φ1(Mz, µz;t) + φ2(Mz, µz, pspin;t) + . . ., (20) where φ0(Mz;t) is the leading order Newtonian solution to the phase evolution, successive terms correspond to small general relativistic corrections, φISCO is the reference phase at ISCO and φn(tISCO) = 0 for all n ≥ 0. The instantaneous GW fre- quency is defined as the time derivative of the GW phase (20), i.e. f = f (t) ≡ dφGW/dt, which changes very slowly compared to the GW phase itself, φGW(t). In practice we use the New- tonian approximation (1), f0(t) = dφ0/dt. Note that equation (20) depends implicitly on the reference time, tmerger, since our time variable t is the look -- back time before tmerger (see § II A 2) The signal measured by LISA is a linear combination of the two polarizations (18), weighted by the antenna beam patterns F I,II for each of the two equivalent interferometers, + defined by (14), resulting in the two observable data -- streams and FI,II × √3 2 [F I,II hI,II(t) = + h+(t) + FI,II × h×(t)], (21) where the factor √3/2 = sin(60◦) comes from the opening an- gle of the LISA arms. The beam patterns are determined by the relative orientation of the source polarizations and the de- tector. Their time-dependence is due to the following three main effects: LISA changes its orientation as it orbits the Sun, LISA changes its relative distance to the source as it orbits the Sun, and the orbital plane of the SMBH binary can precess be- cause of spin-orbit coupling effects. Substituting (18) in (21) and expressing it in complex form, we get hI,II(t) = A(Mz, f ) dL GI,II(Ω, f )eiφGW(pfast,pspin;t), (22) where A(Mz, f )/dL defines the overall amplitude scale, with (23) A(Mz, f ) = 2√3(π f )2/3M5/3 . z The G(Ω, f ) factor defines the angular dependence of the sig- nal, GI,II(Ω, f ) = GI,II A (Ω)eiϕD(Ω, f ), (24) where GA(Ω), the amplitude modulation, captures the vary- ing detector sensitivity with direction and polarizations of the GWs, GI,II A (Ω) = 1 + cos2 θNL 2 + (Ω) - F I,II icosθNLF I,II × (Ω). (25) 5 The additional ϕD(Ω, f ) modulation is the Doppler phase modulation, which is the difference between the phase of the wavefront at the detector and at the barycenter [19]: ϕD(Ω, f ) = 2π f fc sin θN cos γ. (26) There is a non-negligible number of Doppler phase cycles only for a GW frequency satisfying f ≥ fc (see definition of fc below eq. [2] above). However, equation (3) shows that f ≤ fISCO < fc, hence the fc frequency is reached only after ISCO for typical SMBH component masses of m1 = m2 = 106M⊙ and redshift z = 1. Even for smaller 105M⊙ component masses, the total number of cycles, Npm, remains < 1 until the final 5 hr of inspiral. Therefore the Doppler phase (26) is practically negligible for SMBH inspirals. In fact, estimating localization errors without accounting for the Doppler phase affects results by less than a factor of 10- 3 (for m1 = m2 = 106M⊙ at z = 1; S. A. Hughes, private communica- tion). Therefore, in eq. (24), we neglect ϕD(Ω, f ) and restrict ourselves to the approximation GI,II(Ω, f ) ≡ GI,II A (Ω). (27) The explicit frequency-dependence dropped out, and the time evolution of the signal GA is now fully determined by the time evolution of the angles Ω. Note that the amplitude modulation (25), GI,II A (Ω), is tra- ditionally expressed in complex polar notation (e.g. [19]), where the magnitude and argument of the complex number are called polarization amplitude and phase. As we will show, the mode decomposition is simplest in the original Cartesian complex form (25), which already includes both the polariza- tion amplitude and phase; thus, we do not distinguish these two quantities in the following. The function GI,II(Ω, f ) given in (24) also accounts for spin-orbit precession if the orbital orientation (θNL, φNL) in Ω is chosen to be time-dependent, to satisfy the equations for spin-orbit precession, and if an extra precession phase shift, exp(iδP(θNL, φNL)), is introduced (see eq. 2.14 in Lang & Hughes [22]) in addition to the Doppler phase in (24). In our calculations, we neglect spin precession but discuss how the HMD method can be extended to include that effect in § IV C. Finally, we express the measured signal (22) as hI,II(p;t) = hc(pfast, pspin;t)× hI,II m (pslow;t), (28) where hc is the high frequency carrier signal and hm is the slow modulation: hc(pfast, pspin;t) = A(Mz, f (t))eiφGW(pfast,pspin;t) hI,II m (pslow;t) = dL GI,II A (Ω(t)) . (29) (30) Equation (28) shows that the two sets of parameters pslow and {pfast, pspin} are exclusively determined by the low fre- quency modulation and the high frequency carrier, respec- tively. For this reason, we only expect a low level of cross- correlation between these sets of parameters: parameters asso- ciated with very different timescale components should essen- tially decouple. In Sec. VII A and Appendix A, we consider several toy models which allow us to understand the neces- sary conditions, and the extent to which, parameters associ- ated with high and low frequency components decorrelate in the course of an extended, continuous observation. C. Simplifying Assumptions In the present work, we make the following assumptions: 1. We assume that the amplitude modulation can be used to determine the luminosity distance and angular pa- rameters, pslow = {dL, θN, φN, θNL, φNL}, while the other parameters, pfast = {Mz, µz,tmerger, φISCO}, are deter- mined from the high frequency GW phase. We assume no cross-correlations between these two sets of param- eters. This is supported by the results listed in Table 1 of Hughes [7], which shows the full covariance matrix of a Monte Carlo realization of 2PN waveforms. The correlation coefficients are ∼ 0.1 for the above quanti- ties, and the absolute scale of the second set of param- eters is very low in the first place. Berti et al. [12, 13] also report that the sets pfast and pslow are relatively un- correlated for general relativity and even for alternative theories of gravity. In the latter case, the carrier hc(t) in the signal (28) is modified but not the slow modulation, hm(t), so that the general expectation of decoupling is maintained. 2. We assume that there are no additional errors on the detector orientations Φ(0) and Ξ(0). These parame- ters are given by tmerger via eq. (8) and (7), and tmerger itself is determined by the high frequency carrier sig- nal to high precision. Using the full data-stream up to ISCO, δtmerger ∼ 2 sec [5, 7]. Using (8) and (7), we es- timate δΦ(0) = δΞ(0) ≡ ω⊕δtmerger = 4 × 10- 7 rad = 0.08′′. This is so small that we expect the errors δΦ(0) and δΞ(0) to be negligible at any relevant end-of- observation times tf > tISCO, even if the tf-dependence of these errors scale as steeply as (S/N)- 1 (see also Ap- pendix A). 3. We use the circular, restricted post-Newtonian (PN) ap- proximation for the GW waveform, keeping only the leading order (i.e. Newtonian) term in the signal am- plitude. Higher order corrections to the GW amplitude introduce additional structure to the waveform. They improve the parameter estimation uncertainties for high mass binaries [27, 28] and introduce additional corre- lations between the parameters. It will be important to consider these corrections to the amplitude in future in- vestigations. Arbitrary PN corrections to the GW phase only enter via hc in the signal given by eq. (28). Since we neglect correlations between the sets of parameters pslow and (pfast, pspin), all the restricted PN corrections to the phase drop out and become irrelevant for the pslow parameter estimations. 4. We neglect the effects of Doppler phase modulation. This is plausible for SMBH binaries with component 6 masses m > 105M⊙, since the GW wavelength in this case is generally greater than LISA's orbital diameter and Npm < 1 (see eqs. [26] and [1]). 5. We neglect SMBH spins and, in particular, neglect the spin -- orbit precession for angular determinations. This assumption is useful in simplifying our equations and in focusing on the behavior of pure angular modulation. Future studies can incorporate spin -- orbit precession by convolving the angular modulation decomposition with spin -- orbit effects. 6. We fix the start of LISA observations at a look -- back time ti ≡ min{t0( fmin),1 yr} prior to merger. This cor- responds to the time when the GW inspiral frequency f crosses the low frequency noise wall of the detector at fmin = 0.03 mHz, but we limit the initial look -- back time to a maximum of 1 yr before merger. Note that LISA's effective mission lifetime is estimated to be 3 yr. Inte- grated observation times longer (but also shorter) than our assumed ti values are possible in principle, depend- ing on source specifics. In a more elaborate treatment, one could define ti as an a priori random variable. We fix ti here mostly for simplicity and focus on the effects of varying the values of tf (< ti). In § VII A we show that localization errors are primarily determined by the end-of-observation time, tf, and that values of ti > 1 yr do not significantly change the evolution or final local- ization error estimates. If, however, ti ≪ 1 yr (that is, if tmerger is within a few months of the beginning of ob- servation), then localization errors can become signifi- cantly worse than in our results with ti = 1 yr. 7. We neglect finite arm-length effects and we do not make use of the three independent observables of the time de- lay interferometry [29]. This is a valid assumption for SMBH inspirals since here f ≪ c/L = 0.01 Hz. 8. We neglect any finite orbital eccentricities. We note that, for eccentric orbits, higher order harmonics ap- pear in the GW phase. In principle, since these har- monics affect the high frequency GW phase, but not the slow modulation, including finite eccentricities should not significantly affect localization error estimates. For rather eccentric orbits, high-order harmonics with f ≫ fc can have a non-negligible Doppler phase (2), which would lead to an improvement in the determination of θN and φN. Although eccentricity is efficiently damped by gravitational radiation reaction [30], the presence of gaseous circumbinary disks could lead to non-zero eccentricities for at least some LISA inspiral events [31, 32]. 9. We follow Barack & Cutler [21] in calculating the LISA root spectral noise density, Sn( f ), which includes the in- strumental noise as well as galactic/extra-galactic back- grounds. For the instrumental noise [13], we use the ef- fective non-angularly averaged online LISA Sensitivity Curve Generator[38], while we use the isotropic formu- lae for the galactic and extra-galactic background [21]. 10. Our analysis focuses on statistical errors and does not account for possible systematic errors. For example, waveform templates might be inaccurate either due to the imprecision of the theory if the true waveform is not the one predicted by general relativity, or due to practi- cal limitations from necessarily finite realizations of the large template space. Such inaccuracies can introduce new systematic errors. III. HARMONIC MODE DECOMPOSITION In our formalism, the angular information of the LISA in- spiral signal is contained exclusively in the periodic modula- tion due to the detector motion around the Sun, which adds an amplitude modulation to the high frequency waveform. This modulation has a fundamental frequency, f⊕ = 1/ yr, along with upper harmonics j f⊕, where j is an integer. Although it is intuitively clear that the high frequency harmonics will tend to have a vanishing contribution, it is hard to establish this just by looking at eqs. (5 -- 12), which define the time evolution. In this section we show that it is possible to derive surprisingly simple analytical expressions for the amplitude of each har- monic. We provide an outline of the derivation starting from the commonly used Cutler [19] formulae (5 -- 12) and alterna- tively from those in Cornish & Rubbo [23]. We show that the derivation is much simpler in the latter case, in the sense that the Cornish & Rubbo [23] expression is almost already in the desired form. A. Derivation using Cutler [19] We expand the modulating signal (25,30) in a Fourier series hm(pslow(0);t) = GA(Ω(t)) dL(z) = ∞ Xj=- ∞ g j(pslow(0))ei jω⊕t, (31) where g j(pslow(0)) are the mode amplitude coefficients and pslow(0) are the distance and angle variables at t = 0 (see § II A). The coefficients can be obtained as g j(pslow(0)) = 1 2πdLZ 1 yr 0 dt GA(Ω(t))e- i jω⊕t. (32) Substituting the definition of GA(Ω(t)) from eq. (25), using the time evolution of Ω(t), eqs. (5 -- 12), integral (32) can be evaluated. Although conceptually simple, a direct analytical evalua- tion of integral (32) is overly cumbersome. Thus, for practical reasons, we follow an alternative path. We start with the orig- inal Cutler [19] formulae, given by eqs. (14) and (25). First, using general trigonometric identities, we can express cos2x and sin 2x with tan(x) for x = φ′N and x = ψ′. In the second step, we express and substitute for tan φ′N and tan ψ′N with ecliptic variables using (11) and (12). In the third step, we express the trigonometric functions in complex form. After this step, each term in the beam pattern (14) is of the form 7 (33) Pn aneinγ Pm bmeimγ , where the sums over n and m integers are finite, containing only a few terms, and an and bn depend only on the angles (θN, φNL, α). In the fourth step we simplify the product of frac- tions. It turns out that, after combining terms, the denomina- tors drop out exactly, leaving a formula just like (31), except that the largest element in the sum is j = 4. In the fifth step, we substitute in (25), and finally, change back from complex to trigonometric notation for the coefficients, using the half- angles θN/2 and θNL/2. Finally we arrive to the remarkably simple form: 4 Xj=0 hm(pslow) = dL(z)- 1 (LN jD j + L∗N∗j D j + L∗N jD∗j + LN∗j D∗j ), (34) where the functions L, N, and D depend only on the angular momentum angles, sky position angles, and detector angles, respectively: L(θNL, φNL) ≡ sin4(cid:18) θNL N j(θN) ≡ w j cos j(cid:18) θN D j(α, γ) ≡ ie2iαei jγ, 2 (cid:19)e- 2iφNL, 2 (cid:19)sin4- j(cid:18) θN 2 (cid:19) , (35) (36) (37) where w j = 1/16× (9,12√3,18,4√3,1) for j = (0,1,2,3,4), respectively, and we have defined asterisks to refer to the fol- lowing conjugates: L∗(θNL, φNL) ≡ L(π - θNL,- φNL), N∗j (θN) ≡ (- 1) jN j(π - θN), D∗j (α, γ) ≡ - D j(- α,- γ) ≡ D j(α, γ). (38) (39) (40) Note that using these conjugate functions, only the non -- negative terms 0 ≤ j ≤ 4 remain in the sum (34). Substituting the time dependence implicit in γ ≡ γ(0)+ω⊕t, equation (34) becomes hm(pslow(0),t) = dL(z)- 1 4 Xj=- 4 where the coefficients are g j(pslow(0))ei jω⊕t, (41) g j(pslow(0)) =(cid:26) LN jD j(0) + L∗N∗j D j(0) (0) + LN∗ j L∗N jD∗ j D∗ j if j ≥ 0 (0) if j ≤ 0 (42) and the detector functions D j(0) and D∗j (0) refer to their (Note that L,N j,L∗,N∗j are all time- values at t = 0, γ(0). independent.) Since the decomposition (31) is unique, the coefficients (42) that we read off from our result also satisfy eq. (32). B. Derivation using Cornish & Rubbo [23] Our result in (34) can also be derived from the Cornish & Rubbo [23] formulae for the LISA response function. In their paper, these authors use a different set of angles, which relate to ours as follows: θCR = θN, φCR = φN, ψCR ≡ φNL - (π/2), (3π/4), and αCR ≡ γ + φN ≡ Φ. Note λCR ≡ - α+ φN ≡ Φ- Ξ- that our set of angles is very similar to theirs, except that we measure the detector angles relative to the source, φN. This is advantageous given the rotational symmetry around the Earth orbital axis, making angles relative to the source the only ones that should have an effect on the measured GWs; we expect φN to drop out of the equations when using α and γ. Note, once again, that the variables (θN, φN, θNL, φNL, α) are time in- dependent, while γ ≡ γ(t). Writing the Cornish & Rubbo [23] beam patterns for low frequencies, which is fully equivalent to Cutler [19], with our angular variables[39], we get F I,II + F I,II × = - 1 2 = - 1 2 where [cos(2φNL)DI,II [sin(2φNL)DI,II + (t) - sin(2φNL)DI,II × + (t) + cos(2φNL)DI,II × (t)], (43) (t)], (44) D+ = 1 32 {- 36 sin2 θN sin(2γ + 2α) + (3 + cos2θN) × {cos(2φN)[9 sin(- 2α + 2φN) - sin(4γ + 2α + 2φN)] + sin(2φN)[cos(4γ + 2α - 2φN) - 9 cos(- 2α + 2φN)]} 4√3 sin(2θN)[sin(3γ + 2α) - 3 sin(γ + 2α)]}, (45) and D× = 1 8√3 { √3 cos θN[9 cos(- 2α) - cos(4γ + 4α)] 6 sin θN[cos(3γ + 2α) + 3 cos(γ + 2α)]}. (46) One notices instantly that the time dependence here is much simpler than in the original Cutler [19] formula, as it is in- scribed only in the various harmonics of γ. We can identify the highest harmonic present to be 4γ. Expanding the trigono- metric functions using standard identities, we obtain D+ = - 1 32 and D× = - 1 8 ·     · sin(2α) sin(2α + γ) sin(2α + 2γ) sin(2α + 3γ) sin(2α + 4γ)   cos(2α) cos(2α + γ) cos(2α + 2γ) cos(2α + 3γ) cos(2α + 4γ)     , , (47) (48) 9(3 + cos2θN) - 12√3sin 2θN 36 sin2 θN 4√3 sin 2θN 3 + cos2θN - 9 cosθN 6√3 sin θN 2√3 sin θN cos θN 0       where a· b =Pn anbn is the usual dot product. Now, the sec- ond sets of elements carry the time dependence and the de- tector orientation information, while the first sets describe the 8 sky position. Note that the explicit φN dependence dropped out, as expected. Next, we manipulate equations (47,48), sub- stituting complex expressions for the trigonometric ones, and substituting these into eq. (25). We finally arrive at eq. (34) after changing to half -- angles θN/2 and θNL/2. We note that eqs. (34) or (41,42) are fully general ex- pressions, equivalent to the standard LISA inspiral signal in eqns. (14) and (25). The two data-streams are obtained by substituting α = αI and αII, corresponding to the two indepen- dent LISA-equivalent Michelson interferometers (see § II A). To verify our final result, we compare numerically the signals computed using eqs. (14,25) with the signals computed us- ing eqs. (41,42), for a large set of random choices of angles. Agreement is achieved at machine precision levels. The main utility of eq. (34), is that it can be used to "de- construct" parameter error histograms, i.e. to understand how the errors depend on the fiducial values of the parameters. As compared to Cornish & Rubbo [23], our result leads to an ex- plicit decoupling of the signal angular dependence into simple products of one-dimensional functions. In particular, the de- pendence on sky position, angular momentum, and detector angles are separated. Using the special conjugate functions L∗, D∗, N∗, eq. (34) displays the symmetry properties of the signal. Finally, one angular variable, φN is eliminated exactly. IV. ESTIMATING PARAMETER UNCERTAINTIES IN THE HMD FORMALISM Parameter estimations for LISA GW inspiral signals are possible with matched filtering and the expected uncertainties can be forecast using the Fisher matrix formalism [33, 34]. In this section, we apply this approach to the LISA signal derived in § III, with an angular dependence of the signal decomposed into harmonic modes. In § IV B, we consider the simple case of a high frequency carrier signal that is modulated by a low- frequency function, without any cross-correlation between the two sets of relevant parameters. We derive a simple formula for the estimation of modulating parameter uncertainties. In § IV C, we consider a more general post-Newtonian signal and show that parameters related to source localization can still be decoupled from the time evolution and the other source pa- rameters. A. Fisher Matrix Formalism Let us consider a generic real signal described by the func- tion h(x), which depends on N parameters {pa}a∈[1,N]. The measured signal is y(x) = h(x) + n(x), where n(x) is a realiza- tion of the noise specified by a probability distribution. Let us assume that the noise is Gaussian, is statistically station- ary with respect to x, has zero mean value, hn(x)i = 0, where hi represents an ensemble average, and has known variance, σ(x)2 = hn(x)2i. The parameter estimation errors for pa can then be calculated using the Cramer-Rao bound [33] hδ pa δ pbi ≥ hΓ - 1ia b, (49) - - where equality is approached for high S/N signals. Here Γa b is the Fisher matrix defined by Γa b =Z xmax xmin ∂ah(x) ∂bh(x) σ2(x) dx, (50) where ∂a is the partial derivative with respect to the parame- ter pa. Note that σ(x) here is defined as the noise per unit x. In eq. (50), x denotes time t for time-domain samples, or f for frequency-domain samples. The purpose of this work is to study how an arbitrary end of the observation, at xmax (or tf be- low, for time samples) affects the resultant correlation errors hδ pa δ pbi, for a fixed start-of-observation at xmin (or ti below, for time samples). An important quantity for the evolution of parameter esti- mation errors is the signal-to-noise ratio, S/N, defined by N(cid:19)2 (cid:18) S =Z xmax xmin h2(x) σ2(x) dx. (51) For LISA, the noise varies with signal frequency. In this case, the Fisher matrix can be evaluated in Fourier space [33, 34], Γ(tf)a b = ℜ(4Z f (tf) fmin ∂a h( f ) h( f ) ∂b S2 n( f ) d f) , (52) 9 We are only interested in estimating uncertainties for the pslow variables (§ II A), which are determined exclusively by hm(t). Recall from eq. (29) that hc(t) = A so that, for the Fourier transform[40], we have hc(t)2 = 4 hc( f )2(d f /dt). Using these relationships, let us define the instantaneous rela- tive noise amplitude per unit time σ(t) as - 2(t) = 4 σ 2 hc S2 3√5 4 d f0(t) [ f0(t)] n[ f0(t)] (cid:12)(cid:12)(cid:12)(cid:12) dt t- 1/2 M S2 n[ f0(Mz,t)] 5/2 z = = A2[ f0(t)] S2 n[ f0(t)] (56) (cid:12)(cid:12)(cid:12)(cid:12) . The last equality follows from the Newtonian waveform and frequency evolution, eqs. (23) and (1). We point out that the mass dependence is captured entirely by σ(t) and does not appear anywhere else in what follows. By combining eqs. (54), (55), and (56), we arrive at Γ(tf)a b =Z ti tf ℜ[∂ahm(t) ∂bhm(t)] σ2(t) dt. (57) Equation (57) is the special case of (52), where the carrier signal-to-noise ratio and modulation, hm, are conveniently iso- lated. We are now ready to make use of the harmonic mode de- composition. Substituting eq. (31) into (57) gives Γ(tf)a b = ℜ  4 Xj1, j2=- 4 ∂ag j1 ∂bg j2Pj2- , j1(tf)  Pj2- j1(tf) = ei( j2- j1)ω⊕t σ2(t) dt. ti Ztf (58) (59) The function Pj(tf) is shown in Figure 1 for j = 0, to- gether with real and imaginary parts for the j = 4 case, for m1 = m2 = 106M⊙ at z = 1. Since the accumulated signal-to- noise ratio is S/N = P0(t), the figure shows that the instanta- neous signal-to-noise ratio is [d/dt](S/N) = [d/dt]P0(t) ≈ t- 2. The extrapolated signal-to-noise blows up at "merger" (t = 0). Data analysis for such a non-stationary signal-to-noise ratio evolution has several interesting implications, which we study further with toy models in Appendix A. We find that, for such a signal-to-noise ratio evolution, specific combinations of pa- rameters can always be measured to very high accuracy. nation j = j2 - on ( j1, j2) and evaluate one of them independent of time: The time dependence in eq. (59) couples only to the combi- j1. This allows us to rearrange the double sum where Γ(pslow,tf)a b = ℜ  Xj′=- 8 8 [F j(pslow(0))]a b = 8 [F j(pslow(0))]a bPj(tf) Xj=- 8  , (60) ∂ag j+ j′(pslow(0)) ∂bg j(pslow(0)). (61) where h( f ) is the Fourier transform of h(t), the GW signal (28), ∂a is the partial derivative with respect to parameter pa, bars denote complex conjugation, and S2 n( f ) is the one-sided spectral noise density (§ II A). where B. Approximate solution We seek an alternative equivalent form of eq. (52) specific to GW inspirals for which, as in eq. (28), the high frequency carrier signal is decoupled from the slow modulation. In case of SMBH inspirals, with a high frequency signal hc(t) chang- ing its frequency slowly as f0(t) given in eq. (1), and fur- ther modulated by a slowly varying function hm(t) as given in eq. (28), the integral in eq. (52) can be evaluated in the sta- tionary phase approximation, by substituting h( f ) = hm[t0( f )]× hc( f ), (53) where hc( f ) is the Fourier transform of the carrier signal and hm[t0( f )] is the modulating function evaluated at the time when the carrier frequency is f . This can be converted to the time domain, by simply changing the integration variable to t = t0( f ) using the frequency evolution in eq. (2): Γ(tf)a b = ℜ(4Z ti tf and ∂a h(t) h(t) ∂b S2 n[ f0(t)] d f0(t) dt (cid:12)(cid:12)(cid:12)(cid:12) dt), (cid:12)(cid:12)(cid:12)(cid:12) h(t) = hm(t)× hc[ f0(t)]. (54) (55) P0(tf) ℜP4(tf) ℑP4(tf ) 107 106 105 104 103 102 ) f t ( j P m1 = m2 = 106M⊙ z = 1 ti = 340 days 101 0.01 0.1 1 tf [day] 10 100 FIG. 1: The time evolution of the fundamental functions Pj(tf), used to construct the Fisher matrix to forecast localization errors by LISA. The dependence of the Fisher matrix on the look -- back time tf is ob- tained from the 9 fundamental functions with 0 ≤ j ≤ 8. The curves show P0(tf), as well as the real and imaginary parts of P4(tf), for m1 = m2 = 106M⊙ and z = 1. Thin dotted lines represent negative values. Note that P0(tf) ≡ (S/N)2, which is the simple scaling of inverse squared errors, neglecting correlations. The signal-to-noise ratio scales steeply, approximately as S/N ∝ t- 1 . The curve P4(tf) il- lustrates how all the other similar Pj(tf) functions vary, with a relative number j of oscillations, and P- j(tf) ≡ Pj(tf). f Our parameters in the correlation matrix are pa = (dL, θN, φN , θNL, φNL) since we assume that the other param- eters, i.e. {Mz, ηz, φISCO,tmerger, α, γ(0)}, are known from the high frequency carrier signal (§ II C). It is straightforward to compute the derivatives of g j(dL, Ω) using eq. (42) for all pa- rameters pa, except φN. The φN dependence in g j in eq. (42) is implicit in α ≡ α(Ξ(0), Φ(0), φN) and γ(0) ≡ γ(Φ(0), φN) (see § II A). Since we assume that Ξ(0) and Φ(0) are mea- sured to very high precision with the high frequency carrier (§ II C), we can use the chain rule to express the φN derivative as ∂φN g j = ∂αg j - ∂γ0g j. Up to this point we did not make use of the fact that the LISA signal is equivalent to two orthogonal arm interferom- eters rotated by ∆α = π/4 with respect to each other. To ac- count for both data-streams being measured simultaneously, the Fisher matrix is written as the sum of the two Fisher matri- ces corresponding to each individual interferometer. Writing a b(α) = Γa b(α)+ Γa b(α- out only the α dependence, we have Γtot π/4). Finally, according to eq. (49), the parameter error co- variance matrix is the inverse of this total Fisher matrix: hδ pa δ pbi ≥[Γa b(dL, θN, θNL, φNL, α, γ(0)) + Γa b(dL, θN, θNL, φNL, α - π/4, γ(0))]- 1. (62) Equation (62) along with (60) is our final expression, describ- ing the time evolution of parameter estimation uncertainties. We note that after combining both data-streams, the matrices [F j(pslow(0))]a b for 4 < j ≤ 8 modes vanish exactly for all 10 pslow(0). Let us emphasize the most important features of eq. (60): • The parameter dependence is separated from the time dependence. The Fisher matrix, Γa b, is written as a linear combination of matrices F j(pslow) weighted by the scalars Pj(t), where F j(pslow) is independent of time and Pj(t) is independent of the parameters pa. Evaluat- ing F j(pslow) requires the computation of the parameter derivatives ∂ag j. • The evaluation of all integrals Pj(tfn) for different n = 1,2, . . . ,Ntf can be done in the same amount of time as needed for a single integration, since the tf dependence enters only in the integration bound in eq. (59), • Large Monte Carlo (MC) simulations can easily be per- formed since the time evolution is given by a small number of functions, Pj(t), which can be calculated a priori and pre-saved. No integrations at all are neces- sary during the MC simulation for calculating distribu- tions of correlation matrices. In § V below, we estimate the improvement in the computa- tion time provided by the HMD method for calculating distri- butions of parameter errors and their time evolution. C. Generalization to the exact PN signal So far, we only considered the simplest case, assuming no cross-correlation between parameters pslow and (pfast, pspin), for a restricted post-Newtonian waveform. Moreover, we as- sumed the Doppler-phase (24) to be negligible. Including cross-correlations and the Doppler phase would allow us to examine the range of validity of our approximations, and it would allow us to extend computations to the lower compo- nent mass regime, m < 105M⊙, where the Doppler phase be- comes important. Furthermore, including spin precession ef- fects can modify angular localization errors by a factor of ∼ 3, at least for the final errors at tf ≈ tISCO [17, 22]. While we con- tinue to use our initial set of approximations in later sections, we outline here how the HMD formalism could be used to decouple the time -- dependence from the angular parameter -- dependence, even in the case of the most general (arbitrary order) restricted post -- Newtonian waveform. Source sky po- sition angles (θN, φN), detector angles at ISCO (α, γ(0)) and luminosity distance (dL) can all be decoupled even if spin- orbit and spin-spin precessions are included in the waveform, which is potentially a great advantage over the traditional cal- culation methods (see § V for a detailed discussion). Consider a general restricted post-Newtonian signal given by eq. (22), for which we substitute the harmonic mode ex- pansion [41], hI,II( f ) = A 4 Xj=- 4 gI,II j f - 7/6ei[ jω⊕t( f )+ϕD+φGW], (63) where A ≡ A(pfast) is the amplitude in the frequency-domain (eq. 68 in ref. [17]), g j ≡ g j(pslow) is the modulation ampli- tude in eq. (42), ϕD ≡ ϕD(pslow, f ) = cD(pslow) f is the Doppler phase (see eq. 26 above), cD(pslow) = 2π f - 1 c sin θN cos γ), φGW ≡ φGW(p; f ) is the GW phase (20) (e.g. eq. 3.2 in ref. [25]), φGW(p; f ) = N Xn=0 cGW n (pfast, pspin)uGW n ( f ), (64) and time-frequency relationships t( f ) ≡ t(p; f ) can be written as (e.g. eq. 3.3 in ref. [25]) t(p; f ) = N Xn=0 cT n (pfast, pspin)uT n ( f ). (65) n n and cT n the cGW Here, ( f ) and uT coefficients are frequency- independent, while uGW n ( f ) are parameter- independent. They correspond to the various post-Newtonian terms in the post-Newtonian expansion, and N corresponds to the highest order term. The frequency functions are very sim- n ( f ) = f n- (8/3). ple powers of f , i.e. uGW n (Note that neither the cGW ( f ) and uT n ( f ) functions are complex. Every term in eq. (63) is real except for the g j ≡ g j(pslow) coefficients and the complex argument.) ( f ) = f n- (5/3) and uT and cT n coefficients nor the uGW n n Equations (63-65) can be combined into h( f )I,II = 4 Xj=- 4 j (p) f - 7/6eiΨ j(p; f ), AI,II AI,II j (p) = A(pfast)gI,II j (pslow), (66) (67) where and Ψ j(p; f ) = cD f + jω⊕ cT n uT n ( f ) + N Xn=0 N Xn=0 cGW n uGW n ( f )(68) c jkuk( f ), (69) ≡ 2N+1 Xk=0 where in the last step we introduced u ≡ { f ,uT,uGW} and c j ≡ {cD, jω⊕cT,cGW} to collect all f -functions and coefficients in one vector and one matrix, respectively. To compute the Fisher matrix, we need to obtain the partial derivatives of the signal with respect to the parameters: h,a( f ) = 4 Xj=- 4 A j,a + i 2N+1 Xk=0 A jc jk,auk( f )! f - 7/6eiΨ j( f ), (70) where commas in indices denote partial derivatives with re- spect to the parameter following the index. Note, that the pa- rameter index a spans all parameters pslow, pfast, and pspin, and the Fisher matrix accounts for correlations between these pa- rameters. A j1,aA j2 c j2k,bP(1) j1 j2 k(tf) Γ(0) Γ(1) 4 4 i a b(tf) =ℜ Xj1, j2=- 4  a b(tf) =ℜ Xj1, j2=- 4  - i Xj1, j2=- 4 Xj1, j2=- 4 a b(tf) =ℜ  4 4 A j1A j2,b A j1,aA j2,bP(0) , 2N+1 j1 j2(tf)  Xk=0 Xk=0 Xk1,k2=0 c j1k,aP(1) 2N+1 2N+1 j1 j2 k(tf)  , (73) , j1 j2 k1 k2(tf)  (74) Equation (70) can now be substituted in the Fisher matrix in eq. (52). We get 11 Γ(tf)a b = Γ(0) a b(tf) + Γ(1) a b(tf) + Γ(2) a b(tf), where (71) (72) Γ(2) A j1A j2 c j2k1,ac j2k2,bP(2) and where P(1) P(0) fmin j1 j2(tf) = 4Z f (tf) j1 j2 k(tf) = 4Z f (tf) (tf) = 4Z f (tf) fmin fmin P(2) j1 j2 k1 k2 j2)ω⊕t( f ) f - 7/3ei( j1- S2 n( f ) d f , f - 7/3uk( f )ei( j1- S2 n( f ) j2)ω⊕t( f ) d f , (75) f - 7/3uk1( f )uk2( f )ei( j1- j2)ω⊕t( f ) S2 n( f ) d f , are the frequency dependent terms. Equation (71) is our final result, where the localization parameters (i.e. angles and distance) are decoupled from all other parameters (i.e. masses, spins, reference time and phase at ISCO). The equation explicitly shows that, contrary to the traditional methods usually adopted for Monte Carlo computations of random binary orientations and sky positions [5, 7, 8, 13, 17, 19, 20, 21, 22], the localization of a LISA inspiral event and its time -- dependence can be explored with- out the need to evaluate integrals for each realization of the fiducial angles. Note that the only approximation made to ob- tain eq. (71) was to neglect of spin-orbit and spin-spin pre- cession in the general restricted post-Newtonian solution for the Fisher matrix. The time -- dependence is given by the P(tf) functions in eq. (75) and the extrinsic parameter -- dependence is given by the coefficients, A. The P(tf) functions in eq. (75) can be computed a priori, independently of the fiducial angles. Note that P(tf) depends implicitly on the parameters (pfast, pS) through t( f ) in eq. (65), and its inverse f (t). Generally, there are at most (2Jmax + 1)× [1 + (2N + 1) + (1/2)(2N + 1)(2N + 2)] such independent functions. From the general case, we can now deduce the special so- lution in eqs. (60) and (62) valid for a Newtonian evolution, no Doppler phase, and no cross-correlations between pslow and (pfast, pspin). This approximation simply corresponds to the first term Γ(0) ab (tf) in eq. (71), where in eq. (75) the t( f ) - function is computed using the Newtonian formula t0( f ) given by eq. (2). Note also, that the next term in eq. (71), Γ(1) ab (tf), corresponds to the cross-correlation of the amplitude modu- lation with the "high frequency carrier signal" (i.e. Doppler phase and GW phase). The last term, Γ(2) ab (tf), corresponds to the cross-correlations among parameters in the high frequency carrier. j Finally, we briefly consider extensions to include spin-orbit and spin-spin precessions in the signal. Let us refer to the an- gular momentum angles as pL(t) ≡ (θNL(t), φNL(t)), which are now time-dependent. As we briefly show next, in the case of spin precession, the P0,1,2(tf) time-dependent integrals loose the convenient property of being independent of pL, but never- theless, the parameters describing the sky position and detec- tor orientation are still time-frequency independent and they are decoupled in this prescription. Indeed, an extra precession phase exp[iδP(pL, pspin,t)] has to be included in eq. (63) and (p; f ) = Ψ j(p; f )+ δP(pL, pspin;t) j (pslow). We now have Ψprec gI,II instead of eq. (68). Thus, when taking the derivatives of the signal in eq. (70), we will get additional terms propor- tional to AI,II j (p)∂aδP,a and the original A j,a terms will have time variation due to the pL dependence of gI,II . Finally, after j these modifications, the Fisher matrix will be similar to that in eq. (71), except now the pL terms cannot be moved out- side of the frequency-integral but have to be attached to the time -- varying P0,1,2(tf) part [42]. The main advantage we re- tain is therefore that the source position and detector angles (θN, φN, α, γ(0)) will still only be included in the coefficients AI,II j (p) and c jk,a and the time -- evolution can be still be com- puted independently of these parameters. In § V C we show that this indeed reduces computation times by a large factor relative to the traditional methods (e.g. [17, 22]). We leave numerical implementations and explorations of parameter distributions and their time -- dependence, in this case of a general inspiral waveform, to future work. V. COMPUTATION TIME One of the great advantages of introducing the HMD method is the reduction in the computational time needed to evaluate error distributions for the parameters which deter- mine how efficiently LISA can localize SMBH binary inspiral events: sky position, angular momentum orientation and fi- nal detector orientation. In general, this is a computationally very demanding task because of the large dimensionality of the angular parameter space. Mapping the structure of the distribution of correlations in the parameter space of mock LISA measurements requires vast Monte Carlo simulations, which are presently limited by computational resources. Cur- rently, only a small portion of this space has been explored [7, 8, 13, 17, 20, 21, 22]. In this context, it is desirable to tune methods to the specific problem at hand. The HMD method described above is specifically constructed to exploit the struc- ture of LISA inspiral signals. A. Approximate solution 12 The computational time for parameter space exploration, using the HMD method with the approximations described in § IV B, is significantly reduced for the following reasons. The standard approach for estimating parameter errors requires the evaluation of an Np × Np symmetric Fisher matrix, where each matrix element is an integral over the range spanned by the GW frequency during inspiral (see Refs. [33, 34]; and § IV). Here, Np is the number of parameters describing the signal. The number of independent elements in a symmetric matrix is (1/2)Np(Np + 1). Let us assume that the evaluation of a single integral requires to compute the waveform at Nint separate instances. The computation of one integral is suffi- cient also to trace the time evolution at Nint different tf val- ues, if one uses a single trapezoidal integral in frequency from fISCO to fmin and stores results at each intermediate value of f . Since the time evolution of the frequency is known inde- pendently of the angles, we can already get the integral for Nint different tf values. For randomly chosen fiducial angles in a MC simulation of size NMC requires the calculation to be repeated NMC times. To evaluate the evolution of param- eter errors as a function of tf for Ntf different instances re- quires Ntf computations. Therefore, the standard method costs ∆Tstandard = (1/2)Np(Np + 1)NintNMC computational time units. In contrast, with our proposed HMD method (§ IV B), the pslow parameters are decoupled from the pfast parameters (§ II C) and from time. The Np × Np Fisher matrix can be split into two smaller matrices, with Np1 × Np1 and Np0 × Np0 components, where Np = Np1 + Np0. The Np1 × Np1 matrix de- termines the angular errors (which are deduced from the am- plitude modulation of the signal), while the other matrix deter- mines the remaining parameters (e.g. masses, phase and time at ISCO, using the high frequency carrier only). Since we are only interested here in parameters relating to the localization of the source, pslow, it is sufficient for us to consider the corre- sponding Np1 × Np1 matrix only. Since it is symmetric, it has only (1/2)Np1(Np1 + 1) independent elements, but it turns out that the computation of only Np1 elements is sufficient for a single harmonic (we need only the Np1 derivatives of the g j functions, see eq. [58]). Using 2NJ + 1 harmonic modes and a MC simulation with NMC random choices of fiducial parame- ters costs Np1(2NJ + 1)NMC time units. In this method, the time dependence is decoupled, so that parameter dependencies can be taken outside of the integral (see eq. [58] and F j(pslow) in eq. [60]). The MC sampling can therefore be evaluated inde- pendently of time. The time evolution of the signal for each harmonic is known independently of the angles, by construc- tion (Pj(tf), eq. [59]), and this integration for each component can be evaluated a priori, independently of the fiducial param- eter values. For each such mode, we would like to evaluate a number Ntf of integrals. Fortunately, since the different inte- grals differ only via the lower integration bound in the time domain, all integrals can be obtained during a computation of the integral with the largest time domain, tf = tISCO. There- fore for a total of (2NJ + 1) modes, building the time-evolution functions Pj(tf) takes of order (2NJ + 1)Ntf time units. This is generally much faster than building the time-independent co- efficients F j(pslow). In summary, with the HMD method, one only needs ∆THMD = Np1(2NJ + 1)NMC + (2NJ + 1)Ntf units of time. Comparing methods, we find that the computational re- quirements of the HMD method is lower by a factor of ∆T no spin HMD ∆T no spin standard Np1(2NJ + 1)NMC + (2NJ + 1)Ntf (1/2)Np(Np + 1)NintNMC . = (76) Recall from § II A that the number of parameters for a no- spin case is Np0 = 4, Np1 = 5, so that Np = Np0 + Np1 = 9. Choosing Nint = 103, NMC = 104, Ntf = 100, and NJ = 4 for the other parameters in eq. (76) as a representative example, the gain in computational efficiency is ∆Tstandard/∆THMD = (5× 108)/(5× 105) ≈ Nint = 1000. Moreover, the Fisher ma- trix is much smaller, 5× 5, which offers an important further advantage when performing the inversion to obtain the error covariance matrix. Using the HMD method, the inversion of the Fisher matrix is computationally less expensive than gen- erating the matrix. Note that the second term in ∆T no spin HMD , corresponding to the Pj(tf) functions, is negligible in this case and the computation time is dominated by constructing the coefficient matrices. A calculation of the representative MC example above, with a non-optimized implementation of the HMD method, takes less than a minute on a regular worksta- tion. The case for substantial improvement with the HMD method becomes even more compelling when additional pa- rameters (spins and higher order PN terms) are included, as we discuss next. B. Post-Newtonian Signal without Spin Precession We now consider the general HMD method outlined in § IV C, with Nspin ≡ 6 spin components. The spin parame- ters can be grouped as NSM ≡ 2 independent spin magnitudes and NSA ≡ 4 independent spin angles. Since the spins can be oriented arbitrarily, the spin angular parameters have to be randomly chosen, in addition to the other angular param- eters in any Monte Carlo computation. This enlargement of the parameter space of random parameters greatly increases the computational cost, both for the standard method and the HMD method. However, we show next that the incremen- tal cost is much less severe for the HMD method. The HMD method should be considered in future work aimed at comput- ing time-dependent parameter estimation errors in the general case with spins. Here, we neglect the effects of spin preces- sion, so that the angles Ω = (θN, φN, θNL, φNL, α, γ(0)) are de- coupled and, since the signal does not depend on φN, there are only NΩ ≡ 5 independent (spin-unrelated) angles. The larger the parameter space, the larger the sample size must be in a Monte Carlo computation. Let us assume that the MC(cid:17)d sample size is chosen to be NMC =(cid:16)N(1) MC is the effective number of samples for a single parameter, and d = NΩ when spanning the Ω-space only, d = NSA when spanning the spin-angle space only, and d = NΩ + NSA when spanning both. , where N(1) 13 To compute the time-independent matrices, we have to evaluate Np(2NJ + 1) independent A j,a coefficients for the full Ω-space and we have to compute the Np(2NJ + 1)(2NPN + 1) independent c jk,n matrices over a d = ND + NSA dimensional parameter space. (Here ND ≡ 2 denotes the number of param- eters on which the Doppler phase depends, (θN, γ) in eq. [26], using the fact that both cGW and cT also depend on all spin angles in eqs. [64,65].) In § IV C we have shown that there are (2NJ + 1)× (2N2 + 7NPN + 6) independent integrals, where NPN is the number of terms in the post-Newtonian expansion plus the Doppler phase. We have to compute these integrals for all spin angle orientations. Therefore, the computational cost scales as PN ∆T no spin prec HMD = Np(2NJ + 1)(cid:16)N(1) +(2NJ + 1)(2N2 +Np(2NJ + 1)(2NPN + 1)(cid:16)N(1) MC(cid:17)NΩ MC(cid:17)NSA + 7NPN + 6)Ntf(cid:16)N(1) MC(cid:17)ND+NSA PN . (77) For the standard method, reiterating the argument in § V A, we get ∆T no spin prec standard = 1 2 Np(Np + 1)Nint(cid:16)N(1) MC(cid:17)NΩ+NSA , (78) standard /∆T no spin prec where now Np = Np1 + Np0 + Nspin = 15. Taking N(1) MC1 = 10, NPN = 4, NJ = 4, ND = 2, NΩ = 5, NSA = 4, Nint = 103, Ntf = 100 as a representative example, we find that the HMD method is computationally less expensive by a factor ∆T no spin prec = (1 × 1014)/(2× 109) = 7 × 104, as compared to the standard method. Note that the first term in eq. (77) corresponds to our orig- inal approximations in § IV B, i.e. no cross-correlations be- tween pspin and pslow. This approximation indeed leads to much faster computations, since ∆T no spin prec = 9× 106 for the same representative example. /∆T approximate standard HMD HMD C. Post-Newtonian Signal with Spin Precession Accounting for spin precession, the NL = 2 angular momen- tum angles, (θNL, φNL), and the NSA spin angles are now chang- ing with time. In this case, one has to solve a differential equation for the evolution of these angles for each individual Monte Carlo realization. We assume that this can be com- puted independently of the Fisher matrices and that it would take NDE Ntf computation time units to evaluate, at each of the Ntf time instances, for each initial set of angles. The HMD method costs ∆T spin prec HMD MC(cid:17)NΩ MC(cid:17)ND+NSA + 7NPN + 6)Ntf(cid:16)N(1) = Np(2NJ + 1)(cid:16)N(1) +Np(2NJ + 1)(2NPN + 2)(cid:16)N(1) +(2NJ + 1)(2N2 +NDENtf(cid:16)N(1) MC(cid:17)NL+NSA PN , MC(cid:17)NL+NSA (79) where the first term involves constructing the A j,a coefficient matrices, the second term involves constructing the c jk,a coef- ficient matrices, the third term is for computing all three time -- evolution quantities P(tf) in eq. (75), and the fourth term is for solving the precession equations. In the standard method, we need first to solve the precession evolution differential equation and then construct the Fisher matrix [22]. Following the assumptions made above, we esti- mate a cost Lang & Hughes reported angular errors that are a factor of 2 -- 3 lower [35], which are inconsistent with our results at this level. Nevertheless, these discrepancies are still small relative to the typical width of error distributions or to the systematic variations of mean errors with t f , M, and z (from a factor of few to orders of magnitudes, see Fig. 2 below). This success- ful comparison justifies the use of the HMD method to study the dependence of localization errors on look -- back time. 14 ∆T spin prec standard = NDENtf(cid:16)N(1) + 1 2 MC(cid:17)NL+NSA Np(Np + 1)Nint(cid:16)N(1) MC(cid:17)NΩ+NSA Ntf. (80) HMD standard /∆T spin prec Using NL = 2 and all other parameters as in § V B, we find that the HMD method is computationally more efficient by a factor ∆T spin prec = (2×1014)/(6×1010) = 2×103, as compared to the standard method. Since the (θNL, φNL) subspace could no longer be decoupled, the efficiency of the HMD method relative to the traditional method lost a factor of 30, as compared to the no spin precession case in § V B. Nevertheless, the computational advantage remains very sub- stantial. VI. RESULTS Having described the HMD formalism in detail, we now apply it to build MC simulations aimed at studying how RMS source localization errors [43] evolve as a function of look -- back time, tf, before merger. The low computational cost of the HMD method allows us to survey simultaneously the de- pendencies on source sky position, SMBH masses and red- shifts. We carry out MC calculations with 3 × 103 random samples for the angles cos θN,cos θNL, φNL, α, γ(0). Several thousands values of M and z are considered, in the range 105 < M/M⊙ < 108 and 0.1 ≤ z ≤ 7. In addition, we ran specific MC calculations to study possible systematic effects with respect to the source sky position, by fixing θN and φN (on a grid of several hundred values) and varying all the other relevant angles. In all of our computations, we calculate the er- ror covariance matrix for dL, θN, φN, θNL, φNL. Following Lang & Hughes [22], we calculated the major and minor axes of the 2D sky position uncertainty ellipsoid, 2a and 2b, and the equivalent diameter, √4ab. We have verified our HMD implementation and the gen- eral validity of our assumptions by comparing our results at ISCO with those of Lang & Hughes [22] (for m1 = m2 = 105,106,107M⊙ at z = 1 and m1 = m2 = 105,106M⊙ at z = 3, in the no spin precession case). Depending on SMBH masses and redshifts, we found agreement at the 5 -- 30% level for the mean errors on the luminosity distance, major axis, and minor axis. The small discrepancies may be due to differences in the set of assumptions made. Lang & Hughes [22] account for the small cross-correlations between the pslow and {pfast, pspin} parameters and they choose ti to be uniformly distributed be- tween merger time and LISA's mission lifetime. Recently, A. Time dependence of source localization errors We calculate the variation with look -- back time, tf, of the distribution of marginalized parameter errors for a range of values of M,z, θN , θNL, φNL, α, γ(0). Figure 2 shows results for random angles and m1 = m2 = 106M⊙, at z = 1. The top panel shows the luminosity distance error, δdL, while the bottom panel describes the equivalent diameter, 2√ab, of the sky position error ellipsoid with minor and ma- jor axes a and b. The figure displays results for three separate cumulative probability distribution levels, 90%,50%,10%, so that 10% refers to the best 10% of all events, as sampled by the random distribution of angular parameters. The evolution of errors scales steeply with look -- back time for tf ∼> 40 days. In this regime, the improvement of errors is proportional to (S/N)- 1. For smaller look -- back times, errors stop improving in the "worst" (90% level) case, improve with a much shal- lower slope than (S/N)- 1 for the "typical" (50% level) case, and keep improving close to the (S/N)- 1 scaling in the "best" case (10% level among the realizations of fiducial angular pa- rameters). Although Figure 2 shows only the equivalent di- ameter of the 2D sky localization error ellipsoid, we have also computed the evolution of the distribution of the minor and major axes. We find that a ≈ b ≈ √ab initially (i.e. the ellip- soid is circular), but the geometry changes significantly during the last two weeks to merger. For example, in the typical case, the major axis a stops improving at late times, while the minor axis a maintains a steep evolution. Therefore the eccentricity of the 2D angular error ellipsoid changes quickly with look -- back time. This is important because large eccentricities can play a role in assessing observational strategies for EM coun- terpart searches [18]. To map possible systematic effects with respect to source sky position, we carried out MC computations with ran- dom (cosθNL, φNL, α) angles (sample size NMC = 3× 103) but fixed source sky latitude and longitude relative to the detec- tor (θN, γ), for m1 = m2 = 106M⊙ and z = 1. We find no sys- tematic trends with sky position for δdL, for any value of the look -- back time, tf. Neither do we find systematic trends with sky position for the distributions of minor and major axes of the angular ellipsoid, for any value of the look -- back time, tf, as long as θN is not along the equator. The case of equatorial sources, with θN ≈ 90◦ and a short look -- back time tf before merger, is the only nontrivial one we have identified. In that case, we find a minor systematic trend with γ longitude. The error distributions shift periodically up and down, relative to the average, when changing γ from 0 to 2π. In addition, to map dependencies with mass -- redshift -- look -- m1 = m2 = 106M⊙ z = 1 1 0.1 L d / i 2 L d δ h q worst typical 0.01 b est 0.001 0.01 0.1 1 10 tf − tISCO [day] 1000 m1 = m2 = 106M⊙ z = 1 worst 100 typical b e s t 10 0.01 0.1 1 10 tf − tISCO [day] ] i n m c r a [ r o r r e r a l u g n a O C S I t − i t O C S I t − i t 90% 50% 10% 100 90% 50% 10% 100 FIG. 2: Evolution with pre-ISCO look -- back time, tf, of LISA source localization errors, for M = 2 × 106M⊙ and z = 1. The top panel shows luminosity distance errors and the bottom panel shows sky position angular errors (equivalent diameter, 2√ab, of the error el- lipsoid). Best, typical, and worst cases for random orientation events represent the 10%, 50%, and 90% levels of cumulative error distribu- tions, respectively. Errors for worst case events effectively stop im- proving at a finite time before ISCO, even though the signal-to-noise ratio accumulates quickly at late times. Errors for best case events (especially the minor axis) follow the signal-to-noise ratio until the final few hours before merger. back time of localization errors, we carried out MC compu- tations with arbitrary (cosθN,cos θNL, φNL, α, γ) angles, with sample size NMC = 3 × 103, for several thousand pairs of (M,z) values. We find that the evolution with look -- back time of error distributions depends sensitively, and in a complicated way, on the mass-redshift parameters. Gener- ally, localization errors increase with redshift. Firstly, the S/N is approximately proportional to the instantaneous value σ(tISCO) ∝ η3/4[(1 + z)M]5/4/dL(z)Sn( fISCO)- 1 (eq. [56]) and, secondly, the beginning-of-observation time scales as ti ∝ - 1[(1 + z)M]- 5/3 (eq. [2]). For (1 + z)M < 4× 106M⊙, the to- η tal observation time can exceed one year and the second effect 15 is unimportant. We further describe mass -- redshift dependen- cies below, in § VI B, in relation to advance warning times for targeted electromagnetic counterpart searches. The results on localization errors from our extensive explo- ration of the parameter space of potential LISA sources can be summarized as follows: 1. Probability distributions • The error distributions for δdL, 2a, and 2b all have long tails: 1% -- 99% cumulative probability levels are separated by factors of ∼ 100, while the 10% -- 90% levels are separated by factors of ∼ 10. • The δdL distribution is skewed, with a median closer to the best case, a median smaller than the mean, even on a logarithmic scale. On the other hand, sky localization error distributions are roughly symmetric on a logarithmic scale. 2. Fiducial parameter dependencies • The δdL errors are roughly independent of fiducial angles throughout the observation. • For non-equatorial sources, the distribution of sky localization errors, (2a,2b), is independent of sky position, i.e. the distribution does not have a sys- tematic dependence on θN and γ ≡ Φ- φN (for ran- dom α, θNL, φNL). • There is a small systematic trend with γ for equa- torial sources. • There is a complicated dependence of sky lo- calization errors on M, z, and look -- back time tf. For (1 + z)(η/0.25)3/5M ∼< 4 × 106, and long observation times, errors scale with (S/N)- 1 ≈ [(1 + z)M]5/4dL(z)- 1Sn( fISCO(M,z))- 1 fa(tf), where fa(tf) is the tf-scaling shown in Fig 2. For larger redshifted masses, the scaling has a complicated structure in the M,z,tf space that we did not ana- lyze in detail (but see eq. (A10) in the Appendix for scalings in terms of ti and tf.) 3. Time dependence • Luminosity distance and sky localization errors roughly scale with (S/N)- 1 until 2 weeks before ISCO. • For the luminosity distance δdL and the major axis 2a, there is little improvement within the last week before ISCO for the typical to worst cases (i.e. 50% -- 90% levels of cumulative error distri- butions). • For the minor axis 2b, only the worst case events stop improving within the last week. The typical to best cases continuously improve until the last hour. • The eccentricity of the sky localization error el- lipsoid changes with time during the first and last two weeks of observation. The eccentricities are smaller in between these two time intervals. For a detailed discussion of the eccentricity and its im- pact on counterpart searches, see Ref. [18]. • For the luminosity distance δdL, the relative width of error distributions does not change during ob- servation and variations in the difference between the 90% and 10% levels of the cumulative distri- butions do not exceed 10%, except for the initial weeks, when the distribution is much more spread out. • For the sky localization errors, the width of error distributions increases during the final two weeks of observation, by a factor ∼ 2 for the major axis and a factor ∼ 4 for the minor axis. B. Advance warning times for EM searches From the astronomical point of view, being able to identify with confidence, prior to merger, a small enough region in the sky where any prompt electromagnetic (EM) counterpart to a LISA inspiral event would be located, is of great interest. With sufficient "warning time," it would then be possible to trigger efficient searches for EM counterparts as the merger proceeds and during the most energetic coalescence phase. In particular, an efficient strategy to catch such a prompt EM counterpart would be to continuously monitor with a wide- field instrument a single field-of-view (FOV), through coales- cence and beyond. Astronomical strategies for EM counter- part searches are the focus of a second paper in this series [18]. Given an angular scale, θFOV, corresponding to the hypo- thetical FOV of a specific astronomical instrument, it is thus of considerable interest to determine the value of the look -- back time tf at which the major axis, minor axis or equivalent diameter of the sky localization error ellipsoid provided by LISA just reach the relevant θFOV scale. This would allow one to trigger an efficient search for EM counterparts, in a well defined region of the sky that can be monitored. We will here- after refer to this time as the advance warning time. Note that it is important to differentiate the sizes of the major and mi- nor axes of the angular error ellipsoid in this context because the eccentricity can be large, and thus important in assessing optimal strategies for EM counterpart searches [18]. For definiteness, we evaluate advance warning times for an- gular diameters θFOV = 1◦ and 3.57◦ here but generalizations to other θFOV values are obviously possible. The choice of the latter figure is motivated by the 10 deg2 FOV proposed for the future Large Synoptic Survey Telescope, or LSST [36]. Fig- ure 3 shows advance warning times for a fixed source redshift at z = 1 and various values of the total SMBH mass, M. Fig- ure 4 shows similar results for various source redshifts, at a fixed value of M = 2× 106M⊙. In each case, we consider equal mass SMBH binaries and a maximum observation time of 1 yr (or lower if set by the GW noise frequency wall at 0.03 mHz). Each panel in Figs. 3 and 4 shows the values of advance warning times at which 16 2√ab = 1◦ ti −t I S C O z = 1 ] s y a d [ e m i t g n i n r a w e c n a v d A 100 10 1 0.1 b e s t t y pic al 105 106 107 M [M⊙] ti − tISCO 2√ab = 3.57◦ z = 1 b est t y p i c a l t s r o w ] s y a d [ e m i t i g n n r a w e c n a v d A 100 10 1 0.1 105 106 107 M [M⊙] FIG. 3: Advance warning times (in days) for equal mass binary in- spirals at z = 1, as a function of total mass, M (in solar units). Best, typical, and worst cases refer to 10%, 50%, and 90% levels of cu- mulative error distributions for random orientation events, as before. The advance warning times shown correspond to the values of look -- back times when the equivalent diameter, 2√ab, of the error ellip- soid first reaches 1◦ (top panel) or an LSST-equivalent field-of-view (3.57◦, bottom panel). In the top panel, the worst case events are not shown because angular errors are too large even at ISCO. For the largest mass SMBHs, the maximum observation time (and thus ti) is below one year. the equivalent diameter 2√ab of the localization error ellip- soid drops below the reference θFOV value. For each case, we show results for cumulative error distribution levels of 10%, 50%, and 90%, labeled "best", "typical," and "worst" cases, as before. Figure 3 shows that LISA can localize on the sky events at z = 1 to within an LSST FOV at least one month ahead of merger, for 50% of events with masses 2 × 105M⊙ ≤ M ≤ 3 × 106M⊙, and at least 4 days ahead of merger for 90% of events in the same mass range. Fig- ure 4 shows that advance warning times decrease with red- shift, leaving at least 1 day ahead of merger for 50% of events with M = 2 × 106M⊙, as long as z ∼< 1 for θFOV = 1◦ and as 2√ab = 1◦ ti−tISCO M = 2 × 106 M⊙ z b e s t w o r s t t y p i c a l 0 1 2 3 z 4 5 6 2√ab = 3.57◦ ti−tISCO M = 2 × 106 M⊙ z best w o r s t t y p i c a l ] s y a d [ e m i t g n i n r a w e c n a v d A 100 10 1 0.1 ] s y a d [ e m i t i g n n r a w e c n a v d A 100 10 1 0.1 4 3 2 1 0 105 7 6 5 4 3 2 1 0 105 17 m1/m2 = 1 2√ab = 3.57◦ 50% levels 107 108 M m1/m2 = 1 2√ab = 3.57◦ 10% levels 0 .1 1 1 0 100 106 1 10 100 106 107 108 M 0 1 2 3 z 4 5 6 FIG. 4: Same as Fig. 3, for a fixed total mass M = 2× 106M⊙ but various values of the source redshift, z. long as z ∼< 3 for an LSST FOV. For events with this mass scale and the LSST FOV, there is a 10% chance that a 1 day advance warning is possible up to z ∼ 5 -- 6. Figures 3 and 4 display advance warning times for single one dimensional slices of the full (M,z) space of potential LISA events. With the HMD method, however, it is possi- ble to explore the entire parameter space of SMBH inspirals by repeating the calculation on a dense grid of (M,z) val- ues. We construct a uniform grid in the (logM,z) plane, with ∆z = 0.1 and ∆log M = 0.1, and perform MC computations with 3× 103 randomly oriented angles for each grid element. As a result, we obtain a complete description of the time evo- lution of sky localization errors in the large parameter space of potential LISA sources. Figure 5 displays advance warning time contours from this extensive MC calculation, for typical (50%) and best case (10%) events, adopting the LSST FOV as a reference. Advance warning time contours are logarithmically spaced, with solid-red contours every decade and a thick red line high- lighting the 10 day contour. Since advance warning times were computed on a finite mesh, contour levels for arbitrary M FIG. 5: Contours of advance warning times in the total mass (M) and redshift (z) plane with SMBH mass ratio m1/m2 = 1. The contours trace the look -- back times at which the equivalent radius (2√ab) of the localization error ellipsoid first reach an LSST-equivalent field- of-view (3.57◦). The contours correspond to the 50% (top) and 10% (bottom) level of cumulative distributions for random orienta- tion events. The contours are logarithmically spaced in days and 10 days is highlighted with a thick line. and z values were obtained by interpolation. Our interpolated mesh is smooth if tf ∼ 0.1 day, but it gets edgy for short ad- vance warning time approaching ISCO. Figure 5 shows that a 10 day advance warning is possible with a unique LSST-type pointing for a large range of masses and source redshifts, up to M ∼ 3 × 107M⊙ and z ∼ 1.9. The bottom shows how far the advance warning concept can be stretched, by focusing on the 10% best cases of random orientation events. In this case a 10 day advance warning is possible up to z ∼ 3 for masses around M ∼ 106M⊙. Note that, in both cases, allowing for a warning of just one day would extend considerably the range of masses and redshifts for which a unique LSST-type point- ing is sufficient. These results can also be generalized to unequal-mass SMBH binaries. At fixed total mass, M, an unequal-mass bi- nary has an instantaneous signal-to-noise ratio that is reduced because of a lower η value, but it also has a total observation z z 4 3 2 1 0 105 7 6 5 4 3 2 1 0 105 4 3 2 1 0 105 4 3 2 1 z z 18 m1/m2 = 1 2√ab = 3.57◦ 50% levels fmin = 0.1 mHz 3 arms 0 . 0 1 0.1 1 10 1 0 0 106 107 108 M m1/m2 = 10 2√ab = 3.57◦ 50% levels fmin = 0.1 mHz 3 arms 0.1 1 1 0 1 0 0 m1/m2 = 10 2√ab = 3.57◦ 50% levels 0.1 1 1 0 100 106 107 108 m1/m2 = 10 2√ab = 3.57◦ 10% levels 0.1 1 0 1 1 0 0 106 M M 107 108 0 105 106 107 108 M FIG. 6: Same as Fig. 5, except for a SMBH mass ratio of m1/m2 = 10. FIG. 7: Same as the top panels of Figs. 5 and 6, except for a degraded minimum detector frequency of fmin = 0.1 mHz. 2/5 0.25(1+ z) M < 1.8× 107M⊙) are degraded time that is potentially longer. Localization errors for unequal- mass inspiral events with total observation times longer than a month (i.e. with η relative to the equal-mass cases discussed so far. For larger total mass, however, the worsening of errors is mitigated, or even reverted, relative to the equal mass case, thanks to the longer observation time. The error ellipsoid also becomes less eccentric thanks to this additional observation time. Figure 6 summarizes results on advance warning times from the same MC computations as in Fig. 5, but this time for unequal-mass SMBH binaries with mass ratio m1/m2 = 10. Despite a sys- tematic degradation in advance warning times (especially no- ticeable at low M values), the main effect of introducing a mass ratio m1/m2 = 10 is to shift advance warning time con- tours to somewhat larger values of total mass, M. Our main conclusions on advance warning times are not very strongly affected by the inequality of mass components in the popula- tion of SMBH binaries considered. Finally, it is important to understand how sensitive the re- sults are to the LISA detector characteristics. In particular, we examined how advance warning times are affected by increas- ing the minimum frequency noise wall or by loosing one of the arms of the 3-arm constellation. Figure 7 displays results - 5/3 - 8/3 min M z for fmin = 10- 4 Hz, for m1/m2 = 1 and m1/m2 = 10. Increas- ing fmin mostly reduces the total observation time for high ; see eq. [2]) and reduces the mass inspirals (ti ∼ f signal-to-noise ratio by a small factor. As a result, the ad- vance warning time contours primarily shift in the (M,z) plane in the direction of smaller total masses by a factor of ∼ 7, and secondly shift moderately (30 -- 50%) to smaller redshifts. Loosing one LISA arm (i.e. using only one of the two interfer- ometers) most importantly removes the ability of the second datastream to break correlations in localization errors and also reduces the signal-to-noise by a small factor. As a result errors do not improve much during the last ∼ 10 days before merger. Compared to the case with two interferometers, contours rep- resenting an advance warning of less than 10 days are shifted to significantly smaller z (especially for the minor axis of the sky localization ellipsoid), close to the 10 -- day contour, but warning times beyond ∼ 10 days worsen only moderately. We conclude that even if fmin = 0.1 mHz or if only one of the two interferometers is used, LISA still admits 10 -- day advance lo- calizations for a broad range of masses and redshifts, between 105 ∼< M ∼< 2× 106 and z ∼< 1. VII. DISCUSSION We have introduced a novel technique, the HMD method, to compute time -- dependent GW inspiral signals for LISA. The method relies on the fact that LISA's orbital motion induces a modulation on timescales that are long relative to the in- spiral GW frequency. Since this modulation is periodic, with a fundamental frequency of f⊕, it can be expanded in a dis- crete Fourier sum. In the HMD formalism, dependencies on sky position, orbital angular momentum orientation, and de- tector orientation in the LISA signal are inscribed in time- independent coefficients, while time-dependent basis func- tions are independent of these angles. This decomposition helps to reduce the computational cost of Monte Carlo sim- ulations exploring the time-dependence of source localization errors by orders of magnitude. Moreover, the HMD method can be used in conjunction with plausible approximations to further decrease the com- putational cost of explorations of the parameter space of lo- calization errors for LISA inspiral events. In our analysis, we identified two different characteristic frequency constituents of the signal: the high frequency restricted post-Newtonian GW inspiral waveform and the low frequency amplitude mod- ulation resulting from the detector's orbital motion. In the HMD method, these two components separate and parameters that depend only on the low frequency modulation (such as the source position and the orbital angular momentum angle) can be estimated independently of the other source parameters de- termined by the high frequency carrier signal. Our working assumption was that cross-correlations among these two sets of parameters must be much smaller than parameter correla- tions within either set. This hypothesis is valid very generally in the no spin limit for SMBHs, as shown by full Fisher matrix calculations without such approximations for general relativ- ity [7] and alternative theories of gravity [13]. In order to further examine the validity of our assumptions and the ultimate boundaries of our models, and to understand our results, we have constructed illustrative toy models that we now describe in some detail. These toy models show that the separation of parameters into various subsets associ- ated with different characteristic frequencies of the signal is a rather general property, which turns out to be an efficient way of reducing the computational cost of error estimations for the LISA problem. A. Simple toy models In this section, we discuss very simple toy models which capture the essence of the problem posed by the time- evolution of parameter error estimations. We then use these models to answer general questions on the LISA-specific pa- rameter estimation problem. Our harmonic decomposition technique is based on the sim- ple intuition that the angular information can be deduced from the slow periodic modulation of the high frequency GW wave- form. In § III, we have shown that modulation harmonics with frequencies larger than 4 f⊕ vanish exactly. Here, we discuss 19 the general properties of such a modulation. In the case of LISA, the high frequency carrier signal has an effective, cycle- averaged signal-to-noise ratio which monotonically increases with time as SMBH binaries approach merger. To mimic such events, we also assume in all of our toy models that the instan- taneous signal-to-noise ratio continuously improves through- out the observation. We seek answers to the following questions: 1. How do mean errors evolve during the final days of ob- servation? On the one hand, in standard angle-averaged treatments (e.g. [5, 13, 15]), an evolution of errors with the inverse of the signal-to-noise ratio is generally assumed. This would suggest a large improvement during the last day of inspiral. On the other hand, the slow modulation pic- ture suggests just the contrary: not much improvement is expected at late times when there is effectively very little modulation (Finn & Larson 2005, private commu- nication). 2. Does the introduction of additional high frequency components in the signal have any effect on the esti- mations of low frequency parameters? In the GW context, it is of general interest to deter- mine under what circumstances additional high fre- quency signal components, such as higher order post- Newtonian corrections or spin-induced effects, remain decoupled from the determination of angular and dis- tance information based on the signal amplitude modu- lation. 3. Are there combinations of signal parameters for which errors improve rapidly in the last days of observation? If so, what are these combinations? What determines how many such rapidly -- improving combinations there will be? If the distance dL correlates with the angles, then in principle the volume of the 3D error box can be much smaller than the product of the marginalized errors δΩ× δdL would imply. Unfortunately, in practice, this is unlikely to help to reduce the number of false coun- terparts, because the δz error will be dominated by weak lensing [15]. 4. How does the width of parameter error distributions evolve with time? Are the best and worst cases ap- proaching the typical case prior to the final days of ob- servation? How do we expect the eccentricity of local- ization error ellipsoid to evolve with time for LISA? Here, we restrict our discussion to a brief summary of our findings and direct the reader to Appendix A for further details on these toy models. The parameter estimation uncertainties are defined by the correlation error matrix. For Np parameters, this defines an Np-dimensional error-ellipsoid in the Np-dimensional param- eter space, where parameters are constrained at a given con- fidence level. Marginalized errors for a given parameter are then related to the projection of this ellipsoid on the basis vector corresponding to that parameter. Since the principal axes of this error ellipsoid are generally not aligned with the original parameters, the marginalized errors can be substan- tial even if the volume of the error ellipsoid is close to zero. This happens if the ellipsoid is very "thin" but has a large size in at least one direction. Diagonal elements of the correlation matrix provide marginalized squared errors on the parameters, while eigenvalues provide squared errors along the principal axes. We consider three versions of toy signals to understand how a particular harmonic mode contributes to the time- dependence of parameter uncertainties and to find answers to Questions 1 -- 4 above. We start with the simplest toy model and refine this model by adding more details and complexity in the successive models. In each case, we discuss general implications for the model under consideration. 1. Basic toy model In our basic toy model, we assume that the true signal is comprised of a constant carrier signal, which is modulated by a single known-frequency cosine, f⊕: h(t) = c0 + c1 cos(2π f⊕t), (81) ⊕ merger (i.e. tf ∼> 0.1 f - 1 where c0 and c1 are unknown parameters to be estimated. We assume that the noise level is rapidly decreasing during the ob- servation, mimicking the gradual increase in the instantaneous signal-to-noise ratio for LISA inspiral signals. The contradic- tory statements made in relation to Question 1 above can be explored with this model. We find that marginalized param- eter errors scale with the signal-to-noise ratio far away from ) but they quickly converge to their fi- nal values at late times, even though the signal-to-noise ratio keeps accumulating. It is possible to derive analytical formu- lae for the evolution of parameter errors to fully characterize this behavior (see Appendix A). We find that, even though the error ellipse rapidly decreases in volume, as the inverse of the signal-to-noise ratio near merger, the error ellipse only shrinks along one of its dimensions, the semi-minor axis, so that a non-negligible residual uncertainty remains in the orthogonal subspace (e.g. along the semi-major axis). This residual un- certainty carries over to final marginalized errors for both pa- rameters. Therefore, this first toy model verifies the second option in relation to Question 1.: there is no late improvement because there is very little effective signal modulation, mak- ing the signal-to-noise argument largely irrelevant. However, we find below that this model does not carry some essential features of the LISA signal which modify somewhat our final answer to Question 1 (see final toy model below). 2. Second toy model In our toy second model, we modify the single frequency signal by postulating two pairs of unknown amplitudes and 20 phases for two different a priori known frequencies, satisfying f2 ≫ f1, which modulate an otherwise constant signal: h(t) = c0 + s1 sin(2π f1t) + c1 cos(2π f1t) + s10 sin(2π f2t) + c10 cos(2π f2t). (82) The number of unknowns in this model is five: c0,s1,c1,s10 and c10 are the coefficients of the functions 1, sin(2π f1t), cos(2π f1t), sin(2π f2t), and cos(2π f2t). Again, we assume that the noise decreases quickly with time before merger, at t = 0. This model is designed to answer our Question 2 above. In this case, we find that parameter errors are corre- lated only with unique frequency components and the constant 2 . The model thus demon- strates how components associated with very different vari- ation timescales can decouple from each other. Moreover, as for the first toy model, we find that marginalized parameter er- rors effectively stop improving past a finite time before merger (Question 1), which is simply related to their respective fre- quencies. As a result, a nonzero residual error remains again, even though the signal-to-noise ratio continuously increases near merger. signal, all the way to tf ∼ > 0.1 f - 1 3. Final toy model In our final toy model, we insert a few additional fea- tures essential to a realistic LISA data-stream. Firstly, we as- sume 5 low-frequency harmonics, 1, sin(2π f1t), cos(2π f1t), sin(4π f1t), (sin 4π f1t), with unknown amplitudes. We also include a high frequency carrier signal with known fre- quency, f2 ≫ f1, but unknown amplitudes in sin(2π f2t) and cos(2π f2t), for a total of seven free parameters. Secondly, we note that the LISA system is equivalent to two orthogonal arm interferometers with both detectors measuring polariza- tion phases simultaneously (which correspond to the real and imaginary parts of the amplitude modulation, § IV). There- fore, the signal is comprised of 4 simultaneous data-streams. We incorporate this feature by assuming 4 measurements (i.e. 4 corresponding Fisher matrices) of the signal with 4 given phase shifts (ϕs1 i , ϕc1 i , ϕs2 i , ϕc2 i ; 1 ≤ i ≤ 4) so that h(t) = c0 + s1 sin(2π f1t + ϕs1 i ) + c1 cos(2π f1t + ϕc1 i ) i ) + c2 cos(2π f1t + ϕc2 i ) +s2 sin(2π f1t + ϕs2 +s10 sin(2π f2t) + c10 cos(2π f2t), (83) In this case, we find that 4 principal components improve quickly at late times. As in our second toy model, the high frequency parameters decouple from the slow frequency ones, except at very late times when tf ∼ This final toy model allows us to answers all of Questions 1- > 0.1 f - 1 2 . 4 as follows. • Answer 1: Four out of 5 slow principal components of the error ellipsoid are quickly improving with time, while one of them stops improving at tf ∼< 0.1 f - 1 1 . Therefore, any parameter with a large projection along this one poor principal component will stop improving, while parameters nearly orthogonal to it will keep im- proving quickly. Thus, both statements made in relation to Question 1 above can in fact be correct, depending on the connection between a given parameter and the poor principal component. Typically, we expect marginal- ized parameter uncertainties to evolve as (S/N)- 1 for tf ∼> 0.1 f - 1 1 . For smaller tf values, closer to merger, they would continue to improve, albeit with a shallower slope. • Answer 2: We find that the introduction of additional high frequency components does not change the evo- lution of original parameter estimations as long as the time-to-merger is larger than a fraction of the time pe- riod of the additional high frequency components. • Answer 3: As the signal-to-noise ratio increases quickly at late times, rapidly evolving parameter error combi- nations are given by the principal components of the error ellipsoid corresponding to the final situation at merger. With 4 data-streams, there are 4 such best prin- cipal components. Analogously, for the LISA ampli- tude modulation given by eq. (25), we expect that the 2 polarization phases for the 2 beam patterns at ISCO can be best determined: (1 + cos2 θNL)FI,II+ (ΩISCO) and cosθNLF I,II (ΩISCO). (In terms of ecliptic angular vari- × ables, these are the real and imaginary parts of the com- bination given by eq. (34).) • Answer 4: The widths of error distributions for slow parameters do not change significantly as long as tf ∼ > 0.1 f - 1 1 . During this final stretch of time before merger, however, one of the principal components stops improv- ing and the major axis of the error ellipsoid freezes. Since the physical parameters can be considered to be randomly oriented with respect to the ellipsoid axes, distributions of marginalized errors suddenly start 1 , with a worst case relative broadening for tf ∼ orientation leading to very little improvement and a best case relative orientation corresponding to a scaling with (S/N)- 1. < 0.1 f - 1 B. Implications for LISA These simple toy models offer a general interpretation of the time dependence of LISA's parameter estimation errors for source localization. The LISA data stream is described by Np1 = 5 physical parameters, pslow, which are not the harmonic coefficients themselves but determine these coefficients, g j (or conversely, the mode expansion coefficients g j determine the physical parameters pslow; see § III). Neglecting Doppler phase and spin precession effects, 2Jmax + 1 = 9 modes deter- mine the signal by eqs. (41,42). In principle, any Np1 = 5 of the g j mode amplitudes uniquely determine the physical pa- rameters, pslow. However, in the presence of noise, each of these modes are uncertain and the combination of all modes helps in reducing the estimation errors of the pslow variables. 21 The key implication of our toy models for LISA is that the estimation of low frequency g j modes with low j are effectively decoupled from the high frequency signal, unless the merger is within ∼ 0.1 times the cycle time of the fast- oscillating signal. We have shown that the HMD of the or- bital modulation consists purely of low-order harmonics, with j ≤ 4. In comparison, the high frequency GW phase has a much higher frequency, corresponding to j > 1000, and this high frequency signal's cycle time is greater than the time to merger throughout t ∼> tISCO. Hence, physical parameters pslow will remain decoupled from parameters pfast, all the way to ISCO. This finding is independent of details of the wave- form and the modulation, in agreement with the results of Ref. [13] which show that decoupling occurs independently of the details of the hc(t) signals, including the modified in- spiral waveforms of alternative theories of gravity. In terms of post-Newtonian expansions, only terms above second order have cycle times as large as the cycle time of the amplitude modulation. These terms are responsible for the small cross- correlations of the two sets of parameters found by Ref. [7]. We have not considered spin precession effects, but Vec- chio [17] and Lang & Hughes [22] find that spin precession effects can help improve the final localization errors by a fac- tor of ∼ 3. Spin precession cycle times decrease continuously, become of order a few days or less during the last week prior to merger, and of order hours during the last day of inspiral. Therefore, according to our simple models, we expect spin precession effects to improve the source parameter estimation errors especially during the final two weeks before ISCO. Dur- ing that period of time, in the absence of spin effects, param- eter uncertainties (especially the sky position major axis and the luminosity distance) cease to improve when using only the amplitude modulation. 1 (p1): d- 1 L (1 + cos2 θNL)F I,II+ The best-determined parameters at ISCO are, approxi- mately, the independent detector outputs at ISCO, i.e. the real and imaginary parts of hI,II (Ω) and d- 1 L cos θNLFI,II (Ω) (see Appendix A 4). These are the 4 in- × dependent combinations of 5 physical parameters p1 which correspond to the eigenvectors of the error covariance matrix following the steep evolution ∝ (S/N)- 1 all the way to ISCO. We refer to the fifth independent combination, which is or- thogonal to these best eigenvectors, the "worst" eigenvector, since for this combination, the evolution ceases to improve as (S/N)- 1 within ∼ 0.1× (amplitude modulation cycle time) of merger. It is straightforward to obtain this worst combi- nation explicitly by using the 4 other eigenvectors and Gram- Schmidt orthogonalization (but we have not done this in prac- tice). Since the highest frequency harmonic of the slow modu- lation is for j = 4, the corresponding cycle time is yr/4. Thus, we expect errors will stop improving roughly 1 -- 2 weeks prior to merger. Distributions of errors will quickly broaden during these final stages of observation before ISCO. Simply scal- ing errors with (S/N)- 1, as in the angle-averaged formalism (e.g. [13, 15]), is acceptable if one studies the evolution of parameter errors at tf ∼> 2 weeks, or if one only focuses on the best case parameter combinations. In general, the exponent in the (S/N) scaling decreases as one approaches merger time depending on how close the particular combination of angles considered is to the worst combination. Our findings for the eccentricity evolution of LISA's sky lo- calization error ellipsoid can also be understood with the sim- ple toy models. In fact, we found this behavior to be expected for any model signal with relative instantaneous signal ampli- tude increasing quickly with time, e.g. t- α, α ∼> 2. In this case, the principal axes of the general parameter error ellipsoid sep- arate near tf = 0. There are a limited number of principal errors which rapidly decrease to zero near tf = 0, while others "freeze out" at a time related to a fraction of the cycle time of the par- ticular waveform (∼ 0.1Tcycle if ti > Tcycle). For LISA, there are 5 variable parameters, pslow = (dL, θN, φN, θL, φL), and es- timation uncertainties of 4 combinations of these parameters, L (1 + cos2 θNL)F I,II+ d- 1 (Ω), improve quickly with (S/N)- 1. These combinations correspond to the best 4 principal axes of the 5-dimensional error ellipsoid. The re- maining 5th principal axis does not improve as (S/N)- 1, but rather stops improving at a fraction of the last modulation cy- cle time. The two dimensional sky position error ellipsoid is the projection of the general 5-dimensional error ellipsoid on the (θN, φN) plane. This plane will generally not be aligned with the principal axes of the 5-dimensional ellipsoid. In a typical case, therefore, there will be a nonzero projection on the worst principal component and the sky position ellipsoid will stop shrinking along the worst principal component. This explains why the major axis, 2a, ceases to improve and the eccentricity increases close to merger. (Ω),d- 1 L cos θNLF I,II × According to this argument, it is somewhat surprising to find that the minor axis, 2b, can stop improving much before ISCO. Figure 2 shows that this happens in the worst 10% of all cases for randomly chosen source angular parameters. The reason for this is that, in some cases, not all rapidly improv- ing "best" principal components have a small absolute error at ISCO. For example, consider an edge-on binary inspiral (cos θNL ≈ 0). Since two of the quickly improving parame- ters are simply proportional to cosθNL, the errors will be very large for these parameters. Thus, depending on the relative orientation of the detector and the source at ISCO, there can be large absolute errors in some cases even for the best combi- nations of parameters. In short, both axes of the sky position error ellipsoid can stop improving at late times in those cases when LISA is oriented in its least favorable direction at ISCO. VIII. CONCLUSIONS We have developed a new harmonic mode decomposition (HMD) method to study the feasibility of using LISA inspi- ral signals to locate coalescing SMBH binaries in the sky, as the mergers proceed. According to our extensive HMD survey of potential LISA sources, it will be possible to trig- ger large field-of-view searches for prompt electromagnetic counterparts during the final stages of inspiral and coales- cence. Our results indicate, for instance, that for a typical z ∼ 1 merger event with total mass M ∼ 105 - 107M⊙, a 10- day advance notice will be available to localize the source to within a 10 deg2 region of the sky. The advance notice to lo- calize the source to a 10 times smaller area of 1 deg2 is < 1 22 day for the typical event, suggesting that a wide -- field instru- ment of the LSST class, with a 10 deg2 field-of-view, may of- fer significant advantages over a smaller, 1 deg2 field-of-view instrument for observational efforts to catch prompt electro- magnetic counterparts to SMBH binary inspirals. The robust identification of such electromagnetic coun- terparts would have multiple applications, from an alter- native method to measure cosmological parameters to pre- cise measurements of merger geometries in relation to host galaxy properties [8, 15]. If such electromagnetic counter- part searches can be implemented effectively and successfully, LISA could become an extremely valuable instrument for as- trophysics and cosmology, beyond the original general rela- tivistic measurement goals. Given the advance warning time capabilities established here, effective strategies for electro- magnetic counterpart searches, including the concept of par- tially dedicating a ∼> 10 deg2 field-of-view fast survey instru- ment of the LSST class, are considered in detail in a separate investigation [18]. Acknowledgments We thank Samuel Finn and Shane Larson for influen- tial early discussions on this problem and Scott Hughes and Tom Prince for valuable comments which improved our manuscript. BK acknowledges support from a Smithsonian Astrophysical Observatory Predoctoral Fellowship and from Öveges József Fellowship. ZH acknowledges partial support by NASA through grant NNG04GI88G, by the NSF through grant AST-0307291, and by the Hungarian Ministry of Educa- tion through a György Békésy Fellowship. KM was supported in part by the National Science Foundation under Grant No. PHY05-51164 (at KITP). Z.F. acknowledges support from OTKA through grant nos. T037548, T047042, and T047244. APPENDIX A: SIMPLE TOY MODELS 1. Single Frequency Model First, let us consider the following simple model with two unknowns, c0 and c1, h(t) = c0 + c1 cos(2π f⊕t), (A1) where f⊕ ≡ yr- 1 is fixed and assumed to be known prior to the observation. We call t the "look -- back time" before merger. Let us assume that the relative noise continuously decreases during the observation and that the differential squared signal- - 2(t) = t- 2 to-noise ratio (without modulation) is given by σ in eq. (50). Here t = 0 is a proxy for the "merger". Close to merger, the signal-to-noise ratio accumulates very rapidly. We assume that h(t) is measured in the time interval ti ≥ t ≥ tf, where ti is the start of observation, tf is the end of observation (i.e. x = tmerger - t, xmin = tf, and xmax = ti in eq. [50]). We fix ti and examine the dependence of parameter estimation errors as a function of tf, assuming tf ≪ ti. Note that, for the signal (A1), the fiducial values (c0,c1) drop out when calculating the RMS parameter errors ∆c0 and ∆c1 using eq. (50). More generally, this is true for any sig- nal which is a linear combination of the unknown parameters. All our toy models will have this property and the results pre- sented in this section will be general in that respect. First, let us substitute (A1) in (49) and (50), and evaluate the expected covariance matrix numerically. Figure 8 displays the time dependence of marginalized parameter errors and princi- pal errors. The plots show that the parameter errors all de- crease with the signal to noise ratio when the look -- back time before merger is large. However if the end of the observation is within a certain critical time to merger, tf < tc, only one principal component follows the signal-to-noise ratio. Fig- ure 8 shows that tc ∼ 0.1 yr. The start of the observation in Figure 8 was fixed at ti = 5 yr. It is also interesting to examine what happens for general total observation times, do errors stop improving within some time tc before merger? If yes, how does tc depend on the two timescales ti and f - 1 ? We examine this question numerically, ⊕ substituting (A1) in (49) and (50) and now varying both tf/ f - 1 ⊕ and ti/ f - 1 . Let us define the critical end-of-observation, tc, ⊕ as the time when the marginalized squared parameter error is first within a factor of 2 of its final value. Figure 9 plots the result for the two parameters. Figure 9 shows that tc is determined by f - 1 for large ti, but becomes ti-dependent for ⊕ lower ti values. In the limit ti ≪ f - 1 , the critical look -- back ⊕ time is independent of f - 1 , it becomes a constant fraction of ⊕ ti. Note that, in the limit of an observation extending up to merger, at t = 0, the signal becomes h(0) = c0 + c1 and it has infinite instantaneous signal-to-noise ratio. Therefore, this is the best combination of parameters for which the scaling of errors can follow (S/N)- 1 all the way to t = 0. The worst com- bination is c0 - c1, which stops improving before t = 0. For this simple model, the origin of these features can be understood by analyzing the principal errors and the marginal- ized errors in the error covariance matrix. For this purpose, we present an analytical algebraic solution to this problem. To simplify the equations, let us set the time-scale to f - 1 ⊕ /(2π). In this case the Fisher matrix (50) is t- 2dt Γi j(tf,ti) = R ti R ti R ti cos(t)t- 2dt R ti tf tf tf tf cos(t)t- 2dt cos2(t)t- 2dt ! . (A2) The integrals can be evaluated analytically, Γi j(tf,ti) = (cid:18) where Si(x) =R x 0 1 t + Si(t) cos(t) t cos(t) cos(2t)+1 t 2t + Si(t) + Si(2t)(cid:19)(cid:21) tf ti , (A3) sin(x) x dx is the sine integral. In the next two subsections, we find the limiting behavior of marginalized and principal parameter errors in two different limits: f - 1 , respectively. ⊕ ≪ ti and ti ≪ f - 1 ⊕ We consider the case of a total observation time which is not negligible compared to a cycle time, f - 1 , i.e. tf ≪ ti. We next ⊕ examine two possible cases, f - 1 ⊕ ≪ ti, ⊕ ≪ tf ≪ ti and tf ≪ f - 1 23 δh(t) = δc0 + δc1 cos(2πf⊕t) ti = 5f −1 ⊕ 1000 100 i 2 ) f t ( p δ h 10 1 0.1 0.01 0.001 0.01 0.1 tf /f −1 ⊕ δh(t) = δc0 + δc1 cos(2πf⊕t) ti = 5f −1 ⊕ 1000 100 i 2 ) f t ( v δ h 10 1 0.1 0.01 0.001 0.01 0.1 tf /f −1 ⊕ (S/N )−2 hδc2 0i hδc2 1i 1 (S/N )−2 hδv2 0i hδv2 1i 1 FIG. 8: Marginalized parameter errors (top) and principal errors (bot- tom) for the single frequency model. The green curve shows the scal- ing with inverse squared signal-to-noise ratio, (S/N)- 2, for reference on both plots. A total observation of ti = 5 yr is assumed. Marginal- ized errors follow the signal-to-noise ratio for large tf, but they stop improving within tf < tc ∼ 0.1 yr from merger. Only one eigenvalue scales with the signal-to-noise ratio near merger. a. Long Observations ( f - 1 ⊕ ≪ ti) Here, we assume that the signal has been measured for a very long total time and we concentrate on the effects of changing the end of the observation time, tf, near merger. Therefore, we take the limit ti → ∞, for which + Si(tf) + Si(2tf) ! Γi j(tf) = cos(tf) cos(2tf)+1 cos(tf) 2tf tf tf 1 tf + Si(tf) π/2 π/2 π/2(cid:19) . - (cid:18) 0 (A4) δh(t) = δc0 + δc1 cos(2πf⊕t) errors (i.e. diagonal elements) of (A6) become 1 0.1 0.01 1 − ⊕ f / c t 0.001 0.01 hδc2 0i hδc2 1i 0.1 1 10 ti/f −1 ⊕ FIG. 9: Critical look -- back time, tc, at which parameter errors stop improving. Here tc is defined as the time at which marginalized squared errors are within a factor of 2 of their final values for the first time. separately. First let us assume that the merger is still far away in time in units of a cycle period ( f⊕ ≪ tf ≪ ti). We substitute (A4) in (49) and expand Γ - 1(tf) into a t- 1 series: f (Γ - 1)i j ≈ 1 - tf + cos(2tf)- 1 sin(2tf) 2tf t2 f 1 - sin(2tf) 2tf 2 sin(tf) tf 2 sin(tf) tf 2 ! . (A5) Equation (A5) gives the large tf behavior of marginalized er- rors and correlations, which can be compared to Figure 8 in the appropriate regime, tf > 1 yr. In this case, to leading or- der, all of the squared errors scale with tf, which is the scaling of the inverse squared signal-to-noise ratio, (S/N)- 2, for our noise model. Next, let us examine the case when the end-of-observation time is close to merger, i.e. tf ≪ f⊕ ≪ ti. Now, taking the inverse of the matrix and expanding into a tf series around tf = 0 gives (Γ - 1)i j ≈ 2 π 1 + 2 3π t3 f - 1 + t2 2π2 f f - 1 + t2 2π2 1 + π 2 tf ! , (A6) which gives the short timescale behavior of marginalized er- - 1 define the rors and correlations. The eigenvalues of Γ squared length of the individual principal axes of the parame- ter error ellipsoid, in this case (cid:18) hδv2 1i (cid:19) ≈(cid:18) 0i hδv2 + tf 2 4 π tf 2 16 t2 f 16 + 3π +(cid:0) 5π 2 f (cid:19) . π(cid:1)t2 (A7) Note that, in eqs. (A2)-(A7), time is measured in units of f - 1 ⊕ /(2π). In full units, the squared marginalized parameter .   24 (A8) (A9) tf ⊕(cid:19)3# 3√3/(16π2) f - 1 π(cid:20)1 + tf ⊕(cid:21) π2 f - 1 2 1 4 π f - 1 tf/(cid:0) 1 ⊕(cid:21) ⊕(cid:1) π(cid:20)1 + tf  . π2 f - 1 4 For the eigenvalues (A7), we get (cid:18) hδc2 0i hδc2 2 π"1 +(cid:18) 1i(cid:19) =  1i(cid:19) = (cid:18) hδv2 0i hδv2  16π2 f - 1 ⊕ Equation (A8) implies that the evolution of the marginalized squared error on c0 is very flat for small tf, when the second tf ≪ 3q 3 = 0.267 yr, then rises term is negligible, i.e. steeply (∝ t3 f ). The marginalized squared c1 error is also con- π2 f - 1 stant near merger, for tf ≪ 1 ⊕ ≈ 0.1 yr, and it increases ∝ tf ∼∝ (S/N)- 2 for larger tf. Equation (A9) shows that one of the principal errors has a very different time-evolution: it has no constant term proportional to t0 f . Therefore the semi- minor axis of the error ellipsoid can decrease continuously with the signal to noise ratio. On the other hand, the semi- major axis becomes constant for tf ≪ 4 = 0.4 yr. Since the marginalized errors are nontrivial linear combinations of the principal errors, the constant principal error carries over to both marginalized errors and dominates their evolution. All of these findings are in excellent agreement with the numer- ical results shown in Fig. 8 for tf ≪ 1 yr and in Fig. 10 for ti/ f - 1 ⊕ > 1. It is worth emphasizing that, even if the total observation time had been infinite, ti → ∞, the parameters could not have been estimated to infinite precision in this model. It is not very surprising if one recalls that in this model we defined errors to be infinitely large at infinitely early times (σ2(t)∝ t2). For stationary noise, the contribution of the last cycle to the resultant RMS estimation error for a total observation of Ncyc π2 f - 1 ⊕ cycles is 1/pNcyc. In contrast, rather than the total number of cycles, the typical error during the last cycle dominates the determination of noise, for the particular noise model used here. The main conclusion from this toy model analysis is that errors stop improving close to merger, at tc ∼ 0.1 f - 1 . It can ⊕ - 2(t) = t- α be extended to more general noise models, with σ and α 6= 2. Repeating the calculations for larger α values, we find that parameter estimation errors become more and more insensitive to very early times, tf ≪ t ∼ ti, and that marginal- ized parameter estimation errors cease to improve at some tc, which is now an α-dependent fraction of a single cycle time before merger. For α > 2, we find that errors increase more abruptly at tf ∼> tc, which is consistent with the signal-to-noise ratio being a steeper function of time. On the other hand, for lower α values, parameter estimation errors become more and more sensitive to very early times, tf ≪ t ∼ ti. In this case, the marginalized parameter estimation errors are again very slowly changing for 0 ∼ tf < tc, but the approximate time tc at which parameter errors stop decreasing will be primarily - 25 δh(t) = δc0 + δs1 sin(2πf1t) + δc1 cos(2πf1t) +δs2 sin(2πf2t) + δc2 cos(2πf2t) ti = 5f −1 1 0.001 0.01 0.1 tf /f −1 1 δh(t) = δc0 + δs1 sin(2πf1t) + δc1 cos(2πf1t) +δs2 sin(2πf2t) + δc2 cos(2πf2t) ti = 5f −1 1 (S/N )−2 hδc2 0i hδs2 1i hδc2 1i hδs2 2i hδc2 2i 1 (S/N )−2 hδv2 0i hδv2 1i hδv2 2i hδv2 3i hδv2 4i 1 1000 100 i 2 ) f t ( p δ h 10 1 0.1 0.01 1000 100 i 2 ) f t ( v δ h 10 1 0.1 0.01 determined by ti, rather than by the cycle period f - 1 . The ⊕ > tc is not as abrupt, but extends to several cy- transition at tf ∼ cles. The α = 0 case corresponds to a stationary instantaneous signal-to-noise ratio, with errors scaling slowly as 1/√ti - tf. This case is irrelevant to LISA inspiral signals, which have α ∼ 2 to a good approximation for 1day < t < ti in the rele- vant range of SMBH masses. b. Short observations (ti ≪ f - 1 ⊕ ) Let us now examine the opposite limiting case, where the start of observation time is already within the final cycle be- fore merger. This is relevant to LISA signals, since the ob- servation time of SMBH inspirals is often below a full year, especially for (1 + z) M ≥ 4× 106M⊙. We again restrict ourselves to the case with a total observa- tion time that is non-negligible, i.e. tf ≪ ti. Using time units of f - 1 ⊕ /(2π), expanding (A3) into a series of both ti and tf/ti, we get (Γ - 1)i j ≈ 120 i (10 - t3 ti +tf  t2 30- 10t2 i 10- t2 i - 30- 5t2 i 10- t2 i i )(cid:20)(cid:18) 1 - 1 - 1 1 (cid:19)     - 30- 5t2 i 10- t2 i 30- 5 3 t2 i 10- t2 i (A10) Equation (A10) gives the parameter estimation covariance during the final stages of observation before merger for small total observation times. In this case, the final errors strongly depend on the total observation time. The errors reach their final values when the second term becomes negligible in eq. (A10). To leading order, this happens at tc ∼ ti/3 for both parameters, independently of the cycle time, f - 1 . Equa- ⊕ tion (A10) approximates well the ti dependence of tc shown in Fig. 9 for ti/ f - 1 ⊕ < 0.2 2. Double Frequency Model Now consider a more elaborate model with five unknowns c0, s1, c1, s10, and c10: h(t) = c0 + s1 sin(2π f1t) + c1 cos(2π f1t) +s10 sin(2π f2t) + c10 cos(2π f2t). (A11) Here, the signal is comprised of two different characteristic frequencies, f1 and f2, for which we assume f1 ≪ f2. More- over we assume that f1 and f2 are fixed and known prior to the measurement, e.g. we take f1 ≡ 1 yr- 1 and f2 ≡ 10 yr- 1. We again assume an observation in the look -- back time interval ti ≥ t ≥ tf and take the average instantaneous signal-to-noise ratio to increase as σ(t)- 2 = t- 2. Let us substitute in (49) and (50), and evaluate the expected covariance matrix numerically. Figure 10 displays the results. As in the previous model, these plots show that all parame- ter errors decrease with signal to noise ratio until the last cy- cle and all marginalized errors stop improving beyond some 0.001 0.01 0.1 tf /f −1 1 FIG. 10: Marginalized parameter errors (top) and principal errors (bottom) for the double frequency model. The green curve shows the scaling with (S/N)- 2 for reference on both plots. A total observation time ti = 5 yr is assumed. Marginalized errors follow the signal-to- noise ratio for large tf values, but they stop improving after tf ∼ < 0.1 f , for both frequencies. By comparing the two plots, it is clear that high frequency component errors decouple and that they are determined by two corresponding eigenvalues in the bottom panel. nonzero residual error at late times. Thus, the general trends shown in Fig. 10 are very much similar to the ones in the pre- vious simple model (Fig. 8). Again, contrary to the standard 1/pNcyc expectation, the error during the last cycle domi- nates the total error of the accumulated signal. Moreover, comparing Figs. 8 and 10 shows that the presence of addi- tional independent high frequency degrees of freedom practi- cally does not modify the evolution of marginalized parameter errors associated with low frequency components, if ti > f - 1 1 . During the final cycle, the error ellipsoid becomes "thin" and the narrow dimension will not be aligned with any of the pa- rameters. As a result, this bad principal error dominates each of the marginalized parameter errors at late times. (Note that the start-of-observation time in Figure 10 is ti = 5 yr.) 1 − 1 f / c t 1 − 1 f / c t 1 0.1 0.01 0.001 δh(t) = δc0 + δs1 sin(2πf1t) + δc1 cos(2πf1t) +δs2 sin(2πf2t) + δc2 cos(2πf2t) f2 = 10f1 hδc2 0i hδs2 1i hδc2 1i hδs2 2i hδc2 2i 0.01 0.1 1 10 ti/f −1 1 1 0.1 0.01 0.001 δh(t) = δc0 + δs1 sin(2πf1t) + δc1 cos(2πf1t) hδc2 0i hδs2 1i hδc2 1i 0.01 0.1 1 10 ti/f −1 1 FIG. 11: Critical look -- back time, tc (as in Fig. 9), at which marginal- ized parameter errors stop improving. Top: Only (c0, s1, c1) are al- lowed to vary, using the prior (s10, c10) ≡ (0, 0). Bottom: All 5 pa- rameters (c0, s1, c1, s10, c10) are determined from the observation. For > f - 1 1 , estimations of low frequency parameters (c0, s1, c1) stop im- ti ∼ proving at tc ∼ 0.1 f - 1 1 , while improvement for high frequency pa- rameters occurs all the way to tc ∼ 0.1 f - 1 2 . The critical look -- back time, tc, at which this happens is dif- ferent for the different frequency components. The top panel in Fig. 10 shows that tci ∼ 0.1 fi approximately for both sets of components (s1,c1) and (s10,c10), where fi denotes the corre- sponding frequencies f1 = 1 yr- 1 and f2 = 10 yr- 1, respectively. The bottom panel in Fig. 10 shows that the principal errors separate in three groups. There is one best eigenvector that improves continuously until the end, two that stop improving near tc1 ∼ 0.1 f1 and two that stop improving at tc2 ∼ 0.1 f2. The high frequency parameters (s10,c10) totally decouple from the two worst principal components, (v0,v1), and, as a result, decouple from the low frequency parameters (c0,s1,c1) which are primarily determined by (v0,v1). As for our previous model in § A 1, the critical look -- back time is generally different for different ti values. The bottom panel in Figure 11 shows the time tc at which the squared er- 26 1 rors first double, as a function of ti/ f - 1 1 , as in Fig. 9. Fig. 11 justifies the rule-of-thumb scaling tci ∼ 0.1 fi if the observation time is at least one cycle period, f - 1 1 . The central question for the present analysis is how sensi- tive is the time evolution of low frequency modulation errors to the presence of high frequency components. We can ex- amine this question by computing the critical look -- back time, tc, when the high frequency terms are totally neglected. The top panel in Fig. 9 shows that, if one limits the parameters to (c0,s1,c1), and the total observation time is not smaller than the long-period cycle time, ∼ f - 1 1 , the resulting tc value for pa- rameters c0 and c1 is unchanged at the few percent level. How- ever, if the high frequency components are introduced, the s1 error evolves differently since it asymptotes already at much larger tc values (∼ 0.1 f - 1 rather than ∼ 0.03 f- 1). The reason is that, for small t, with a noise level decreasing quickly, the corresponding function s1 sin(2π f1t) ≈ 2π f1s1t is linearly in- dependent of, and thus uncorrelated with, the functions c0 and c1 cos(2π f1t) which are both constant to first order. Hence, if there are no more unknowns than (c0,s1,c1), then c0 and c1 are correlated while s1 is decoupled and can be determined independently of the other parameters. However, if we add any parameters which are not constant for t ≪ f - 1 1 , then s1 becomes correlated with those. This is exactly what happens in the bottom panel of Fig. 9, when considering the high fre- quency modulations: the estimation on s1 becomes limited for 1 due to the correlations with s10 and c10. Quite similarly, if one introduces any other low-frequency function that is not constant to first order, like s2 sin(4π f1t), then the correlations with this parameter will limit the improvement of estimation errors for s1 at tc ∼ 0.1 f1, even when neglecting the high frequency components. As we shall see, this is the case for LISA: there are generally more than one sin and cos low-frequency modes. In this case, the evolution of estima- tion errors for low frequency parameters can be obtained with the high frequency modes (like s10 and c10) priored out. This justifies our simple intuition: once the signal is decomposed into different time-scale components, the parameter estima- tion problem becomes separable and the evolution of param- eter errors corresponding to different such time-scales can be estimated independently from each other. t ∼< tc1 ∼ 0.1 f - 1 Rather than going through an analytical derivation as in § A 1, we answer one remaining question here: what combina- tion of the original parameters (c0,s1,c1,s10,c10) corresponds to the best principal component, v0, which can be determined extremely accurately at late times, tf → 0? At t = 0, the noise drops to zero. Therefore, the quantity we can measure us- ing the t = 0 information is simply h(t = 0). Looking back at eq. (A11), this is c0 + c1 + c2. It will be interesting to look for similar "best determined combinations" of physical parame- ters for the case of the LISA's realistic signals. 3. Four data-stream models For our final toy model, we insert additional features of a realistic LISA data-stream. We consider five low frequency unknowns, c0, s1, c1,s2, c2, and a high frequency carrier signal 10 1 0.1 i 2 ) f t ( p δ h 10 1 0.1 i 2 ) f t ( v δ h 0.01 0.001 with additional unknowns s10, and c10. Moreover we consider the simultaneous measurement of four data-streams. The sig- nal is h(t) = c0 + s1 sin(2π f1t + ϕs1 i ) + c1 cos(2π f1t + ϕc1 i ) i ) + c2 cos(2π f1t + ϕc2 i ) +s2 sin(2π f1t + ϕs2 +s10 sin(2π f2t) + c10 cos(2π f2t), (A12) i where ϕc1,s1,c2,s2 (i = 1 . . .4) are fixed at a priori randomly cho- sen numbers defining the relative phases of the various modes which are being simultaneously measured. We compute in- dependent Fisher matrices for each four set of ϕc1,s1,c2,s2 . We assume that f1 ≪ f2 and that f1 and f2 are fixed and known prior to the measurement. We choose f2 = 10 f1 and find the evolution of marginalized errors and principal errors in two limits: i (i) neglecting cross-correlations with the high frequency parameters by assuming a prior δs10 = δc10 = 0, and (ii) accounting for these high frequency parameters. We again assume an observation in the look -- back time inter- val ti ≥ t ≥ tf and take the average instantaneous signal-to- noise ratio to increase as σ(t)- 2 = t- 2. The results for these models are shown in Figure 12. Th marginalized errors (top) and principal errors (bottom) are shown for both cases (i) and (ii) above. The figures show that, in agreement with our previous model, uncertainties on the low frequency parameters are not affected by the high fre- quency parameters, except during the final 0.1 cycle time of the high frequency component, 0.1 f - 1 2 . The figures also show that the four principal components of the error ellipsoid im- prove quickly at late times. Marginalized parameter errors improve quickly if they have negligible projection on the bad directions of the error ellip- soid. As a result, our expectation is that errors will typically not stop improving abruptly, but that there will be a shallower evolution in the final two weeks. In the worst case for a given parameter, if it is aligned with the bad ellipsoid principal com- ponent, it will stop improving near merger. In the best case, if the parameter is orthogonal to the bad ellipsoid principal component, it will improve quickly throughout the final days of inspiral. Therefore, we understand that the distribution of errors broadens for tf ≪ 0.1 f - 1 1 . 4. Best Determined Parameters In the previous section, we have shown that, if the noise decreases quickly like t2 near merger (at t = 0), the best- determined parameters are the eigenvectors of the error co- variance matrix that improve with (S/N)- 1. Near merger, these are the independent detector outputs at t = 0. In the case of LISA inspirals, the observation only extends down to ISCO. In this case, the best determined combination of physical parameters p1 at ISCO are the real and imaginary parts of hI,II 1 (p1). To prove this, we have to show that these are uncorrelated and decrease with (S/N)- 1. The functions 27 δh(t) = δc0 + δs1 sin(2πf1t + ϕ1) + δc1 cos(2πf1t + ϕ2) +δs2 sin(4πf1t + ϕ3) + δc2 cos(4πf1t + ϕ4) [+δs10 sin(2πf2t) + δc10 cos(2πf2t)] f2 = 10f1 ti = 2f −1 1 (S/N )−2 hδc2 0i hδs2 1i hδc2 1i hδs2 2i hδc2 2i 0.01 0.1 1 tf /f −1 1 δh(t) = δc0 + δs1 sin(2πf1t + ϕ1) + δc1 cos(2πf1t + ϕ2) +δs2 sin(4πf1t + ϕ3) + δc2 cos(4πf1t + ϕ4) [+δs10 sin(2πf2t) + δc10 cos(2πf2t)] f2 = 10f1 ti = 2f −1 1 (S/N )−2 hδv2 0i hδv2 1i hδv2 2i hδv2 3i hδv2 4i hδv2 5i hδv2 6i 0.01 0.1 1 tf /f −1 1 0.01 0.001 FIG. 12: Marginalized parameter errors (top) and principal errors (bottom) for the four data-stream model. Pairs of curves with the same line style show results for cases with five and seven param- eters. The extra two parameters correspond to high frequency ( f2) components, which affect errors on the other parameters through cor- relations only slightly (factor of 2 . The green curve shows the scaling with inverse squared signal-to-noise ratio, (S/N)- 2, for reference on both plots. A total observation time ti = 2 yr is as- sumed. Marginalized errors follow the signal-to-noise ratio for large tf values. Four principal errors scale with the signal-to-noise ratio near merger. < 2) if tf ∼ ∼ < 0.1 f - 1 hI(t) and hII(t) are uncorrelated by construction, since they correspond to the two independent Michelson detector out- puts (see § II B and Cutler [19]). The real and imaginary parts of one of the detectors, ℜhI 1, are uncorre- lated since they are the coefficients of the high frequency car- rier, sin φGW and cos φGW , for which correlation over one φGW cycle (during which the detector noise is approximately con- stant) is zero. Another way to see this is to focus on the real part in the definition of the Fisher matrix (57), which is expressed as the integral of ℜ[∂ahI,II 1 (t)]. The term in brackets is purely imaginary for the cross correlation of 1(t) and ℑhI 1 (t)∂bhI,II 1 1(t) and ℑhI ℜhI 1, hence the real part is always zero. There- fore, the correlation matrix for ℜhI,II is diagonal. For diagonal terms, the derivatives are 1 and the integrals be- - 2dt, which is exactly (S/N)2. The RMS es- come simplyR σ timation uncertainty of ℜhI,II 1 (p1) follows the (S/N)- 1 all the way down to ISCO. These best combinations are d- 1 1 (p1) and ℑhI,II L (1 + cos2 θNL)FI,II+ 1 ,ℑhI,II (Ω) and d- 1 The evolution of an arbitrary combination of angles will be determined by the projection of this combination on the co- variance matrix eigenvectors. A linear combination of good eigenvectors leads to similarly quick improvement of errors with (S/N)- 1. However, as soon as there is a nonzero projec- tion on the fifth eigenvector, the estimation uncertainty will stop improving at ∼ 0.1Tcycle which, for the highest j = 4 har- monic, is between 1 -- 2 weeks. L cos θNLF I,II × (Ω). APPENDIX B: ANGULAR VARIABLES Here we define the relative angles θNL and φNL, using the polar angles (θN, φN) and (θL, φL) and the corresponding unit vectors N and L. Let us write a rotation around z and y as Oz(φ) and Oy(θ), respectively. Then, z = Oy(- θN)Oz(- φN) N and we define 28   sin(θNL) cos(φNL) sin(θNL) cos(φNL) cos(θNL)   ≡ Oy(- θN)Oz(- φN) L. (B1) This uniquely defines θNL and φNL, which correspond to the relative latitude and longitude, respectively. More explicitly, we get = arccos[sin θN sin θL cos(φL - φN) + cosθN cos θL] , θNL = arccos( N· L) = φNL =(cid:26) 2π - φ0 φ0 where otherwise if (φL - φN)/π ∈ [- 1,0]S[1,2] φ0 = arccos(cid:18) cos θN sin θL cos(φL - φN) - sinθN cosθL sin θNL , (B2) (B3) (cid:19) . (B4) [1] O. Dreyer, B. Kelly, B. Krishnan, L. S. Finn, D. Garrison & R. .../K.Danzmann/KD_LISASymp04.pdf (2004). Lopez-Aleman, Class. Quantum Grav. 21, 787 (2004). [2] M. C. Miller, ApJ 618, 426 (2005). [3] S. A. Hughes & K. Menou, ApJ 623, 689 (2005). [4] E. Berti, V. Cardoso & C. M. Will, Phys. Rev. D73, 064030 (2006). [5] K. G. Arun, Phys. Rev. D74, 024025 (2006). [6] B. F. Schutz, Nature 323, 310 (1986). [7] S. A. Hughes, Mon. Not. Roy. Astron. Soc. 331, 805 (2002). [8] D. E. Holz & S. A. Hughes, ApJ 629, 15 (2005). [9] M. C. Begelman, R. D. Blandford & M. J. Rees, Nature 287, 307 (1980). [10] J. E. Barnes & L. Hernquist, Ann. Rev. Astron. Astrop. 30, 705 (1992). [11] K. Menou, Z. Haiman, & V. K. Narayanan, ApJ 558, 535 (2001). [12] E. Berti, A. Buonanno & C. M. Will, Class. Quant. Grav. 22 S943 (2005). [13] E. Berti, A. Buonanno & C. M. Will, Phys. Rev. D71, 084025 (2005). [14] M. Milosavljevic & E. S. Phinney, ApJ 622, L93 (2005). [15] B. Kocsis, Z. Frei, Z. Haiman, & K. Menou, ApJ 637, 27 (2006). [16] M. Dotti, R. Salvaterra, A. Sesana, M. Colpi & F. Haardt, Mon. Not. Roy. Astron. Soc. 372, 869 (2006). [17] A. Vecchio, Phys. Rev. D70, 042001 (2004). [18] B. Kocsis, Z. Haiman, K. Menou, & Z. Frei, in prep. (2007). [19] C. Cutler, Phys. Rev. D57, 7089 (1998). [20] T. A. Moore & R. W. Hellings, Phys. Rev. D65, 062001 (2002). [21] L. Barack & C. Cutler, Phys. Rev. D69, 082005 (2004). [22] R. Lang & S. A. Hughes, Phys. Rev. D74, 122001 (2006). [23] R. Cornish & L. J. Rubbo, Phys. Rev. D67, 022001 (2003). [24] C. W. Misner, K. S. Thorne, & J. A. Wheeler, Gravitation (San Francisco: Freeman) (1973) [25] E. Poisson & C. M. Will, Phys. Rev. D52, 848 (1995). [26] K. Danzmann, LISA and LISA Pathfinder (Noordwijk: ESA), www.rssd.esa.int/SP/SP/docs/LISASymposium... [27] C. Van Den Broeck and A. S. Sengupta, Class.Quant.Grav. 24 155 (2007). [28] C. Van Den Broeck and A. S. Sengupta, Class.Quant.Grav. 24 1089 (2007). [29] T. A. Prince, M. Tinto, S., L. Larson, & J. W. Armstrong, Phys. Rev. D66, 122002 (2002). [30] P. C. Peters, Phys. Rev. B136, 1224 (1964). [31] P. J. Armitage & P. Natarajan, ApJ 634, 921 (2005). [32] J. C. Papaloizou, R. P. Nelson & F. Masset, A&A 366, 263 (2001). [33] L. S. Finn, Phys. Rev. D46, 5236 (1992). [34] C. Cutler & È. E. Flanagan, Phys. Rev. D49, 2658 (1994). [35] R. Lang & S. A. Hughes, Phys. Rev. D75, 089902E (2007). [36] J. A. Tyson & the LSST collaboration, Proc. SPIE Int. Soc. Opt. Eng. 4836, 10, astro-ph/0302102 (2002). [37] Note that, contrary to our convention for redshifted mass pa- rameters, we drop the z index for f and t because we never consider comoving frequencies or times. [38] www.srl.caltech.edu/∼shane/sensitivity/ [39] Cornish & Rubbo [23] combine the √3/2 factor with the beam patterns FCR,I,II 2 F I,II, but we follow the original definition, where √3/2 appears only when taking the linear combination of GW polarizations (21). = √3 + [40] The reason for the factor 4 is that the mean squared of cos(x) or sin(x) is 1/2 in (18), and since we use one-sided signals in frequency domain ( f > 0), responsible for another factor of 1/2 in comparison. [41] Note that this expression includes both the polarization am- plitude and the polarization phase, as both of these terms are accounted for by the complex harmonic mode coefficients gI,II j (pslow). [42] Note that if pL was intricately coupled to the other angular pa- then it would be impossible to detach the pL part and attach it to P(0),(1),(2)(tf). Fortunately, these terms j (p) rameters in AI,II j from AI,II j were originally included exclusively in the coefficients AI,II in eq. (67) and in a very simple way: the pL terms are found only in the L and L∗ coefficients in gI,II j (pslow) (see eqs. [41,42,67]). The precession phase terms, ∂aδP(pL, pspin,t), can also be sim- ply attached to P(0),(1),(2)(tf). Due to the precession phase, the index k now spans the range 0 ≤ k ≤ 2N + 2. [43] The Fisher matrix method yields phδ p2 i i RMS error for each set of fiducial angles. As an approximation, we identify the dis- tribution of errors with the distribution of RMS errors. 29
0812.1284
2
0812
2008-12-09T07:35:32
Multiple scattering of waves by a pair of gravitationally stratified flux tubes
[ "astro-ph" ]
We study the near-field coupling of a pair of flux tubes embedded in a gravitationally stratified environment. The mutual induction of the near-field {\it jackets} of the two flux tubes can considerably alter the scattering properties of the system, resulting in sizable changes in the magnitudes of scattering coefficients and bizarre trends in the phases. The dominant length scale governing the induction zone turns out to be approximately half the horizontal wave length of the incident mode, a result that fits in quite pleasantly with extant theories of scattering. Higher-$\beta$ flux tubes are more strongly coupled than weaker ones, a consequence of the greater role that the near-field jacket modes play in the such tubes. We also comment on the importance of incorporating the effects of multiple scattering when studying the effects of mode absorption in plage and interpreting related scattering measurements. That the near-field plays such an important role in the scattering process lends encouragement to the eventual goal of observationally resolving sub-wavelength features of flux tubes using techniques of helioseismology.
astro-ph
astro-ph
Multiple scattering of waves by a pair of gravitationally stratified flux tubes Max-Planck-Institut-fur-Sonnensystemforschung, 37191 Katlenburg-Lindau, Germany Shravan M. Hanasoge1 Centre for Stellar and Planetary Astrophysics, Monash University, Victoria 3800, Australia Paul S. Cally ABSTRACT We study the near-field coupling of a pair of flux tubes embedded in a gravitationally stratified environment. The mutual induction of the near-field jackets of the two flux tubes can considerably alter the scattering properties of the system, resulting in sizable changes in the magnitudes of scattering coefficients and bizarre trends in the phases. The dominant length scale governing the induction zone turns out to be approximately half the horizontal wave length of the incident mode, a result that fits in quite pleasantly with extant theories of scattering. Higher-β flux tubes are more strongly coupled than weaker ones, a consequence of the greater role that the near- field jacket modes play in the such tubes. We also comment on the importance of incorporating the effects of multiple scattering when studying the effects of mode absorption in plage and interpreting related scattering measurements. That the near-field plays such an important role in the scattering process lends encouragement to the eventual goal of observationally resolving sub-wavelength features of flux tubes using techniques of helioseismology. Subject headings: Sun: helioseismology -- Sun: interior -- Sun: oscillations -- waves -- hydrodynamics 1. INTRODUCTION An outstanding issue in solar physics concerns the accurate constraining of the internal constitution of sunspots. Since we are unable directly image the interior, we study the solar acoustic wave field in and around sunspots and attempt to comprehend these observations through theories of wave interactions. An example of such an effort is the first putative detection of downflows underneath sunspots by Duvall et al. (1996), who analyzed the solar wave field using methods of time-distance helioseismology (Duvall et al. 1993). In the Sun, observations have almost always been more plentiful than theory. Of late, the importance of developing theoretical and computational methods to aid the interpretation of observations of solar magnetism has risen to the fore. A considerable body of computational work has recently focused on understanding the nature of wave interactions in magnetized environments (e.g. Khomenko & Collados 2006; Cameron et al. 2007; Hanasoge 2008). Complementary to such efforts, we attempt here to develop further the theory of flux tube related multiple scattering, probably ubiquitously present on the Sun but has, for the most part, been studiously ignored in the past due to the many challenges involved in modeling such interactions. On another front, it is important to place bounds on the degree of wave absorption and scattering by the plage for it tells us how much energy is transmitted to the corona and could help explain the complex frequency dependence of acoustic mode linewidths (e.g. Bogdan et al. 1996; Hindman & Jain 2008). Because plage 1Visiting academic at Indian Institute of Astrophysics, Bangalore, India -- 2 -- comprises ensembles of compactly packed thin flux tubes, the interaction with waves possibly lies in the multiple scattering regime. In linear theory, when a wave encounters an anomaly of some sort, it is scattered at constant frequency, with the resultant wave field being broadly classified into near- and far-field components. The far-field consists of propagating modes that transport some fraction of the incident mode energy away from the scatterer. The near-field is more complex, comprising a number of non-propagating horizontally evanescent waves, that arise when the displacements of the anomaly due to the external wave buffeting cannot be matched by the set of eigenfunctions of the propagating modes. In the case of stratified flux tubes in the Sun, the set of p-mode eigenfunctions at constant frequency is an incomplete basis, requiring a supplementary set, an uncountably infinite continuum in fact, of these evanescent near-field functions to complete the basis. The problem is exacerbated when the displacement eigenfunctions of the scatterer gain complexity. The mathematics required to address the near-field in the case of thin flux tubes was set down by Bogdan & Cally (1995), who termed this non-propagating sheath of waves as an "acoustic jacket" that envelopes flux concentrations. The uncountable continuum of jacket modes arises as a consequence of an infinitely deep lower boundary, required in order to allow tube modes to disappear into the solar interior. Unfortunately, numerically computing the near-field jacket in this scenario is all but impossible because of various formidable integrals in the equations. In order to arrive at an analogous but more tractable problem, Barnes & Cally (2000) introduced an artificial lower boundary, leading to a discrete and countably infinite number of near-field modes. More recently, Hanasoge et al. (2008) adopted the model of Barnes & Cally (2000) and estimated the magnitude of the near field jacket and the single scattering by an isolated thin flux tube. They attempted to model the observations of Duvall et al. (2006), who characterized the scattering of f modes by magnetic flux elements. However, magnetic elements are known to consist of a number of tightly packed flux tubes, all likely within the near-fields of each other. Thus the single scattering assumption may not be entirely accurate when interpreting these measurements. The presence of a large body of theory to draw upon makes it easier to proceed towards an understanding of the importance of multiple scattering. When a pair of flux tubes lie in the proximity of each other, their near-fields communicate and depending on the separation, can dramatically alter the nature of the scatter. Thus accounting for mutual induction of near-fields is rather important when studying plage or other closely spaced scatterers. Bogdan & Fox (1991), when considering a pair of flux tubes at a series of separations in an unstratified medium, found evidence for three different scattering regimes that they termed multiple, coherent, and incoherent. The nomenclature points to differences in the degree of coupling between the two flux tubes, with the incoherent regime no different from isolated body scattering and the multiple regime, substantially different. Subsequently Keppens et al. (1994) studied ensembles of flux tubes and found that the degree of absorption was greater in "spaghetti" models than monoliths, pointing to a way of discerning the differences between the two. These efforts have been restricted to unstratified media, mainly due to the considerable mathematical complexity that stratification injects. With the addition of gravity, the external driver and the flux tube displacement eigenfunctions assume distinct and more complicated forms, possibly destroying a number of resonances hitherto possible in the unstratified case. Furthermore, purely analytical techniques cease to be of utility, even when considering single scattering, let alone its multiple counterpart. When studying mode mixing due to thin flux tubes, Hanasoge et al. (2008) were forced to apply a number of methods of linear algebra in order to reliably estimate scattering coefficients and the near-field jacket. On the other hand, multiple scattering is a bit more of a challenge since one must simultaneously determine the wave fields of a number of disparate scatterers, all the while keeping in mind that each scattering coefficient is defined -- 3 -- according to a coordinate system centered on that specific scatterer. Fortunately, the theoretical machinery to study these sorts of problems has been developed decades ago, in the context of fluid mechanics, by e.g., Kagemoto & Yue (1986); Linton & Evans (1990). We adopt these methods in our calculations in order to determine the degree of mode mixing and scattering from a two-tube system. Please note that in the discussions below, the terms "near field", "acoustic jacket" and "envelope of evanescent modes" are used interchangeably. The plan of this paper is as follows. We describe the stratification and basic aspects of the flux tube tube model in §2. The method of Kagemoto & Yue (1986) in the context of interacting thin flux tubes is discussed in §3. The scattering coefficients derived through the application of these techniques for the two-tube system for different incident modes are presented in §4. Finally we summarize and conclude in §5. 2. MODEL The background structure in this calculation, adapted from Bogdan et al. (1996); Hanasoge et al. (2008), is an adiabatically stratified, truncated polytrope with index m = 1.5, gravity g = −2.775 × 104 cm s−2z, reference pressure p0 = 1.21 × 105 g cm−1 s−2, and reference density ρ0 = 2.78 × 10−7 g cm−3, such that the pressure and density variations are given by, and z p(z) = p0(cid:18)− ρ(z) = ρ0(cid:18)− z0(cid:19)m+1 z0(cid:19)m z , (1) (2) . We utilize a right-handed cylindrical co-ordinate system in our calculations, with coordinates x = (r, θ, z) and corresponding unit vectors (r, θ, z). The photospheric level of the background model is at z = 0, with the upper boundary placed at a depth of z0 = 392 km. Following Barnes & Cally (2000), we introduce a lower boundary at a depth of 98 Mm. The displacement potential Ψ(x, t) describing the oscillation modes (t is time) is required to enforce zero Lagrangian pressure perturbation boundary conditions at both boundaries. This upper boundary condition is reflective in nature and therefore, possibly not very realistic. The incoming pn-mode, a plane wave, which expanded in cylindrical coordinates (e.g. Gizon et al. 2006) has a displacement eigenfunction, Ψinc, of the form: Ψinc = ∞ Xm=−∞ where, Φp(κp imJm(kp nr)Φp(κp n; s)ei[mθ−ωt], n; s) = s−1/2−µNn(cid:20)Cp nMκp n,µ(cid:18) sν2 n (cid:19) + Mκp κp n,−µ(cid:18) sν2 n (cid:19)(cid:21) . κp The various symbols in equations (3) and (4) are: µ = m − 1 2 , ν2 = mω2z0 g , kp n = ν2 2κp nz0 , (3) (4) (5) ω the angular frequency of oscillation, s = −z/z0, Jm(w), the Bessel function of order m and argument w and Mκ,µ(w), the Whittaker function (e.g. Whittaker & Watson 1963) with indices κ, µ and argument w. The eigenvalue κp n characterizing the mode are obtained through the procedure n > 0 and constant Cp -- 4 -- described in appendix A of Hanasoge et al. (2008). The n = 0 mode corresponds to the surface gravity or f mode, while n > 0 represents the acoustic pn mode. Note that the lower boundary results in a finite sized box and hence places a restriction on the number of p modes that can fit in this domain. The term Nn is the normalization constant for the mode, defined as Nn ="Z ∞ 1 (cid:20)Cp nMκp n,µ(cid:18) ν2s n (cid:19) + Mκp κp n,−µ(cid:18) ν2s n (cid:19)(cid:21)2 κp ds#−1/2 . (6) The near-field eigenfunctions are also solutions to the same differential equation that governs the prop- agating modes: ζn(κJ n; s) = s−1/2−µ(cid:20)CJ n M−iκJ n,µ(cid:18) iν2 κJ n s(cid:19) + M−iκJ n,−µ(cid:18) iν2 κJ n s(cid:19)(cid:21) . (7) As can be seen, the only difference between the form of the propagating and evanescent mode eigenfunctions is the fact that the roots are now imaginary. In order to determine the roots, we perform a high resolution search for eigenvalues κp,J n ; the task is relatively easy for the propagating mode parameters but is reasonably difficult for the jacket modes because the eigenvalues may be very finely spaced (appendix A of Hanasoge et al. 2008). Subsequently, tables of the propagating and jacket modes are pre-computed at a range of frequencies. Computations of Whittaker functions over large parameter spaces are fairly non-trivial; we use a number of CERNLIB routines to accomplish all these tasks. n and constants Cp,J 2.1. FLUX TUBE Applying the approximations listed in §2 of Bogdan et al. (1996), a thin flux tube carrying a magnetic flux of Φf = 3.88×1017Mx, with constant plasma-β everywhere inside the tube is embedded in the polytrope. The thin flux tube approximation, b(s) ≈s 8πp(s) 1 + β , πR2(s) ≈ Φf b(s) , (8) where b(s) and R(s) are the magnetic field and the radius of the tube at depth s, is shown to be accurate to better than a percent in the truncated polytrope situated below z = −z0 or s = 1 (Bogdan et al. 1996). Note that the magnetic flux associated with the tube is held constant - different values of β therefore result in different b(s) and R(s). The constant-β property of the tube follows from the assumption that thin flux tubes are for all practical purposes in thermal and radiative equilibrium with the external medium. 2.2. Oscillations of the tube For the cases addressed here, we treat only horizontal kink motions of the flux tube (ξ⊥(ω, s)), caused by impinging m = ±1 modes (e.g. Bogdan et al. 1996). The m = ±1 modes affect the tube oscillations according to the differential equation: (cid:20)ω2z0 + 2gs (1 + 2β)(m + 1) ∂2 ∂s2 + g 1 + 2β ∂ ∂s(cid:21) ξ⊥ = 2(1 + β) 1 + 2β ω2z0 ∂Ψinc ∂x , (9) where x = r cos θ. The scattered wave field is computed by matching the horizontal components of the motion of the flux tube to the external oscillation velocities. The manner in which this is accomplished is detailed in the following section. -- 5 -- 3. METHOD Consider a system of randomly distributed flux tubes. The scattered wave field around tube i with the origin of the coordinate system located at the center of the upper boundary of the tube is given by: φS i = − 1 Xm=−1" nP Xn=0 αi mnΦn(κp n; s)H (1) m (kp nri)eimθi + N Xn=nP +1 αi mnζn(κJ n; s)Km(kJ nri)eimθi# , (10) where ζn are near-field eigenfunctions, Φn describe the propagating p-modes, H (1) m (x) and Km(x) are Hankel and K-Bessel functions of order m acting on argument x respectively. nP denotes the number of propagating mode eigenfunctions (a finite number due to the presence of the lower boundary) and the rest corresponding to the evanescent jacket modes. Also note that the m summation is truncated, since thin flux tube theory applies only to interactions with m ≤ 1 waves. Following Kagemoto & Yue (1986), we write this in matrix form: AT inΨS in, (11) φS i =Xn where Ain is a vector (of size 3 × 1) of scattering coefficients for tube i and mode n. The matrix Ψin (size 3 × nz, with nz being the number of points in the z grid) contains the partial wave expansions in terms of H (1) m , Km. In particular, the elements of A, Ψ are: Ain = − ( αi in)cd = H (1) c−2(kp in)cd = Kc−2(kJ (ΨS (ΨS −1n αi 0n αi 1n )T , nri)Φn(κp nri)ζn(κJ n; sd) n; sd) (n ≤ nP ), (n > nP ), (12) (13) (14) where sd is the dth point along the s axis, and the indices c, d run from [1, 3] and [1, nz], respectively. As described in the introductory section, the challenge in computing the wave field interactions lies in simultaneously solving for the scattering coefficients of all the tubes. Tubes that are placed sufficiently far from each other can only interact via the far-field propagating modes since the evanescent jacket has a spatial decay scale of a wave length or so. Thus the coupling is reasonably weak since the amplitude of the expanding far-field modal ring falls rapidly with distance from the scatterer. Stronger interactions occur when tubes lie within each other's near-field induction zones, for the evanescent modes of one influence the tube oscillations on the other and vice-versa. Fairly significant changes in the scattering cross-sections are a consequence of this phenomenon. Thus, to capture this effect, we must compute the wave field around each tube and project its influence on the neighboring flux elements. Since the wave field around each tube is written in a co-ordinate system centered along its axis, suitable co-ordinate transformations must be performed. For this purpose, we employ Graf's addition formulae, which transform cylindrical wave functions between co-ordinate systems (e.g. Abramowitz & Stegun 1964): H (1) m (kp nri)eim(θi−γil) = ∞ Xd=−∞ H (1) m+d(kp nRil)Jd(kp nrl)eid(π−θl+γil), Km(kJ n ri)eim(θi−γil) = ∞ Xd=−∞ Km+d(kJ n Ril)Id(kJ n rl)eid(π−θl+γil), (15) (16) where Im(x) is a Bessel-I function of order m acting on argument x. These equations show how to expand an outgoing scattered wave function from tube i in terms of the incident mode waves of tube l. Here Ril is -- 6 -- Fig. 1. -- The transformation parameters of equations (15) and (16) shown graphically. The tubes i, l are separated by a distance Ril, with the line joining their centers inclined at angle γil to the x-axis of tube i. Note that γli = π − γil. the distance between the centers and γil the angle that the line between the centers of the tubes subtends at the x-axis of the tube i (shown graphically in Figure 1). This relation shows how the scattered wave field from i acts as an incident wave on l. Written in matrix form: ΨS in = Tn ilΨI ln, (17) where T is the transformation operator that relates the scattered wave field of tube i to the resultant incident field on l. We use equations (15) and (16) to build T; the precise distribution of elements is listed in appendix A. Note that the index n that appears in equation (17) denotes a specific modal order, while the superscripts I, S represent the incident and scattered wave expansions. Recalling that equation (11) contains the contributions of various near- and far-field waves and collecting all the n's, we have: AT inTn ilΨI ln. φil =Xn (18) Equation (18) tells us how much the scattered wave field of tube i acts as an incident field on l. Summing up the contributions from the scattered wave fields of all the other flux tubes (except itself, of course) and the zeroth order incident wave, the total incident wave field at l is given by: φI l = Xn = Xn N  φ0ln +  aT ln + Xi=1,i6=l Xi=1,i6=l N ln  ln, lnΨI ln, where aT and φ0ln = aT ln is a vector of coefficients representing the zeroth order incident wave mode of order n, acting on tube l. The determination of the amplitudes of the off-axis incident modes is described in appendix B. Note that aln and Aln have the same matrix sizes; the former refers to the zeroth order incident wave whereas the latter contains the net scattering terms. Finally, to close the equations, we use the diffraction transfer matrix approach of e.g. Kagemoto & Yue (1986) to relate the total incident coefficients AT inTn ilΨI AT inTn il  ΨI (19) (20) -- 7 -- to the scattered ones. In effect we generate a matrix Bl that acts on the incident wave field at l and produces the scattering coefficients. We know the total incident wave field at l: it is given by equation (20). And since Al is the vector of scattering coefficients as seen by the co-ordinate system centered on l, the following relation must hold: or, Al = BlφI l , where we have the following relations: Al = Bl N Xi=1,i6=l TT il Ai  , al + al = ( al0 al1 al2 ... )T , Al = ( Al0 Al1 Al2 Til = T 0 il 0 0 ... 0   0 T 1 il 0 ... 0 0 0 T 2 il ... ... ... )T , 0 ... 0 ... ... 0 ... ... 0 T N il .   (21) (22) (23) (24) (25) (26) The matrix Bl is constructed as follows. The scattering into all other modes is computed for each incident mode (m, n). This includes both propagating pn modes and near-field type incident waves (the Bessel-I functions). Thus Bl contains a full description of the scattering of any given incident mode (m, n) into any (m′, n′) for tube l. More details are described in appendix A. Finally, we discuss the means applied to study the simple case of two interacting thin flux tubes 1, 2. Proceeding from equation (22) we have: A1 = B1(cid:0)a1 + TT A2 = B2(cid:0)a2 + TT A1 = B1(cid:0)a1 + TT 12(cid:3) A1 = B1a1 + B1TT 21A2(cid:1) , 12A1(cid:1) , 21B2(cid:2)a2 + TT 21B2a2. 12A1(cid:3)(cid:1) , 21B2TT (cid:2)I − B1TT (27) (28) (29) (30) We pre-compute the B matrices for a number of different incident mode frequencies and plasma-β values. Finally, having constructed the transformation matrices T at a series of tube separations, we apply the standard MATLAB least-squares algorithm (backslash command) to solve equation (30) for A1 and then for A2. 4. RESULTS First we consider the interaction of a pair of identical flux tubes at different separations with incident waves. The angle between the tubes is set to γ12 = 0, meaning that the tubes lie on the x-axes of the coordinate systems of each other (see Figure 1). Undoubtedly there will be large changes in the scattering coefficients at different angles; for now we stick to this simple case, and leave further exploration of the parameter space to future endeavors. A pictorial representation of an incoming f -mode with respect to the -- 8 -- n, where 2π/kp flux tubes 1,2 is shown in Figure 2. The results of these interactions are shown in Figure 3. A number of conclusions may be drawn: (1) the strongest coupling is between the f mode and the flux tube; there is a rapid fall off with increasing radial order, (2) the length scale of the region occupied by the near-field region well approximated by π/kp n is the horizontal wave length of the incident pn mode, (3) the scattering coefficients attain large values at small flux tube separations; this has a number of consequences for wave absorption in plage, (4) the phases of the scattered waves exhibit highly unpredictable trends, even more so than the magnitudes of the scattering coefficient (Figure 4), and (5) higher-β flux tubes have more extended jackets than their lower-β counterparts; this may be attributed to the fact that when matching tube displacement eigenfunctions to the external waves, the jacket modes play a more significant role in the former case. The last point is further illustrated by the upper two panels of Figure 5 which show the isolated body displacements ξ⊥ of the tubes at plasma-β = 0.1, 1. The fact that a larger number of kinks are seen in the higher-β tube means a larger number of jacket modes must crowd the near field region, resulting in stronger coupling between the tubes. The lower two panels of Figure 5 demonstrate the stark differences between the isolated body and near-field coupled tube displacements; the structure, the number of nodes, and amplitude of the eigenfunction are seen to greatly increase when the jackets of the tubes are able to communicate. This is the very premise of multiple scattering. We also briefly investigate the interactional behavior of a pair of non-identical flux tubes in Figure 6. Because the magnetic flux in each tube is held constant (see §2.1), the two tubes which have differing plasma- β of 0.1,1 also are of different radii, commensurate with equation (8). The scattering coefficient trends are unremarkable, showing differences in structure from those of Figure 3 but do not possess any noticeable features. Despite differences in the flux tube geometries, the coupling remains strong and continues to adhere to the half-wavelength rule-of-thumb near-field dimension. Lastly, in Figure 7, we graph the components of the wave field with the upper panels showing the tubes strongly interacting (a separation distance of 1.085 Mm) while the isolated body case is displayed in the lower panel. In the strong interaction case, the near-field lobes of the two tubes are seen to be in communication, the scattering in one tube significantly influencing the other through this medium. The far-field in both cases are outward spirals, transporting some of the incident mode energy away from the scene of scattering. 5. CONCLUSIONS The work of Bogdan & Zweibel (1987) was among the first efforts to characterize and study the multiple scattering of waves by flux tubes. Since then, further activity in this area was restricted to the study of mul- tiple scattering by unstratified flux tubes (e.g. Bogdan & Fox 1991; Keppens et al. 1994) and subsequently, single scattering by gravitationally stratified tubes (e.g. Bogdan et al. 1996; Hanasoge et al. 2008). Including stratification when attempting to solve the full multiple scattering problem introduces manifold difficulties, especially without the right set of techniques. However, much progress in this regard and the availability of the technique of Kagemoto & Yue (e.g. 1986) has allowed us to attempt this problem. There has been the general thinking that not only does gravity inhibit resonant absorption but may in fact prevent strong interactions between closely spaced flux tubes. The reason for this is the disparity in the depth structure between the eigenfunctions of external modes and the tube displacement; gravity introduces strong structural differences between these two quantities, almost certainly ruling out a strong and direct matching between the two. Thus resonant absorption in stratified magnetized environments may be largely ruled out; but what about multiple scattering and spikes in scattering cross sections and phases? Is the theory of single scattering sufficient to describe the interaction of waves with clustered magnetic elements -- 9 -- Incident m = 1, 4mHz f mode 1 2 0.5 0.4 0.3 0.2 0.1 0 −0.1 −0.2 −0.3 −0.4 −0.5 −4 −3 −2 −1 0 x [Mm] 1 2 3 4 5 ] m M [ y 5 4 3 2 1 0 −1 −2 −3 −4 −5 −5 Fig. 2. -- Example of an incident m = 1 wave: Ψinc = iJ1(kp 0r)eiθ. Tube 1 always sees the incident wave, whereas with increasing separation, tube 2 sees less and less of it. Notice that the presence of the second tube kills the m = ±1 symmetry when the tubes are close to each other. The symmetry is restored for wide tube separations, at which point, both tubes are practically isolated bodies. in the Sun? Certainly not, as demonstrated in §4. Fairly dramatic changes in the scattering coefficients are observed at close separations, of the order of several hundred kilometers, not unlike the distances between flux tubes in plage. The fact that scatter by a pair non-identical flux tubes also exhibits a similar trend (Figure 6) demonstrates the robustness of multiple scattering type interactions. The scattering coefficient is a proxy for the degree of absorption and mode mixing exhibited by the system; if nothing else, the jumps in the coefficients point to a loss of coherence of the incident modes, introducing a mechanism for wave damping. The coefficients behave in a quirky manner, rather similar to those obtained by Bogdan & Fox (1991). Thus drawing out larger behavioral properties from these interactions is somewhat difficult without a more detailed search of the parameter space. Part of this quirkiness may be attributed to the interference between the sizable numbers of modes competing in the excitation of the flux tubes. Through analyses of observations of thousands of small magnetic elements, Duvall et al. (2006) succeeded in estimating the detailed scattering properties of these features, concluding that the m = ±1 waves couple quite strongly with the magnetic tubes. Subsequently, Hanasoge et al. (2008) modeled these measurements in the single scattering limit in an attempt to constrain properties of the average magnetic element. However, these elements consist of a number of thin flux tubes, the scattering presumably in the strongly interacting regime due to their proximity. Merely with two tubes at a sequence of separations, the scattering coefficients and phases display remarkably intricate behavior; extending this to a number of flux tubes at various random locations is arguably a difficult task. The near field in Figure 7 still remains to be directly detected in observations. One possible route towards this goal is to look for statistically significant auto-correlation signals in the vicinity of these small magnetic elements. This follows from the spatially stationary nature of the near-field modes; they merely pulse at the frequency of the incident wave in a thin envelop around the scatterer. However it is unclear how to use this information in a productive manner; evidently a greater understanding of these evanescent x 10−3 f−f scatter, ν = 2 mHz β = 0.1, m = −1 β = 0.1, m = 1 β = 1, m = −1 β = 1, m = 1 4 3.5 3 2.5 2 1.5 1 0.5 -- 10 -- x 10−4 10 −f scatter, ν = 2 mHz p 1 8 6 4 2 2000 4000 6000 8000 10000 2000 4000 6000 8000 10000 f−f scatter, ν = 4 mHz −f scatter, ν = 4 mHz p 1 t i n e c i f f e o c g n i r e t t a c S i t n e c i f f e o c g n i r e t t a c S 10−1 10−2 0 500 1000 1500 2000 2500 Separation distance [km] 10−1 10−2 10−3 3000 3500 4000 0 500 1000 1500 2000 2500 3000 3500 4000 Separation distance [km] Fig. 3. -- Mode-mixing and scattering computed for f and p1 incident waves at two different frequencies and azimuthal orders m = ±1 for a pair of flux tubes at a number of separations. These are the coefficients written according to the co-ordinate system centered around tube 1, the tube that encounters the untransformed zeroth order incident mode. Note that the symmetry of the m = ±1 scattering coefficients are destroyed due to the presence of tube 2; this symmetry is regained when the separation becomes large enough that tube 1 can be considered an isolated body. Clear signs of multiple scattering are observed for separations comparable to about half the horizontal wave length at the surface (∼ π/kp n) of the incident waves, which are 11.04 and 2.76 Mm for the f modes and 25.76 and 6.44 Mm for the p1 modes at 2 and 4 mHz respectively. The degree of scatter exhibits a complex behavior at small separations and more importantly, shows rather large deviations from the isolated body values. Note also the differences between the m = ±1 modes; the presence of the second flux tube destroys any symmetries with respect to these two m waves. Also of interest is the fact that the β = 1 flux tubes couple via the near-field more strongly than the low-β tubes; this is presumably because the displacement eigenfunction exhibits more depth-structure in the former, requiring a greater participation of the near-field modes, and thereby resulting in stronger coupling (see Figure 5). waves is needed. The question relating to sunspot structure, as to whether one can expect to be able to helioseismically discern a monolith from a jelly fish, still shows promise, for it would appear that multiple scattering, despite the presence of strong gravitational stratification, still plays a major role. Our preliminary conclusions, derived from Figures 3 and 4 echo those of Keppens et al. (1994), who suggested that ensembles of flux tubes can absorb quite effectively, but not cause coherent phasing of the scattered waves. This work represents a small first step towards comprehending the action of multiple scattering in stratified magnetized environments. Much work remains to be done in terms of characterizing and under- -- 11 -- f−f scatter, ν = 4 mHz −f scatter, ν = 4 mHz p 1 250 200 150 100 ] s e e r g e d [ e s a h p e d o m d e r e t t a c S 50 1000 2000 3000 Separation [km] β = 0.1, m = −1 β = 0.1, m = 1 β = 1, m = −1 β = 1, m = 1 300 250 200 150 100 50 4000 5000 1000 2000 3000 Separation [km] 4000 5000 Fig. 4. -- Phases of the scattered f mode. The parameter space is similar to that explored in Figure 3. No clear pattern is seen, indicating that interpreting the phases in a multiply scattered wave field is quite a non-trivial affair. standing the interactions of clusters of randomly located flux tubes and the impact on mode linewidths and observations of scattering in plage. The idea for this work occurred over conversations at the CSPA, Monash University, where S. M. H. was a visiting academic in 2008. Much of the computation was performed on the Stanford University solar group machines; thanks to Phil Scherrer for the use of these resources. Some part of this work was accomplished while S.M.H. was at Stanford, he wishes to acknowledge support from NASA grant HMI NAS5-02139. A. THE T and B MATRICES The procedure outlined here is adapted from Kagemoto & Yue (1986). In order to construct Til, we have the following formula for each m, m′ ≤ 1, n: (Til)cd = ei(m−m′)γilH (1) m−m′(knRil) for n ≤ nP , where c = 3n + m′ + 2 and d = 3n + m + 2, and (Til)cd = ei(m−m′)γil(−1)m′−1Km−m′(knRil) (A1) (A2) for n > nP , where c, d are as above. The means of constructing Bj are as follows. For every tuplet (m ≤ 1, n) an incident wave with the appropriate eigenfunction and radial behaviour of unit amplitude is chosen. If n ≤ nP , the incident mode is propagating and has a Jm(kp nr) type horizontal behavior, whereas if n > nP , the horizontal part of the eigenfunction is given by Im(kJ nr). For each incident mode, the scattering into all other modes (at constant frequency and m) is computed. Having thus computed the amplitudes of all the scattered waves for each incident mode, we fill up the matrix Bj as follows: where β is the scatter into the mode (m, n′) by incident wave (m, n). The indices are c = 3n′ + m + 2, d = 3n + m + 2. We only consider m ≤ 1 since the tube is insensitive to other azimuthal orders in this theory. (Bj)cd = βm n′,n, (A3) -- 12 -- Normalized tube displacement eigenfunctions for a 4 mHz, f incident mode β = 0.1, ∆ = 3.25 Mm 10 20 30 40 50 60 70 80 90 100 β = 1, ∆ = 3.25 Mm 10 20 30 40 50 60 70 80 90 100 β = 1, ∆ = 1.20 Mm 10 20 30 40 50 Depth [Mm] 60 70 80 90 100 0.5 0 −0.5 t r a p l a e R −1 0 0.4 0.2 0 −0.2 −0.4 0 10 5 0 −5 −10 0 t r a p l a e R t r a p l a e R Fig. 5. -- Real parts of the normalized tube displacement eigenfunction (ξ⊥/kp n) for flux tube 1 for β = 0.1, 1, and separations, ∆ = 1.20, 3.25 Mm. At this frequency, tubes at a separation distance of 3.25 Mm can be considered isolated bodies. The first aspect take note of is that the displacement eigenfunction of the isolated body β = 0.1 tube has fewer nodes and therefore less structure in depth than the β = 1 tube (upper two panels). Consequently, a greater near-field participation is required to successfully match the β = 1 tube displacements with the external wave eigenfunctions, resulting in a more diverse near-field than the β = 0.1 case. Secondly, the lower two panels show how complicated the tube displacement eigenfunction becomes when the near-fields of the two tubes are in communication. Also, it is important to note that the displacement amplitudes in the strong interaction case are larger by a factor of 10 or more than the isolated body counterpart. This aspect underscores the very premise of multiple scattering. B. TRANSFORMATION OF THE INCIDENT WAVE FIELD We use Graf's addition formula (e.g. Abramowitz & Stegun 1964): Jm(kp nri)eim(θi−γij ) = ∞ Xd=−∞ Jm+d(kp nRij)Jd(kp nrj )eip(π−θj +γij ). (B1) -- 13 -- f−f scatter, ν = 4 mHz x 10−3 −f scatter, ν = 4 mHz p 1 m=−1 m=1 t i n e c i f f e o c g n i r e t t a c S 0.045 0.04 0.035 0.03 0.025 0.02 0.015 0.01 0.005 0 0 1000 7 6 5 4 3 2 1 5000 6000 1000 2000 3000 4000 Separation [km] 2000 3000 Separation [km] 4000 5000 Fig. 6. -- Scattering coefficients of the β = 1 tube for an interacting pair of non-identical β = 0.1, 1 tubes. The behavior is unremarkable, not greatly differing from that in Figure 3. Note that because the magnetic flux in the tube is held constant, changes in β necessarily imply a change in flux tube radius. The incident f and p1 modes have horizontal wavelengths of 2.76 and 6.44 Mm respectively. An m = 0 wave seen at the axis center of tube i produces the following components at flux tube l whose center is located at distance Ril (the m = [−1, 0, 1] waves): ( J1(knRil)eiγil J0(knRil) J−1(knRil)e−iγil ) . Similarly m = ±1 waves seen at tube i produce at l: ( J2(knRil)e2iγil ( J0(knRil) J−1(knRil)e−iγil J−2(knRil)e−2iγil ) , J1(knRil)eiγil J0(knRil) ) , (B2) (B3) (B4) respectively. Since in our theory, the tube is insensitive to m > 1, we do not include these terms in the transformation. Abramowitz, M., & Stegun, I. A. 1964, Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables, ninth dover printing, tenth gpo printing edn. (New York: Dover) REFERENCES Barnes, G., & Cally, P. S. 2000, Sol. Phys., 193, 373 Bogdan, T. J., & Cally, P. S. 1995, ApJ, 453, 919 Bogdan, T. J., & Fox, D. C. 1991, ApJ, 379, 758 Bogdan, T. J., Hindman, B. W., Cally, P. S., & Charbonneau, P. 1996, ApJ, 465, 406 Bogdan, T. J., & Zweibel, E. G. 1987, ApJ, 312, 444 Cameron, R., Gizon, L., & Daiffallah, K. 2007, Astronomische Nachrichten, 328, 313 Duvall, Jr., T. L., Birch, A. C., & Gizon, L. 2006, ApJ, 646, 553 -- 14 -- Strongly interacting far−field Strongly interacting near−field 0.03 h λ / y h λ / y 1.5 1 0.5 0 −0.5 −1 −1.5 1.5 1 0.5 0 −0.5 −1 −1.5 −1.5 −1 −0.5 0 0.5 1 1.5 Isolated body far−field −1.5 −1 −0.5 0 x / λ h 0.5 1 1.5 0.02 0.01 0 −0.01 −0.02 −0.03 0.01 0.005 0 −0.005 −0.01 1.5 1 0.5 0 −0.5 −1 −1.5 1.5 1 0.5 0 −0.5 −1 −1.5 −1.5 −1 −0.5 0 0.5 1 1.5 Isolated body near−field −1.5 −1 −0.5 0 x / λ h 0.5 1 1.5 x 10−3 8 6 4 2 0 −2 −4 −6 −8 x 10−3 5 0 −5 Fig. 7. -- Real parts of the near and far scattered wave fields shown pictorially for a pair of β = 1 flux tubes (upper panels) and the isolated body case (lower panels), both systems lit up by an incident 4 mHz f mode. The axes are x/λh, y/λh, where λh is the horizontal wave length of the incident f mode, 2.76 Mm. The far-field is seen to spiral out whereas the near-field is bound to within a distance of λh/2 of the scatter. Both fields pulsate with a period of 4.16 minutes (4 mHz wave). The tubes are separated by a distance of 1.085 Mm in the upper panels. There may be hope of observationally deducing the near-field modes by studying the wave-field auto-correlations and using this information to determine properties of the scatterer. Duvall, Jr., T. L., Jefferies, S. M., Harvey, J. W., & Pomerantz, M. A. 1993, Nature, 362, 430 Duvall, T. L. J., D'Silva, S., Jefferies, S. M., Harvey, J. W., & Schou, J. 1996, Nature, 379, 235 Gizon, L., Hanasoge, S. M., & Birch, A. C. 2006, ApJ, 643, 549 Hanasoge, S. M. 2008, ApJ, 680, 1457 Hanasoge, S. M., Birch, A. C., Bogdan, T. J., & Gizon, L. 2008, ApJ, 680, 774 Hindman, B. W., & Jain, R. 2008, ApJ, 677, 769 Kagemoto, H., & Yue, D. K. P. 1986, Journal of Fluid Mechanics, 166, 189 Keppens, R., Bogdan, T. J., & Goossens, M. 1994, ApJ, 436, 372 -- 15 -- Khomenko, E., & Collados, M. 2006, ApJ, 653, 739 Linton, C. M., & Evans, D. V. 1990, Journal of Fluid Mechanics, 215, 549 Whittaker, E. T., & Watson, G. N. 1963, A course of modern analysis (Cambridge: University Press, -- c1963, 4th ed.) This preprint was prepared with the AAS LATEX macros v5.2.
astro-ph/9607150
1
9607
1996-07-29T09:45:28
Superheated Microdrops as Cold Dark Matter Detectors
[ "astro-ph", "hep-ph", "nucl-ex" ]
It is shown that under realistic background considerations, an improvement in Cold Dark Matter sensitivity of several orders of magnitude is expected from a detector based on superheated liquid droplets. Such devices are totally insensitive to minimum ionizing radiation while responsive to nuclear recoils of energies ~ few keV. They operate on the same principle as the bubble chamber, but offer unattended, continuous, and safe operation at room temperature and atmospheric pressure.
astro-ph
astro-ph
Phys. Rev. D54 (1996) 1247 Superheated Microdrops as Cold Dark Matter Detectors J. I. Collar Department of Physics and Astronomy University of South Carolina Columbia, South Carolina 29208, USA and Groupe de Physique des Solides (UA 17 CNRS) Universites Denis Diderot (Paris 7) and P. & M. Curie (Paris 6) 2, Place Jussieu, 75251 Paris Cedex 05, FRANCE electronic address: [email protected] PACS numbers: 95.35, 29.40. Abstract: It is shown that under realistic background considerations, an improvement in Cold Dark Matter sensitivity of several orders of magnitude is expected from a detector based on superheated liquid droplets. Such devices are totally insensitive to minimum ionizing radiation while responsive to nuclear recoils of energies ~ few keV. They operate on the same principle as the bubble chamber, but offer unattended, continuous, and safe operation at room temperature and atmospheric pressure. 1 6 9 9 1 l u J 9 2 0 5 1 7 0 6 9 / h p - o r t s a A number of experimental efforts aim at the detection of nuclear recoils produced by the elastic scattering of Weakly Interacting Massive Particles (WIMPs) off target nuclei [1]. The next generation of Cold Dark Matter (CDM) < 1 recoil / kg / day to discover or rule out the detectors will require a sensitivity ~ neutralino, a CDM candidate arising from supersymmetric extensions of the Standard Model [2]. To achieve this, future devices must have the ability to distinguish nuclear recoils in the ~ keV energy range from similar energy depositions by minimum ionizing radiations, still present in ultra-low background underground detectors [3]. Several schemes have been proposed in this respect, such as the simultaneous detection of ionization and phonons [4]; their realisation has proven to be a non-trivial task. In addition to this, large detector masses are necessary for unambiguous WIMP identification through the small modulations > 100 kg is needed to characteristic of the WIMP recoil signal [5-7]: a target mass ~ detect the ~ 5 % yearly change in recoil rate of ref. [7] in a modest period of time (few years). Planned experiments are far from simultaneously meeting these demands. Ideally, a WIMP detector should be sensitive only to recoils like those produced by fast neutrons. One such detector exists; superheated liquid droplet neutron detectors, also known as "bubble detectors" (BDs) and first proposed by Apfel [8], are strictly sensitive to high-Linear Energy Transfer (LET) radiation (the unrestricted LET is the amount of energy dissipated by a radiation per unit path length, dE dx ). Energetic muons, gamma rays, X-rays and beta particles > 200 have a LET well below the activation threshold of BDs, which is typically ~ keV / m m at room temperature. Bubble detectors are totally insensitive to gamma rays of energy < 6 MeV at operating temperatures T < 30(cid:176) C [8-10]. This and several other advantages can make of them an optimal device for neutralino matter searches. A bubble detector consists of a dispersion of droplets (radius 25-75 m m) of a superheated liquid, fixed in a viscous polymer or aqueous gel [8,11]. Several techniques exist for the production of BDs [12,13]. The metastable superheated state is maintained indefinitely at room temperature and 1 atm. Mechanical energy is stored in each droplet, which behaves like a miniature bubble chamber. The energy deposition of a particle can release this energy, triggering the vaporization of the droplet and forming a visible gas bubble (diameter ~ 1 mm). Depending on the viscoelasticity of the medium, the bubbles remain fixed in it, providing a visual record of the radiation dose and simple optical reading, or rise 2 to collect above it, where the volume of evolved gas can be measured. Alternatively, the sound released by the sudden vaporization can be registered by a piezoelectric transducer [12,14]. The performance of BDs has been measured over a period of four years, without a significant loss of sensitivity [9]. Polymer- based BDs do not display shock-induced bubbles when dropped from heights of several meters [15]. The difficulties inherent to the handling of large volumes of superheated liquid are absent in BDs; practically, conventional bubble chambers used in nuclear experiments are "dirty" in the sense that there is homogeneous nucleation caused by large scale statistical density fluctuations in the bulk of the liquid and heterogeneous nucleation from contact with the container walls, gaskets, etc. This boiling tends to recompress the liquid and the bubble chamber needs to be pressure-cycled every few seconds. The low event rate necessary for a competitive dark matter experiment seems difficult to obtain in bubble chamber proposals [16]. Safety concerns such as explosive boiling of a large volume of superheated liquid (contact vapour explosion) disappear; the vaporization of a single droplet does not generate avalanche effects and BDs operate in a passive, unattended fashion, with no external power supply. The probability of spontaneous vaporization of droplets is extremely small; a droplet of radius as large as ~ 1 mm is stable against homogeneous nucleation for > 106 y at T as high as 50(cid:176) C [13,17]. The rate of spontaneous bubble formation in commercial BD personal dosimeters is accounted for by the cosmic neutron flux [9,18]. The volume filling factor of the superheated liquid is kept low (~1 %) in commercial BDs to avoid an excessive response to this background. The mechanism of bubble nucleation under irradiation has been studied by several authors [10,12,17,19,20] and is based on the thermal spike model of Seitz [21]; an intense energy deposition along a particle's path can provide enough localized heating to create bubbles of a critical size or larger. If a vapour bubble grows larger than a critical radius rc (T) (~ few tens of nm [19]), it becomes thermodynamically unstable and continues to expand evaporating all of the droplet's liquid. The conditions necessary for radiation induced nucleation are two [17,19]: (1) The total energy deposited must be larger than a critical energy for bubble formation, Ec (T), computed from the sum of the thermodynamically reversible processes of vaporization, formation of bubble surface and bubble expansion against the gel, and 3 (2) the stopping power of the particle in the target material must be such that Ec (T) is deposited within a small distance L (T) of order rc : (dE / dx) L(T) ‡ Ec (T) . (1) This second condition is responsible for the insensitivity of BDs to low-LET radiations. Note that the response of BDs does not depend on droplet size, as long as they are not smaller than rc and large enough to produce visible bubbles if optical reading is used. Several liquids have been tested in BDs [10]. Freon®-12 ( CCl2F2) [22] is by 5 far the best for a CDM search due to its very low critical energy; e.g., Ec (30(cid:176) C) » keV, Ec (20(cid:176) C) » 16 keV, increasing to Ec (0(cid:176) C) » 200 keV [19]. The response of Freon-12 BDs to thermal, fission, and monochromatic neutrons has been investigated and is in good agreement with theoretical models [10,12,17,19]. Freon-12 is the main concern of this paper. The value of L(T) for Freon-12 has been measured using a 252Cf neutron source [17,19]. It varies in the range 1.0 rc - 1.5 rc for T between 0(cid:176) C - 35(cid:176) C. Figure 1 displays the requirements for bubble nucleation in Freon-12 at different T. The total stopping power of alpha particles and recoiling atoms in the liquid, obtained from the code TRIM92 [23], is also shown. While a Freon-12 BD is sensitive to Cl recoils down to Ec in the temperature range envisioned for a CDM search (~20(cid:176) C - 30(cid:176) C), this is not the case for F or C recoils. For spin-independent coherent WIMP interactions, where the scattering cross section varies with the square of the number of nucleons in the target nucleus, this feature does not reduce the expected WIMP bubble production rate by much, Cl being the heaviest of all Freon-12 components. However, one advantage of this detector resides in its high fluorine content; this nucleus has the largest expected interaction rate for axial-vector (spin-dependent) neutralino scattering [24]. Operating temperatures close to 30(cid:176) C must be used in order to maximize the sensitivity to this sector of the neutralino parameter space. Above ~ 27(cid:176) C the radiopurity of the gel becomes a concern [18], due to the possibility that alpha particles from U and Th contaminants produce bubble nucleation. The absence of alpha-induced nucleations below 25 - 30(cid:176) C has been recently demonstrated using actinide-doped BDs [25]. nucleation [10]: (1) elastic scattering, (2) inelastic scattering, (3) Several types of neutron interactions meet both requirements for bubble 35Cl(n, p)35S, 4 35Cl(n,a )32P . The last two are predominant for thermal neutrons. A and (4) simple model has been developed to predict the response of BDs to neutrons [10,12,17,20]; this energy-dependent response function, R (En ) , is calculated as: R (En ) = y (En ) V Ni (En )Sij (En , T), ij (2) i j where En is the incident neutron energy, (En ) is the neutron fluence, V is the total volume of superheated liquid, Ni is the number of atoms per unit volume of ij(En ) is the neutron cross section of the jth the ith atomic species in the liquid, Sij(En , T) is the "superheat factor", interaction with the ith atomic species, and i.e., the fraction of the recoil nuclei with kinetic energy above the minimum (threshold) energy, Ethr(T), that satisfies both requirements for bubble nucleation. Calibrations using monochromatic neutrons are in excellent agreement with this model's predictions [10]. The Freon-12 response function has an advantageous feature for a CDM search: it drastically drops from an approximately constant value of P(En ) ~ 10 En ~ P(En ) ~ 10 the underground neutron spectrum by simply surrounding the detector with - 1 (in units of bubbles / neutron per cm2 / cm3 of Freon-12) for - 6 for En ~ few tens of keV, slowly increasing to - 4 for thermal En [10]. This allows for very efficient shielding of > 200 keV, to P(En ) ~ 10 - 3- 10 water, which can then be used to simultaneously regulate the operating temperature (as in a double-bath). Fast neutrons, to which Freon-12 is most sensitive, fall after moderation in the region of diminished P(En ) . The predictability of the response of BDs to neutron recoils makes one confident that equally reliable predictions can be generated for WIMP-induced recoils. The temperature dependence of the threshold energy Ethr allows for the measurement of a differential recoil energy spectrum by running the detector at different temperatures. This differential rate depends strongly on the mass and coupling of the scattered particle and can be used to differentiate a true WIMP signal from the backgrounds discussed below. Elastic scattering is by far the most important mode of WIMP interaction and Eq. (2) is thereby simplified by the removal of the jth index. The recoil energy, Erec , transferred to the ith atomic species is determined from the differential cross , of the particular WIMP candidate under consideration. A heavy section, ds dErec 5 s (cid:229) (cid:229) y s "neutrino", i.e., a generic massive neutral particle is used in this calculation, having the differential cross section for elastic scattering [26]: ds dErec (cid:181) G N2 2 mR max (mc ) Erec F(q2) , (3) max (mc ) is the maximum recoil energy for a given WIMP mass mc Erec where G is a coupling constant, mR is the reduced mass of the WIMP-nucleus system, , and N is the number of neutrons in the target nucleus. The term F(q2) is a form factor accounting for the loss of coherence for very massive WIMPs. The exponential approximation in ref. [3], F(q2 ) = exp - 2(cid:215) M (cid:215) Erec 3(cid:215) h2 e 2 , (4) is adequate even for the heaviest nuclei [27] ( e is the nuclear radius and M is the mass of the recoiling nucleus). The choice of a heavy "neutrino" as the WIMP is justified by the simplicity in the computation of its rate of interaction and energy transfer to the nucleus (neutralino cross sections are far more parameter- intensive). Moreover, the scalar neutralino coupling from Higgs boson exchange 2 (N + Z)2, Z is the atomic number) prevails over spin-dependent [28] (s ~ mR channels for most neutralinos with a Zino-Higgsino mixture [29,30]; their differential scattering cross-section depends on Erec only through the same form factor of Eq. (4) [27,30]. In this simple case, F(q2) alone defines the energy distribution of recoiling nuclei. The collision kinematics for these neutralinos and heavy "neutrinos" are then the same to a good approximation, and the rates of bubble nucleation presented below apply to this important neutralino sector after scaling by the coupling constant in Eq. (3). The rate of interaction at a given recoil energy is computed by weighted ds dErec integration of as is customary in WIMP direct searches [3]. The relevant halo parameters used here are vEarth= 260 km / s (annual average of Earth's = 270 km / s (dispersion in the galactic CDM speeds), speed through the halo), vdis » 0.4 GeV / c2 / cm3 vesc (local halo density) [3]. The superheat factor for WIMP recoils, Si , is equal to the = 550 km / s (local galactic escape velocity) and halo 6 (cid:215) (cid:230) Ł (cid:231) (cid:246) ł (cid:247) r recoil rate between Ethr and shown in fig. 2 for different T; note that Ethr for carbon is so large for T ~ (fig. 1) that few nucleations from carbon recoils are expected. The large change in max (mc ), divided by the total recoil rate. This Si is Erec < 25(cid:176) C SF from T= 25(cid:176) C to 30(cid:176) C is due to the very different Ethr values for F (~110 keV and 5 keV, respectively). The equivalent of Eq. (2) can be employed now to obtain the rate of WIMP bubble nucleation, using the averaged values of WIMP fluence and cross section over their velocity distribution in the Earth's frame [3]. A max for slightly more accurate method is to integrate the recoil rate from Ethr to each atomic species, weighting the results by the species' abundance and summing Erec them. The so obtained response function is plotted in fig. 3 for a heavy Dirac neutrino. At T= 25(cid:176) C, the seasonal change in bubble production from the yearly modulation of ref. [7] is ~ 5% for (~ mc = 10 GeV / c2 ) [18]. > 100 GeV / c2 and larger for lighter ~ 30% for mc mc Perhaps the simplest implementation of BDs as WIMP detectors will consist of modular containers of dimension ~1 m x 1 m x 0.1 m, filled with BD, and immersed in a temperature-controlled water double-bath in an underground laboratory. Such a volume contains, even at the low commercial filling factor, a target mass ~ 1 kg. With large enough BD masses the technique can be extrapolated to the detection of solar and stellar collapse neutrinos [18], taking advantage of the coherent enhancement to their neutral-current scattering cross section [31]. Even in such an environment, certain sources of high-LET background radiation will be unavoidable. A typical [32] underground neutron flux has been n ~ 3.8 (cid:215) 10 - 6 cm measured in the Gran Sasso laboratory (depth ~ 3800 meters of water equivalent, < 25 MeV . This m.w.e.) [33]. This flux is in rocks, is expected - 1 [34] and the attenuation length for fast neutrons in Nn ~ 2.5(cid:215) 10 = 30 g / cm2 [35]; the flux inside a (thick) 4p "shield" (rock walls) granite, from the typical neutron production rate - 7 neutrons g-1 s - 2 s-1 for 10 - 3 eV < En granite is approximately given by: » Nn l ~ 7.5(cid:215) 10 - 6 cm - 2 s-1, n (5) in rough agreement with the observed value. Folding the measured differential n of ref. [33] with the response P(En ) in ref. [10], one obtains the expected rate 7 F l F F of bubble production by the unmoderated neutron flux in a typical underground laboratory (fig. 3). The water-moderated flux, 'n , is calculated by means of the mass removal cross section, i.e., l - 1 [35]: = F 'n - t /l n e (6) where t is the distance travelled in water and = 10.1 g / cm2 [35]. Strictly H2O - 6 cm - 2 s-1 for En ~ speaking, Eq. (6) applies only to the fast component of the neutron flux (~ 0.7 (cid:215) 10 > 200 keV [33], i.e., in the energy region of maximum P(En ) ); the moderated neutrons can still produce nucleations, albeit with a largely - 4 (see above) is assumed for diminished P(En ) . Here a conservative P(En ) = 10 those. The obtained expected rate of bubble nucleation is shown in fig. 3 for a modest t ~ 70 g / cm2 . A more precise Monte Carlo simulation of the moderated neutron spectrum, able to determine the optimal value of t, is underway [18]. - 1 day n ~ 3.5(cid:215) 10 - 11cm Another concern is the neutron flux from U and Th impurities in the water shielding itself; the contribution from (a , n) reactions is ~ 5 times larger than that 238U [34,36]. Concentrations of U and Th as low as ~ from spontaneous fission of 0.01 ppb are common even in tap water [37]. This gives an equilibrium alpha - 1. Taking a representative (a emission of ~ 300 kg per 106 alphas [38], a production rate of Nn ~ 3.5(cid:215) 10 obtained. Use of Eq. (5) for a 4p , n) yield of ~ 1 neutron - 12 neutrons g-1 s - 1 is water shield gives a neutron flux at the BD of - 2 s-1 from water radioimpurities. Allowing the maximum - 4 bubbles / kg of response to these neutrons, P(En ) ~ 10 Freon-12 / day (fig. 3). Finally, the neutron production by muons in rock is - 1 at 3200 m.w.e. [34]; from the systematics of the Nn ~ 8 (cid:215) 10 underground rate of production of neutrons via (m ,n) in different materials [39], - 4 this production rate should be down by a factor ~ 4 in water, resulting in ~ 1(cid:215) 10 bubbles / kg of Freon-12 / day from this channel (fig. 3). According to the classical theory of homogeneous nucleation [40], a temperature T > 90 (cid:176) C would be needed to provoke this same rate spontaneously in superheated Freon-12 at 1 atm. - 1, results in ~ 2(cid:215) 10 - 12 neutrons g-1 s Two other sources of background, internal to Freon-12, must be contemplated. First, a recoiling daughter from alpha emission in U and Th 206Pb alpha-recoil has an impurities is able to produce nucleation. For instance, a energy ~ 103 keV, range ~ 0.08 m m and dE dx ~ 1.8 MeV / m m in liquid Freon-12 - 1 is expected for an equilibrium [23]. An alpha-decay rate of ~ 3(cid:215) 104 kg - 1day 8 F F l F concentration of U and Th of 1 ppb. Fortunately, radioimpurities in cryogenic - 15g/g [41]; two alpha decays have been liquids can be frozen-out to levels < 10 registered in a 6 kg liquid Xe detector after ~ 100 days [42]. A similar radiopurity in Freon-12 yields a nucleation rate comparable to that from the moderated neutron flux of fig. 3. At this level of radiopurity, a contribution from fission 36S are high-LET = 3(cid:215) 105 y) is expected to be present in small concentrations in Freon-12. The K-shell binding energy in S fragments is entirely negligible. Finally, Auger electrons from radiations, able to produce bubble nucleation; 36Cl ( T1/2 (2.5 keV) is an upper bound to the energy deposited by this process. This is below Ec (30(cid:176) C) and does not interfere with this search. Second order processes such as fission fragments originating in the gel and alpha reaction products will be treated elsewhere [18]. An expected WIMP exclusion plot can be generated by taking the largest of all the above background estimates (,i.e., moderated neutrons) as the signal detected. This is shown in fig. 4, together with present exclusions from underground Ge detectors [3]. Bubble detectors will be able to explore, in principle, a vast region of the neutralino parameter space. In conclusion, BDs are mature enough to offer an excellent alternative for WIMP direct detection. The simplicity and low cost of these detectors, together with their inherent background rejection and the possibility of using large target masses, promise a great leap in CDM sensitivity. The development of a dedicated bubble detector has started in the framework of the Paris-Zaragoza-Lisbon-South Carolina collaboration. Acknowledgements: I have profited from conversations with F.T. Avignone, T. Girard, A. Morales, J. Morales and G. Waysand. I thank B. Sur for calling my attention to recent work by Zacek [16] before the completion of this work. I am indebted to the Fondation Robert Schuman for support during the 1994-1995 academic year and to the Groupe de Physique des Solides for their hospitality during the completion of this work. References: [1] [2] P. F. Smith and J. D. Lewin, Phys. Rep. 187 (1990) 203. G. Jungman, M. Kamionkowski and K. Griest, hep-ph /9506380 (Phys. Rep., in press). 9 [3] S.P. Ahlen et al, Phys. Lett. B195 (1987) 603; E. Garcia et al, Phys. Rev. [4] [5] D51 (1995) 1458, and refs. therein. T. Shutt et al, Phys. Rev. Lett. 69 (1992) 3531. J. I. Collar and F. T. Avignone, III, Phys. Lett. B275 (1992) 181; Phys. Rev. D47 (1993) 5238; Astropart. Phys. 3 (1995) 37. [6] J. R. Primack, D. Seckel and B. Sadoulet, Ann. Rev. Nucl. Part. Sci. 38 (1988) 751. [7] A.K. Drukier, K. Freese and D.N. Spergel, Phys. Rev. D33 (1986) 3495; K. Freese, J. Frieman, and A. Gould, Phys. Rev. D37 (1988) 3388. [8] R. E. Apfel, Nucl. Instr. & Meth. 162 (1979) 603; Nucl. Instr. & Meth. 179 (1981) 615. [9] J. A. Sawicki, Nucl. Instr. & Meth. A336 (1993) 215. [10] Y.-C. Lo and R. Apfel, Phys. Rev. A38 (1988) 5260. [11] H. Ing and H. C. Birnboim, Nucl. Tracks and Rad. Meas. 8 (1984) 285. [12] Y.-C. Lo, Ph. D. dissertation, Yale University, 1987. [13] R. E. Apfel, U.S. Patent No. 4,143,274 (Mar. 6, 1979). [14] R. E. Apfel and S. C. Roy, Rev. Sci. Instrum. 54 (1983) 1397. [15] T. Cousins, K. Tremblay and H. Ing, IEEE Trans. Nucl. Sci. 37 (1990) 1769. [16] V. Zacek, Nuovo Cimento A107 (1994) 291. [17] M. J. Harper, Ph. D. dissertation, University of Maryland, 1991. [18] J. I. Collar et al, (in preparation). [19] M. J. Harper and J. C. Rich, Nucl. Instr. & Meth. A336 (1993) 220. [20] R. E. Apfel, S. C. Roy and Y.-C. Lo, Phys. Rev. A31 (1985) 3194. [21] F. Seitz, Phys. Fluids 1 (1958) 2. [22] Freon-12 is a trademark of the Du Pont Co. [23] J. F. Ziegler, J. P. Biersack and U. Littmark, The Stopping and Range of Ions in Solids (Pergamon Press, Oxford, 1985). [24] J. Ellis, Nuovo Cimento A107 (1994) 1091. [25] C. K. Wang, W. Lim and L. K. Pan, Nucl. Instr. & Meth. A353 (1994) 524. [26] M. W. Goodman and E. Witten, Phys. Rev. D31 (1985) 3059. [27] J. Engel, Nucl. Phys. B (Proc. Suppl.) 28A (1992) 310. [28] R. Barbieri, M. Frigeni and G. F. Guidice, Nucl. Phys. B313 (1989) 725. [29] G.B. Gelmini, P. Gondolo and E. Roulet, Nucl. Phys. B351 (1991) 623. [30] A. Bottino et al, Phys. Lett. B295 (1992) 330. 10 [31] J.N. Bahcall, Neutrino Astrophysics (Cambridge University Press, Cambridge, 1989). [32] For depths > few tens of m.w.e., the largest contribution to n comes from natural radioactivity in the rock. The neutron flux in the Frejus tunnel < 5 MeV is similar to the equivalent Gran (1780 m.w.e.) for 1 MeV < En Sasso measurement (M. Chapellier, private comm.). A consistent flux has been observed at 3200 m.w.e. [33]. [33] P. Belli et al, Nuovo Cimento A101 (1989) 959. [34] S. R. Hashemi-Nezhad and L.S. Peak, Nucl. Instr. & Meth. A357 (1995) 524. [35] C. F. Cook and T. R. Strayhorn, Fast Neutron Physics, part I, in Monographs and Texts in Physics and Astronomy, vol IV, Ed. by J. B. Marion and J. L. Fowler (Interscience, New York, 1960). [36] A. Da Silva et al, Nucl. Instr. & Meth. A354 (1995) 553. [37] V.Bansal, A.Rawat, P.J.Jojo and R.Prasad, Health Phys. 62 (1992) 257. [38] Y. Feige, B.G. Oltman and J. Kastner, J. Geophys. Res. 73 (1968) 3135. [39] G. V. Gorshkov, V. A. Zyabkin and R. M. Yakovlev, Sov. J. Nucl. Phys. 13 (1971) 450; B. V. Pritychenko, preprint DKM.59 (1992). [40] M. Blander and J. L. Katz, AIChE Journal 21 (1975) 833. [41] G. Ranucci et al (Borexino collab.), Nucl. Instr. & Meth. A315 (1992) 229. [42] A. Incicchitti (BPRS-Lyon-DAMA collaboration), private comm. [43] E. Fiorini, J. Low Temp. Phys 93 (1993) 189. 11 F Fig. 1 Linear Energy Transfer of recoil atoms and alpha particles in liquid Freon-12 r = 1.3 g / cm3 ), as extracted from TRIM92 [23]. Horizontal and vertical bars ( mark the minimum LET and critical energy, Ec , respectively, necessary for bubble nucleation (see text). At a given temperature, only recoils falling in the upper right quadrant inscribed by the corresponding lines can produce droplet vaporization; the intersection of the curves and the boundaries of the quadrants defines the threshold energy, Ethr(T), for each atomic species. Fig. 2 Superheat factor for recoils in Freon-12 (dotted lines = F, dash-dot = C, solid = Cl) from WIMP scattering via a scalar (spin-independent) coupling, as a function of the WIMP mass. These curves are common for heavy neutrinos and neutralinos with a Zino-Higgsino mixture. The operating temperature of the bubble detector is indicated for each curve. Fig. 3 Bubble nucleation rate per kg of Freon-12 produced by a heavy Dirac neutrino » 0.4 GeV / c2 / cm3). The response to other composing the galactic halo ( particles with predominant spin-independent couplings is scaled down by their halo coupling constant, G (Eq. (3)). The characteristic variation of the response with operating temperature allows for the identification of the WIMP mass, . The mc horizontal lines mark the nucleation rates expected from high-LET backgrounds existing in an underground laboratory (see text). For comparison purposes, the sea-level response to cosmic neutrons is ~ 8 (cid:215) 103 bubbles / kg / day. Bubbles produced by a -particles emitted in the gel are not included (they produce nucleation only at T ~ Fig. 4 > 27(cid:176) C [25]). Expected exclusion plot for WIMPs with scalar couplings from a Freon-12 BD after ~ 1 kg-y of data acquisition (solid lines, T as indicated). The water-moderated underground neutron spectrum of fig. 3 is conservatively assumed to be the predominant background. Freon cross sections are normalized to Ge for comparison purposes. Also shown: a) present exclusion limits from underground Ge experiments (UZ/PNL/USC collaboration [3]), b) expected improvement that could be obtained with a similar Ge detector able to reject 99 % of the low-LET background, c) cross section for a heavy Dirac neutrino (coherence loss included), 12 r and (shaded) regions populated by some neutralino candidates; d) minimal SUSY, e) GUT [43]. 13 ) m m / v e k ( x d / E d fig. 1 fig. 2 1 03 30(cid:176) C 25(cid:176) C 20(cid:176) C Cl F C 1 02 20(cid:176) C 25(cid:176) C 30(cid:176) C 1 00 1 01 1 02 1 03 kinetic energy (keV) 1 00 i S , r o t c a f t a e h r e p u s 1 0-1 1 0-2 3 0(cid:176) 3 0(cid:176) 2 5(cid:176) 2 0(cid:176) 1 01 3 0(cid:176) 2 5(cid:176) 1 02 1 03 1 04 mc (GeV / c2) 1 05 14 a fig. 3 fig. 4 y a d / 2 1 - n o e r F g k / s e l b b u b 1 02 30(cid:176) C 1 01 25(cid:176) C 1 00 20(cid:176) C 1 0-1 1 0-2 1 0-3 1 0-4 ambient neutrons ambient + 70 cm H2O (a ,n), (m ,n), nfission 1 01 1 02 1 03 1 04 mc (GeV / c2) 1 05 ) 2 m c ( n o i t c e s s s o r c 1 0-32 1 0-33 1 0-34 1 0-35 1 0-36 1 0-37 c ) 20(cid:176) C 25(cid:176) C a) b) e) d) 1 01 1 02 1 03 1 04 mc (GeV / c2) 1 05 15
astro-ph/0210360
1
0210
2002-10-16T09:10:44
XMM-Newton observations of the nearby brown dwarf LP 944-20
[ "astro-ph" ]
The nearby (d=5.0 pc) brown dwarf LP944-20 was observed with the XMM-Newton satellite on 07 January 2001. The target was detected with the Optical Monitor (V=16.736$\pm$0.081), but it was not detected during the $\approx 48$ ks observation with the X-ray telescopes. We determine a $3\sigma$ upper limit for the X-ray emission from this object of $L_{X}<3.1 \times 10^{23}$ $ergs \cdot s^{-1}$, equivalent to a luminosity ratio upper limit of $\log{(L_{X}/L_{bol})} \le -6.28$. This measurement improves by a factor of 3 the previous \emph{Chandra} limit on the quiescent X-ray flux. This is the most sensitive limit ever obtained on the quiescent X-ray emission of a brown dwarf. Combining the XMM data with previous \emph{ROSAT} and \emph{Chandra} data, we derive flare duty cycles as a function of their luminosities. We find that very strong flares [Log$(L_X/L_{bol})>-$2.5] are very rare (less than 0.7% of the time). Flares like the one detected by Chandra [Log$(L_X/L_{bol})=-$4.1] have a duty cycle of about 6%, which is lower than the radio flare duty cycle ($\sim$13%). When compared with other M dwarfs, LP944-20 appears to be rather inactive in X-rays despite of its relative youth, fast rotation and its moderately strong activity at radio wavelengths.
astro-ph
astro-ph
XMM-Newton observations of the nearby brown dwarf LP 944-20 Institute for Astronomy, 2680 Woodlawn Drive, Honolulu Hawai'i, 96822, USA Eduardo L. Mart´ın 2 0 0 2 t c O 6 1 1 v 0 6 3 0 1 2 0 / h p - o r t s a : v i X r a [email protected] and Herv´e Bouy E.S.O, Karl Schwarzschildstrasse 2, D-85748 Garching, Germany [email protected] ABSTRACT The nearby (d = 5.0 pc) brown dwarf LP 944 -- 20 was observed with the XMM -- Newton satellite on 07 January 2001. The target was detected with the Optical Monitor (V =16.736±0.081), but it was not detected during the ≈ 48 ks observation with the X-ray telescopes. We determine a 3σ upper limit for the X-ray emission from this object of LX < 3.1 × 1023 ergs · s−1, equivalent to a luminosity ratio upper limit of log (LX/Lbol) ≤ −6.28. This measurement improves by a factor of 3 the previous Chandra limit on the quiescent X-ray flux. This is the most sensitive limit ever obtained on the quiescent X-ray emission of a brown dwarf. Combining the XMM -- Newton data with previous ROSAT and Chandra data, we derive flare duty cycles as a function of their luminosities. We find that very strong flares [Log(LX/Lbol) > −2.5] are very rare (less than 0.7% of the time). Flares like the one detected by Chandra [Log(LX /Lbol) = −4.1] have a duty cycle of about 6%, which is lower than the radio flare duty cycle (∼13%). When compared with other M dwarfs, LP 944 -- 20 appears to be rather inactive in X-rays despite of its relative youth, fast rotation and its moderately strong activity at radio wavelengths. Subject headings: individual: LP 944 -- 20 -- stars : brown dwarfs, low-mass -- stars : X-ray emission -- 2 -- 1. Introduction LP 944 -- 20 (=BRI 0337-3535) is an isolated nearby brown dwarf (Mass<0.075 M⊙) first identified by Luyten & Kowal (1975). Spectroscopic observations reported by Tinney (1998) give a spectral type of dM9, an age of about 500 Myr and a mass of about 0.065 M⊙ using the lithium test proposed by Magazz`u, Mart´ın, & Rebolo (1993). Because of its substellar mass and its proximity to the Sun, this object is a benchmark in the study of very low-mass objects. LP 944 -- 20 was observed in the X-Ray regime for the first time with the ROSAT satellite by Neuhauser et al. (1999), but was not detected. It has been detected with the Chandra satellite (Rutledge et al. 2000) in december 1999 during an X-ray flare of duration 1 ∼ 2 hours. Outside the flare, LP 944 -- 20 was not detected with a 3 σ upper limit on the emission at LX < 1.0 × 1024 ergs · s−1. Both quiescent and flaring radio emission have been observed with the VLA (Berger et al. 2001) . X-ray emission is widespread among fully-convective M dwarfs (spectral type M3 and later), and it is frequently variable (Fleming et al. 1993; Marino, Micela & Peres 2000; Feigelson et al. 2002). X-ray observations of ultracool dwarfs (spectral type M7 and later) are still very scarce. Strong variability has been observed in a few objects (Rutledge et al. 2000; Fleming et al. 2000; Schmitt & Liefke 2002). It is thought that the X-ray photons are emitted from a hot corona. The properties of coronae (permanent or transient) in ultracool dwarfs are not well understood. LP 944 -- 20 was targeted for observations with XMM -- Newton for two reasons mainly, first to try to catch it during a flare, which would have allowed to obtain spectroscopy if a flare had occurred, and second to search for quiescent X-ray emission. Unfortunately, we did not detect LP 944 -- 20 at all. Nevertheless, we improve the value of the upper limit on the quiescent X-ray emission from LP 944 -- 20, and we derive X-ray flare rates as a function of X-ray luminosity that may be useful for planning future observations of this brown dwarf. 2. Observations XMM -- Newton observed LP 944 -- 20 on 2001 January 07-08, between 14:21:05 and 04:43:52 UTC for 51767 s. The pointing position was R.A. = 03h39min34.60sec and Dec = −35◦25′′51.0′ Epoch 2000, according to the Simbad database.1 1Simbad database, operated at CDS, Strasbourg, France -- 3 -- 2.1. Optical Data and Source Identification Besides its three X-ray telescopes, XMM -- Newton also has a 30-cm optical-UV telescope, providing the possibility to observe simultaneously in the X-ray and optical-UV regimes. The Optical Monitor (OM ) was used with the V filter for four exposures of 5000 s each in imaging mode. The OM detector has a format of 2048 × 2048 pixels, each 0.5′′ × 0.5′′. The field of view is therefore 17′ × 17′. We used the data given by the OM to confirm the presence of the source in the field of view. The coordinates we used to identify LP 944 -- 20 are different from the pointing coordinates. We used the more recent coordinates given by the astrometry of Rutledge et al. (2000) after their observation of LP 944 -- 20 with Chandra during a X-ray flare in 1999 December. These coordinates are: R.A. = 03h39min35.16sec ± 0.1′′ and Dec = −35◦25′′44.0′ ± 0.1′′, epoch 1999.95. We identified LP 944 -- 20 in the field of view by plotting these coordinates on the four images given by the OM. The result is shown in Figure 1. We found a point-source at a distance of 0.4′′ ± 0.1′′ from the expected coordinates of LP 944 -- 20. This offset is small in comparison with the < 2′′ known systematic uncertainties in the XMM -- Newton preprocess- ing analysis astrometry (XMM -- Newton User's Handbook 2000). Moreover a small offset was also expected because of the proper motion of LP 944 -- 20 since it was observed by Rut- ledge et al. (2000) more than one year ago. The offset in the optical position is consistent with the known proper motion of LP 944 -- 20. We were also able to identify the three other sources in the field of view simply by direct comparison with the image obtained by Rutledge et al. (2000). We therefore identify this optical source with the brown dwarf LP 944 -- 20 on the basis of positional coincidence with the expected coordinates. We also used the four exposures given by the OM to look for optical variability. The magnitudes of the source identified as LP 944 -- 20 and of two other stars in the field of view were calculated using the IRAF phot task. This allowed us to plot the evolution of the magnitude during the whole observation. The result is shown in Figure 2. The behavior of LP 944 -- 20's magnitude is different than that of the two other stars. We notice that whereas the magnitudes of the two reference stars are roughly correlated, LP 944 -- 20's magnitudes are not. We measure a mean magnitude of V =16.736 and a standard deviation of 0.081 for LP 944 -- 20. For reference star 1 we measure a mean magnitude of V =16.681 and a standard deviation of 0.028. For reference star 2 we measure a mean magnitude of V =16.770 and a standard deviation of 0.067. We conclude that LP 944 -- 20 may be variable in the OM data. The variability of LP 944 -- 20, however, is only slightly higher than that of reference star 2. We cannot make a definitive claim that LP 944 -- 20 is variable in our data because of the lack of enough reference stars in the field of view. -- 4 -- 2.2. X-ray Data and Analysis The European Photon Imaging Camera (EPIC ) MOS1 & MOS2 were used. The total ontarget exposure time in each camera was 48724 s in prime full window mode with 2.5 s time resolution. The EPIC PN was used in prime full window mode with 73.4 ms time resolution and ontarget exposure 46618 s. Each EPIC instrument was used with the thin filter, which suppresses the optical contamination up to mV of 18 (MOS) or mV of 17 (PN). Both MOS & PN detectors have a circular field of view of 30′ diameter. For more details on XMM -- Newton and its instruments, we refer to XMM -- Newton User's Handbook (2000). The first step was to search for LP 944 -- 20 in the field of view of each EPIC MOS & PN detector. We plotted the coordinates of LP 944 -- 20 on each image obtained by each detector, but we did not detect any source. To be sure that we were using the most sensitive exposure, we used the image given by the pipeline which combines data from all EPIC's MOS & PN in the entire energy range (0.1 − 12.0 keV ), as shown in Figure 3. On these figures, the X-ray source closest to the expected position is offset by δ(R.A.) = 2.5′′±0.1′′ and δ(Dec) = −28.37′′±0.1′′. This offset is large in comparison with the 0.4′′±0.1′′ offset observed with the optical data. So at this point we already knew that LP 944 -- 20 was not detected. We performed differential astrometry between the XMM -- Newton and Chandra images. This allowed us to identify all the X-ray sources in our field of view in an area of 20′′ × 10′′ centered on the expected coordinates of LP 944 -- 20. We were then able to conclude that LP 944 -- 20 was not detected during the XMM -- Newton observation. Since the target was not detected, the first conclusion is that there was no significant X-ray flare during this 48 ks observation. Nevertheless, it is interesting to calculate the upper limit on the time average quiescent X-ray luminosity. We did the calculation of the upper limit of X-ray flux with each of the 3 images of the preprocessing products, corresponding to the 3 exposures obtained with the EPIC MOS1, MOS2 & PN detectors on the entire energy range (0.1 − 12.0 keV). To calculate a 3σ flux upper limit, we considered an area of twice the FWHM (= 2 pixels) around the expected position of LP 944 -- 20. In this area we calculated the mean value of counts · pixel−1 and the corresponding standard deviation σ using the IRAF2 phot task, which gives count rates corrected for background. The results are listed in Table 1. As the EPIC PN camera is the most sensitive detector, we therefore only consider its result to calculate the flux upper limit. 2IRAF is distributed by National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy, Inc., under contract with the National Science Foundation. -- 5 -- To convert this count rate to flux, we followed XMM -- Newton User's Handbook (2000) instructions, by using the PIMMS3 software. As LP 944 -- 20 was not detected during the observation, we were not able to fit its spectrum in order to know its spectral type. We adopted the results of Rutledge et al. (2000) summarized in Table 2. The HI column density was calculated using the nH 3 software: nH = 1.31 × 1020cm−2. The results for the upper limit of flux and X-ray luminosity (3σ, assuming Gaussian statistics) are listed in Table 3. The uncertainties on these values are Poisson (counting statistics) plus ∼ 10% spectral uncertainty, according to Rutledge et al. (2000). As their best fit on the spectrum was obtained with the Raymond -- Smith thermal plasma model, we finally kept the corresponding 3 σ flux as the upper limit on the time average quiescent X-ray luminosity for LP 944 -- 20: LX < 3.1 × 1023ergs · s−1, equivalent to a luminosity ratio upper limit of log (LX/Lbol) ≤ −6.28. 3. Discussion The first observation of LP 944 -- 20 in the X-ray regime by Neuhauser et al. (1999) with the ROSAT satellite placed an upper limit of log (LX/Lbol) ≤ −4.17 after a 220 ks ROSAT -- HRI observation. The second X-ray observation (Rutledge et al. 2000) placed a better upper limit of log (LX /Lbol) ≤ −5.70 after the 44 ks Chandra -- ACIS-S observation before the flare. Our new value improves this upper limit of quiescent X-ray emission at log (LX /Lbol) ≤ −6.3. The corresponding upper limit of X-ray luminosity (LX ≤ 3.1 × 1023ergs · s−1) is thus improved by a factor 3. This value is close to the solar X-ray emission level during its maxima of activity. In Figure 4 we display the dependence of X-ray activity versus spectral type for M dwarfs. The upper envelope of X-ray activity is remarkably independent of M subclass. There is no obvious connection between age and X-rays. The M dwarfs in the ChaI, ρOph and Taurus star-forming regions show X-ray emission comparable to much older field dwarfs (Neuhauser et al. 1999; Mokler & Stelzer 2002). We conclude that there is no evidence for a decline in coronal emission with increasing age in fully convective M dwarfs. We have estimated the rate of X-ray flare occurrence in LP 944 -- 20 by adding 287.5 ks of ROSAT data analyzed by Neuhauser et al. (1999), 43.8 ks of Chandra data analyzed by Rutledge et al. (2000) and 51.7 ks of XMM -- Newton data (this paper). The duration of the 3nH and PIMMS (Portable, Interactive, Multi-Mission Simulator) are distributed by the NASA's HEARSAC -- 6 -- only flare so far observed was 2.76 ks Rutledge et al. (2000). Thus, we get an X-ray flare frequency of 0.72% for flares with luminosity Log(LX/Lbol) > −2.5. Weaker flares could not have been detected by ROSAT. Thus, we derive the duty cycle of flares with luminosity comparable with the flare detected by Chandra [Log(LX/Lbol) = −4.1] to be about 6%. The published VLA radio data have a duration of 79.9 ks (Berger et al. 2001). Three flares were detected with a total duration of ∼10.4 ks. Hence, we estimate a radio flare frequency of about 13%. It is not known to what X-ray luminosity the radio flares correspond, but we can set an upper limit using the XMM -- Newton data. If we assume a flare duty cycle of 13%, the flare X-ray luminosity has to be Log(LX/Lbol) < −5.39 to be consistent with our non detection. LP 944 -- 20 is a relatively inactive object. Tinney (1998) reported Hα weakly in emis- sion. Berger et al. (2001) estimated a surface magnetic field strength of B∼5G assuming synchrotron emission during the radio flares. For comparison, active dM stars have magnetic field strengths of a few kG, solar flares have B∼100G, and Jupiter has an average B∼10G (Hide & Stannard 1976). Our low upper limit on the quiescent X-ray emission of LP 944 -- 20 is consistent with a weak magnetic field. The detection of low-amplitude photometric vari- ability in LP 944 -- 20, and the low-level of X-ray emission in this object, suggests the presence of surface thermal inhomogeneities due to nonmagnetic clouds, as discussed by Tinney & Tolley (1999) for LP 944 -- 20 and by Mart´ın, Zapatero Osorio, & Lehto (2001) for the M9.5 dwarf BRI0021-0214. Recent theoretical works (Gelino et al. 2002; Mohanty et al. 2002) show that in ultracool dwarfs there should not be much interaction between the gas and the magnetic field because the Reynolds number is very low throughout the photosphere. LP 944 -- 20 appears to be an object with substellar mass, weak magnetic field, fast rotation and cloudy atmosphere. We speculate that the radio activity may be enhanced by the presence of a close planetary-mass companion (Io-type), which provides a continuous supply of ionized particles to the ionosphere of LP 944 -- 20. Radial velocity variability has been detected by Guenther & Wuchterl (2003) which could be due to a planetary-mass close companion, but they did not have enough data to derive an orbital period. -- 7 -- This research has made use of the Simbad 3 database, operated at CDS, Strasbourg, France. Support for this project has been provided by NASA grant NAG5-9992. We thank our collaborators Gibor Basri, Lars Bildsten and Bob Rutledge, and an anonymous referee for helpful discussions. REFERENCES Berger, E., Ball, S., Becker, K. M., Clarkes, M., et al. 2001, Nature 410, 338 Feigelson, E. D., Broos, P., Gaffney, J. A., Garmier, G., Hillenbrand, L. A., Pravdo, S. H., Townsley, L., & Tsuboi, Y. 2002, ApJ574, 258 Fleming, T. A., Giampapa, M. S., Schmitt, J. H. M. M., & Bookbinder, J. A. 1993, ApJ410, 387 Fleming, T. A., Schmitt, J. H. M. M., & Giampapa, M. S. 1995, ApJ450, 401 Fleming, T. A., Giampapa, M. S., Schmitt, J. H. M. M. 2000, ApJ533, 372 Gelino, C. R., Marley, M. S., Holtzman, J. A., Ackerman, A. S., & Lodders, K. 2002, astroph-005305 Giampapa, M. S., Rosner, R., Kashyaf, V., Fleming, T. A., Schmitt, J. H. M. M., & Book- binder, J. A. 1996, ApJ463, 707 Guenther, E. & G. Wuchterl 2003, in IAU Symposium 211: Brown Dwarfs, E. L. Mart´ın (ed.), in press Hide, R. & Stannard, D. 1976, in Jupiter, T. Gehrels (ed.), The University of Arizona Press, Tucson Luyten, W. J., & Kowal, C. T. 1975, Proper Motion Survey with the 48 inch Schmidt Tele- scope. XLIII. One Hundred and Six Faint Stars with Large Proper Motion (University of Minnesota, Minneapolis). Magazz`u, A., Mart´ın, E. L., & Rebolo, R. 1993 ApJ, 404, L17 Marino, A., Micela, G., & Peres, G. 2000, A&A353, 177 Mart´ın, E. L., Zapatero Osorio, M. R. & Lehto, H. 2001, ApJ, 557, 822 3http://simbad.u-strasbg.fr/ -- 8 -- Mohanty, S., Basri, G., Shu, F., Allard, F., & Chabrier, G. 2002, ApJ, 571, 469 Mokler, F., & Stelzer, B. 2002, A&A, in press Neuhauser, R., et al. 1999, A&A, 343, 883 Rutledge, R. E., Basri, G., Mart´ın, E. L, Bildsten, L., 2000, ApJ, 538L, 141 Schmitt, J. H. M. M. & Liefke, C. 2002, A&A, 382, L9 Tinney, C. G. 1996, MNRAS. 281, 644 Tinney, C. G. 1998, MNRAS. 296, L42 Tinney, C. G. & Tolley 1999, MNRAS. 304, 119 XMM -- Newton SOC Team, XMM -- Newton User's Handbook Issue 2, 03.31.2000 This preprint was prepared with the AAS LATEX macros v5.0. -- 9 -- Table 1. Count rates Instrument Counts/pixel (average) Standart deviation σ (counts/pixel) 3σ (counts/pixel/s) EPIC PN EPIC MOS1 EPIC MOS2 4.53 1.59 1.73 2.00 1.18 1.24 1.3 × 10−4 7.3 × 10−5 7.6 × 10−5 Table 2. Specral parameters of LP 944 -- 20 Model Associated Feature Black Body Power Law Thermal Plasma kB × T = 0.26 keV kB × T = 0.17 keV αphoton = 2.6 References. -- Rutledge et al. (2000) Table 3. X-ray luminosity upper limit Model Flux (10−17erg.cm−2.s−1 ) log LX (ergs.s−1 ) Black Body Power Law Thermal Plasma ≤ 1.1 ≤ 1.3 ≤ 1.0 ≤ 23.50 ≤ 23.59 ≤ 23.49 log ( LX Lbol ) ≤ −6.27 ≤ −6.19 ≤ −6.28 References. -- Tinney (1996) for Lbol ∼ 6 × 1029ergs.s−1 -- 10 -- Fig. 1. -- Indentification of LP 944 -- 20 in the field of view of the Optical Monitor Data. The plus indicate the expected coodinates of LP 944 -- 20. The diffuse bright area on the top of the image is an artifact. O p t i c a l M o n i t o r . F i g . 2 . -- O p t i c a l m a g n i t u d e o f L P 9 4 4 -- 2 0 a n d o f 2 r e f e r e n c e s t a r s i n t h e fi e l d o f v i e w o f t h e e d u t i n g a M 16.85 16.8 16.75 16.7 16.65 16.6 Magnitude in V vs. Time XMM Optical Monitor LP944-20 Ref. Star 1 Ref. Star 2 -- 1 1 -- 5000 10000 time (s) 15000 20000 -- 12 -- EPIC MOS1 + MOS2 + PN (0.1−12 keV) DEC −35:10 −35:20 −35:30 −35:40 3h41m 3h40m 3h39m RA Fig. 3. -- Combined image from all EPICs MOS & PN instruments in the energy range (0.1 − 12.0 keV ). The figure shows the entire field of view. The square chips of the EPIC PN and the circular field of view of MOS1 & 2 appear on this figure. A plus indicate the expected position of LP 944 -- 20. -- 13 -- -1.5 -2 -2.5 -3 -3.5 -4 -4.5 -5 -5.5 -6 -6.5 Field Flares Taurus ρ−Oph Cha I Sun at maximum Sun at maximum Sun at maximum Sun at maximum Sun at maximum LH0426+17 LHS 2065 VB 8 VB 10 CFHT-BD-Tau 4 Gl 234B Cha Hα 1 LP944-20 GY141 LHS 3003 ROSAT Chandra XMM-Newton )  X  L L ( l o b g o L M0 M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 Spectral Type Fig. 4. -- Distribution of X-ray luminosity of dM dwarfs as a function of spectral type. The squares represent values obtained for field dwarfs (Fleming et al. (1993); Fleming, Schmitt, & Giampapa (1995); Fleming et al. (1996); Rutledge et al. (2000)). The circles, diamonds and triangles denote young objects in star-forming regions. The stars represent the flaring objects at their maximum with the corresponding value or upper limit on the quiescent emission linked by a dotted line. To avoid confusion, the Taurus object's spectral types have been shifted by 0.1 subclass. (2000); Giampapa et al.
astro-ph/9906016
2
9906
1999-09-03T20:02:04
Metallicity Effects in NLTE Model Atmospheres of Type Ia Supernovae
[ "astro-ph" ]
We have calculated a grid of photospheric phase atmospheres of Type Ia supernovae (SNe Ia) with metallicities from ten times to one thirtieth the solar metallicity in the C+O layer of the deflagration model, W7. We have modeled the spectra using the multi-purpose NLTE model-atmosphere and spectrum-synthesis code, PHOENIX. We show models for the epochs 7, 10, 15, 20, and 35 days after explosion. When compared to observed spectra obtained at the approximately corresponding epochs these synthetic spectra fit reasonably well. The spectra show variation in the overall level of the UV continuum with lower fluxes for models with higher metallicity in the unburned C+O layer. This is consistent with the classical surface cooling and line blocking effect due to metals in the outer layers of C+O. The UV features also move consistently to the blue with higher metallicity, demonstrating that they are forming at shallower and faster layers in the atmosphere. The potentially most useful effect is the blueward movement of the Si II feature at 6150 Angstrom with increasing C+O layer metallicity. We also demonstrate the more complex effects of metallicity variations by modifying the 54Fe content of the incomplete burning zone in W7 at maximum light. We briefly address some shortcomings of the W7 Finally, we identify that the split in the Ca H+K feature produced in W7 and observed in some SNe Ia is due to a blending effect of Ca II and Si II and does not necessarily represent a complex abundance or ionization effect in Ca II. amodel when compared to observations.
astro-ph
astro-ph
Metallicity Effects in NLTE Model Atmospheres of Type Ia Supernovae Eric J. Lentz, E. Baron, David Branch Department of Physics and Astronomy, University of Oklahoma, 440 West Brooks, Norman, OK 73019-0225 baron,branch,[email protected] Peter H. Hauschildt Department of Physics and Astronomy & Center for Simulational Physics, University of Georgia, Athens, GA 30602 [email protected] and Peter E. Nugent Lawrence Berkeley National Laboratory, Berkeley, CA 94720 [email protected] ABSTRACT We have calculated a grid of photospheric phase atmospheres of Type Ia super- novae (SNe Ia) with metallicities from ten times to one thirtieth the solar metallicity in the C+O layer of the deflagration model, W7. We have modeled the spectra using the multi-purpose NLTE model-atmosphere and spectrum-synthesis code, PHOENIX. We show models for the epochs 7, 10, 15, 20, and 35 days after explosion. When compared to observed spectra obtained at the approximately corresponding epochs these synthetic spectra fit reasonably well. The spectra show variation in the overall level of the UV continuum with lower fluxes for models with higher metallicity in the unburned C+O layer. This is consistent with the classical surface cooling and line blocking effect due to metals in the outer layers of C+O. The UV features also move consistently to the blue with higher metallicity, demonstrating that they are forming at shallower and faster layers in the atmosphere. The potentially most useful effect is the blueward movement of the Si II feature at 6150 A with increasing C+O layer metallicity. We also demon- strate the more complex effects of metallicity variations by modifying the 54Fe content of the incomplete burning zone in W7 at maximum light. We briefly address some shortcomings of the W7 model when compared to observations. Finally, we identify that the split in the Ca H+K feature produced in W7 and observed in some SNe Ia is due to a blending effect of Ca II and Si II and does not necessarily represent a complex abundance or ionization effect in Ca II. -- 2 -- Subject headings: line: formation -- nuclear reactions, nucleosynthesis, abundances -- radiative transfer -- stars: atmospheres -- supernovae: general 1. Introduction Type Ia supernovae (SNe Ia) are recognized for their near uniformity as standard candles. This has led to the use of SNe Ia in cosmology (Branch 1999; Garnavich et al. 1998; Riess et al. 1998; Perlmutter et al. 1999). The development of empirical calibrations between peak brightness and light curve shape (Phillips 1993; Hamuy et al. 1996; Riess et al. 1996; Perlmutter et al. 1997a,b; Tripp & Branch 1999), have enhanced the usefulness of SNe Ia as distance indicators. Nevertheless theoretical modelers have yet to agree on the source of these apparently systematic variations. A primary concern is the evolutionary lifetime of SNe Ia progenitors and the possibility of significant deviations of distant SNe Ia from their well observed counterparts in the local galactic neighborhood. If galactic chemical evolution occurs slowly, then more distant SNe arise from a younger, metal poor population. On the other hand, the metallicity variations in the local sample may already span the range of the entire observational sample. We probe the possible effects of progenitor metallicity variations on the observed spectra, by modifying the parameterized deflagration model, W7 (Nomoto et al. 1984; Thielemann et al. 1986). Using base fits to observations (§ 2.3) we have scaled all elements heavier than oxygen in the unburned C+O layer of W7 to simulate the effects of various metallicities in the progenitor system. Hoflich et al. (1998) explored this question by modifying the pre-explosion metal content of a particular SNe Ia model and noted the differences in final composition, light curves, and spectra. They found that the composition of the partially burned layers of the ejecta yielded larger quantities of 54Fe. A similar effect can be seen in the lowered 54Fe abundance when W7 is calculated using a pure C+O mixture without other metals (Iwamoto et al. 1999). We have modified the 54Fe abundance of the partially burned layers of W7, to replicate this effect. While Hoflich et al. (1998) focused mainly on the effects of metallicity variations on the light curve and energetics, here we concentrate exclusively on its effects on the observed spectra, particularly at early times where the formation of the spectrum occurs in the unburned C+O layer, which is most sensitive to initial progenitor metallicity. Our computational methods are given in § 2. In § 3 we show the effects of metallicity modifi- cation of the C+O layer on the synthetic spectra at each epoch. In § 4 we show the effects of 54Fe on day 20 and 35 model spectra. In § 5.1 we discuss the evolution of the Si II feature at 6150 A. -- 3 -- 2. Methods 2.1. Abundance Modification To model the effects of metallicity on SNe Ia, we have modified the base W7 model at several epochs. W7, while clearly not the complete model for SNe Ia, is a good starting point for spectral modeling. The composition structure of W7 as a function of velocity does reasonably well in repro- ducing the observed spectra of SNe Ia (Harkness 1991a,b; Branch et al. 1985; Nugent et al. 1997). By making separate calculations to consider the direct (C+O metallicity) and the indirect explosive nucleosynthetic consequences (54Fe enhancement), our computational methods allow us to probe these effects of progenitor metallicity independently. Hoflich et al. (1998) changed the composition before computing the hydrodynamics and explosive nucleosynthesis of the model. This gives them consistent nucleosynthesis and energetics within the context of their chosen input physics. How- ever, it is difficult to separate effects in spectra that arise from different consequences of the initial metallicity variation. To modify the metallicity of the unburned C+O layer we have scaled the number abundances of elements heavier than oxygen in the velocity range 14800 -- 30000 km s−1 by a constant factor ζ, such that for all species i heavier than oxygen, the new number abundance n′ i, is given by The mass fractions are then renormalized, n′ i = ζni. in each layer. For all epochs we have used C+O metallicity factors, ζ, of 10, 3, 1, 1/3, 1/10, and 1/30. These effects are discussed in detail in § 3. X Xi = 1, The nucleosynthesis results of Hoflich et al. (1998) show that modifying the metallicity of the progenitor white dwarf changes the quantity of 54Fe produced in the incomplete burning zone in a manner approximately proportional to the metallicity. This is due to excess neutrons in the progenitor over a pure C+O mixture, particularly 22Ne (see for example Arnett 1996; Nomoto et al. 1984; Iwamoto et al. 1999, and references therein). To simulate this effect in W7, we have scaled the 54Fe abundance in the incomplete burning zone of W7, 8800 -- 14800 km s−1, in the same manner as the C+O layer metallicity, n′ 54Fe = ξn54Fe. The mass fractions are then renormalized in each layer. We have used 54Fe abundance factors, ξ, of 3, 1, 1/3, 1/10, and 1/30 for the epochs 20 and 35 days after explosion. For the remaining epochs we have only computed models for 1/10 54Fe abundance (ξ = 0.1). These effects are discussed in detail in § 4. Finally, we simulate the full effects of progenitor metallicity on the output spectra, by com- bining the two effects, C+O layer metallicity and incomplete burning zone 54Fe abundance, into -- 4 -- a single series of models. For each model we have used the same factor for both C+O metallicity and 54Fe abundance (ξ = ζ). These factors are the same as for the 54Fe abundance modifications above. These effects are discussed in detail in § 4. 2.2. Computational Methods of PHOENIX The calculations were performed using the multi-purpose stellar atmospheres program PHOENIX 9.1 (Hauschildt & Baron 1999; Baron & Hauschildt 1998; Hauschildt et al. 1997a,b, 1996). PHOENIX solves the radiative transfer equation along characteristic rays in spherical symmetry including all special relativistic effects. The non-LTE (NLTE) rate equations for several ionization states are solved including the effects of ionization from non-thermal electrons from the γ-decay energy of the 56Ni core. The atoms and ions calculated in NLTE are: He I (11 levels), He II (10), C I (228), O I (36), Na I (3), Mg II (18), Si II (94), S II (85), Ti II (204), Fe II (617), and Co II (255). Each model atom includes primary NLTE transitions, which are used to calculate the level populations and opacity, and weaker secondary NLTE transitions which are are included in the opacity and implicitly affect the rate equations via their effect on the solution to the transport equation. In addition to the NLTE transitions, a number of LTE line opacities for atomic species not treated in NLTE are treated with the equivalent two-level atom source function, using a thermalization parameter, α = 0.1, as in Nugent et al. (1997) for LTE lines for all models in this paper. The atmospheres are iterated to energy balance in the co-moving frame, γ r2 ∂(r2H) ∂r γβ ∂J ∂r + + γ(cid:20) β r γ 2 ∂β ∂r (3J − K) + (J + K + 2βH)(cid:21) = S 4π , (1) S is the deposited energy from radioactive decay, γ and β have their usual meaning and where, c = 1. This equation neglects the explicit effects of time dependence in the radiation transport equation, however the term on the right hand side implicitly includes these effects. The models are parameterized only by the day, which determine the radii and amount of radioactive decay, and by the luminosity parameter, η, which is defined as, Lbol = ηLabs γ , γ where Labs is the instantaneous deposition of radioactive decay γ-ray energy. The deposition of γ-rays is determined by solving a grey transport equation (Nugent 1997; Nugent et al. 1997; Sutherland & Wheeler 1984) as a function of time. This approach has been shown to be of sufficient -- 5 -- accuracy for our purposes when compared to detailed Monte-Carlo calculations (Swartz et al. 1995). We use two different inner boundary conditions for our models. For days 7, 10, and 15, which are optically thick at the core, we use a diffusive inner boundary condition. For days 20 and 35 which have lower total optical depth we use a 'nebular' inner boundary. This boundary condition takes the downward intensity of any ray and assumes that it passes unimpeded through the core of the atmosphere, becoming the upward intensity on the other side, with only the necessary Lorentz transformation. Additional details on the use of PHOENIX in modeling SNe Ia atmospheres can be found in Nugent (1997). 2.3. Baseline Models To perform numerical experiments on the metallicity effects in the spectra of SNe Ia, we need reasonable base models for each epoch. For each day, an observed spectrum of SN 1994D was fitted with a W7 model of appropriate age. Several models were computed with various luminosity parameters, η, to fine tune the spectral shape. A separate paper fitting W7 to SN 1994D photospheric era spectra (Lentz et al. in preparation) shows the comparisons to observations. Hoflich & Khokhlov (1996) find η (their Q) in the range 0.7 < η < 1.8. When η > 1 stored radiative energy is being released and when η < 1 radiative energy is being stored. The maximum light (epoch day 0 for the observations) fit for SN 1994D to W7 (with η = 1.0) at day 20 after the explosion is the same as in Nugent et al. (1997). For the two model fits at days 10 (which we compare to day -9 with respect to B maximum) and 15 (compared to day -4 with respect to B maximum) for W7 we have used luminosity parameters of η = 0.4 and η = 0.8 respectively. For these two models the fits are generally good, but with all of the pre-maximum spectra, the red edge of the Si II feature at 6150 A does not extend far enough to the red. For model day 7 (fit to day -12) we chose the the model with luminosity η = 0.1. This model fits the luminosity and most features well; however, like the day 10 models the Ca H+K feature not extend blueward enough when compared to the earliest observations. Hatano et al. (1999) have modeled the same spectra and found that the Ca H+K feature requires ionized calcium with velocities up to ∼ 40000 km s−1. We have conducted experiments extending the density profile of W7. This did not provide enough optical depth in the Ca H+K feature to affect the line profile. Neither do our models reproduce the secondary feature at 4700 A that Hatano et al. (1999) ascribe to Fe II absorption in the C+O layer. For our post-maximum model, day 35 (15 days after maximum), we fit the observation with a model with luminosity, η = 1.5. This model fits generally; however, the large feature at 4800 A (probably Fe II) is not seen in the observations. The observed Na I D-line feature is missing, probably due to deficiencies in the sodium model atom, which has already been improved in the latest version of PHOENIX. Here, we are interested in small differential effects so it was essential that all models be run with the code version frozen. -- 6 -- 3. Metallicity of Unburned C+O Layer The simplest effect of metallicity on the spectral formation in SNe Ia is the change in the metal content of the unburned C+O layer neglecting changes in the density structure and deeper layers. In this section we examine these effects independently of other effects of progenitor metallicity on SNe Ia. Since ongoing supernova searches are expected to discover SNe early, and early spectra probe the outermost layers only, this approach is sensible and yields physical insight that is somewhat model independent. When we look at the overall UVOIR synthetic spectra (Figures 1, 6, 8, 11, and 13) of the models with variations in the C+O layer metallicity, we see two consistent and significant effects: shifts in the UV pseudo-continuum level (expanded view for day 7 in Figure 2) and variations in the Si II line at 6150 A (expanded views in Figure 5). The general effect in the UV is the increase in the UV pseudo-continuum level with decreasing metallicity. Simultaneous is the redward (blueward) shift of most UV features with decreasing (increasing) metallicity. In the UV, the line-forming region is in the C+O layer. As the metallicity decreases, the line forming region must reach deeper into the atmosphere to have the same line opacity, resulting in smaller line velocities. Modification of the C+O layer metal abundance gives a classic surface cooling effect, lower temperatures for higher metallicity. The higher temperatures of the lower metallicity C+O atmospheres give higher thermal fluxes, moving the UV pseudo- continuum higher with lower metallicity. The surface cooling and the resulting shifts in UV pseudo- continuum are evident at every epoch. There is the complementary effect of additional metals increasing the line blocking. We make no attempt to separate these two effects in this paper. The Si II line at 6150 A (Figure 5) shifts blueward with increases in metallicity for epochs through day 20. These shifts demonstrate that some line formation in this feature takes place in the C+O layer. The earlier epochs show large variations in the total depth of the feature which implies that the line forms less in the incomplete burning zone with its large, unchanging silicon abundance, and more in the C+O layer where the silicon abundance changes. At later epochs these conditions are reversed, resulting in smaller changes with C+O layer metallicity variation. These effects are discussed further in § 5.1. The Mg II "h+k" feature at 2600 A (Figure 2) does not move to the blue or red as the metallicity varies. The decrease in Mg II h+k feature strength with increasing metallicity is caused by the increasing UV line blanketing from background line opacity in the C+O layer. The Mg II absorption occurs mostly in the deeper, partially burned layer that is highly enriched in magnesium. We have confirmed this hypothesis by calculating diagnostic output spectra without using any background opacity (see Baron et al. 1999, for a discussion of the method). These diagnostic spectra suggest that this feature (and the feature near 2800 A) may actually be due to a complicated blend of Mg II UV 1-4 transitions: h+k (UV 1) λ2798, (UV 2) λ2796, (UV 3) λ2933, and (UV 4) λ2660. -- 7 -- 3.1. Day 7 Our grid of synthetic spectra of W7 on day 7 with C+O metallicity variations are shown in Figure 1. The case ζ = 10 is likely extreme and is shown for illustrative purposes only, we don't consider it further at this epoch. The UV spectra (Figure 2), show variations in the pseudo- continuum due to surface cooling, line blocking, and feature shifts due to changes in the feature formation depth (velocity). The pseudo-continuum level in the optical and IR vary with metallicity. Lower metallicities have lower pseudo-continuum fluxes. Figures 3 and 4 display optical and infrared spectra which illustrate this effect more clearly. This is due to backwarming. Figure 3 shows that the line features at 3650 A (Ca II H+K), 4250 A (Fe II, Mg II, & Si III), 4700 A (Fe II), and 5650 A (Si II) nearly disappear as the metallicity drops to 1/30 of normal. The 4700 A Fe II feature is weak and doesn't extend far enough to the blue in the base model fits (§ 2.3). Increasing the metallicity by even a factor of 10 does not extend the feature blueward enough (to sufficiently large velocities) to match the observations. The same is true of the Ca II H+K lines. This indicates that more mass extending to higher velocities is needed to fit this feature. The Si II feature at 6150 A (Figure 5a) shows the blueward movement of the feature minimum and blue-edge wall with increasing metallicity. The slope of the red edge of the feature changes and at the lowest metallicities the feature is not strong enough to form the red emission wing of the P Cygni feature. The infrared spectra (Figure 4) show several features that weaken with lower metallicity, Mg II λ9226 A and Si I complexes at λ10482 and λ10869 A (cf. Millard et al. 1999), as well as two features that strengthen with lower metallicity. These two O I features (7400 A and 8150 A) weaken as the oxygen abundances decreases in favor of higher Z metals. 3.2. Day 10 Our grid of synthetic spectra of W7 on day 10 with metallicity variations in the C+O layer (Figure 6) display similar variations to those in the day 7 spectra (Figure 1). The UV variations again show the blue-shifted features and lower continua with increasing metallicity. The optical pseudo-continuum effects (Figure 7) are still present, but smaller. We again see lines that deepen with increased metallicity such as 3650 A (Ca II H+K), 4250 A (Fe II, Mg II, & Si III), and 6100 A (Si II), but these effects are less dramatic than at day 7. The Si II feature (Figure 5b) still shows the changing depth and slopes of the line edges with metallicity, but, at day 10 the depth of the Si II feature in the lowest metallicity model is larger than in the same model on day 7, relative to the highest C+O layer metallicity in the respective epoch. These effects on the optical lines indicate that line formation for certain strong lines is now taking place in layers below the C+O layer. The effects of line formation below the C+O layer becomes more important as the supernova atmosphere expands and becomes less opaque. -- 8 -- 3.3. Day 15 Our grid of synthetic spectra of W7 on day 15 with C+O metallicity variations (Figure 8) are again similar to those in the day 7 and 10 spectra (Figures 1 & 6). The decreasing effects of C+O layer metallicity are apparent. The UV pseudo-continuum variation with metallicity remains strong. The optical (Figure 9) and the near infrared (Figure 10) show the backwarming optical/IR pseudo-continuum flux effect which occurred in the optical at day 10, and extended well into the IR at day 7. Features which increased in strength with increasing C+O metallicity are a blend of Fe II, Mg II, & Si III at 4250 A, the multi-species blend at 3300 A, Mg II at 8850 A, and an unidentified feature at 10400 A (possibly Si I). The feature at 3300 A is a blend of weak lines that forms in the C+O layer. Figure 9 illustrates the blue shifting of features as changes in metal content move the depth (and thus velocity) of line formation. The two component feature at 3700 -- 3900 A is usually labeled Ca II H+K. Nugent et al. (1997) have identified this 'Ca split' as arising from a blend of Ca II H+K with Si II λ3858. We have calculated a series of diagnostic spectra at this epoch using the same temperature structure, but without background opacity. We have confirmed the identification of the blue wing with the Si II line. When spectra are calculated without any other line opacities, Ca II forms a pair of features. The blue Ca II absorption forms from the λ3727 line, the red absorption from the H+K lines. The Si II feature falls on the peak between the two Ca II absorptions. As the Si II feature strengthens, the flux displaced by its absorption 'fills-in' the Ca II H+K absorption creating a 'split'. With enough Si II only the blue absorption feature remains, while the red feature becomes an inflection. The lack of a split in some SNe Ia indicates that either the Si II feature is weaker or the Ca II H+K is stronger, preventing the formation of a split. This helps to illustrate that in supernovae of all types, features are often the result of more than one multiplet, or even ionic species. The Si II feature at 6150 A (Figure 5c) shows much smaller, but still significant, effects due to C+O metallicity. The blue edges are parallel and the red edges nearly so. The depth contrast is now much smaller than before. This is strong evidence that the feature is forming primarily below the C+O layer at this epoch. Figure 10 shows the displacement of oxygen by metals in the O I lines at 7450 A, 8100 A, and 8250 A, the latter two of which are superimposed on the stronger Ca II 'IR-triplet' which, like the H+K absorption, forms in the calcium rich incomplete burning zone and is unaffected by the C+O metallicity. 3.4. Day 20 Our grid of synthetic spectra of W7 on day 20, approximately maximum B magnitude, with C+O metallicity variations (Figure 11) shows the continuing reduction in importance of the C+O layer metallicity as the atmosphere becomes more optically thin. The UV pseudo-continuum and feature shift effects are still strong. The only remaining features which display C+O layer metallicity effects in the optical (Figure 12) are the Ca II H+K, the 4250 A (Fe II, Mg II, & Si III), and -- 9 -- 6150 A Si II features. As at day 15, the 'Ca split' remains, the blue component increasing in strength and the red component decreasing with increasing C+O metallicity. The Si II feature at 6150 A (Figure 5d) now shows parallel edges (both blue and red) for the various C+O layer metallicities. The relatively small changes in the line strength indicate that the feature forms mostly in the deeper, silicon rich layers, but some measurable effects due to C+O layer metallicity are still evident. 3.5. Day 35 Our grid of synthetic spectra of W7 on day 35 with C+O metallicity variations are plotted in Figure 13. The UV features still show the same pseudo-continuum and line shifting behavior seen in the previous epochs. This illustrates that even at this epoch the pseudo-continuum formation in the UV is still in the unburned C+O layer. The Si II feature at 6150 A has only small variations which can not be separated from the pseudo-continuum effects. It should be noted that the Si II feature at this epoch fits the observations rather poorly. In the optical we see the blue-shifting of the 4800 A (Fe II) feature with increasing metallicity without significant changes to the overall width, shape, or depth of the feature. This indicates that it forms mostly at deeper layers, with smaller effects from the C+O layer. When looking for a base fit for this epoch, we find that while the line shapes near 5000 A are better with somewhat hotter models, those models had poor overall spectral shape or color. To diagnose this sudden change in model behavior, we have plotted the temperature profile of the three models with the lowest metallicity, the density profile, and the carbon abundance (Figure 14). We can see that the ζ = 1/30 model has a definite temperature inversion. This inversion corresponds to the position of the density spike in W7 arising from the deflagration wave. The higher density coincides with the change from the more efficient cooling available by intermediate mass elements to that of the less efficient cooling of C+O creating the temperature inversion. In models with higher metal content the effectiveness of cooling by metals in the C+O layer provides the needed cooling to prevent the formation of an inversion. 4. 54Fe Content of Incomplete Burning Zone 4.1. Pre-Maximum Light Epochs We have computed models with 54Fe abundance reductions of 1/10 in the intermediate burning zone and models with the combined 1/10 C+O metallicity and 1/10 54Fe abundance reductions for the epochs 7, 10, and 15 days. When we compared the 54Fe reduced abundance models with the related models containing the same C+O metallicities (normal and 1/10 normal, respectively) we found no changes in feature strength. Some slight differences in pseudo-continuum levels were -- 10 -- seen in the day 10 and 15 comparisons. These may be due to weaker versions of the 54Fe abun- dance caused backwarming described in § 4.2. Since each comparison pair includes two models with different composition in the incomplete burning zone this may slightly affect the temperature structure. 4.2. Day 20 Our grid of synthetic spectra of W7 on day 20, approximately maximum B magnitude, with incomplete burning zone 54Fe abundance variations is shown in Figure 15. Some small vertical displacements of the UV flux, without blue- or red-shift, can be understood by the effects of surface cooling/backwarming in the incomplete burning zone. As intermediate mass elements are replaced by 54Fe, the larger line opacity of iron cools the 54Fe rich models. Since this does not affect the temperature gradient in the C+O layer, the temperature shift remains constant throughout the C+O layer. In the optical (Figure 16) we can see a few features that decrease with increasing 54Fe, such as the Si II and S II in the 4850 A feature, the S II "W" at 5300 A, and the Si II feature at 6150 A. This is caused by displacement of the species forming the line by additional 54Fe. Several features can be seen to strengthen with greater 54Fe abundance. This indicates that they are formed at least in part by iron in the incomplete burning zone. These features included the feature at 4100 A that erodes the peak of a neighboring feature and the red wing of the 4300 A feature. Our grid of synthetic spectra of W7 on day 20 with incomplete burning zone 54Fe abundance variations and C+O layer metallicity are shown in Figure 17. The effects of the combined modi- fications mostly separate into the effects seen in Figures 11 & 15. The UV displacements in flux and the effects on the Si II 6150 A feature in the combined models reflect the effects of metallicity variation in the C+O layer alone. The S II "W" at 5300 A and the "peak erosion" line at 4100 A show the primary effect of 54Fe abundance on the optical spectra (Figure 18) when both the 54Fe abundance and the C+O layer metallicity are varied simultaneously. The effect of the combined modifications on the Ca II H+K feature are small. The 4350 A feature in the combined modifica- tion has the combination effects of deeper red wing strength with increasing 54Fe abundance and a deeper blue wing (Mg II) with increasing C+O layer metallicity. The changes in the C+O layer metallicity and in the 54Fe abundance of the incomplete burning zone each have effects which are separate from one another and combined effect is essentially the sum of the two effects. 4.3. Day 35 Our grid of synthetic spectra of W7 on day 35 with incomplete burning zone 54Fe abundance variations are shown in Figure 19. The UV pseudo-continuum varies due to the 'surface' cooling and additional line blocking in the incomplete burning zone. The O I feature at 8000 A and the -- 11 -- Si II feature at 6150 A become weaker as oxygen and silicon are displaced by 54Fe. There are several significant effects in the infrared, however, observations which extend far enough into the IR to compare with real SNe Ia are not available. Our grid of synthetic spectra of W7 on day 35 with combined incomplete burning zone 54Fe abundance and C+O layer metallicity variations are shown in Figure 20. The effects of the combined modifications mostly separate in the the effects shown in Figures 13 & 19. The UV pseudo- continuum variation, changes near the 5000 A Fe II feature, and the changes to the 11000A feature reflect the C+O metallicity modification alone. The small shift in the Si II feature at 6150 A and the surrounding pseudo-continuum show the contributions of 54Fe abundance modification. 5. Discussion 5.1. Evolution of Si II Figure 5 shows the evolution of the Si II 6150 A feature to maximum light. The feature grows stronger and steeper as line formation of the Si II λ6355 feature moves into the silicon rich layers of W7. As line formation moves into the silicon rich zones, the effects of C+O layer metallicity on line formation, are reduced, but not eliminated. The blue-shift velocities of the deepest points of the Si II feature are plotted for these models in Figure 21. Except for the extreme case, ζ = 10, the blue-shift velocities increase monotonically with C+O layer metallicity through day 20. The increasing opacity from the C+O layer moves the feature blueward. The velocity shifts due to metallicity are degenerate with changes that could be expected from silicon rich material extending to higher velocities. However, the effects of primary line formation in the C+O versus silicon rich layers can be distinguished by line shapes. These general trends and spreads in blueshift velocities are similar to those seen in the data by Branch & van den Bergh (1993). They found that the slower blueshift velocities tended to be found in earlier galaxy types. A similar study correlating blueshift velocities, peak magnitude (or a suitable proxy), and a more quantitative estimate of the pre-supernova environment metallicity will be the subject of future work. 5.2. Other Features and Effects The UV pseudo-continuum shows the effects of metallicity through the surface cooling of the C+O layer, additional line blocking and the shifting of lines which form at faster moving layers with higher metallicity. The UV pseudo-continuum displacement is relatively constant and still present at 35 days after explosion. The UV displacement over the entire range of models is typically ∼ 0.5 dex. For the more likely range of metallicities, 1/3 to 3 times solar, the change is up to ∼ 0.2 dex or 0.5 magnitude. Unfortunately, we cannot use near-UV flux as a metallicity indicator. The UV flux is diagnostic of the temperature, density, and radius (velocity) of the C+O layer, but is -- 12 -- not uniquely determined for any one quantity. The related backwarming causes pseudo-continuum shifts in the optical and near infra-red for the early epoch spectra. This backwarming shift in continua fades in strength as pseudo-continuum formation moves to deeper layers. A smaller surface cooling/backwarming effect exists in the partially burned layers from changes in 54Fe abundance, beginning at day 20. Hoflich et al. (1998) also report a change in UV pseudo-continuum for models with different metallicity by noting decreases in the U band magnitude with increasing metallicity. They show a change in flux in the UV spectra presented, but the change occurs in the opposite direction with metalicity to that which we find here. This is due to the difference in the density structures between delayed detonation models (DDs) they employ and that of the parameterized deflagration model W7 that we use here. In the DD model the lower metallicity model forms the UV pseudo-continuum at a smaller radius and so the flux is lower for smaller metallicities, since the radial effect wins out over the opacity. Some features which had significant impact from C+O metallicity at early epochs are nearly unaffected at later epochs. At day 7 we find that most of the optical features nearly disappear when the metal content is dropped to 1/30 normal. This indicates that much of line formation for these features takes place in the C+O layer. In later epochs the impact of the C+O layer metallicity on most optical features becomes quite small, since line formation occurs mostly in the deeper layers. The influence of 54Fe abundance in the incomplete burning zone is very small at early epochs, since significant spectrum formation occurs in the C+O layer. At maximum light and later, 54Fe abundance variations change Fe feature strengths, have some small temperature based effects on the pseudo-continuum, and change the strength of certain features from species, e.g. sulfur, which are displaced by 54Fe abundance changes. We have confirmed the Nugent et al. (1997) identification of Si II λ3858 as a component of the Ca II 'split'. To the blue of the Ca II H+K feature is the emission peak from another Ca II line. When the Si II line forms, the Si II absorption falls on that peak and the emission from the Si II line forms a peak in the center of the 'split'. For models with a strong Si II feature, the absorption minimum of the red wing of the Ca II H+K will seem to disappear. For W7, changes in metallicity alter the strength of the Si II line and thus, the shape of the Ca 'split' feature. In real SNe Ia, other factors may prevent the formation of a 'split' such as stronger Ca II H+K, weaker Si II, or temperature effects. Jha et al. (1999) show spectra for several normal SNe Ia. The Ca II H+K feature in the observations display a range of morphologies similar to those seen in our synthetic spectra. We do not require any abundance or ionization effect to produce this feature. While this interpretation is strictly correct in the model W7, we suspect it also produces the observed feature in SNe Ia, however, detailed comparisons of the calculated velocities to those of the observed features are required. -- 13 -- 6. Conclusions By calculating a series of model atmospheres with abundance variations around the base W7 model for SNe Ia, we have demonstrated unexpected and complex effects on the output spectra. The UV spectra show lower flux and blueshifting lines as surface cooling, additional line blocking, and outward movement of the line-forming region occur with higher metallicity in the unburned C+O layer. We have demonstrated, at epochs well before maximum light that line formation occurs largely in the C+O layer for species that will later form in newly synthesized material. The 'splitting' of the Ca II H+K feature we can now better understand as a blend of a Si II feature with the stronger Ca II lines, without abundance or ionization effects in the W7 model. We have shown that the strength, profile, and velocity of the SNe Ia characteristic Si II 6150 A feature are affected by C+O layer metallicity. This provides a mechanism for the variation in the blueshift of the Si II feature without variations in the explosion energy. For exploding white dwarfs of different metallicities Hoflich et al. (1998) and Iwamoto et al. (1999) find changes in nucleosynthesis. We tested these effects on spectra by varying the 54Fe abundance in the incomplete burning zone. We have found that the 54Fe abundance has negligible effect on pre-maximum spectra and relatively little effect afterwards. The effects of progenitor metallicity variations can mostly be separated into effects due to 54Fe, and those due to C+O metallicity. Previous spectral studies (Branch et al. 1985; Harkness 1991a,b; Nugent et al. 1995, 1997) have shown that W7 has the appropriate composition structure to reproduce photospheric era spectra. In preparing the base fits for this paper we have found that the quality of the fits decreases away from maximum light. The temperature structure inversion in the lowest metallicity model at day 35 demonstrates the importance of changes (≈ 20%) in the temperature structure. We see that density/temperature structures are important in fitting spectral features. In forthcoming work we will present the results of numerical experiments to alter the density structure of W7 to make the model better correspond to the observations. We have shown through parameterized abundance modifications of the SNe Ia model W7 that pre-explosion metallicity can have detectable effects on the output spectra at every epoch. Due to the uncertainty in hydrodynamic models and severe blending of lines in SNe Ia spectra, we cannot give a prescriptive analysis tool for measuring the pre-explosion metallicity of SNe Ia. However, hopefully this can contribute to the understanding of the diversity of SNe Ia, and the ways that various progenitor metallicity effects can affect SNe Ia. Studies computing detailed spectra of hydrodynamic models that include progenitor metallicity in the evolution of the star and the hydrodynamics and nucleosynthesis of the supernova should show what effects overall progenitor metallicity have in creating SNe Ia diversity. This work was supported in part by NSF grants AST-9731450 and AST-9417102; NASA grant NAG5-3505, an IBM SUR grant to the University of Oklahoma; and by NSF grant AST-9720704, NASA ATP grant NAG 5-3018 and LTSA grant NAG 5-3619 to the University of Georgia. Some -- 14 -- of the calculations presented in this paper were performed at the the San Diego Supercomputer Center (SDSC), supported by the NSF, and at the National Energy Research Supercomputer Center (NERSC), supported by the U.S. DOE. We thank both these institutions for a generous allocation of computer time. REFERENCES Arnett, W. D. 1996, Supernovae and Nucleosynthesis (Princeton, NJ: Princeton University Press) Baron, E., Branch, D., Hauschildt, P. H., Filippenko, A. V., & Kirshner, R. P. 1999, ApJ, in press Baron, E. & Hauschildt, P. H. 1998, ApJ, 495, 370 Branch, D. 1999, Ann. Rev. Astr. Ap., 36, 37 Branch, D., Doggett, J. B., Nomoto, K., & Thielemann, F.-K. 1985, ApJ, 294, 619 Branch, D. & van den Bergh, S. 1993, AJ, 105, 2231 Drell, P. S., Loredo, T. J., & Wasserman, I. 1999, ApJ, submitted Garnavich, P. M. et al. 1998, ApJ, 493, L53 Hamuy, M., Phillips, M. M., Schommer, R. A., Suntzeff, N. B., Maza, J., & Aviles, R. 1996, AJ, 112, 2391 Harkness, R. 1991a, in Proceedings of the ESO/EPIC Workshop on SN 1987A and other Super- novae, ed. I. J. Danziger & K. Kjar (Munich: ESO), 447 -- . 1991b, in Supernovae, ed. S. E. Woosley (New York: Springer-Verlag), 454 Hatano, K., Branch, D., Baron, E., & Fisher, A. 1999, ApJ, 525, in press Hauschildt, P. H. & Baron, E. 1999, J. Comp. Applied Math., in press Hauschildt, P. H., Baron, E., & Allard, F. 1997a, ApJ, 483, 390 Hauschildt, P. H., Baron, E., Starrfield, S., & Allard, F. 1996, ApJ, 462, 386 Hauschildt, P. H., Schwarz, G., Baron, E., Starrfield, S., Shore, S., & Allard, F. 1997b, ApJ, 490, 803 Hoflich, P. & Khokhlov, A. 1996, ApJ, 457, 500 Hoflich, P., Wheeler, J. C., & Thielemann, F.-K. 1998, ApJ, 495, 617 Iwamoto, K., Brachwitz, F., Nomoto, K., Kishimoto, N., Hix, W. R., & Thielemann, F.-K. 1999, ApJ, submitted -- 15 -- Jha, S. et al. 1999, ApJ Suppl., astro-ph/9906220, in press Millard, J., Branch, D., Baron, E., Hatano, K., Fisher, A., Filippenko, A. V., Kirshner, R. P., Challis, P. M., Fransson, C., Panagia, N., Phillips, M. M., Sonneborn, G., Suntzeff, N. B., Wagoner, R. V., & Wheeler, J. C. 1999, ApJ, in press Nomoto, K., Thielemann, F.-K., & Yokoi, K. 1984, ApJ, 286, 644 Nugent, P., Baron, E., Branch, D., Fisher, A., & Hauschildt, P. 1997, ApJ, 485, 812 Nugent, P., Baron, E., Hauschildt, P., & Branch, D. 1995, ApJ, 441, L33 Nugent, P. E. 1997, PhD thesis, University of Oklahoma Perlmutter, S. et al. 1997a, in Thermonuclear Supernovae, ed. P. Ruiz-Lapuente, R. Canal, & J. Isern (Dordrecht: Kluwer), 749 Perlmutter, S. et al. 1997b, ApJ, 483, 565 -- . 1999, ApJ, 517, 565 Phillips, M. M. 1993, ApJ, 413, L105 Riess, A. et al. 1998, AJ, 116, 1009 Riess, A. G., Press, W. H., & Kirshner, R. P. 1996, ApJ, 473, 88 Sutherland, P. & Wheeler, J. C. 1984, ApJ, 280, 282 Swartz, D., Sutherland, P., & Harkness, R. 1995, ApJ, 446, 766 Thielemann, F.-K., Nomoto, K., & Yokoi, K. 1986, A&A, 158, 17 Tripp, R. & Branch, D. 1999, ApJ, in press This preprint was prepared with the AAS LATEX macros v5.0. -- 16 -- Fig. 1. -- Models with various metallicities in C+O layer at 7 days after explosion. Thick solid denotes 10 times the normal C+O metallicity, thick dotted -- 3 times, short dashed -- normal, long dashed -- 1/3, dot-dashed -- 1/10, and thin solid -- 1/30. -- 17 -- Fig. 2. -- Expansion of ultraviolet region of Figure 1 -- 18 -- Fig. 3. -- Expansion of optical region of Figure 1 -- 19 -- Fig. 4. -- Expansion of infrared region of Figure 1 -- 20 -- Fig. 5. -- Expansion of Si II 6150 A region of a. Figure 1 (Day 7), b. Figure 6 (Day 10), c. Figure 8 (Day 15), and d. Figure 11 (Day 20). -- 21 -- Fig. 6. -- Models with various metallicities in C+O layer at 10 days after explosion. Labels are the same as Figure 1. -- 22 -- Fig. 7. -- Expansion of optical region of Figure 6 -- 23 -- Fig. 8. -- Models with various metallicities in C+O layer at 15 days after explosion. Labels are the same as Figure 1. -- 24 -- Fig. 9. -- Expansion of blue regions of Figure 8 -- 25 -- Fig. 10. -- Expansion of near infrared region of Figure 8 -- 26 -- Fig. 11. -- Models with various metallicities in C+O layer at 20 days after explosion. Labels are the same as Figure 1. -- 27 -- Fig. 12. -- Expansion of optical region of Figure 11 -- 28 -- Fig. 13. -- Models with various metallicities in C+O layer at 35 days after explosion. Labels are the same as Figure 1. -- 29 -- Fig. 14. -- Panels from top to bottom display carbon abundance, density, and temperature profiles for low metallicity models at day 35. The line styles for the temperature profiles are the same as for the corresponding spectra in Figure 13. -- 30 -- Fig. 15. -- Models with various 54Fe abundances in the incomplete burning layer at 20 days after explosion. Solid denotes 3 times the normal 54Fe abundance in the incomplete burning zone, thick dotted -- normal, short dashed -- -- 1/3, long dashed -- 1/10, and dot-dashed -- 1/30. -- 31 -- Fig. 16. -- Expansion of optical region of Figure 15 -- 32 -- Fig. 17. -- Models with various 54Fe abundances in the incomplete burning layer and metallicities in the C+O layer at 20 days after explosion. These models combine the effects in the models in Figures 11 & 15. Solid denotes 3 times the normal 54Fe abundance in the incomplete burning zone and C+O layer metallicity, thick dotted -- normal, short dashed -- -- 1/3, long dashed -- 1/10, and dot-dashed -- 1/30. -- 33 -- Fig. 18. -- Expansion of optical region of Figure 17 -- 34 -- Fig. 19. -- Models with various 54Fe abundances in the incomplete burning layer at 35 days after explosion. Labels are the same as in Figure 15. -- 35 -- Fig. 20. -- Models with various 54Fe abundances in the incomplete burning layer and metallicities in the C+O layer at 35 days after explosion. These models combine the effects in the models in Figures 13 & 19. Labels are the same as in Figure 17. -- 36 -- Fig. 21. -- Blueshift velocities of Si II 6150 A feature as a function of epoch. Down triangle symbols represent 1/30 normal C+O layer metallicity, left triangles 1/10 normal, up triangles 1/3 normal, diamonds normal, squares 3 times normal, and circles 10 times normal metallicity. Data are from lines shown in Figure 5.
astro-ph/9903380
2
9903
2000-03-29T02:04:33
Deviations from the Fundamental Plane and the Peculiar Velocities of Clusters
[ "astro-ph" ]
We have fit the Fundamental Plane of Ellipticals (FP) to over 400 early-type galaxies in 20 nearby clusters (4000 < cz < 11000 km/s), using our own photometry and spectroscopy as well as measurements culled from the literature. We find that the quality-of-fit, r.m.s.[log(sigma)], to the average fundamental plane <FP> varies substantially among these clusters. A statistically significant gap in r.m.s.[log(sigma)] roughly separates the clusters which fit <FP> well from those that do not. In addition, these two groups of clusters show distinctly different behavior in their peculiar velocity (PV) distributions. Assuming galaxies are drawn from a single underlying population, cluster PV should not be correlated with r.m.s.[log(sigma)]. Instead, the clusters with below average scatter display no motion with respect to the cosmic microwave background (CMB) within our measurement errors (~250 km/s), while clusters in the poor-fit group typically show large PVs. Furthermore, we find that all X-ray bright clusters in our sample fit the <FP> well, suggesting that early-type galaxies in the most massive, virialized clusters form a more uniform population than do cluster ellipticals as a whole, and that these clusters participate in a quiet Hubble flow.
astro-ph
astro-ph
Deviations from the Fundamental Plane and the Peculiar Velocities of Clusters R. A. Gibbons1 and A. S. Fruchter2 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218; [email protected]; [email protected] G. D. Bothun2 Department of Physics, University of Oregon, 120 Willamette Hall, Eugene, OR 97403; [email protected] ABSTRACT We have fit the Fundamental Plane of Ellipticals (FP) to over 400 early-type galaxies in 20 nearby clus- ters (cz ∼ 4000 − 11000 km s−1), using our own photometry and spectroscopy as well as measurements culled from the literature. We find that the quality-of-fit, r.m.s.[log(σ)], to the average Fundamental Plane (FP) varies substantially among these clusters. A statistically significant gap in r.m.s.[log(σ)] roughly separates the clusters which fit FP well from those that do not. In addition, these two groups of clusters show distinctly different behavior in their peculiar velocity (PV) distributions. Assuming galaxies are drawn from a single underlying population, cluster PV should not be correlated with r.m.s.[log(σ)]. Instead, the clusters with below average scatter display no motion with respect to the cosmic microwave background (CMB) within our measurement errors (∼ 250 km s−1), while clusters in the poor-fit group typically show large PVs. Furthermore, we find that all X-ray bright clusters in our sample fit the FP well, suggesting that early-type galaxies in the most massive, virialized clusters form a more uniform population than do cluster ellipticals as a whole, and that these clusters participate in a quiet Hubble flow. Subject headings: galaxies: fundamental parameters -- galaxies: clusters -- cosmology: cosmic microwave background -- cosmology: distance scale 1. Introduction 1Department of Astronomy, University of Maryland, College Park, MD 20742 2Visiting Astronomer, Kitt Peak National Observatory, National Optical Astronomy Observatories, which is operated by the Association of Universities for Research in Astronomy, Inc. (AURA) under cooperative agreement with the National Science Foundation. -- 2 -- 1.1. The Fundamental Plane of Ellipticals The high correlation of the structural and kinematic properties of ellipticals suggests that these galaxies formed by similar processes and are largely virialized. Assuming structural symmetry, isotropic velocities, hv2i = σ2, a constant mass-to-light ratio (M/L), and structural homology, a tight correlation is expected as a natural result of virialization : σ2 ∝ hSBir, (1) where σ2 = the velocity dispersion within the galaxies, r = a fiducial radius, and hSBi the average surface brightness within r. In log space this expression relating basic physical properties collapses to a plane and, hence, is called the Fundamental Plane of Ellipticals (Djorgovski & Davis 1987; Lynden-Bell et al. 1988; Lucey, Bower, & Ellis 1991). In terms of observables the FP becomes log(re) = α log(σ) + β hµie + Const, (2) with α = 2, β = 0.4 representing the virial plane. Standard units of measure are km s−1 for the central velocity dispersion, σ, arcseconds for the r1/4−law half-light radius, re(effective radius), and magnitudes per arcsec2 for the average surface brightness within that radius, hµie(effective surface brightness). Empirically, ellipticals occupy a narrow range within this three-dimensional parameter space indicating they are a homologous family. S0 galaxies also lie on the FP due to the fact that, while errors in the estimated re and hµie are made by fitting r1/4−law profiles to galaxies with disk components, the errors cancel such that the FP quantity, "mboxlog(re) − β hµie", is left remarkably unaffected (Saglia et al. 1997; Scodeggio et al. 1998; Kelson et al. 1999). The observed FP, however, is misaligned with the virial prediction. Ellipticals are better described by, α ∼ 1.4 and β ∼ 0.3. This tilt of the FP implies the assumptions used in Equation 1 are not quite true. Nonetheless, the uniformity of ellipticals makes these galaxies important standard candles since the observed re provides a direct indication of distance (Dressler 1987; Dressler et al. 1987; Lucey & Carter 1988). The small observed scatter about the FP translates to ∼ 19% error in the distance to individual galaxies (Jørgensen, Franx, & Kjaergaard 1996; Hudson et al. 1997; present work). The FP is thus comparable as a distance indicator to the Tully-Fisher (TF) relation applied to spirals (de Carvalho & Djorgovski 1989). 1.2. Motivation & Goal for this Study The FP, therefore, should be a powerful tool for measuring relative cluster distances as well as deviations from the Hubble flow. However, surveys of large scale motions using the FP and TF have found significantly different cosmic flows. For example, the recent FP "Streaming Motions of Abell Clusters" survey (SMAC) (Hudson et al. -- 3 -- 1999) reveals a large bulk peculiar motion on the galaxy cluster scale. Yet, while SMAC agrees with the 15 cluster TF survey (LP10K) of Willick et al. 1999, most TF surveys have shown little peculiar motion on these scales. The large and somewhat deeper TF cluster survey of Dale et al. 1999a limits the motion to less than 200 km s−1, agreeing with TF surveys going back to the Aaronson et al. 1986 TF survey. To further confuse the issue, Lauer & Postman 1994, using a novel method employing brightest cluster galaxies in 119 clusters, detected a large bulk flow of the "Abell Cluster Inertial Frame" (ACIF). But the direction is in conflict with the results of the aforementioned surveys, none of which have the large sky coverage of the ACIF survey. As can be seen from the compilation of previous measurements (Table 1), the results of these surveys are clearly discrepant or their errors have been underestimated. Indeed, some of the possible sources of systematic errors, as summarized by Jacoby et al. (1992), are still not well understood. Moreover, as emphasized by Strauss et al. 1995, if flows on this large a scale exist, then they can not be accommodated by any present cosmological model. These models assume the average random motion of clusters atop the Hubble flow is driven by gravity and the underlying mass distribution and thereby provides a constraint on the density parameter, Ω (Borgani et al. 1997; Watkins 1997; Bahcall & Fan 1998). To address this problem, we examine the FP and rigorously test the assumption that the FP is invariant from cluster to cluster. If this assumption is violated, then previous surveys, which have assumed a uniform fundamental plane, could contain significant systematic errors. We have compiled a sample of over 400 early-type galaxy in 20 clusters. Measured distances to clusters should be more accurate than to individual galaxies by the root number of galaxies observed per cluster, and thus typical distance errors to individual clusters should be of order ∼ 5%. Hence, flows on cluster scales can in principle be accurately determined. Furthermore, this sample allows us to investigate the effects of possible systematic errors on a cluster by cluster basis. The rest of this paper is organized as follows: in §2 we briefly describe our data as a thorough presentation will be the subject of a subsequent paper; we discuss our FP fitting procedure in §3; results and statistical analysis are presented in §4; we finish with further discussion and implications of our result in §5; and summarize our conclusions in §6. 2. The Data Our observations have yielded useful spectroscopy and photometry of 132 E+S0 galaxies in 8 Abell clusters. Spectra have been obtained with the Nessie multi fiber instrument on the KPNO Mayall 4m, and imaging with a large format (20482) Tek CCD on the KPNO 0.9m. Redshifts and central velocity dispersions are measured by matching the galaxy spectra with G and K class stellar templates, using the Fourier-quotient technique (Sargent et al. 1977; Tonry & Davis 1979) adapted for use in IRAF by Kriss (1992). Aperture corrections are applied to the velocity dispersions, -- 4 -- due to the fact that the fibers, which are of fixed diameter, sample larger portions of galaxies at larger distances, systematically lowering measured velocity dispersions. For model ellipticals, Jørgensen, Franx, & Kjaergaard 1995 find the velocity dispersion measured within an aperture varies with aperture size as a power law of index of −0.04. The applied aperture correction to measured velocity dispersions is then calculated using cluster redshift as a first order estimate of distance and the angular diameter−distance relation with q0 = 0.5. Photometry has been corrected for atmospheric extinction, Galactic absorption (Burstein & Heiles 1984), (1+z)4 cosmological dimming, and k-correction. The k-correction to flux scales as (1 + 1.089z), calculated for an average elliptical galaxy spectrum (Coleman, Wu, & Weedman 1980) through our CCD and the R filter. We work in R band to help minimize the effects of cluster differences in age and metallicity which can cause shifts in the FP at shorter wavelengths (Guzm´an & Lucey 1993; Gregg 1995). Effective radii are calculated by fitting an r1/4−law to isophotes, which have been obtained using the ELLIPSE task in IRAF. A grid of seeing corrections to re and hµie is calculated from models convolved with the PSFs of the images. Merging our data with recently published observations, increases the number of galaxies to 428 in 20 clusters. Table 2 lists the literature sets as well as abbreviations used hereafter. Velocity dispersions, effective radii, and surface brightnesses have been normalized to the SLHS97 system by minimizing residuals of each of these parameters for 56 galaxies in four clusters common to both samples. The details of our observations and this conversion including seeing and aperture corrections will be described in detail in Gibbons, Fruchter, & Bothun (2000c). For some clusters, e.g. A400, our sample size is limited because we exclude objects with strong emission lines and evidence of a strong disk component. We retain only galaxies which are clearly cluster members and which have excellent spectroscopic S/N. Average galaxy FP errors for clusters are comparable between the three data sets except for A400, whose spectra are of lower quality than the other clusters. However, excluding this cluster from our analysis does not alter our results. 3. Fitting to All Data Simultaneously Once the corrections described in §2 are applied to the measured σ's and hµie's, shifts in the FP should be due purely to the change in apparent galaxy size with distance. The FP is then fit to all the galaxy data simultaneously by letting cluster distance be a free parameter. Specifically, we solve a set of equations with 428 galaxies in 20 clusters, log(re) = α log(σ) + β hµie + γi , i = 1, 20 (3) where shifts in the intercepts, ∆γi, reflect the offsets in log(re), which translate into the relative distances between clusters. Because one is seeking high precision in relative distances it may seem most prudent to minimize the residuals -- 5 -- about the FP in the direction of the distance dependent parameter, log(re). This particular projection also yields the smallest observed scatter due to the strong correlation between re and hµie. However, Lucey, Bower, & Ellis (1991) as well as Jørgensen, Franx, & Kjaergaard (1996) have recognized that minimizing the residuals of log(re) will introduce a bias into the fit because the errors in log(re) and hµie are correlated, since hµie is a function of re (see also Akritas & Bershady 1996). Therefore, like SLHS97, we isolate the velocity dispersion, σ, as the only parameter with independent errors. Explicitly, we write the FP solely as a function of re, log(σ) = f (re) = (log(re) − β hµie − γi)/α , i = 1, 20 and solve for the coefficients, which minimize the absolute residuals, X i=1,20 X j=1,ni log(σ)j − fij(re) (4) (5) where ni = the number of galaxies per cluster. When the absolute residuals are minimized, galaxies which lie far off the main relation have effectively less weight than they would under least squares. Our use of σ as the indepen- dent variable avoids the strong bias caused by the correlated errors in re and hµie when re is instead treated as the independent variable. We solve this system of equations allowing 22 free parameters: 20 cluster intercepts, γi's, plus common α and β. In this way, we find the average fundamental plane, FP, which best fits the entire sample of galaxies, further reducing the influence of anomalous galaxies. To ensure the best solution within a parameter space of such complexity (22 dimensions) requires an efficient searching algorithm. We adopt simulated annealing, because it efficiently converges towards the global minimum while avoiding becoming trapped in local minima (Metropolis et al. 1953; Kirkpatrick, Gelatt, & Vecchi 1983; Vanderbilt & Louie 1984). The convergence is quite rapid and most of the computing time is spent sampling the region near best fit. 4. Results 4.1. The FP Derived from the Present Sample The best-fit FP, α = 1.37±0.04, β = 0.331±0.004, is shown in Figure 1a and is in agreement with that found by Hudson et al. (1997). We estimate the errors on α and β by fixing γi, varying α and β, and constructing maps of chi-square versus α and β. Our chi-square assumes the average total FP scatter (intrinsic plus measurement error) is well represented by the measured average r.m.s. scatter in log(σ) about FP, hΣi. Then χ2 = n X j=1 [log(σj) − FP]2/hΣ2i. (6) -- 6 -- The one sigma errors, corresponding to ∆χ2 = 1, are read directly from these curves. For our fit, the average r.m.s. scatter in log[σ], hΣi = 0.065, is equivalent to a 20% error in distance to individual galaxies. Assuming the average scatter applies to all clusters, errors in cluster distances range from about 2 − 8%, depending on the number of galaxies fit per cluster. However, 9 of the 20 clusters, about half the total sample, show significantly more scatter than the other 11 (hereafter Σ denotes FP r.m.s. scatter). When these clusters are fit separately, they also fail to define a tight FP. We must therefore investigate the significance of the larger scatter: i.e., whether or not the high-Σ fits are truly poor. The degree to which the FP describes the individual clusters can be seen in Figure 1b. Assuming galaxies in a cluster are chosen randomly from a single population, we calculate for each cluster the significance, or probability, P , of a chi-square this large about the FP and plot this value against Σ. Immediately evident is a break in P coincident with a separation in Σ about hΣi. While the clusters should uniformly fill probability space, a gap this large anywhere in P has less than a 0.1% chance of occurring (Similarly, along the Σ axis, one would expect the points to be clustered about hΣi rather than showing the deficit that is observed). However, as this estimate of probability relies on the assumption that the FP is gaussian, we have also estimated the probability of this gap by simulating many trial sets of clusters from the total sample of observed galaxies and calculating the distribution of the maximum gap. Doing so again implies the observed situation is improbable, with an expected occurrence of a gap this size of 0.2%. This strongly suggests that our sample has not been drawn from a single galaxy distribution. We therefore refit for FP based on the 11 best-fit clusters (Figure 2a). Although the FP coefficients are only slightly different, α = 1.39±0.04, β = 0.335±0.005, and hΣi changes by ∼ 10%, the 9 clusters with Σ > hΣi clearly appear to be outliers, falling in the upper 5% tail in probability space (Figure 2b). If truly deviant, these clusters may not be reliable objects for FP distance work. But do the high-Σ clusters distinguish themselves from the rest in any other way? PV is a second, independent dimension in which we can examine the behavior of this sample and is after all the quantity we aim to measure. 4.2. Measuring Distances and Peculiar Velocities To derive peculiar velocities, we must first fix the scale between FP relative distances and absolute distances. That is, we must set the conversion between γ and cluster redshift. We begin with the assumption that the clusters are at rest with respect to the Hubble flow, i.e. at rest in the CMB frame. Cluster line-of-sight velocities are translated from heliocentric redshifts, as determined from our spectra, to redshifts in the CMB frame using a vector in the direction l = 264.◦4, b = 48.◦4 in Galactic coordinates with a magnitude of 369.5 km s−1, (Kogut et al. 1993). Taking into account (the small) cosmological corrections, we then find the angular diameter − distance zero-point which -- 7 -- minimizes the one-dimensional r.m.s. peculiar velocity, σ1D, of the clusters. This normalization does not change significantly if we minimize σ1D for the eleven best-fit clusters instead of for all twenty. Interestingly, this procedure places the Coma cluster nearly at rest, so we would have recovered a similar result had we anchored the physical scale by assuming no peculiar motion for Coma (which has traditionally been done). Finally, plotting PV against Σ, we uncover an astonishing result (Figure 3). There is an obvious discontinuity in the scatter in cluster PV which occurs across the same region of Σ-space as the gap in probability already discussed. The clusters which fit FP well are at rest with respect to the CMB with reduced χ2 < 1 while the poor-fit clusters with Σ > hΣi show large scatter in their PVs, χ2 = 4.4. 3 Results from both the 20 cluster FP and 11 cluster FP are similar as can be seen in Figure 4. While one might naively expect a correlation between Σ and the measurement error of PV, this is not the case. If cluster ellipticals do indeed form one population, then the sample mean and sample standard deviation of a randomly drawn subset will be uncorrelated. Therefore, PV should not reflect the value of Σ. Similarly PV errors do not correlate with Σ; we therefore scale the errors by hΣi. As we will show in the following section, this null hypothesis is heavily disfavored as the observed non-uniformity in σ1D is highly significant. 4.3. Significant Variation in σ1D To examine the significance of the variation in σ1D, we have used two approaches, linear regression and analysis of variance (ANOVA). Simple regression favors a strong correlation between PV and Σwhether we use hΣi to estimate the PV errors (null hypothesis), or use the individual Σ. Straightforward weighted regressions of "PV on Σ" and "Σ on PV" yield slopes which are nonzero with 99.9% confidence. The true slope lies somewhere between the forward and inverse regressions. But because of the two-way errors, it is difficult to decide significance analytically, so the problem can also be dealt with numerically. Resampling by complete permutation along the Σ axis is a measure of the likelihood of the observed correlation. Permutation ignores errors in Σ and complete mixing of the 20 points without replacement is allowed. Under the assumption of no correlation, the expected slope is zero, negative and positive slopes being equally likely. The resultant distribution shows the observed slope to be unlikely, with a probability of less than one percent. 4 The null hypothesis likewise fails ANOVA tests. The measured σ1D for the two subsamples of clusters (defined 3 Fractional errors in distance are predicted based on hΣi and the FP coefficient, α, and scale inversely with √ngal. Estimated PV errors are distance errors added in quadrature with redshift errors. 4 Note, however, if two separate populations do not show such a correlation, this test is inappropriate, in which case ANOVA is more meaningful. -- 8 -- by the gap in Σ) are not consistent with a single PV distribution. The ratio of chi-squares disfavors the two groups being drawn from the same distribution with probability 99.94%. 5 Again permuting the points along Σ−space allows us to estimate the probability of the present result without an assumption about the nature of the distribution. Using this bootstrapping procedure one finds the significance for the ratio of observed chi-squares is again above the 99th percentile. The behavior of the sample both in Σ−space as well as in PV−space tends to contradict at very high levels of significance the assumption that we are observing a single population of clusters. Taken together, these statistics are convincing evidence that we are seeing two populations. If this bimodality is an intrinsic property of clusters of galaxies, then all current FP samples potentially mix together these two apparently different populations. 5. Errors in FP Distances or Peculiar Velocities? We have established that our sample of clusters divides into two groups defined by significantly different intrinsic scatter about the FP. Additionally, we find that these two groups have radically different PV distributions. The obser- vation that only the high-Σ clusters show peculiar motions implies that such clusters can inject significant error into the measure of σ1D. This could have the consequence of producing a spuriously large PV signal for the entire sample. Alternatively, these two groups could indeed be subject to truly different kinematics. Either scenario, suggests that the nature of the inferred PV field is sample dependent. We find, however, no evidence that this apparent separation of the clusters into two populations is due to system- atic problems with our data. There is no correlation of Σ with cluster distance (e.g. more distant clusters do not have poorer fits). We see no effect due to data quality or data source; the clusters which we observed ourselves and the sample which was culled from the literature are equally divided between the two Σ groups, and with the exception of A400, the quality of the imaging and spectra of all the clusters are comparable. We are also confident that our cluster samples are not significantly contaminated by field ellipticals or S0s; we are sampling the centers of clusters over an area of approximately one half square degree and our redshift criterion is z−zmean ∼< 3σ, where σ = cluster velocity dispersion. We calculate that, with the possible exception of 7S21, the expected number of field interlopers per cluster in our sample is far less than one. Addition of an Mg2 line index term to diminish cluster to cluster differences in the age/metallicity of the stellar populations also does not improve the fits. We find no strong evidence for curvature along the FP. In addition, the galaxies in each cluster cover similar ranges in log(σ) so that we expect any error introduced by sampling different portions of the FP is at most a minor effect in our final fits and distance measurements. The 5 This is not the traditional F-test because the comparison is not between two independent fits. For this problem, the number of degrees of nclust × nsubgroup. Specifically for nclust = 20, the d.o.f. are 8.55 and 10.45 for nsubgroup = 9 and 11 freedom for the subgroups is nclust−1 respectively. Furthermore, the ratio of χ2 is no longer purely F-distributed, so we have computed the proper cumulative distribution function. -- 9 -- high-Σ clusters do on average have a smaller number of observed members than low-Σ clusters. However, we would expect a small sample size to only increase the scatter of Σ and not bias its value. Bootstrap tests on subsamples of galaxies chosen at random from Coma show this to be the case. Rather, the high-Σ clusters are also on average the least rich clusters, and as we discuss below, there is good reason to believe that Σ is a function of cluster richness. If our study is relatively free from systematic errors, then our statistical analysis shows that different clusters have different amounts of intrinsic log(σ) scatter about the FP. But what property of a cluster might be driving this observed difference? We have looked at many cluster properties as predictors of whether a cluster will be a high- or low-Σ cluster and have found only one strong indicator : cluster X-ray luminosity, as can clearly be seen in Figure 5. The four brightest X-ray clusters, those with Lx > 1044 erg s−1, all have low Σ. This result suggests that early-type galaxies within the most massive, well-virialized clusters form a more homogeneous population than cluster ellipticals as a whole. A weaker, but similar correlation, can be seen with cluster spiral fraction, which is also an indirect indicator of the dynamical evolutionary state of a cluster (see for example Dressler et al. 1997). Clusters with Σ > 0.08 exhibit a spiral fraction of 55 ± 1% compared to 42 ± 3%. This trend of Σ with spiral fraction is shown in Figure 6. Similarly, Σ appears to be related to the cluster velocity dispersion, which is itself, of course, closely tied to cluster X-ray luminosity (Figure 7). With one exception, clusters with velocity dispersions less than 400 km s−1 tend to have high Σ whereas those with dispersions greater than 800 km s−1 tend to have low Σ. There are also indications that substructure may be an important component of the increased scatter observed in some clusters. Take the case of the relatively spiral rich, low velocity dispersion clusters A2151 and A400. In A2151, the early type galaxies in the core have a mean velocity offset of −900 km s−1 from that of the spirals (Zabludoff et al. 1993, Dale et al. 1999b; this paper). Indeed, A2151 and A2147 are clearly entangled in a mutual interaction (e.g. Bird et al. 1995; Maccagni et al. 1995), with a component of peculiar velocity reflecting infall and/or merging of subclusters. A similar offset between mean elliptical and cluster velocity has been observed in A400 by Beers et al. (1992). In addition, both A400 and A2151 contain many examples of interacting/peculiar galaxies (see Bothun & Schommer 1982), indicating the prevalence of low velocity dispersion subgroups within the overall structure. Furthermore, substructures that are separated by ∼ 10 Mpc or less will not be adequately resolved in kinematic distance space by the current relative indicators. An increase in Σ can be introduced due to a range in distance along the line of site. Indeed, such a spread in distances has already been proposed for one rich, but high Σ cluster, A2634 (see West & Bothun 1990; Scodeggio et al. 1995). One reason for the fact that highly X-ray luminous clusters do not show significant peculiar velocity may be that X-ray luminosity coincides with the virialization of infalling structures. Whatever the source of the consistent behavior of these clusters, it appears that X-ray luminosity provides a good pre-filter for the selection of clusters to be -- 10 -- used in PV studies. As will be shown in paper II of this series (Gibbons, Fruchter, & Bothun 2000b), such a pre-filter produces a sample that exhibits no large scale flow. 6. Conclusions Using high-quality data and a robust fitting technique, we have determined that clusters of galaxies show a large range in quality-of-fit of their member galaxies to the FP. Statistical tests described in §4 argue that the observed separation in Σ is significant at the ∼ 10−3 level. Additionally, an examination of the distribution of clusters along the PV axis provides further direct support for this separation. If clusters are truly random realizations of galaxies drawn from a single population, then the total sample should have, on average, a uniform σ1D. Instead, when the sample is divided on the basis of the observed scatter about the FP, the two subsamples have drastically different properties in the independent dimension provided by the PV. Again, this effect is statistically strong at more than the 99% level. Simply put, high-Σ clusters have large PV and low-Σ clusters have essentially no PV, within the observational errors. Taken together, these two results cannot be reconciled with the idea that these clusters are randomly drawn from a single population. The behavior of Σ appears to be an intrinsic cluster property. We have found that a subset of the clusters, the X-ray luminous clusters, exhibit a well-defined FP. This suggests that the most massive and virialized clusters possess a homogeneous population of ellipticals and therefore provide the most accurate measures of relative distance. About half of the cluster sample possess Σ below the mean. When using this half of the sample, we find the clusters to be at rest within the CMB frame (see details in the forthcoming Gibbons et al. 2000b). The high-Σ clusters provide the bulk of the positive PV signal when all the clusters are averaged together. Overall, our results strongly suggest that making a cut on the r.m.s. scatter about the FP, or more strictly on cluster X-ray luminosity, will lead to a more reliable sample from which bulk flow properties can be established. The current level of disagreement between various observations of large scale flows (e.g. Table 1) may possibly reflect the fact that less reliable samples have been used. Aaronson, M., Bothun, G. D., Mould, J. R., Huchra, J. P., Schommer, R. A., & Cornell, M. E. 1986, ApJ, 302, 536 REFERENCES Akritas, M. G. & Bershady, M. A. 1996, ApJ, 470, 706 Bahcall, N. A. & Fan, X. 1998, ApJ, 504, 1 Bird, C. M., Davis, D. S., & Beers, T. C. 1995, AJ, 109, 920 -- 11 -- Bothun, G. D. & Schommer, R. A. 1982, AJ, 87, 1368 Borgani, S., da Costa, L. N., Freudling, W., Giovanelli, R., Haynes, M. P., Salzer, J., & Wegner, G. 1997, ApJL, 482, 121 Burstein, D. & Heiles, C. 1984, ApJS, 54, 33 Beers, T. C., Gebhardt, K., Huchra, J. P., Forman, W., Jones, C., & Bothun, G. D. 1992, ApJ, 400, 410 Coleman, G. D., Wu, C-C, & Weedman, D. W. 1980, ApJS, 43, 393 Courteau, S., Faber, S. M., Dressler, A., & Willick, J. A. 1993, ApJL, 412, 51 Carvalho, de R. R. & Djorgovski, S. 1989, ApJ Lett., 341, 37 Carvalho, de R. R. & Djorgovski, S. 1992, ApJ Lett., 389, 49 Dale, D. A., Giovanelli, R., Haynes, M. P., Campusano, L. E., Hardy, E., & Borgani, S. 1999a, ApJL, 510, 11 Dale, D. A., Giovanelli, R., Haynes, M. P., Campusano, L. E., & Hardy, E. 1999b, AJ, 118, 1489 Djorgovski, S. & Davis, M. 1987, ApJ, 313, 59 Dressler, A. 1987, ApJ, 317, 1 Dressler, A., Oemler Jr., A., Couch, W. J., Smail, I., Ellis, R. S., Barger, A., Butcher, H., Poggianti, B. M., & Sharples, R. M. 1997, ApJ, 490, 577 Dressler, A., Lynden-Bell, D., Burstein, D., Davies, R. L., Faber, S. M., Wegner, G., & Terlevich, R. 1987, ApJ, 313, 42 Franx, M., Illingworth, G., & Heckman, T. 1989, ApJ, 344, 613 Gibbons, R. A., Fruchter, A. S., & Bothun, G. D. 2000b, in prep Gibbons, R. A., Fruchter, A. S., & Bothun, G. D. 2000c, in prep Giovanelli, R., Haynes, M. P., Herter, R., Vogt, N. P., Wegner, G., Salzer, J. J., da Costa, L. N., & Freundling, W. 1997, AJ, 113, 22 Giovanelli, R., Haynes, M. P., Wegner, G., da Costa, L. N., Freundling, W., & Salzer, J. J. 1996, ApJ Lett., 464, 99 Gregg, M. D. 1995, ApJ, 443, 527 Guzm´an, R. & Lucey, J. R. 1993, MNRAS, 263, 47 Hudson, M. J., Lucey, J. R., Smith, R. J., & Steel J. 1997, MNRAS, 291, 461 Hudson, M. J., Smith, R. J., Lucey, J. R., Schlegel, D. J., & Davies, R. L. 1999, ApJL, 512, 79 Jacoby, G. H., Branch, D., Ciardullo, R., Davies, R. L., Harris, W. E., Pierce, M. J., Pritchet, C. J., Tonry, J. L., & Welch, D. L. 1992, PASP, 104, 599 -- 12 -- Jørgensen, I., Franx, M., & Kjaergaard, P. 1995a, MNRAS, 273, 1097 Jørgensen, I., Franx, M., & Kjaergaard, P. 1995b, MNRAS, 276, 1341 Jørgensen, I., Franx, M., & Kjaergaard, P. 1996, MNRAS, 280, 167 Kogut, A. et al. 1993, ApJ, 419, 1 Kelson, D. D., Illingworth, G. D., van Dokkum, P. G., & Franx, M. 1999, astro-ph/9911065 (accepted for publication in the ApJ) Kirkpatrick, S., Gelatt, C. D., Vecchi, M. P. 1983, Science, 220, 671 Kriss, J. 1992, IRAF package stsdas.contrib.redshift Lauer, T. & Postman, M. 1994, ApJ, 425, 418 Lucey, J. R., Bower, R. G., & Ellis, R. S. 1991, MNRAS, 249, 755 Lucey, J. R. & Carter, D. 1988, MNRAS, 235, 1177 Lucey, J. R., Guzm´an, R., Smith, R. J., & Carter, D. 1997, MNRAS, 287, 899 Lynden-Bell, D., Faber, S. M., Burstein, D., Davies, R. L., Dressler, A., Terlevich, R. J., & Wegner, G. 1988, ApJ, 326, 19 Maccagni, D., Garilli, B., & Tarenghi, M. 1995, AJ, 109, 465 Metropolis, N., Rosenbluth, A., Rosenbluth, M., Teller, A., & Teller, E. 1953, J. Chem. Phys., 21, 1087 Sargent, W. L. W., Schechter, P. L., Boksenberg, A., & Shortridge, K. 1977, ApJ, 212, 326 Scodeggio, M., Giovanelli, R., & Haynes, M. P. 1998, AJ, 116, 2728 Scodeggio, M., Solanes, J. M., Giovanelli, R., & Haynes, M. 1995, ApJ, 444,41 Smith, R. J., Lucey, J. R., Hudson, M. J., & Steel J. 1997, MNRAS, 291, 461 Strauss, M. A., Cen, R., Ostriker, J. P., Lauer, T. R., & Postman, M. 1995, ApJ, 444, 507 Tonry, J. L., & Davis, M. 1979, AJ, 84, 1511 Vanderbilt, D. & Louie, S. G. 1984, J. Comp. Phys., 56, 259 Watkins, R. 1997, MNRAS, 292, 59 West, M. J. & Bothun, G. D. 1990, ApJ, 350, 36 Whitmore, B. C., Gilmore, D. M., & Jones, C. 1993, ApJ, 407, 489 White, D. A., Jones, C., & Forman, W. 1997, MNRAS, 292, 419 Willick, J. A., Courteau S., Faber, S. M., Burstein, D., & Dekel, A. 1995, ApJ, 446, 12 -- 13 -- Zabludoff, A. I., Geller, M. J., Huchra, J. P., & Vogeley, M. S. 1993, AJ, 106, 1273 This preprint was prepared with the AAS LATEX macros v4.0. -- 14 -- Survey ACIF SMAC LP10K Dale et al. 1999 a&b Watkins 1997 Method BCG FP IRTF IRTF IRTF PV w.r.t. CMB (km s−1) Survey Depth (km s−1) σ1D (km s−1) 689 ± 178 (l = 343 b = 52) 630 ± 200 (l = 260 b = −1) 720 ± 280 (l = 266 b = 19) < 200 -- 15,000 12,000 12,000 18,000 12,000 ∼ 400 ∼ 350 ∼ 415 341 ± 93 265+106 −75 Table 1: Recent Surveys : These surveys disagree in both the magnitude and the direction of their derived flows. References appear in the text. N.b. The Watkins analysis is of subsets of the Giovanelli et al. (1997) and Willick et al. (1995) surveys. cz(km/s) 7215 6708 9923 10406 10188 4636 9425 4714 8947 4417 9132 5312 5139 5517 5122 8615 11471 4873 8473 3976 l(deg) 57.6 170.2 203.1 12.9 31.6 130.2 150.3 126.8 62.9 136.6 103.5 140.7 150.5 113.8 142.2 195.6 240.3 317.4 25.3 269.6 b(deg) 88.0 -44.9 67.8 49.7 44.5 -27.0 -34.4 -30.3 43.7 -25.1 -33.7 -18.1 -13.7 -40.0 -62.9 -17.6 -22.7 31.0 -75.8 26.5 cluster σclust 952 A1656 326 A400 A1185 509 651 A2063 A2151 393 358 HMS0122 638 J8 436 Pisces A2199 636 540 A262 781 A2634 768 A347 A426 1364 285 7S21 747 A194 891 A539 A3381 171 535 A3574 A4038 769 644 A1060 1 = this work; 2 = Lucey & Carter 1988 (LC88); 3 = Jørgensen, Franx & Kjaergaard 1995a,1995b (JFK95a,JFK95b); 4 = Lucey, Guzm´an, Smith & Carter 1997 (LGSC97); 5 = Smith, Lucey, Hudson & Steel 1997 (SLHS97) Data Ref. 1,4 1 1 1 1 5 5 5 4 1,5 1,4 5 1,5 5 3,2 3 3 3 3,2 3,2 Σ FP 0.059 0.112 0.045 0.088 0.086 0.058 0.087 0.055 0.063 0.054 0.073 0.059 0.062 0.092 0.074 0.053 0.084 0.090 0.057 0.056 Σ αIN D βIN D FPIN D 0.052 1.54 0.86 0.149 0.057 1.62 0.096 1.88 1.50 0.070 0.068 1.87 0.096 1.37 0.047 1.10 1.32 0.059 0.061 1.67 0.072 1.37 0.061 1.26 1.45 0.065 0.130 1.94 0.069 1.04 0.055 1.46 1.72 0.076 0.115 1.49 0.061 1.36 1.58 0.056 0.305 0.301 0.267 0.299 0.257 0.293 0.340 0.369 0.359 0.308 0.311 0.348 0.357 0.179 0.333 0.327 0.227 0.352 0.342 0.347 ngal 79 6 10 17 9 9 13 25 36 15 38 8 49 7 19 22 14 7 27 18 Table 2: The Cluster Sample : column (1) cluster name; (2-3) cluster position in Galactic coordinates; (4) heliocentric redshift; (5) cluster velocity dispersion; (6) number of observed galaxies; (7) data references as defined above; (7) scatter about the average FP; (8-9) FP parameters for the fits to individual clusters. -- 15 -- Fig. 1. -- (a) The FP for the entire sample of 428 galaxies. This projection shows the scatter about log(re). (b) The significance of χ2 about FP for each cluster vs. FP scatter. The gap seen in probability space is highly significant, making it unlikely that all clusters have been drawn from a single population. -- 16 -- Fig. 2. -- (a) The FP for the 11 best-fit clusters. (b) Same as Figure 1b but for the 11 cluster FP fit. When fit this way the high-Σ clusters are outliers, falling in the upper 5% tail in probability space, while the rest are consistent with being uniformly distributed. -- 17 -- Fig. 3. -- Cluster PV vs. Σ. Each point represents a cluster containing ∼ 10 − 80 observed galaxies. The x-axis is the scatter within the cluster with respect to FP. The abscissa is the PV as derived from redshift minus FP distance. The reduced χ2 about zero PV for the entire sample is shown at the top of the figure. hΣi (vertical line) roughly marks the separation of the sample into two subgroups, over which χ2 rises dramatically (values shown at the bottom of the panel). The errors on Σ are shown in the bottom panel for clarity. One can see a paucity of clusters about hΣi, where one would expect the density of clusters to be highest if clusters are drawn from a single population. -- 18 -- Fig. 4. -- Comparison of the PV distribution for the 20 and 11 cluster fits. As is readily seen, use of one fit or the other does not significantly change our results. -- 19 -- Fig. 5. -- Σ vs. X-ray luminosity, Lx. Note that the massive, virialized clusters fit the FP well, while the quality-of-fit for the lower mass clusters cannot be predicted. The most X-ray bright clusters (Lx ∼> 1044 erg s−1) are labeled. The clusters for which there exist no X-ray data are shown along the left side of the panel. Lx from White, Jones, & Forman (1997). Fig. 6. -- Σ vs. cluster spiral fraction (SF). A weak correlation between these two variables is evident and is in the same sense as expected from the X-ray data. Spiral fractions from Dressler 1980. -- 20 -- Fig. 7. -- Σ vs. cluster velocity dispersion. The high dispersion (∼> 800 km s−1) clusters have low Σ.
astro-ph/0011122
1
0011
2000-11-06T15:59:05
The Turbulent Shock Origin of Proto--Stellar Cores
[ "astro-ph" ]
The fragmentation of molecular clouds (MC) into protostellar cores is a central aspect of the process of star formation. Because of the turbulent nature of super-sonic motions in MCs, it has been suggested that dense structures such as filaments and clumps are formed by shocks in a turbulent flow. In this work we present strong evidence in favor of the turbulent origin of the fragmentation of MCs. The most generic result of turbulent fragmentation is that dense post shock gas traces a gas component with a smaller velocity dispersion than lower density gas, since shocks correspond to regions of converging flows, where the kinetic energy of the turbulent motion is dissipated. Using synthetic maps of spectra of molecular transitions, computed from the results of numerical simulations of super--sonic turbulence, we show that the dependence of velocity dispersion on gas density generates an observable relation between the rms velocity centroid and the integrated intensity (column density), Sigma(V_0)-I, which is indeed found in the observational data. The comparison between the theoretical model (maps of synthetic 13CO spectra), with 13CO maps from the Perseus, Rosette and Taurus MC complexes, shows excellent agreement in the Sigma(V_0)-I relation. The Sigma(V_0)-I relation of different observational maps with the same total rms velocity are remarkably similar, which is a strong indication of their origin from a very general property of the fluid equations, such as the turbulent fragmentation process.
astro-ph
astro-ph
The Turbulent Shock Origin of Proto -- Stellar Cores Harvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138 Paolo Padoan1, Helsinki University Observatory,Tahtitorninmaki, P.O.Box 14, SF-00014 University of Helsinki, Finland Mika Juvela2, Harvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138 Alyssa A. Goodman3 and Ake Nordlund4 Astronomical Observatory and Theoretical Astrophysics Center, Juliane Maries Vej 30, DK-2100 Copenhagen, Denmark ABSTRACT The fragmentation of molecular clouds (MC) into proto -- stellar cores is a central aspect of the process of star formation. Because of the turbulent nature of super -- sonic motions in MCs, it has been suggested that dense structures such as filaments and clumps are formed by shocks in a turbulent flow. In this work we present strong evidence in favor of the turbulent origin of the fragmentation of MCs. The most generic result of turbulent fragmentation is that dense post shock gas traces a gas component with a smaller velocity dispersion than lower density gas, since shocks correspond to regions of converging flows, where the kinetic energy of the turbulent motion is dissipated. Using synthetic maps of spectra of molecular transitions, computed from the results of numerical simulations of super -- sonic turbulence, we show that the dependence of velocity dispersion on gas density generates an observable relation between the rms velocity centroid and the integrated intensity (column density), σ(V0) − I, which is indeed found in the observational data. The comparison between the theoretical model (maps of synthetic 13CO spectra), with 13CO maps from the Perseus, Rosette and Taurus MC complexes, shows excellent agreement in the σ(V0) − I relation. The σ(V0)−I relation of different observational maps with the same total rms velocity are remarkably similar, which is a strong indication of their origin from a very general property of the fluid equations, such as the turbulent fragmentation process. Subject headings: Rosette, Taurus); radio astronomy: interstellar: lines turbulence -- ISM: kinematics and dynamics -- individual (Perseus, [email protected] [email protected] [email protected] [email protected] 1 1. Introduction The importance of turbulence in the process of star formation was recognized long ago (Von Weizsacker 1951), and was discussed in a seminal paper by Lar- son (1981). Several successive works have tried to use the observational data to relate different properties of MCs with the physics of laboratory and numer- ical turbulence, such as power spectra of kinetic en- ergy 5 , probability distributions of velocity and veloc- ity differences 6, intermittency (Falgarone & Phillips 1990; Falgarone & Puget 1995; Falgarone, Pineau Des Forets, & Roueff 1995), and self -- similarity 7. During the last decade, numerical simulations of transonic turbulence (Passot & Pouquet 1987; Passot, Pouquet & Woodward 1988; L´eorat, Passot & Pou- quet 1990; Lee, Lele & Moin 1991 ; Porter, Pouquet & Woodward 1992 ; Kimura & Tosa 1993; Porter, Woodward & Pouquet 1994; Vazquez -- Semadeni 1994; Passot, Vazquez -- Semadeni & Pouquet 1995) and highly super -- sonic turbulence (Passot, V´azquez -- Semadeni, Pouquet 1995; V´azquez -- Semadeni, Passot & Pouquet 1996; Padoan & Nordlund 1997, 1999; Stone, Os- triker & Gammie 1998; MacLow et al. 1998; Ostriker, Gammie & Stone 1999; Padoan, Zweibel & Nordlund 2000; Klessen, Heitsch & MacLow 2000), on relatively high -- resolution numerical grids, have become avail- able, and very detailed comparisons between observa- tional data with turbulence models of MCs have been performed (Padoan, Jones & Nordlund 1997; Padoan et al. 1998; Padoan & Nordlund 1999; Padoan et al. 1999; Rosolowsky et al. 1999; Padoan, Rosolowsky & Goodman 2000). Recent numerical studies of super -- sonic magneto -- hydrodynamic (MHD) turbulence have brought new understanding of the physics of turbulence. The most important results are: • Super -- sonic turbulence decays in approximately one dynamical time, independent of the mag- 5See Leung, Kutner & Mead 1982; Myers 1983; Quiroga 1983; Sanders, Scoville & Solomon 1985; Goldsmith & Arquilla 1985; Dame et al. 1986; Falgarone & P´erault 1987. 6 See Scalo 1984; Kleiner & Dickman 1985, 1987; Hobson 1992; Miesch & Bally 1994; Miesch & Scalo 1995; Lis et al. 1996; Miesch, Scalo & Bally 1999. 7See Beech 1987; Bazell & D´esert 1988, Scalo 1990; Dickman, Horvath & Margulis 1990; Falgarone, Phillips & Walker 1991; Zimmermann, Stutzki & Winnewisser 1992; Henriksen 1991; Hetem & Lepine 1993; Vogelaar & Wakker 1994; Elmegreen & Falgarone 1996. netic field strength (Padoan & Nordlund 1997, 1999; MacLow et al. 1998; Stone, Ostriker & Gammie 1998; MacLow 1999). • The probability distribution of gas density in isothermal turbulence is well approximated by a Log -- Normal distribution, whose standard de- viation is a function of the rms Mach number of the flow (Vazquez -- Semadeni 1994; Padoan 1995; Padoan, Jones & Nordlund 1997; Scalo et al. 1998; Passot & Vazquez -- Semadeni 1998; Nordlund & Padoan 1999; Ostriker, Gammie & Stone 1999). • Super -- sonic isothermal turbulence generates a complex system of shocks which fragment the gas very efficiently into high density sheets, fil- aments, and cores (this is the general result of any numerical simulation of super -- sonic turbu- lence). • Super -- Alfv´enic turbulence provides a good de- scription of the dynamics of MCs, and an ex- planation for the origin of dense cores with magnetic field strength consistent with Zeeman splitting observations. (Padoan & Nordlund 1997, 1999).8 We call "turbulent fragmentation" the process of generation of high density structures by turbulent shocks. Since random super -- sonic motions are ubiq- uitous in MCs, turbulent fragmentation cannot be avoided: it is a direct consequence of the observa- tional evidence. In some analytical studies it is tac- itly assumed that turbulent fragmentation can be ne- glected if the kinetic energy of random motions is in rough equipartition with the magnetic energy. This assumption is wrong, because motions along magnetic field lines are unavoidable, and so turbulent fragmen- tation occurs via super -- sonic compressions along the magnetic field lines, as discussed by Gammie & Os- triker (1996) and Padoan & Nordlund (1997, 1999). Numerical simulations of turbulence have been used to discuss the turbulent origin of MC structures by Passot & Pouquet (1987). Vazquez -- Semadeni, 8This particular result is supported almost exclusively by our work. A significant fraction of the astrophysical community still favors the traditional idea that a rather strong magnetic field supports MCs against their gravitational collapse. An example of numerical work that favors the traditional idea of magnetic support is Ostriker, Gammie & Stone (1999). 2 Passot & Pouquet (1996) and Ballesteros -- Paredes, Hartmann & Vazquez -- Semadeni (1999) have used two dimensional simulations to argue that MCs are formed by turbulence. Padoan & Nordlund (1997, 1999) have shown that super -- sonic and super -- Alfv´enic turbulence can explain the origin of magnetized cores in MCs, including the observed field strength -- density (B − n) relation (Myers & Goodman 1988; Fiebig & Gusten 1989; Crutcher 1999). The idea that proto -- stellar cores and stars are formed in turbulent shocks has been previously discussed by Elmegreen (1993). In that work, an analysis of the gravitational instabil- ity of the cores can be found. Here, we present new strong observational evidence in favor of the turbulent shock origin of proto -- stellar cores. Such evidence is based on the fact that dense post shock gas traces a gas component with a smaller velocity dispersion than lower density gas, since it maps regions of converging flows, where the kinetic energy of the turbulent mo- tion is dissipated. In §2 and 3, numerical simulations and observa- tional data, used in this work, are briefly described. In §4 we compute the rms flow velocity as a func- tion of the gas density, in simulations of super -- sonic turbulence, and show that it decreases for increasing values of the gas density. In §5 we show that such general property generates an observable relation be- tween the rms velocity centroid and the integrated in- tensity (roughly proportional to the surface density), for the J=1-0 13CO transition, and in §6 the same relation is found in the observational data. Results are discussed in §7, and conclusions are summarized in §8. 2. Numerical Models The numerical models used in this work are based on the results of numerical simulations of super -- Alfv´enic and highly super -- sonic MHD turbulence, run on a 1283 computational mesh, with periodic bound- ary conditions. As in our previous works, the initial density and magnetic fields are uniform; the initial velocity is random, generated in Fourier space with power only on the large scale. We also apply an exter- nal random force, to drive the turbulence at a roughly constant rms Mach number of the flow. This force is generated in Fourier space, with power only on small wave numbers (1 < k < 2), as the initial velocity. The isothermal equation of state is used. Descriptions of the numerical code used to solve the MHD equations 3 can be found in Galsgaard & Nordlund (1996); Nord- lund, Stein & Galsgaard (1996); Nordlund & Gals- gaard (1997); Padoan & Nordlund (1999). In this work we neglect the effect of self -- gravity, although that can be described with a different ver- sion of our code (Padoan et al. (2000)). Here we compare the relative velocity of regions of MC com- plexes as a function of their gas density or column density. Such regions are distributed across the full extension of the MC complexes, that is several pc. In our numerical models, driven continuously by an ex- ternal force on the large scale, self -- gravity is respon- sible for the collapse of gravitationally bound cores formed by the turbulent flow, but does not affect sig- nificantly the large scale flow. Since we assume that large scale motions in MC complexes are due to tur- bulence, and not to a gravitational collapse, we can neglect self -- gravity. Similarly, we have neglected the effect of ambipolar drift, since it is not relevant for motions on the scale of several pc, although that is computed in a different version of our code, using the strong coupling approximation (see Padoan, Zweibel & Nordlund 2000). In order to scale the models to physical units, we use the following empirical Larson type relations, as in our previous works: M = 2.0(cid:18) L 1pc(cid:19)0.5 , (1) where M is the rms sonic Mach number of the flow, and a temperature T = 10 K is assumed, and hni = 2.0 × 103(cid:18) L 1pc(cid:19)−1 , (2) where the gas density n is expressed in cm−3. The rms sonic Mach number is an input parameter of the numerical simulations, and can be used to re-scale them to physical units. When comparing theoretical models with observations, M is in fact the only pa- rameter that we need to match (or its two dimensional equivalent on the maps, σv -- see §3). In the absence of self -- gravity and magnetic field, statistical properties of turbulent flows (very large Reynolds number) with the same value of M should be universal, and inde- pendent of the average density (for isothermal flows without self -- gravity). If the magnetic field is present, the rms Alfv´enic Mach number of the flow is also an input parameter of the numerical simulations (it de- termines the magnetic field strength). The physical unit of velocity in the code is the isothermal speed of sound, Cs, and the physical unit of the magnetic field is Cs(4πhρi) 2 (cgs). 1 In this work we use two numerical models with rms velocity 3.4 and 1.7 km/s, which corresponds to M ≈ 13.0 and 6.5 respectively. Using the Larson type relations (1) and (2) we get L ≈ 42 and 10.5 pc and hni ≈ 48 and 190 cm−3. The one dimensional rms velocity for the two models is σv = 2.0 and 1.0 km/s, which are also recovered from the analysis of the syn- thetic spectral maps computed with these models. The values of σv have been chosen for the appropriate comparison with the observational data presented in § 3.9 The magnetic field strength is B ≈ 5 µG in both models. Maps of synthetic spectra of molecular transitions are computed, using a non -- LTE Monte Carlo radia- tive transfer code (Juvela 1997, 1998), from the three dimensional density and velocity fields generated in the numerical MHD experiments. The method of computing synthetic spectra was presented in Padoan et al. (1998). For the purpose of this work we use only one molecular transition, namely J=1-0 13CO . Uniform temperature, T = 10 K is assumed for these radiative transfer calculations, in agreement with the isothermal equation of state used in the MHD calcu- lations. We are presently computing thermal equilib- rium models of MCs that we will use in a future work to study the effect of realistic temperature variations in molecular spectra. Since the spectral noise resulting from uncertain- ties in our radiative transfer calculations is always much smaller than the typical observational noise, the comparison of synthetic spectra with observational data can be done only after adding noise to the syn- thetic spectra, and the effect of noise on the statis- tical properties of the spectra need to be quantified (see §4). 3. Observational Data We choose to use the J=1-0 13CO transition be- cause it samples the range of values of column den- sity we are here interested in, and also because several 9 The Taurus MC complex has σv ≈ 1.0 km/s, and L ≈ 40 pc, while the Larson type relation (1) would give L ≈ 12.6 pc. For the purpose of this work we are interested in comparing the observations with numerical models with similar rms Mach number, and we do not try to match the physical extension of each MC complex. 4 large maps of molecular clouds are available in this transition. We compare maps of J=1-0 13CO syn- thetic spectra with some of the largest observational J=1-0 13CO spectral maps in the literature from the following Galactic regions: the Perseus MC complex (Billawala, Bally & Sutherland 1997), the Taurus MC complex (Mizuno et al. 1995), the Rosette MC com- plex (Blitz & Stark 1986; Heyer et al. 2000). These MC complexes have an extension of approximately 30 -- 50 pc, and radial velocity dispersions in the range 1 -- 2.4 km/s. We define the total rms velocity of a map as the rms velocity weighted with the total spectrum of the map, T (v): where σv =sPv (v − ¯v)2 T (v) dv Pv T (v) dv ¯v = Pv v T (v) dv Pv T (v) dv . , (3) (4) The Blitz & Stark map of the Rosette MC complex has σv = 2.4 km/s, but, limited to the region that matches the more recent Heyer et al. map, the value is σv = 2.0 km/s (for both maps). The full map of the Perseus MC complex also yields a value of σv = 2.0 km/s. Taurus is instead much less turbulent, despite its large spatial extent, with σv = 1.0 km/s. The angular resolution is inversely proportional to the diameter of the antenna: 4 m for the Taurus map, 7 m for the Perseus and the Blitz & Stark Rosette maps, and 14 m for the Heyer et al. map of Rosette. Assuming a distance of 140 pc for Taurus, 300 pc for Perseus, and 1600 pc for Rosette, the spatial reso- lution of the maps is 0.1 pc for Taurus, 0.15 pc for Perseus, 0.84 pc for the Blitz & Stark map of Rosette, and 0.42 pc for the Heyer et al. map of Rosette. The spectral resolution is 0.1 km/s for Taurus, 0.273 km/s for Perseus, 0.68 km/s for the Blitz & Stark map of Rosette, and 0.06 km/s for the Heyer et al. map of the same cloud. Rms noise N and average spectrum quality hQi (see Padoan, Rosolowsky & Goodman 2000) also vary from map to map. The spectrum quality, Q, is related to the signal -- to -- noise, S/N . It is defined as the ratio of the rms signal (over the whole spectrum or inside a velocity window) to the rms noise, N : Q = pPv T (v)2dv N (5) The usual definition of S/N is based on Gaussian fits of the spectra, which we prefer to avoid because the J=1-0 13CO transition typically yields spectra with significant non -- Gaussian shape and multiple compo- nents. The spectrum quality is a sort of signal -- to -- noise weighted over the whole spectrum. The relation between Q and S/N is discussed in Padoan, Good- man & Rosolowsky (2000). We call average spectrum quality hQi the value of Q averaged over the whole map. Values of hQi, N , resolutions and σv are listed in Table 1 for all the observational maps. In the fol- lowing, when different maps are compared with each other, we make noise and velocity resolution equal in the different maps, by adding noise and reducing the velocity resolution where necessary. 4. Velocity Dispersion Versus Gas Density in Super -- Sonic Turbulence Turbulent fragmentation generates a complex sys- tem of dense post shock sheets, filaments and cores, reminiscent of structures observed in molecular cloud (MC) maps. An example of a two dimensional pro- jection of the three dimensional density field from a simulation of super -- sonic turbulence is shown in Fig- ure 1. In previous works (Padoan, Jones & Nord- lund 1997; Padoan & Nordlund 1997, 1999), we have shown that, besides this morphological similarity, the density field of numerical super -- sonic turbulence has important statistical properties in agreement with the density field of observed MCs. Figure 2 (left panel) is a two dimensional section (no projection) of the same three-dimensional density field used in Figure 1. A complex system of filaments is apparent, with a number of high density cores in- side the filaments. In three dimensional super -- sonic turbulence, filaments are formed by two dimensional compressions (at the intersection of sheets) and the densest cores are formed by three dimensional com- pressions. Most of the "filaments" in two dimensional sections like Figure 2, are instead two dimensional cuts through sheets, and most of the cores are local density maxima, due to fluctuations in the shock ve- locity (they usually corresponds to strongly curved segments of filaments). Such density maxima are of- ten unstable to gravitational collapse, and are the ori- gin of proto -- stellar cores. Since the dense gas originates in shocks, that is in regions of converging flows, it should move with sig- nificantly lower velocity than the lower density tur- bulent flow. This is illustrated in the right panel of Figure 2, which shows the magnitude of the flow ve- locity on the same plane as the left panel. Dark blue is low velocity, and dark red high velocity. It is clear that high density filaments trace regions of low veloc- ity, at the intersections of high velocity "blobs". This general property of super -- sonic turbulence is quanti- fied by the dependence of the rms flow velocity, σ(v), on the gas density: σ(v) = h(v − ¯v)2ρi, (6) We repeat for different values of ρ, and than plot the result versus the gas density. Expression (6) means that the rms flow velocity is obtained as an average over the whole computational box, using only posi- tions where the gas density has a value contained in an interval centered around ρ. The average is repeated for different intervals of values of ρ, to span the whole range of densities. The plot is shown in Figure 3, at time t/tdyn = 0.07 (square symbols) and t/tdyn = 1.0 (asterisks), where t/tdyn is the time in units of the dynamical time (de- fined as the size of the computational box divided by the rms flow velocity), and the density is uni- form in the initial conditions, at t/tdyn = 0.0. At t/tdyn = 1.0, regions of ρ ≈ hρi have σ(v) comparable to the total rms velocity, σv = 3.0 km/s, while re- gions with 10 times higher density have much smaller rms velocity, σ(v) ≈ 2.3 km/sec, and at ρ = 100 hρi σ(v) ≈ 0.9 km/s. The rms velocity conditioned to the gas density decreases sharply with increasing gas density already at a very early time, t/tdyn = 0.07, when the density field is still very smooth. 5. Rms Velocity Centroid versus Integrated Intensity Observational spectral maps of MCs, do not pro- vide a direct estimate of the three dimensional ve- locity field and gas density. Only radial velocity and velocity -- integrated intensity (which is roughly pro- portional to the surface density), averaged along each line of sight, are directly available from the data. However, the dependence of the rms velocity cen- troid on the integrated intensity should resemble the σ(v) − ρ relation, since lines of sight of high intensity are usually dominated by one or more dense cores. The velocity centroid, V0, is the average velocity along an individual line of sight x on the map: V0(x) = Pv v T (v, x)dv Pv T (v, x)dv , (7) 5 where T (v, x) is the signal (antenna temperature), at the velocity v and position x, and dv is the width of the velocity channel. The integrated intensity, I(x), is: I(x) =Xv T (v, x)dv. (8) We compute the rms velocity centroid, σ(V0), con- ditioned on I: σ(V0) = h(V0 − ¯V0)2Iix, (9) (spatial average) and plot it against I. Expression (9) means that the rms velocity centroid is obtained as an average over the whole map, selecting all lines -- of -- sight on the map with values of I inside a given interval, [I, I + dI]. This rms velocity centroid is not equivalent to a local line width, because it is ob- tained as an average over the whole map. The plot is shown in Figure 4, for two maps of J=1-0 13CO synthetic spectra with different total line widths σv (defined in §3). Figure 4 shows that σ(V0) decreases significantly for increasing values of I. The general property of super -- sonic turbulence, namely high den- sity gas moves relatively slowly, is therefore apparent also in the observable relation σ(V0) − I. The rela- tion σ(V0) − I is affected by noise and by the width of the velocity channels, which must be taken into account when comparing different maps. Noise has been added to the synthetic spectra used for com- puting the plot in Figure 4, to yield a value of the spectrum quality comparable to a typical value found in the observational data used in this work, hQi = 3.5 (see §3 for the definition of hQi). In Figure 5 we show the effect of noise (left panel) and spectral resolution (right panel). The effect of increasing the noise (decreasing the value of hQi) is that of making the σ(V0)− I relation steeper, because the uncertainty in the determination of the velocity centroids due to noise contributes to the dispersion of velocity centroid values, and the effect increases at de- creasing values of integrated intensity I. Lower spec- tral resolution further increases the same effect. How- ever, we have verified that if the noise is low enough, hQi > 20, both noise and velocity resolution have no effect on the σ(V0) − I relation, and the squared sym- bols in the left panel of Figure 5 correspond therefore to the intrinsic relations. We can conclude that this observable relation truly originates from the three di- mensional σ(v) − ρ relation, with only a partial con- tribution from noise. 6 We have also verified that spatial resolution does not affect our results. The spatial resolution can be decreased significantly, by rebinning the map to a smaller number of spectra, without any appreciable variation in the σ(V0) − I relation. As the spatial res- olution is decreased, however, the statistical sample (number of spectra) decreases, and statistical fluctua- tions (deviations around the high resolution σ(V0) − I relation) become progressively more important. 6. Observational σ(V0) − I Relation We have shown in the previous section that the σ(V0) − I relation is sensitive to the value of the rms noise and to the spectral resolution. In the following plots, where we compare different spectral maps from observations and models, we have therefore added noise to the spectra and increased the velocity chan- nel width dv to match the map with the largest noise and dv. We have not modified the spatial resolution in any map, since that has no systematic effect on the σ(V0) − I relation, as commented above. The left panel of Figure 6 shows the σ(V0) − I re- lation for the maps of MC complexes introduced in §3. For the Rosette MC complex we have used the portion of the full Blitz & Stark map that matches the Heyer et al. map. The two models used for the comparison have σv = 2.0 km/s, similar to Rosette and Perseus, and σv = 1.0 km/s, similar to Taurus. It is remarkable that the Rosette and the Perseus MC complexes have indistinguishable σ(V0) − I relations, which are also coincident with the theoretical predic- tion (square symbols), for the same value of σv. The result for Taurus is also in good agreement with the σv = 1.0 km/s model. Horizontal shifts could be ex- pected in the plot, since different MC complexes can have different surface density. However, MCs and MC complexes are known to approximately follow the Lar- son relation between density and size (equation (2)), which implies roughly constant surface density, and small horizontal shifts in the plot. To compute the plots in the left panel of Figure 6, all maps have been treated to make their rms noise and velocity resolution equal. This is achieved by re- binning spectral profiles into a smaller number of ve- locity channels, and by adding noise, when necessary. In order to check that the σ(V0) − I relations from different maps treated in this way are really compa- rable, we have computed the σ(V0)−I relation for the Rosette MC complex using both the Heyer et al. map, Maps of synthetic spectra contain a lot of infor- mation about the kinematic, the structure and the thermal properties of molecular clouds, and can be analyzed with different statistical tools. In this work we have presented a new statistical method to an- alyze spectral -- line maps, which is very useful be- cause it probes directly a very general property of super -- sonic turbulence, that is the fact that dense gas traces a gas component with a smaller velocity disper- sion than lower density gas. This property arises be- cause the gas density is enhanced in regions where the large scale turbulent flow converges (compressions) and the kinetic energy of the turbulent flow is dis- sipated by shocks. If local compressions are instead due to local instabilities (e.g. gravitational instabil- ity, or gravitational instability mediated by ambipo- lar diffusion), and the large scale motions are only the consequence of local instabilities (as it should be in a self -- consistent picture), gas density increases with the flow velocity dispersion (see for example the model by Indebetouw & Zweibel 2000), contrary to the obser- vational evidence presented in this work. The main conclusion of this work is that every model for the origin of molecular cloud structure and proto -- stellar cores should be tested against the σ(V0) − I relation. Turbulent fragmentation provides a realistic scenario for the origin of proto -- stellar cores, which satisfies this new observational constraint. We are grateful to Edith Falgarone and Phil My- ers for discussions that stimulated this work, and to the referee for a number of useful comments. This work was supported by NSF grant AST-9721455. Ake Nordlund acknowledges partial support by the Danish National Research Foundation through its establish- ment of the Theoretical Astrophysics Center. and the portion of the Blitz & Stark map that matches the region covered by the Heyer et al. map. The re- sult is plotted in the right panel of Figure 6. The velocity resolution of the Heyer et al. map has been decreased from dv = 0.06 km/s to dv = 0.68 km/s, and noise has been added, to match exactly the veloc- ity resolution and rms noise in the Blitz & Stark map. As can be seen in the right panel of Figure 6, after this drastic treatment of the higher resolution map, the σ(V0) − I relations for the two maps are prac- tically indistinguishable from each other, which sup- port the validity of the comparison of different maps (left panel of Figure 6. In the right panel, the case of the full Blitz & Stark map is also plotted (square symbols). The full map has a higher total rms ve- locity, σv = 1.4 km/s, than the portion that matches the Heyer et al. map, and its σ(V0) − I relation is therefore steeper, as expected. 7. Discussion and Conclusions The origin of proto -- stellar cores is a fundamen- tal problem in our understanding of the process of star formation. Models that have been proposed to describe proto-stellar cores are based on i) static or quasi -- static equilibrium (e.g. Curry & Mckee 2000; Jason & Pudritz 2000), ii) thermal instability (e.g. Yoshii & Sabano 1980; Gilden 1984; Graziani & Black 1987), iii) gravitational instability through ambipolar diffusion (e.g. Basu & Mouschovias 1994; Nakamura, Hanawa, & Nakano 1995; Indebetouw & Zweibel 2000; Ciolek & Basu 2000), iv) non -- linear Alfv´en waves (e.g. Carlberg & Pudritz 1990; Elmegreen 1990, 1997, 1999), v) clump collisions (e.g. Gilden 1984; Kimura & Tosa 1996), vi) super -- sonic turbu- lence (e.g Elmegreen 1993; Klessen, Burkert and Bate 1998; Klessen, Heitsch, & Mac Low 2000; Padoan et al. 2000). Many detailed comparisons between observational data and models, which support the idea of the turbu- lent origin of the structure and kinematics of molec- ular clouds, have been presented in our previous pa- pers (Padoan, Jones & Nordlund 1997; Padoan et al. 1998; Padoan et al. 1999; Padoan & Nordlund 1999; Padoan, Goodman & Rosolowsky 2000). Models of numerical turbulence can be compared with the ob- servations by computing i) synthetic stellar extinction measurements, ii) synthetic spectral maps of molec- ular transitions, iii) synthetic Zeeman splitting mea- surements, iv) synthetic polarization maps. 7 REFERENCES Leung, C. M., Kutner, M. L., Mead, K. N. 1982, ApJ, Ballesteros-Paredes, J., Hartmann, L., V´azquez- Semadeni, E. 1999, ApJ, 527, 285 Bazell, D., D´esert, F. X. 1988, ApJ, 333, 353 Beech, M. 1987, Ap&SS, 133, 193 Crutcher, R. M. 1999, ApJ, 520, 706 Dame, T. M., Elmegreen, B. G., Cohen, R. S., Thad- deus, P. 1986, ApJ, 305, 892 Dickman, R. L., M., M., Horvath, M. A. 1990, ApJ, 365, 586 Elmegreen, B. G. 1993, ApJ, 419, 29 Elmegreen, B. G., Falgarone, E. 1996, ApJ, 471, 816 Falgarone, E., P´erault, M. 1987, in G. Morfil, M. Scholer (eds.), Physical Processes in Interstellar Clouds, Reidel, Dordrecht, 59 262, 583 Mac Low, M. 1999, ApJ, 524, 169 Mac Low, M., Smith, M. D., Klessen, R. S., Burkert, A. 1998, ApJS, 261, 195 Mac Low, M.-M. 1998, in J. Franco, A. Carraminana (eds.), Interstellar Turbulence, Cambridge Univer- sity Press Myers, P. C. 1983, ApJ, 270, 105 Myers, P. C., Goodman, A. A. 1988, ApJ, 326, L27 Nordlund, A., Galsgaard, K. 1997, A 3D MHD Code for Parallel Computers, technical report, Astro- nomical Observatory, Copenhagen University Nordlund, A., Stein, R. F., Galsgaard, K. 1996, in Lecture Notes in Computer Science 1041, ed. J. Wazniewsky, (Heidelberg: Springer), 450 Falgarone, E., Phillips, T. G. 1990, ApJ, 359, 344 Ostriker, E. C., Gammie, C. F., Stone, J. M. 1999, Falgarone, E., Phillips, T. G., Walker, C. 1991, ApJ, ApJ, 513, 259 378, 186 Falgarone, E., Pineau des Forets, G., Roueff, E. 1995, A&A, 300, 870 Padoan, P., Bally, J., Billawala, Y., Juvela, M., Nord- lund, A. 1999, ApJ, 525, 318 Padoan, P., Jones, B., Nordlund, A. 1997, ApJ, 474, Falgarone, E., Puget, J. 1995, A&A, 293, 840 730 Fiebig, D., Gusten, R. 1989, A&A, 214, 333 Galsgaard, K., Nordlund, A. 1996, J. Geophys. Res., 101(A6), 13445 Padoan, P., Juvela, M., Bally, J., Nordlund, A. 1998, ApJ, 504, 300 Padoan, P., Nordlund, A. 1997, astro -- ph/9706176 Gammie, C. F., Ostriker, E. C. 1996, ApJ, 466, 814 Padoan, P., Nordlund, A. 1999, ApJ, 526, 279 Goldsmith, P. F., Arquilla, R. 1985, ApJ, 297, 436 Padoan, P., Nordlund, A., Rognvaldsson, O., Good- Henriksen, R. N. 1991, ApJ, 377, 500 Hetem, A., J., Lepine, J. R. D. 1993, A&A, 270, 451 Indebetouw, R. ., Zweibel, E. G. 2000, ApJ, 532, 361 man, A. A. 2000, ApJ (submitted) Padoan, P., Rosolowsky, E, W., Goodman, A. A. 2000, ApJ (in press) Padoan, P., Zweibel, E., Nordlund, A. 2000, ApJ, 540, Kimura, T., Tosa, M. 1993, ApJ, 406, 512 332 Klessen, R. S., Heitsch, F., MacLow, M. M. 1999, Passot, T., Pouquet, A. 1987, J. Fluid Mech., 181, astro -- ph/9911068 441 Larson, R. B. 1981, MNRAS, 194, 809 Passot, T., Pouquet, A., Woodward, P. 1988, 197, 228 Lee, S., Lele, S. K., Moin, P. 1991, Phys. Fluids A, 3, Passot, T., V´azquez-Semadeni, E., A., P. 1995, ApJ, 657 455, 536 8 Porter, D. H., Pouquet, A., Woodward, P. R. 1994, Phys. Fluids, 6, 2133 Quiroga, R. J. 1983, Ap. Sp. Sci., 93, 37 Rosolowsky, E, W., Goodman, A. A., Wilner, D. J., Williams, J. P. 1999, ApJ, 524, 887 Sanders, D. B., Scoville, N. Z., Solomon, P. M. 1985, ApJ, 289, 372 Scalo, J. M. 1990, in R. Capuzzo-Dolcetta, C. Chiosi, A. D. Fazio (eds.), Physical Processes in Fragmen- tation and Star Formation, (Kluwer : Dordrecht, p.151 Scalo, J. M., V´azquez-Semadeni, E., Chappell, D., Passot, T. 1998, ApJ, 504, 835 Stone, J. M., Ostriker, E. C., Gammie, C. F. 1998, ApJ, 508, L99 V´azquez-Semadeni, E. 1994, ApJ, 423, 681 V´azquez-Semadeni, E., Passot, T., Pouquet, A. 1996, ApJ, 473, 881 Vogelaar, M. G. R., Wakker, B. P. 1994, A&A, 291, 557 von Weizsacker, C. F. 1951, ApJ, 114, 165 Zimmermann, T., Stutzki, J., Winnewisser, G. 1992, in Evolution of Interstellar Matter and Dynamics of Galaxies, 254 This 2-column preprint was prepared with the AAS LATEX macros v4.0. 9 TABLE AND FIGURE CAPTIONS: Table 1: Cloud name; maximum spatial extension; total rms velocity; telescope beam size; velocity channel width; rms noise; average spectrum quality (signal -- to -- noise); bibliographic reference. Figure 1: Two dimensional projection of the three dimensional density field from a simulation of isothermal super -- sonic turbulence with rms Mach number ∼ 10. Figure 2: Left panel: Two dimensional section (no projection) of the same density field used for Figure 1. Most filaments are sections of sheets, and most dense cores are density maxima inside curved segments of filaments, formed by fluctuations in the shock velocity. Right panel: Modulus of the three dimensional flow velocity on the same two dimensional plane as in the left panel. Dark blue is low velocity, and dark red high velocity. The dense filaments on the left panel are commonly found in regions of small flow velocity, at the intersections of patches of high velocity. Figure 3: Rms flow velocity, conditioned to gas density, versus the gas density, computed from a simulation of super -- sonic turbulence with rms Mach number ∼ 10. Squared symbols are for an early time, just 7% of the dynamical time after an initial condition with uniform density; asterisks are for a time equal to a dynamical time. Density values are binned over 20 logarithmic intervals between the average and the highest density. Within each interval, the density grows by a factor of 1.04 at the early time, and 1.27 at one dynamical time. Figure 4: Rms velocity centroid, conditioned to integrated intensity, versus the integrated intensity (see text for details). Two maps of synthetic spectra from MHD turbulence simulations are used, with different values of the total rms radial velocity (intensity weighted), σv = 1.0 km/s (diamonds), and σv = 2.0 km/s (triangles). The model with larger rms velocity has also a slightly larger maximum intensity (surface density), although both models are rescaled into physical units using the Larson relations (see §2). Integrated intensity values are binned over 12 linear intervals between the lowest and the highest values in the map. Within each interval, the density increment is ≈ 1 K km/s (for the σv = 1.0 km/s model) and ≈ 2 K km/s (for the σv = 2.0 km/s model). Figure 5: Effect of noise and spectral resolution. Left panel: Conditioned rms velocity centroid versus intensity. Triangle and diamond symbols are the same as in Figure 4, while squared symbols have higher spectrum quality hQi (signal -- to -- noise). Different values of hQi are obtained by adding different levels of noise to the synthetic spectra. hQi = 3.5 is typical of the observational maps used in this work. The plot is not sensitive to the decreasing noise, for hQi > 20. Right panel: Same as left panel, but the squared symbols are for lower spectral resolution (the velocity channels of the synthetic spectra is increased from dv = 0.2 km/s to dv = 0.6 km/s). Clearly, both the level of noise and the spectral resolution must be taken into account when comparing different maps. Integrated in- tensity values are binned over 12 linear intervals between the lowest and the highest values in the map, as in Figure 4. Figure 6: Observational σ(V0) − I relation. Left panel: Conditioned rms velocity centroid versus intensity for different MC complexes: Rosette, Perseus, and Taurus. Square symbols are for maps of synthetic spectra with total rms velocity σv = 1.0 km/s and 2.0 km/s. Right panel: Comparison of the two maps of the Rosette MC complex (see text for details). Integrated intensity values are binned over 10 linear intervals between the lowest and the highest values in the map. 10 L [pc] σv[km/s] dx [pc] 38 27 52 36 36 1.0 2.0 2.4 2.2 2.0 0.10 0.15 0.84 0.84 0.42 dv [km/s] N [K] 0.24 0.24 0.20 0.19 0.12 0.10 0.27 0.68 0.68 0.06 reference hQi 2.3 Mizuno et al. (1995) 2.8 Billawala et al. (1997) 2.1 Blitz & Stark (1986) 4.0 Blitz & Stark (matching region) 3.8 Heyer et al. (2000) MC Taurus Perseus Rosette Rosette Rosette Table 1: 11 Fig. 1. -- 12 Fig. 2. -- 13 Fig. 3. -- 14 Fig. 4. -- 15 Fig. 5. -- 16 Fig. 6. -- 17
astro-ph/0310166
1
0310
2003-10-06T22:10:23
Astrochemistry of Sub-Millimeter Sources in Orion: Studying the Variations of Molecular Tracers with Changing Physical Conditions
[ "astro-ph" ]
Cornerstone molecules (CO, H_2CO, CH_3OH, HCN, HNC, CN, CS, SO) were observed toward seven sub-millimeter bright sources in the Orion molecular cloud in order to quantify the range of conditions for which individual molecular line tracers provide physical and chemical information. Five of the sources observed were protostellar, ranging in energetics from 1 - 500L_sun, while the other two sources were located at a shock front and within a photodissociation region (PDR). Statistical equilibrium calculations were used to deduce from the measured line strengths the physical conditions within each source and the abundance of each molecule. In all cases except the shock and the PDR, the abundance of CO with respect to H_2 appears significantly below (factor of ten) the general molecular cloud value of 10^-4. {Formaldehyde measurements were used to estimate a mean temperature and density for the gas in each source. Evidence was found for trends between the derived abundance of CO, H_2CO, CH_3OH, and CS and the energetics of the source, with hotter sources having higher abundances.} Determining whether this is due to a linear progression of abundance with temperature or sharp jumps at particular temperatures will require more detailed modeling. The observed methanol transitions require high temperatures (T>50 K), and thus energetic sources, within all but one of the observed protostellar sources. The same conclusion is obtained from observations of the CS 7-6 transition. Analysis of the HCN and HNC 4-3 transitions provides further support for high densities n> 10^7 cm^-3 in all the protostellar sources.
astro-ph
astro-ph
Astronomy&Astrophysicsmanuscript no. 3536ms (DOI: will be inserted by hand later) November 20, 2018 Astrochemistry of Sub-Millimeter Sources in Orion Studying the Variations of Molecular Tracers with Changing Physical Conditions Doug Johnstone1,2, Annemieke M. S. Boonman3, and Ewine F. van Dishoeck3 1 National Research Council Canada, Herzberg Institute of Astrophysics, 5071 West Saanich Road, Victoria, B.C., V9E 2E7, Canada e-mail: [email protected] 2 Department of Physics & Astronomy, University of Victoria, Victoria, BC, V8P 1A1, Canada 3 Leiden Observatory, P.O. Box 9513, 2300 RA Leiden, The Netherlands Abstract. Cornerstone molecules (CO, H2CO, CH3OH, HCN, HNC, CN, CS, SO) were observed toward seven sub-millimeter bright sources in the Orion molecular cloud in order to quantify the range of conditions for which individual molecular line tracers provide physical and chemical information. Five of the sources observed were protostellar, ranging in energetics from 1 − 500 L⊙, while the other two sources were located at a shock front and within a photodissociation region (PDR). Statistical equilibrium calculations were used to deduce from the measured line strengths the physical conditions within each source and the abundance of each molecule. In all cases except the shock and the PDR, the abundance of CO with respect to H2 appears significantly below (factor of ten) the general molecular cloud value of 10−4. Formaldehyde measurements were used to estimate a mean temperature and density for the gas in each source. Evidence was found for trends between the derived abundance of CO, H2CO, CH3OH, and CS and the energetics of the source, with hotter sources having higher abundances. Determining whether this is due to a linear progression of abundance with temperature or sharp jumps at particular temperatures will require more detailed modeling. The observed methanol transitions require high temperatures (T > 50 K), and thus energetic sources, within all but one of the observed protostellar sources. The same conclusion is obtained from observations of the CS 7-6 transition. Analysis of the HCN and HNC 4-3 transitions provides further support for high densities n > 107 cm−3 in all the protostellar sources. The shape of the CO 3 -- 2 line profile provides evidence for internal energetic events (outflows) in all but one of the protostellar sources, and shows an extreme kinematic signature in the shock region. In general, the CO line and its isotopes do not signif- icantly contaminate the 850 µm broadband flux (less than 10%); however, in the shock region the CO lines alone account for more than two thirds of the measured sub-millimeter flux. In the energetic sources, the combined flux from all other measured molecular lines provides up to an additional few percent of line contamination. Key words. astrochemistry -- ISM:molecules(CO,H2CO,CH3OH,HCN) -- ISM:individual(Orion A:OMC1,OMC2,OMC3 -- Orion B:NGC2071) -- stars:formation 3 0 0 2 t c O 6 1 v 6 6 1 0 1 3 0 / h p - o r t s a : v i X r a 1. INTRODUCTION The study of structure within the star-forming regions of molecular clouds has benefitted significantly from observations of many molecular lines, each tracing specific chemical and physical conditions. Most studies either focus on surveys using particular tracers, such as carbon monoxide, N2H+, or ammo- nia, in order to map the column density and kinematics of the gas (Myers 1999), or have focused on individual sources, pro- ducing detailed molecular catalogues in several to dozens of species designed to constrain the physical and chemical mor- phology of the region (van Dishoeck & Blake 1998). Send offprint requests to: D. Johnstone Several molecules have emerged as particularly good di- agnostics of the conditions and chemistry near young stellar objects. For example, H2CO has many lines that are readily observed at sub-millimeter wavelengths, and whose ratios are either good temperature or density tracers (e.g., Hurt, Barsony, & Wooten 1996, Mangum, Wooten, & Barsony 1999, Mitchell et al. 2001). Analysis of dust continuum and line emission has shown that temperature and density gradients exist across the protostellar envelopes, with temperatures varying from the in- ner to the outer region from > 100 K to less than 20 K and den- sities from > 107 cm−3 to ∼ 104 cm−3 (e.g., van der Tak et al. 2000a, Shirley, Evans, & Rawlings 2002, Jørgensen, Schoier, & van Dishoeck 2002). The chemistry responds to these changes. Molecules freeze-out onto the grains in the cold outer parts of 2 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion the clouds, where they can form new species through grain- surface reactions. The composition of icy mantles can be deter- mined through infrared absorption, with species such as H2O, CO, CO2 and CH3OH known to have high ice abundances (e.g., Gibb et al. 2000). In the inner region close to the protostar, the grains are heated and the ices are observed to evaporate back into the gas (e.g., van Dishoeck & Helmich 1996; Boonman & van Dishoeck 2003). High temperature gas-phase reactions be- tween evaporated species can subsequently lead to high abun- dances of complex organic molecules observed in high-mass hot cores (e.g., Rodgers & Charnley 2001). Depending on the evolutionary state of the source, different chemical characteris- tics become more prominent. In this paper we consider a selection of cornerstone molecules and study a variety of independent locations within the Integral Shaped Filament (ISF) in Orion A, along with NGC 2071IR in Orion B, in order to quantify the range of condi- tions for which individual molecular line tracers provide phys- ical or chemical information. This is primarily a morphologi- cal study, comparing the differences in molecular line emission across sources of varying physical conditions using simplified equilibrium modeling, in order to search for obvious diagnos- tic features within the data. The list of sources observed in Orion A was compiled from a sensitive dust continuum survey of the ISF (Johnstone & Bally 1999) and includes both highly enshrouded sub-millimeter bright sources, possibly protostel- lar, through more evolved protostars, a bright PDR knot, and a shock front. Many of the young stellar objects are of inter- mediate mass and thus the observations bracket sub-millimeter studies of high-mass (e.g. Hatchell et al. 1998; van der Tak et al. 2000a) and low-mass (e.g. Blake et al. 1994, 1995; van Dishoeck et al. 1995; Buckle & Fuller 2002; Jørgensen et al. 2002) star-forming regions. The selection of cornerstone molecules and molecular transitions was chosen based on pre- vious experience in detailed studies of individual objects. 2. The Orion Integral Shaped Filament The ISF (Bally et al. 1987) in the northern portion of the Orion A molecular cloud is the nearest site with active high-mass through low-mass star formation. The region is associated with the Orion Nebula and the Trapezium Cluster of 700 young stars (Hillenbrand 1997; Hillenbrand & Hartmann 1998), con- tains the OMC1 cloud core immediately behind the Nebula, and two other extensively studied star-forming cores, OMC2 and OMC3 located about 15′ and 25′ to the north (Castets & Langer 1994). To the south the region becomes much qui- eter with OMC4 (Johnstone & Bally 1999) showing no bright sub-millimeter sources. The ISF has spawned several thousand young stars in the past few million years and contains dozens (possibly hundreds) of embedded young stellar objects (Chini et al. 1997, Johnstone & Bally 1999) which power dozens of molecular outflows, Herbig-Haro objects (Reipurth, Bally, & Devine 1997), and molecular hydrogen emitting shocks and jets (Yu, Bally, & Devine 1997). The abundance of star for- mation, at various stages of evolution, within a single, nearby, molecular cloud provides a unique setting for conducting tests of astrochemical diagnostics. 2.1. TheIndividualSources The sources chosen for this study were primarily drawn from the 450 and 850 µm dust emission maps of the ISF (Johnstone & Bally 1999; Fig. 1). Supplemental information on the prop- erties of the sources comes from measurements at 350 µm (Lis et al. 1998), 1300 µm (Chini et al. 1997), 3.6 cm VLA data (Reipurth, Rodriguez, & Chini 1999), and the IRAS point source catalogue. Additional 2 µm molecular hydrogen obser- vations (Yu et al. 1997) of the ISF provide evidence of local heating expected within shocks, and delineate the location of molecular jets. The seven sources were chosen to cover a wide selection of physical conditions in order to test the hypothesis that molecular line observations provide sensitive diagnostics. The coldest sub-millimeter source in our sample is MMS6, Td ∼ 15 K as measured by the sub-millimeter fluxes from 350 to 1300 µm (Lis et al. 2001). MMS6 is also the bright- est 850 µm source within the OMC3 region of the ISF, with a peak flux of 7.4 Jy within a 15 arcsecond beam (Johnstone & Bally 1999). The total mass within the envelope is estimated to be 36 M⊙ (Chini et al. 1997). There is no corresponding IRAS point source, although the background brightness of the region makes point source detection difficult, suggesting that there is no strong internal heating source. The sub-millimeter luminosity, energy emitted at wavelengths greater than 350 µm, is 1.2 L⊙; however, the lack of far-infrared emission measure- ments provides only an upper limit on the total bolometric lu- minosity of < 60 L⊙ (Chini et al. 1997). Despite the very cold appearance of MMS6 in the sub-millimeter, and thus the possi- bility that it is a pre-stellar clump heated only from the outside by the strong interstellar radiation field (ISRF) in Orion, MMS6 has been classified as a Class 0 protostar (Chini et al. 1997). Yu et al. (1997) observed knots of H2 directly north and south of MMS6 which are interpreted as bow shocks within a molecular outflow from the continuum source. Chini et al. (1997) report substantial high-velocity red-shifted emission in 12CO centered on MMS6, although no corresponding blue-shifted lobe was observed. More recent observations by Aso et al. (2000) as- sociate the CO outflow with a different sub-millimeter source. VLA observations at 3.6 cm (Reipurth et al. 1999) locate a ra- dio source coincident with MMS6. Such coincidences are com- mon among Class 0 sources and have been interpreted as ther- mal Bremsstrahlung radiation produced in shocks within the inner part of the molecular outflow. Thus, while MMS6 is pos- sibly a pre-collapse clump, it is likely that the sub-millimeter source enshrouds a very young Class 0 protostar. Despite being three times fainter than MMS6 at 850 µm, MMS9 appears to be the source of the most powerful out- flow in the OMC2/3 region (Yu et al. 2000). The spectral in- dex for MMS9 is less steep than observed in MMS6, suggest- ing a warmer temperature in the dust envelope, Td ∼ 20 K (Chini et al. 1997). The total envelope mass is estimated to be 10 M⊙, considerably lower than the envelope mass of MMS6. The IRAS flux limits constrain the luminosity of the embedded source to be < 94 L⊙ (Chini et al. 1997). Despite having no coincident IRAS point source, MMS9 is expected to contain a well-advanced Class 0 protostar. Reipurth et al. (1999) ob- served 3.6 cm continuum emission at this location. The rising Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 3 infrared background toward MMS9 may be responsible for the lack of an IRAS point source detection. The only IRAS point source within the OMC3 region is IRAS 05329-0505, coincident with MMS7. From the wave- length of the peak emission, λmax ∼ 60 µm, the warm com- ponent of the dust can be estimated, Td ∼ 50 K. In the sub- millimeter MMS7 appears quite similar to MMS9, although the envelope temperature is somewhat higher Td ∼ 26 K (Chini et al. 1997). The total envelope mass is ∼ 8M⊙, comparable to the envelope around MMS9, and the bolometric luminosity is ∼ 76 L⊙. Reipurth et al. (1999) observed 3.6 cm continuum emission at this location, providing further evidence for an em- bedded Class 0 source. MMS7 appears to be in a somewhat more advanced stage of star formation, powering a giant bipo- lar Herbig-Haro flow (Reipurth et al. 1997) and harboring both a central FU Ori-like source (Reipurth & Aspin 1997) and an optical reflection nebula (Wolstencroft et al. 1986). Further south along the ridge, near the center of the OMC2 molecular core, lies the far infrared source FIR4 (Mezger, Zylka, & Wink 1990), also observed as an IRAS point source. The region around FIR4 harbors a cluster of near-infrared sources (Johnson et al. 1990) although none are explicitly co- incident with the far-infrared source. FIR4 is the brightest sub- millimeter source in the OMC2 region and contains a VLA ra- dio source (Reipurth et al. 1999), leading to identification as a Class 0 object. Its proximity to the near IR cluster suggests, however, that it might be a more evolved envelope. The pres- ence of far-infrared emission provides evidence for warm dust Td ∼ 40 K (Mezger et al. 1990), although the sub-millimeter emission traces the much colder outer envelope Td ∼ 15 K. The total envelope mass is ∼ 34 M⊙ and the luminosity is ∼ 400 L⊙ (Mezger et al. 1990). The final protostellar source in our sample is NGC2071IR in the Orion B North molecular cloud (Johnstone et al. 2001; Motte et al. 2001; Harvey et al. 1979). Significant far-infrared emission from the source (Butner et al. 1990) provides evi- dence of an internal source and a warm dust Td > 50 K com- ponent providing a reprocessed luminosity ∼ 500 L⊙. The total envelope mass, derived from the sub-millimeter dust emission, is 12 − 30 M⊙ (Johnstone et al. 2001). NGC2071 contains one of the most powerful outflows known (Bally & Lada 1983). Along with the five protostellar candidates, two sub- millimeter bright regions which are not associated with indi- vidual protostars were also observed. A sub-millimeter bright location, PDR1, along the warm photodissociation region (the Orion Bar) separating the HII region M42 (the Orion Nebula) from the Orion A molecular cloud was included to monitor changes in the molecular line diagnostics in regions of en- hanced UV fluxes and higher excitation temperatures (see also Hogerheijde, Jansen, & van Dishoeck 1995). Also, a strong shock front, SK1, located via bright 2 µm H2 emission (Yu et al. 1997), was included to test for shock diagnostics. The shocked region is also visible in the spectral index map of the ISF (Johnstone & Bally 1999) as a region with anoma- lously shallow slope, requiring either a lower temperature than the surrounding material, a shallower dust emissivity profile, or contamination of the 850 µm continuum measurement by molecular lines. A summary of the physical properties of the seven sources is presented in Table 1. The sources are ordered from the cold- est, least energetic, to the warmest, most energetic, with the PDR and shock front noted separately. 3. Observations 3.1. ChoiceofMolecules Eight different molecules and their isotopes have been ob- served in order to find useful diagnostics to study physical and chemical differences between the sources in our sample. Among these are carbon-, oxygen-, nitrogen-, and sulphur- bearing species, allowing for a study of differentiation in the chemical abundances. Abundant CO is observed toward all sources. Nevertheless, it is a useful molecule to study deple- tion effects, which are expected to be largest for the coldest, less evolved sources. In those sources where CO is not too depleted, it quickly becomes optically thick. In that case, the optically thin isotope C17O can be used to derive H2 column densities. Formaldehyde, H2CO, is a slightly asymmetric rotor molecule, providing it with many of the temperature and den- sity diagnostic features of purely symmetric rotors. Mangum & Wootten (1993) and van Dishoeck et al. (1993) list many of the strengths of this molecule as a probe of the physical conditions in dense molecular clouds. Methanol, CH3OH, is a complex molecule with many transitions in the sub-millimeter. Since these lines are concentrated in complex bands spanning relatively small wavelength ranges, this allows for efficient ob- servations even at relatively high spectral resolution. Methanol is also chemically important, since it has been detected both in the gas-phase and on grains (e.g. Dartois et al. 1999). The high gas-phase abundances of CH3OH most likely result from evap- oration off of grains (Charnley, Tielens, & Millar 1992; van der Tak, van Dishoeck, & Caselli 2000b; Schoier et al. 2002). One of the most abundant nitrogen-bearing species is HCN. Chemical models predict that the HCN abundances increase with temperature (Rodgers & Charnley 2001), thus provid- ing a useful evolutionary indicator. Observations of massive star-forming regions confirm this (Lahuis & van Dishoeck 2000), although the situation for low-mass sources is less clear (Jørgensen et al. in prep.). Schilke et al. (1992) show that the abundance ratio of HCN and HNC depends on the density and kinetic temperature of the gas, providing a physical diagnostic. The chemical equilibrium of HCN, HNC, and CN also depends on the hardness of the radiation field. Finally, sulphur-bearing species have been associated with shocks and/or outflows (e.g. Orion; Blake et al. 1987). However, recent studies of sulphur-bearing species in massive star-forming regions other than Orion cannot distinguish be- tween an origin in the circumstellar envelope or the outflow (Keane et al. 2001a; van der Tak et al. 2003). Thus, it is still debatable whether sulphur-bearing species are good diagnos- tics of outflows. On the other hand, CS transitions have been used to derive detailed density structure within the molecular envelopes of star-forming regions, making this species a useful indicator of physical properties (van der Tak et al. 2000a). The 4 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion H2S/SO2 ratio has also been proposed as a chemical clock of hot core regions (Charnley 1997), although observational re- sults are still ambiguous (Hatchell et al. 1998; van der Tak et al. 2003). 3.2. Sub-millimeterObservations The molecular line observations were taken at the 15-meter James Clerk Maxwell telescope (JCMT)1 on Mauna Kea in Hawaii. The data were obtained during three separate observ- ing runs: January 1999, September 1999, and December 2000. Table 2 provides details of the instruments and settings used during each run, along with typical receiver conditions. For each source, Table 3 lists the molecular lines observed and the results of fitting one or two Gaussian components to the inte- grated line intensity, the full width at half maximum, and the peak intensity. The data were obtained in a beam switching mode, with a 10 arcminute switch in right ascension. The integrated line data were transfered from the antenna temperature scale T ∗ A to the main-beam brightness scale Tmb by dividing by the main- beam efficiency ηmb. For the A band receiver (210 -- 270 GHz) the measured efficiency is ηmb = 0.69, while for the B band receiver (330 -- 370 GHz) ηmb = 0.60. The relevant beam sizes for the JCMT are 21 and 14 arcseconds at respectively 230 and 345 GHz. The velocity resolution ranged from 0.2 −0.5 km s−1. The rms noise after a typical integration time of twenty minutes was 0.30 K per channel for A band observations and 0.15 K per channel for B band observations. Further details on the JCMT and the heterodyne receivers can be found on the JCMT home- page2. 3.3. RepresentativeLineProfiles Many important differences between the sources are readily ob- served simply by looking at the spectra. The extremely strong, and almost always self-absorbed, CO 3 -- 2 transition reveals in- formation on the gas kinematics within each source (Fig. 2). Single component Gaussian fits to the profiles are only pos- sible for MMS6 (line width 4 km s−1) and PDR1 (3 km s−1). MMS9 requires both a narrow component (6 km s−1), providing two thirds of the integrated intensity, and a broad component (15 km s−1). FIR4 appears to be a scaled-up version of MMS9, with a narrow component (9 km s−1) containing three quarters of the integrated intensity and a broad component (21 km s−1). Both the shock region and NGC2071 are poorly fit by Gaussian components and are dominated by the extended wings. Eight of the methanol lines are observed in a single spectral setting and can be plotted for each source (see examples in Fig. 3). These spectra also contain the sole observed SO transition. In agreement with the results found by van der Tak et al. (2003) there is an apparent lack of correlation between the strength 1 The JCMT is operated by the Joint Astronomy Centre on behalf of the Particle Physics and Astronomy Research Council of the UK, the Netherlands Organization for Scientific Research, and the National Research Council of Canada. 2 http://www.jach.hawaii.edu/JACpublic/JCMT of the SO line and the methanol lines. As well, differences in the relative strengths of the methanol lines are observable be- tween MMS6 and FIR4, indicating the more extreme temper- ature and density conditions in the energetic infrared source. Finally, the FIR4 methanol spectra show evidence for two com- ponents, narrow and broad, similar to the CO spectra. The HCN and HNC 4 -- 3 profiles are also observed in a sin- gle spectral setting (Fig. 4). A range of relative line strengths is seen in the spectra; however, optical depth considerations make simple abundance determinations more complicated (see Sect. 4.5.1). The substantial optical depth of the transitions allows for more detailed determinations of the source kinematics. In all sources except PDR1, the HCN line profile is broader than the HNC line profile or has evidence for extended line wings. 4. Molecular Line Survey 4.1. StatisticalEquilibriumCalculations The analysis of the data is done with statistical equilibrium calculations, using the numerical code RADEX3. The adopted method is described in detail by Jansen, van Dishoeck, & Blake (1994), van Dishoeck et al. (1993), and Jansen et al. (1995). This method assumes that the source is a homogeneous spheri- cal region. The decoupling of radiative transfer and level popu- lations is done with the escape probability method. The optical depth τ at the line center is proportional to the ratio of the total column density and the line width of the molecule. The optical depth and the line intensity depend only on this ratio. All vari- ables can be calculated, using a geometrical assumption that they are constant throughout the region. The radiative excita- tion of rotational lines in the sub-millimeter region is usually dominated by the 2.73 K cosmic background radiation, which peaks at 1.871 mm. Therefore, this external radiation field is in- cluded in the calculations. No central radiation field is included nor infrared pumping by dust. Since the sources have rather low dust temperatures, this latter effect is expected to be negligible. The kinetic temperature Tkin and the density n(H2) for the homogeneous region can be estimated from line ratio calcu- lations where available. The column density of molecule X, N(X), can then be estimated using the temperature and the den- sity as found from line ratio calculations, such that the line width, integrated line strength, and main beam temperature of the emission lines can be fitted. Due to the simplistic treatment of the envelope structure the derived abundances are only roughly determined. Rising temperature and density gradients toward the envelope centers are expected due to heating by the central protostellar sources and these conditions can influence the observed line strengths significantly. Additionally, these Orion sources reside within an enhanced external radiation field, due to the proximity of many nearby massive stars (Li, Goldsmith, & Menton 2003). Thus, the outer envelopes are kept warm, Td ∼ 30 K (Johnstone & Bally 1999). Modeling of such envelopes (Li et al. 2003, Jørgensen et al. in prep.) reveal an inward decreasing tempera- ture gradient near the surface, encompassing much of the mass 3 This program was originally written by J.H. Black. Here a more recent version by D.J. Jansen is used. Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 5 of the envelope, coupled to an inward rising temperature gradi- ent once the central heating source dominates. The large, externally warmed, envelopes surrounding the protostellar sources in Orion should allow for reasonable abun- dance estimates from the observed line intensities using statis- tical equilibrium calculations, except for those transitions re- quiring either very high densities n > 107 cm−3 or high temper- atures T > 50 K. As a guide to the accuracy of the calculations presented in this paper, detailed modeling of the protostellar source IRAS 16293-2422, including careful derivation of the physical structure, has shown that abundance estimates from this simpler procedure are typically only accurate to within a factor of two or three (Schoier et al. 2002). 4.2. IsotopesofCarbonMonoxide Within star-forming regions of molecular clouds, comparison between the line strengths of low-lying CO transitions and the intensity of the sub-millimeter dust continuum often suggest that there is a depletion of CO with respect to the dust, and by extension the molecular hydrogen abundance (see for ex- ample Mitchell et al. 2001, Jørgensen et al. 2002, Bergin et al. 2002, Bacmann et al. 2002). The most common explanation for this observational discrepancy between CO line strength and dust continuum emission is that the CO has depleted through freeze-out onto the grains. The time scale for depletion scales inversely with the density of the region, and is shorter than 105 yr for densities above a few ×104 cm−3. Recent laboratory ex- periments on CO and H2O ices show that CO will remain com- pletely frozen out if the grain temperatures stay below ∼20 K. Above 20 K, CO starts to evaporate back into the gas phase in a complicated pattern depending on the ice structure, with a frac- tion of the CO remaining trapped in the H2O ice matrix until the entire mantle evaporates at ∼100 K (Collings et al. 2003, Fraser et al. 2001). 4.2.1. Line strengths Table 4 details the observed line strengths of CO, 13CO, and C17O 3 -- 2 transitions obtained at the JCMT, along with an es- timate of the column of carbon monoxide along the line-of- sight. Using RADEX, the column is deduced from the inte- grated line strength of the C17O 3 -- 2 line, assuming that the 13CO peak temperature provides a reasonable estimate of the kinetic temperature of the gas. The RADEX calculations for CO include energy levels up to J = 25 and use collisional rate coefficients by Schinke et al. (1985). The results are not signif- icantly changed if the new rate coefficients by Flower (2001) are adopted (see Jørgensen et al. 2002). For NGC2071, where the 13CO line was not observed, the temperature was assumed to be 30 K, similar to the temperature of FIR4. The cosmic abundance ratio for isotopes of Oxygen, [16O]/[17O]∼ 1800 (Wilson & Rood 1994), was used to convert the calculated C17O column to an equivalent column of CO. For the expected conditions within the sub-millimeter clumps, n > 104 cm−3, there is only a slight density dependence on the calculated column of the gas. In most cases the CO peak tem- perature was larger than the 13CO temperature; however, the optical thickness of CO is extreme and hence the line profile may be dominated by a small fraction of hot gas associated with each source. There is still uncertainty comparing the thicker 13CO with the thinner C17O. We estimate the reliability of the computed physical parameters by comparing the RADEX com- puted optical depth of the 13CO line, τmod, with the observation- ally determined optical depth τobs, measured by comparing the peak temperature of the C17O against the 13CO peak tempera- ture and assuming [13CO]/[C17O]= 25 (Wilson & Rood 1994). This calculation confirms that the 13CO line is optically thick in all cases (τ13CO > 1.3). While there are deviations between the two determinations of the optical depth, the order unity dis- persion provides an equivalent estimate of the reliability of the measured column density of CO. 4.2.2. Determination of CO abundance The column density of molecular hydrogen within each en- velope is best determined from the sub-millimeter flux mea- surements. Assuming that the 850 µm emission observed by Johnstone & Bally (1999) and Johnstone et al. (2001) traces the bulk of the envelope and that the dust emission is optically thin, the mean column density through the center of each clump is related to the peak brightness: N(H2) = S850Ωbm B850(Td)κ850 µ mH2 , (1) where Ωbm is the effective solid angle of the SCUBA beam (in steradians), B850(Td) is the Planck function at 850 µm for dust radiating at Td, κ850 is the dust emissivity at 850 µm (assumed here to be κ850 = 0.02 cm2 g−1), mH2 is the molecular hydrogen mass, and µ is the mean molecular weight (in units of mH2). As shown in Appendix A, significant CO line flux can con- taminate the broadband continuum measurements; therefore, it is important to correct the continuum measurements for the presence of line emission. Having both line and continuum measurements for each of the sources in Orion allows for a di- rect measure of the level of this line contamination. At 850 µm, 1 Jy within the JCMT 14 arcsecond beam is equivalent to a brightness temperature of 51 mK. However, the broad SCUBA 850 µm bandpass, 30 GHz wide, dilutes the importance of indi- vidual spectral line features against the continuum. The equiv- alent line flux required to mimic a 1 Jy signal with SCUBA on the JCMT is 1540 K km s−1. For a typical peak gas tempera- ture of ∼ 25 K, the Gaussian component of the CO line would need to be ∼ 6 km s−1 wide to produce a ten percent, 0.1 Jy, error in the continuum measurement. Thus, for strong contin- uum sources, S > 2 Jy, line contamination should be minimal. Table 5 presents the CO line contamination for each of the six Orion sources. For completeness, the combined line contami- nation due to all other observed lines lying within the 850 µm bandpass is also shown. With the exception of the shock knot SK1, the total line contamination is always less than about ten percent. Note that SK1 has a high peak brightness temperature for the CO 3 -- 2 transition and has very extended line wings, consistent with its identification as a hot shock front by Yu et al. (1997) on the basis of bright H2 emission. 6 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion After accounting for line contamination, it is straightfor- ward to compute the column density of H2, derived from the strength of the remaining continuum emission, and the frac- tional abundance of CO for each source. These values are also presented in Table 5. Within the PDR and the shock front, the abundance of gas-phase CO agrees with the expected value of 10−4 to within the accuracy of the computations (a factor of a few). However, toward each of the sub-millimeter proto- star regions the CO gas-phase abundance is smaller than ex- pected by about a factor of 5 - 10, further evidence that al- most all of the CO has been depleted onto grain surfaces in the dense envelopes. This result is in qualitative agreement with the more detailed modeling results presented by Jørgensen et al. (2002), where low-mass Class 0 and pre-stellar sources were observed to have similarly depleted CO. Here, however, the outer envelope is warm, Td ∼ 30 K and the bulk of the de- pletion likely occurs in an intermediate zone between the hot, centrally heated deep interior and the externally heated surface layer. The higher abundance of CO in MMS9 is likely due to the smaller envelope, and thus a fractionally smaller interme- diate zone. Detailed modeling of externally heated protostellar envelopes supports this hypothesis (Jørgensen et al. in prep.). 4.3. Formaldehyde One of the advantages of H2CO is the ability to observe, in the same side-band of a single spectrum, two transitions which to- gether probe the temperature of the gas under a large variation in density conditions. This feature removes many of the obser- vational uncertainties inherent in sub-millimeter spectroscopy such as beam effects, weather conditions, and instrumental gain. The RADEX calculations for H2CO include energy lev- els up to J = 10 and use collisional rate coefficients by Green (1991). In the present study we observed three para-formaldehyde lines. Two of the lines were simultaneously observed in the 218 GHz atmospheric window, 303 − 202 and 322 − 221. Because transitions between the different Kp ladders are radiatively for- bidden, the ratio of the line strengths from transitions in dif- ferent ladders depends only on the kinetic temperature of the gas, operating as a finely tuned thermometer over an extended density range (Mangum & Wootten 1993). A third line was ob- served at 362 GHz, 505 − 404, within the same Kp ladder as the 303 − 202 transition. Within a given Kp ladder the excitation en- ergies do not vary dramatically and thus the ratio of these lines provides a useful density indicator at low to moderate densities (nH < 3 × 106 cm−3) and a second temperature probe at higher densities. The first panel in Fig. 5 plots the expected line ratios against temperature and density. The physical parameters derived from the formaldehyde lines are presented in Table 6, and are plotted graphically for each source in Fig. 5. The spread of line ratios R33 and ki- netic temperature of the gas inferred from the 303 − 202 and 322 − 221 transitions ranges from R33 < 2, T > 300 K, at the location of PDR1, to R33 > 10, T < 30 K at the location of the sub-millimeter source MMS9. As expected, the inter- mediate mass star-forming regions NGC2071 and FIR4 show moderately high temperatures T > 70 K, consistent with the observability of infrared dust continuum emission. Mangum & Wooten (1993) also observed FIR4 and NGC2071, obtaining temperatures of 85 K and 80 K and with a range of probable temperatures bounding the results of this study. The only in- frared source in OMC3, MMS7, shows no evidence for high temperatures. The strength of the 505 − 404 transition constrains the den- sity of the environment from which the emission is observed. A complication arises from the difference in the beam size at the telescope for the 303 − 202 and 505 − 404 lines. If the line emission fills the larger 218GHz beam, then there are no diffi- culties in taking the ratio of the brightness temperatures of the two lines. However, if the source of the formaldehyde emission is smaller than the 362GHz beam the ratio of the beam sizes should be taken into account when comparing the brightness temperatures to account for the effect of beam dilution. Source sizes intermediate between the 218GHz and 363GHz beam re- quire a varying normalization. Thus, the measured line ratio R53, computed from the 505 − 404 and 303 − 202 lines, has an inherent uncertainty of ∼ 3 due to the possible complications from intermediate beam dilution, with the value given in Table 6 strictly an upper limit. For all sources the measured ratio R53 is larger than 0.3, im- plying a high density, n > 106 cm−3. The relatively evolved far infrared source FIR4 shows two kinematically distinct compo- nents, a narrow-line region (∆V ∼ 1.5 km s−1) requiring an ef- fective density near 106 cm−3 and broad-line component (∆V ∼ 6 km s−1) requiring a much higher density. While MMS6 and NGC2071 also show distinct velocity components, the den- sity does not obviously vary between the components for these sources. Intriguingly, in all three cases the temperature deduced from both the narrow and broad-line components are quite sim- ilar. With the temperature and density estimated, the column density of para-formaldehyde can be deduced from the inte- grated line strength of the 303 −202 transition. The high temper- ature equilibrium ratio of the para to ortho forms (1:3) was used to convert this result into a total formaldehyde column density. The inferred abundance of formaldehyde was then measured against both the H2 column, deduced from the sub-millimeter dust continuum emission, and the measured CO column. The results are presented in Table 6. The two infrared sources show the largest abundance of formaldehyde measured against the CO. For these sources the abundance ratios lie in the range [H2CO]/[CO] = 2.5 − 10.0 × 10−5. The sub-millimeter sources MMS6 and MMS9 show a lower abundance ratio [H2CO]/[CO] ∼ 1 × 10−5. Disentangling the importance of CO depletion in these data complicates any analysis unless the formaldehyde depletes in a similar fashion to CO. However, comparing the formaldehyde abundance to H2 produces a clear relation (see Table 6 and Fig. 8). The hot PDR region and the infrared sources, NGC2071 and FIR4, have the largest formaldehyde abundance, approximately 5-10 times greater than observed for the sub-millimeter sources. At least half of the increase in the abundance of formaldehyde in the two infrared sources is due to the broad-line component. Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 7 In low-mass Class 0 sources, typical [H2CO]/[CO] abun- dance ratios in the outer region are a few ×10−5, similar to those observed here, but jumps in the H2CO abundance by more than a factor of 100 have been found in the inner warm envelopes, consistent with ice evaporation (e.g., Ceccarelli et al. 2000, Schoier et al. 2002, Maret et al. in prep.). H2CO is expected to be a product of grain surface hydrogenation of CO and has been tentatively detected in interstellar ices with an abundance of a few percent with respect to H2O ice, corresponding to ∼ 10−6 with respect to H2 (Keane et al. 2001b). Thus, the trend observed here for the protostellar sources could be consistent with evaporation of H2CO-containing ices in the warm interior regions, with more energetic sources containing a larger frac- tional mass of envelope at high enough temperatures to evapo- rate such ices. For high-mass sources, H2CO shows a less clear relation with temperature (van der Tak et al. 2000b). 4.4. Methanol Methanol emission lines have been detected in complex bands at 241 and 338 GHz, with excitation energies ranging upward from Tex ∼30 K. The detection of many CH3OH lines in our spectra allows for the construction of a rotation diagram for each source in order to get a first indication of the tempera- ture and column density. Described in detail by Blake et al. (1987), this method assumes that the lines are optically thin and that the excitation can be characterized by a single tem- perature. The resulting rotation diagrams, however, are often misleading. Direct computation of the level populations using RADEX shows that the higher excitation lines require increas- ingly larger critical densities and thus the derived temperature from rotation diagram fitting is heavily dependent on the den- sity of the medium in which the methanol resides (see also the discussion by Bachiller et al. 1995). Figure 6 reveals the ef- fect on the rotation diagram produced by increasing the critical density of the medium while holding the kinetic temperature constant at 100 K and the total column density at NCH3OH = 1014 cm−3. Even though reasonable linear fits are produced, the derived temperature ranges from 25 K for nH2 = 107 cm−3 to 80 K for nH2 = 109 cm−3. The derived column densities also overestimate the actual amount of methanol by factors of a few. For a protostellar source with both density and tempera- ture gradients a proper analysis becomes even more complex. Given that higher excitation lines tend to have higher critical densities, it is likely that these lines are dominated by emission solely from the deep interior of the envelope. Indeed, there is some evidence that the source size decreases as the required excitation of the line increases (Hatchell et al. 1998). The bulk properties of the envelope derived from methanol, however, are expected to be only slightly weighted toward the highest den- sity and temperature regions due to the small angular extent of these extreme locations. Complicating even a detailed analy- sis, there is evidence for an enhanced abundance of methanol at high temperatures, T > 90 K, in protostellar envelopes (van der Tak et al. 2002b, Schoier et al. 2002). Despite the above concerns, one might hope that methanol may be used both as a temperature and density indicator by treating all of the lines through a proper statistical equilibrium calculation. The RADEX calculations for methanol include en- ergy levels up to 260 K and use collisional rate coefficients by Lees & Haque (1974) and Walmsley (private communication). For this paper, all the measured methanol lines were compared with RADEX computations to produce a chi-squared fit as a function of temperature and density. The results are presented graphically in Fig. 7 and are tabulated in Table 7. For FIR4, the methanol measurements can be separated into narrow and broad line components. The narrow line region requires very high density and temperature n > 108 cm−3, T ∼ 70 K, while the broad line component requires n ∼ 108 cm−3, T ∼ 80 K. The broad line region appears to be beam diluted, at least with respect to the larger 241GHz beam, requiring a normalization factor between the two sets of measurements to produce rea- sonable chi-squared fits. These results are not consistent with the formaldehyde measurements in that higher densities are re- quired to account for the methanol observations. Given a range of densities within the surroundings of FIR4, however, the methanol observations would be most sensitive to the highest density gas. In contrast, the other infrared source, NGC2071, is best fit by a moderate density, n ∼ 5 × 106 cm−3, and a mod- erate temperature, T ∼ 60 K, consistent with the formaldehyde measurements. The two sub-millimeter sources show some variation in the inferred temperature. MMS6 requires a moderate temperature T ∼ 50 K while MMS9 requires a low temperature T < 30 K. For both sources, the methanol appears to probe a dense in- ner region, ∼ 107 cm−3 compared with the formaldehyde de- rived envelope densities ∼ 106 cm−3. This is likely due to the higher critical density required to excite the methanol transi- tions. Across all the sources there is a reasonable correlation between the methanol and formaldehyde temperatures with the sub-millimeter sources being the coldest, the infrared sources being moderately warm, and the PDR having the highest tem- peratures. The derived abundance of methanol is similar to that of formaldehyde. The warm infrared sources have the highest abundance with respect to CO. However, when compared with the total column of hydrogen deduced from the sub-millimeter continuum measurements (Table 5), the PDR region also shows an enhanced abundance of methanol. Considering that the methanol observations are predominantly probing higher den- sity regions within each envelope, it is likely that these abun- dance estimates are strictly lower limits. Differences in the abundance derivations across sources may be affected by the extent of the warm interior within each envelope. Like H2CO, CH3OH is expected to be a product of the hy- drogenation of CO on grain surfaces (Tielens & Charnley 1997) and is detected in grain mantles, albeit with significant varia- tions between different sources (Dartois et al. 1999). CH3OH abundance jumps of at least a factor of 100 have been found in various high-mass objects (van der Tak et al. 2000b), whereas the well-studied low-mass Class 0 source IRAS 16293 -- 2422 shows a jump by nearly three orders of magnitude in the inner warm envelope (Schoier et al. 2002). Such high CH3OH abun- dances can only be produced by evaporation from grain man- tles. Enhanced CH3OH abundances have also been detected in 8 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion outflows at positions offset from the protostar, e.g., in L1157 (Bachiller & P´erez-Guti´errez 1997), where the CH3OH is re- leased from the grain mantles by shocks rather than thermal evaporation. The general trend of an increased CH3OH abun- dance with temperature observed for our sources is consis- tent with this picture. The presence of both narrow and broad CH3OH line profiles suggests that both evaporation mecha- nisms play a role in some objects (e.g., NGC 2071). 4.5. Nitrogen-bearingSpecies 4.5.1. HCN and HNC Identical lines of both HCN and HNC are observed simulta- neously with a single spectrograph setting at the JCMT and thus, assuming the collisional cross sections for the various ex- citations are identical across species, the relative abundance of the two molecules can be simply determined if the transitions are optically thin. To test the optical depth conditions, the ra- tios of line strengths between HCN and H13CN, and HNC and HN13C were computed. In all cases the HCN line was found to be moderately to strongly optically thick (τHCN ∼ 3 − 6) however, only the most energetic sources FIR4 and NGC2071 were found to have optically thick HNC lines (τHNC ∼ 2 − 3). The [HCN]/[HNC] ratio tabulated in Table 8 was determined by accounting for these optical depth effects. For the four protostellar regions observed, the integrated [HCN]/[HNC] abundance ratio ranges from 5 -- 10. Using the chemical models by Schilke et al. (1992), this would imply low temperatures T < 30 K and high densities n > 107 cm−3, or high temperatures T > 50 K and moderate to low densities n < 106 cm−3. For PDR1, the ratio rises to 70, requiring both high densities n > 106 cm−3 and high temperatures T > 50 K (Schilke et al. 1992). It is worth noting that the main obser- vational difference between the least energetic sub-millimeter sources, MMS6 and MMS9, and the most energetic sources, FIR4 and NGC2071, is in the peak brightness of the lines. Thus, it would appear that the conditions probed by the HCN and HNC molecules in all four sources are nearly identical but that the solid angle covered in the most energetic sources is somewhat larger. RADEX calculations support this hypoth- esis. For the determined physical conditions of the optically thick HCN lines the excitation temperature is expected to be relatively large Tex > 10 K. The least energetic sources have observed peak intensities of only Tpeak ∼ 1 K, while the in- frared sources show Tpeak ∼ 15 K. Care should be taken in us- ing chemistry to constrain temperature, however, since the low- mass sources observed by Jørgensen et al. (2003; in prep.) have high HCN/HNC abundance ratios at low temperatures, which are not readily explained by the Schilke et al. (1992) models. RADEX calculations for HCN, including energy levels up to J = 12 and using collisional coefficients by Green & Thaddeus (1974), have been performed to derive HCN abun- dances using the estimated physical conditions derived from formaldehyde. The resulting abundances with respect to CO are ∼5×10−5 for the sub-millimeter sources and PDR1; however, the production of HCN appears to be significantly enhanced in the far infrared sources FIR4 and NGC2071. The HCN abun- dances with respect to H2 vary from 0.6 × 10−10 to 40 × 10−10 (Table 8). These latter abundances show a clear trend with in- creasing source energetics. HCN has not been detected in inter- stellar ices due to a lack of clear spectroscopic signatures, but is expected to freeze-out in cold regions like most molecules. The HCN abundance is known to be enhanced by orders of magni- tude in the warm inner regions of massive protostars (Lahuis & van Dishoeck 2000; Boonman et al. 2001), but the high temper- ature routes to HCN do not become effective until T > 200 K (e.g., Doty et al. 2002). It remains to be determined whether the intermediate mass sources studied here have sufficient amounts of gas at such high temperatures to enhance HCN significantly. The trends observed here are more likely due to general evap- oration of ices with temperatures between 20 and 100 K, en- hancing the overall gas-phase abundances. Detailed consideration of the HCN line shows that it is usu- ally wider than the HNC line. This requires that in the broad line region the ratio of the two molecules is much higher than the ratio of the total integrated intensities, implying higher tem- peratures and/or higher densities if chemical equilibrium is as- sumed (Schilke et al. 1992). Another possible explanation for the general lack of broad line HNC may be found in the shock model solution of Schilke et al. (1992) where HNC is pref- erentially destroyed within the interaction region, reforming downstream from the shock front. In particular, the broad HCN emission, coupled with little or no HNC emission, found in the shock knot SK1 is likely best explained in this manner. Interferometer data of the low-mass source IRAS 16293 -- 2422 also show that HNC avoids the warm inner region, in contrast with HCN (Schoier et al. 2003; in prep.). 4.5.2. CN Observations of the CN hyperfine structure at 340 GHz pro- vides a set of constraints for the modeling of CN (see Table 3). In the optically thin limit, the ratio between the strongest and weakest hyperfine components should be ∼ 15; however, this is only observed in the spectrum of PDR1. In all other regions, except possibly MMS6, the main CN component is optically thick, (τCN ∼ 3 − 6), while the weaker components remain rea- sonably optically thin. Assuming that the line flux is predom- inantly from the outer envelope surrounding the sources, the column of CN derived from the optically thin hyperfine lines is presented in Table 8. The RADEX calculations for CN include energy levels up to J = 15. As found for the abundance of HCN, only the coldest, least energetic source, MMS6 has a dra- matically low abundance of CN. However, a clear correlation is found between the ratio [HCN]/[HNC] and [HCN]/[CN]. Assuming that all CN is locked up in these three molecules, the total abundance of CN can be readily determined. Table 8 displays the results of this calculation. With the exception of the sub-millimeter source (MMS6) the derived total abundance of CN is much more constant than any of the individual molec- ular species. Results of systematic studies of CN in low- or high-mass sources are not yet available, but like HNC, CN seems to avoid the inner warm region in the low-mass protostar IRAS 16293 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 9 -- 2422. The radical is known to be abundant in the outer regions of PDRs (e.g., Fuente et al. 1998), where it is produced by rapid gas-phase chemistry of CH and C2 with N. Alternatively, it can be a photodissociation product of HCN. Photodissociation of CN itself requires photons with energies higher than 12 eV. The observed lack of variation of the CN abundance in the pro- tostellar sources observed here implies that they are exposed to similar radiation fields. As mentioned earlier, the Orion region is bathed in a significantly increased external radiation field due to the close proximity of massive stars. The much lower abun- dance within the massive envelope of MMS6, however, is not easily explained via this hypothesis unless the structure of the envelope is such that the interstellar radiation field is unable to penetrate through the surface layer and into the significant interior. 4.6. Sulphur-bearingSpecies 4.6.1. CS Both CS and the rarer C34S isotope were observed in the J = 5 -- 4 transition. In all cases the line ratio was less than the iso- topic ratio of 22 implying moderately optically thick conditions within the CS line but optically thin conditions for the isotope. Taking the formaldehyde physical conditions for the tempera- ture and the density, the column density of CS was determined using RADEX (Table 9). The RADEX calculations for CS in- clude energy levels up to J = 12. The abundance with respect to H2 trends with the source energetics, rising from a few 10−11 to a few 10−10, in the protostellar sources. This is a stronger trend than seen for HCN suggesting that the more energetic sources are providing more heating to their outer envelope. The abun- dance rises sharply to a few 10−9, in the warm PDR1 source. For four of the sources the J = 7 -- 6 transition of C34S was also detected. In all observed cases the ratio of the J = 7 -- 6 versus J = 5 -- 4 lines implies a warm, (T > 50 K) and dense (n > 106) environment (Table 9). The C34S-derived temper- ature is higher than that derived from H2CO and more likely refers to the inner warm and dense region rather than the outer envelope. If more transitions are observed, a detailed density structure for each of the sources could be determined. 4.6.2. SO The only observed SO transition was unfortunately blended with a methanol line and thus difficult to accurately measure. Despite this complication, three sources, spanning a range of energetics and conditions, have unambiguous detections for SO (MMS6, NGC2071, and PDR1). Assuming the formalde- hyde derived properties for the physical conditions of the gas, RADEX was used to determine the required column density of SO for each of the sources, The RADEX calculations for SO include energy levels up to 580 K. The results are presented in Table 9, along with the abundance of SO in each source and the SO to CS abundance ratio. Observations of more SO lines and other sulphur-bearing species, such as SO2 and H2S, are needed to study possible differences in the sulphur-chemistry in these sources. 5. Discussion and Conclusion This molecular line study was undertaken to determine if mor- phological clues and qualitative indicators were observable across a range of environmental conditions, from pre-stellar and young protostellar envelopes Lbol ∼ 1 − 100L⊙ through infrared bright energetic sources Lbol ∼ 400L⊙, a PDR knot and a shock front. Several trends are apparent, especially in the derived abundances of many molecular species. As well, there are a number of spectral hints that protostellar sources reside within the sub-millimeter clumps, excluding the PDR and the shock knot. 5.1. Abundances Despite the inherent danger in assuming a single temperature and density for the environments of each source, general trends in abundance are observed. It is worth noting that detailed mod- eling (van der Tak et al. 2000a, Schoier et al. 2002, Doty et al. 2002, Jørgensen et al. 2002) provides a much more accurate determination of relative abundances, especially for molecules which are excited only in parts of the envelope. Also, the pres- ence of abundance gradients or 'jumps' can be established for some molecules. Such models require a determination of the density and temperature profiles of the sources from the dust continuum emission, which has not yet been completed for this study. Schoier et al. (2002) show that the inferred abun- dances using a constant temperature and density do not differ by more than a factor of a few from the detailed analysis as long as the adopted conditions are appropriate for the partic- ular molecule or line, which should be the case for this work. However, detailed modeling requires significant input as to the density and temperature profile of the source, as well as relying on additional assumptions such as the dust emissivity profile. Additionally, the Orion sources are bathed in a strong external radiation field (Li et al. 2003) requiring careful consideration of the exterior conditions of each source envelope where the dust and gas temperatures reach T ∼ 30 K. This study is concerned primarily with the constraints on physical conditions provided by the variations in the molecular tracers without resorting to detailed modeling of individual sources. Some strong trends are observed across the source list. The peak brightness and integrated line strength of both CO and 13CO 3 -- 2 lines follow the energetics and warm dust tempera- tures of the envelopes. Despite the observed depletion of CO in the protostellar source envelopes (typically the abundance is a factor of 10 lower than the mean molecular cloud value), in all cases the objects are visible in each of the observed CO lines. The formaldehyde derived temperatures, T(H2CO), do not match the envelope dust temperatures, measured in the sub- millimeter. However, for the sub-millimeter sources MMS6, MMS7, and MMS9 T(H2CO) is consistent with the gas tem- perature in the outer envelope as seen in the 13CO line. For the infrared sources, FIR4 and NGC2071, T(H2CO) requires in- ternal heating through a significant fraction of each envelope. Detailed modeling of other sources provides evidence that the formaldehyde abundance increases in the warm, dense interi- ors of protostellar envelopes, and that grain mantle evaporation 10 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion may be important in producing the enhancement. However, the relatively low lying lines observed in this study are more accu- rate tracers of the extended envelope conditions, where most of the mass resides (Jørgensen et al. in prep.). While more uncer- tain, the methanol derived temperatures follow the same trend as the formaldehyde temperatures. For the remainder of the dis- cussion we adopt the formaldehyde conditions as representa- tive of the bulk conditions within each source. Despite uncertainties of order unity, the abundance of both formaldehyde and methanol correlate well against the source energetics (Fig. 8). The changing abundance may reflect jump conditions in formaldehyde and methanol abundance with dust temperature in the inner warm part. Ignoring the broad (out- flow) methanol component in NGC2071, the abundance of both formaldehyde and methanol may be plotted as a step func- tion with cold sources, T < 50 K, having a low abundance and warm sources, T > 50 K, having about five times higher abundance. Although less observational data from the study are available, the CS abundance also appears to correlate with source energetics. The abundance of HCN is also sensitive to source con- ditions; however, this is mostly determined by extreme cases MMS6 and PDR1. In contrast to the other observed species, little variation of abundance was found for HCN among the energetic sources, FIR4 and NGC2071, and the weak sub- millimeter source MMS9. Considering the three dominant CN- bearing species (HCN, HNC, and CN), the total abundance of these three molecules is quite constant across a wide range of conditions, varying by less than thirty percent between MMS9, FIR4, and NGC2071 and only varying by a factor of two when the PDR region is included. Only the cold, dense sub- millimeter source MMS6 appears to have a severely depleted abundance of CN-bearing species. The enhanced external radi- ation field in Orion may be responsible for this apparent equi- librium. The general observed trend of increased abundance with source energetics is consistent with a scenario in which freeze- out of molecules occurs in the cooler intermediate zone within the envelope and evaporation of ices occurs in the warmer inte- rior and possibly exterior regions. Similar trends have been ob- served for samples of low-mass (Jørgensen et al. 2002, 2003) and high-mass (van der Tak et al. 2000a,b; Boonman et al. 2003) sources without external heating, although not for all molecules. The only other intermediate-mass source that has been studied in some detail chemically is AFGL 490 (Schreyer et al. 2002) which has a luminosity of 103 L⊙. This source has a large envelope, with abundances comparable to those of the warmer sources observed here. Strong solid-state features of various species indicative of freeze-out are also detected. None of the observed sources shows the characteristic crowded line spectra of a 'hot core', such as found for high- mass protostars like G34.3+0.15 (Macdonald et al. 1996), G327.3 -- 0.6 (Gibb et al. 2000) or W3(H2O) (Helmich & van Dishoeck 1997). Deeper integrations are needed to determine whether complex organic molecules like CH3OCH3 are present in the warmer sources in our sample, e.g. FIR4 or NGC 2071. These molecules are expected to be produced by gas-phase chemistry between evaporated ices, but have so far only been observed in high-mass sources. The different dynamical time scales of the hot core gas in low- versus high-mass objects com- pared with the chemical timescales of ∼ 104 yr may prevent formation of such second generation species. The PDR1 position differs chemically from the other sources by having the highest gas-phase abundances overall and the largest [HCN]/[HNC] ratio. PDRs also generally have high abundances of radicals such as CN and C2H and ions like CO+ (Jansen et al. 1995). Indeed, a study of a set of more evolved intermediate-mass sources by Fuente et al. (1998) has identified CN as a particularly good diagnostic of the ultravio- let radiation. In our sources, this trend is not so evident, likely because all sources are located in the Orion region which is bathed in intense radiation from nearby O and B stars. Few molecules are observed in the SK1-OMC3 shock and only CO shows truly broad line wings at that position. Sulphur- containing species like H2S, SO and SO2 are predicted to be abundant in shock models (e.g., Leen & Graff 1988) and are seen to be enhanced in the Orion-KL plateau gas with broad line wings, but are not prominent here; however, the source is quite weak and the lines may be below our detection limits. In NGC 2071, molecules like CH3OH are present in the outflow but this results from grain evaporation in the shock rather than high-temperature chemistry. Chernin, Mason, & Fuller (1994) have observed broad SO emission in NGC 2071 with some abundance enhancement. Deeper searches for shock diagnos- tics and accurate determinations of abundances at shocked and non-shocked positions in these and other sources are needed. 5.2. ProtostellarSourceDiagnostics A large number of sub-millimeter continuum surveys have now been completed in star-forming molecular clouds, and hun- dreds of new sub-millimeter bright envelopes have been enu- merated. However, an outstanding question is which fraction of these objects surround protostellar sources and which are pre-stellar. Within the present study there were no clear pre- stellar objects although MMS6 appear somewhat ambiguously defined as a Class 0 source. Despite this lack of an obvious pre- stellar candidate, it is worth considering if any of the observed molecular signatures require an embedded heating source. For many of the sub-millimeter objects found within sur- veys, 13CO observations have provided no indication of a co- incident CO peak (Mitchell et al. 2001). Thus, the clear mea- surement of CO isotopes in this sample, despite the depletion, may provide a clue to which objects contain embedded sources, perhaps by warming and evaporating CO back into the envi- ronment from which it had frozen out during an earlier epoch. Alternatively, the presence of a strong external radiation field may bias the results in Orion. The broad CO line wings provide clear evidence of enhanced kinematics within these condensa- tions and, along with the ubiquitous outflows in star-forming regions, act as a sign post for embedded sources. The appearance of molecular lines with relatively high excitation temperatures, and preferably also with high criti- cal densities such that the warm region might be inferred to be deeper within the envelope, should provide a signature to Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 11 Class 0 and later sources. Thus, the methanol lines are likely the strongest indicator of a warm, dense region within the envelope. For all sources observed in this study, except the known outflow source MMS9, the required internal tempera- ture deduced from the methanol lines is > 50 K, a temperature unattainable without an energetic internal source. The CS 7 -- 6 line also provides evidence of a warm, dense interior region in all sources except MMS9. It is interesting to note that all the protostellar sources considered in this paper require high densities in their interior. Standard models for star formation expect a power-law density distribution toward the clump cen- ter during the collapse phase (Shu 1977; Henriksen, Andr´e, & Bontemps 1997) The high densities measured here may be a result of the stage of evolution of the individual clumps. Formaldehyde observations also provide important evi- dence for a large warm envelope. While the low-lying lines observed in this study are predominantly excited in the mas- sive outer envelope, the derived temperature places constraints on the required heating source. In particular, envelope temper- atures above 20 K are difficult to reconcile without either a warming source inside or an external heating source. All the sources observed in this study have two components, narrow and broad, associated with the formaldehyde measurements. However, this is not always the case for sub-millimeter bright regions (Mitchell et al. 2001; Tothill, private communication). It is possible that the broad component of the formaldehyde, often implying a much denser zone, is tracing an inner region within the envelope which is undergoing collapse or has be- come much more turbulent. Not all heating is due to protostellar sources and it is there- fore important to distinguish external heating (such as in a PDR) from internal heating. Protostellar sources should have central condensations which show up both in sub-millimeter dust continuum maps and maps of molecular lines with high critical densities (e.g., CS 7-6). Chemically, PDRs are best rec- ognized by a high [HCN]/[HNC] ratio and by high abundances of radicals such as CN and C2H and ions like CO+ (see above discussion). 5.3. LineContamination One of the original motivations for this molecular line study was to determine the importance of line contamination within the broad SCUBA 850 µm passband. While theoretical calcu- lations (Appendix A) show that the influence of CO contami- nation can become exceedingly large in particular situations, it is also clear that typical conditions within molecular clouds are not so extreme. The observational evidence presented (Table 5) does show that the contamination, while typically less than 10%, occasionally dominates the continuum flux. However, this only occurs in regions with warmer molecular gas temper- atures and large velocity gradients which allow for enhanced integrated line strengths in CO and its isotopes. Such con- ditions occur most often at isolated shock fronts within the cloud, for example in the knots associated with protostellar jets. Contamination may also arise from lines of other molecules, especially around energetic sources. The best observed exam- ple for this is Orion KL, for which a forest of lines, primarily from SO and SO2, produce between 28% and 50% line con- tamination at 850 µm (Serabyn & Weisstein 1995; Groesbeck 1994). Consideration of Table 5 shows that in our sample the line contamination is never dominated by lines other than CO; however, for the most energetic sources line emission from HCN, HNC, CN, and methanol provides a significant fraction (> 40 %) of the total line contamination. Acknowledgements. We wish to thank Jes Jørgensen, Floris van der Tak, John Bally and Heather Scott for helpful comments during the creation of this manuscript. The careful review by the anonymous ref- eree produced many improvements to the paper. DJ wishes to thank the Sterrewacht Leiden for its kind hospitality during the past three summers, at which time most of the data analysis was conducted. DJ also acknowledges the support of an NSERC grant and support from NOVA. This work was supported by the Netherlands Organization for Scientific Research (NWO) through grant 614-041-003 and a Spinoza grant. References Aso, Y., Tatematsu, K., Sekimoto, Y., Nakano, T., Umemoto, T., Koyama, K., & Yamamoto, S. 2000, ApJS, 131, 465 Bachiller, R. & P´erez-Guti´errez M. 1997, ApJ, 487, 93L Bachiller, R., Liechti, S., Walmsley, C. M., & Colomer, F. 1995, A&A, 295, 51 Bacmann, A., Lefloch, B., Ceccarelli, C., Castets, A., Steinacker, J., & Loinard, L. 2002, A&A, 389, 6 Bally, J. & Lada, C. J. 1983, ApJ, 265, 824 Bally, J., Langer, W. D., Stark, A. A., & Wilson, R. W. 1987, ApJ, 313, L45 Bergin, E.A., Alves, J., Huard, T., & Lada, C.J. 2002, ApJ, 570, 101L Blake, G.A., Sutton, E.C., Masson, C.R., & Phillips, T.G. 1987, ApJ, 315, 621 Blake, G.A., van Dishoeck, E.F., Jansen, D.J., Groesbeck, T.D. & Mundy, L.G. 1994, ApJ, 428,680 Blake, G.A., Sandell, G., van Dishoeck, E.F., Groesbeck, T.D., Mundy, L.G. & Aspin, C. 1995, ApJ, 441, 689 Boonman, A.M.S., Stark, R., van der Tak, F.F.S., van Dishoeck, E.F., van der Wal, P.B., Schafer, F., de Lange, G., Laauwen, W.M. 2001, ApJ, 553, 63L Boonman, A.M.S. & van Dishoeck, E.F. 2003, A&A, 403, 1003 Boonman, A.M.S., Doty, S.D., van Dishoeck, E.F., Bergin, E.A., Melnick, G.J., Wright, C.M., & Stark, R. 2003, A&A, 406, 937 Buckle, J.V. & Fuller, G.A. 2002, A&A, 381, 77 Butner, H.M., Evans, N.J.II, Harvey, P.M., Mundy, L.G., Natta, A., & Randich, M.S. 1990, ApJ, 364, 164 Castets, A. & Langer, W. D. 1995, A&A, 294, 835 Charnley, S.B. 1997, ApJ, 481, 396 Charnley S.B., Tielens A.G.G.M., & Millar T.J. 1992, ApJ, 399, L71 Ceccarelli, C., Loinard, L., Castets, A., Tielens, A.G.G.M., & Caux, E. 2000, A&A, 357, 9L Chernin, L.M., Masson, C.R., & Fuller, G.A. 1994, ApJ, 436, 741 Chini, R., Reipurth, B., Ward-Thompson, D., Bally, J., Nyman, L.-Å., Sievers, A., & Billawala, Y. 1997, ApJ, 474, L135 Collings M.P., Dever, J.W., Fraser H.J., & McCoustra M.R.S., 2003, ApJ, 583, 1058 Dartois, E, Schutte, W., Geballe, T.R., Demyk, K., Ehrenfreund, P., & D'Hendecourt, L. 1999, A&A, 342, L32 Doty, S.D., van Dishoeck, E.F., van der Tak, F.F.S., & Boonman, A.M.S. 2002, A&A, 389, 908 12 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Flower, D.R. 2001, J. Phys. B, 34, 2731 Fraser, H.J., Collings, M.P., McCoustra, M.R.S., & Williams, D.A. Schinke, R., Engel, V., Buck, U., Meyer, H., & Diercksen, G.H.F. 1985, ApJ 299, 939 2001, MNRAS, 327, 1165 Schoier, F.L., Jørgensen, J.K., van Dishoeck, E.F., & Blake, G.A. Fuente, A., Martin-Pintado, J., Bachiller, R., Neri, R., & Palla, F. 1998, 2002, A&A, 390, 1001 A&A, 334 Gibb, E.L., Whittet, D.C.B., Schutte, W.A., Boogert, A.C.A., Chiar, J.E., Ehrenfreund, P., Gerakines, P.A., Keane, J.V., Tielens, A.G.G.M., van Dishoeck, E.F., & Kerkhof, O. 2000, ApJ, 536, 347 Green, S. 1991, ApJS, 76, 979 Green, S. & Thaddeus, P. 1974, ApJ, 191, 653 Groesbeck, T.D. 1994, Ph.D. thesis, Caltech Harvey, P. M., Campbell, M. F., Hoffmann, W. F., Thronson, H. A., Jr., & Gatley, I. 1979, ApJ, 229, 990 Hatchell, J., Thompson, M.A., Millar, T.J., MacDonald, & G.H. 1998, A&AS, 133, 29 Henriksen, R., Andr´e, P., & Bontemps, S. 1997, A&A, 323, 549 Helmich, F.P. & van Dishoeck, E.F. 1997, A&AS, 124, 205 Hillenbrand, L. A. 1997, AJ, 113, 1733 Hillenbrand, L. A., & Hartmann, L. W. 1998, ApJ, 492, 540 Hogerheijde, M.R., Jansen, D.J. & van Dishoeck, E.F. 1995, A&A, 294, 792 Hurt, R. L., Barsony, M., & Wootten, A. 1996, ApJ, 456, 686 Jansen, D.J., van Dishoeck, E.F., & Black, J.H. 1994, A&A, 282, 605 Jansen, D.J., Spaans, M., Hogerheijde, M.R., van Dishoeck, E.F. 1995, A&A, 303, 541 Schoier, F.L., Jørgensen, J.K., van Dishoeck, E.F., & Blake, G.A. 2003, in Chemistry as a Diagnostic of Star Formation, ed. C.L. Curry & M. Fich, in press Schreyer, K., Henning, Th., van der Tak, F.F.S., Boonman, A.M.S., & van Dishoeck, E.F. 2002, A&A, 394, 561 Serabyn, E. & Weisstein, E.W. 1995, ApJ, 451, 238 Shirley, Y.L., Evans, N.J.II, Rawlings, J.M.C. 2002, ApJ, 575, 337 Shu, F. 1977, ApJ, 214, 488 Tielens, A.G.G.M. & Charnley, S.B. 1997, Origins of Life and Evolution of the Biosphere, 27, 23 Tilanus, R.P.J., van der Werf, P.P., & Israel, F.P. 2000, in ASP Conf. Ser. Imaging at Radio through Submillimeter Wavelengths, ed. J.G. Mangum & S.J.E. Radford, 217, 177 van der Tak, F.F.S., van Dishoeck, E.F., & Caselli, P. 2000b, A&A, 361, 327 van der Tak, F.F.S., van Dishoeck, E.F. Evans, N.J.II, & Blake, G.A. 2000a, ApJ, 537, 283 van der Tak, F.F.S., Boonman, A.M.S., Braakman, R., van Dishoeck, E.F., 2003, A&A, in press van Dishoeck, E.F. & Blake, G.A. 1998, ARAA, 36, 317 van Dishoeck, E.F. & Helmich, F.P. 1996, A&A, 315, L177 van Dishoeck, E.F., Jansen, D.J., & Phillips, T.G. 1993, A&A, 279, Johnson, J. J., Gehrz, R. D., Jones, T. J., Hackwell, J. A., & Grasdalen, 541 van Dishoeck, E.F., Blake, G.A., Jansen, D.J. & Groesbeck, T.D. 1995, ApJ, 447, 760 Wilson, T.L. & Rood, R. 1994, ARAA, 32, 191 Wolstencroft, R. D., Scarrott, S. M., Warren-Smith, R. F., Walker, H. J., Reipurth, B., & Savage, A. 1986, MNRAS, 218, 1 Yu, K. C., Bally, J., & Devine, D. 1997, ApJ, 485, L45 Yu, K. C., Billawala, Y., Smith, M. D., Bally, J., & Butner, H. M. 2000, AJ, 120, 1974 G. L. 1990, AJ, 100, 518 Johnstone, D. & Bally, J. 1999, ApJ, 510, L49 Johnstone, D., Fich, M., Mitchell, G. F., Moriarty-Schieven, G. 2001, ApJ, 559, 307 Jørgensen, J.K., Schoier, F.L., & van Dishoeck, E.F. 2002, A&A, 389, 908 Jørgensen, J.K., Schoier, F.L., & van Dishoeck, E.F. 2003, in Chemistry as a Diagnostic of Star Formation, ed. C.L. Curry & M. Fich, in press Keane, J.V., Boonman, A.M.S., Tielens, A.G.G.M., & van Dishoeck, E.F. 2001a, A&A, 379, 588 Keane, J.V., Tielens, A.G.G.M., Boogert, A.C.A., Schutte, W.A., & Whittet, D.C.B. 2001b, A&A, 376, 254 Lahuis, F. & van Dishoeck, E.F. 2000, A&A, 355, L699 Leen T.M. & Graff, M.M. 1988, ApJ, 325, 411 Lees, R. & Haque, S. 1974, Canadian J. of Physics, 52, 2250 Li, D., Goldsmith, P.F., & Menten, K. 2003, ApJ, 587, 262 Lis, D. C., Serabyn, E., Keene, Jocelyn, Dowell, C. D., Benford, D. J., Phillips, T. G., Hunter, T. R., & Wang, N. 1998, ApJ, L299 MacDonald, G.H., Gibb, A.G., Habing, R.J., & Millar, T.J. 1996, A&AS, 119, 333 Mangum, J. G. & Wootten, A. 1993, ApJS, 89, 123 Mangum, J. G., Wootten, A., & Barsony, M. 1999, ApJ, 526, 845 Mezger, P. G., Zylka, R., & Wink, J. E. 1990, A&A, 228, 95 Mitchell, G., Johnstone, D., Moriarty-Schieven, G., Fich, M., & Tothill, N. F. H. 2001, ApJ, 513, 139 Motte, F., Andr´e, P., Ward-Thompson, D., & Bontemps, S. 2001, A&A, 372, L41 Myers, P.C. 1999, in The Origin of Stars and Planetary Systems, ed. C.J. Lada & N.D. Kylafis, Kluwer Academic Publishers, 67 Reipurth, B. & Aspin, C. 1997, AJ, 114, 2700 Reipurth, B., Bally, J., & Devine, D. 1997, AJ, 114, 2708 Reipurth, B., Rodriguez, L. F., & Chini, R. 1999, AJ, 118, 983 Rodgers S.D. & Charnley, S.B. 2001, ApJ, 546, 324 Schilke, P., Walmsley, C. M., Pineau Des Forets, G., Roueff, E., Flower, D. R., & Guilloteau, S 1992, A&A, 256, 595 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 13 Table 1. Source List and Sub-millimeter Diagnostics Name R.A.[a] (J2000) Dec.[a] (J2000) Td (K) Menv (M⊙) L (L⊙) Object MMS6 − OMC3 5 : 35 : 23.5 −05 : 01 : 32 MMS9 − OMC3 5 : 35 : 26.2 −05 : 05 : 46 MMS7 − OMC3 5 : 35 : 26.4 −05 : 03 : 59 26 − 50 5 : 35 : 26.7 −05 : 09 : 59 15 − 40 FIR4 − OMC2 NGC2071IR 5 : 47 : 04.4 +00 : 21 : 49 20 − 50 15 20 36 10 8 34 30 1.2 − 60 Class0 0.6 − 90 Class0 Class0 Class0/I Class0/I 76 400 500 PDR1 − OMC1 5 : 35 : 25.3 −05 : 24 : 34 SK1 − OMC3 5 : 35 : 17.0 −05 : 06 : 03 60 10 PDR Shock a Positions are taken from the sub-millimter maps of the ISF (Johnstone & Bally 1999) and Orion B North (Johnstone et al. 2001). 14 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Table 2. Calendar of Observations and Conditions Date Receiver νupper (GHz) ∆V νlower (GHz) (km/s) Tsys (K) Lines 1999 January RxA3 1999 September RxB3 1999 September RxB3 1999 September RxB3 2000 December RxA3 2000 December RxA3 2000 December RxA3 2000 December RxB3 2000 December RxB3 2000 December RxB3 2000 December RxB3 218.35 226.48 330.53 338.53 338.53 346.53 354.40 362.40 233.00 241.00 233.75 241.75 236.94 244.94 337.25 345.25 340.45 348.45 345.67 353.67 354.60 362.60 0.2 0.5 0.5 0.5 0.4 0.4 0.4 0.5 0.5 0.5 0.5 400 − 700 1200 − 1700 1200 − 1700 1200 − 1700 350 − 400 350 − 400 375 − 385 530 − 550, 710 − 740 700 − 800 490 − 510 590 − 620 H2CO, CH3OH 13CO, CH3OH CH3OH, SO HCN, HNC C34S CH3OH CS C17O, C34S CN, HN13C CO HCN, HNC, H2CO Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 15 Table 3. Observed Spectra and Gaussian Fit Parameters Source Species MMS6-OMC3 CO 13CO C17O ν (GHz) 345.796 330.588 337.061 J 3 -- 2 3 -- 2 3 -- 2 W (K km s−1) 93.00 4.22 40.70 4.78 ∆V (km s−1) -- 0.57 2.24 1.08 H2CO 218.222 303 -- 202 H2CO H2CO 218.475 362.736 322 -- 221 505 -- 404 CH3OH 218.440 CH3OH 241.700 CH3OH 241.767 CH3OH 241.791 CH3OH 241.904 CH3OH 338.345 CH3OH 338.404 338.409 CH3OH 338.583 CH3OH 338.615 CH3OH 338.722 42E-31E 50E-40E 5−1E-4−1E 50A+-40A+ 5±2E-4±2E 7−1E-6−1E 76E-66E 70A+-60A+ 73E-63E 71E-61E 7±2E-6±2E CN HNC HCN H13CN CS C34S C34S SO 340.248 362.630 354.505 345.340 244.936 241.016 337.396 346.528 3-2 4-3 4-3 4-3 5-4 5-4 7-6 8,9-7,8 2.81 2.52 0.53 1.80 1.76 0.44 0.18 0.95 1.27 0.34 0.84 0.34 0.96 0.51 0.46 0.45 2.95 4.56 9.65 0.49 2.31 2.23 0.65 0.54 1.28 0.88 2.40 1.73 0.91 2.92 1.41 0.82 1.63 1.65 2.80 1.56 0.99 3.45 4.81 2.05 1.69 1.41 1.06 3.81 2.05 0.93 2.85 1.58 1.19 1.23 TMB (K) -- 7.00 17.1 4.15 3.00 0.99 0.29 1.85 0.57 0.29 0.21 0.55 0.72 0.11 0.50 0.32 0.26 0.10 0.21 0.25 1.97 4.04 2.38 0.23 2.34 0.74 0.39 0.43 0.98 16 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Table 3. Observed Spectra and Gaussian fit parameters (cont'd) W (K km s−1) 2.03 0.10 ∆V (km s−1) 1.38 0.55 Source Species MMS7-OMC3 MMS9-OMC3 H2CO H2CO CO 13CO C17O ν (GHz) 218.222 218.475 345.796 330.588 337.061 J 303 -- 202 322 -- 221 3 -- 2 3 -- 2 3 -- 2 H2CO 218.222 303 -- 202 H2CO H2CO 218.475 362.736 322 -- 221 505 -- 404 CH3OH 218.440 CH3OH 241.700 CH3OH 241.767 CH3OH 241.791 CH3OH 338.345 42E-31E 50E-40E 5−1E-4−1E 50A+-40A+ 7−1E-6−1E CN HNC HCN 340.248 362.630 354.505 H13CN 345.340 CS 244.936 3-2 4-3 4-3 4-3 5-4 225.00 59.10 2.88 1.42 1.65 0.28 0.94 0.23 0.14 0.51 0.66 0.30 4.65 2.97 1.05 3.57 0.41 2.73 TMB (K) 1.39 0.18 -- 23.2 2.10 1.99 0.73 0.36 0.78 0.22 0.16 0.40 0.72 0.50 2.46 2.85 1.21 0.98 0.07 -- 2.40 1.29 0.67 2.11 0.74 1.12 0.98 0.82 1.21 0.86 0.55 1.78 0.98 0.82 3.40 5.23 1.50 1.71 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 17 Table 3. Observed Spectra and Gaussian fit parameters (cont'd) J 3 -- 2 3 -- 2 3 -- 2 W (K km s−1) 442.00 76.20 24.70 1.27 2.70 ∆V (km s−1) -- 2.57 12.5 0.90 1.65 Source Species FIR4-OMC2 CO 13CO ν (GHz) 345.796 330.588 C17O 337.061 H2CO 218.222 H2CO 218.475 H2CO 362.736 303 -- 202 322 -- 221 505 -- 404 TMB (K) -- 27.8 1.85 1.32 1.54 6.21 2.35 1.85 0.73 4.09 4.50 3.29 1.38 1.65 0.95 3.44 1.60 4.11 1.81 1.43 1.10 4.12 3.85 4.14 4.46 2.59 1.69 2.20 3.00 0.45 0.32 1.67 1.69 2.12 4.03 3.79 3.73 1.67 5.60 1.58 15.0 0.97 0.48 0.24 0.34 0.97 1.10 9.82 15.70 2.96 4.49 5.24 29.3 3.43 7.49 2.39 5.36 4.73 9.33 5.93 10.10 2.35 6.65 5.86 21.70 5.31 25.20 1.58 13.30 5.37 11.30 1.44 1.96 2.24 10.20 9.44 11.60 29.20 7.48 15.50 9.63 11.50 163.00 8.66 0.72 1.31 2.64 <9.00 6.70 1.49 6.28 1.50 5.81 1.20 6.02 0.98 5.10 1.36 5.33 1.29 5.50 1.36 5.25 1.54 5.68 1.34 5.28 1.21 5.30 0.57 7.37 2.29 3.54 3.03 5.70 1.26 5.68 4.19 2.70 7.23 1.88 8.68 1.62 6.83 10.2 8.43 1.40 5.18 7.39 6.00 6.00 CH3OH 218.440 42E-31E CH3OH 241.700 50E-40E CH3OH 241.767 5−1E-4−1E CH3OH 241.791 50A+-40A+ CH3OH 241.904 5±2E-4±2E CH3OH 338.345 7−1E-6−1E CH3OH 338.404 76E-66E CH3OH 338.513 74A±-64A± CH3OH 338.542 73A±-64A± CH3OH 338.583 73E-63E CH3OH 338.615 71E-61E CH3OH 338.640 CH3OH 338.722 72A+-620A+ 7±2E-6±2E CN 340.248 HNC 362.630 HCN H13CN 354.505 345.340 C34S C34S SO 241.016 337.396 346.528 3-2 4-3 4-3 4-3 5-4 7-6 8,9-7,8 18 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Table 3. Observed Spectra and Gaussian fit parameters (cont'd) Source Species NGC2071IR CO C17O ν (GHz) 345.796 337.061 J 3 -- 2 3 -- 2 W (K km s−1) 770.00 2.96 4.25 ∆V (km s−1) -- 1.73 7.31 H2CO 218.222 303 -- 202 H2CO 218.475 322 -- 221 H2CO 362.736 505 -- 404 CH3OH 218.440 CH3OH 241.700 42E-31E 50E-40E CH3OH 241.767 5−1E-4−1E CH3OH 241.791 50A+-40A+ CH3OH 241.879 CH3OH 241.888 CH3OH 241.904 51E-41E 52A+-42A+ 5±2E-4±2E CN 340.248 HNC 362.630 HN13C HCN 345.340 354.505 H13CN 345.340 CS 244.936 C34S C34S 241.016 337.396 3-2 4-3 4-3 4-3 4-3 5-4 5-4 7-6 12.00 14.00 3.75 3.67 10.90 10.70 3.12 2.39 5.36 6.51 6.18 4.93 8.12 1.45 0.74 1.09 2.98 14.50 14.50 12.30 10.70 1.30 52.10 31.20 5.51 10.30 25.50 0.59 3.86 0.53 3.57 2.21 9.12 2.95 16.3 2.07 7.50 2.30 1.36 5.33 2.38 12.1 1.80 7.25 1.88 2.24 1.52 4.66 2.61 8.63 1.92 6.44 3.89 3.95 14.3 10.1 1.87 6.29 0.85 3.45 3.70 3.74 TMB (K) -- 1.60 0.55 5.10 1.45 1.19 0.21 4.93 1.34 1.28 1.65 0.94 2.56 0.48 2.58 1.05 0.73 0.31 0.67 0.60 5.20 1.58 6.00 1.56 0.31 12.4 2.05 0.51 5.17 3.81 0.66 1.05 0.13 0.89 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 19 Table 3. Observed Spectra and Gaussian fit parameters (cont'd) Source Species PDR1-OMC1 CO 13CO ν (GHz) 345.796 330.588 C17O 337.061 J 3 -- 2 3 -- 2 3 -- 2 W (K km s−1) 535.00 129.00 90.10 10.20 ∆V (km s−1) -- 1.39 1.28 1.38 H2CO 218.222 303 -- 202 H2CO 218.475 322 -- 221 H2CO 362.736 505 -- 404 CH3OH 241.700 CH3OH 241.767 CH3OH 241.791 CH3OH 241.904 CH3OH 338.345 CH3OH 338.404 CH3OH 338.722 50E-40E 5−1E-4−1E 50A+-40A+ 5±2E-4±2E 7−1E-6−1E 76E-66E 7±2E-6±2E CN HNC HCN H13CN C34S C34S SO CO 13CO 340.248 362.630 354.505 345.340 241.016 337.396 346.528 345.796 330.588 C17O 337.061 H2CO H2CO CN HNC HCN 218.222 218.475 340.248 362.630 354.505 3-2 4-3 4-3 4-3 5-4 7-6 8,9-7,8 3 -- 2 3 -- 2 3 -- 2 303 -- 202 322 -- 221 3-2 4-3 4-3 0.38 5.11 0.23 1.76 3.80 0.33 0.89 1.20 0.35 1.14 1.85 0.74 23.80 3.83 41.70 2.11 7.32 5.74 4.82 404.00 5.33 69.50 3.31 0.57 <0.20 0.94 0.23 0.43 1.74 0.49 1.88 0.49 1.32 1.56 1.83 1.87 1.83 1.71 1.43 1.86 2.03 2.18 1.57 2.39 1.47 1.65 1.28 1.64 -- 1.12 2.49 1.60 1.45 0.49 3.49 1.25 1.24 14.6 SK1-OMC3 TMB (K) -- 87.0 66.0 6.94 0.72 2.55 0.43 1.25 2.29 0.17 0.45 0.61 0.19 0.75 0.94 0.34 10.3 2.29 16.4 1.35 4.16 4.21 2.77 -- 4.49 26.2 1.94 0.37 0.37 0.25 0.17 0.32 0.11 20 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Table 4. Physical Conditions Discerned from Carbon- Monoxide Source MMS6 MMS9 FIR4 NGC2071 PDR1 SK1 CO Tpeak (K) 23. 28. 42. 67. 165 52. T ∆V (K km s−1) 93. 225 442 770 535 404 13CO Tpeak (K) 21. 23. 28. -- 105 26. T ∆V (K km s−1) 45. 59. 101 -- 220 75. C17O Tpeak (K) 3.6 2.1 2.7 2.2 7.2 1.9 T ∆V N(CO) (K km s−1) (1018 cm−2) 4.8 3.0 4.1 7.3 10. 3.3 4.6 2.7 3.4 5.7 11. 2.8 τmod 13CO 2.1 1.3 1.8 1.5 2.0 1.5 τobs/τmod 2.4 1.5 1.3 1.0 1.1 0.9 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 21 Table 5. Measured Total Line Intensity within the 850 µm Band and Contribution to Continuum Flux Source CO Line Strength[a] Total Line Strength[b] Contamination N(H2) [CO]/[H2] (K km s−1) (K km s−1) MMS6 MMS9 MMS7 FIR4 NGC2071 PDR1 SK1 143 287 547 777 765 483 170 301 934 945[c] 853 487 % 1.5 8.2 8.0 8.3 11.1 63. (1022 cm−2) 104 14 22 47 62 10 6 (10−4) 0.04 0.19 0.07 0.09 1.10 0.47 a CO line strengths include contribution from CO, 13CO, and C17O J = 3 -- 2 transitions. b Total line strength includes all measured lines within the 850µm passband. c The total line strength for NGC2071 does not include contribu- tions from methanol. Comparison with FIR4 suggests that an ad- ditional few percent of line contamination might be produced by these unobserved lines. 22 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Table 6. Physical Conditions and Abundances Discerned from Formaldehyde Source C[a] R[b] 33 R[b] 53 n (106 cm−3) N(H2CO) (1013 cm−2) [H2CO]/[CO] [H2CO]/[H2] (10−5) (10−10) MMS6 MMS6 MMS6 MMS9 MMS7 FIR4 FIR4 FIR4 NGC2071 NGC2071 NGC2071 PDR1 PDR1 PDR1 n b t t t n b t n b t n b t 4.0 4.0 8.0 2.2 2.0 12.0 20.0 32.0 15.0 17.0 32.0 0.8 6.0 6.8 1.7 0.8 9.4 4.3 0.6 0.8 1.6 0.9 6.5 5.1 6.8 T (K) < 50 < 50 > 5 > 5 0.6 0.6 1 − 3 1 − 3 10 0.3 30 1 − 3 > 7 - < 40 - 3.3 3.4 3.2 3.8 1.8 2.8 0.7 1.8 0.9 0.8 - 0.7 90 80 90 70 0.5 − 2 2 − 10 1 − 3 1 − 3 > 200 120 - 0.5 − 2 a The letter designations refer to the velocity components with nar- row (n), broad (b), and total (t). b R33 is the ratio of the 303 -- 202 transition versus the 322 -- 221 transi- tion and R53 is the ratio of the 505 -- 404 transition versus the 303 -- 202 transition. Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 23 Table 7. Physical Conditions and Abundances Discerned from Methanol Source C[a] MMS6 MMS9 FIR4 FIR4 FIR4 NGC2071 PDR1 t t n b t t t T (K) 50 < 30 70 80 60 > 100 n (107 cm−3) N(CH3OH) (1013 cm−2) [CH3OH]/[CO] [CH3OH]/[H2] (10−5) (10−10) 1.0 0.5 > 10 5 0.5 5.0 3.0 1.5 30 95 125 30 6.0 0.7 0.6 37 5.6 0.6 0.3 1.1 6.1 19 25 5.1 5.8 a The letter designations refer to the velocity components with nar- row (n), broad (b), and total (t). 24 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Table 8. Abundances Discerned from HCN, HNC, and CN Source [HCN]/[HNC] N(HCN) (1014 cm−2) MMS6 MMS9 FIR4 NGC2071 PDR1 6.3 5.4 11. 6.0 33. 0.6 1.3 8.0 6.4 4.0 N(CN) [HCN]/[CN] [HCN]/[H2] [CN(total)]/[H2] (1014 cm−2) <2.0 2.8 6.0 19. 2.0 (100) >0.3 0.5 1.3 0.3 2.0 (10−10) 0.6 9.3 16. 10. 40. (10−10) <2.4 32. 30. 41. 72. Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 25 Table 9. Physical Conditions and Abundances Discerned from CS and SO Source MMS6 MMS9 FIR4 NGC2071 PDR1 N(CS) (1013 cm−2) [CS]/[H2] (10−10) 3.5 1.5 14. 19. 29. 0.3 1.1 2.9 3.1 29. TCS (K) >50 >100 > 50 > 50 N(SO) (1014 cm−2) <2.6 <4.0 [SO]/[H2] (10−10) <2.5 <8.2 [SO]/[CS] (100) <7.4 <2.9 0.8 8.0 0.3 26 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion −5:00 05 10 15 20 25 ) 0 0 0 2 ( . c e D 50 40 30 20 10 5:35:00 50 40 R.A. (2000) Fig. 1. The Integral Shaped Filament (ISF) at the north end of the Orion A molecular cloud, observed in dust continuum emission at 850 microns. The grey-scale ranges from 0 to 5 Jys with a logarithmic stretch. The positions of the six sources located in the ISF are indicated by arrows pointing at the peak continuum flux locations. Details of the sources are found in Table 1 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 27 Fig. 2. The CO J = 3 -- 2 line profiles. Note the self-absorption in all sources except PDR1, and the extended line wings (dashed lines) in all sources except MMS6, and PDR1. 28 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Fig. 3. A selection of methanol lines observable with a single spectral setting, along with an SO NJ =89 → 78 line. Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 29 Fig. 4. The HCN and HNC J = 4 -- 3 transitions. 30 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Fig. 5. Determination of the equilibrium physical properties of molecular gas from formaldehyde line ratios. The top left panel presents the expected line ratio for the 303 − 202 and 322 − 221 transitions as functions of temperature and density (solid con- tours) against the expected line ratio for the 505 − 404 versus the 303 − 202 transitions (dashed contours). The remaining panels show the range of physical properties obtainable for the sources in the present study, with the solid lines denoting the 303 − 202 versus 322 − 221 ratio and the dashed lines providing the loca- tion of the 505 − 404 versus 303 − 202 ratio. Two sets of dashed lines are shown to account for possible beam dilution (see text). Also, the dotted lines provide an indication of the range of un- certainty on the calculations. Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 31 32 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion (a) (b) (c) Fig. 6. Rotation diagrams for methanol. The input data are taken from the result of RADEX equilibrium calculations at a temperature of 100 K. (a) Calculation with n = 107 cm−3. The slope of the data at high energies is significantly affected by the sub-thermal excitation, and the derived temperature from the rotation diagram does not correspond to the physical con- dition. The low energy lines are much closer to being ther- malized, however, and the column density derived is close to the input column density of 1014 cm−3. (b) Calculation with n = 108 cm−3. The slope of the data at high energies is still affected by the sub-thermal excitation, and the derived temper- ature from the rotation diagram continues to underestimate the physical condition. (c) Calculation with n = 109 cm−3. At this density the rotation diagram produces a reasonable fit to the input physical conditions. Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 33 Fig. 7. Chi-squared confidence fits to the physical parameters of sources using the observed methanol lines and the radiative transfer code RADEX. The sources are (Top Left) FIR4 narrow component, (Top Center) FIR4 broad component, (Top Right) NGC2071, (Bottom Left) MMS6, (Bottom Center) MMS9, (Bottom Right) PDR1. Note that there is often a strong degen- eracy between the best fit temperature and density due to the lack of thermalization of the higher energy transitions. 34 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Fig. 8. The range in abundances of the various molecular species investigated against T(H2CO). The abundances have been derived assuming a constant temperature and density en- velope. Diamonds represent sub-millimeter sources, stars rep- resent infrared sources, and the cross represents the PDR, with symbol size approximating the order unity uncertainty ex- pected in the measurements. For all protostellar sources the CO abundance is depleted by about an order of magnitude. For all species the infrared sources show a remarkable similarity in abundances, however, in the case of methanol the very broad > 10 km s−1 component from FIR4 has not been included. For most molecules the sub-millimeter sources have lower abun- dances while the PDR has the highest abundance. Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion 35 where possible. However, as long as the line is not significantly broadened, the CO contamination should be minor. Freeze-out of the CO onto dust grains decreases the importance of CO contamination for the densest and coldest regions. Appendix A: Contamination of sub-millimeter continuum by molecular lines Comparison of 850 µm continuum against CO 3 -- 2 measure- ments in the spiral arms of external disk galaxies has revealed that as much as 50% of the emission within the SCUBA band- pass is contamination due to this single line (Tilanus, van der Werf, & Israel 2000). This result is not surprising since the en- ergy radiated in an optically thin CO j-i line by a column of gas with gas temperature Tg and number density n is given by S ji = h ν ji 4π ! A ji N(CO) Xj(Tg, n) erg s−1cm−2sr−1, (A.1) where the column of CO is N(CO) = x(CO) N(H2), and the fractional population of the jth level is X j. For the CO 3 -- 2 line, ν32 = 3.458 × 1011 s−1 and A32 = 2.48 × 10−6 s−1. Alternatively, for optically thin dust the energy radiated across a passband centered at νd and with an effective width ∆ν is S d = Bνd(Td) NH2 µ mH κνd ∆ν erg s−1cm−2sr−1, (A.2) where κνd is the emissivity of the dust at νd and µ is the mean molecular weight of the gas. Since the dust radiation in molec- ular clouds is usually near the Rayleigh-Jeans limit, one can replace the Planck function Bν(Td) with 2 k Td fν(Td)/λ2 ν, where fν(Td) accounts for the difference between the Planck law and the Rayleigh-Jeans limit at frequency ν. Thus, in the optically thin limit the ratio of the line to con- tinuum is independent of the column of hydrogen, depend- ing only on the temperature and density within the region, the abundance of CO in the gas phase, and the dust emissivity. R = fνd (Td) x(CO) X3(Tg, n).(A.3) 1 8π hν ji kTd! A ji ∆ν!  λ2 νd µ mH κνd Computing the ratio of the CO 3 -- 2 line against the JCMT 850 µm continuum band, where ∆ν = 3 × 1010 s−1, yields R850 = 28 (cid:18) Td 20 K(cid:19) −1 κ850 0.02cm2/g!−1 x(CO) 1 × 10−4! f850(Td) X3(Tg, n).(A.4) The ratio depends strongly on the population of the upper CO energy level X3 which is a function of both the gas temper- ature and the density. Tables A.1 and A.2 show X3 and R850 for a range of internal gas conditions, assuming that the gas dust temperatures are equal. The values shown in Table A.1 only apply in the optically thin limit for the CO 3 -- 2 transition (N(H2) < 1021 cm−3). At higher columns, as are expected deep within molecular clouds, the continuum emission continues to increase linearly while the line emission is severely damped and the contamination is expected to be less influential. For ex- tragalactic studies, where the spatial resolution does not permit the separation of thick and thin portions of the cloud, it is not surprising that the CO emission produces a significant contam- ination. The degree to which the CO line contamination affects sub-millimeter continuum measurements at high optical depth within molecular clouds, as investigated in this paper, depends sensitively on the width of the CO 3 -- 2 line and the strength of the extended line wings, and should be measured independently 36 Doug Johnstone et al.: Astrochemistry of Sub-Millimeter Sources in Orion Table A.1. Fractional population of the 3rd CO energy level for different physical conditions Temperature (K) 10 20 30 50 100 500 Density (cm−3) 103 2.51e-3 1.23e-2 2.21e-2 3.66e-2 5.61e-2 8.44e-2 104 2.23e-2 9.02e-2 1.41e-1 1.95e-1 2.38e-1 2.53e-1 105 5.51e-2 1.68e-1 2.13e-1 2.24e-1 1.91e-1 1.13e-1 Table A.2. Ratio of CO line strength to 850 µm continuum with SCUBA for different physical conditions, assuming [CO]/[H2] = 10−4 and κ850 = 0.02 cm2 g−1. Temperature (K) 10 20 30 50 100 500 Density (cm−3) 103 0.4 0.8 0.6 0.5 0.4 0.1 104 3.3 4.0 3.6 2.6 1.5 0.3 105 8.2 7.5 5.4 3.0 1.2 0.1
astro-ph/0208504
1
0208
2002-08-28T14:16:57
Similarities and Differences between Relativistic Electron-Photon Cascades Developed in Matter, Photon Gas and Magnetic Field
[ "astro-ph" ]
We investigate properties of astrophysical electromagnetic cascades developed in matter, photon gas and magnetic fields, and discuss similarities and differences between characteristics of electron-photon showers developed in these 3 substances.
astro-ph
astro-ph
Similarities and Differences between Relativistic Electron-Photon Cascades Developed in Matter, Photon Gas and Magnetic Field F.A. Aharonian a,1 A.V. Plyasheshnikov b,2 aMPI fur Kernphysik, Sauperfercheckweg 1, D-69117 Heidelberg, Germany bAltai State University, Dimitrov street 66, Barnaul, 656099, Russia Abstract We investigate the properties of astrophysical electromagnetic cascades in matter, photon gas and magnetic fields, and discuss similarities and differences between characteristics of electron-photon showers developed in these 3 substances. We ap- ply the same computational technique based on solution of the adjoint cascade equations to all 3 types of cascades, and present precise numerical calculations of cascade curves and broad-band energy spectra of secondary electrons and photons at different penetration lengths. Key words: Electromagnetic Cascades, High Energy Processes, Gamma-Ray Sources 1 Introduction Relativistic electrons – directly accelerated, or being secondary products of various hadronic processes – may result in copious γ-ray production caused by interactions with ambient targets in forms of gas (plasma), radiation and magnetic fields. In different astrophysical environments γ-ray production may proceed with high efficiency through bremsstrahlung, inverse Compton scat- tering and synchrotron (and/or curvature) radiation, respectively. Generally, γ-ray production is effective when the cooling time that characterizes the rate of the process does not significantly exceed (i) the source (accelerator) age (ii) 1 E-mail: [email protected] 2 E-mail: [email protected] Preprint submitted to Elsevier Preprint 25 November 2017 the characteristic time of non-radiative losses caused by adiabatic expansion of the source or particle escape and (iii) the cooling time of competing radia- tion mechanisms that result in low-energy photons outside the γ-ray domain. As long as the charged particles are effectively confined in the γ-ray produc- tion region, at some circumstances these condition could be fulfilled even in environments with a relatively low gas and photon densities or weak mag- netic field. More specifically, the γ-ray production efficiency could be close to 1 even when trad ≫ R/c (R is the characteristic linear size of the pro- duction region, c is the speed of light). In such cases the secondary γ-rays escape the source without significant internal absorption. Each of the above mentioned γ-ray production mechanisms has its major "counterpart" - γ-ray absorption mechanism of same electromagnetic origin resulting in electron- positron pair production in matter (the counterpart of bremsstrahlung), in photon gas (the counterpart of inverse Compton scattering), and in magnetic field (the counterpart of synchrotron radiation). The γ-ray production mecha- nisms and their absorption counterparts have similar cross-sections, therefore the condition for radiation trad ≥ R/c generally implies small optical depth for the corresponding γ-ray absorption mechanism, τabs ≤ 1. But in many astro- physical scenarios, in particular in compact galactic and extragalactic objects with favorable conditions for particle acceleration, the radiation processes pro- ceed so fast that trad ≥ R/c. At these conditions the internal γ-ray absorption becomes unavoidable. If the γ-ray production and absorption processes oc- cur in relativistic regime, namely when (i) Eγ,e ≥ 103mec2 in the hydrogen ec4 in photon gas (often called Klein-Nishina regime; ǫ is gas, (ii) Eγ,eǫ ≫ m2 the average energy of the target photons), or (iii) (Eγ,e/mec2)(H/Hcr) ≫ 1 in the magnetic field (often called quantum regime; Hcr ≃ 4.4 × 1013 G is the so-called critical strength of the magnetic field), the problem cannot be reduced to a simple absorption effect. In this regime, the secondary electrons produce new generation of high energy γ-rays, these photons again produce electron-positron pairs, so the electromagnetic cascade develops. The characteristics of electromagnetic cascades in matter have been compre- hensively studied, basically in the context of interactions of cosmic rays with the Earth's atmosphere (see e.g. [1]) as well as for calculations of performance of detectors of high energy particles (e.g. [2]). The theory of electromagnetic cascades in matter can be applied to some sources of high energy cosmic radiation, in particular to the "hidden source" scenarios like massive black holes in centers of AGN or young pulsars inside the dense shells of recent supernovae explosions (see e.g. [3]). Within another, so-called "beam dump" models (see e.g. Ref. [3,5]) applied to X-ray binaries, protons accelerated by the compact object (a neutron star or black hole), hit the atmosphere of the normal companion star, and thus result in production of high energy neu- trinos and γ-rays [4]. In such objects, the thickness of the surrounding gas can significantly exceed 100 g/cm2, thus the protons produced in the central source would initiate (through production of high energy γ-rays and electrons) 2 electromagnetic showers. These sources perhaps represent the "best hope" of neutrino astronomy, but they are generally considered as less attractive targets for gamma-ray astronomy. However, the γ-ray emission in these objects is not fully suppressed. The recycled radiation with spectral features determined by the thickness ("grammage") of the gas shell, should be seen in γ-rays in any case, unless the synchrotron radiation of secondary electrons dominates over the bremsstrahlung losses and channels the main fraction of the nonthermal energy into the sub gamma-ray domains. The development of electromagnetic cascades in photon gas and magnetic fields is a more common phenomenon in astrophysics. In photon fields such cascades can be created on almost all astronomical scales, from compact ob- jects like accreting black holes [6–10], fireballs in gamma-ray bursts [11] and sub-pc jets of blazars [12] to large-scale (up to ≥ 100 kpc) AGN jets [13,14] and ≥ 1 Mpc size clusters of galaxies [15]. Electromagnetic cascades in the inter- galactic medium lead to formation of huge (≥ 10 Mpc) nonthermal structures like hypothetical electron-positron pair halos [16]. Finally, there is little doubt that the entire Universe is a scene of continuous creation and development of electromagnetic cascades. All γ-rays of energy ≥ a few GeV emitted by astro- physical objects have a similar fate. Sooner or later they terminate on Hubble scales due to interactions with the diffuse extragalactic background. Since the energy density of 2.7 K CMBR significantly exceeds the energy density of intergalactic magnetic fields, these interactions initiate electron-photon cas- cades [17]. The superposition of contributions of γ-rays from these cascades should constitute a significant fraction of the observed diffuse extragalactic background. Bonometto and Rees [18] perhaps where the first who realized the astrophys- ical importance of development of electron-photon cascades supported by γ-γ pair-production and inverse Compton scattering in dense photon fields. When the so called compactness parameter [19] l = (L/R)(σT/mec3) (L is the lu- minosity and R is the radius of the source) is less than 10, then the cascade is developed in the linear regime, i.e. when the soft radiation produced by cascade electrons do not have a significant feedback effect on the cascade development. In many cases, including the cascade development in compact objects, this approximation works quite well. The first quantitative study of characteristic of linear cascades in photon fields has been performed using the method of Monte Carlo simulations [20]. Gen- erally, the kinetic equations for the cascade particles can be solved only nu- merically. However, with some simplifications it is possible to derive useful analytical approximations [7,8] which help to understand the features of the steady-state solutions for cascades in photon fields. The cascade development in the magnetic field is a key element to understand 3 the physics of pulsar magnetospheres [21,22], therefore it is generally treated as a process associated with very strong magnetic fields. However, such cascades could be triggered in many other (at first glance unusual) sites like the Earth's geomagnetic field [23–25], accretion disks of massive black holes [26], etc. In general, the pair cascades in magnetic fields are effective when the product of the particle (photon or electron) energy and the strength of the field becomes close to the "quantum threshold" of about Hcritmec2 ≃ 2 × 107 TeV · G, unless we assume a specific, regular field configuration. An approximate approach, similar to the so-called approximation A of cascade development in matter [1], has been recently applied by Akhiezer et al. [27]. Although this theory quite satisfactorily describes the basic features of photon-electron showers, it does not provide an adequate accuracy for quantitative description of the cascade characteristics [24]. Note that in both studies approximate cross-sections for magnetic bremsstrahlung and magnetic pair-production have been used, thus reducing the validity of the results to the limit EH ≫ 107 TeV · G (E is the minimum energy of secondary particles being under consideration). As long as we are interested in the one-dimensional cascade development (which seems to be quite sufficient for many astrophysical purposes), all 3 types of cascades can be described by the same integro-differential equations like the ones derived by Landau and Rumer [28], but in each case specifying the cross-sections of relevant interaction processes. The solution of these equa- tions in a broad range of energies is however not a trivial task. In this paper we present the results of our recent study of cascade characteristics in 3 sub- stances - matter, photon gas and magnetic field - with emphasis on the analysis of similarities and differences between these 3 types of cascades. For quanti- tative studies of these characteristics we have chosen the so-called technique of adjoint cascade equations. Although this work has been initially motivated by methodological and pedagogical objectives, some results are rather original and may present practical interest in certain areas of high energy astrophysics. 2 Technique of adjoint cascade equations The results of this study are based on numerical solutions of the so-called adjoint cascade equations. The potential of this computational technique, in particular its possible applications to different problems of cosmic ray physics has been comprehensively described in Ref.[29]. Below we briefly discuss the main features of solution of the adjoint cascade equations. For our purposes it is quite sufficient to consider only the longitudinal development of electromag- netic cascades neglecting the emission angles of secondary particles and, thus, assuming that all cascade particles are moved along the shower axis. Besides, we neglect the differences in interactions of electrons and positrons, i.e. do not consider positron annihilation, both in flight and after thermalization in the 4 ambient plasma. This is a quite good approximation as long as we are inter- ested in γ-ray energies exceeding α−1 f mec2 ≃ 70 MeV (αf = 1/137 is the fine structure constant). Under these assumptions, the system of adjoint cascade equations reads ∂f ∂t ∂g ∂t + f /λe − + g/λγ − ε Z εth ε Z εth Wee(ε, ε′)f (t, ε′)dε′ − Wγe(ε, ε′)f (t, ε′)dε′ − ε Z εth ε Z εth Weγ(ε, ε′)g(t, ε′)dε′ = F (t, ε),(1) Wγγ(ε, ε′)g(t, ε′)dε′ = G(t, ε).(2) The adjoint functions f (t, ε) and g(t, ε) in Eqs. (1) and (2) describe the contri- butions (averaged over random realizations of the cascade development) from cascade initiated by a primary electron (f ) or a photon (g) of energy ε (all energies are expressed in units mec2); t is the cascade penetration depth; εth is the minimum energy of cascade particles being under consideration. The parameter λα in Eqs. (1) and (2) defines the mean free path length of cascade particles of type α (α = e or γ). It can be expressed through the total cross-sections (σα) of interaction with the ambient medium. In the most general case, when the medium consists of matter (M), magnetic field (F ) and photon gas (G), λα = hn(M ) 0 σ(M ) α + σ(F ) α + n(G) 0 σ(G) α i−1 (3) where n(M ) ambient photons, respectively. The parameters and n(F ) 0 0 are the number density of the matter atoms and the Wαβ(ε, ε′) = n(M ) 0 w(M ) αβ (ε, ε′) + w(F ) αβ (ε, ε′) + n(G) 0 w(G) αβ (ε, ε′) (4) are differential cross-sections over the energy ε′ of the secondary particle of type β (β = e or γ). They are normalized as Z Wαβ(ε, ε′)dε′ = n(M ) 0 ¯n(M ) αβ σ(M ) α + ¯n(F ) αβ σ(F ) α + n(G) 0 ¯n(G) αβ σ(G) α (5) where ¯n(M ) of type β produced by a particle of type α. αβ and ¯n(G) αβ , ¯n(F ) αβ are the mean multiplicities of secondary particles Properties of the particle "detector" located at the depth t are defined by the right hand side functions F and G in Eqs. (1) and (2) and by the boundary 5 conditions f (t = 0, ε) and g(t = 0, ε). For example, if the "detector" measures the number of cascade electrons above some threshold energy, ε ≥ εth, then F (t, ε) = G(t, ε) = 0; f (t = 0, ε) = H(ε − εth), g(t = 0, ε) = 0 (6) where H(x) = 1 for x < 0. Analogously, if the "detector" measures the number of cascade photons with energy ε ≥ εth, then x ≥ 0 and H(x) = 0 for F (t, ε) = G(t, ε) = 0; f (t = 0, ε) = 0, g(t = 0, ε) = H(ε − εth). (7) For solution of the adjoint cascade equations we use the numerical method proposed in Ref. [30]. To apply this approach to Eqs. (1) and (2) we introduce an increasing subsequence {εk} = ε0 = εth, ε1, . . . , εk, . . . of energy points, and corresponding values of adjoint functions {fk(t), gk(t)}. Let's write now the Lagrange polynomial interpolation formulae for an approximate presentation of functions f and g inside the energy intervals ∆εk = (εk−1, εk) for k = 1, 2, . . . through their values at the support points {εk}, f (ε, t) ≃ f (ε, t) = k Xj=k−n fj(t)Ln kj(ε), g(ε, t) ≃ g(ε, t) = k Xj=k−n gj(t)Ln kj(ε); Ln kj(ε) = n Yr=0 (ε − εk−r)(εj − εk−r)−1, r 6= k − j, ε ∈ ∆εk (8) where n ≡ n(k) = min(N, k) and N is the maximum available power of the Lagrange polynomial. For the first k energy intervals ∆εk the support values from f0, g0 to fk, gk are used; in this case the polynomial power n is equal to k. For intervals with k ≥ N the support values fk−N , gk−N , fk−N +1, gk−N +1, . . ., fk, gk are involved in the interpolation with use of the power n = N. Let's adopt now ε = εk in Eqs.(1) and (2), and present the integral members of these equations in the form of the sum over the energy intervals ∆εi (i = 1, 2, . . . , k): εk Z ε0 Wαβ(εk, ε) · · · dε = k Xi=1 εi Z εi−1 Wαβ(εk, ε) · · · dε, α, β = e, γ . (9) Then, after some simple calculations, we find the following equations ∂fk ∂t + Akfk − Bkgk = F ′ k, ∂gk ∂t 6 − Ckfk + Dkgk = G′ k (10) where Ak = 1/λe(εk) − akk, Bk = bkk, Ck = ckk, Dk = 1/λγ(εk) − dkk,(11) k−1 k−1 F ′ k = Fk + Xj=0 [ajkfj + bjkgj], G′ k = Gk + [cjkfj + djkgj]. (12) Xj=0 The coefficients ajk, bjk, cik and djk can be expressed through cross-sections of relevant processes: k Xi=1 εi Z εi−1 Wee(εk, ε)Ln ij(ε) i Xs=i−n δjsdε , εi Z Weγ(εk, ε)Ln ij(ε) i Xs=i−n δjsdε , etc. (13) ajk = k bjk = Xi=1 εi−1 with δjs = 1 if j = s, and δjs = 0 if j 6= s. Thus, this method allows us to reduce the integro-differential Eqs. (1) and (2) to the system of ordinary differential equations (10). The solution of Eq. (10) can be obtained in two steps: (1) for ε = ε0 = εth in formulae (1) and (2), we find the following equations for functions f0(t) and g0(t) ∂f0 ∂t + f0(t)/λe(ε0) = F (t, ε0), ∂g0 ∂t + g0(t)/λγ(ε0) = G(t, ε0) (14) (2) after solving Eqs. (14) we can calculate the functions f1(t), g1(t), because for k = 1 Eq. (10) contains only f0(t), g0(t), f1(t) and g1(t); after that one can find by the same way the functions f2(t), g2(t), f3(t), g3(t), etc. For a fixed value of k, we introduce an increasing subsequence of the depth values {tl} (t0 = 0, tl = tl−1+τl) and correspondingly fk(tl) = fk,l, gk(tl) = gk,l. For each interval (tl−1, tl) one can evaluate fk,l, gk,l by solving Eq. (10) for which the values fk,l−1, gk,l−1 serve as boundary conditions. This leads to fk,l = (λ0k − λ1k)−1 1 Xν=0 (−1)ν{exp(λνkτl)[(Dk + λνk)fk,l−1 + Bkgk,l−1]+ + tl Z tl−1 exp[λνk(tl − τ )][(Dk + λνk)F ′ k(τ ) + BkG′ k(τ )]}dτ, (15) 7 gk,l = (λ0k − λ1k)−1 (−1)ν{exp(λνkτl)[(Ak + λνk)gk,l−1 + Ckfk,l−1]+ Xν=0 1 + tl Z tl−1 exp[λνk(tl − τ )][(Ak + λνk)G′ k(τ ) + CkF ′ k(τ )]}dτ (16) where λνk = 1 2 {−(Ak + Dk) + (−1)ν[(Ak − Dk)2 + 4BkCk]1/2}. (17) Note that for the homogeneous environment these relations are valid for an arbitrary value of τl, otherwise they can be applied only for τl ≪ tl. The knowledge of fk,0 and gk,0 (these quantities can be calculated on the basis of boundary conditions like Eq. (6) or Eq. (7) ) and the multiple application of Eqs (15) – (17) allow to calculate quantities fk,l, gk,l for l = 1, 2, etc. The approach of solution of adjoint cascade equations described above provides an accuracy better than a few per cent and gives results for an arbitrary region (εth, εmax) of the primary energy in just one run of calculations. Also we notice that the computational time consumed by this method does not exceed on average a few minutes for a 1 GHz PC type computer and increases only weakly (logarithmically) with the primary energy. In summary, the important feature of this technique is its flexibility to describe the cascade processes in 3 different substances. It allows large number of calculations with a good accuracy throughout very large energy intervals of both primary and secondary energies. This is an important condition for the quantitative analysis and for clear understanding of similarities and differences in cascade development in environments dominated by matter, photon gas or magnetic field. 3 Elementary processes The above described technique requires specification of elementary processes that initiate and support the cascade development. If we are interested in the longitudinal development of cascades, the input should consist of total cross-sections of interactions and the differential cross-sections as functions of energy but integrated over the emission angles of secondary particles. In many astrophysical situations, especially at very high energies, this is a fair approximation, given the very small (of order ∼ mec2/E ≪ 1) emission angles of secondary products. The one-dimensional treatment of the cascade develop- ment perfectly works, if electrons move along the lines of the regular magnetic 8 field. Otherwise, the deviation of particles from the cascade core is determined by reflection of secondary electrons by magnetic inhomogeneities rather than the emission angles. In these cases the diffusion effects must be appropriately incorporated into the cascade equations. This question is beyond the present paper. All processes involved in the electromagnetic cascades are well known. The cross-sections of these processes have been calculated and comprehensively studied with very high precision within the quantum electrodynamics. In the case of cascades in matter we take into account the ionization losses and bremsstrahlung for electrons (positrons), and the pair production, Comp- ton scattering and photoelectric absorption for photons. At extremely high energies the so-called Landau-Pomeranchuk-Migdal (LPM) effect that results in suppression of the bremsstrahlung and pair-production cross-sections, may have significant impact on the cascade development. The energy region where the LPM effect becomes noticeable, depends on the atomic number of the am- bient matter. In the hydrogen dominated medium this energy region appears well beyond E ≥ 1020 eV, therefore in many astrophysical scenarios the LPM effect can be safely ignored. Two pairs of coupled processes – the inverse Compton scattering and γ-γ pair production in the photon gas and the magnetic bremsstrahlung (synchrotron radiation) and magnetic pair production in the magnetic field – determine the basic features of cascade produced in radiation and magnetic fields. At extremely high energies the higher order QED processes may compete with these basic channels. Namely, when the product of energies of colliding cascade particles (electrons or photons) E and the background photons ǫ significantly exceed 105 − 106m2 ec4, the processes γγ → e+e−e+e−[31] and eγ → eγe+e− [32] dominate over the single (e+, e−) pair production and the Compton scat- tering, respectively. For example, in the 2.7 K CMBR the first process stops the linear increase of the mean free path of highest energy γ-rays around 1021 eV, and puts a robust limit on the mean free path of γ-rays of about 100 Mpc. Analogously, above 1020 eV the second process becomes more important than the conventional inverse Compton scattering. Because γγ → 2e+2e− and eγ → eγe+e− channels result in production of 2 additional electrons, they significantly change the character of the pure Klein-Nishina pair cascades. However, an effective realization of these processes is possible only at very specific conditions with an extremely low magnetic field and narrow energy distribution of the background photons. Because this study has methodological objectives, below we do not include these processes in calculations of cascade characteristics. This allows us to avoid unnecessary complications and make the analysis more transparent. For the same reason we do not include in this study the effect of the photon 9 splitting [33] which becomes quite important in pulsars with magnetic field close to Hcrit = 4.4 × 1013 G (see e.g. Ref. [22]). 3.1 Total cross-sections In Figs. 1,2 and 3 we show the photon- and pair-production cross-sections in hydrogen gas and in radiation and magnetic fields, respectively. The energies of electrons and γ-rays are expressed in units of mec2. 0 σ / p σ , 0 σ / r b σ 10 1 -1 10 -2 10 10 2 10 3 10 4 10 5 10 0 ε Fig. 1. Total cross-sections of the bremsstrahlung (σbr) and pair production (σp) pro- cesses in hydrogen gas normalized to the asymptotic value (σ0) of the pair produc- tion cross-section at ε0 → ∞. The bremsstrahlung cross-sections are calculated for secondary γ-rays produced with energies exceeding (1) the pair-production thresh- old, εth = 2; (2) the critical energy, εth ≃ 700; (3) half of the energy of the primary electron, εth = ε0/2. In Fig. 1 the total cross-sections for matter are presented. They are normalized to the asymptotic value of the pair production cross-section at ε0 → ∞: σ0 = 7/9 × 4αfr2 e Z(Z + 1) ln(183Z −1/3) 1 + 0.12(Z/82)2 (18) where Z is the charge of the target nucleus, re is the classical electron radius. This actually implies introduction of the so-called radiation length X (M ) 0 = 7/9hn(M ) 0 σ0i−1 (19) 10 Table 1 The asymptotic behaviour of the total cross-sections of electron and photon inter- actions at high energies. Environment Matter Photon gas Magnetic field σe σγ ∼ lnε0 ∼ (κ0 · lnκ0)−1 ∼ const ∼ (κ0 · lnκ0)−1 0 ∼ χ−1/3 ∼ χ−1/3 0 0 X (M ) has a meaning of the average distance over which the ultrarelativis- tic electron loses all but 1/e of its energy due to bremsstrahlung. The same parameter approximately corresponds to the mean free path of γ-rays. There- fore, the cascade effectively develops at depths exceeding the radiation length. Usually the radiation length is expressed in units g/cm2. For the hydrogen gas X (M ) 0 = 63 g/cm2. The second important parameter that characterizes the cascade development is the so-called critical energy below which the ion- ization energy losses dominate over bremsstrahlung losses. In the hydrogen gas εcr ≃ 700. Effective multiplication of particles due to the cascade processes is possible only at energies ε ≥ εcr. At lower energies electrons dissipate their energy by ionization rather than producing more high energy γ-rays which would support further development of the electron-photon shower. The bremsstrahlung differential cross-section has a 1/εγ type singularity at εγ → 0 (εγ is the energy of the emitted photon), but because of the hard spectrum of bremsstrahlung photons the energy losses of electrons are con- tributed mainly by emission of high energy γ-rays. In Fig. 1 we show the bremsstrahlung total cross-sections calculated for 3 different values of mini- mum energy of emitted photons: εth = 2, εcr and εe/2. The first value cor- responds to the cross-section of production of all γ-rays capable to produce electron-positron pairs. The second value corresponds to the cross-section of production of γ-rays produced above the critical energy, and thus capable to support the cascade. And finally, the third value corresponds to the cross- section of production of the "most important" γ-rays which play the major role in the cascade development. It is seen from Fig. 1 that while for εth = 2 the total cross-section of pair-production is an order of magnitude lower com- pared to the bremsstrahlung cross-section, for εth = εe/2 the cross-sections of two processes become almost identical at energies ε ≥ 100. Both cross-sections grow significantly with energy until ε0 ∼ 103. At higher energies the pair-production cross-section becomes energy-independent, but for εth = 2 the bremsstrahlung cross-section continues to grow, although very slowly (logarithmically). In Fig. 2 we show the total cross-sections of the inverse Compton scatter- ing and pair-production for the mono-energetic and Planckian isotropic pho- 11 1 T σ σ / -1 10 -2 10 -3 10 -4 10 -2 10 -1 10 1 10 2 10 3 10 5 10 4 10 κ 0=ε ω 0 0 Fig. 2. Total cross-sections of the inverse Compton scattering and pair production. Two spectral distributions for the ambient photon gas are assumed - mono-energetic with energy ω0 (curves 1 and 3) and Planckian with the same mean photon energy ω0 ≃ 3kT /mec2. ton fields, normalized to the Thompson cross section σT = 8/3πr2 e = 6.65 × 10−25 cm2. Both cross sections depend on the product κ0 = ε0ω0 of the pri- mary (ε0) and ambient (ω0 = ǫ/mec2) photon energies. At κ0 → 0 the inverse Compton cross-section σIC → σT. At high energies it decreases with κ0, as ∝ (κ0 · lnκ0)−1. In the mono-energetic radiation field the pair production has a strict kinematic threshold at κ0 = 1. At small values of κ0 ≤ 2 the cross-section rapidly increases with κ0 achieving its maximum of about ≃ 0.2σT at κ0 ≃ 3.5−4, and decreases with further increase of κ0 as ∝ (κ0 · lnκ0)−1. Thus in this regime the pair-production cross-section behaves quite similar to the Compton scattering in the Klein-Nishina regime, but its absolute value is higher by a factor of 1.5. In the case of interactions of electrons and photons with the magnetic field it is more convenient to introduce, instead of standard total cross-sections, the interaction probabilities [24]. But in the literature this parameter is still formally called as cross-section. These probabilities normalized to the strength of the magnetic field are shown in Fig. 3. The probabilities of both processes – photon production by electrons (synchrotron radiation) and pair production by photons – depend only on the parameter χ0 = ε0H/Hcr, where H is the component of magnetic field intensity perpendicular to the velocity of the particle. The parameter χ0 = H/Hcrε0 is an apparent analog of the parameter 12 1 - ) G x m c ( , H σ / -6 10 -7 10 10 -8 -1 10 1 10 2 10 3 10 χ 4 10 10 ε 0=H/Hcr 0 5 Fig. 3. Total cross-sections of the synchrotron radiation and magnetic pair produc- tion. Solid curves – our calculations; points – calculations of Anguelov et.al. [24]. Table 2 The singularities in the differential cross-sections of cascade processes. Elementary process Singularity point Behaviour Bremsstrahlung Inverse Compton scattering Pair production in photon gas Pair production in photon gas Synchrotron radiation εγ = 0 εγ = ε0 εe = 0 εe = ε0 εγ = 0 ∼ ε−1 γ ∼ (ε0 − εγ)−2 ∼ ε−2 e ∼ (ε0 − εe)−2 ∼ ε−2/3 γ κ0. While the probability of the synchrotron radiation at χ0 ≪ 1 is constant, the probability of the pair production drops dramatically below χ0 = 1. After achieving its maximum at χ0 ≃ 10, the probability of the pair production decreases with χ as ∝ χ−1/3 . The same behaviour at large χ has also the probability of synchrotron radiation, but the absolute value of the latter always exceeds by a factor of 3 the probability of the pair production. 0 13 Table 3 The mean portion of primary energy (¯εγ/ε0) transferred to the secondary photon in the inverse Compton scattering (ics) and the synchrotron radiation (syn) processes via the values of parameters κ0 = ε0ω0 and χ0 = ε0H/Hcr, respectively. κ0, χ0 0.01 0.1 1 102 104 106 (¯εγ /ε0)ics 0.014 0.099 0.358 0.760 0.867 0.910 (¯εγ /ε0)syn 0.44 · 10−2 0.033 0.118 0.241 0.250 0.250 Table 4 The mean portion of primary energy (¯εe/ε0) transferred to the leading secondary particle for the processes of pair production in a mono-energetic photon gas (G) and in the magnetic field (F ) via the values of parameters κ0 = ε0ω0 and χ0 = ε0H/Hcr, respectively. κ0, χ0 1 3 10 102 104 106 (¯εe/ε0)G 0.500 0.701 0.797 0.891 0.948 0.966 (¯εe/ε0)F 0.634 0.693 0.746 0.782 0.824 0.825 3.2 Differential cross-sections of cascade processes The differential cross-sections of cascade processes are presented in Figs. 4,5 and 6. The bremsstrahlung and pair-production cross-sections are from Ref. [34,35]. The cross-sections for the inverse Compton scattering and pair pro- duction in the mono-energetic isotropic photon field are from Ref. [35] and Ref. [36], respectively. The cross-sections of processes in the magnetic field are from Ref. [27]. All three cross-sections of pair-production are symmetric functions in respect to the point x = εe/ε0 = 0.5. The photon production processes are asymmetric functions. The bremsstrahlung, synchrotron radiation, the inverse Compton scattering and the pair production in the photon gas have singularities. The location of singularities and the cross-section behaviour near the singularity points are summarized in Table 2. The character of cascade development is largely determined by the fraction of energy of electrons and photons lost per interaction. In Table 3 we present mean fractions of the electron energy transferred to the secondary photons in the inverse Compton and synchrotron radiation processes at different values of κ0 and χ0. In the classical regime (κ0 ≪ 1 and χ0 ≪ 1) the mean energy lost by the electron per interaction is very small, ∆ε/ε ≪ 1. In the "quantum" regime (κ0 ≫ 1 and χ0 ≫ 1) the interactions have a catastrophic charac- ter; the secondary photons get a significant fraction of the energy of parent 14 10 2 10 1 s t i n u . b r a , ) γ ε , 0 ε ( r b w ε 0 -1 10 0 1 s t i n u . b r a , ) e ε , , 0 0.2 0.4 0.6 0.8 x=εγ/ε 1 0 ε ( p w ε 0 10 -1 0 0.2 0.4 0.6 0.8 x=ε 1 e/ε 0 Fig. 4. Differential cross-sections of the bremsstrahlung (upper panel) and pair pro- duction (bottom panel) processes in hydrogen. The cross-sections are normalized to one radiation length. The energies of primary electrons and γ-rays ε0 are indicated at the curves. electrons. In the photon gas at κ0 ≫ 1 this fraction exceed 0.5, approach- ing asymptotically to 1. In the magnetic field the energy transfer is smaller; at χ0 ∼ 1 it is approximately 0.1, and asymptotically approaches to 1/4 at extremely large χ0. The pair production processes in all 3 substances have (by definition) catas- trophic character (the photon always disappears). Since the differential spectra of secondary electrons are quite flat with increase towards the maximum en- ergy εe → εγ (in the ultrarelativistic regime), these processes proceed with formation of leading electron with energy x = εe/εγ → 1 in the photon gas and somewhat smaller (≃ 0.8) in the magnetic field (see Table 4). 15 s t i n u y r a r t i b r a , ) γ ε , ε ( 0 s c i w s t i n u y r a r t i b r a , ) e ε , 0 ε ( p 10 1 -1 10 -2 10 0 10 1 0.2 0.4 0.6 0.8 1 x=εγ/εγmax w -1 10 0 0.2 0.4 0.6 x=(ε 1 emin) 0.8 e-ε emin)/(ε emax-ε Fig. 5. Differential cross-sections of the inverse Compton (upper panel) and pair production (bottom panel) processes for the case of a mono-energetic gas of ambient photons. The parameters εγmax, εemin and εemax are defined as εγmax = 4ε0(κ0/1 + 4κ0) and εemin, emax = 0.5ε0(1 ∓p1 − 1/κ0). Different values of the parameter κ0 = ε0ω0 are indicated at the curves. 4 Cascades In this section we discuss and compare the so-called cascade curves and the en- ergy spectra of electrons and photons for showers produced in matter, photon gas and magnetic field. 16 ) 0 ε ( n y s σ / ) γ ε , 0 ε ( n y s w ε 0 ) 0 ε ( p σ / ) ε , e 0 ε ( p w ε 0 10 1 -1 10 -2 10 1 -1 10 0 0.2 0.4 0.6 0.8 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 x=εγ/ε 1 0 0.9 x=ε e/ε 0 Fig. 6. Differential cross-sections of synchrotron radiation (upper panel) and the magnetic pair production (bottom panel) normalized to the total cross-sections of these processes. Different values of parameter χ0 = H/Hcrε0 are indicated at the curves. 4.1 Cascade curves The cascade curve Nβ(t, ε0, εth) describes the average number of cascade elec- trons (β = e) or photons (β = γ) above εth, as a function of the penetration depth t. 4.1.1 Matter In Fig. 7 we show the cascade curves for electrons calculated for different values of the primary (ε0) and threshold (εth) energies using the adjoint equation technique. For comparison we also show the cascade curves calculated within 17 the approximation A of the analytical theory of electromagnetic cascades [1] (valid for εth ≫ εcr), as well as the cascade curves obtained by Monte Carlo simulations using the ALTAI code[37]. The results obtained by 3 different methods are in a good agreement with each other. One of the basic parameters characterizing the cascade development is the depth tm (expressed in units of radiation length X (M ) ) at which the cascade curve achieves its maximum. Generally tm grows logarithmically with the pri- mary energy. In particular, for light materials (with the nucleus charge number Z ≤ 10) and for ε0 ≫ εcr, the parameter tm for electrons of a cascade initiated by a primary photon can be approximated as [1]: 0 te m ≃ ln(ε0/εth) for εth ≫ εcr; tm ≃ ln(ε0/εcr) for εth → 0. (20) The primary electron interacts with matter somewhat earlier than the primary γ-ray (see Fig. 1), therefore tγ m ≈ te m + 0.7. Another important parameter is the total number of electrons at the shower maximum: Nmax = Ne(tm, ε0, εth). This number is approximately proportional to the primary energy. In the energy region εth ≫ εcr the electron number Nem ∼ ε−1 th ; however it does not depend on εth, if εth → 0. In Fig. 8 we present the ratio of the number of the cascade photons to electrons as a function of the penetration depth. For εth ≫ εcr, the number of cascade photons Nγ is comparable to the electron number Ne. At small threshold energies, (εth ≪ εcr) the number of γ-rays considerably exceeds the number of electrons. This is explained by the break of symmetry between the electrons and γ-rays caused by ionization losses of electrons. 4.1.2 Photon gas For description of the cascade development in the mono-energetic photon gas it is convenient to introduce, analogously to the cascade in matter, the radiation length X (G) in the following form 0 X (G) 0 = h4πn(G) 0 r2 ei−1 κ0 , (21) 0 is the number density of photons. Apparently, X (G) where n(G) corresponds, with accuracy up to logarithmic terms, to the mean free path of γ-rays in the photon field at κ0 ≫ 1. 0 Note that unlike to the radiation length in matter, the primary energy en- ters, through the parameter κ0 = ε0ω0, in X (G) . Obviously, for the ambient 0 18 10 3 e N 10 2 10 1 0 5 10 15 20 25 t, r.l. Fig. 7. Cascade curves of electrons for showers initiated by primary electrons (solid curves) and photons (dashed curves). The calculations are performed for the follow- ing primary energies ε0 = 2 · 108 (curve 1), 2 · 107 (curve 2), 2 · 104 (curve 3) and the ratio εth/εcr = 125 (curves 1 and 2), 0.05 (curve 3). For comparison, the results derived from the analytical cascade theory [1] (boxes) and by simulations with the ALTAI code [37] (triangles) are also shown photon gas with a broad band energy distribution this parameter becomes meaningless. At the same time in a narrow band radiation fields, e.g. with Planckian distribution, this parameter can work effectively by substituting ¯ω0 ≈ 3kT /mec2. Note that for the black-body radiation the density of pho- tons is determined by temperature, therefore X (BB) 0 = 3/πα3 f r0(kT /mec2)−2ε ≃ 6.9 × 10−7(kT /mec2)−2ε cm (22) In particular, in the 2.7 K CMBR X (BB) 0 ≃ 3.3 × 1012ε cm. The results shown in Fig. 9 are cascade curves of electrons and photons in the blackbody photon gas calculated for the fixed value of κth = εth ¯ω0 = 1 corresponding to the threshold energy of cascade particles, εth = mec2/3kT , but for different values of the parameter κ0, i.e. for different primary photons energies ε0 = κ0 · mec2/3kT . We can see that the depth tm of the shower maximum measured in units of radiation length X (G) depends rather weakly on the primary energy ε0. It ranges within ∼ 1 ÷ 2. This means that in geometrical units of length tm is approximately a linear function of ε0. This is explained by the approximately 0 19 e N / γ N 12 10 8 6 4 2 0 0 5 10 15 20 25 t, r.l. Fig. 8. The depth dependence of the Nγ/Ne ratio for cascades initiated by primary electrons (solid curves) and photons (dashed curves). Primary energy ε0 = 2 · 108; εth/εcr = 0 (curves 1), 125 (curves 2). linear decrease of the cross-sections and correspondingly by the increase of mean free path of electrons and γ-rays interacting with the ambient photons in the Klein-Nishina (quantum) regime (see Fig. 2). The number of cascade particles in the Klein-Nishina regime increases with ε0 slowly. Even near the shower maximum this number does not exceed a few particles. This is explained by an extremely high efficiency of conversion of en- ergy of the photon to the leading electron at γ-γ interactions, and vice versa - the electron energy to the upscattered photon at the Compton scattering (see Figure 5 and Tables 3,4). As a result, the energy of the second (secondary) particle appears too small for noticeable contribution to the cascade develop- ment. Since in this regime the cross-sections of the Compton scattering and photon-photon pair production are quite similar, the cascade process in the zeroth approximation can be considered as propagation of a single composite γ/e′′ particle which spends 2/5 of its time in the "γ state" and 3/5 in the "elec- tron state" (these times are determined by the ratio 1.5 of the corresponding cross-sections in the Klein-Nishina limit). Fig. 10 illustrates dependence of the cascade curve on the threshold energy εth with the corresponding parameter κth below and above 1. With reduction of κth, the cascade curves of both γ-rays and electrons increase, especially in the regime of κth ≤ 1. This has a simple explanation. Although in this (Thompson) regime of scattering the cross-section is large, the energy transferred to the 20 secondary photon is quite small (see Table 3). Consequently, the mean free path of these electrons increases with reduction of their energy as 1/ε. These electrons produce large number of γ-rays with energy below the effective pair- production threshold 3 , which therefore penetrate very large distances without interacting with the ambient background radiation. Consequently, the number of these photons remains almost constant. e N γ N 5 4 3 2 1 0 5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 0 1 2 3 4 5 6 7 t, r.l. 0 1 2 3 4 5 6 7 t, r.l. Fig. 9. Cascade curves of electrons (upper panel) and photons (bottom panel) for showers produced in the Planckian gas of ambient photons by high energy photons. The results are obtained for κth = εth ¯ω0 = 1, and several values of κ0 = ε0 ¯ω0 indicated at the curves. 3 In the case of the black-body radiation of background photons there is no strict kinematic threshold; γ-rays with energy ε ≤ mec2/3kT may still interact with the background photons from the Wien tail of distribution. 21 e N 10 2 10 1 -1 10 0 1 2 3 4 5 6 7 8 9 t, r.l. γ N 10 4 10 3 10 2 10 1 10 -1 0 1 2 3 4 5 6 7 8 9 t, r.l. Fig. 10. Cascade curves of electrons (upper panel) and photons (bottom panel) for showers produced in the Planckian gas of ambient photons by high energy photons. The results are obtained for κ0 = ε0 ¯ω0 = 103 and for values of κth = εth ¯ω0 indicated at the curves. 4.1.3 Magnetic field Fig. 11 shows the cascade curves of electrons in the magnetic field obtained with the adjoint equation technique. The comparison with the Monte Carlo simulations [24] shows nice agreement between results obtained by two differ- ent methods. Following to Ref. [24] we express the depth of penetration of cascade particles in units of the radiation length in the magnetic field 0 = 0.207 · 10−7 Hcr X (F) H χ1/3 0 , (23) 22 where χ0 = ε0H/Hcr. Similar to the cascade in the photon gas, the radiation length in the magnetic field depends on the primary energy, but in this case the dependence is slower, X (F) 0 ∼ ε1/3 . 0 e N 10 5 10 4 10 3 10 2 10 1 -1 10 0 1 2 3 4 5 6 7 8 t, r.l. Fig. 11. Cascade curves of electrons for showers initiated by primary electrons in the magnetic field. Different values (indicated at the curves) of the ratio of primary and threshold energies are assumed for the fixed χth = εthH/Hcr = 103. For comparison, the results obtained in Ref. [24] are also shown (dashed curves). Fig. 11 illustrates dependence of cascade curves on the primary energy in a deep quantum regime with the threshold value of the parameter χth = 103. It is seen that the location of the cascade curve maximum tm ranges within 2 ÷ 3 radiation lengths. In geometrical units this implies that the position of the cascade curve maximum increases with energy proportional to ε1/3 . The maximum electron number, Nm, grows with ε0 as ∼ ε0.7 0 . Thus, the cascade development in the magnetic field in the quantum regime is somewhat inter- mediate between the cascades in matter and the photon gas. 0 In Fig. 12 we present the cascade curves of electrons and photons for different values of the threshold energy εth. It is seen that in the regime χth ≪ 1 the cascade curves have a behaviour quite similar to that for the cascade curves in the photon gas shown in Figure 10. This can be explained by similarities of the synchrotron radiation and the Compton scattering in the non-quantum regime. In both cases the electrons suffer a large number of collisions but with small energy transfer to secondary photons. In addition, in both cases γ-rays stop to interact effectively with the ambient medium. Table 5 demonstrates the dependence of the Nγ/Ne ratio on χth and t. It 23 e N 10 5 10 4 10 3 10 2 10 1 -1 10 10 7 10 6 10 5 10 4 10 3 10 2 10 1 -1 10 γ N 0 2 4 6 8 10 12 t, r.l. 0 2 4 6 8 10 12 t, r.l. Fig. 12. Cascade curves of electrons (upper panel) and photons (bottom panel) for showers initiated by primary electrons (solid curves) and photons (dashed curves) in the magnetic field. Different values of the threshold energy (εth) are assumed for the fixed ε0 = 2 · 107 and H = 1011 G. The ε0/εth ratios are indicated at the curves. The corresponding values for χth are presented in Table 5. Table 5 The ratio Nγ/Ne for cascade curves presented in Figure 12. ε0/εth χth t =0.3 r.l. t =0.9 r.l. t =3 r.l. t =6 r.l. 102 4.5 · 102 0.76 0.94 1.00 1.03 104 4.5 0.92 1.09 1.17 1.19 105 106 108 4.5 · 10−1 4.5 · 10−2 4.5 · 10−4 2.66 4.65 18.6 140 5.95 1.92 45.2 455 1.44 2.73 11.2 107 24 behaves quite similar to cascades in matter. Particularly, for large threshold energies εth the electron and photon numbers are comparable; for small εth the photon number is considerably larger, and the ratio Nγ/Ne increases with the depth. 4.2 Energy spectra of cascade particles Compared to the cascade curves, the energy spectra of electrons and photons at different stages of the cascade propagation contain more circumstantial information about the shower characteristics. ε d / N d 2 ε 10 7 10 6 10 5 10 4 -1 10 1 10 ε/ε cr Fig. 13. Differential energy spectra of electrons (solid curves) and photons (dashed curve) for cascades initiated in the hydrogen gas by a primary photon with energy ε0 = 2 · 108. Different values of depths in units of the cascade maximum tm are shown at the curves. In Fig. 13 we show the energy spectra of cascade electrons and photons at different penetration depths in the hydrogen gas. At energies exceeding the critical energy, ε ≫ εcr, the spectra of both electrons and photons are de- scribed by power-law dN/dε ∼ ε−α, where the spectral index is function of the penetration depth. Near the shower maximum, α ≃2 for both electron and photon spectra. With depth the spectra become steeper. Below the critical energy, the cross-sections of the bremsstrahlung and pair- production processes are not sufficiently large to support the cascade devel- opment against the dissipative processes like ionization and Compton scatter- ing. Thus the multiplication of cascade particles is dramatically reduced. The 25 ε d / e N d 2 ε ε d / γ N d 2 ε 10 3 10 2 10 1 -1 -2 -3 10 10 10 10 4 10 3 10 2 10 1 -4 10 -3 10 -2 10 -1 10 1 10 2 10 εω 0 10 10 -1 -2 -7 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 10 1 10 2 10 εω 0 Fig. 14. Differential energy spectra of cascade electrons (upper panel) and photons (bottom panel) in the cascade initiated by a primary photon (κ0 = 103) in the radiation field with Planckian spectral distribution. The spectra are calculated for different penetration depths indicated (in units of radiation lengths) at the curves. cooling time of electrons due to ionization losses is proportional to energy, therefore below the critical energy this process leads to significant hardening of the electron, and correspondingly also the photon spectrum that behaves as dN/dε ∝ ε−1. In Fig. 14 we show the differential energy spectra of cascade particles in the radiation field with Planckian type spectral distribution. These spectra are quite different from the ones that appear in the cascades developed in matter. In the high energy (Klein-Nishina) region, κ = εω0 ≫ 1, and for not very large depths (t ≤ 10 r.l.) the shape of the differential spectra of electrons and photons is quite insensitive to the depth (contrary to the cascade in matter). At these energies the differential spectra are characterized by slopes with indices between 1 and 1.5. 26 At κ ∼ 1 there exists a pronounced transition region where the energy spectra undergo dramatic changes. The reason for the appearance of the transition re- gion is the change of the character of Compton scattering (from the Thompson to the Klein-Nishina regime) and the sharp reduction of the pair production cross-section. Obviously, a broader energy distribution of background photons would make this transition smoother and less pronounced. ε d / e N d 2 ε ε d / γ N d 2 ε 10 5 10 4 10 3 10 2 10 7 10 6 10 5 10 4 10 3 10 2 -4 10 -3 10 -2 10 -1 10 1 2 10 χ=H/HCR 10 ε -7 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 10 2 1 10 χ=H/HCR 10 ε Fig. 15. Differential energy spectra of cascade electrons (upper panel) and photons (bottom panel) in showers initiated in the magnetic field of intensity H = 1011 G by a primary photon of energy ε0 = 2 · 107. The spectra are calculated at different depths indicated (in units of radiation lengths) at the curves. At low energies (κ ∼ 1) the cascade development is not supported by pair- production. Here we deal with an ensemble of photons not interacting with the environment and an ensemble of electrons continuously cooling down with the characteristic (Thompson) time tT ∼ 1/ε. This results in standard electron spectra with spectral index α = 2. During propagating into deeper layers of the photon gas, these electrons produce large number of low energy photons 27 with energy spectrum dNγ/dε ∝ ε−1.5. For any finite depth t, electrons do not have enough time to be cooled down to energies ε → 0. Therefore the electron spectrum drops dramatically below some energy εe which decreases with depth as ∼ 1/t. This effect is clearly seen Fig. 14. The corresponding response in the photon spectrum is also quite distinct. Below the break energy around ∼ ω0ε2 e, the photon spectrum becomes extremely hard, dNγ/dε ≈ const. The energy spectra of electrons and photons produced during the cascade development in the magnetic field are quite similar to the spectra of electro- magnetic cascades in the radiation field with Planckian distribution of target photons. These spectra are shown in Fig. 15. All features discussed above in the context of cascading in the photon gas, are clearly seen also in cascades developed in the magnetic field, if we express the penetration depths in units of radiation lengths, and the energies of electrons and photons in the form of products εω0 and ε(H/Hcr). The cascade spectra in radiation and mag- netic fields are not, however, identical. For example, because of significant differences in the asymptotics of relevant cross-sections, γ-ray spectra in the magnetic field in the quantum regime are flatter than the corresponding γ-ray spectra in the photon gas in the Klein-Nishina regime (compare Figs. 14 and 15). 5 Mixed environment In order to reveal peculiarities of the cascade development in different sub- stances, in previous sections we limited our discussion to the "clean" environ- ments dominated by matter, radiation or magnetic fields. The "pure cascade" concept is not only a convenient theoretical approximation. In fact, under cer- tain realistic conditions and within limited energy regions, this could be the most likely realization of particle interactions with then ambient medium. The relativistic electron-photon cascades in the Earth's atmosphere, in the inter- galactic medium and in pulsar magnetospheres are 3 characteristic examples of "pure" cascade developments in matter, photon gas and magnetic field, respectively. In some cases, however, "parallel" interactions of electrons and photons with 2 or 3 substances can proceed simultaneously and with comparable efficien- cies. The outcome of the interference of several competing processes could be quite different and complex depending on relative densities of the ambi- ent plasma, radiation and magnetic fields, as well as on the energy of pri- mary particles. For example, interactions of ≥ 1020 eV protons with 2.7 K CMBR leads to production of secondary electrons, positrons and γ-rays of 28 Table 6 Parameters of the mixed environment used for calculations of the cascade curves shown in Fig. 16. H is the strength of the magnetic field, T is the temperature of the black-body radiation, wF and wG are energy densities of the magnetic field and black-body radiation, respectively. Parameter Combination 1 H, Gauss kT , eV wF , erg/cm3 wG, erg/cm3 2 0 3 0 3 100 0.3 400 4 100 3 400 5 10 0.3 4 100 – 400 0 ≃ 104 ≃ 1 ≃ 104 ≃ 1 energy ≥ 1019 eV, which in their turn trigger electromagnetic cascades in the same radiation field. However, due to the synchrotron cooling of elec- trons this process would be significantly suppressed if the intergalactic mag- netic field exceeds 10−10 G. The characteristic energy of synchrotron photons ∼ 5 × 108(B/10−10 G)(E/1019 eV)2 eV is too small for interactions with the diffuse photon fields on the Hubble scales. However, if the energy of electrons and/or photons injected into the intergalactic medium exceeds ≥ 1021 eV (this could be the case, for example, of secondary products from decays of the so- called topological defects), then the synchrotron photons appear in the TeV energy range. The TeV γ-rays effectively interact with the diffuse extragalac- tic infrared radiation, and trigger new, low energy electron-photon cascades. In both cases the synchrotron radiation changes the character of the cascade development in the radiation field, but cannot support its "own" cascade in the magnetic field. Below we briefly discuss a more interesting scenario, when the cascade de- velops both in the radiation and magnetic fields. Let's assume that a γ-ray photon of energy Eγ = 1020 eV is injected into a highly magnetized low-density plasma with thermal radiation density comparable to the energy density of the magnetic field H 2/8π. In principle such a situation can occur in the vicinity of central engines of AGN. Fig. 16 illustrates the impact of the magnetic field and the temperature of blackbody radiation on cascade curves. Five different combinations of these pa- rameters presented in Table 6 have been analyzed. The minimum particle en- ergy for the cascade curves was taken 5×1011 eV, i.e. comparable or larger (for the assumed radiation temperatures) to the effective pair production threshold in the black-body radiation , E(G) ec4/kT ≃ 1011(kT /1 eV)−1 eV. th ∼ m2 For the chosen parameters, the radiation length in the magnetic field is much smaller than the mean free path of primary γ-rays in the photon field. At ab- sence of magnetic field (parameter combination 2) primary γ-rays penetrates 29 5 10 6 10 7 10 8 10 9 10 10 10 10 11 10 12 10 13 10 14 10 15 10 16 10 17 18 19 10 10 t, cm e N 10 4 10 3 10 2 10 1 γ N 10 6 10 5 10 4 10 3 10 2 10 1 5 10 6 10 7 10 8 10 9 10 10 10 10 11 10 12 10 13 10 14 10 15 10 16 10 17 18 19 10 10 t, cm Fig. 16. The cascade curves of electrons (upper panel) and photons (bottom panel) with energy E ≥ Eth = 5 · 1011 eV for showers initiated by a primary photon of energy E0 = 1020 eV in a mixed environment consisting of the magnetic field and the blackbody radiation. The calculations are performed for 5 combinations of pa- rameters characterizing the target radiation and magnetic fields. These parameters are listed in Table 6. very deep into the source without interacting with the ambient photon gas. Consequently, the cascade starts very late and remains as underdeveloped. The presence of a strong magnetic field makes the cascade development much more effective, which now is supported by magnetic pair production and syn- chrotron radiation. The cascade develops in this regime until the energy of γ-rays is reduced down to the effective threshold of the magnetic pair pro- duction, E(F) th ≃ 1017(H/100 G)−1 eV. Therefore, in the case of pure magnetic field we have quite simple and predictable cascade curves (parameter combi- nation 1). 30 th , E(F ) The presence of photon gas changes significantly the character of cascade de- velopment. At energies above E(F ) th both electrons and γ-rays interact mainly with magnetic field. When the cascade particles enter the energy interval de- termined by the pair-production thresholds in the radiation and magnetic fields, [E(G) th ], the γ-rays interact effectively with the photon gas, while the electrons continue to interact mainly with the magnetic field. Although for kT = 3 eV, the blackbody radiation density considerably exceeds the energy density of 100 G magnetic field (see Table 6), because of the Klein-Nishina effect the interactions of electrons with photon gas become significant only when the electrons are cooled down to energies ≤ 100 GeV. The mean free path of γ-rays in the black-body radiation field has minimum around E(G) th . Therefore, γ-rays start to interact intensively with the ambient photons at depths comparable with the mean free path Λmin ∼ Λγ(E(G) th ). This results in a rapid growth of the electron cascade curve and reduction of the high energy photon cascade curve (parameter combinations 3-5). Due to the rapid synchrotron cooling, the mean free paths of very energetic electrons in the magnetic field are very short, therefore for each given depth the main source of these electrons are γ-rays which can penetrate much deeper. On the other hand, due to the same synchrotron losses, the electrons cannot support repro- duction of very high energy γ-rays. Therefore, the at depths ≫ Λmin both the electron and photon curves are suppressed. The increase of the temperature increases the energy density of the photon gas (∝ T 4) and makes interactions of cascade particles with radiation more intensive. This results in faster absorption of cascade particles (parameter combination 4). The reduction of the magnetic field (parameter combination 5) leads to the increase of the magnetic radiation length, as well as to the increase of the magnetic pair production threshold. As a result, the cascade develops slower. Figure 17 illustrates the spectral evolution of cascade γ-rays calculated for H = 100 G and two different temperatures of the black-body radiation, kT = 0.3 and 3 eV (parameter combinations 3 and 4). It is seen that at small depths (t ≤ 1010 cm for the upper panel and t ≤ 108 cm for the bottom panel) the spectra are fully determined by interactions with the magnetic field. At these depths a "standard" power-law spectrum with α = 1.5 is formed in a broad energy region below E(F ) th . At larger depths the interactions with the photon gas start to deform the shape of the spectrum. These interactions lead to absorption of photons above E(G) th . At the same time, synchrotron radiation of the photo-produced pairs appears in the energy region E ≤ E(G) th . Thus, only the photons with energy below E(G) can survive at large depths. The th spectrum of these photons is close to power-law with photon index 1.8-1.9, i.e. steeper than the the canonical ε−1.5 cascade photon spectrum formed in the pure photon gas or pure magnetic field. This is explained by the break of symmetry between electrons and γ-rays. While at late stages of the cas- 31 cade development γ-rays continue to produce high energy electron-positron pairs, the synchrotron cooling of these electrons does not anymore support the cascade, but rather destroys it. Finally, in Fig. 18 we show the spectral evolution of cascade γ-rays calculated for a less extreme combination of model parameters. Namely, compared to Fig. 17, we assume a smaller energy for primary γ-rays (E = 1017 eV) and weaker magnetic field (H = 1 G). At these conditions the primary photons do not interact with the magnetic field. Instead, the cascade development starts with pair production in the radiation field. Nevertheless, we can see that many basic features, in particular the ε−2 type spectrum at low energies, are quite similar to the spectra shown in Fig. 17. V e , γ E d / γ N d γ 2 E V e , γ E d / γ N d γ 2 E 10 19 10 18 10 17 10 16 10 19 10 18 10 17 10 16 7 10 8 10 9 10 10 10 11 10 12 10 13 10 14 10 15 10 16 17 10 10 Eγ, eV 7 10 8 10 9 10 10 10 11 10 12 10 13 10 14 10 15 10 16 17 10 10 Eγ, eV Fig. 17. Differential energy spectra of γ-rays of the cascade initiated by a primary photon of energy E0 = 1020 eV in the compound environment consisting of a black- body radiation of temperature T and a homogeneous magnetic field H=100 G. Upper panel – kT = 0.3 eV ; bottom panel – kT = 3 eV. The spectra are calculated at different depths (in cm) indicated at the curves. 32 γ E d / γ N d γ 2 E 10 16 10 15 10 14 10 13 10 12 10 11 10 10 8 10 9 10 10 10 11 10 12 10 13 10 14 10 16 15 10 10 Eγ, eV Fig. 18. The same as in Fig. 17, but for the energy of primary photons E0 = 1017 eV and the magnetic field H = 1 G. The temperature of the black-body radiation is assumed 0.3 eV. 6 Summary In this study, the technique of adjoint cascade equations has been applied to investigate properties of electron-photon cascades in hydrogen gas and in ambient radiation and magnetic fields. We also have inspected the main fea- tures of cross-sections of relevant processes that initiate and support cascade developments in these substances. The cascade curves of electrons and photons in the photon gas and magnetic field have features quite different from the cascade curves in matter. The energy spectra of cascade particles are also considerably different from the conventional cascade spectra in matter. The spectra for the magnetic field have properties intermediate between those for cascade spectra in matter and in the photon field. Although for certain astrophysical scenarios the development of cascades in "pure" environments can be considered as an appropriate and fair approximation, at some conditions the interference of processes associated with interactions of cascade electrons and γ-rays with the ambient photon gas and magnetic field (or matter) can significantly change the character of cascade development, and consequently the spectra of observed γ-rays. The impact is very complex and quite sensitive to the choice of specific parameters. Therefore each practical case should be subject to independent studies. 33 Acknowledgements We are grateful to A.Timokhin for fruitful discussions. AVP thanks Max- Planck-Institut fur Kernphysik (Heidelberg) for hospitality and support during his work on this paper. References [1] B. Rossi, K. Greisen, Rev. Mod. Phys. 13 (1941) 419; J.Nishimura, Handbuch der Physik Bd.XLVI/2 (1967) 1; I.P Ivavenko, Electromagnetic Cascade Processes (1968), Moscow State University Press (in Russian); T.K. Gaisser, Cosmic Rays and Particle Physics (1990), Cambridge University Press. [2] W.R. Nelson, H. Hirayama, D. Rogers, Preprint SLAC-265 (1985), Standford University. [3] V.S. Berezinsky, S.V. Bulanov, V.A. Dogiel, V.L. Ginzburg, V.S. Ptuskin Astrophysics of cosmic rays (1991), Amsterdam: North-Holland. [4] V.S. Berezinsky, in Roberts A. (editor), Proc. 1976 DUMAND Summer Workshop (1976), FNAL, Batavia, p. 229; D. Eichler, W.T. Westrand, Nature 307 (1984) 613. [5] F. Halzen, D. Hooper, Rep. Prog. Phys. 65 (2002) 102. [6] F.A. Aharonian, V.V. Vardanian, V.G. Kirillov-Ugryumov, Astrophysics (tr. Astrofizika) 20 (1984) 118; F.A. Aharonian, V.V. Vardanian, Astr. Sp. Sci. 115 (1985) 31. [7] A.A. Zdziarski, Astrophys. J. 335 (1988) 786. [8] R. Svensson, MNRAS 227 (1987) 403. [9] P.S. Coppi, R.D. Blandford, MNRAS 245 (1990) 453. [10] A. Mastichiadis, R.J Protheroe, A.P. Szabo, MNRAS 266 (1994) 910. [11] G. Cavallo, M.J. Rees, MNRAS 183 (1978) 359; E.V. Derishev, V.V. Kocharovsky, Vl.V. Kocharovsky, Astron. Astrophys. 372 (2001) 107; C.D. Dermer, Astrophys. J. 574 (2002) 65. [12] K. Mannheim, Phys. Rev. D 48 (1993) 2408; R.D. Blandford, A. Levinson, Astrophys. J. 441 (1995), 79; A. Atoyan, C.D. Dermer, Phys. Rev. Lett. 87 (2001) 221102; A. Muecke, R.J. Protheroe, R. Engel, J.P. Rachen, T. Stanev, Astropart. Phys. (2002), in press; [13] P.L. Biermann, P.A. Strittmatter, Astrophys. J. 322 (1987) 643; K. Mannheim, P.L. Biermann, W.M. Kruells, Astron. Astrophys. 251 (1991) 723. 34 [14] A. Neronov, D . Semikoz, F. Aharonian, O. Kalashev, Phys. Rev. Lett. 89 (2001) 1101. [15] F. A. Aharonian, MNRAS 332 (2002) 215. [16] F.A. Aharonian, P.S. Coppi, H.J. Volk, Astrophys. J. 423 (1994) L5. [17] F.A. Aharonian, A.M. Atoyan, Sov. Phys. JETP 62 (1985) 189; R.J. Protheroe, MNRAS 221 (1986) 769; F.A. Aharonian, B.L. Kanevsky, V.V. Vardanian, Astr. Sp. Sci 167 (1990) 93; I.P. Ivanenko, A.A. Lagutin, Proc 22nd ICRC (Dublin), 1991, vol. 1, p. 121; F.A. Aharonian, P. Bhattacharjee, D.N. Schramm, Physical Review D 46 (1992) 4188; R.J. Protheroe, T. Stanev, MNRAS (264) 191, 1993; R.J. Protheroe, T. Stanev, V.S. Berezinsky, Nucl. Phys. B 43 (1995) 62; R.J. Protheroe, T. Stanev, Phys. Rev. Letters 77 (1996) 3708; P.S. Coppi, F.A. Aharonian, Astrophys. J. 423 (1997) L9; G. Sigl, S. Lee, D. Schramm, P. Coppi, Phys. Letters B 392 (1997) 129; S. Lee, Phys. Rev. D 58 (1998) 043004, Z. Fodor, S.D. Katz, Phys. Rev. D 63 (2001) 023002; F.A. Aharonian, A.N. Timokhin, A.V. Plyasheshnikov, Astronon. Astrophys. 384 (2002) 834; O. Kalashev, V. Kuzmin, D. Semikoz, G. Sigl, Phys. Rev. D 65 (2002) 103003. [18] S. Bonometto and M. J. Rees, MNRAS 152 (1971) 21. [19] P.W. Guilbert, A. C. Fabian, M.J. Rees, MNRAS 205 (1983) 593. [20] F.A. Aharonian, V.G. Kririllov-Ugriumov, V. V. Vardanian, Astr. Sp. Sci. 115 (1985) 201. [21] P.A. Sturrock, Astrophys. J. 164 (1971) 529. [22] M.G. Baring, A.K. Harding, Astrophys. J. 547 (2001) 529. [23] B.L. Kanevsky and A.I. Goncharov, Voprosy atomnoy nauki i techniki 4 (1999) p. 1 (in Russian). [24] V. Anguelov, H. Vankov, J. Phys. G: Nucl. Part Phys. 25 (1999) 1755. [25] A.V. Plyasheshnikov, F.A. Aharonian, . J. Phys. G: Nucl. Part Phys. 28 (2002) 267. [26] W. Bednarek, MNRAS 285 (1997) 69. [27] A.I. Akhiezer, N.P. Merenkov, A.P. Rekalo J.Phys.G: Nucl. Part. Phys. 20 (1994) 1499. [28] L. D. Landau, G. Rumer, Proc. R. Soc A 166 (1938) 213. [29] V.V. Uchaikin, V.V. Ryzhov The Stochastic Theory of Transport of High Energy Particles (1998), Novosibirsk, Nauka (in Russian) [30] A.V. Plyasheshnikov, A.A. Lagutin, V.V. Uchaikin, Proc. of 16th ICRC (1979), Kyoto, vol. 7, p.1 . [31] R.W. Brown, W.F. Hunt, K.O. Mikaeilian, I.J. Muzinich, Phys. Rev. D 8 (1973) 3083. 35 [32] A. Mastichiadis, MNRAS 253 (1991) 235; C.D. Dermer, R. Schlickeiser, Astron. Astrophys. 252 (1991) 414. [33] S.L. Adler, Ann. Phys. 67 (1971) 599. [34] A.I. Akhiezer, V.B. Berestetskii, Quantum Electrodynamics, 1965, Interscience, New York. [35] G.R. Blumenthal, R.G. Gould, Review Mod. Phys. 42 (1971) 237. [36] F.A. Aharonian, A.M. Atoyan, A.M. Nagapetyan, Astrophysics (tr. Astrofizika) 19 (1983) 187. [37] A.K. Konopelko, A.V. Plyasheshnikov, Nucl. Instr. Meth. 450 (2000) 419. 36
astro-ph/9604184
1
9604
1996-05-01T00:54:32
Status of Cosmological Parameters: $\Omega_0\approx 0.3$ vs. $\Omega=1$
[ "astro-ph", "hep-ph" ]
The cosmological parameters that I discuss are the Hubble parameter $H_0 \equiv 100 h$ km s$^{-1}$ Mpc$^{-1}$, the age of the universe $t_0$, the average density $\Omega_0$, and the cosmological constant $\Lambda$. To focus the discussion, I concentrate on the the value of $\Omega_0$ in currently popular models in which most of the dark matter is cold, especially Cold + Hot Dark Matter (CHDM) and flat ($\Omega_0 + \Omega_\Lambda=1$) low-$\Omega$ CDM with a Cosmological Constant ($\Lambda$CDM). The evidence would favor small $\Omega_0 \approx 0.3$ if (1) the Hubble parameter actually has the high value $h \approx 0.75$ favored by many observers, and $t_0 \geq 13$ Gy; or (2) the baryonic/total mass ratio in clusters of galaxies is actually $\sim 15$\%, about 3 times larger than expected for standard BBN in an $\Omega=1$ universe, $\Omega_b \approx 0.0125 h^{-2}$, despite the recent measurement by Tytler of $D/H=2.4\times 10^{-5}$ in two high-redshift Lyman limit systems, implying $\Omega_b\approx 0.024 h^{-2}$. The evidence would favor $\Omega=1$ if (1) the POTENT analysis of galaxy peculiar velocity data is right, in particular regarding outflows from voids or the inability to obtain the present-epoch non- Gaussian density distribution from Gaussian initial fluctuations in a low- $\Omega$ universe; or (2) the preliminary LSND report indicating neutrino mass $\gsim 2.4$ eV is right, since that would be too much hot dark matter to allow significant structure formation in a low-$\Omega_0$ $\Lambda$CDM model. Statistics on gravitational lensing of quasars provide an upper limit on $\Lambda$, and the preliminary results on the deceleration parameter $q_0=\Omega_0/2-\Omega_\Lambda$ on very large scales from high-redshift Type Ia supernovae suggest that $\Omega_0 \sim 1$ and $\Omega_\Lambda$ is small.
astro-ph
astro-ph
Status of Cosmological Parameters: Ω0 ≈ 0.3 vs. Ω = 1 ∗ Joel R. Primack Santa Cruz Institute for Particle Physics, University of California, Santa Cruz, CA 95064 Abstract The cosmological parameters that I will discuss are the traditional ones: the Hubble parameter H0 ≡ 100h km s−1 Mpc−1, the age of the universe t0, the average density Ω0 ≡ ¯ρ/ρc in units of critical density ρc, and the cosmological constant Λ (or ΩΛ ≡ Λ/(3H 2 0 )). To focus the discussion, I will concentrate on the issue of the value of Ω0 in currently popular cosmological models in which most of the dark matter is cold, especially Cold + Hot Dark Matter (CHDM), and flat (Ω0 + ΩΛ = 1) low- Ω CDM with a Cosmological Constant (ΛCDM). The evidence would favor a small Ω0 ≈ 0.3 if (1) the Hubble parameter actually has the high value h ≈ 0.75 favored by many observers, and the age of the universe t0 ≥ 13 Gy; or (2) the baryonic/total mass ratio in clusters of galaxies is actually ∼ 15%, about 3 times larger than expected for standard Big Bang Nucleosynthesis in an Ω = 1 universe, and standard BBN is actually right in predicting that the density of ordinary matter is Ωb ≈ 0.0125h−2, based mainly on 4He and 7Li data, despite the recent measurement by Tytler of D/H = 2.4 × 10−5 in two high-redshift Lyman limit systems, implying Ωb ≈ 0.024h−2. The evidence would favor Ω = 1 if (1) the POTENT analysis of galaxy peculiar velocity data is right, in particular regarding outflows from voids or the inability to obtain the present-epoch non-Gaussian density distribution from Gaussian initial fluctuations in a low-Ω universe; or (2) the preliminary report from LSND indicating a neutrino mass >∼ 2.4 eV is right, since that would be too much hot dark matter to allow significant structure formation in a low-Ω0 ΛCDM model. Statistics on gravitational lensing of quasars provide a strong upper limit on Λ. It also appears to be possible to measure the deceleration parameter q0 = Ω0/2 − ΩΛ on very large scales using high-redshift Type Ia supernovae; the preliminary results suggest that Ω0 ∼ 1 and ΩΛ is small. The era of structure formation is another important discriminant between these alternatives, low Ω0 favoring earlier structure formation, and Ω = 1 favoring later formation with many clusters and larger-scale structures still forming today. Reliable data on all of these issues is becoming available so rapidly today that there is reason to hope that a clear decision between these alternatives will be possible within the next few years. ∗To appear in International School of Physics "Enrico Fermi", Course CXXXII: Dark Matter in the Universe, Varenna 1995, eds. S. Bonometto, J.R. Primack, A. Provenzale. 1 1 Introduction As I write this in spring 1996 [1], there is still concern about a crisis in cosmology. The first article [2] using HST observations of Cepheid variable stars to determine a distance to a relatively distant galaxy, the beautiful face-on spiral M100, was published about a year ago. The distance obtained was 17.1 ± 1.8 Mpc. With the additional assumptions that M100 lies in the core of the Virgo cluster and that the recession velocity of Virgo corrected for infall is about 1400 km s−1, the value obtained for the Hubble parameter is at the high end of recent estimates: H0 = 80 ± 17 km s−1 Mpc. Using h = 0.8 gives, for Ω = 1 and a vanishing cosmological constant Λ = 0, a very short age for the universe t0 = 8.15 Gyr, almost certainly younger than the ages of Milky Way globular clusters and even some nearby white dwarfs. Even with Ω0 = 0.3, about as low as permitted by observations, and with ΩΛ ≡ Λ/(3H 2 0 ) = 0.7, perhaps even higher than current data allow, t0 = 11.8 Gyr for h = 0.8, which is also uncomfortably short. Is this a crisis? Does it undermine the strong evidence for the standard Big Bang? I don't think so. Given the considerable uncertainties reflected in the large quoted error on H0, I think even Ω = 1 models are not excluded. But this Cepheid measurement of the distance to M100 bodes well for the success of the HST Key Project on the Extragalactic Distance Scale, which seeks to measure H0 to 10% within a few years. There has also been recent progress in using Type Ia supernovae as distance indicators, for measuring both H0 and the deceleration parameter q0 = Ω0/2 − ΩΛ. The expectation that accurate measurements of the key cosmological parameters will soon be available is great news for theorists trying to construct a fundamental theory of cosmology, and helps motivate the present summary. In addition to the Hubble parameter H0 ≡ 100h km s−1 Mpc−1, I will discuss the age of the universe t0, the average density Ω0, and the cosmological constant Λ. But there are several additional cosmological parameters whose values are critical for modern theories: the densities of ordinary matter Ωb, cold dark matter Ωc, and hot dark matter Ων, and, for primordial fluctuation spectra P (k) = Aknp, the index np and the amplitude A, or equivalently (for a given model) the bias parameter b ≡ 1/σ8, where σ8 ≡ (δM/M)rms on a scale of 8 h−1 Mpc. A full treatment of these parameters would take a much longer article than this one, so to focus the discussion I will concentrate on the issue of the value of the density Ω0 in currently popular cosmological models in which most of the dark matter is cold. Although much of the following discussion will be quite general, it will be helpful to focus on two specific cosmological models which are perhaps the most popular today of the potentially realistic models: low-Ω Cold Dark Matter with a Cosmological Constant (ΛCDM, discussed as an alternative to Ω = 1 CDM since the beginning of CDM [3,4], and worked out in detail in [5]), and Ω = 1 Cold + Hot Dark Matter (CHDM, proposed in 1984 [6], and first worked out in detail in 1992-3 [7,8]). I will begin by summarizing the rationale for these models. 2 2 Models with Mostly Cold Dark Matter Let me begin here by recalling the definitions of "hot" and "cold" dark matter. These terms describe the astrophysically relevant aspects of candidate dark matter particles. The fact that the observational lower bound on Ω0 -- namely 0.3 <∼ Ω0 -- exceeds the most conservative upper limit on baryonic mass Ωb <∼ 0.03h−2 from Big Bang Nucleosynthesis [9] is the main evidence that there must be such nonbaryonic dark matter particles. About a year after the big bang, the horizon surrounding any point encompassed a mass of about 1012M⊙, the mass now in the dark matter halo of a large galaxy like the Milky Way. The temperature then was about a kilovolt. We define cold dark matter as particles that were moving sluggishly, and hot dark matter as particles that were still relativistic, at that time. As Kim Griest and Antonio Masiero discussed in their Varenna lectures, the lightest superpartner particle (LSP neutralino) and the axion remain the best motivated cold dark matter candidates, although of course many other possibilities have been suggested. The three known neutrino species νe, νµ, and ντ are the standard hot dark matter candidates. Their contribution to the cosmological density today is Ων = Pi m(νi) 94h2 eV Since Ων < Ω0 <∼ 2, each neutrino's mass must be much less than a keV, so they were certainly moving at relativistic speeds a year after the big bang. Any of these neutrinos that has a cosmologically significant mass (>∼ 1 eV) is therefore a hot dark matter particle. If a horizon-sized region has slightly higher than average density at this time, cold dark matter -- moving sluggishly -- will preserve such a fluctuation. But light neutrinos -- moving at nearly the speed of light -- will damp such fluctuations by "free streaming." For example, two years after the big bang, the extra neutrinos will have spread out over the now-larger horizon. The smallest fluctuations that will not suffer this fate are those that come into the horizon when the neutrinos become nonrelativistic, i.e. when the temperature drops below the neutrino mass. In a universe in which most of the dark matter is hot, primordial fluctuations will damp on all scales up to superclusters (with mass ∼ 1016M⊙), leading to a sequence of cosmogony (cosmological structure formation) in which galaxies form only after superclusters. But this is contrary to observations, which show galaxies to be old but superclusters still forming. Indeed, with fluctuations on large scales consistent with the COBE upper limit, standard HDM models (i.e. with the dark matter being mostly neutrinos, and a Zel'dovich spectrum of Gaussian adiabatic fluctuations) cannot form a significant number of galaxies by the present. Thus most current comparisons of cosmological models with observations have focused on models in which most of the dark matter is cold. The standard CDM model [3] assumed a Zel'dovich (i.e. np = 1) spectrum of primor- dial Gaussian adiabatic fluctuations with Ω = 1. It had the great virtues of simplicity and 3 predictive power, since it had only one free parameter, the amplitude or bias b. Moreover, for a while it even looked like it agreed with all available data, with b ≈ 2.5. One early warning that all was not well for CDM was the cosmic background dipole anisotropy, indicating a large velocity of the local group with respect to the cosmic background ra- diation rest frame, about 600 km s−1. I confess that I and many other theorists did not immediately appreciate its possibly devastating impact. However, as evidence began to accumulate, starting in 1986, that there were large-scale flows of galaxies with such veloc- ities [10], it became clear that standard CDM could fit these large-scale galaxy peculiar velocities (i.e. motions in addition to the general Hubble expansion) only for b ≈ 1. Standard CDM had various problems for any value of b; for example, the CDM matter correlation function, and hence also the galaxy and cluster correlations, are negative on scales larger than about 30 h−1 Mpc, while observations on these large scales show that the cluster correlations are at least ∼ 3σ positive [11]. A low value of the bias parameter subsequently also turned out to be required by the COBE DMR data, which was first an- nounced in April 1992. But for such a small b <∼ 1, CDM produces far too many clusters and predicts small-scale galaxy velocities that are much too large [12]. Thus standard CDM does not look like a very good match to the now-abundant observational data. But it did not miss by much: if the bias parameter b is adjusted to fit the COBE data, the fluctuation amplitude is too large on small scales by perhaps a factor of ∼ 2 − 3. In the wake of the discovery of the existence of large-scale galaxy peculiar velocities, I suggested that Jon Holtzman (then a UCSC graduate student whose planned Ph.D. research based on HST observations had been indefinitely postponed by the Challenger explosion) improve the program that George Blumenthal and I had written to do linear CDM calculations, and use it to investigate a variety of models in which the dark matter was mostly cold. He ultimately worked out a total of 94 such models, about half of them including some hot dark matter, and (since this was the largest such suite of interesting models all worked out the same way) his thesis [13] provided the basis for the COBE- DMR interpretation paper [14]. Meanwhile, in a follow-up paper [15], we showed that of all these CDM-like models the ones that best fit the available data -- especially the cluster correlations -- were Ω = 1 Cold + Hot Dark Matter (CHDM), and low-Ω Cold Dark Matter with a Cosmological Constant (ΛCDM). Since both of these models turned out to fit all available data rather well when their fluctuation amplitudes were normalized to COBE observations, they remain perhaps the most popular models for galaxy formation and large scale structure. Moreover, since CHDM works best for h ≈ 0.5 while ΛCDM works best for higher h, they will serve nicely for this review as representatives of these two opposing alternatives. 3 Age of the Universe t0 The strongest lower limits for t0 come from studies of the stellar populations of globular clusters (GCs). Standard estimates of the ages of the oldest GCs are 14-18 Gyr [16], and 4 a conservative lower limit on the age of GCs is 13 ± 2 Gyr [17], which is then a lower limit on t0. The main uncertainty in the GC age estimates comes from the uncertain distance to the GCs: a 0.25 magnitude error in the distance modulus translates to a 22% error in the derived cluster age [18]. Stellar mass loss is a recent idea for lowering the GC t0[19], but observations constrain the reduction in t0 to be less than ∼ 1 Gyr. Allowing ∼ 1 − 2 Gyr for galaxy and GC formation, we conclude that t0 >∼ 11 Gyr from GCs, with t0 ≈ 13 Gyr a "likely" lower limit on t0, obtained by pushing many but not all the parameters to their limits. The GC age estimates are of course based on standard stellar evolution calculations. But the solar neutrino problem reminds us that we are not really sure that we understand how even our nearest star operates; and the sun plays an important role in calibrating stellar evolution, since it is the only star whose age we know independently (from radioac- tive dating of early solar system material). What if the GC age estimates are wrong for some unknown reason? The only independent estimates of the age of the universe come from cosmochronome- try -- the chemical evolution of the Galaxy -- and white dwarf cooling. Cosmochronom- etry age estimates are sensitive to a number of uncertain effects such as the formation history of the disk and its stars, and possible actinide destruction in stars [20]. Age esti- mates also come from the cooling of white dwarfs in the neighborhood of the sun. The key observation is that there is a lower limit to the luminosity and therefore also the tempera- ture of nearby white dwarfs; although dimmer ones could have been seen, none have been found. The only plausible explanation is that the white dwarfs have not had sufficient time to cool to lower temperatures, which initially led to an estimate of 9.3 ±2 Gyr for the age of the Galactic disk [21]. Since there is evidence that the stellar disk of our Galaxy is about 2 Gyr younger than the oldest GCs [22], this in turn gave an estimate of the age of the universe of t0 ∼ 11 ± 2 Gyr. More recent analyses [23] conclude that sensitivity to disk star formation history, and to effects on the white dwarf cooling rates due to C/O separation at crystallization and possible presence of trace elements such as 22Ne, allow a rather wide range of ages for the disk of about 10 ± 4 Gyr. The latest determination of the white dwarf luminosity function, using white dwarfs in proper motion binaries, leads to a somewhat lower minimum luminosity and therefore a somewhat higher estimate of the age of the disk of ∼ 10.5+2.5 −1.5 Gyr; it follows that t0 ≥ 11.5 Gyr [24]. Suppose that the GC stellar age estimates that t0 >∼ 13 Gyr are right. Fig. 1 shows that t0 > 13 Gyr implies that H0 ≤ 50 km s−1 Mpc−1 for Ω = 1, and that H0 ≤ 81 km s−1 Mpc−1 even for Ω0 as small as 0.2 (in flat cosmologies with Ω0 + ΩΛ = 1). 4 Hubble Parameter H0 The Hubble parameter H0 ≡ 100h km s−1 Mpc−1 remains uncertain by about a factor of two: 0.4 <∼ h <∼ 1. Sandage has long contended that h ≈ 0.5, and he still concludes [25,26] that the latest data are consistent with h = 0.55 ± 0.1. de Vaucouleurs long 5 t0 (Gy) 35 30 25 20 15 10 5 40 Ω=1.0 0 Ω =0.1 0.2 0.3 0.4 Inflation Inspired Ω + Ω = 1Λ 0 50 60 70 80 90 Hubble Parameter (km/s/Mpc) Figure 1. Age of the universe t0 as a function of Hubble parameter H0 in inflation inspired models with Ω0 + ΩΛ = 1, for several values of the present-epoch cosmological density parameter Ω0. contended that h ≈ 1. A majority of observers currently favor a value intermediate between these two extremes (recent reviews include [27,28,29,30]). The Hubble parameter has been measured in two basic ways: (A) Measuring the distance to some nearby galaxies, typically by measuring the periods and luminosities of Cepheid variables in them; and then using these "calibrator galaxies" to set the zero point in any of the several methods of measuring the relative distances to galaxies. (B) Us- ing fundamental physics to measure the distance to some distant object directly, thereby avoiding at least some of the uncertainties of the cosmic distance ladder [31]. The difficulty with method (A) is that there are so far only a handful of calibrator galaxies close enough for Cepheids to be resolved in them. However, the success of the HST Cepheid measure- ment of the distance to M100 [2] shows that the HST Key Project on the Extragalactic Distance Scale can significantly increase the set of calibrator galaxies within a few years. Adaptive optics from the ground may also be able to contribute to this effort, although I am not very impressed by the first published result of this approach [32]. The difficulty with method (B) is that in every case studied so far, some aspect of the observed system or the underlying physics remains somewhat uncertain. It is nevertheless remarkable that the results of several different methods of type (B) are rather similar, and indeed not very far from those of method (A). This gives reason to hope for convergence. 6 4.1 (A) Relative Distance Methods One piece of good news is that the several methods of measuring the relative distances to galaxies now mostly seem to be consistent with each other [28,29]. These methods use either (1) "standard candles" or (2) empirical relations between two measurable properties of a galaxy, one distance-independent and the other distance-dependent. The old favorite standard candle is Type Ia supernovae; a new one is the apparent maximum luminosity of planetary nebulae [28]. Sandage [26] and others [33] get low values of h ≈ 0.55 from HST Cepheid distances to SN Ia host galaxies, including the seven SNe Ia with well- observed maxima that lie in six galaxies with HST Cepheid distances. There are claims that taking account of an empirical relationship between the SN Ia light curve shape and maximum luminosity leads to higher h [34], but Sandage and Tammann counter that any such effect is small [35,26] since they have not used those "abnormal" SNe Ia that would be significantly affected. The old favorite empirical relation used as a relative distance indicator is the Tully-Fisher relation between the rotation velocity and luminosity of spiral galaxies (and the related Faber-Jackson or Dn − σ relation); a newer one is based on the decrease in the fluctuations in elliptical galaxy surface brightness on a given angular scale as galaxies are seen at greater distances [36]. 4.2 (B) Fundamental Physics Approaches The fundamental physics approaches involve either Type Ia or Type II supernovae, the Sunyaev-Zel'dovich (S-Z) effect, or gravitational lensing. The 56Ni radioactivity method for determining H0 using Type Ia SN avoids the uncer- tainties of the distance ladder by calculating the absolute luminosity of Type Ia supernovae from first principles using a plausible but as yet unproved physical model. The first result obtained was that h = 61 ± 10 [37]; however, another study [38] found that uncertainties in extinction (i.e., light absorption) toward each supernova increases the range of allowed h. Demanding that the 56Ni radioactivity method agree with an expanding photosphere approach leads to H0 = 60+14 −11 [39]. The expanding photosphere method compares the expansion rate of the SN envelope measured by redshift with its size increase inferred from its temperature and magnitude. This approach was first applied to Type II SN; the 1992 result h = 0.6 ± 0.1 [40] was subsequently revised upward by the same authors to h = 0.73 ± 0.06 ± 0.07 [41]. However, there are various complications with the physics of the expanding envelope [42]. The S-Z effect is the Compton scattering of microwave background photons from the hot electrons in a foreground galaxy cluster. This can be used to measure H0 since properties of the cluster gas measured via the S-Z effect and from X-ray observations have different dependences on H0. The result from the first cluster for which sufficiently detailed data was available, A665 (at z = 0.182), was h = (0.4−0.5)±0.12 [43]; combining this with data on A2218 (z = 0.171) raises this somewhat to h = 0.55 ± 0.17 [44]. Early results from the ASCA X-ray satellite gave h = 0.47 ± 0.17 for A665 (z = 0.182) and 7 h = 0.41+0.15 −0.12 for CL0016+16 (z = 0.545) [45]. A few S-Z results have been obtained using millimeter-wave observations, and this promising method should allow many more such measurements soon [46]. Corrections for the near-relativistic electron motions will raise these estimates for H0 a little [47], but it seems clear that the S-Z results favor a smaller value than many optical astronomers obtain. However, since the S-Z measurement of H0 is affected by the orientation of the cluster ellipticity with respect to the line of sight, this will only become convincing with observations of a significant number of additional clusters. Fortunately, this now appears to be possible within the next several years. Several quasars have been observed to have multiple images separated by a few arc seconds; this phenomenon is interpreted as arising from gravitational lensing of the source quasar by a galaxy along the line of sight. In the first such system discovered, QSO 0957+561 (z = 1.41), the time delay ∆t between arrival at the earth of variations in the quasar's luminosity in the two images has been measured to be 409 ± 23 days [48], although other authors found a value of 540 ± 12 days [49]. The shorter ∆t has now been confirmed by the observation of a sharp drop in Image A of about 0.1 mag in late December 1994 [50], followed by a similar drop in Image B about 405-420 days later (R. Schild and E.L. Turner, private communications). Since ∆t ≈ θ2H −1 0 , this observation allows an estimate of the Hubble parameter, with the results h = 0.50 ± 0.17 [51], or h = 0.63 ± 0.21 (h = 0.42 ± 0.14) including (neglecting) dark matter in the lensing galaxy [52], with additional uncertainties associated with possible microlensing and unknown matter distribution in the lensing galaxy. However, recent deep images have allowed mapping of the gravitational potential of the cluster (at z = 0.36) in which the lensing galaxy lies using weak gravitational lensing, which leads to the conclusion that h ≤ 0.70(1.1yr/∆t) [53]. Also, detailed study of the lensed QSO images (which include a jet) constrains the lensing and implies h = 0.82(1 − κ)(1.1yr/∆t) < 0.82, where the upper limit follows because the convergence due to the cluster κ > 0, or alternatively h = 0.82(σ/322 km s−1)2(1.1yr/∆t) without uncertainty concerning the cluster if the one- dimensional velocity dispersion σ in the core of the giant elliptical galaxy responsible for the lensing can be measured [54]. Although the uncertainty in H0 remains rather large, it is reassuring that this method gives results consistent with the other determinations. The time-delay method is promising, and when delays are reliably measured in several other multiple-image quasar systems, that should lead to a reliable value for h. 4.3 Correcting for Virgocentric Infall What about the recent HST Cepheid measurement of H0, giving h ≈ 0.8 [2]? This calculated value is based on neither of the two methods (A) or (B) above, and I do not regard it as being very reliable. Instead this result is obtained by assuming that M100 is at the core of the Virgo cluster, and dividing the sum of the recession velocity of Virgo, about 1100 km s−1, plus the calculated "infall velocity" of the local group toward Virgo, about 300 km s−1, by the measured distance to M100 of 17.1 Mpc. (These recession and infall velocities are both a little on the high side, compared to other values one finds in the 8 literature.) Adding the "infall velocity" is necessary in this method in order to correct the Virgo recession velocity to what it would be were it not for the gravitational attraction of Virgo for the Local Group of galaxies, but the problem with this is that the net motion of the Local Group with respect to Virgo is undoubtedly affected by much besides the Virgo cluster -- e.g., the "Great Attractor." For example, in our CHDM supercomputer simulations (which appear to be a rather realistic match to observations) Anatoly Klypin and I have found that galaxies and groups at about 20 Mpc from a Virgo-sized cluster often have net outflowing rather than infalling velocities. Note that if the net "infall" of M100 were smaller, or if M100 were in the foreground of the Virgo cluster (in which case the actual distance to Virgo would be larger than 17.1 Mpc), then the indicated H0 would be smaller. The authors of Ref. [2] gave an alternative argument that avoids the "infall velocity" uncertainty: the relative galaxy luminosities indicate that the Coma cluster is about six times farther away than the Virgo cluster, and peculiar motions of the Local Group and the Coma cluster are relatively small corrections to the much larger recession velocity of Coma; dividing the recession velocity of the Coma cluster by six times the distance to M100 again gives H0 ≈ 80. However, this approach still assumes that M100 is in the core rather than the foreground of the Virgo cluster; and in deducing the relative distance of the Coma and Virgo clusters it assumes that the galaxy luminosity functions in each are comparable, which is uncertain in view of the very different environments. More general arguments by the same authors [55] lead them to conclude that H0 = 73 ± 11 regardless of where M100 lies in the Virgo cluster. But Tammann et al. [26], using all the available HST Cepheid distances and their own complete sample of Virgo spirals, conclude that H0 ≈ 54. To summarize, many observers, using mainly method (A), favor a value h ≈ 0.6 − 0.8 although Sandage's group and some others continue to get h ≈ 0.5 − 0.6, while the methods I have grouped together as (B) typically lead to h ≈ 0.4 − 0.7. The fact that the latter measurements are mostly of more distant objects has suggested [56] that the local universe may actually be underdense and therefore be expanding faster than is typical. But in reasonable models where structure forms from Gaussian fluctuations via gravitational instability, it is extremely unlikely that a sufficiently large region has a density sufficiently smaller than average to make more than a rather small difference in the measured value of h [57]. Moreover, the small dispersion in the corrected maximum luminosity of distant SNe Ia found by the LBL Supernova Cosmology Project [58] compared to nearby SNe Ia shows directly that the local and cosmological values of H0 are approximately equal. There has been recent observational progress in both methods (A) and (B), and I think it likely that the Hubble parameter will be known reliably to 10% within a few years. But at present the uncertainty remains rather large 0.4 < h < 0.8, and we must keep an open mind. 9 5 Cosmological Constant Λ, and t0 Again Inflation is the only known solution to the horizon and flatness problems and the avoidance of too many GUT monopoles. And inflation has the added bonus that with no extra charge (except the perhaps implausibly fine-tuned adjustment of the self-coupling of the inflaton field to be adequately small), simple inflationary models predict a near- Zel'dovich spectrum (i.e., with np ≈ 1) of adiabatic Gaussian primordial fluctuations -- which seems to be consistent with observations. All simple inflationary models predict that the curvature constant k is vanishingly small, although inflationary models that are extremely contrived (at least, to my mind) can be constructed with negative curvature and therefore Ω0 <∼ 1 without a cosmological constant [59]. Thus most authors who consider inflationary models impose the condition k = 0, or Ω0 + ΩΛ = 1 where ΩΛ ≡ Λ/(3H 2 0 ). This is what is assumed in ΛCDM models, and it is what was assumed in Fig. 1. (I hope it has been clear from the foregoing that I use Ω to refer only to the density of matter and energy, not including the cosmological constant, whose contribution in the Ω units is ΩΛ.) I know of no one (except possibly Lev Kofman) who actually finds the idea of a nonvanishing Λ intrinsically attractive. There is no known physical reason why Λ should be so small (from the viewpoint of particle physics), though there is also no known reason why it should vanish. The most unattractive features of Λ 6= 0 cosmologies are the fact that Λ must become important only at relatively low redshift -- why not much earlier or much later? -- and also that ΩΛ >∼ Ω0 implies that the universe has recently entered an inflationary epoch (with a de Sitter horizon comparable to the present horizon). The main motivations for Λ > 0 cosmologies are (1) reconciling inflation with observations that seem to imply Ω0 <∼ 1, and (2) avoiding a contradiction between the lower limit t0 >∼ 13 Gyr from globular clusters and t0 = (2/3)H −1 0 = 6.52h−1 Gyr for the standard Ω = 1, Λ = 0 Einstein-de Sitter cosmology, if it is really true that h > 0.5. The cosmological effects of a cosmological constant are not difficult to understand [60,61]. With a positive Λ, there is a repulsion of space by space. In the early universe, the density of energy and matter is far more important than Λ on the r.h.s. of the Friedmann equation. But the average matter density decreases as the universe expands, and at a rather low redshift z ∼ 0.3 the Λ term finally becomes dominant. If it has been adjusted just right, Λ can almost balance the attraction of the matter, and the expansion nearly stops: for a long time, the scale factor a ≡ (1 + z)−1 increases very slowly, although it ultimately starts increasing exponentially as the universe starts inflating under the influence of the increasingly dominant Λ term. The existence of a period during which expansion slows while the clock runs explains why t0 can be greater than for Λ = 0, but this also shows that there is a increased likelihood of finding galaxies at the redshift interval when the expansion slowed, and a correspondingly increased opportunity for lensing of quasars (which mostly lie at higher redshift z >∼ 2) by these galaxies. The frequency of such lensed quasars is about what would be expected in a standard 10 Ω = 1, Λ = 0 cosmology, so this data sets fairly stringent upper limits: ΩΛ ≤ 0.70 at 90% C.L. [62,63], with more recent data giving even tighter constraints: ΩΛ < 0.66 at 95% confidence [64]. This limit could perhaps be weakened if there were (a) significant extinction by dust in the E/S0 galaxies responsible for the lensing or (b) rapid evolution of these galaxies, but there is much evidence that these galaxies have little dust and have evolved only passively for z <∼ 1 [65]. A weaker but independent constraint comes from the cosmic background radiation data. In standard Ω = 1 models, the quantity ℓ(ℓ + 1)Cℓ (where Cℓ =< a2 ℓ,m >m is the average of squared coefficients of the spherical harmonic expansion of the CMB data) is predicted to be roughly constant for 2 ≤ ℓ <∼ 10 (with an increase for higher multiples toward the Doppler peak at ℓ ∼ 200), while in models with Λ > 0 ℓ(ℓ + 1)Cℓ is predicted to dip before rising toward the Doppler peak. Comparison with the two-year COBE data, in which such a dip is not seen, implies that ΩΛ ≤ 0.78 at the 90% C.L. [66]. Yet another constraint comes from number counts of bright E/S0 galaxies in HST images [67], since these galaxies appear to have evolved rather little since z ∼ 1 [65]. The number counts are just as expected in the Ω = 1, Λ = 0 Einstein-de Sitter cosmology. Even allowing for uncertainties due to evolution and merging of these galaxies, this data would allow ΩΛ as large as 0.8 in flat cosmologies only in the unlikely event that half the Sa galaxies in the deep HST images were misclassified as E/S0. This approach may be very promising for the future, as the available deep HST image data and our understanding of galaxy evolution both increase. A model-dependent constraint comes from a detailed simulation of ΛCDM [68]: a COBE-normalized model with Ω0 = 0.3 and h = 0.7 has far too much power on small scales to be consistent with observations, even allowing for seemingly unrealistic antibiasing of galaxies with respect to dark matter. For ΛCDM models, the only solution appears to be raising Ω0, lowering H0, and tilting the spectrum (np < 1) [68], though of course one could alternatively modify the primordial power spectrum in other ways. Fig. 1 shows that with ΩΛ ≤ 0.7, the cosmological constant does not lead to a very large increase in t0 compared to the Einstein-de Sitter case, although it may still be enough to be significant. For example, the constraint that t0 ≥ 13 Gyr requires h ≤ 0.5 for Ω = 1 and Λ = 0, but this becomes h ≤ 0.70 for flat cosmologies with ΩΛ ≤ 0.66. 6 Measuring Ω0 6.1 Very Large Scale Measurements Although it would be desirable to measure Ω0 and Λ through their effects on the large-scale geometry of space-time, this has proved difficult in practice since it requires comparing objects at higher and lower redshift, and it is hard to separate the effects of the evolution of the objects from those of the evolution of the universe. For example, in "redshift-volume" tests involving number counts of galaxies per redshift interval, how 11 can we tell whether the galaxies at redshift z ∼ 1 correspond to those at z ∼ 0? Several galaxies at higher redshift might have merged, and galaxies might have formed or changed luminosity at lower redshift. Eventually, with extensive surveys of galaxy properties as a function of redshift using the largest telescopes such as Keck, it should be possible to perform these classical cosmological tests at least on a particular class of galaxies -- that is one of the goals of the Keck DEEP project. At present, perhaps the most promising technique involves searching for Type Ia su- pernovae (SNe Ia) at high-redshift, since these are the brightest supernovae and the spread in their intrinsic brightness appears to be relatively small. Saul Perlmutter, Gerson Gold- haber, and collaborators have recently demonstrated the feasibility of finding significant numbers of such supernovae [69], but a dedicated campaign of follow-up observations of each one will be required in order to measure Ω0 by determining how the apparent bright- ness of the supernovae depends on their redshift. This is therefore a demanding project. It initially appeared that ∼ 100 high redshift SNe Ia would be required to achieve a 10% measurement of q0 = Ω0/2−ΩΛ. However, using the correlation mentioned earlier between the absolute luminosity of a SN Ia and the shape of its light curve (e.g., slower decline correlates with higher peak luminosity), it now appears possible to reduce the number of SN Ia required. The Perlmutter group has now analyzed seven high redshift SN Ia by this method, with a tentative result that q0 ∼ 0.5 and in any case q0 >∼ 0 [70]. In November 1995 they discovered an additional 11 high-redshift SN Ia, and they have just discov- ered an additional seven. Preliminary results from the next 11 SN Ia are consistent with q0 ∼ 0.5, and there are no peculiar or anomalous ones in the dataset (Perlmutter, private communication). Two other groups, collaborations from ESO and MSSSO/CfA/CTIO, are also searching for high-redshift supernovae to measure q0, and have found at least a few. There has also been recent progress understanding the physical origin of the SN Ia luminosity-light curve correlation. At the present rate of progress, a reliable answer may be available within perhaps a year or so if a consensus emerges from these efforts. 6.2 Large-scale Measurements Ω0 has been measured with some precision on a scale of about ∼ 50 h−1 Mpc, using the data on peculiar velocities of galaxies, and on a somewhat larger scale using redshift surveys based on the IRAS galaxy catalog. Since the results of all such measurements to date have recently been reviewed in detail [71], I will only comment briefly on them. The analyses such as "POTENT" that try to recover the scalar velocity potential from the galaxy peculiar velocities are looking increasingly reliable, since they reproduce the observed large scale distribution of galaxies -- that is, many galaxies are found where the converging velocities indicate that there is a lot of matter, and there are voids in the galaxy distribution where the diverging velocities indicate that the density is lower than average. The comparison of the IRAS redshift surveys with POTENT and related analyses typically give fairly large values for the parameter βI ≡ Ω0.6 0 /bI (where bI is the biasing parameter for IRAS galaxies), corresponding to 0.3 <∼ Ω0 <∼ 3 (for an assumed bI = 1.15). It is not 12 clear whether it will be possible to reduce the spread in these values significantly in the near future -- probably both additional data and a better understanding of systematic and statistical effects will be required. A particularly simple way to deduce a lower limit on Ω0 from the POTENT peculiar velocity data has recently been proposed [72], based on the fact that high-velocity outflows from voids are not expected in low-Ω models. Data on just one void indicates that Ω0 ≥ 0.3 at the 97% C.L. This argument is independent of assumptions about Λ or galaxy formation, but of course it does depend on the success of POTENT in recovering the peculiar velocities of galaxies. However, for the particular cosmological models that I am focusing on in this review -- CHDM and ΛCDM -- stronger constraints are available. This is because these models, in common with almost all CDM variants, assume that the probability distribution function (PDF) of the primordial fluctuations was Gaussian. Evolution from a Gaussian initial PDF to the non-Gaussian mass distribution observed today requires considerable gravitational nonlinearity, i.e. large Ω. The PDF deduced by POTENT from observed velocities (i.e., the PDF of the mass, if the POTENT reconstruction is reliable) is far from Gaussian today, with a long positive-fluctuation tail. It agrees with a Gaussian initial PDF if and only if Ω is about unity or larger: Ω0 < 1 is rejected at the 2σ level, and Ω0 ≤ 0.3 is ruled out at ≥ 4σ [73]. 6.3 Measurements on Scales of a Few Mpc On smaller length scales, there are many measurements that are consistent with a smaller value of Ω0 [74]. For example, the cosmic virial theorem gives Ω(∼ 1h−1 Mpc) ≈ 0.15[σ(1h−1 Mpc)/(300 km s−1)]2, where σ(1h−1 Mpc) here represents the relative velocity dispersion of galaxy pairs at a separation of 1h−1 Mpc. Although the classic paper [75] which first measured σ(1h−1 Mpc) using a large redshift survey (CfA1) got a value of 340 km s−1, this result is now known to be in error since the entire core of the Virgo cluster was inadvertently omitted [76]; if Virgo is included, the result is ∼ 500 − 600 km s−1 [77,76], corresponding to Ω(∼ 1h−1 Mpc) ≈ 0.4 − 0.6. Various redshift surveys give a wide range of values for σ(1h−1 Mpc) ∼ 300 − 750 km s−1, with the most salient feature being the presence or absence of rich clusters of galaxies; for example, the IRAS galaxies, which are not found in clusters, have σ(1h−1 Mpc) ≈ 320 km s−1 [78], while the northern CfA2 sample, with several rich clusters, has much larger σ than the SSRS2 sample, with only a few relatively poor clusters [79,76,80]. It is evident that the σ(1h−1 Mpc) statistic is not a very robust one. A standard method for estimating Ω on scales of a few Mpc is based on applying virial estimates to groups and clusters of galaxies to try to deduce the total mass of the galaxies including their dark matter halos from the velocities and radii of the groups; roughly, GM ∼ rv2. (What one actually does is to assume that all galaxies have the same mass-to-light ratio M/L, given by the median M/L of the groups, and integrate over the luminosity function to get the mass density [81,82,83].) The typical result is 13 that Ω(∼ 1 h−1 Mpc) ∼ 0.1 − 0.2. However, such estimates are at best lower limits, since they can only include the mass within the region where the galaxies in each group can act as test particles. In CHDM simulations, my colleagues and I [84] have found that the effective radius of the dark matter distribution associated with galaxy groups is typically 2-3 times larger than that of the galaxy distribution. Moreover, we find a velocity biasing [85] factor in CHDM groups bgrp v ≡ vgal,rms/vDM,rms ≈ 0.75, whose inverse squared enters in the Ω estimate. Finally, we find that groups and clusters are typically elongated, so only part of the mass is included in spherical estimators. These factors explain how it can be that our Ω = 1 CHDM simulations produce group velocities that are fully consistent with those of observed groups, even with statistical tests such as the median rms group velocity vs. the fraction of galaxies grouped [86,84]. This emphasizes the point that local estimates of Ω are at best lower limits on its true value. Another approach to estimating Ω from information on relatively small scales has been pioneered by Peebles [87]. It is based on using the least action principle (LAP) to reconstruct the trajectories of the Local Group galaxies, and the assumption that the mass is concentrated around the galaxies. This is a reasonable assumption in a low-Ω universe, but it is not at all what must occur in an Ω = 1 universe where most of the mass must lie between the galaxies. Although comparison with Ω = 1 N-body simulations showed that the LAP often succeeds in qualitatively reconstructing the trajectories, the mass is systematically underestimated by a large factor by the LAP method [88]. Unexpectedly, a different study [89] found that the LAP method underestimates Ω by a factor of 4-5 even in an Ω0 = 0.2 simulation; the authors say that this discrepancy is due to the LAP neglecting the effect of "orphans" -- dark matter particles that are not members of any halo. Shaya, Peebles, and Tully [90] have recently attempted to apply the LAP to galaxies in the local supercluster, again getting low Ω0. The LAP approach should be more reliable on this larger scale, but the method still must be calibrated on N-body simulations of both high- and low-Ω0 models before its biases can be quantified. 6.4 Estimates on Galaxy Halo Scales Recent work by Zaritsky and White [91] and collaborators has shown that spiral galax- ies have massive halos. A classic paper by Little and Tremaine [97] argued that the avail- able data on the Milky Way satellite galaxies required that the Galaxy's halo terminate at about 50 kpc, with a total mass of only about 2.5 × 1011M⊙. But by 1991, new data on local satellite galaxies, especially Leo I, became available, and the Little-Tremaine es- timator increased to 1.25 × 1012M⊙. A recent, detailed study finds a mass inside 50 kpc of (5.4 ± 1.3) × 1011M⊙ [98]. Zaritsky and collaborators have collected data on satellites of isolated spiral galaxies, and conclude that the fact that the relative velocities do not fall off out to a separation of at least 200 kpc shows that massive halos are the norm. The typical rotation velocity of ∼ 200 − 250 km s−1 implies a mass within 200 kpc of 2 × 1012M⊙. A careful analysis taking into account selection effects and satellite orbit uncertainties concluded that the indicated value of Ω0 exceeds 0.13 at 90% confidence, 14 with preferred values exceeding 0.3 [91]. Newer data suggesting that relative velocities do not fall off out to a separation of at least 300 kpc will raise these Ω0 estimates [92]. However, if galaxy dark matter halos are really so extended and massive, that would imply that when such galaxies collide, the resulting tidal tails of debris cannot be flung very far. Therefore, the observed merging galaxies with extended tidal tails such as NGC 4038/39 (the Antennae) and NGC 7252 probably have halo:(disk+bulge) mass ratios less than 10:1 [93], unless the stellar tails are perhaps made during the collision process from gas that was initially far from the central galaxies (J. Ostriker, private communication, 1996); the latter possibility can be checked by determining the ages of the stars in these tails. A direct way of measuring the mass and spatial extent of many galaxy dark matter halos is to look for the small distortions of distant galaxy images due to gravitational lensing by foreground galaxies. This technique was pioneered by Tyson et al. [94]. Though the results were inconclusive, powerful constraints could perhaps be obtained from deep HST images or ground-based images with excellent seeing. Such fields would also be useful for measuring the correlated distortions of galaxy images from large-scale structure by weak gravitational lensing; although a pilot project [95] detected only a marginal signal, a reanalysis detected a significant signal suggesting that Ω0σ8 ∼ 1 [96]. Several groups are planning major projects of this sort. 7 Clusters 7.1 Cluster Baryons vs. Big Bang Nucleosynthesis A recent review [9] of Big Bang Nucleosynthesis (BBN) and observations indicating primordial abundances of the light isotopes concludes that 0.009h−2 ≤ Ωb ≤ 0.02h−2 for concordance with all the abundances, and 0.006h−2 ≤ Ωb ≤ 0.03h−2 if only deuterium is used. For h = 0.5, the corresponding upper limits on Ωb are 0.08 and 0.12, respectively. The recent observations [99] of a possible deuterium line in a hydrogen cloud at redshift z = 3.32 indicating a deuterium abundance of ∼ 2 × 10−4 (and therefore Ωb ≤ 0.006h−2) are contradicted by similar observations by Tytler and collaborators [100] in systems at z = 3.57 and z = 2.504 but with a deuterium abundance about ten times lower, consistent with solar system measurements of D and 3He and implying Ωbh2 = 0.024 ± 0.05, or Ωb in the range 0.08-0.11 for h = 0.5. (If these represent the true D/H, then the earlier observations [99] were most probably of a Lyα forest line. However, Rugers and Hogan [101] argue that the width of their z = 3.32 absorption features is better fit by deuterium, although they admit that only a statistical sample of absorbers will settle the issue. There is a new possible detection of D at z = 4.672 in the absorption spectrum of QSO BR1202- 0725 [102] and at Z = 3.086 toward Q 0420-388 [103]. But Tytler [100] argues that the two systems he and his colleagues have analyzed are much more convincing as real 15 detections of deuterium, and it is surely significant that they measure the same D/H in both systems.) White et al. [104] have emphasized that recent X-ray observations of clusters, es- pecially Coma, show that the abundance of baryons, mostly in the form of gas (which typically amounts to several times the total mass of the cluster galaxies), is about 20% if h is as low as 0.5. For the Coma cluster they find that the baryon fraction within the Abell radius is fb ≡ Mb Mtot ≥ 0.009 + 0.050h−3/2, 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 = 0.3(cid:18) Ωb fb ! . 0.06(cid:19) 0.2 A recent review of gas in a sample of clusters [105] 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 [104] (cf. also [106]). The gas mass would also be overestimated if large tangled magnetic fields provide a significant part of the pressure in the central regions of some clusters [107]; this can be checked by observation of Faraday rotation of sources behind clusters [108]. 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 [109] of the gas at the measured X-ray temperature (which is surprising since they agree [110]), or (c) that the BBN upper limit on Ωb is wrong. It is interesting that there are indications from weak lensing [111] 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 [84,112,113]. 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 higher than usually estimated). It is of course possible that the solution is some combination of alternatives (a), (b), and (c). If none of the alternatives is right, then the only conclusion left is that Ω0 ≈ 0.3. The cluster baryon problem is clearly an issue that deserves very careful examination. 7.2 Cluster Morphology Richstone, Loeb, and Turner [114] showed that clusters are expected to be evolved -- i.e. rather spherical and featureless -- in low-Ω cosmologies, in which structures form at 16 relatively high redshift, and that clusters should be more irregular in Ω = 1 cosmologies, where they have formed relatively recently and are still undergoing significant merger activity. There are very few known clusters that seem to be highly evolved and relaxed, and many which are irregular -- some of which are obviously undergoing mergers now or have recently done so (see e.g. [115]). This disfavors low-Ω models, but it remains to be seen just how low. Recent papers have addressed this. In one [116] a total of 24 CDM simulations with Ω = 1 or 0.2, the latter with ΩΛ = 0 or 0.8, were compared with data on a sample of 57 clusters. The conclusion was that clusters with the observed range of X-ray morphologies are very unlikely in the low-Ω cosmologies. However, these simulations have been criticized because the Ω0 = 0.2 ones included rather a large amount of ordinary matter: Ωb = 0.1. (This is unrealistic both because h ≈ 0.8 provides the best fit for Ω0 = 0.2 CDM, but then the standard BBN upper limit is Ωb < 0.02h−2 = 0.03; and also because observed clusters have a gas fraction of ∼ 0.15(h/0.5)−3/2.) Another study [117] using dissipationless simulations and not comparing directly to observational data found that ΛCDM with Ω0 = 0.3 and h = 0.75 produced clusters with some substructure, perhaps enough to be observationally acceptable. Clearly, this important issue deserves study with higher resolution hydrodynamic simulations, with a range of assumed Ωb, and possibly including at least some of the additional physics associated with the galaxies which must produce the metallicity observed in clusters, and perhaps some of the heat as well. Better statistics for comparing simulations to data may also be useful [118]. 7.3 Cluster Evolution There is evidence for strong evolution of clusters at relatively low redshift, both in their X-ray properties [119] and in the properties of their galaxies. In particular, there is a strong increase in the fraction of blue galaxies with increasing redshift (the "Butcher- Oemler effect"), which may be difficult to explain in a low-density universe [121]. Field galaxies do not appear to show such strong evolution; indeed, a recent study concludes that over the redshift range 0.2 ≤ z ≤ 1.0 there is no significant evolution in the number density of "normal" galaxies [120]. This is compatible with the predictions of CHDM with two neutrinos sharing a total mass of about 5 eV [122] (see below). 8 Early Structure Formation In linear theory, adiabatic density fluctuations grow linearly with the scale factor in an Ω = 1 universe, but more slowly if Ω < 1 with or without a cosmological constant [74]. As a result, if fluctuations of a certain size in an Ω = 1 and an Ω0 = 0.3 theory are equal in amplitude at the present epoch (z = 0), then at higher redshift the fluctuations in the low-Ω model had higher amplitude. Thus, structures typically form earlier in low-Ω models than in Ω = 1 models. 17 Since quasars are seen at the highest redshifts, they have been used to try to constrain Ω = 1 theories, especially CHDM which because of the hot component has additional sup- pression of small-scale fluctuations that are presumably required to make early structure (e.g., [123]). The difficulty is that dissipationless simulations predict the number density of halos of a given mass as a function of redshift, but not enough is known about the na- ture of quasars -- for example, the mass of the host galaxy -- to allow a simple prediction of the number of quasars as a function of redshift in any given cosmological model. A recent study [124] concludes that very efficient cooling of the gas in early structures, and angular momentum transfer from it to the dark halo, allows for formation of at least the observed number of quasars even in models where most galaxy formation occurs late. Observers are now beginning to see significant numbers of what appear to be the central regions of galaxies in an early stage of their formation at redshifts z = 3 − 3.5 [125] -- although, as with quasars, a danger in using systems observed by emission is that they may not be typical. As additional observations clarify the nature of these objects, they can perhaps be used to constrain cosmological parameters and models. Another sort of high redshift object which may hold more promise for constraining theories is damped Lyman α systems (DLAS). DLAS are dense clouds of neutral hydrogen, generally thought to be protogalactic disks, which are observed as wide absorption features in quasar spectra [126]. They are relatively common, seen in roughly a third of all quasar spectra, so statistical inferences about DLAS are possible. At the highest redshift for which data is published, z = 3 − 3.4, the density of neutral gas in such systems in units of critical density is Ωgas ≈ 0.6%, comparable to the total density of visible matter in the universe today [127]. Several recent papers [128] pointed out that the CHDM model with Ων = 0.3 could not produce such a high Ωgas. However, my colleagues and I showed that CHDM with Ων = 0.2 could do so [129], since the power spectrum on small scales is a very sensitive function of the total neutrino mass in CHDM models. This theory makes two crucial predictions [129]: Ωgas must fall off at higher redshifts z, and the DLAS at z >∼ 3 correspond to systems of internal rotation velocity or velocity dispersion less than about 100 km s−1 (this can be measured from the Doppler widths of the metal line systems associated with the DLAS). Preliminary reports regarding the amount of neutral hydrogen in such systems deduced from the latest data at redshifts above 3.5 appear to be consistent with these predictions [130]. But a possible problem is the large velocity widths of the metal line systems associated with the highest-redshift DLAS yet reported [131], at z = 4.4; if these actually indicate that a massive disk galaxy is already formed at such a high redshift, and if discovery of other such systems shows that they are not rare, that would certainly disfavor CHDM and other Ω = 1 theories with relatively little power on small scales. Other interpretations of such data which would not cause such problems for theories like CHDM are perhaps more plausible, though [132]. One of the best ways of probing early structure formation would be to look at the main light output of the stars of the earliest galaxies, which is redshifted by the expansion of the universe to wavelengths beyond about 5 microns today. Unfortunately, it is not 18 possible to make such observations with existing telescopes; since the atmosphere blocks almost all such infrared radiation, what is required is a large infrared telescope in space. The Space Infrared Telescope Facility (SIRTF) has long been a high priority, and it would be great to have access to the data such an instrument would produce. In the meantime, an alternative method is to look for the starlight from the earliest stars as extragalactic background infrared light (EBL). Although it is difficult to see this background light directly because our Galaxy is so bright in the near infrared, it may be possible to detect it indirectly through its absorption of TeV gamma rays (via the process γ γ → e+ e−). Of the more than twenty active galactic nuclei (AGNs) that have been seen at ∼ 10 GeV by the EGRET detector on the Compton Gamma Ray Observatory, only two of the nearest, Mk421 and Mk501, have also been clearly detected in TeV gamma rays by the Whipple Atmospheric Cerenkov Telescope [133]. Absorption of ∼ TeV gamma rays from (AGNs) at redshifts z ∼ 0.2 has been shown to be a sensitive probe of the EBL and thus of the era of galaxy formation [134]. 9 Neutrino mass There are several experiments which suggest that neutrinos have mass. In particu- lar, the recent announcement of the observation of ¯νµ → ¯νe oscillations at the Liquid Scintillator Neutrino Detector (LSND) experiment at Los Alamos suggests that δm2 ≡ m(νµ)2 − m(νe)2 ≈ 6 eV2 [135], and the observation of the angular dependence of the atmospheric muon neutrino deficit at Kamiokande [136] suggests νµ → ντ oscillations are occurring with an oscillation length comparable to the depth of the atmosphere, which requires that the muon and tau neutrinos have approximately the same mass. If, for ex- ample, m(νe) ≪ m(νµ), then this means that m(νµ) ≈ m(ντ ) ≈ 2.4 eV [137,138]. Clearly, discovery of neutrino mass in the few eV range favors CHDM; and, as I mentioned above, this total neutrino mass of about 5 eV is just what seems to be necessary to fit the large scale structure observations [129]. Dividing the mass between two neutrino species re- sults in somewhat lower fluctuation amplitude on the scale of clusters of galaxies because of the longer neutrino free streaming length, which improves agreement between CHDM normalized to COBE and observations of cluster abundance [137]. Of course, one cannot prove a theory since contrary evidence may always turn up. But one can certainly disprove theories. The minimum neutrino mass required by the preliminary LSND result [135] δm2 = 6 eV2 is 2.4 eV. This is too much hot dark matter to permit significant structure formation in a low-Ω universe; for example, in a ΛCHDM model with Ω0 = 0.3, the cluster number density is more than two orders of magnitude lower than observations indicate [137]. Thus if this preliminary LSND result is correct, it implies a strong lower limit on Ω0, and a corresponding upper bound on Λ, in ΛCDM models that include light neutrinos. 19 10 Conclusions The main issue that I have tried to address is the value of the cosmological density parameter Ω. Strong arguments can be made for Ω0 ≈ 0.3 (and models such as ΛCDM) or for Ω = 1 (for which the best class of models that I know about is CHDM), but it is too early to tell for sure which is right. The evidence would favor a small Ω0 ≈ 0.3 if (1) the Hubble parameter actually has the high value H0 ≈ 75 favored by many observers, and the age of the universe t0 ≥ 13 Gyr; or (2) the baryonic fraction fb = Mb/Mtot in clusters is actually ∼ 15%, about 3 times larger than expected for standard Big Bang Nucleosynthesis in an Ω = 1 universe. This assumes that standard BBN is actually right in predicting that the density of ordinary matter Ωb lies in the range 0.009 ≤ Ωbh2 ≤ 0.02. High-resolution high-redshift spectra are now providing important new data on primordial abundances of the light isotopes that should clarify the reliability of the BBN limits on Ωb. If the systematic errors in the 4He data are larger than currently estimated, then it may be wiser to use the deuterium upper limit Ωbh2 ≤ 0.03, which is also consistent with the value Ωbh2 ≈ 0.024 indicated by the only clear deuterium detection at high redshift, with the same D/H≈ 2.4 × 10−5 observed in two different low-metallicity quasar absorption systems [100]; this considerably lessens the discrepancy between fb and Ωb. Another important constraint on Ωb will come from the new data on small angle CMB anisotropies -- in particular, the height of the first Doppler peak [139], with the latest data consistent with low h ≈ 0.5 and high Ωb ≈ 0.1. The evidence would favor Ω = 1 if (1) the POTENT analysis of galaxy peculiar velocity data is right, in particular regarding outflows from voids or the inability to obtain the present-epoch non-Gaussian density distribution from Gaussian initial fluctuations in a low-Ω universe; or (2) the preliminary report from LSND indicating a neutrino mass ≥ 2.4 eV is right, since that would be too much hot dark matter to allow significant structure formation in a low-Ω ΛCDM model. The statistics of gravitational lensing of quasars is incompatible with large cosmological constant Λ and low cosmological density Ω0. Discrimination between models may improve as additional examples of lensed quasars are searched for in large surveys such as the Sloan Digital Sky Survey. It now appears to be possible to measure the deceleration parameter q0 = Ω0/2 − ΩΛ on very large scales using the objects that may be the best bright standard candles: high- redshift Type Ia supernovae. It is very encouraging that the Perlmutter group [70] now has discovered ∼ 25 high-redshift Type Ia supernovae, that other groups are also succeeding in finding such supernovae, and that a theoretical understanding of the empirical correlation between SN Ia light curve shape and maximum luminosity may be emerging. If the high value q0 ∼ 0.5 in the preliminary report [70] is right, ΩΛ is probably small and Ω0 ∼ 1. The era of structure formation is another important discriminant between these alter- natives, low Ω favoring earlier structure formation, and Ω = 1 favoring later formation with many clusters and larger-scale structures still forming today. A particularly critical 20 test for models like CHDM is the evolution as a function of redshift of Ωgas in damped Lyα systems. Reliable data on all of these issues is becoming available so rapidly today that there is reason to hope that a clear decision between these alternatives will be possible within the next few years. What if the data ends up supporting what appear to be contradictory possibilities, e.g. large Ω0 and large H0? Exotic initial conditions (e.g. "designer" primordial fluctuation spectra) or exotic dark matter particles beyond the simple "cold" vs. "hot" alternatives (e.g. decaying intermediate mass neutrinos) could increase the space of possible inflation- ary theories somewhat. But it may ultimately be necessary to go outside the framework of inflationary cosmological models and consider models with large scale spatial curvature, with a fairly large Λ as well as large Ω0. This seems particularly unattractive, since in addition to implying that the universe is now entering a final inflationary period, it means that inflation did not happen at the beginning of the universe, when it would solve the flatness, horizon, monopole, and structure generation problems. Therefore, along with most cosmologists, I am rooting for the success of inflation-inspired cosmologies, with Ω0 + ΩΛ = 1. With the new upper limits on Λ from gravitational lensing of quasars, number counts of elliptical galaxies, and high-redshift Type Ia supernovae, this means that the cosmological constant is probably too small to lengthen the age of the universe significantly. So I am hoping that when the dust finally settles, H0 and t0 will both turn out to be low enough to be consistent with General Relativistic cosmology. But of course the universe is under no obligation to live up to our expectations. ACKNOWLEDGEMENTS. I have benefited from conversations or correspondence with S. Bonometto, S. Borgani, E. Branchini, D. Caldwell, L. Da Costa, M. Davis, A. Dekel, S. Faber, G. Fuller, K. Gorski, K. Griest, J. Holtzman, A. Klypin, C. Kochanek, K. Lanzetta, P. Lilje, R. Mushotzky, R. Nolthenius, J. Ostriker, P.J.E. Peebles, S. Perlmutter, D. Richstone, A. Sandage, R. Schild, J. Silk, L. Storrie-Lombardi, R.B. Tully, E.L. Turner, M. Turner, A. Wolfe, E. Wright, D. York, and D. Zaritsky, and fruitful interactions with UCSC graduate students R. Dav´e, M. Gross, and R. Somerville. This research was supported by NASA, NSF, and UC research grants at UCSC. References [1] This is a considerably revised version of my article in Particle and Nuclear Astro- physics and Cosmology in the Next Millenium, eds. E. Kolb and R. Peccei (Singapore: World Scientific, 1995), pp. 85-98. A popular version is J. Roth and J.R. Primack, Sky and Telescope, 91 (1), 20 (January 1996). [2] W.L. Freedman et al., Nature 371, 757 (1994). 21 [3] G.R. Blumenthal, S.M. Faber, J.R. Primack, and M.J. Rees, Nature 311, 517 (1984); Erratum: 313, 72 (1985). M. Davis, G. Efstathiou, C.S. Frenk, and S.D.M. White, Astrophys. J. 292, 371 (1985). [4] P.J.E. Peebles, Astrophys. J. 284, 439 (1984). [5] L.A. Kofman, N.Y. Gnedin, and N.A. Bahcall, Astrophys. J. 413, 1 (1993); R. Cen, N.Y. Gnedin, and J.P. Ostriker, Astrophys. J. 417, 387 (1993); R. Cen and J.P. Ostriker, Astrophys. J. 429, 4 (1994). [6] S.A. Bonometto and R. Valdarnini, Phys. Lett. 103A, 369 (1984); L.Z. Fang, S.X. Li, S.P. Xiang, Astron. Astrophys. 140, 77 (1984); A. Dekel and S.J. Aarseth, Astroph. J. 283, 1 (1984); Q. Shafi and F.W. Stecker, Phys. Rev. Lett. 53, 1292 (1984). [7] M. Davis, F. Summers, and D. Schlegel, Nature 359, 393 (1992). [8] A. Klypin, J. Holtzman, J.R. Primack, and E. Regos, Astrophys. J. 416, 1 (1993). [9] C.J. Copi, D.N. Schramm, M.S. Turner, Science 267, 192 (1995). Cf. also R.A. Malaney and G.J. Mathews, Phys. Rep. 229, 145 (1993); L.M. Krauss and P.J. Ker- nan, Phys. Lett. B, 347, 347 (1995); N. Hata et al., Phys. Rev. Lett. 75, 3977 (1995); C.J. Copi, D.N. Schramm, and M.S. Turner, Phys. Rev. Lett. 75, 3981 (1995); B.D. Fields and K. Olive, Phys. Lett. B, 368, 103 (1996). [10] A. Dressler et al., Astrophys. J. 313, L37 (1987); D. Lynden-Bell et al., Astrophys. J. 326, 19 (1988). [11] See e.g. S. Olivier, J.R. Primack, G.R. Blumenthal, and A. Dekel, Astroph. J. 408, 17 (1993); A. Klypin and G. Rhee, Astroph. J. 428, 399 (1994); and references therein. [12] S.D.M. White, G. Efstathiou, and C.S. Frenk, Mon. Not. R. Astron. Soc. 262, 1023 (1993). [13] J. Holtzman, Astrophys. J. Supp. 71, 1 (1989). [14] E.L. Wright, Astrophys. J. 396, L13 (1992). [15] J.A. Holtzman and J.R. Primack, Astrophys. J. 405, 428 (1993). These results were presented (in J.R. Primack and J.A. Holtzman, in Gamma Ray -- Neutrino Cosmol- ogy and Planck Scale Physics: Proc. 2nd UCLA Conf., February 1992, D. Cline, ed. (World Scientific, 1993), pp. 28-44) before the anouncement of the COBE discovery. Cf. also T. van Dalen and R.K. Schaefer, Astrophys. J. 398, 33 (1992); R.K. Schaefer and Q. Shafi, Phys. Rev. D 47, 1333 (1993). [16] M. Bolte and C.J. Hogan, Nature, 376, 399 (1995). 22 [17] Note however that while R. Jimenez et al., astro-ph/9602132, Mon. Not. R. Astron. Soc. in press (1996) gives a best estimate for the ages of the oldest GCs of 13.5±2 Gyr, they also give a lower limit of 9.7 Gyr. A new analysis by M. Salaris, S. Degl'Innocenti, A. Weiss, astro-ph/9603092 (1996), also gives 13 Gyr as the best estimate of the ages of the three old GCs that they studied (M15, M68, M92), and says that a lower age is possible. [18] B. Chaboyer, Astrophys. J. 444, L9 (1995) includes a table showing the effects on the GC ages of more than 20 possible changes in input physics and abundances; a Monte Carlo study (B. Chaboyer et al. 1995, astro-ph/9509115) of variations in these parameters gives a 95% CL lower limit on the ages of the oldest GCs of 12 Gyr. [19] X. Shi, Astrophys. J. 446, 637 (1995). [20] R.A. Malaney, G.J. Mathews, and D.S.P. Dearborn, Astrophys. J. 345, 169 (1989); G.J. Mathews and D.N. Schramm, Astrophys. J. 404, 468 (1993). [21] D.E. Winget et al., Astrophys. J. 315, L77 (1987). [22] D.A. VandenBerg et al., Astron. J. 100, 445 (1990); ibid., 102, 1043 (1991). [23] J.W. Yuan, Astron. Astrophys. 224, 108 (1989); 261, 105 (1992). V. Weidemann, Ann. Rev. Astron. Astrophys. 28, 103 (1990). M.A. Wood, Astrophys. J. 386, 539 (1992). M. Hernanz et al., Astrophys. J. 434, 652 (1994). [24] T.D. Oswalt, J.A. Smith, and M.A. Wood, Nature, submitted. [25] A. Sandage, in Practical Cosmology: Inventing the Past, 23rd Sass Fee lectures, ed. B. Binggeli and R. Buser (Berlin: Springer, 1995). A. Sandage and G.A. Tammann, in Advances in astrofundamental physics, Proc. Int. School of Physics "D. Chalonge", eds. N. Sanchez and A. Zichichi (World Scientific, 1995). [26] G.A. Tammann et al., astro-ph/9603076, to be pub. in Science with the HST - II (1996). [27] J.P. Huchra, Science, 256, 321 (1992). S. Van den Bergh, Science 258, 421 (1992). [28] G.H. Jacoby et al., Publ. Astron. Soc. Pac., 104, 599 (1992). R. Ciradullo, G.H. Jacoby, and J.L. Tonry, Astrophys. J. 419, 479 (1993). [29] M. Fukugita, C.J. Hogan, and P.J.E. Peebles, Nature 366, 309 (1993). [30] R.C. Kennicutt, W.L. Freedman, and J.R. Mould, Astron. J. 110, 1476 (1995). [31] M. Rowan-Robinson, The Cosmic Distance Ladder (1985). 23 [32] M.J. Pierce et al., Nature 371, 385 (1994). [33] S. van den Bergh, astro-ph/9509117 (1995); D. Branch et al., astro-ph/9604006 (1996). Cf. B. Schaefer, Astrophys. J. 459, 438 (1996). [34] M.M. Phillips, Astrophys. J. 413, L105 (1993); M. Hamuy et al., Astron. J. 109, 1 (1995). A.G. Riess, W.H. Press, and R.P. Kirshner, Astrophys. J. 438, L17 (1995); they get h = 0.67 ± 0.07. [35] G.A. Tammann and S. Sandage, Astrophys. J. 452, 16 (1995). [36] J. Tonry, Astrophys. J. 373, L1 (1991). [37] W.D. Arnet, D. Branch, and J.C. Wheeler, Nature 314, 337 (1985); D. Branch, Astroph. J. 392, 35 (1992). For more recent results see A. Fisher et al., Astrophys. J. 447, L73; S. van den Bergh, astro-ph/9509117, Astrophys. J. in press. [38] B. Leibundgut and P.A. Pinto, Astroph. J. 401, 49 (1992). But cf. T.E. Vaughan et al., Astrophys. J. 439, 558 (1995). [39] P. Nugent, D. Branch, et al., Phys. Rev. Lett. 75, 394; Erratum 75, 1874 (1995). Reviews: D. Branch and A.M. Khokhlov, Phys. Rep. 256, 53 (1995); D. Branch et al., astro-ph/9601006, to appear in Proc. NATO Advanced Studies Institute on Thermonuclear Supernovae, Aiguablava, Spain, June, 1995, eds. R. Canal, P. Ruiz- Lapuente, and J. Isern (1996). [40] B.P. Schmidt, R.P. Kirschner, and R.G. Eastman, Astrophys. J. 395, 366 (1992). [41] B.P. Schmidt et al., Astrophys. J. 432, 42 (1994). See also Robert Kirschner's Varenna lectures. [42] M. Best and R. Wehrse, Astron. Astroph. 284, 507 (1994). P. Ruiz-Lapuente et al., Astrophys. J. 439, 60 (1995). [43] M. Birkinshaw, J.P. Hughes, and K.A. Arnoud, Astrophys. J. 379, 466 (1991). [44] M. Birkinshaw and J.P. Hughes, Astrophys. J. 420, 33 (1994). [45] K. Yamashita, in New Horizon of X-ray Astronomy -- First Reslts from ASCA, eds. F. Makino and T. Ohashi (Universal Academy Press, Tokyo, 1994), p. 279. [46] T.M. Wilbanks et al., Astrophys. J. 427, L75 (1994), and parallel session talk at Snowmass 94. [47] Y. Rephaeli, Ann. Rev. Astron. Astrophys 33, 541 (1995). [48] J. Pelt et al., Astron. Astroph. 286, 775 (1994). 24 [49] W.H. Press, G.B. Rybicki, and J.N. Hewitt, Astrophys. J. 385, 416 (1992). [50] T. Kundic et al., Astrophys. J. 455, L5 (1995). [51] G.F.R.N. Rhee, Nature 350, 211 (1991). [52] D.H. Roberts et al., Nature 352, 43 (1991). [53] H. Danle, S.J. Maddox, and P.B. Lilje, Astrophys. J. 435, L79 (1994). Also further analyses in prep. [54] N.A. Grogan and R. Narayan, Astrophys. J. in press (1996). [55] J. Mould et al., Astrophys. J. 449, 413 (1995). Cf. also N.R. Tanvir et al., astro- ph/9509160, which uses new HST Cepheids in M96 to calibrate early-type galaxies and deduce a distance to the Coma cluster, leading to H0 = 69 ± 8. [56] E.L. Turner, R. Cen, and J.P. Ostriker, Astron. J. 103, 1427 (1992); X.-P. Wu et al., preprint (1995). [57] Y. Suto, T. Suginohara, and Y. Inagaki, Prog. Theor. Phys. 93, 839 (1995). [58] A. Kim et al., astro-ph/9602123, to appear in Thermonuclear Supernovae (NATO ASI), eds. R. Canal, P. Ruiz-LaPuente, and J. Isern (1996). [59] M. Sasaki, et al. Phys. Lett. B 317, 510 (1993). B. Ratra and P.J.E. Peebles, As- trophys. J. 432, L5 (1994); Phys. Rev. D 52, 1837 (1995). M. Kamionkowski et al., Astrophys. J. 434, L1 (1994). M. Sasaki, T. Tanaka, and K. Yamamoto, Phys. Rev. D 51, 2979 (1995). M. Bucher, A.S. Goldhaber, and N. Turok, Phys. Rev. D 52, 3314 (1995). [60] O. Lahav, P. Lilje, J.R. Primack, and M.J. Rees, Mon. Not. R. Astron. Soc. 251, 128 (1991). [61] S.M. Carroll, W.H. Press, and E.L. Turner, Ann. Rev. Astron. Astrophys, 30, 499 (1992). [62] D. Maoz and H.W. Rix, Astrophys. J. 416, 425 (1993). [63] C. Kochanek, Astrophys. J. 419, 12 (1993). [64] C.S. Kochanek, astro-ph/9510077, Astrophys. J. in press (1996). [65] C. Steidel, M. Dickinson, and S.E. Persson, Astrophys. J. 437, L75 (1994); S. Lilly et al., Astrophys. J. 455, 108 (1995); D. Schade et al., astro-ph/9604032 (1996). 25 [66] E. Bunn and N. Sugiyama, Astrophys. J. 446, 49 (1995). However, this limit may be stronger than is justified by a more complete analysis of the COBE data (E. Wright and K. Gorski, private communications). [67] S. Driver et al., astro-ph/9511141, Astrophys. J. in press (April 1996). [68] A. Klypin, J.R. Primack, and J. Holtzman, astro-ph/9510042, Astrophys. J. in press (July 1996). [69] G. Goldhaber et al., Nucl. Phys. B, S38, 435 (1995). A. Goobar and S. Perlmutter, Astrophys. J. 450, 14 (1995). S. Perlmutter et al., Astrophys. J. 440, L41 (1995). [70] S. Perlmutter, astro-ph/9602122, to appear in Thermonuclear Supernovae (NATO ASI), eds. R. Canal, P. Ruiz-Lapuente, and J. Isern (1996). [71] A. Dekel, Ann. Rev. Astron. Astroph. 32, 371 (1994). M.A. Strauss and J.A. Willick, Phys. Rep. 261, 271 (1995). The latest POTENT results on the power spectrum of mass fluctuations are in S. Zaroubi et al., astro-ph/9603068 (1996) and T. Kolatt and A. Dekel, astro-ph/9512132 (1995). [72] A. Dekel and M.J. Rees, Astrophys. J. 422, L1 (1994). [73] A. Nusser and A. Dekel, Astrophys. J. 405, 437 (1993). Similar constraints have been derived using a perturbative technique by F. Bernardeau, R. Juskiewicz, A. Dekel, and F.R. Bouchet, Mon. Not. R. Astron. Soc. 274, 20 (1995). [74] P.J.E. Peebles, Principles of Physical Cosmology (Princeton Univ. Press, 1992), esp. §20. [75] M. Davis and P.J.E. Peebles, Astrophys. J. 267, 465 (1983). [76] R. Somerville, M. Davis, and J.R. Primack, astro-ph/9604041, Astrophys. J. submit- ted (1996). [77] H.J. Mo, Y.P. Jing, and G. Borner, Mon. Not. R. Astron. Soc. 264, 825 (1993). W. Zurek et al., Astrophys. J. 431, 559 (1994). [78] K.B. Fisher, et al., Mon. Not. R. Astron. Soc. 267, 927 (1994). [79] R.O. Marzke, M.J. Geller, L.N. da Costa, and J.P. Huchra, Astron. J. 110, 477 (1995). [80] R. Somerville, J.R. Primack, and R. Nolthenius, astro-ph/9604051, Astrophys. J. submitted (1996). [81] R. Kirschner, A. Oemler, and P. Schechter, AJ 84, 951 (1979). [82] J.P. Huchra and M.J. Geller, Astrophys. J. 257, 423 (1982). 26 [83] M. Ramella, M.J. Geller, and J.P. Huchra, Astrophys. J. 344, 57 (1989). [84] R. Nolthenius, A. Klypin, and J.R. Primack, astro-ph/9410095, Astrophys. J., in press (1996). [85] R.G. Carlberg and H.M.P. Couchman, Astrophys. J. 340, 47 (1989). [86] R. Nolthenius, A. Klypin, and J.R. Primack, Astrophys. J. 422, L45 (1994). [87] P.J.E. Peebles, Astrophys. J. 344, 53 (1989); 362, 1 (1990); 429, 43 (1994). [88] E. Branchini and R.G. Carlberg, Astrophys. J. 434, 37 (1994). [89] A.M. Dunn and R. Laflamme, Astrophys. J. 443, L1 (1995). [90] E.J. Shaya, P.J.E. Peebles, and R.B. Tully, Astrophys. J. 454, 15 (1995). [91] D. Zaritsky et al., Astrophys. J. 405, 464 (1993); D. Zaritsky and S.D.M. White, Astrophys. J. 435, 599 (1994). [92] D. Zaritsky, private communication (1995). [93] J. Dubinski, J.C. Mihos, and L. Hernquist, astro-ph/9509010, Astrophys. J. in press (1996). [94] J.A. Tyson et al., Astrophys. J. 281, L59 (1984); cf. I. Kovner and M. Milgrom, Astrophys. J. 321, L113 (1987). [95] J. Mould et al., Mon. Not. R. Astron. Soc. 271, 31 (1994). [96] J. Villumsen, astro-ph/9507007 (1995). [97] B. Little and S. Tremaine, Astrophys. J. 320, 493 (1987). [98] C.S. Kochanek, astro-ph/9505068, submitted to Astrophys. J. (1995). [99] A. Songaila et al., Nature 368, 599 (1994); R.F. Carswell et al. Mon. Not. R. Astron. Soc. 268, L1 (1994). [100] D. Tytler and X. Fan, Bull. Am. Astron. Soc. 26, 1424 (1994); D. Tytler, X. Fan, and S. Burles, astro-ph/9603069, submitted to Nature (1996). S. Burles and D. Tytler, astro-ph/9603070, submitted to Science (1996). [101] M. Rugers and C.J. Hogan, Astrophys. J. 459, L1 (1996). cf. C.J. Hogan, astro- ph/9512003 (1995). [102] E.J. Wampler et al., astro-ph/9512084, submitted to Astron. Astroph. (1995). 27 [103] R.F. Carswell et al., Mon. Not. R. Astron. Soc. 278, 506 (1996). [104] S.D.M. White and C.S. Frenk, Astrophys. J. 379, 52 (1991). S.D.M. White et al. Nature 366, 429 (1993). Cf. Lubin et al., Astrophys. J. 460, 10 (1996); A.E. Evrard, C.A. Metzler, and J.F. Navarro, FERMILAB-Pub-95/337-A (1995); S. Schindler, As- tron. Astroph. 305, 756 (1996). Review: G. Steigman and J.E. Felten, Space Science Reviews 74, 245 (1995). [105] D.A. White and A.C. Fabian, Mon. Not. R. Astron. Soc., 273, 72 (1995). Cf. R. Mushotzky et al., in Dark Matter, ed. S.S. Holt and C.L. Bennett, AIP Conf. Proc. 336, 231 (1995); D.A. Buote and C.R. Canizares, Astrophys. J. 457, 565 (1996); S. Bardelli et al., Astron. Astroph. 305, 435 (1996). However, recent ASCA data on several rich clusters imply lower baryon fractions, less than 0.1 for h = 0.5 (R. Mushotzky private communication, March 1996), which may call into question the "fair sample" hypothesis. [106] K.F. Gunn and P.A. Thomas, astro-ph/9510082 (1995) show that the baryon frac- tion could be overestimated by a factor of 2 or more if the cluster gas is a multiphase medium. [107] A. Loeb and S. Mao, Astrophys. J. 435, L109 (1994). P. Biermann (private commu- nication) has suggested that such tangled fields might extend throughout clusters; cf. J.E. Felten, in Clusters, Lensing, and the Future of the Universe, eds. V. Trimble and A. Reisenegger, ASP Conf. Series 88, 271 (1996). [108] P.P. Kronberg, Reports on Progress in Physics 57, 325 (1994). [109] C. Balland & A. Blanchard, astro-ph/9510130 (1995). [110] N.A. Bahcall and L.M. Lubin, Astrophys. J. 426, 513 (1994). M. Bartelmann and R. Narayan, in Dark Matter, College Park, MD, October 1994 (AIP Conference Proceedings 336, 1995) 307. [111] G. Falhman et al. Astrophys. J. 437, 56 (1994), reviewed in N. Kaiser et al., astro- ph/9407004 (1994). There are several similar examples. However, R.G. Carlberg, H.K.C. Yee, and E. Ellingson, astro-ph/9512087 (1995) show that the velocities of galaxies around 16 galaxy clusters are probably not consistent with such large dark matter halos. This puzzling discrepancy needs further study. [112] This is shown visually (with accompanying video; high-resolution frames and low- resolution mpeg excerpts from the video can be viewed or downloaded from our WWW page http: //physics.ucsc.edu/groups/cosmology.html) in my group's CHDM simulations in D. Brodbeck et al., Astrophys. J. in press (1996); cf. also A. Klypin, R. Nolthenius, and J.R. Primack, astro-ph/9502062, Astrophys. J. in press (1996). 28 Observed cluster temperatures and even X-ray luminosities appear to be well fit, according to the first CHDM hydrodynamic cluster simulation, G. Bryan et al., As- trophys. J. 437, L5 (1994). [113] L. Kofman, A. Klypin, D. Pogosyan, and J.P. Henry, astro-ph/9509145 (1995). [114] D. Richstone, A. Loeb, and E.L. Turner, Astrophys. J. 393, 477 (1992). [115] J.O. Burns et al., Astrophys. J. 427, L87 (1994). [116] J.J. Mohr, A.E. Evrard, D.G. Fabricant, and M.J. Geller, Astrophys. J. 447, 8 (1995). [117] Y.P. Jing, H.J. Mo, G. Borner, and L.Z. Fang, in International Workshop on Large Scale Structure in the Universe, Potsdam, 1994 (World Scientific, 1995); Mon. Not. R. Astron. Soc. 276, 417 (1995). [118] E.g., D.A. Buote and J.C. Tsai, Astrophys. J., 458, 27 (1996), and J.C. Tsai and D.A. Buote, astro-ph/9510057, Mon. Not. R. Astron. Soc. in press (1996). [119] J.P. Henry et al., Astrophys. J. 386, 408 (1992). F.J. Castander et al., Nature 377, 39 (1995). H. Ebeling et al., to appear in Roentgenstrahlung from the Universe, Wurzberg, Sept. 1995. [120] C.C. Steidel, M. Dickinson, and S.E. Persson, Astrophys. J. 437, L75 (1994). [121] G. Kauffmann, Mon. Not. R. Astron. Soc. 274, 161 (1995). [122] A. Klypin et al., in prep. (1996). [123] M.G. Haehnelt, Mon. Not. R. Astron. Soc. 265, 727 (1993). [124] N. Katz, T. Quinn, E. Bertschinger, and J.M. Gelb, Mon. Not. R. Astron. Soc. 270, L71. Cf. D.J. Eisenstein and A. Loeb, Astrophys. J. 443, 11 (1995). [125] C.C. Steidel et al., astro-ph/9602024 (1996), and M. Giavalisco, C.C. Steidel, and P.D. Macchetto, astro-ph/9603062 (1996). [126] A.M. Wolfe, in Relativistic Astrophysics and Particle Cosmology, eds. C.W. Ackerlof, M.A. Srednicki (New York, New York Academy of Science, 1993), p. 281. [127] K.M. Lanzetta, Publ/ Astron. Soc. Pac., 105, 1063, (1993); K.M. Lanzetta, A.M. Wolfe and D.A. Turnshek, Astrophys. J. 440, 435 (1995). [128] H.J. Mo and J. Miralda-Escude, Astrophys. J. 430, L25 (1994); G. Kauffmann and S. Charlot, Astrophys. J. 430, L97 (1994); C.-P. Ma and E. Bertschinger, Astrophys. J. 434, L5 (1994). 29 [129] A. Klypin, S. Borgani, J. Holtzman, and J.R. Primack, Astrophys. J. 444, 1 (1995). [130] L.J. Storrie-Lombardie, R.G. McMahon, M.J. Irwin, and C. Hazard, Astrophys. J. 427, L13-16 (1994); -- -, Proc. ESO Workshop on QSO Absorption Lines, ed. G. Meylan (Springer, 1995). Newer data suggest that Ωgas ≈ 3 × 10−3 for z ≈ 2 − 3.5, and either declining or at most remaining constant for higher z ∼ 4 − 4.5 (private communications from Lisa Storrie-Lombardi and Art Wolfe, December 1995). [131] L.M. Lu et al., Astrophys. J. 457, L1 (1996). [132] M. Haehnelt, M. Steinmetz, and M. Rauch, astro-ph/9512118 (1996). [133] J. Quinn et al., Astrophys. J. 456, L83 (1996); M.S. Schubnell et al., astro- ph/9602068, Astrophys. J. in press (1996). [134] D. MacMinn and J.R. Primack, in TeV Gamma Ray Astrophysics, ed. Heinz Volk and F. Aharonian, Space Science Reviews, 75, 413 (1996). Cf. Ground Based Gamma- Ray Astronomy, in Particle and Nuclear Astrophysics and Cosmology in the Next Millenium, eds. E. Kolb and R. Peccei (Singapore: World Scientific, 1995), p. 295. [135] D.O. Caldwell, in Trends in Astroparticle Physics, Stockholm, Sweden 22-25 September 1994, eds. L. Bergstrom, P. Carlson, P.O. Hulth and N. Snellman, Nucl. Phys. B, Proc. Suppl., 43, 126 (1995). C. Athanassopoulos et al., Phys. Rev. Lett. 75, 2650 (1995). In the latest preprint from LSND, available at http://nu1.lampf.lanl.gov/∼lsnd/, the 1995 data strengthens the statistics implying oscillations, but perhaps weakens the case for δm2 = 6 eV2. [136] Y. Fukuda, Phys. Lett. B 335, 237 (1994). [137] J.R. Primack, J. Holtzman, A. Klypin, and D.O. Caldwell, in Trends in Astroparticle Physics, Stockholm, Sweden 22-25 September 1994, eds. L. Bergstrom, P. Carlson, P.O. Hulth and N. Snellman, Nucl. Phys. B, Proc. Suppl., 43, 133 (1995), and Phys. Rev. Lett. 74, 2160 (1995); J.R. Primack, J. Holtzman, and A. Klypin, Moriond 1995, in press. Cf. D.Yu. Pogosyan and A.A. Starobinsky, in International Workshop on Large Scale Structure in the Universe, Potsdam, 1994 (World Scientific, 1995) and Astrophys. J. 447, 465 (1995); K.S. Babu, R.K. Schaefer, and Q. Shafi, Phys. Rev. D53, 606 (1996); A.R. Liddle et al., astro-ph/9511057 (1995). [138] However, an inverted neutrino mass hierarchy with νe the most massive appears to be required if all the current experimental evidence for neutrino mass (solar and atmospheric neutrino deficits, and LSND) are valid, and r-process nucleosynthesis (responsible for production of the heavy elements) takes place in the hot ν bubble a few hundred km above the neutron star in Type II supernovae; see G.M. Fuller, J.R. Primack, and Y. Qian, Phys. Rev. D 52, 1288 (1995). 30 [139] See e.g. S. Dodelson, E. Gates, and A. Stebbins, astro-ph/9509147 (1995); G. Jung- man et al., astro-ph/9512139 (1995); M. Tegmark, astro-ph/9601077, Astrophys. J. Lett. in press (1996). 31
0708.0086
2
0708
2007-08-24T10:59:59
Neutron star cooling after deep crustal heating in the X-ray transient KS 1731-260
[ "astro-ph" ]
We simulate the cooling of the neutron star in the X-ray transient KS 1731-260 after the source returned to quiescence in 2001 from a long (>~ 12.5 yr) outburst state. We show that the cooling can be explained assuming that the crust underwent deep heating during the outburst stage. In our best theoretical scenario the neutron star has no enhanced neutrino emission in the core, and its crust is thin, superfluid, and has the normal thermal conductivity. The thermal afterburst crust-core relaxation in the star may be not over.
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 1 -- 5 (2007) Printed October 26, 2018 (MN LATEX style file v2.2) Neutron star cooling after deep crustal heating in the X-ray transient KS 1731 -- 260 P. S. Shternin1, D. G. Yakovlev1, P. Haensel2, and A. Y. Potekhin1 1 Ioffe Physico-Technical Institute, Politechnicheskaya 26, 194021 Saint-Petersburg, Russia 2 N. Copernicus Astronomical Center, Bartycka 18, PL-00-716 Warsaw, Poland Accepted 2007 August 23. Received 2007 August 22; in original form 2007 August 01 ABSTRACT We simulate the cooling of the neutron star in the X-ray transient KS 1731 -- 260 after the source returned to quiescence in 2001 from a long (& 12.5 yr) outburst state. We show that the cooling can be explained assuming that the crust underwent deep heating during the outburst stage. In our best theoretical scenario the neutron star has no enhanced neutrino emission in the core, and its crust is thin, superfluid, and has the normal thermal conductivity. The thermal afterburst crust-core relaxation in the star may be not over. Key words: X-rays: individual: KS 1731 -- 260 -- stars: neutron. 1 INTRODUCTION KS 1731 -- 260 is a neutron star X-ray transient whose observational history has been described recently by Cackett et al. (2006). The source was discovered in the ac- tive state in August 1989 by the Kvant orbital observatory; subsequent analysis showed that it had also been active in October 1988 (Sunyaev et al. 1990). It remained a bright X-ray source showing type I X-ray bursts for about 12.5 years. It is believed that this activity was powered by accre- tion onto the neutron star (through an accretion disk) from its low-mass companion in a compact binary. Adopting a distance to the source of D = 7 kpc, Cackett et al. (2006) report a characteristic 2 -- 10 keV luminosity of KS 1731 -- 260 in the active state of ∼ 1036 erg s−1. The source remained active till the beginning of 2001 and then returned to qui- escence. The last detection in the active state was made on January 21, 2001 with RXTE, but on February 7, 2001, RXTE already failed to detect KS 1731 -- 260 in the active state (Wijnands et al. 2001). The first detection of KS 1731 -- 260 in quiescence was made by Wijnands et al. (2001) with Chandra on March 27, 2001. For D =7 kpc, the 0.5 -- 10 keV luminosity was ∼ 1033 erg s−1, three orders of magnitude lower than in the active state. The radiation spectrum contains a component that can be interpreted as the thermal emission from the neu- tron star surface. Since then the source has been observed several times with Chandra and XMM-Newton as summa- rized by Cackett et al. (2006). Its X-ray light curve faded over time scale ∼ 2 years showing a trend to flattening (with the residual luminosity of ∼ 2 × 1032 erg s−1). According to observations, the accretion in quiescent states of X-ray transients is stopped or strongly suppressed. The nature of quiescent X-ray emission is a subject of de- bates (see, Cackett et al. 2006 and references therein, for a list of possible hypotheses). Here, we focus on the hypoth- esis of deep crustal heating of neutron stars proposed by Brown, Bildsten & Rutledge (1998). It states that, when a neutron star accretes, its crust is heated by nuclear trans- formations (mainly by beta captures and pycnonuclear re- actions) in the accreted matter sinking within the crust un- der the weight of newly accreted material. The star remains sufficiently warm after an accretion episode, producing qui- escent surface emission. The sequence of nuclear transfor- mations and associated energy generation rates were cal- culated by Haensel & Zdunik (1990) assuming that the ac- creted matter burns to 56Fe in the neutron star surface layers so that, initially, before sinking within the deep crust, the matter is composed of 56Fe. Later Schatz et al. (2001) calcu- lated explosive nucleosynthesis in the neutron star surface layers and showed that the explosive burning can proceed to much heavier elements. Accordingly, Haensel & Zdunik (2003) proposed new deep crustal heating scenarios (start- ing with heavier elements, particularly, with 106Pd). In all the cases Haensel & Zdunik (1990, 2003) obtained similar deep crustal energy release, ∼ 1 − 1.5 MeV per one accreted nucleon, sufficient to power quiescent thermal emission in X-ray transients. Recently Gupta et al. (2007) have recon- sidered the heating starting with multicomponent matter (ashes of explosive burning in the surface layers). They have shown that the heating of the deep outer crust can be higher because beta captures can produce daughter nuclei in ex- cited states; their deexcitation can generate extra heat. The onset of the quiescent state of KS 1731 -- 260 was recognized as an outstanding phenomenon from the very beginning. The majority of other X-ray transients undergo 2 P. S. Shternin et al. short accretion episodes (days to months) in which the deep crustal heating cannot break the crust-core thermal coupling and make the crust much hotter than the stellar core. How- ever, it is possible in KS 1731 -- 260 because of the long accre- tion stage (Rutledge et al. 2002). Therefore, observations of its quiescent thermal emission can help to understand how the crustal heat spreads over the entire star, that is useful for exploring the neutron star structure. The first modelling of the KS 1731 -- 260 cooling was made by Rutledge et al. (2002) soon after the quiescence onset. The authors based on previous simulations by Ushomirsky & Rutledge (2001) of the crust-core relaxation in a neutron star with a heated crust. Rutledge et al. (2002) proposed several cooling scenarios based on the deep crustal heating model of Haensel & Zdunik (1990) and different crust and core microphysics. They predicted that the neu- tron star can reach the crust-core relaxation and associated flattening of the quiescent soft X-ray light curve in 1 -- 30 years. Cackett et al. (2006) have compared the new observa- tions of KS 1731 -- 260 with the predictions of Rutledge et al. (2002) and conclude that the star should have high thermal conductivity in the crust and enhanced neutrino emission in the core. Here we present new cooling calculations and discuss their consistency with the observations of KS 1731 -- 260. 2 COOLING MODEL AND PHYSICS INPUT are similar to those cooling simulations of Our Ushomirsky & Rutledge (2001) and Rutledge et al. (2002). We assume that the neutron star crust underwent deep crustal heating during the long accretion stage. We employ the model of deep crustal heating of Haensel & Zdunik (1990), but modify the energy release due to sequences of pairs of beta-captures in the crust. Specifically, we assume that daughter nuclei after a primary beta capture are produced in excited states and deexcite before a secondary beta capture, heating thus the matter (instead of wasting extra energy into neutrino emission). In this way the distribution of heating sources remains the same as in Haensel & Zdunik (1990) but the sources in the outer crust become stronger, resembling those obtained by Gupta et al. (2007). The source positions and strengths, calculated by Haensel & Zdunik (2007), are shown in Fig. 1. The overall energy release is 1.9 MeV per accreted nucleon. To simulate the neutron star cooling we use our gen- eral relativistic cooling code (Gnedin, Yakovlev & Potekhin 2001). It solves the thermal diffusion problem within the star (at densities ρ > ρb) and uses a predetermined quasi- stationary relation Ts − Tb (Potekhin, Chabrier & Yakovlev 1997) between the effective surface temperature Ts and the temperature Tb at the base (ρ = ρb) of a thin heat- blanketing envelope (ρ 6 ρb). Now we shift ρb from previ- ously used values ∼ 1010 − 1011 g cm−3 to ρb = 108 g cm−3. This allows us to put all heat sources into the region of ρ > ρb and to reduce the time of heat propagation through the blanketing layer from ∼ 1 yr to ∼ 1 d (enabling the code to trace short-term -- 1 day -- surface temperature vari- ations). To explore the sensitivity of calculations to the crust physics, we employ two models of the neutron star crust, composed of ground-state (GS) or accreted (A) matter. The ground-state crust (e.g., Haensel, Potekhin & Yakovlev 2007) has been used in our previous simulations. The model of accreted crust (Haensel & Zdunik 1990) is consistent with the adopted model of deep crustal heating. The accreted crust is composed of lighter nuclei with lower atomic num- bers. Deep in the inner crust, at ρ & 1013 g cm−3, composi- tion is similar to the ground-state one, with &80% of nucle- ons constituting a neutron gas (Haensel & Zdunik 1990). We employ the electron thermal conductivity in the crust, κ, limited by electron-ion (Gnedin et al. 2001) and electron-electron (Shternin & Yakovlev 2006) scattering. It will be called normal. We will also use the model electron thermal conductivity proposed by Brown (2000). It corre- sponds to an amorphous crust (e.g., Jones 2004) and will be called low. Actually, it gives the lowest limit on κ in the crust. Several model thermal conductivities as functions of density in the crust for two values of temperature (T = 108 and 107 K) are plotted in Fig. 1. In the inner crust, we take into account the effects of neutron superfluidity on the heat capacity of free neutrons (e.g., Yakovlev, Levenfish & Shibanov 1999). A representa- tive set of models for superfluid neutron gaps in the in- ner crust, which determine superfluid critical temperature profiles Tc(ρ), is collected by Lombardo & Schulze (2001). The collection includes a well defined gap provided by the pure BCS theory of singlet-state neutron pairing (with a maximum of Tc ∼ 2 × 1010 K within the crust) and a number of gaps calculated using various neutron polariza- tion models (with the maxima of Tc approximately three times lower). BCS superfluidity very strongly suppresses the neutron heat capacity in the inner crust; this super- fluidity will be called strong. The effects of other superfluid models are weaker and more or less similar. For illustra- tion of the latter effects, we will use the model proposed by Wambach, Ainsworth & Pines (1993); such superfluidity will be called moderate. We calculate the neutrino emission in the crust and in the core according to Yakovlev et al. (2001). In our cooling models the neutron star stays not too hot, so that crustal neutrino emission (including that due to Cooper pairing of neutrons) is insignificant. In the neutron star core, we use an equation of state of dense matter (containing nucleons, electrons, and muons) constructed by Akmal, Pandharipande & Ravenhall (their model Argonne V18+δv+UIX∗). Specifi- (1998) cally, we adopt its convenient parametrization proposed by Heiselberg & Hjorth-Jensen (1999) and described as APR I by Gusakov et al. (2005). In this case, the maximum grav- itational mass of stable neutron stars is Mmax = 1.923M⊙ and the direct Urca process of powerful neutrino emission opens at M > 1.828M⊙. We will mainly use two neutron star models, with masses M = 1.6 and 1.4 M⊙, where di- rect Urca process is forbidden; both stars demonstrate slow neutrino cooling via the modified Urca process. The 1.4 M⊙ star has the central density ρc = 9.4 × 1014 g cm−3, the circumferential radius R = 12.14 km, and the crust thick- ness ∆R = R − Rcore = 1.16 km (where Rcore is the core radius corresponding to ρ = 1.5 × 1014 g cm−3). The 1.6 M⊙ star is more compact, with thinner crust, and has ρc = 1.16 × 1015 g cm−3, R = 11.88 km, and ∆R = 890 m. The thermal conductivity of the neutron star core is described following Baiko, Haensel & Yakovlev (2001) and Neutron star cooling in KS 1731 -- 260 3 cooling time scales 1 -- 10 kyr until the next accretion episode. The extra heat deposited to the core is mainly emitted over those long cooling time scales via core neutrino emission. Our cooling curves in Fig. 2 are calculated for differ- ent neutron star masses, microphysics in the crust, mass accretion rates (and Etot), and T ∞ s0 (as shown in the fig- ure and Table 1). Our aim is to explain the observed tem- poral evolution T ∞ s (t) of KS 1731 -- 260 in the quiescent state. A successful explanation should also be consistent with the observational constraint on the mass accretion rate, M . 5 × 10−9M⊙ yr−1 (for D = 8 kpc, see Table 3 in Yakovlev, Levenfish & Haensel 2003), which translates into Etot . 2.4 × 1044 erg for the adopted deep heating model. Table 1 shows that all presented cooling models roughly sat- isfy this requirement. Fig. 2a refers to the 1.6 M⊙ neutron star model, while Fig. 2b is for the 1.4 M⊙ star. All curves in Figs. 2a and b are calculated assuming the initial surface temperature to be T ∞ s0 = 0.8 MK (so that the internal temperature is ∼ 8 × 107 K, as in a cooling isolated neutron star which is ∼ 105 years old). This is a typical surface temperature of the neutron star provided by the last three observational points. Thus, in Figs. 2a and b we (following Cackett et al. 2006) tacitly assume that the crust-core equilibrium is re- established in two years after the quiescence onset. In all curves in Figs. 2a and b, but in curve 6, Etot has been cho- sen in such a way for the surface temperature at the first quiescent observation to be consistent with data. Curve 1 in Fig. 2a seems to be the best. It corresponds to the accreted crust with the normal thermal conductivity and moderate neutron superfluidity. It naturally explains the thermal relaxation of KS 1731 -- 260 with the standard physics input. The maximum internal temperature raise to T ∼ 4 × 108 K takes place at t = 0 near the bound- ary between the outer and the inner crust. The core-crust relaxation takes ∼ 2 years. The star would need ∼ 103 years to emit all the heat pumped into the core during the outburst and reach the same thermal state as before the outburst. This is in good agreement with the estimate of Rutledge et al. (2002) for a similar cooling model. Using the same physics as for curve 1 but the ground- state crust (with lower conductivity) we obtain slower re- laxation (curve 3). It is acceptable but less consistent with the observations; it requires lower Etot because it is easier to heat the crust with smaller thermal conductivity. Taking the latter cooling model 3 and neglecting superfluidity in the inner crust we obtain curve 2. A non-superfluid crust has larger (neutron) heat capacity which noticeably delays the thermal relaxation making it much less consistent with the data. Returning to our best model 1 but assuming strong superfluidity in the crust, we stronger suppress the heat ca- pacity of neutrons and obtain curve 4; it shows faster and quite acceptable relaxation. The effects of strong and moder- ate superfluidity are actually very close, although the pres- ence of superfluidity greatly improves the agreement with the data. Now if we return to model 1 but assume low ther- mal conductivity, we come to much longer crust-core relax- ation (over several hundred years, curve 5). It is inconsistent with the observations, in agreement with the conclusion of Cackett et al. (2006). Finally, if we take the best model 1 but assume the same (lower) mass accretion rate as in model 2, we get curve 6. Therefore, the latter mass accretion rate, Figure 1. Density dependence of the electron thermal conductiv- ity κ (left vertical scale) in the neutron star crust with accreted (A) or ground-state (GS) matter at two temperatures (log T [K]=7 and 8, numbers next to curves). The thin lower curve is for the model of low κ while other curves are for normal κ. Vertical bars show positions and power (right vertical scale) of the heat sources. Initial layer is assumed to consist of 56Fe, as in Haensel & Zdunik (1990), but neutrino losses in electron captures are suppressed, following Gupta et al. (2007). Shternin & Yakovlev (2007). For simplicity, the effects of nu- cleon superfluidity in the core are neglected. 3 RESULTS AND DISCUSSION s We have calculated (Fig. 2) a number of cooling curves which give the effective surface temperatures T ∞ , as detected by a distant observer, versus time t; t = 0 refers to February 1, 2001, the date near which KS 1731 -- 260 turned in qui- escence. We compare the curves with seven observational points presented by Cackett et al. (2006); the values of T ∞ were inferred from the observed X-ray spectra (employ- ing non-magnetic neutron star hydrogen atmosphere models from the Xspec database and assuming D = 7 kpc, R = 10 km, and M = 1.4 M⊙). We doubled the reported 1σ obser- vational error bars to enlarge statistical significance, that would make our analysis more realistic. s To start any cooling calculation, we have taken a neu- tron star model with the thermally relaxed interior and some initial surface temperature T ∞ s0 . Then we switch on deep crustal heating produced by a constant mass accretion rate M over 12.5 years. In that period a certain amount of heat Etot is deposited into the crust. The crust is heated and its thermal balance with the thermally inertial core is violated. Then we switch off accretion (deep crustal heat- ing) and the crust cools down regaining thermal equilibrium with the core. Some (typically small) part of Etot diffuses to the surface and radiates away via thermal surface emis- sion. The rest is carried by thermal conduction to the core. The core temperature stays almost unchanged because of the high core thermal conductivity and heat capacity. The crust-core thermal relaxation takes 1 -- 100 years, depending on the neutron star model. After this relaxation is over, the surface temperature nearly reaches its initial value T ∞ s0 . The star cools down further with isothermal interior over typical 4 P. S. Shternin et al. Table 1. Cooling curves in Fig. 2 Curve T ∞ Crust s0 MK model Conduction Superfluid in crust in crust Etot 1044 erg 1a 2a 3a 4a 5a 6a 1b 2b 3b 1c 2c 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.67 0.63 A GS GS A A A A GS GS GS GS normal normal normal normal low normal normal normal normal normal normal moderate none moderate strong moderate moderate moderate none moderate none none 2.6 1.9 1.8 2.6 0.6 1.9 2.3 1.7 1.5 2.4 2.4 being used for the microphysics of model 1, is insufficient to explain high values of T ∞ in the beginning of the quiescent state. s Curves 1 -- 3 in Fig. 2b are analogous to curves 1 -- 3 in Fig. 2a, but are calculated for a less massive star, with thicker crust. The thicker crust produces longer thermal relaxation, less consistent with the observations. We have also performed many other cooling calcula- tions varying physics input. In particular, we have varied the distribution of heat sources within the crust. We have obtained that it is much easier to explain the observations by placing the sources into the outer crust. These models naturally give short thermal relaxation and efficient heat- ing of the surface. In contrast, were all sources located in the deep inner crust, the star would show longer thermal relaxation and one would need too much energy to heat the surface because the heat would be pumped into the core. In connection to this, the improved model of deep crustal heating used here, where the heat release in the crust is en- hanced compared to the original model of Haensel & Zdunik (1990) (due to switching-off neutrino losses associated with electron captures, Gupta et al. 2007), is more favorable for explaining the observations. In addition, we have artificially varied the thermal con- ductivity in different places of the crust and found high sen- sitivity of the cooling curves to these variations. The conduc- tivity strongly affects both, the thermal relaxation time and the efficiency of surface heating. Taking the conductivity a few times lower than the normal conductivity of accreted or ground-state crust produces too long crust-core relaxation which disagrees with the data. Furthermore, we have taken different neutron star mod- els (different equations of state in the core, and different masses). In particular, we have used the models of massive neutron stars whose core neutrino emission is strongly en- hanced by the nucleonic direct Urca process (e.g., 1.9 M⊙ model for the equation of state employed in Fig. 2). We have found that we need unrealistically intense crustal heating (too high Etot) to explain the high observed values of T ∞ s (t) in the beginning of quiescence. Moreover, such a star has too short global cooling time scale (years to decades), compara- ble to the crust-core relaxation time. The crust-core relax- ation becomes coupled to the global thermal relaxation; the cooling curves do not show the observed flattening at t & 2 years. Hence, we cannot reconcile theory with observations if the neutrino emission of the star is enhanced by the direct Urca process. Nevertheless, we think that it may be pos- sible to explain the observations if the neutrino emission is enhanced by a less efficient mechanism (e.g., by pion or kaon condensation in the stellar core) or if the direct Urca process operates but is strongly suppressed by nucleon superfluid- ity (e.g., Yakovlev & Pethick 2004, Page, Geppert & Weber 2006). Finally, we remark that the thermal crust-core relax- ation in KS 1731 -- 260 may be still not over. This is illustrated in Fig. 2c, where we present two new cooling curves for our 1.6 M⊙ and 1.4 M⊙ neutron star models. They are calcu- lated without imposing the constraint that T ∞ s0 = 0.8 MK. We have intentionally taken the physics input (ground-state, non-superfluid crust with normal conductivity) which gives too long thermal relaxation to explain the data for the sce- narios in Figs. 2a and b. Now we take lower T ∞ s0 and reach consistency with the current observations (and get a rather low crustal heat release Etot). We see that the crust-core re- laxation in KS 1731 -- 260 can really last longer than 2 years, and this possibility widens the class of cooling models con- sistent with the data. It will hopefully be checked in future observations of KS 1731 -- 260. 4 CONCLUSIONS We have simulated the cooling of the neutron star in the quiescent state of KS 1731 -- 260 employing the model of deep crustal heating of the star in the outburst state. We have used the model of deep crustal heating (Haensel & Zdunik 1990) updated by switching-off neutrino losses in the crust (Gupta et al. 2007). Our main conclusions are: (i) One can explain current observations of KS 1731 -- 260 using a model of deep crustal heating and a standard micro- physics of the neutron star. (ii) If the crust-core thermal relaxation in the neutron star is reached in ∼ 2 years, the most successful cooling model implies the model of accreted crust with normal ther- mal conductivity and neutron superfluidity; the neutron star should be sufficiently massive (to have a thinner crust), but the neutrino emission in its core cannot be too high (e.g., it can be provided by the modified Urca process). All these factors shorten the crust-core thermal relaxation. (iii) The model of low thermal conductivity (amorphous crust) gives too long crust-core relaxation, inconsistent with the data. (iv) The enhanced neutrino cooling via the direct Urca process in the neutron star core gives too fast cooling of the entire star and requires too intense crustal heating, incon- sistent with the data. (v) The crust-core thermal relaxation can be not reached yet. If so, the data can be explained by a wider class of neutron star models. We stress that the thermal crust-core relaxation of the neutron star in KS 1731 -- 260 is much more sensitive to the physics of the crust than the core. We employed the mod- els of non-superfluid core just for simplicity. Core superflu- Neutron star cooling in KS 1731 -- 260 5 Figure 2. Theoretical cooling curves for (a) M = 1.6 M⊙ and (b) 1.4 M⊙ neutron stars, and (c) for stars with both M compared with observations. The curves are explained in Table 1 and in text. idity can change the core heat capacity and neutrino lu- minosity, but the principal conclusions will be the same. Our calculations are not entirely self-consistent. For in- stance, the surface temperature was inferred from observa- tions (Cackett et al. 2006), assuming neutron star masses and radii different from those used in our cooling models. This inconsistency cannot affect our main conclusions, but it would be desirable to infer T ∞ for our neutron star mod- els. The thermal relaxation in the quiescent state has been observed also (Cackett et al. 2006) for another neutron star X-ray transient, MXB 1659 -- 29. We hope to analyse these data in the next publication. s 5 ACKNOWLEDGMENTS We are grateful to A. I. Chugunov, O. Y. Gnedin, K. P. Levenfish and Yu. A. Shibanov for useful discussions, and to the referee, Ulrich Geppert, for valuable remarks. The work was partly supported by the Russian Foundation for Basic Research (grants 05-02-16245, 05-02-22003), by the Federal Agency for Science and Innovations (grant NSh 9879.2006.2), and by the Polish Ministry of Science and Higher Education (grant N20300632/0450). One of the au- thors (P.S.) acknowledges support of the Dynasty Founda- tion and perfect conditions of the Nicolaus Copernicus As- tronomical Center in Warsaw, where this work was partly performed. P.H. and A.P. thank the Institute for Nuclear Theory at the University of Washington for hospitality and the U.S. Department of Energy for partial support during the completion of this work. References Akmal A., Pandharipande V. R., Ravenhall D. G.,1998, Phys. Rev. C, 58, 1804 Baiko D. A., Haensel P., Yakovlev D. G., 2001, A&A, 374, 151 Brown E. F., 2000, ApJ, 531, 988 Brown E. F., Bildsten L., Rutledge R. E., 1998, ApJ, 504, L95 Cackett E. M., Wijnands R., Linares M., Miller J. M., Homan J., Lewin W. H. G., 2006, MNRAS, 372, 479 Gnedin O. Y., Yakovlev D. G., Potekhin A. Y., 2001, MN- RAS, 324, 725 Gupta S., Brown E. F., Schatz H., Moller P., Kratz K.-L., 2007, ApJ, 662, 1188 Gusakov M. E., Kaminker A. D., Yakovlev D. G., Gnedin O. Y., 2005, MNRAS, 363, 555 Haensel P., Zdunik J. L., 1990, A&A, 227, 431 Haensel P., Zdunik J. L., 2003, A&A, 404, L33 Haensel P., Zdunik J. L., 2007, A&A, in preparation Haensel P., Potekhin A. Y., Yakovlev D. G., 2007, Neutron Stars. 1. Equation of State and Structure, Springer, New- York Heiselberg H., Hjorth-Jensen M., 1999, ApJ, 525, L45 Jones P. B., 2004, MNRAS, 351, 956 Lombardo U., Schulze H.-J., 2001, Lecture Notes Phys, 578, 30 Page D., Geppert U., Weber F., 2006, Nucl. Phys. A, 777, 497 Potekhin A. Y., Chabrier G., Yakovlev D. G., 1997, A&A, 323, 415 Rutledge R. E., Bildsten L., Brown E. F., Pavlov G. G., Zavlin V. E., Ushomirsky G., 2002, ApJ, 580, 413 Schatz H. et al., 2001, Phys. Rev. Lett., 86, 3471 Shternin P. S., Yakovlev D. G., 2006, Phys. Rev. D, 74, 043004. Shternin P. S., Yakovlev D. G., 2007, Phys. Rev. D, 75, 103004 Sunyaev R. A. et al., 1990, SvA Lett., 16, 59 Ushomirsky G., Rutledge R. E., 2001, MNRAS, 325, 1157 Wambach J., Ainsworth T. L., Pines D., 1993, Nucl. Phys. A, 555, 128 Wijnands R., Miller J. M., Markwardt C., Lewin W.H.G., van der Klis M., 2001, ApJ, 560, L159 Yakovlev D. G., Pethick C. J., 2004, ARA&A, 42, 169 Yakovlev D. G., Levenfish K. P., Haensel P., 2003, A&A, 407, 265 Yakovlev D. G., Levenfish K. P., Shibanov Yu. A., 1999, Physics -- Uspekhi, 42, 737 Yakovlev D. G., Kaminker A. D., Haensel P., Gnedin O. Y., 2001, Physics Reports, 354, 1
0802.2604
1
0802
2008-02-19T20:21:21
Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies
[ "astro-ph" ]
Tori of Active Galactic Nuclei are made up of a mixture of hot and cold gas, as well as dust. In order to protect the dust grains from destruction by the hot gas as well as by the energetic radiation of the accretion disk, the dust is often assumed to be distributed in clouds. In our new 3D model of AGN dust tori, the torus is modelled as a wedge-shaped disk in which dusty clouds are randomly distributed, by taking the dust density distribution of the corresponding continuous model into account. We especially concentrate on the differences between clumpy and continuous models in terms of the temperature distributions, the surface brightness distributions and interferometric visibilities, as well as spectral energy distributions. To this end, we employ radiative transfer calculations with the help of the 3D Monte Carlo code MC3D. In a second step, interferometric visibilities are calculated from the simulated surface brightness distributions, which can be directly compared to observations with the MIDI instrument. The radial temperature distributions of clumpy models possess significantly enhanced scatter compared to the continuous cases. Even at large distances, clouds can be heated directly by the central accretion disk. The existence of the silicate 10 micron-feature in absorption or in emission depends sensitively on the distribution, the size and optical depth of clouds in the innermost part of the torus, due to shadowing effects of clouds there. This explains failure and success of previous modelling efforts of clumpy tori. After adapting the parameters of our clumpy standard model to the circumstances of the Seyfert 2 Circinus galaxy, it can qualitatively explain recent mid-infrared interferometric observations performed with MIDI, as well as high resolution spectral data.
astro-ph
astro-ph
Astronomy&Astrophysicsmanuscript no. schartmannclumpy November 1, 2018 c(cid:13) ESO 2018 8 0 0 2 b e F 9 1 ] h p - o r t s a [ 1 v 4 0 6 2 . 2 0 8 0 : v i X r a Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies M. Schartmann1 ,⋆, K. Meisenheimer1, M. Camenzind2, S. Wolf1, K. R. W. Tristram1, and Th. Henning1 1 Max-Planck-Institut fur Astronomie (MPIA), Konigstuhl 17, D-69117 Heidelberg, Germany 2 ZAH, Landessternwarte Heidelberg, Konigstuhl 12, D-69117 Heidelberg, Germany Received / Accepted ABSTRACT Context. Tori of Active Galactic Nuclei (AGN) are made up of a mixture of hot and cold gas, as well as dust. In order to protect the dust grains from destruction by the surrounding hot gas as well as by the energetic (UV/optical) radiation from the accretion disk, the dust is often assumed to be distributed in clouds. Aims. A new three-dimensional model of AGN dust tori is extensively investigated. The torus is modelled as a wedge-shaped disk within which dusty clouds are randomly distributed throughout the volume, by taking the dust density distribution of the corresponding continuous model into account. We especially concentrate on the differences between clumpy and continuous models in terms of the temperature distributions, the surface brightness distributions and interferometric visibilities, as well as spectral energy distributions. Methods. Radiative transfer calculations with the help of the three-dimensional Monte Carlo radiative transfer code MC3D are used in order to simulate spectral energy distributions as well as surface brightness distributions at various wavelengths. In a second step, interferometric visibilities for various inclination as well as position angles and baselines are calculated, which can be used to directly compare our models to interferometric observations with the MIDI instrument. Results. We find that the radial temperature distributions of clumpy models possess significantly enhanced scatter compared to the continuous cases. Even at large distances, clouds can be heated directly by the central accretion disk. The existence of the silicate 10 µm-feature in absorption or in emission depends sensitively on the distribution, the size and optical depth of clouds in the innermost part of the dust distribution. With this explanation, failure and success of previous modelling efforts of clumpy tori can be understood. The main reason for this outcome are shadowing effects of clouds within the central region. We underline this result with the help of several parameter variations. After adapting the parameters of our clumpy standard model to the circumstances of the Seyfert 2 Circinus galaxy, it can qualitatively explain recent mid-infrared interferometric observations performed with MIDI, as well as high resolution spectral data. Key words. Galaxies: active - Galaxies: nuclei - Galaxies: Seyfert - Radiative transfer - ISM: dust, extinction - Galaxies: individual: Circinus 1. Introduction and motivation According to today's knowledge, Active Galactic Nuclei (AGN) are powered by accretion onto a supermassive black hole (106 − 1010 M⊙, e.g. Shankar et al. 2004) residing in their centres. Thereby, gravitational energy is converted into heat by viscous processes within the surrounding accretion disk, which extends from the marginally stable orbit up to several thousands of Schwarzschild radii. The emitted UV/optical light illuminates the attached, toroidally shaped dust reservoir. The concept of this obscuring torus was introduced in order to unify mainly two classes of observed spectral energy distributions (SEDs): one shows a peak in the UV-region with overlayed broad and narrow optical emission lines, the other class shows only narrow optical emission lines. This can be interpreted as an inclination angle dependence. For viewing angles within the dust-free cone of the torus (type 1 sources), direct signatures of the accretion disk (a peak in the UV-range) and the region close to the centre within the funnel of the torus show up. This is where gas moves fast and, therefore, produces broad emission lines (the region is hence called the Broad Line Region (BLR) of the nucleus). For edge- on lines of sight (type 2 sources), the direct view onto the centre Send offprint requests to: M. Schartmann ⋆ e-mail: [email protected] is blocked and optical emission lines can only be detected from gas beyond the torus funnel. Being further away from the centre, it moves slower and hence produces narrow emission lines only. This is the so-called Unified Scheme of Active Galactic Nuclei (Antonucci 1993; Urry & Padovani 1995). First evidence for this scenario came from spectropolarimetric observations of type 2 sources (Miller & Antonucci 1983), clearly displaying type 1 signatures in the polarised light, which is scattered by electrons and tenuous dust within the funnel above the torus. The opening angle of the torus can be estimated with the help of statistics of the different types of Seyfert galaxies. Maiolino & Rieke (1995) find a ratio between Sy 2 to Sy 1 galaxies of 4:1 in their sam- ple, which results in an opening angle of the light cones of 74◦, in concordance with many observations of ionisation cones of individual galaxies. Direct support for the idea of geometrically thick tori comes from recent interferometric observations in the mid-infrared (e.g. Jaffe et al. 2004; Tristram et al. 2007). These tori are made up of at least three components: (i) hot ionised gas, (ii) warm molecular gas and (iii) dust. Krolik & Begelman (1988) proposed that the dusty part has to be organ- ised in a clumpy structure in order to prevent the grains from being destroyed by the hot surrounding gas (with temperatures of the order of 106 K) in which the clouds are supposed to be embedded. Another hint for the clumpy nature of the obscuring 2 M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies material -- in this case mainly for the distribution of neutral gas -- comes from X-ray measurements of the absorbing column den- sity. Risaliti et al. (2002) claim that the observed variability of these measurements on timescales from months to several years can be explained by a clumpy structure of the torus. Combining X-ray absorbing column densities with spectral information fur- ther strengthens the claim for a clumpy distribution of the dust (Shi et al. 2006). Earlier work on torus simulations concentrated mostly on smooth dust distributions (e.g. Pier & Krolik 1992a; Granato & Danese 1994; van Bemmel & Dullemond 2003; Schartmann et al. 2005). This was mainly caused by the lack of appropriate (3D) radiative transfer codes and computational power. Nevertheless, such models are good approximations for the case that the clumps that build up the torus are small compared to the total torus size, as is also shown in a parameter study described in Sect. 4.2. These continuous models are able to describe the gross observable features of these objects (see e. g. Schartmann et al. 2005). However, problems arose from too strong emission features of silicate dust compared to the observations, when looking directly onto the inner rims of the model structures (face-on views). They had never been observed before that time, although almost all models showed them for the face-on view. Therefore, much theoretical effort was undertaken in order to find models showing no silicate feature at all in the face-on case, while retaining the silicate absorption feature in the edge-on case. Manske et al. (1998) for example succeeded in avoiding silicate emission features with a flared dust disk of high optical depth in combination with an anisotropic radiation characteristic of the central illuminating source. A very promising idea was to solve the problem naturally by splitting the dust distribution into single clouds. This was first attempted by Nenkova et al. (2002). A one-dimensional code for the simulation of radiative transfer through single clumps was used and, in a second step, the torus and its emitted SED was assembled by adding many clouds of different aspect angles with the help of a statistical method. With this approach, they could show that a clumpy dust distribution of this kind can significantly smear out the prominent silicate emission feature of the SEDs of type 1 objects at 10 µm for a large range of parameter values. No more fine-tuning was needed, as in the previously proposed solutions with the help of special continuous models. Subsequently, real two-dimensional radiative transfer calculations were undertaken by Dullemond & van Bemmel (2005). Clouds were modelled as concentric rings. A direct comparison between these kinds of clumpy models and the corresponding continuous models did not show evidence for a systematic suppression of the silicate feature in emission in the clumpy models. Meanwhile, silicate features in emission were found with the help of the Infrared Spectrograph (IRS) onboard the Spitzer space telescope (e. g. Siebenmorgen et al. 2005; Hao et al. 2005; Sturm et al. 2005; Weedman et al. 2005). For these kinds of studies, Spitzer is superior to other available facilities, due to its high sensitivity and the coverage of a wavelength range includ- ing both silicate features (at 9.7 µm and 18.5 µm) and the sur- rounding continuum emission. Silicate emission features were found in different levels of AGN activity, ranging from very lu- minous quasars down to weak LINERS. These findings are in good agreement with a geometrical unification by an optically thick dusty torus, as silicate emission features can be produced even in the simplest models. But one has to be cautious, as due to the large beam of the Spitzer space telescope and the low tem- peratures measured, it is unclear whether these silicate features result from dust emission in the innermost parts of the torus or from optically thin regions surrounding them. Very detailed simulations of clumpy tori were undertaken re- cently by Honig et al. (2006). They apply a similar method as Nenkova et al. (2002), but use a 2D radiative transfer code for the simulation of SEDs of individual spherical clumps at var- ious positions in the torus and with various illumination pat- terns: directly illuminated and/or illuminated by reemitted light of surrounding clouds. In a second step, these clouds are dis- tributed according to physical models by Vollmer et al. (2004) and Beckert & Duschl (2004). A comparison of the resulting SEDs and images with spectroscopic and interferometric obser- vations shows good agreement. This model is characterised by a large number of small clouds with a very large optical depth, es- pecially close to the centre. We compare our models with these models in Sect. 5.2. Despite the detection of geometrically thick dust tori in nearby Seyfert galaxies (e.g. Jaffe et al. 2004; Tristram et al. 2007), many questions remain: How are these tori formed? How are they stabilised against gravity? Do steady torus solutions ex- ist? Several attempts to answer these questions have been made. For example Krolik & Begelman (1988) and Beckert & Duschl (2004) support the scale-height of their tori with the help of discrete clumps, moving at supersonic velocities, maintained by mainly elastic collisions with the help of strong magnetic fields. Other groups replace the torus by a magnetically-driven wind solution (Konigl & Kartje 1994). The most recent sug- gestion comes from Krolik (2007), building up on an idea of Pier & Krolik (1992b), where the scale-height of tori can be maintained with the help of infrared radiation pressure, as shown with an idealised analytical model. A more detailed review of possible solutions and their drawbacks is given in Krolik (2007). Another possible scenario, where the effects of stellar feed- back from a nuclear cluster play a major role, is discussed in Schartmann et al. (2008). In this paper, we address the implications of clumpiness on the temperature structure, the infrared spectral energy distribu- tions, surface brightness distributions as well as interferometric visibilities by implementing fully three-dimensional radiative transfer calculations through a clumpy dust distribution and discuss the possible mechanisms causing this behaviour. In Sect. 2, a description of our model is given, before we present the basic results for our standard model (Sect. 3) and for several parameter studies (Sect. 4) and discuss the findings (Sect. 5), as well as differences and similarities to other models. In Sect. 6 we interpret our results in terms of MIDI interferometric observations and compare them to data for the Circinus galaxy. Finally we draw our conclusions in Sect. 7. 2. The model 2.1. Assemblyofourclumpystandardmodel We apply a very simple, wedge-like torus geometry with a half opening angle of 45◦ in order to gain resolution. In our previous two-dimensional continuous TTM-models (Schartmann et al. 2005), the simulation of the whole θ-range was necessary, due to the radial as well as θ-dependence of the dust distribution. It resulted from an equilibrium between turbulent pressure forces and forces due to an effective potential. The latter is mainly made up of gravitational forces due to the central black hole and the central stellar distribution, as well as rotation. The cloudy dust distribution is set up on a spherical three-dimensional grid M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies 3 Table 1. Main model parameters for our continuous and clumpy standard model. both models inner radius of the torus outer radius of the torus half opening angle of the torus total optical depth in equatorial plane exponent of continuous density distribution number of grid cells in r direction number of grid cells in θ direction number of grid cells in φ direction additional in clumpy model number of clumps exponent of clump size distribution constant of clump size distribution optical depth of each clump average number of cells per clump Rin Rout θopen equ 9.7 µmEφ Dτ α Nclump β a0 clump 9.7 µm τ 0.4 pc 50 pc 45◦ 2.0 -0.5 97 31 120 400 1.0 0.2 pc 0.38 272 r = (r, θ, φ), which is linear in θ and φ and logarithmic in r. To obtain the clumpy density structure, the following proce- dure is applied: A random number generator (RAN2 taken from Press et al. 1992) determines the radial coordinate of the clump centre, which is equally distributed between the inner and outer radius. The θ and φ coordinates are chosen such that the result- ing points are equally distributed on spherical shells. In a second step, the spatial distribution found so far is coupled to the dust density distribution of the continuous model: ρcont(r, θ, φ) = ρ0 r 1 pc!α . (1) The radii of individual clumps aclump vary with distance from the centre according to a distribution 1 pc !β aclump = a0 rclump . (2) All cells within this clump radius are homogeneously filled with clump 9.7 µm, measured dust. All clumps possess the same optical depth τ along a radial ray through the clump centre. We further require that clumps always have to be completely contained within the model space, but are allowed to intersect. Such a combination of intersecting clumps will be called a cloud from now on. Thus, a cloud may contain overdensities, where the intersection happens. A clump size distribution as described above seems to be rea- sonable, as shear forces due to the differential rotation increase towards the centre. Therefore, clouds are more easily disrupted in the inner part of the torus. Furthermore, clouds become com- pressed when moving towards the centre due to the increasing ambient pressure in a deeper potential well. All other routines and algorithms used in this paper are iden- tical to the modelling described in Schartmann et al. (2005) and will only be mentioned briefly in Sect. 2.2. The main model parameters of the continuous and clumpy distributions are summarised in Table 1, where the numerical values refer to our clumpy and continuous standard model. The dust density distribution for the clumpy case is shown in Fig. 1. The torus possesses a volume filling factor of 30% and the dust mass was chosen such that the optical depth of the torus within the equatorial plane (averaged over all angles φ) reaches Fig. 1. 3D rendering of the clump distribution of our standard clumpy torus model. The chosen inclination angle corresponds to a Seyfert 1 type (face-on) view onto the torus. a value of two at 9.7 µm. With this value, the resulting absorp- tion column densities are in concordance with observations of Seyfert type 2 galaxies obtained with the IRS spectrometer on- board Spitzer (see e. g. Shi et al. 2006), and the modelled silicate absorption feature depth compares well with observations. If not stated otherwise, the optical depth always refers to a wavelength of 9.7 µm throughout this paper. equ As in Nenkova et al. (2002), all clumps possess the same op- tical depth in our standard model. A total optical depth within the equatorial plane of Dτ = 2.0 results in an optical depth of clump 9.7 µm = 0.38 along a radial ray through the centre of the clump. τ The corresponding continuous model has the same geometrical structure, continuously filled with dust according to the density distribution given in equation 1. 9.7 µmEφ 2.2. Summaryofmethods A very brief overview of the dust composition, the heating source and the numerical method of radiative transfer will be given in this section. Although several hints (Maiolino et al. 2001a,b; Jaffe et al. 2004) point to the possibility that dust in the nuclear regions of AGN is dominated by large grains, we will limit our present in- vestigation to the classic MRN-model (Mathis et al. 1977) for three reasons: first, and most important, we aim for comparabil- ity with our earlier paper on continuous tori (Schartmann et al. 2005). Second, we have tested that our essential results about the change in grain size distribution (Sect. 3.9 in Schartmann et al. 2005) remain unchanged when distributing the dust in a clumpy structure. Third, our approach, which explicitly takes into ac- count the size-dependent sublimation radius, is generically more robust against changes in the grain distribution than calculations that ignore this effect. For our current simulations, we represent the MRN-model by three different grain species with 5 different grain sizes each. Taking different sublimation radii of the various grains into account then partially accounts for the destruction of small grains in the harsh environment of the quasar, as they pos- sess larger sublimation radii. 4 M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies 2.3. Resolutionstudy In Fig. 2a, we show SEDs for our standard clumpy model (solid line, 5 · 106 monochromatic photon packages) and for the same model, but with twice as many photon packages (107) used for the simulation of the temperature distribution (dotted graphs, identical with solid lines). Despite slight differences in the tem- perature distributions of single grains, we find an almost iden- tical behaviour in the displayed SEDs, with differences smaller than the thickness of the lines. Maps at 12 µm display the same distribution with slight problems along the projected torus axis, which are not visible in the single surface brightness distribu- tions and without any noticeable effect on the interferometric visibility distributions calculated from these maps. In Fig. 2b, the solid curves displays the SEDs for our high spatial resolu- tion standard model and the dotted lines refer to a model with a factor of roughly three less grid cells. Only very small deviations are visible at short wavelengths. Fig. 2 clearly shows that the re- sults and conclusions we draw from our simulations are neither affected by photon noise nor by too low spatial resolution. 3. Analysis of our standard model In most of the SEDs discussed in this paper, only pure dust ree- mission SEDs are shown and an azimuthal viewing angle of 45◦ is used, if not stated otherwise. 3.1. Temperaturedistribution In Fig. 3, the temperature distribution of all cells in all θ and φ directions for the smallest silicate grain component is plotted a) for our clumpy standard model and b) for the corresponding continuous model. The red curves show the radial temperature Fig. 2. a) SEDs for a photon number study. The solid curves re- fer to our standard model and the dotted graphs (identical with the solid curves) result after doubling the number of photon packages. b) SEDs for a resolution study: high resolution (solid curves -- our standard model) and a factor of 3 reduced number of grid cells (dotted curves). Shown are the cases for inclination angles 0◦ and 90◦. The dust distribution is heated by a point-like, central accre- tion disk with the SED of a mean quasar spectrum (see Fig. 3b in Schartmann et al. 2005). The radiation characteristic is chosen to follow a cos(θ) law for all wavelengths. For the simulations shown in this paper, the accretion disk SED is normalised to a bolometric luminosity of 1.2 × 1011 L⊙, except for the compari- son with the Circinus galaxy. In order to obtain the temperature, the SEDs and the sur- face brightness distributions of the dusty torus, we use the three- dimensional radiative transfer code MC3D1 (Wolf et al. 1999; Wolf 2003). We apply the Monte Carlo procedure mainly for the calculation of temperature distributions and the scattering part whenever necessary, whereas SEDs and surface brightness maps for dust reemission are obtained with the included raytracer. The main advantage compared to other codes is MC3D's capability to cope with real three-dimensional dust density distributions, needed for a realistic modelling of the dust reemission from a clumpy torus. For this paper, we implemented the automatic determination of the sublimation surfaces of the various grain species in three dimensions. As we expect the sublimation to happen along irregularly shaped surfaces in a three dimensional, discontinuous model, a raytracing technique is used to solve the (1D) radiative transfer equation approximatively in all directions of the model space. For further information on the radiative transfer proce- dure used and the other preconditions (mainly primary source and dust composition), see Wolf et al. (1999), Wolf & Henning (2000), Wolf (2001, 2003) and Schartmann et al. (2005). 1 MC3D (Monte Carlo 3D) has been tested extensively against other radiative transfer codes for 2D structures (Pascucci et al. 2004) and we also performed a direct comparison for the special case of AGN dust tori with the simulations of Granato & Danese (1994), one of the standard torus models for comparison, calculated with his grid based code (see Fig. 4 in Schartmann et al. 2005). Fig. 3. Comparison between radial temperature distributions (for the smallest silicate grains) in all directions of the clumpy stan- dard model (panel a) with the temperature distribution of the cor- responding continuous model (panel b). The red curve indicates the temperature averaged over all angles θ and φ. M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies 5 profile, averaged over all (θ, φ) directions. It is evident that the spread of temperature values for a given distance from the pri- mary radiation source is much larger for the clumpy models than for the continuous ones. Higher temperatures are possible even in parts of the torus further out, as dust free or optically thin lines of sight exist far out, depending on the distribution of sin- gle clumps. Therefore, a direct illumination of clouds is possible even at large radii. Concerning the continuous model, the scat- ter decreases significantly from 2 pc outwards. Further in, the θ dependent radiation characteristic of the primary source causes greater scatter due to higher temperatures further away from the midplane. 3.2. Viewingangledependence Fig. 4a shows the dependence of the spectral energy distribu- tions on the inclination of the torus. One can only see a clear distinction between lines of sight within the dust-free funnel (0◦ and 30◦ inclinations) and those within the wedge-shaped disk (60◦ and 90◦). This was already reported by Granato & Danese (1994), based on their continuous wedge models. In our case, it is caused by the relatively large volume filling fraction and the large clouds in the outer part of the torus. Therefore, only a weak dependence of the dust density distribution on the po- lar angle exists, which we chose for simplicity. In our previ- ous (2D) modelling (see Schartmann et al. 2005), we obtained the expected smooth transition in the polar direction. In Fig. 4b, the azimuthal angle is varied for a constant inclination angle of 60◦. Nearly identical SEDs result, which is understandable when considering our large volume filling factors. The largest devia- tions appear at the shortest wavelengths, where the emission re- sults from the hottest parts of the torus, which are also the most centrally concentrated parts. Therefore, this wavelength range is most sensitive to changes of the optical depth along the direct line of sight towards the centre. Fig. 5. Inclination angle study of images of clumpy (first two rows) and continuous models (third and fourth row) with two different dust masses at 12 µm. Shown are the extreme cases with half of the mass of the standard model (first and third row) and eight times the mass (second and fourth row). For details of the mass study, see Sect. 4.3. The inclination angles i = 0◦, 60◦, 90◦ are shown in different columns. The scaling is logarithmic with a range between the maximum value of all images and the 10−6th fraction of it (excluding the central point source). Labels are sizes in pc. The dependence on the inclination angle of images is shown in Fig. 5 (upper two rows). It is especially interesting that the different inclination angles look very similar, which was not the case for the continuous model (see lower two rows of Fig. 5). There, the images at larger inclination angles are dominated by the boundaries of the disk, which are not so well defined in the clumpy case. In the zoomed-in images (Fig. 7, upper row), the basic features of our model are directly visible, as one can see the different illumination patterns of clouds: Clouds in the inner- most part are fully illuminated and, therefore, show bright inner rims and cold outer parts. Other clouds are partly hidden behind clouds further in and appear as bright spots only. 3.3. Wavelengthdependency Fig. 4. Dependence of the SEDs on the viewing angle: a) dif- ferent inclination angles for a common azimuthal angle φ = 45◦. Inclination angles shown are: 0◦ (solid line), 30◦ (dotted line), 60◦ (dashed line), 90◦ (dashed-dotted line). b) Different az- imuthal angles φ for a common inclination angle of i = 60◦. Azimuthal angles shown are: 0◦ (solid line), 45◦ (dotted line) and 180◦ (dashed line). Fig. 6 shows the wavelength dependency of our standard model. At short wavelengths, the hottest inner parts dominate the bright- ness distribution. Further out, a few more directly illuminated clumps are visible as bright spots. At the longest wavelengths, emission arises from clumps all over the torus, as colder dust emits strongest at these wavelengths. This dust is spread over a larger volume, due to the steeply decreasing temperature dis- 6 M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies At 0◦ inclination angle (upper row), we observe nearly iden- tical SEDs. The dashed line (corresponding to the fourth col- umn) shows a slightly enhanced flux at short wavelengths, as a larger number of clouds are close to the central source. In the third column, the cloud number density in the central part is the lowest of the three examples. For the case of the middle row (i = 60◦), the largest devi- ations are visible for the case of the dotted line (third column). Here, the silicate feature even appears in emission. This is visi- ble in the surface brightness distributions, as more directly illu- minated clouds are visible on unobscured lines of sight, result- ing in a brighter central region compared to the other two maps. At an inclination angle of 90◦ (third row), absorption along the line of sight increases drastically from the second to the fourth column, visible in a deepening of the silicate absorption feature and the darkening of the central region of the surface brightness distributions. 4.2. Differentvolumefillingfactors Starting from the standard model with a volume filling factor of 30% and 400 clumps within the whole model space, we halved it once by distributing only 160 clumps within the calculation domain and doubled it, for which 1500 clumps were needed due to the applied procedure of randomly distributing clumps. The resulting surface brightness distributions at λ = 12.0 µm are shown in Fig. 8 with the three models given in differ- ent rows and for three different inclination angles: i = 0◦, 60◦, 90◦. In the case of the lowest volume filling factor, individual clouds are visible. The distribution of the surface brightness of these individual clouds reflects the temperature structure within single clumps. The directly illuminated clumps are hotter and, therefore, appear brighter. When adding more and more clumps Fig. 8. Different volume filling factors: 15% (upper row), 30% (middle row), 60% (lower row) for λ = 12.0 µm. From left to right, the inclination angle changes: i = 0◦, 60◦, 90◦. The scaling is logarithmic with a range between the maximum value of all images and the 10−6th fraction of it (excluding the central point source). Length scales are given in pc. Fig. 6. Wavelength dependence of the surface brightness distri- butions: λ = 4.6 µm (upper row), λ = 9.7 µm (second row), λ = 12.0 µm (third row) and λ = 30.2 µm (lower row). Within the rows, the inclination angle changes from face-on view (left- most panel) over 60◦ to 90◦ (rightmost panel). The images are given in logarithmic scaling with a range of values between the global maximum of all images and the 10−5th fraction of it (cen- tral source excluded). Labels are in pc. tribution at small radii. Furthermore, the extinction curve has dropped by a large factor at these wavelengths and, therefore, the torus becomes optically thin and the whole range of cloud sizes is visible. 4. Parameter variations 4.1. Differentrealisationsoftheclumpydistribution As already discussed in Schartmann et al. (2005), SEDs of dust reemission depend strongly on the distribution of dust in the in- nermost region. Changing the random arrangements of clumps -- as done in this section -- therefore is expected to cause sig- nificant changes of the SEDs, especially for the case of a small number of clouds. The second important parameter is the opti- cal depth of the single clumps. The larger it is, the stronger is the dependence of the SEDs on the dust distribution in the in- nermost region. For the case of our modelling, the small number of clouds is expected to cause large differences in the observed SEDs. But this effect is partially compensated -- in most of the simulations -- by optically thin individual clumps, resulting in a more similar behaviour of the SEDs. Looking at the simulated SEDs and matching them with the images, the following results can be seen (compare to Fig. 7): M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies 7 Fig. 7. Different random arrangements of clumps. The rows show three different inclination angles: 0◦ (first row), 60◦ (second row), 90◦ (third row). Given in columns are the SEDs (first column) and the images at 12 µm for the three different random cloud arrangements (column 2 to 4), all having the same parameters (see Table 1). Here, the solid line in the SED corresponds to the first column of images, the dotted line to the second and the dashed line to the third column. Images are given in logarithmic color scale ranging from the maximum of all images to the 10−4th part of it (excluding the central source). Labels denote distance to the centre in pc. (increasing the volume filling factor), the chance of directly illu- minating clumps further out decreases and at higher filling fac- tors it is only possible for clumps close to the funnel. This is clearly visible at higher inclination angles: the higher the vol- ume filling factor, the clearer the x-shaped feature appears, as only clumps within or close to the funnel can be directly illu- minated. At a volume filling factor of 60%, the surface bright- ness distribution looks very similar to that of the corresponding continuous model (compare to Fig. 5). For large volume filling factors and close to edge-on, substructure is only visible from clouds in a viewing direction towards the dust-free cones. The corresponding SEDs are shown in Fig. 9. With increas- ing filling factor, more and more flux at short wavelengths ap- pears for the face-on case, as seen in Schartmann et al. (2005). The shape of the clumpy model SEDs resemble the correspond- ing continuous model most (compare to Fig. 10, right column) for the highest volume filling factor. Concerning the silicate fea- ture, it increases slightly in emission as the amount of dust at the appropriate temperature increases as well (visible at the tran- sition from the lowest to the medium volume filling factor). The silicate absorption feature at higher inclinations strongly de- pends on the viewing angle (compare Fig. 9b and c) especially for the model with the least number of clumps (dotted line). Thus, this study shows the validity of the simplification of using a smooth dust distribution in the case of very high torus volume filling factors, as was assumed in previous simulations. 4.3. Dustmassstudy Fig. 9. Different volume filling factors: 15% (dotted line), 30% (solid line), 60% (dashed line). Rows show SEDs for the three inclination angles: i = 0◦, 60◦, 90◦. The viewing angle in φ- direction is always 45◦. To study the dependence of the SEDs on the optical depth of the torus, we carried out a study with 0.5, 1, 2, 4 and 8 times the dust mass in the standard model. This leads to an opti- cal depth at 9.7 µm within the equatorial plane, averaged over = 1, 2, 4, 8, 16. Single clumps equ all angles of φ of Dτ 9.7 µmEφ then change from optically thin to optically thick (τ 0.19, 0.38, 0.76, 1.52, 3.04). clump 9.7 µm = The resulting behaviour of the SEDs is shown in Fig. 10, where it is also compared to the corresponding continuous mod- 8 M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies Fig. 10. SEDs for different enclosed dust masses. The left column shows the case for the clumpy models and a face-on view (upper row) as well as an edge-on view (lower row). In the right column, continuous models are displayed. The solid line corresponds to the standard model, the dotted to half of the mass, the dashed double the mass, the dash-dotted to four times the mass and the dash-triple-dotted to eight times the mass of the standard model. tivated shape. Furthermore, it also involves very deep and so far unobserved silicate absorption features for the edge-on case (see lower right panel in Fig. 10). Concerning the clumpy model, the explanation for the flattening of the silicate feature in the edge- on case with increasing dust mass of the torus will be given in Sect. 5.1. The Wien branches show different behaviour when looking face-on. For the case of the clumpy torus model, increasing the optical depth means that the Wien branch moves to larger wave- lengths, as expected for the edge-on case. This is understandable when most of the directly illuminated surfaces of the clouds are then hidden behind other clouds, an argument which is not valid if the clouds are too optically thin in the inner part. For the edge-on case, we qualitatively obtain a comparable be- haviour as in the continuous case, because of the large number of clumps and the same optical depth within the equatorial plane. But a very important difference can be seen in the appearance of the silicate feature in absorption: when we want to have only very weak silicate emission features in the face-on case, a large optical depth is needed, resulting in an unphysically deep silicate feature in absorption in the edge-on case of the continuous mod- els, whereas the silicate feature remains moderate for many lines of sight for the clumpy model, where we see a large scatter for different random arrangements of clumps (compare to Fig. 7). Concerning surface brightness distributions (see Fig. 5), one can see that the objects appear smaller at mid-infrared wave- lengths for the case of higher dust masses: the larger the op- tical depth, the brighter the inner region and the dimmer the outer part. This is caused by a steepening of the radial dust tem- perature distribution with increasing mass of the objects, as the probability of photon absorption increases in the central region. Especially for the continuous case, the asymmetry at intermedi- ate inclination angles becomes visible for larger optical depths caused by extinction on the line of sight due to cold dust in the outer parts of the torus. 4.4. Concentrationofclumpsinradialdirection As already described in the model section (2), clump posi- tions are also chosen in accordance with the density distribu- tion of the corresponding continuous model. Therefore, chang- Fig. 11. Close-up of the spectrum between 7 and 15 µm of the face-on case of the dust mass study for the clumpy model in linear display. The lines are defined as in Fig. 10. els. Concerning the silicate feature (in emission) for the face-on case (top row), we see a very similar behaviour of the SEDs of the clumpy and continuous model. Increasing the mass and with it the total optical depth leads to a flattening of the SED around the silicate feature, even more pronounced in the con- tinuous case. In addition to that, a slight shift of the maximum of the silicate feature towards longer wavelengths is visible for the case of the highest dust mass, apparent in the zoom-in of Fig. 10 around the silicate feature (see Fig. 11). This is due to the increasing underlying continuum towards longer wavelengths. Although the principal behaviour of the silicate feature is iden- tical for our clumpy and our continuous model, the reasons dif- fer: In the case of a continuous wedge-like torus, the inner, di- rectly illuminated walls are only visible through a small amount of dust. From step to step, the walls become more opaque and shield the directly illuminated inner rim better, decreasing the height of the silicate feature. This was not the case for our con- tinuous TTM-models in Schartmann et al. (2005). With them, it was not possible to significantly reduce the silicate feature height within reasonable optical depth ranges. This was caused by the fully visible, directly illuminated inner funnel. Therefore, in the wedge-shaped continuous models, the reduction of the feature is an artefact caused by the unphysical, purely geometrically mo- M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies 9 The same behaviour is visible in the corresponding SEDs shown in Fig. 13. Decreasing the amount of dust in the centre near the heating source leads to decreasing flux at near-infrared wavelengths, whereas the flux at far-infrared wavelengths in- creases (reflecting the enhancement of dust in the outer part). Fig. 12. Surface brightness distributions for various slopes of the density distribution in the corresponding continuous model (ρcont ∝ rα) leading to different concentrations of clumps in the radial direction in the clumpy model. The slopes are: α = 0.0 (upper row), α = −0.5 (middle row), α = −1.0 (lower row). From left to right, the inclination angle increases from face-on to edge-on: i = 0◦, 60◦, 90◦. Shown are images at 12 µm with a logarithmic color scale ranging from the maximum of all images to the 10−6th part of it. Labels denote the distance to the centre in pc. ing the slope of this radial density distribution, defined to be ρcont(r, θ, φ) = ρ0 (cid:16) r 1 pc(cid:17)α, leads to a different concentration of clumps along radial rays. In this section, we vary the slope of the distribution α from a homogeneous dust distribution (α = 0.0) over α = −0.5 (our standard model) to α = −1.0. Decreasing α leads to an enhancement of the clump number density towards the central region. In order to keep the volume filling fraction at a constant level of 30%, we need to increase the number of clumps, as their size decreases towards the central region. All clumps possess the same optical depth. In order to have a con- stant mean optical depth in the midplane, the total dust mass has to be decreased. For an overview of the modified parameters see Table 2. The change of clump concentration can be seen directly from the simulated images at 12.0 µm in Fig. 12, especially in the face-on case (first column). In the upper panel, single reemitting clumps are visible in the central region. This changes more and more to a continuous emission for the case of the highest cloud concentration in the centre due to multiple clumps along the line of sight and intersecting clumps. At higher inclination angles, the higher concentration leads to a sharper peak of the surface brightness. Table 2. Varied parameters of the clump concentration study. Parameter No. of clumps Dust mass [M⊙] α = 0 250 22950 α = −0.5 α = −1 400 12562 900 6418 Fig. 13. SEDs for the clump concentration study. The varied slopes of the underlying density distribution are: α = 0.0 (dotted line), α = −0.5 (solid line), α = −1.0 (dashed line): a) face-on, b) edge-on. 4.5. Dependenceontheclumpsizedistribution In our clumpy standard model, a radially changing clump size proportional to the radial distance to the centre was chosen. In this section, we test the effects of decreasing the slope β of the radial size distribution aclump = a0 (cid:16) rclump 1 pc (cid:17)β of the clumps. This is done in a way that the volume filling fraction as well as the op- tical depth in the midplane, averaged over all azimuthal angles φ, remain constant. It is achieved by changing the proportional- isation constant of the clump size distribution a0 and the total dust mass of the torus. Doing this results in very well resolved clumps in the inner part. Beyond a distance of approximately 25 pc, the number of grid cells per clump drops below the value of our standard model. Surface brightness distributions for the extreme case of a constant clump size are shown in Fig. 14. Due to the large clump radius of 5 pc even in the central region and that fractions of clumps at the model space boarder are prohibited, it leads to a density distribution with a quite large, unevenly shaped central cavity, as can be seen in the face-on view (left panel of Fig. 14). The inner rim is given by only a few intersecting clumps, instead of the otherwise defined spherical central cavity. Therefore, in the edge-on case, the surface brightness distribution shows an in- ner boundary, which is bent towards the centre (convex shaped). In these models, due to the large clump size in the inner region, many clumps intersect, producing a nearly continuous dust dis- tribution at the inner boundaries, which lets the -- typical for con- tinuous models -- x-shaped structure appear again. For the same reason, the extinction band due to the cos(θ)-radiation char- 10 M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies acteristic is visible in the edge-on view. Especially at the 90◦ inclination angle, single clumps are directly visible (above and below the centre). In these cases, their shading directly shows the illumination pattern due to the primary source (accretion disk), emission from other clumps and extinction from the foreground dust distributions. The corresponding SEDs (Fig. 15) mainly reflect the in- crease of the inner cavity and, therefore, the lack of flux at short wavelengths. The convex shape of that region causes a larger di- rectly illuminated area at the funnel walls and, therefore, slightly strengthens the silicate emission feature in a face-on view. A dif- ferent appearance (emission/absorption) of the 10 µm feature at i = 60◦ (middle panel) is seen. This is due to the lower number density of clumps in the inner part, enforced by the restriction of having only whole clumps within the model space. A dust mass study for the case of the large, constant diameter clump model (β = 0) reveals the same behaviour as discussed in Sect. 4.3 when looking edge-on onto the torus. However, the face-on case differs: only the relative height of the silicate feature changes slightly. This was already observed in our TTM-models in Schartmann et al. (2005) and is due to the now inwardly bent inner walls of the funnel (see Fig. 14, right panel), caused here by the very large and spherical clumps in the innermost torus region. 5. Discussion 5.1. Explanationforthereductionofthesilicatefeature The results shown in the subsections before can be explained with the following model, which was partially discussed by Nenkova et al. (2002). It is illustrated in Fig. 16, where yellow denotes directly illuminated clump surfaces. Many of the ex- plained features can also be seen in the zoomed-in versions of the surface brightness distributions (Fig. 7) for the face-on case (upper row). As already pointed out in Schartmann et al. (2005), the SEDs of dust tori in the mid-infrared wavelength range are mainly de- termined by the inner few parsecs of the toroidal dust distribu- tion. In each of the central clouds of the clumpy model, the dust temperature drops from the inner directly illuminated edge to- wards the cloud's outer surface. With an inclination angle close to i = 90◦, we expect -- for realistic volume filling factors -- comparable behaviour of the SED as in the continuous case. The silicate feature has a smaller Fig. 15. Dependence on the clump size for different inclination angles (rows: 0◦, 60◦, 90◦). The solid line corresponds to our standard model with β = 1. For the dashed line, clumps have equal size (aclump = 5 pc, β = 0), independent of the radial po- sition and the dotted line corresponds to an intermediate model with β = 0.5. depth, as discussed in Sect. 4.3. But the situation changes with decreasing inclination angle. Here, one has to distinguish be- tween different cases: 1. With a relatively high volume filling factor and a not too small extension of the clouds in the central region of the torus, it is likely that the directly illuminated part of most of the clouds is hidden from direct view by other clouds. Fig. 14. Images at 12 µm and inclination angles i = 0◦ and 90◦ for a clumpy distribution with a constant clump size, indepen- dent of the radial position. The scaling is logarithmic with a range between the maximum value of all images and the 10−6th fraction of it (excluding the central point source). Fig. 16. Sketch of our clumpy torus model. Indicated in yellow are directly illuminated surfaces of the clumps. i is the inclina- tion angle, θopen is the half opening angle of the torus. M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies 11 Therefore, the directly illuminated surface area is reduced compared to the corresponding continuous model. As this area is responsible for the emission fraction of the silicate feature within the SED, it shows less silicate emission. In order to produce such a shadowing effect, clouds have to possess a large enough optical depth, which means that they have to be either small or massive (τclump ∝ mclump a−2 clump, where mclump is the mass of the clump). 2. For the case of a too small optical depth of the clumps in the innermost region, we expect the silicate feature to appear in strong emission. 3. Another possibility of producing silicate features in emission is when the model possesses a small number of clumps in the inner part, making the shadowing effect inefficient. 4. A further effect on the silicate feature strength arises from the grain dependent sublimation implemented in our models. As graphite grains possess a higher sublimation temperature as silicate grains, they are able to partially shelter the silicate grains from direct irradiation. Thus, the strength of the silicate feature is mainly deter- mined by the distribution, size and optical depth of the clouds in the direct vicinity of the sublimation surface of the dust. We will see in the next section that this finding well explains the fact that Nenkova et al. (2002) and Dullemond & van Bemmel (2005) come to different conclusions concerning the reduction of the strength of the silicate feature due to clumpiness, as the distribution (and size) of their clouds differ. 5.2. Comparisonwithothertorusmodels The results of Nenkova et al. (2002) - the pioneering work in the field of clumpy tori - are broadly consistent with the explanations given in Sect. 5.1 of this paper. Dullemond & van Bemmel (2005) model 2D clumps in the form of rings with a two-dimensional radiative transfer code. In contradiction to all other simulations, no systematic reduction of the silicate feature due to clumpiness is found. The reason for this is understandable with the explanations given in Sect. 5.1, as their model features a small clump number density in the central region and shadowing effects are rather small. Therefore, they find both strengthening of the silicate feature and reduction, de- pending on the random ring distribution. A comparison of our clumpy standard model and two other random cloud distributions with simulations by Honig et al. (2006) is shown in Fig. 17. They follow a different, multi- step approach: 2D radiative transfer calculations of individual clouds at different positions and with various illumination pat- terns within the torus are carried out. In a second step, the SED of the total system is calculated. The cloud distribution and parame- ters such as optical depth or size arise from an accretion scenario of self-gravitating clouds close to the shear limit (Vollmer et al. 2004; Beckert & Duschl 2004). The advantage of this approach is that resolution problems can be overcome easily, as only 2D real radiative transfer calculations of single clumps are needed. Characteristics for their modelling are small cloud sizes with very high optical depths in the inner part of the torus and a large number of clumps. For comparison, a cloud at the sublimation radius of their model has a radial size of Rcloud = 0.02 pc with clump an optical depth of τ 0.55 µm ≈ 250. In our standard model, clouds at the sublimation radius are four times larger and possess an clump optical depth of only τ 0.55 µm ≈ 3. The large optical depth in the innermost part in combination with the large number den- sity there reduces the silicate feature significantly by shadowing with respect to their single clump calculations. Their finding that the silicate emission feature can be reduced further by increas- ing the number density of clumps in the innermost part perfectly fits our explanation presented in Sect. 5.1. Deviations between the two approaches (see Fig. 17) are mainly due to the approxi- mately eight times larger primary luminosity and the larger op- tical depth, at least in the midplane of the Honig et al. (2006) modelling compared to our standard model. Furthermore, in our simulations only dust reemission SEDs are shown. This leads to relatively higher fluxes at short wavelengths compared to long wavelengths for the i = 30◦ case (Fig. 17a) and to more ex- tinction within the midplane and, therefore, a shift of the Wien branch towards longer wavelengths in the edge-on case (lower panel). 6. MIDI interferometry Even with the largest single-dish mid-infrared telescopes, it is impossible to directly resolve the dust torus of the nearest Seyfert galaxies. Therefore, interferometric measurements are needed. Recently, Jaffe et al. (2004) succeeded for the first time to resolve the dusty structure around an AGN in the mid-infrared wavelength range. In this case, they probed the active nucleus of the nearby Seyfert 2 galaxy NGC 1068 with the help of the MID- infrared interferometric Instrument (MIDI, Leinert et al. 2003). It is located at the European Southern Observatory's (ESO's) Very Large Telescope Interferometer (VLTI) laboratory on Cerro Paranal in Chile. Its main objective is the coherent combination of the beams of two 8.2 m diameter Unit Telescopes (UTs) in order to obtain structural properties of the observed objects at high angular resolution. A spatial resolution of up to λ/(2 B) ≈ 10 mas at a wavelength of λ = 10 µm can be obtained for the largest possible separation of two Unit Telescopes of B ≈ 120 m. Operating in the N-band (8 − 13.5 µm), it is perfectly suited to detect thermal emission of dust in the innermost parts of nearby Fig. 17. Comparison of our clumpy standard model and two other random realisations of the clump distribution (blue lines) with simulations done by S. F. Honig (private communication, described in Honig et al. 2006), shown by the yellow lines, for 10 different random realisations of their model. The latter are scaled with a factor of 2.2 in order to give rough agreement be- tween the two models (see text for further explanation). 12 M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies Fig. 18. Visibilities of our continuous standard model at a wave- length of 12 µm plotted against the projected baseline length. Colour of the visibility distributions refers to different position angles of the projected baseline w. r. t. the torus axis. Each panel shows a different inclination angle, as indicated in the upper right corner. Fig. 19. Visibilities of our clumpy standard model at a wave- length of 12 µm plotted against the projected baseline length. Colour of the visibility distributions refers to different position angles of the baseline w. r. t. the torus axis. Each panel shows a different inclination angle, as indicated in the upper right corner. Seyfert galaxies. MIDI is designed as a classical Michelson in- terferometer. Being a two-element beam combining instrument, it measures so-called visibility amplitudes. Visibility is defined as the ratio between the correlated flux and the total flux. Its in- terpretation is not straightforward, since no direct image can be reconstructed. Therefore, a model has to be assumed, which can then be compared to the visibility data. MIDI works in dispersed mode, which means that visibilities for the whole wavelength range are derived. The dust emission is probed depending on the orientation of the projected baseline. Point-like objects result in a visibility of one, as the correlated flux equals the total flux. The more extended the object, the lower the visibility. With the help of a density distribution, surface brightness distributions in the mid-infrared can be calculated by applying a radiative trans- fer code. A Fourier transform of the brightness distribution then yields the visibility information, depending on the baseline ori- entation and length within the so-called U-V-plane (or Fourier- plane). The main goal of the following analysis is to investigate whether MIDI can distinguish between clumpy and continuous torus models of the kind presented above. Furthermore, we try to derive characteristic features of the respective models and show a comparison to data obtained for the Circinus galaxy. 6.1. Modelvisibilities In Fig. 18, calculated visibilities for four inclinations of our con- tinuous standard model at a wavelength of λ = 12 µm are shown. Various orientations of the projected baseline are colour coded (the given position angle is counted anti-clockwise from the pro- jected torus axis). Due to the axisymmetric setup, all lines coin- cide for the face-on case. For all other inclination angles, visibili- ties decrease until a position angle of 90◦ is reached and increase symmetrically again. This means that the torus appears elon- gated perpendicular to the torus axis at this wavelength. Fig. 19 shows the same study, but for the corresponding clumpy model. The basic behaviour is the same, but the visibilities show fine structure and the scatter is much greater, especially visible in the comparison of the i = 0◦ cases. Furthermore, while all of the curves of the continuous model monotonically decrease with baseline length, we see rising and falling values with increasing baseline length for the same position angle in the clumpy case. In addition, for the continuous models, curves do not intersect, in contrast to our clumpy models. However, to detect such fine structure in observed MIDI data, a very high accuracy in the vis- ibility measurements of the order of σv ≈ 0.02 and a very dense sampling is required. In Fig. 20, the wavelength dependence of the visibilities is shown. The first two panels represent the case of the clumpy standard model and the third and fourth the continuous model. Each of the two panels of the respective model visualises a dif- ferent position angle (counted anti-clockwise from the projected torus axis). An inclination angle of 90◦ is used in all panels. Three different wavelengths are colour coded: 8.2 µm at the be- ginning of the MIDI-range (black dotted line), 9.8 µm within the silicate feature (blue) and 12.6 µm at the end of the MIDI wave- length range (yellow), outside the silicate feature. While the con- tinuous model results in smooth curves (see also Fig. 18), much fine structure is visible for the case of the clumpy model. The differences between the displayed wavelengths relative to the longest wavelength are smaller for the clumpy models than in the continuous case. Fig. 21 shows visibilities for our clumpy standard model at 12 µm, plotted against the position angle (counter-clockwise from the projected torus axis). Baselines are colour coded be- M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies 13 Fig. 20. Visibilities of our clumpy (first two panels) and continuous (last two panels) standard model at different wavelengths (colour coded), plotted against the projected baseline length for the two position angles 0◦ (in torus axis direction) and 90◦ (along the midplane) for an edge-on view onto the torus (i = 90◦). long computational times of the order of 30 to 40 hours per in- clination angle (including calculation of the temperature distri- bution, the SED and surface brightness distribution). Therefore, we applied the following procedure: From our experience with modelling the SED of the Circinus galaxy with our previously used continuous Turbulent Torus Models (see Schartmann et al. 2005), we adopt the size of the object used there. Furthermore, we tried to stay as close to our clumpy standard model as possi- ble (for the parameters of the clumpy standard model, compare to Table 1) and copied the parameters α, β and a0. The rest of the parameters were changed, in order to obtain the best pos- sible adaptation to the data, within the investigated parameter range. The comparison of our current clumpy Circinus model as described above and in Table 3 (yellow stars) to interferomet- ric observations with MIDI (Tristram et al. 2007) of the Circinus galaxy (black) is shown in Fig. 22. In contrast to the presenta- tion of continuous visibility curves above, single measurements of combinations of various baseline lengths and position angles are displayed in this plot. Position angle now refers to the an- gle on the sky measured from north in a counter-clockwise di- rection. The rotation axis of our simulated torus has a position angle of approximately −45◦ according to this definition. The black numbers denote the length of the projected baseline (given in m) of the corresponding data point. From the approximate correspondence of the model values with the data, one can see that the size of the emitting region at the two wavelengths is re- produced quite well. Most of the local extrema of the curve can be reproduced for the case of 9.1 µm. Larger deviations are vis- ible for λ = 12.0 µm. The good adaptation partly is due to the changes in baseline length. Longer baselines naturally result in smaller visibilities, as we are probing smaller and smaller struc- tures (see also Fig. 19). Greater visibilities for shorter or equal Table 3. Circinus model parameters: For an explanation of the parameters see Sect. 2 and Table 1. MBH is the mass of the cen- tral black hole (from Greenhill et al. 2003) and Ldisk/Ledd is the Eddington luminosity ratio resulting for the assumed luminosity of the central source. Parameter Rin Rout θopen equ 9.7 µmEφ Dτ MBH α Value 0.6 pc 30 pc 65◦ 3.9 1.7 · 106 M⊙ -0.5 Parameter Value 500 Nclump 1.0 β a0 clump τ 9.7 µm Ldisk/Ledd 0.2 pc 0.96 30% Fig. 21. Visibilities of our clumpy standard model at different inclination angles (as annotated in the upper right corner) plotted against the position angle for various projected baseline lengths (colour coded) and a wavelength of 12 µm. tween 20 m and 100 m in steps of 20 m. A longer baseline means that structures are better resolved, leading to decreasing visibili- ties. For the case of inclination angles close to edge-on, the vis- ibility distribution changes from more or less flat to a character- istic oscillating distribution at longer baselines (from 60 m on- wards) with minima around 100◦ and 300◦. This means that our torus model seems to be more elongated within the equatorial plane and has the smallest width along the projected torus axis at this wavelength. But this only applies for the innermost part; the torus as a whole looks approximately spherically symmetric. At small inclination angles no such favoured size distribution is visible. 6.2. ComparisonwithMIDI-datafortheCircinusgalaxy Unfortunately, a fitting procedure involving a large parameter study is not possible with our current model, due to the very 14 M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies baselines and similar position angle, therefore, have to be due to those curves in Fig. 19 with increasing visibility with base- line or a very inhomogeneous distribution of dust with position angle. Both can be interpreted as signs of clumpiness. The SED of the same Circinus model is plotted over current high resolu- tion data in Fig. 23. The NIR (near-infrared) data points were ob- tained with the NACO camera at the VLT and corrected for fore- ground extinction by AV = 6 mag (Prieto et al. 2004). Different symbols refer to various aperture sizes (see figure caption). The thick green line shows the MIDI spectrum (Tristram et al. 2007) and the black line is our Circinus model as discussed above for an aperture of 0.4′′ in radius; the yellow line denotes the same model, but calculated for the whole simulated model space. Both modelled SEDs include the direct radiation of the central source (calculated with real Monte Carlo radiative transfer), which in these examples dominates over dust reemission for the small wavelength part from about 2 to 3 µm downwards and shows some noise, due to the low photon packet numbers used. In con- tradiction to our continuous Circinus model in Schartmann et al. (2005), enough nuclear radiation can be observed in order to ex- plain the turnover of the SED at small wavelengths and we do not need to assume scattering by material (dust and electrons) within the torus funnel. As can be seen from these figures, our model is able to qualitatively explain the SED as well as the visi- bility information. However, as we are not able to investigate the whole parameter range of our models, we cannot exclude that a different parameter set can describe the data equally well. This degeneracy problem was already pointed out by Galliano et al. (2003) for the case of SED fitting. Adding new clumpiness pa- rameters will even strengthen this degeneracy. On the other hand, adding more data such as more visibility information will place more constraints and will weaken this problem. 7. Conclusions In this paper, we implemented a new clumpy torus model in three dimensions. For computational reasons, a wedge-like shaped disk is used. In the discussion of our results, we place special emphasis on the comparison with continuous models and their differentiation using interferometric observations, such as with MIDI. In Schartmann et al. (2005), we had found that the SEDs of AGN tori in the mid-infrared wavelength range are mainly de- termined by the innermost part of the torus. With the presented clumpy torus models, this claim can be further strengthened. According to the new simulations, the silicate feature strength is mainly determined by the number density and distribution, as well as the optical depth and size of the clumps in the inner re- gion. With a sufficiently high optical depth of the clouds in the inner part, shadowing effects become important, which hide the illuminated cloud surfaces from direct view and, thereby, reduce the silicate feature in emission. At the same time, enough lines of sight with low optical depth remain so that only weak absorp- tion features result for the edge-on case. Continuous models with special and unrealistic morphologies (like the wedge-shaped tori used here) are also able to weaken the silicate emission fea- ture for the face-on view when applying an anisotropic radiation characteristic, but fail to simultaneously account for moderate absorption features, when looking edge-on to the torus. Due to the large clumps in our model, appreciable scatter in SEDs for different random realisations of the torus are ex- pected. A contrary effect is caused by the small optical depth of the single clumps and also of many dust-free lines of sight towards the centre. Direct comparison between calculated inter- ferometric visibilities for clumpy and the corresponding contin- uous models show that clumpy models naturally possess more fine structure, which can partly be resolved by MIDI. We also showed that these kinds of models are able to quali- tatively describe the available interferometric visibility and high resolution spectroscopic data of the Circinus galaxy at the same time. Currently, it is one of the best studied Seyfert galaxies in terms of mid-infrared visibility measurements (Tristram et al. 2007). The decreasing slope of the SED at short wavelengths can be described with our clumpy model, whereas it was at odds with the continuous model described in Schartmann et al. (2005). Acknowledgements. We would like to thank the anonymous referee for com- ments, as well as C. P. Dullemond for useful discussions and S. F. Honig for pro- viding some of his torus models for the comparison with our work. S. W. was supported by the German Research Foundation (DFG) through the Emmy Noether grant WO 857/2. References Antonucci, R. 1993, Ann. Rev. Astron. Astrophys., 31, 473 Beckert, T. & Duschl, W. J. 2004, Astron. Astrophys., 426, 445 Dullemond, C. P. & van Bemmel, I. M. 2005, Astron. Astrophys., 436, 47 Galliano, E., Alloin, D., Granato, G. L., & Villar-Mart´ın, M. 2003, Astron. Astrophys., 412, 615 Granato, G. L. & Danese, L. 1994, Mon. Not. R. Astron. Soc., 268, 235 Greenhill, L. J., Booth, R. S., Ellingsen, S. P., et al. 2003, Astrophys. J., 590, 162 Hao, L., Spoon, H. W. W., Sloan, G. C., et al. 2005, Astrophys. J., Lett., 625, L75 Honig, S. F., Beckert, T., Ohnaka, K., & Weigelt, G. 2006, Astron. Astrophys., 452, 459 Jaffe, W., Meisenheimer, K., Rottgering, H. J. A., et al. 2004, Nature, 429, 47 Konigl, A. & Kartje, J. F. 1994, Astrophys. J., 434, 446 Krolik, J. H. 2007, Astrophys. J., 661, 52 Krolik, J. H. & Begelman, M. C. 1988, Astrophys. J., 329, 702 Leinert, C., Graser, U., Waters, L. B. F. M., et al. 2003, in Interferometry for Optical Astronomy II. Edited by Wesley A. Traub . Proceedings of the SPIE, Volume 4838, pp. 893-904 (2003)., ed. W. A. Traub, 893 -- 904 Maiolino, R., Marconi, A., & Oliva, E. 2001a, Astron. Astrophys., 365, 37 Maiolino, R., Marconi, A., Salvati, M., et al. 2001b, Astron. Astrophys., 365, 28 Maiolino, R. & Rieke, G. H. 1995, Astrophys. J., 454, 95 Manske, V., Henning, T., & Men'shchikov, A. B. 1998, Astron. Astrophys., 331, 52 Mathis, J. S., Rumpl, W., & Nordsieck, K. H. 1977, Astrophys. J., 217, 425 Miller, J. S. & Antonucci, R. R. J. 1983, Astrophys. J., Lett., 271, L7 Nenkova, M., Ivezi´c, Z., & Elitzur, M. 2002, Astrophys. J., Lett., 570, L9 Pascucci, I., Wolf, S., Steinacker, J., et al. 2004, Astron. Astrophys., 417, 793 Pier, E. A. & Krolik, J. H. 1992a, Astrophys. J., 401, 99 Pier, E. A. & Krolik, J. H. 1992b, Astrophys. J., Lett., 399, L23 Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. 1992, Numerical recipes in FORTRAN. The art of scientific computing (Cambridge: University Press, 1992, 2nd ed.) Prieto, M. A., Meisenheimer, K., Marco, O., et al. 2004, Astrophys. J., 614, 135 Risaliti, G., Elvis, M., & Nicastro, F. 2002, Astrophys. J., 571, 234 Schartmann, M., Meisenheimer, K., Camenzind, M., Wolf, S., & Henning, T. 2005, Astron. Astrophys., 437, 861 Schartmann, M., Meisenheimer, K., Klahr, H., et al. 2008, Mon. Not. R. Astron. Soc., in preparation Shankar, F., Salucci, P., Granato, G. L., De Zotti, G., & Danese, L. 2004, Mon. Not. R. Astron. Soc., 354, 1020 Shi, Y., Rieke, G. H., Hines, D. C., et al. 2006, Astrophys. J., 653, 127 Siebenmorgen, R., Haas, M., Krugel, E., & Schulz, B. 2005, Astron. Astrophys., 436, L5 Sturm, E., Schweitzer, M., Lutz, D., et al. 2005, Astrophys. J., Lett., 629, L21 Tristram, K. R. W., Meisenheimer, K., Jaffe, W., et al. 2007, Astron. Astrophys., 474, 837 Urry, C. M. & Padovani, P. 1995, Publ. Astron. Soc. Pac., 107, 803 van Bemmel, I. M. & Dullemond, C. P. 2003, Astron. Astrophys., 404, 1 Vollmer, B., Beckert, T., & Duschl, W. J. 2004, Astron. Astrophys., 413, 949 Weedman, D. W., Hao, L., Higdon, S. J. U., et al. 2005, Astrophys. J., 633, 706 Wolf, S. 2001, Ph.D. Thesis Wolf, S. 2003, Astrophys. J., 582, 859 Wolf, S. & Henning, T. 2000, Computer Physics Communications, 132, 166 Wolf, S., Henning, T., & Stecklum, B. 1999, Astron. Astrophys., 349, 839 M. Schartmann et al.: Three-dimensional radiative transfer models of clumpy tori in Seyfert galaxies 15 Fig. 22. Comparison of model visibilities (yellow stars and lines) for an azimuthal viewing angle of φ = 225◦ with MIDI observations for two different wavelengths. The base- line length for all data points is given above the data. Data courtesy of Tristram et al. (2007). Fig. 23. Comparison of model SEDs with data for the Circinus galaxy. Different symbols refer to various aperture radii: blue stars -- 0.38′′ (NACO), red rectangle -- 0.1′′ (HST/NICMOS) and black triangle -- 1.0′′ (ESO/TIMMI2). Data compilation by Prieto et al. 2004. The thick green line shows the total MIDI spectrum (Tristram et al. 2007). Our model (see model parameters in Table 3) is calculated for an aperture radius of 0.4′′ (black line) and the total model space (yellow line).
astro-ph/9809132
3
9809
1998-10-21T17:26:37
The Distance to the Large Magellanic Cloud from the Eclipsing Binary HV2274
[ "astro-ph" ]
The distance to the Large Magellanic Cloud (LMC) is crucial for the calibration of the Cosmic Distance Scale. We derive a distance to the LMC based on an analysis of ground-based photometry and HST-based spectroscopy and spectrophotometry of the LMC eclipsing binary system HV2274. Analysis of the optical light curve and HST/GHRS radial velocity curve provides the masses and radii of the binary components. Analysis of the HST/FOS UV/optical spectrophotometry provides the temperatures of the component stars and the interstellar extinction of the system. When combined, these data yield a distance to the binary system. After correcting for the location of HV2274 with respect to the center of the LMC, we find d(LMC) = 45.7 +/- 1.6 kpc or DM(LMC) = 18.30 +/- 0.07 mag. This result, which is immune to the metallicity-induced zero point uncertainties that have plagued other techniques, lends strong support to the ``short'' LMC distance scale as derived from a number of independent methods.
astro-ph
astro-ph
TO APPEAR IN ASTROPHYSICAL JOURNAL LETTERS Preprint typeset using LATEX style emulateapj v. 04/03/99 THE DISTANCE TO THE LARGE MAGELLANIC CLOUD FROM THE ECLIPSING BINARY HV2274 E.F. GUINAN, E.L. FITZPATRICK, L.E. DEWARF, F.P. MALONEY, P.A. MAURONE Department of Astronomy & Astrophysics, Villanova University, Villanova, PA 19085 Departament d'Astronomia i Meteorologia, Universitat de Barcelona, I. RIBAS E-08028 Barcelona, Spain J.D. PRITCHARD Mount John University Observatory and Department of Physics & Astronomy, University of Canterbury, Private Bag 4800, Christchurch, New Zealand D.H. BRADSTREET Department of Physical Science, Eastern College, St. Davids, PA 19087 A. GIMÉNEZ Laboratorio de Astrofisica Espacial y Fisica Fundamental, Villafranca, Spain To appear in Astrophysical Journal Letters ABSTRACT The distance to the Large Magellanic Cloud (LMC) is crucial for the calibration of the Cosmic Distance Scale. We derive a distance to the LMC based on an analysis of ground-based photometry and HST-based spec- troscopy and spectrophotometry of the LMC eclipsing binary system HV2274. Analysis of the optical light curve and HST/GHRS radial velocity curve provides the masses and radii of the binary components. Analy- sis of the HST/FOS UV/optical spectrophotometry provides the temperatures of the component stars and the interstellar extinction of the system. When combined, these data yield a distance to the binary system. After correcting for the location of HV2274 with respect to the center of the LMC, we find dLMC = 45.7 ± 1.6 kpc or (V0 - Mv)LMC = 18.30 ± 0.07 mag. This result, which is immune to the metallicity-induced zero point uncertainties that have plagued other techniques, lends strong support to the "short" LMC distance scale as derived from a number of independent methods. Subject headings: Binaries: Eclipsing - Stars: Distances - Stars: Fundamental Parameters - Stars: Individual (HV2274) - Galaxies: Magellanic Clouds - Cosmology: Distance Scale 1. INTRODUCTION We present the first accurate distance determination to the Large Magellanic Cloud (LMC) using an eclipsing binary sys- tem, resulting from an international collaboration studying the physical properties of these important objects. This program includes both ground-based and HST observations and is aimed primarily at determining the stellar properties (masses, radii, luminosities, and ages) and the distances of about a dozen se- lected +14 < V < +16 mag eclipsing binaries in the Clouds (see Guinan et al. 1996). It is well known that studies of eclipsing binaries yield the most fundamental determinations of the ba- sic stellar properties. We demonstrate here their great value as "standard candles" for distance determination. The distance to the LMC is a critical rung on the cosmic distance ladder and numerous independent methods (utilizing, for example, RR Lyrae stars, Cepheids, SN1987a, and "red clump" stars) have been employed to determine it. Unfortu- nately, and despite the use of the eagerly-awaited Hipparcos parallaxes, there remains considerable disagreement about this important measurement and, as summarized by Cole (1998; see also Westerlund 1997), the current uncertainty is about 10-15%. In this paper, we show that the analysis of several eclipsing LMC binaries could reduce the uncertainty in the LMC distance to level of several percent. These first results deal with the LMC eclipsing binary Har- vard Variable 2274 (hereafter HV2274; < V >max≃ 14.2; B1-2 IV-III + B1-2 IV-III; P = 5.73 days). This system is relatively bright, lightly reddened, and has complete CCD light curves showing it to have deep eclipses and uncomplicated out-of- eclipse light variations (Watson et al. 1992). Moreover, it is a detached system with nearly identical bright stars located in an uncrowded field. To provide an accurate distance to the system and, hence, the LMC, absolute radii, temperatures, and redden- ing corrections are needed (Guinan 1993; Giménez et al. 1994; Guinan et al. 1996). The radii are obtained from the light and radial velocity curves and the temperatures and reddening from UV/optical spectrophotometry. The following sections present the HST and ground-based observations used in this study (§2), describe our analysis techniques (§3), and present our distance determination for the LMC along with a discussion of the un- certainties in this result (§4). 2. OBSERVATIONS The ground-based V band light curve data for HV2274 are taken from Watson et al. (1992) and are shown in the lower panel of Figure 1 ("plus signs"). The light curve is well-defined and shows two minima of about equal depth (∼0.7 mag) and nearly constant light outside of the eclipses. The more shallow minimum (secondary eclipse) occurs at orbital phase 0.53, in- dicating an eccentric orbit (e = 0.136). We also adopt the very recently available V and B - V measurements from Udalski et al. (1998), i.e., V = 14.16 and B - V = - 0.13. The spectrophotometry and radial velocity curves needed to complete the analysis of HV2274 were obtained by HST as part of our larger project on eclipsing binaries in the Clouds. Four FOS spectra, covering four different wavelength regions, were acquired, calibrated, and merged to produce a single spectrum 1 2 Distance to the LMC spanning 1150 Å to 4820 Å. These observations were made at orbital phases outside of the eclipses, when both stars were completely unobscured. the radial velocity curve. We also obtained 16 HST/GHRS medium resolution (R = 23000) spectra at a number of selected orbital phases to construct The spectra covered 34 Å and were centered at two different wavelengths: one near λ1305 Å and the other near λ1335 Å. These regions contain strong UV photospheric lines in early B stars (see Massa 1989). Stellar Fe III λ1292.2, Si III λλ1300 triplets (1294.5, 1296.7 and 1298.9 Å), and C II λ1323.9 lines were fit with double gaus- sians, which gave the radial velocities for each component. The individual line measurements for each component were aver- aged to obtain its radial velocity at each phase. In the upper panel of Figure 1, these radial velocities are plotted against or- bital phase as determined by Watson et al. (1992; filled sym- bols). The mean uncertainties in the radial velocity measures are approximately ±15 km s- 1. 3. ANALYSIS 3.1. Modeling the light and radial velocity curves The radial velocity and light curve data were analyzed using an improved Wilson-Devinney light curve analysis code (Wil- son & Devinney 1971; Wilson 1990) that includes Kurucz at- mosphere models for the computation of the stellar radiative parameters (Milone et al. 1994). The Wilson-Devinney code is a standard tool in the analysis of eclipsing binaries. The radial velocity and the light curve solutions for HV2274 are shown in the top and bottom panels of Figure 1, respectively (solid curves). For the light curve, the differences between ob- served and computed values at V ("(O-C)V ") are also shown. Tables 1 lists the values determined for the orbital period (P), eccentricity (e), orbital inclination (i), longitude of periastron (ω), velocity semi-amplitudes (K), systemic velocity (γ), tem- perature ratio (TB/TA), luminosity ratio (LB/LA), absolute stellar radii (r), masses (M), and surface gravities (g). The accurate measurement of the radii (r) of the components is critically important for deriving the distance to HV2274. The radii are determined by combining the fractional radii (r f = r/a) obtained from the light curve analysis with the orbital semi- major axis (a) derived chiefly from the spectroscopic orbit. Thus, the absolute radius of each star is r = r f × a and the cal- culated uncertainties in the radii are about 2.3%. The precision achievable in the measurement of fundamental properties, such as radius, is one of the keys to the usefulness of eclipsing bina- ries as standard candles. 3.2. Modeling the spectrophotometry The observed energy distribution of HV2274, fλ⊕, as ob- tained by the FOS between 1150 Å and 4800 Å is shown in Figure 2 (filled circles; the lower of the two full spectra). The observed fluxes depend both on the surface fluxes of the bi- nary's components and on the attenuating effects of distance and interstellar extinction. This relationship can be expressed as: fλ⊕ = 1 d2 [r2 AF A λ + r2 BF B λ ] × 10- 0.4A(λ) or, substituting for the total extinction, A(λ), fλ⊕ = (cid:16) rA d (cid:17)2 [FA λ + (rB/rA)2F B λ ] (1) (2) × 10- 0.4E(B- V)[k(λ- V )+R] where Fi λ {i = A,B} are the emergent fluxes at the surfaces of the two components of the binary, the ri are the radii of the com- ponents, and d is the distance to the binary. A(λ) is the total extinction along the line of sight to the system at each wave- length λ and is expressed in Eq. 2 as a function of E(B - V ), the normalized extinction curve k(λ - V ) ≡ E(λ - V )/E(B - V), and the ratio of selective to total extinction in the V band R ≡ A(V )/E(B - V ). Our analysis consists of modeling the observed energy distribution of HV2274 via a non-linear least squares procedure to derive values of the parameters (rA/d)2, λ, E(B - V ), and k(λ - V ) in Eq. 2. Ultimately, the distance to Fi the binary will be determined from (rA/d)2. We represent the stellar surface fluxes Fi λ by ATLAS9 atmo- sphere models from R.L. Kurucz, which each depend on four parameters: effective temperature (Te f f ), surface gravity (g), metallicity (normally z), and microturbulence velocity (µ). The eight parameters needed to define the surface fluxes of the two binary components are constrained by the results of the light and radial velocity curve analyses and we adopt the effective temperature ratio TB/TA and log g values listed in Table 1. In ad- dition, we assume that both components of the binary have the same metallicity and the same microturbulence velocity. The ratio of the stellar radii (rB/rA)2 in Eq. 2 is determined from the light and radial velocity curve solutions (see Table 1), and we adopt the standard Galactic value of R = 3.1 (see §4). The normalized extinction curve k(λ - V ) is modeled using the results of Fitzpatrick & Massa (1990), who showed that the functional form of the extinction wavelength dependence in the UV is tightly constrained. The parameters describing the shape of the curve are determined from the fitting procedure. The method of smoothly joining UV and optical extinction curves is discussed by Fitzpatrick (1999). This technique of solving simultaneously for both the stellar parameters of a reddened star and the shape of the UV/optical extinction curve is described in detail by Fitzpatrick & Massa (1999; FM99). They demonstrate that the new ATLAS9 models can reproduce the observed UV/optical continua of unreddened main sequence (and slightly evolved) B stars to a level consis- tent with the uncertainties in currently available spectrophoto- metric data. They further show that both the model atmosphere parameters and the wavelength dependence of interstellar ex- tinction can be extracted from analysis of UV/optical spectra of reddened stars. This is possible because the spectral "signature" of interstellar extinction is very different from the "signatures" of temperature, surface gravity, metallicity, or microturbulence. The dereddened energy distribution of HV2274 is shown as the upper spectrum in Figure 2 (small filled circles with obser- vational error bars) superimposed with the best-fitting model (solid "histogram" curve). The inset shows a blowup of the Balmer jump region. The dereddened V and B data from Udal- ski et al. (1998), converted to flux units, are shown by the large filled circles. The excellence of this fit (the reduced χ2 is close to 1) is illustrative of the fits achieved by FM99 for B stars in general. The parameters of the fit are listed in Table 1. For B-type stars, the "metallicity" measured from modeling the UV/optical flux is most heavily influenced by absorption from iron-group elements in the UV region. Therefore in Table 1 we refer to the derived metallicity as [Fe/H], and the result is quite reasonable for a LMC star. The uncertainties indicated in Ta- ble 1 are 1-σ internal errors and incorporate the full covariance Guinan et al. 3 (or interdependence) of all the parameters. I.e, if any of the parameters is changed by ± 1-σ, the total χ2 of the best fit -- after all the other parameters are reoptimized -- increases by +1. The small size of these uncertainties is testament to the lack of covariance among the parameters and the quality of the data. The best-fit model found in this paper differs significantly from that reported in an earlier version of this work (Guinan et al. 1998). This results entirely from the inclusion here of the V and B - V measurements from Udalski et al. (1998). The ear- lier work incorporated very uncertain optical photometry from Watson et al. (1992), but these data were weighted so low that results were based essentially entirely on the FOS data. This had an important effect because, unless the observations extend through the 4400 -- 5500 Å region, a possible degeneracy ex- ists between the best-fit values of E(B - V ) and the normalized extinction curve k(λ - V ). I.e. very similar relative extinction corrections, E(B - V ) × k(λ - V ), can be obtained with a range of E(B - V) values. In Guinan et al. (1998) the degeneracy re- gion was found to be E(B - V) ≃ 0.08 - 0.12, with the best fit given by E(B - V ) ≃ 0.08 -- the solution reported in that pa- per. This result appeared reasonable since, combined with the Watson et al. photometry (B - V = - 0.18), it yielded an intrin- sic color of (B - V )0 ≃ - 0.26, compatible with the spectral type of HV2274. Using the more accurate (and thus more heavily weighted) Udalski et al. photometry, the reddening degeneracy disappears and a reasonable fit to the entire observed spectrum can only be achieved with E(B - V ) = 0.12 -- a value at the opposite end of the degeneracy "valley" from that reported in Guinan et al. (1998), as expected from Murphy's Law. 4. THE DISTANCE TO THE LMC The distance to HV2274 is computed from the values of (rA/d)2 and rA derived above, and the uncertainty in this re- sult is found by propagating the uncertainties in each quantity. The total uncertainty in (rA/d)2 has internal and external com- ponents. The internal 1-σ fitting error is listed in Table 1. The external effects include uncertainty in the the extinction ratio R, which we take to be 1-σ(R) = ±0.3, and uncertainty in the zero point of the FOS spectrophotometry, which we assume to be 1-σ(FOS) = ±2-3% (Bohlin 1996; Bless & Percival 1996). Quadratically combining these independent influences yields a total uncertainty of σ[(rA/d)2] ≃ 4.5%. The uncertainty in rA is taken from the Wilson-Devinney analysis and is listed in Ta- ble 1. From these results we derive dHV2274 = 46.8 ± 1.6 kpc, or, (V0 - Mv)HV2274 = 18.35 ± 0.07 mag. This result is 0.14 mag smaller than that reported by Guinan et al. (1998) due to the resolution of the aforementioned reddening degeneracy. To obtain the distance to the center of the LMC, the above re- sult must be corrected for the position of HV2274 in the LMC. The line-of-sight to the HV2274 binary system projects onto the end of the western region of the central bar of the LMC. The HST/GHRS spectra show strong ISM lines from LMC gas in O I λ1302.2 Å and Si II λ1304.4 Å. These lines show only one component with a Heliocentric radial velocity of 285 km s- 1, similar to the HV2274 systemic velocity of 312 km s- 1. These data suggest that HV2274 lies in or close to the disk of the LMC. To transform the distance of HV2274 to the LMC opti- cal center we assumed the position angle of the line of nodes is 168◦ and that the disk inclination is 38◦ (Schmidt-Kaler & Gochermann 1992). Simple trigonometry then implies that HV2274 is located on the far-side of the LMC, about 1100 pc behind the center. The distance of the center of the LMC is then dLMC = 45.7 ± 1.6 kpc or (V0 - Mv)LMC = 18.30 ± 0.07 mag. This result lends strong support to the "short" LMC distance scale (see Cole 1998). Its importance, however, lies not only in the value of the distance itself, but in the precision that appears possible from using eclipsing binaries. This precision can be understood by noting two facts: 1) the values of log g for the two binary members are very well determined by the light and radial velocity curve analysis; and 2) the best fitting model at- mosphere reproduces the HV2274 energy distribution over the entire observed range (1200 -- 5500 Å) to within the small ob- servational uncertainties (see Fig. 2). These imply that the stel- lar temperature must -- like the gravity -- be well-determined, since features such as the Balmer jump are sensitive to both properties. Further, a correct temperature could not be found unless the distorting effects of interstellar extinction were prop- erly estimated and removed, since the possibility of a degener- acy in the reddening solution has been eliminated by the inclu- sion of V band data in the analysis (§3.2) . Finally, we note that this technique is immune to the metallicity-induced zero point uncertainties that have plagued many other LMC distance esti- mates because the determination of the stellar metallicity is an explicit part of the analysis and because the derived distance is actually extremely insensitive to the metallicity. The result from a single binary system -- as precise as it ap- pears to be -- does not by itself resolve the LMC distance issue. Unanticipated external systematic effects, such as a peculiar lo- cation of the star within the LMC, could compromise the de- rived distance. Such effects can only be addressed through the analysis of more systems, preferably with a variety of locations within the LMC and covering a range of stellar properties. The ultimate precision of the LMC distance will best be evaluated by the range of results derived from such analyses. The results presented here, however, demonstrate the power of eclipsing bi- naries to address fundamental astrophysical issues, such as the derivation of basic stellar properties, and cosmologically im- portant issues, such as the LMC distance. This work was supported by NASA grants NAG5-7113 and HST GO-06683, and NSF/RUI AST-9315365. REFERENCES Bless, R.C., & Percival, J.W. 1997, in Proc. IAU Symp. 189, "Fundamental Stellar Properties", eds. Bedding, Booth, and Davis, in press Bohlin, R. 1996, AJ, 111, 1743 Cole, A.A., 1998, ApJ, 500, L137 Fitzpatrick, E.L. 1999, PASP, in press (astro-ph/9809387) Fitzpatrick, E.L., Massa, D., 1990, ApJS, 72, 163 Fitzpatrick, E.L., Massa, D., 1999, in preparation (FM99) Giménez, A., Clausen, J.V., Guinan, E.F., Maloney, F.P., Bradstreet, D.H., Storm, J., Tobin, W., 1994, Experimental Astronomy, 5, 181 Guinan, E.F., 1993, in "Pacific Rim Colloquium on New Frontiers in Binary Star Research", eds. K.C. Leung and I. Nha, A.S.P. Conf. Series, vol. 38, p. 1 Guinan, E.F., Bradstreet, D.H., DeWarf, L.E., 1996, in "The Origins, Evolutions, and Destinies of Binary Stars in Clusters", eds. E.F. Milone and J.-C. Mermilliod, A.S.P. Conf. Series, vol. 90, p. 197 Guinan, E.F., Ribas, I., Fitzpatrick, E.L., and Pritchard, J.D. 1998, in Ultraviolet Astrophysics Beyond the IUE Final Archive, eds. R. Gonzalez-Riestra, R., W. Wamsteker, and R. Harris (ESA SP-413), p. 315 Massa, D., 1989, A&A, 224, 131 Milone, E.F., Stagg, C.R., Kallrath, J., Kurucz, R.L., 1994, BAAS, 184, 0605 Schmidt-Kaler, T., Gochermann, J., 1992, in "Variable Stars and Galaxies", ed. B. Warner, A.S.P. Conf. Series, vol. 30, p. 203 Udalski, A., Pietrzynski, G., Wozniak, P., Szymanski, M., Kubiak, M., Zebrun, K. 1998 (astro-ph/9809346) 4 Distance to the LMC Watson, R.D., West, S.R.D., Tobin, W., Gilmore, A.C., 1992, MNRAS, 258, Westerlund, B.E., 1997, The Magellanic Clouds, (Cambridge University Press, 527 Cambridge) Wilson, R.E., Devinney, E.J., 1971, ApJ, 166, 605 Wilson, R.E., 1990, ApJ, 356, 613 TABLE 1 ORBITAL AND STELLAR PARAMETERS Orbital Solution and Stellar Propertiesa P = 5.726006(12) days i = 89◦.6(1.3) ω(1990.6) = 73◦.3(1.5) KA = 166.2(5.9) km s- 1 TB/TA = 1.005(5) rA = 9.84(24)R⊙ MA = 12.1(4)M⊙ loggA = 3.54(3) Stellar Properties from UV/Optical Spectrophotometryb e = 0.136(12) a = 38.58(93) R⊙ γ = +312(4) km s- 1 KB = 177.3(5.8) km s- 1 LB/LA(V ) = 0.844(5) rB = 9.03(20)R⊙ MB = 11.4(4)M⊙ log gB = 3.58(3) TA = 23000(180)K [Fe/H]AB = - 0.45(6) E(B - V) = 0.120(9) mag TB = 23110(180)K µAB = 1.9(7) km s- 1 (rA/d)2 = 2.249(63) × 10- 23 Distance to HV2274 d = 46.8(1.6) kpc V0- Mv =18.35(0.07)mag Distance to Center of LMC d = 45.7(1.6) kpc V0- Mv =18.30(0.07) mag aThe uncertainties in the parameters resulting from the light curve and radial velocity curve analyses were set at three times the r.m.s. scatter of several solutions obtained from different initial conditions. bThe uncertainties in the parameters derived from anal- ysis of the UV/optical spectrophotometry are 1-σ internal errors derived from a non-linear least squares fitting proce- dure (see §3.2). Guinan et al. 5 FIG. 1. -- The upper panel shows the radial velocity measurements of the two components of the HV2274 binary system as derived from medium resolution HST/GHRS spectra (filled symbols) and the radial velocity curve solution derived from the Wilson-Devinney analysis (solid curves). The discontinuous jumps seen in the model curves, known as the "Rossiter effect," occur when rotating stars are partially eclipsed. The lower panel shows the ground-based light curve data from Watson et al. 1992, based on differential photometry (Variable - Comparison) in the V band ("V-Cp"; plus signs). The fit to the light curve is shown (solid curve) as well as the Observed - Computed residuals ("O-C"). 6 Distance to the LMC FIG. 2. -- The UV/optical energy distribution of HV2274, in units of ergs cm- 2s- 1Å- 1. The lower full spectrum shows the observed HST/FOS energy distribution; the upper spectrum shows the extinction-corrected energy distribution, superimposed with the best-fitting ATLAS9 atmosphere model (plotted in histogram style). Vertical lines through the data points indicate the 1-σ observational errors. Crosses indicate data points excluded from the fit due to contamination by interstellar absorption lines. The large filled circles show dereddened B and V photometry from Udalski et al. 1998. An expanded view of the fit near the Balmer jump is shown within the inset.
astro-ph/9703086
1
9703
1997-03-12T16:43:38
Quantum theory of frequency shifts of an electromagnetic wave interacting with a plasma
[ "astro-ph", "gr-qc" ]
In the paper we calculate the frequency shift induced on a photon by the interaction with a low density electronic plasma. The technique is the standard perturbation theory of quantum electrodynamics, taking into account the many body character of the plasma. The shift in non relativistic approximation is shown to be blue. Besides the quantum shift also the known classical effects and the correct temperature dependence are obtained. Finally the limits of the approximations used are discussed.
astro-ph
astro-ph
I. INTRODUCTION The propagation of electromagnetic waves through plasmas has been extensively studied from a classical viewpoint. This is usually justified in most cases by the physical conditions regarding the plasma temperature and density at least when the considered wavelengths are not too short. In the classical approach the plasma operates as a sort of active filter both absorbing and distorting electromagnetic waves in various ways depending on the plasma parameters, its homogeneity, the presence of an external magnetic field and of course the frequency of the wave. To have an idea of this complicated methodology and of the kind of results one obtains it is useful to have a look to ref. [1]. It is remarkable that, according to the different conditions, various frequency shifts arise. Another classical approach, particularly fit for optical, or shorter, wavelengths is that which considers the interaction of the electromagnetic wave with a fluctuating medium as a scattering by inhomogeneities inside it. This analysis has been made in refs. [2], [3], [4] and leads again to frequency shifts of the incoming radiation depending now also on the scattering angle. In particular a blue shift is found for forward scattering, as a manifestation of the Rayleigh scattering. In principle however the quantum aspects of the interaction of the wave with the plasma should not be overlooked. This is likely to be especially true in some situations of astrophysical or even cosmological interest, where the density of the plasma is rather low and the distances are so high to allow also small effects to pile up. In this paper we shall precisely investigate the quantum effects using standard quantum electrodynamics and a perturbative treatment, as outlined in section II. The method will produce also the known classical features of the propagation of electromagnetic waves through a plasma, as we show in section IV. On the quantum side, we take into account in section III the many body nature of a plasma that entrains the appropriate fermionic statistic. In the non relativistic limit we find, for a low density locally homogeneous (i.e. homogeneous on the scale of a few wavelengths) plasma, a blue shift of the photon frequency (this has the same sign as that of the classical results in [2], [3], [4], though now the effect has a completely different origin). Finally in section V we discuss the validity conditions of our approximations, in particular those concerning the possibility of overlooking relativistic corrections. Just to fix ideas and exemplify our low density plasma, we refer, in various parts of the paper, to numerical values of the parameters of the order of those valid for the solar corona, i.e. a number density ∼ 106 electrons/cm3 and a temperature ∼ 106 K. II. BASIC ASSUMPTIONS AND OUTLINE OF THE METHOD We consider a situation where an electromagnetic wave, plane and monochromatic, coexists with a plasma of electrons with a numerical density distribution n(x). The propagation is along the x axis. The "unperturbed" state is obtained when the coupling between the wave and the plasma is set to zero; consequently the energy distribution of the electrons is that of a fermionic plasma of temperature T and density n restrained into a potential well and the wave has a frequency ω. Let us now switch on the coupling and, assuming the interaction energy to be small, we determine the shift in the energies as a perturbation of the "initial" (i.e. uncoupled) situation. Supposing that the interaction is set up in a finite time lapse and comparing the situations at −∞ in time with that at +∞, we mimic the actual process of an incoming plane wave of frequency ω and an outgoing one of frequency ω + δω. The technique actually used to compute the shift in the frequency of the wave is that of the time independent perturbations. The interaction Hamiltonian is: HI = − e cZ j · A dr (1) where A is the vector potential operator of the wave and j is the current density operator appropriate for this problem; the volume integral is limited to the confinement region of the plasma. The complete non relativistic expression for j when an electromagnetic interaction is present is [5]: 7 9 9 1 r a M 2 1 1 v 6 8 0 3 0 7 9 / h p - o r t s a : v i X r a Cast in the second quantization formalism, the interaction Hamiltonian is [6]: j = − e mc Φ†ΦA i¯h 2m(cid:0)Φ†∇Φ − Φ∇Φ†(cid:1) − HI = H ′ I + H ′′ I 1 (2) (3) H′I :=Xk gk q,q ′ a†qaq′ + h.c. H′′I := e2 2mc2 δk−k′,q−q′ √ωω′ b†k′bk ⊗ a†q′aq ′ ¯h bk ⊗Xq,q V Xk,k′,q,q ms ¯h3 e ′ (4) (5) (6) where we defined gk q,q ′ = q′ · ukδk,q′−q 2V ω The b†'s and the b's are the bosonic creation and annihilation operators associated to the photons; the a's are fermionic (electronic) operators. The energy of the photon is ¯hω, its momentum is ¯hk; of course ω = ck; the polarization of the photon is expressed by the unitary vector uk; k's and q's are respectively photonic and electronic wave vectors; primes denote intermediate state variables; finally, the unessential spin and polarization indices have been dropped, though an average over them in the final formulae is performed. The last term in (2) is usually omitted under the explicit, and some times implicit, assumption it gives contributions small with respect to those coming from the first. The frequency shift of the external photon can be written in the form δω = δω′ + δω′′ where δω′ and δω′′ are respectively the contributions due to H′I and H′′I . In this paper we calculate δω for a low density plasma, with a perturbative treatment up to order α = e2/¯hc (this is equivalent to keep the first order term in H′′I and the second in H′I ). III. THE QUANTUM FREQUENCY SHIFT We first consider the contributions to the shift due to H′I . Let Hem and HP be the electromagnetic field and the plasma free Hamiltonians; Hem and HP are respectively the bosonic Fock space associated to the photons and the fermionic Fock space associated to the electrons; Φk ∈ Hem and Ψq ∈ HP are the one particle electronic and photonic wave functions. We calculate the shift ∆Eω,q of the value of the energy of the state Φk ⊗ Ψq of Hem ⊗ HP . Since Φk ⊗ Ψq is an eigenvector of Hem + HP we can use the second order perturbation theory (the first order energy shift is zero). The energy shift of the state i is ∆Ei =Xj6=i (HI )ij (HI )ji Ei − Ej (1 − νj) (7) where j labels any (normalized) eigenvector of Hem + HP and νj is the probability that the state j is occupied by an electron. The only non vanishing contributions are those with respect to j states of the form j1 := Φ0 ⊗ Ψq′ Φk′ ⊗ Φk(cid:17) ⊗ Ψq′, (Φ0 is the electromagnetic vacuum). It is easy to show that Φk ⊗ Φk′(cid:17) ⊗ Ψq′ or j3 :=(cid:16) 1√2 (cid:16) 1√2 , j2 := (H′I )j1i = gk q,q′ So the energy shift induced by H′I on Φk ⊗ Ψq is [7] (H′I )j2i = (H′I )j3i = 1 √2 gk′ q′,q ∆E′ω,q = ∆E(1) ω,q + ∆E(2) ω,q (8) 2 where ∆E(1) ω,q :=Xq ′ (cid:16)1 − νq′(cid:17)" (cid:12)(cid:12)gk q,q′(cid:12)(cid:12)2 ¯hω + ǫq − ǫq′ − ¯hω + ǫq′ − ǫq# q′,q(cid:12)(cid:12)2 (cid:12)(cid:12)gk q′,q(cid:12)(cid:12)(cid:12) (cid:12)(cid:12)(cid:12)gk′ 2 ǫq − ǫq′ − ¯hω′ ∆E(2) ω,q :=Xk′ Xq′ (1 − νq′ ) where, for a Fermi gas, in low density approximation: νq = 1 z eβǫq + 1 ≃ ze−βǫq 1 In order to obtain the energy shift, we have to take the mean value of ∆E′ω,q with respect to the possible states of the electrons Ψq. We have ∆E′ω = ∆E(1) ω + ∆E(2) ω ∆E(1) ∆E(2) ω :=Xq ω :=Xq νq ∆E(1) ω,q νq ∆E(2) ω,q (12) (13) We shall consider the contribution to the shift ∆E(1) The term ∆E(2) ω in the next section. ω gives a divergent contribution [7]. In the following we shall show that, if we subtract the second order self energy of the electrons ∆E0 , the contribution to the resulting term is finite and independent from the normalization volume. This self energy is Thus, the observable contribution to the shift is: ∆E0 =Xq νqXq′k′ 2 q′,q(cid:12)(cid:12)(cid:12) (cid:12)(cid:12)(cid:12)gk′ ǫq − ǫq′ − ¯hck′ Using (10) and (14), ∆eE(2) ω = −Xk′ Xqq′ ∆eE(2) ω = ∆E(2) ω − ∆E0 νqνq′ 2 q′,q(cid:12)(cid:12)(cid:12) (cid:12)(cid:12)(cid:12)gk′ ǫq − ǫq′ − ¯hck′ (9) (10) (11) (14) (15) We now show that 1 Since in low density conditions νqνq′ ∼= z2 exp [−β (ǫq′ + ǫq)], we have, using the explicit expression (1) for the gk′ ω has a finite value, independent from the normalization volume . q′,q, Summing over q′ and using 1 ¯h3 2 1 V 2 Xk q q′ (q′ · uk)2 k exp (−β (ǫq′ + ǫq)) ¯hck + ǫq′ − ǫq δk+q′,q 1 V V ∆eE(2) ω = z2 e2 ∆eE(2) m2c V 2Pqk → 1 ω = z2 e2 ∆eE(2) m2c 1 V (2π)6R dkR dq, we get 2 (2π)6Z dqZ dk ¯h3 3 (q · uk)2 k exp(cid:2)−β(cid:0)ǫq−k + ǫq(cid:1)(cid:3) ¯hck + ǫq−k − ǫq (16) The integral over k can be computed in spherical coordinates (r, θ, ϕ) choosing q as polar axis; the polarization vector (normal to the polar axis) is defined by ϕ = 0. We have (µ = cos θ): ¯h2 ǫq−k ± ǫq = (q · uk)2 = q2 cos2 ϕ sin2 θ = q2 cos2 ϕ(cid:0)1 − µ2(cid:1) 2m(cid:0)q2 + k2 − 2qkµ ± q2(cid:1) dqq4Z 1 dkZ +∞ −1 0 dµ(cid:0)1 − µ2(cid:1) exp(cid:2)−(cid:0)2q2 + k2 − 2qkµ(cid:1) /k2 T(cid:3) 1 + ¯h mc(cid:0) k 2 − qµ(cid:1) (17) Calculating the integral (16) over ϕ, then integrating over all the possible directions of q: z2e2 m2c2 ¯h2 25π4Z +∞ 0 ω = 1 V ∆eE(2) with kT :=q 2mkB T ¯h2 . In order to calculate the integrals over q and k , we notice that the integrand is not exponentially zero only when the positive definite quantity 2q2 + k2 − 2qkµ is smaller then a few k2 T 's. This condition is satisfied only if k and q are small with respect to some kT 's. Then, if the plasma temperature is not higher than ∼ 106 K , we can neglect 2 − qµ(cid:1) in the denominator of (17) with respect to 1 (in fact ¯h mc(cid:0) k 2 − qµ(cid:1) = O(cid:0) ¯h ¯h mc(cid:0) k T ∼ 106 K,q kB T mc2 ∼ 10−2). Now the integrals are Gaussian, and they can be easily evaluated . This gives mc kT(cid:1) = O(cid:18)q kB T mc2(cid:19), and, if Finally in the low density assumption (z ≪ 1) the density n of the plasma is: z2e2 m2c2 ¯h2 29π3 k6 T 1 V ω = ∆eE(2) n = z · k3 T 1 23π 3 2 As a consequence, we have 1 V (18) (19) If the plasma is not homogenous, this energy shift depends on the position via the electronic plasma thermodynamic parameters. As we noticed in the introduction, the dependence on the position is negligible over distances of the order of the wavelength. Thus, the observed energy shift is simply the volume average (see section V below) of ∆eE(2) ω : ¯hδω(2) := dx n2 (x) (20) n2e2 m2c2 ¯h2 23 = n2 m2c ¯h3 23 α ω = ∆eE(2) ω (x) = dx ∆eE(2) 1 V ZV αZV ¯h3 = 23m2c IV. CLASSICAL CONTRIBUTIONS TO THE SHIFT We now consider the contribution to the shift due to ∆E(1) ω . In low density conditions we have, after summing with respect to q′ and taking the continuous limit, ∆E(1) ω = e2 m2 ¯h3 2ω z 1 (2π)3Z dqe−βǫq (q · uk)2(cid:20) 1 ¯hω + ǫq − ǫq+k − 1 ¯hω + ǫq−k − ǫq(cid:21) The integral can be calculated in the same approximation that leads to (19). We obtain 4 ∆E(1) ω = e2 m ¯h 2ck kBT mc2 n Once again, if the plasma is not homogeneous, ¯hδω(1) := 1 ¯hV ZV dx ∆E(1) ω (x) = e2 m 1 2ck kBT mc2 N V (21) (22) where N is the total number of electrons in V . Let us now determine the contribution coming from H Using again a perturbative treatment we see that the first order term is no longer zero and its magnitude in terms I of (3). ′′ of powers of the coupling parameter α is the same as that of (20), it cannot then be a priori overlooked. The correction we are now studying is: e2 ¯h (23) ¯hδω′′ = νqDΦω ⊗ Ψq(cid:12)(cid:12)(cid:12)bH′′I(cid:12)(cid:12)(cid:12) Φω ⊗ ΨqE ¯hδω′′ =Xq νq′*Φω ⊗ Ψq(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12) Xq,q′ ,k,k′ ωXq′ where bH′′I is given by eq. (5). Thus V Xq′ b†q′bq ⊗ a†k′ak(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12) In factPq′ vq′ is, by definition, the total number N of particles, consequently N V is nothing else than the average density hni of the plasma in the given volume. This gives δω′′ = e2 hni /(2mck), which is precisely the classical correction to the dispersion relation due to the presence of a low density homogeneous electron plasma (whose constant density is hni), when the temperature is zero. In fact, when it is ω2 Φω ⊗ Ψq+ = mc2 , we have (see [8]) c2k2 >> kB T δq−q′,k−k′ e2 2mc √kk′ νq′ = 2mc2 2mc2 N V (24) ¯h V ¯h k e2 1 = (25) P + c2k2(cid:18)1 + m n is the plasma frequency. For ωp ≪ kc and kB T ω2 = ω2 kBT mc2 ω2 P ω2(cid:19) where ωp =q e2 mc2 ≪ 1 ( non relativistic approximation) this gives 1 kBT ω = kc + 2(cid:18)1 + mc2(cid:19) ω2 This is of course true when T ∼ 106 K, which implies kB T The last term in (26) coincides with δω′′ + δω(1), where δω(1) gives the first order temperature correction. Like all the classical terms, it depends on N/V (V is the normalization volume of the electromagnetic field). Hence, when V → ∞, such a term gives a vanishing contribution to the frequency shift. = kc +(cid:18)1 + mc2(cid:19) e2n mc2 ∼ 10−4. 2mck p kc kBT (26) V. DISCUSSION The dependency of the classical terms on N/V rules them out when the measuring apparatus is outside of the plasma and the normalization volume for the electromagnetic wave is infinite. This is not the case for the quantum contribution (20), which is always different from zero: it will be the only contribution to the frequency shift observable in astrophysical or even cosmological conditions. An important remark on (20) is that the volume over which one integrates cannot be the volume of the whole universe; in fact it extends from the source to the receiver (∼ from −∞ to +∞ ) along the line of sight, but transversely it should not be more than the distance over which the plasma may practically be thought of as infinite for quantum mechanical calculations. This transversal extension is a sort of coherence length for the plasma. It should be less than the wavelength of plasmons, the Debye length and the 5 screening length of the plasma. On the other hand its square should of course be much greater than the Compton scattering cross section: our approach has indeed nothing to do with individual scattering phenomena. Assuming for simplicity a Gaussian distribution along the line of sight, such as n (x) = n0e− x 2R2 2 (20) gives: δω(2) = ¯h2 23m2c αn2 0π3/2RL2 (27) L is the transverse "coherence length": it may in fact be used as a phenomenological parameter. In our model relaxation phenomena in the plasma play no rbole. This is because we consider a steady state situation and, furthermore, our plasma is modelled as a reservoir, that can exchange with the photon any amount of energy without changing its thermodynamic state. Thus any dynamic process due to the transit of the photon is neglected. Of course, this approximation is good only if the plasma is very extended, its density is very low and the electromagnetic field is not too strong, but all of these conditions are indeed satisfied in our case. The result we found has been obtained under some assumptions that need a careful consideration. First of all, the calculation is based on the perturbation theory. Usually this theory treats single electrons interacting with an external field or with a bath of photons. Here we are dealing with one photon interacting with a gas of electrons. The fermionic nature of electrons comes into play through the factor νqνq′ in (15), or simply, when the approximation is appropriate, through the fugacity z of the plasma. When the fugacity is small enough, it is easy to see that the relevant terms of the perturbative series are weighed not only by α and its powers, but rather by products of powers of α and powers of z: this fact may change the relative importance of the different contributions. Actually (20) is of order αz2. If relativistic effects had been taken into account, a second order contribution had and can be come also from processes such as virtual pair creation. Such a contribution is proportional to αz(cid:0) ¯hω mc2(cid:1)2 overlooked when mc2(cid:19)2 (cid:18) ¯hω << z To sum up we conclude that the blue shift found in this paper applies to waves whose frequency satisfies to the condition ωp < ω << mc2 ¯h √z For higher frequencies relativistic contributions must be included. To give an idea of the numbers, consider that in the physical conditions of vast portions of the solar corona z can be as low as 10−17; consequently the upper frequency is ∼ 1011 Hz. [1] V.L.Ginzburg, Propagation of electromagnetic Waves in Plasma, Gordon and Breach Pub., New York (1961) [2] J. T. Foley, E. Wolf, Phys. Rev. A, 40, 579, (1989) [3] J. T. Foley, E. Wolf, Phys. Rev. A, 40, 588, (1989) [4] D. F. V. James, E. Wolf, Phys. Lett. A, 146, 167, (1990) [5] L. Landau, E. Lifshitz, M´ecanique Quantique, 510, MIR, Moscow (1967) [6] H. Haken, Quantum field theory of solids, North Holland, Amsterdam (1976), section 44b) [7] L. Accardi, A. Laio, F. Cardi, G. Rizzi, in Proc. 11th Italian conf. on General Relativity, Trieste, 1994 [8] Page 119 of ref. [1] 6
astro-ph/0503344
1
0503
2005-03-16T13:44:25
Abundance Analyses of Field RV Tauri Stars, VI: An Extended Sample
[ "astro-ph" ]
An abundance analysis is presented and discussed for a sample of 14 RV Tauri stars. The present abundance data and those from our previous papers and by other workers are combined in an attempt to further understanding of the dust-gas separation process which afflicts many RV Tauri variables. We propose that a star's intrinsic (i.e., initial) metallicity is given by the photospheric zinc abundance. Variables warmer that about 5000 K and with an initial metallicity [Fe/H] $\geq$ $-$1 are affected by dust-gas separation. Variables of all metallicities and cooler than about $T_{\rm eff} \simeq 5000$ K are unaffected by dust-gas separation. The RV Tauri variables show a spread in their C abundances with the lower boundary of the points in the C versus Zn plane falling close to the predicted trend for giants after the first dredge-up. The upper boundary is inhabited by a few stars that are carbon-rich. The O abundances in the mean follow the predicted trend from unevolved stars in line with the expectation that photospheric O abundance is unaffected by the first dredge-up. An evolutionary scenario involving mass loss by a first ascent or early-AGB red giant, the primary star of a binary, is sketched.
astro-ph
astro-ph
Abundance Analyses of Field RV Tauri Stars, VI: An Extended Sample Sunetra Giridhar Indian Institute of Astrophysics; Bangalore, 560034 India [email protected] David L. Lambert The W.J. McDonald Observatory; University of Texas; Austin, TX 78712-1083 [email protected] Bacham E. Reddy1 Indian Institute of Astrophysics; Bangalore, 560034 India [email protected] Guillermo Gonzalez Department of Physics and Astronomy; Iowa State University; Ames, IA 50011-3160 [email protected] David Yong Department of Astronomy; University of Texas; Austin, TX 78712-1083 [email protected] ABSTRACT An abundance analysis is presented and discussed for a sample of 14 RV Tauri stars. The present abundance data and those from our previous papers and by other workers are combined in an attempt to further understanding of the dust-gas separation process which afflicts many RV Tauri variables. We propose that a star's intrinsic (i.e., initial) metallicity is given by the photospheric zinc abundance. Variables warmer that about 5000 K and with an initial metallicity [Fe/H] ≥ −1 are affected by dust-gas separation. Variables of all metallicities and cooler than about Teff ≃ 5000 K are unaffected by dust-gas separation. The RV Tauri variables show a spread in their C abundances with the lower boundary of the points in the C versus Zn plane falling close to the predicted trend for giants after the first dredge-up. The upper boundary is inhabited by a few stars that -- 2 -- are carbon-rich. The O abundances in the mean follow the predicted trend from unevolved stars in line with the expectation that photospheric O abundance is unaffected by the first dredge-up. An evolutionary scenario involving mass loss by a first ascent or early-AGB red giant, the primary star of a binary, is sketched. Subject headings: stars:abundances -- stars:AGB and post-AGB -- stars: vari- ables (RV Tauri) 1. Introduction In this series of papers, we have been exploring the chemical compositions of RV Tauri variables in the Galactic field. This exploration concludes here with abundance analyses reported for fourteen variables. Of these, twelve are analysed for the first time, one (LR Sco) was previously analysed by us before its status as a RV Tauri variable was appreciated by Lloyd Evans (1999), and one (DY Ori) is analysed more thoroughly than in our earlier attempt. Beginning with the analysis of the southern RV Tauri variable IW Car (Giridhar, Rao, & Lambert 1994, Paper I), we have shown that the atmospheric composition of a RV Tauri star may be abnormal in the sense that the photospheric abundance anomalies are roughly correlated with the predicted condensation temperature for low pressure gas of solar com- position. In particular, elements (e.g., Al, Ca, Ti, and Sc) with the highest condensation temperatures (∼ 1600 K) may be seriously underabundant relative to their abundance ex- pected from the abundances of elements (e.g., S and Zn) of low condensation temperature (Gonzalez, Lambert & Giridhar 1997a, Paper II; Gonzalez, Lambert & Giridhar 1997b, Pa- per III; Giridhar, Lambert & Gonzalez 1998, Paper IV; Giridhar, Lambert & Gonzalez 2000, Paper V). Our findings were confirmed by independent analyses of several RV Tauri stars by Van Winckel et al. (1998), Maas, Van Winckel, & Waelkens (2002), and Maas (2003). These abundance anomalies imply that the RV Tauri's photosphere is deficient in those elements which condense most readily into dust grains. We refer to the principal operation necessary to achieve the deficiencies as dust-gas separation. If such a separation is to affect the photospheric composition, three conditions must be met. First (condition A), a site for dust formation must exist near the star. Two proposed locations compete to meet this condition : the wind off the star or a circumstellar or circumbinary disk. Second (condition B), a mechanism must be identified to separate 1Visiting Observer, Cerro Tololo Inter-American Observatory, which is operated by the Association of Universities for Research in Astronomy, Inc. under contract with the US National Science foundation. -- 3 -- dust from gas. It has been supposed that radiation pressure on the dust grains drives them through the gas and away from the star(s). Third (condition C), the dust-free gas accreted by the star must become the dominant constituent of the star's photosphere. Here the principal issue is that the photosphere is a part of a convective envelope. In order for the photosphere to assume a composition dominated by dust-free gas, the mass of the envelope must be small relative to the mass of accreted gas. Our earlier analyses have shown that the severity of the atmospheric abundance anoma- lies differs from one RV Tauri to another. These differences presumably are clues to the circumstances under which the above conditions are or are not met. Taking them in inverse order, the following may be noted. Relevant to condition C, the coolest RV Tauri variables independent of their metallicity are free of the abundance anomalies. One interpretation is that these stars have deep convective envelopes which dilute accreted gas and prevent the appearance of abundance anomalies even when gas but not dust is accreted (Paper V). Stars of intrinsically low metallicity such as the variables in globular clusters (Gonzalez and Lambert 1997; Russell 1998) and some high-velocity variables in the field appear immune to the effects of a dust-gas separation. This fact is relevant to condition B: it is likely a consequence of the inability of radiation pressure on dust grains to force a separation of dust from gas when the mass fraction of dust is very low, as it is in a metal-poor cool environment. Accretion under these circumstances will not change the surface composition of the star even if it has a shallow envelope. With respect to condition A, the extremely metal-deficient A-type post-AGB stars are evidence that dust-gas separation does not have to occur in a stellar wind (Van Winckel 2003). The prototypical example is HR 4049 (Teff = 7600 K) with an extensive infrared excess from dust but its stellar wind, if it exists, is surely too hot to be the site of dust formation. Yet, dust-gas separation has provided a photospheric abundance of [Fe/H] ≃ −4.8 for a star with an initial abundance [Fe/H] ≈ −0.4. Van Winckel, Waelkens, & Waters (1995) showed that HR 4049 and other similar metal-deficient stars were spectroscopic binaries. The dust-gas separation is presumed to occur in a circumbinary dusty disk. Superposition of the pulsational velocity variation on an orbital variation complicates a demonstration that all RV Tauri stars affected by dust-gas separation are binaries. Certainly, several RV Tauri stars are known to be spectroscopic binaries. An assessment of the direct and indirect evidence (Van Winckel et al. 1998) led Van Winckel (2003) to write that 'binarity may very well be a common phenomenon among RV Tauri stars'. The question of dust-gas separation occurring in a wind off a cool star remains open. An aim of this paper was to enlarge the sample of RV Tauri variables in order to test in more depth previous deductions about the dust-gas separation processes. We present -- 4 -- abundances for the fourteen variables listed in Table 1. This new sample particularly increases the representation of high-velocity RV Tauri stars. 2. Observations and Abundance Analyses The program stars and dates of observation are listed in Table 1 along with the measured radial velocity, the pulsational period, the Preston spectroscopic type A, B, or C (Preston et al. 1963), and the photometric type a or b. The photometric type b indicates that the light curve shows a long-term modulation. A majority of the stars were observed with the McDonald Observatory's 2.7m Harlan J. Smith reflector with the CCD-equipped '2dcoud´e' spectrograph (Tull et al. 1995). A spectral resolving power R = λ/∆λ ≃ 60000 was used and a broad spectral range was covered in a single exposure. A S/N ratio of 80-100 over much of the spectral range was achieved. Figure 1 illustrates the quality of typical spectra. The stars LR Sco and AZ Sgr were observed at CTIO, Chile with the 4m Blanco telescope equipped with a cross-dispersed echelle Cassegrain spectrograph and a CCD of 2048 × 2048 pixels. The spectrograph was set to record the wavelength interval 4900 A to 8250 A in 45 orders. Spectral coverage was complete between these limits. The resolving power R ≃ 35,000 was achieved, as measured from the Th lines in the Th-Ar comparison spectrum. (Emission was almost always present at Hα.) Spectra were rejected if they showed line doubling, markedly asymmetric lines, or strong emission at Hβ. It is presumed that the spectra not showing these characteristics represent the atmosphere at a time when standard theoretical models may be applicable. This presumption should be tested by analysis of a series of spectra taken over the pulsational cycle. This remains to be done but in previous papers we have analysed several stars using spectra taken at different phases and obtained consistent results. A striking example was given in Paper III where three observations of SS Gem gave widely different effective temperatures (4750, 5500, and 6500 K) but similar results for the composition. The abundance analyses were performed as described in earlier papers of this series. The 2002 version of the spectrum synthesis code MOOG (Sneden 1973) was used with ATLAS model atmospheres ( Kurucz 1993). Molecule formation was taken into account in computing the line spectrum. Hyperfine splitting was considered for the lines of relevant atoms and lines ( e.g. Sc, Mn). Atmospheric parameters were determined in the usual way from the -- 5 -- Fe i and Fe ii lines by demanding excitation and ionization equilibrium, and that the iron abundance be independent of the equivalent width. Ionization equilibrium is also satisfied for Si, Ti, and Cr: the mean abundance difference (in dex) between neutral and ionized lines is +0.08 ± 0.16 for Si from 11 stars, −0.02 ± 0.16 for Ti from 7 stars, and 0.00 ± 0.11 for Cr from 13 stars. The adopted parameters listed in Table 2 were used for the full abundance analysis. Abundances are referred to the solar abundances given by Lodders (2003, her Table 1). Results are given in Table 3 for the stars obviously carrying the signatures of dust-gas separation and in Table 4 for the other stars. 3. The Chemical Compositions The chemical composition of a RV Tauri variable may be a blend of several signatures. (i) initial composition of the star, (ii) the effects of deep mixing during stellar evolution on the composition, and (iii) the effects of the dust-gas separation, and (iv) if the RV Tauri stars are the primary of a spectroscopic binary, a change of composition may have resulted from mass transferred from the companion. The initial composition may be anticipated from abundances of elements thought to be essentially unaffected in the course of evolution as single or binary star and also by dust-gas separation. Here, S and Zn are deemed to qualify as such elements. Published abundance analyses of main sequence stars show that a star's initial composition is generally predictable to within a small uncertainty from a determination of the abundance of one element -- see Goswami & Prantzos (2000 -- their Figure 7) for a graphical summary. We adopt S and Zn as the reference elements here, from which we predict initial abundances of other elements to high accuracy. Although the prehistory of RV Tauri variables is unknown in detail, it may be assumed that they have experienced the first dredge-up which brings CN-cycled material into the atmosphere. This reduces the C abundance and increases the N abundance (Iben 1967). If RV Tauri variables have evolved from the AGB on which they may have experienced thermal pulses and the third dredge-up, they are expected to be enriched in C and possibly also in the s-process elements (Busso, Gallino and Wasserburg 1999). This last statement will require modification if a star has accreted substantial amounts of gas from a companion directly or through transfer from a circumbinary disk. If the RV Tauri stars are the primary of a binary, a change of composition may have resulted from mass transferred from the companion. When gas cools and dust grains form, the abundances of the elements in the gas phase are reduced below their initial abundances. Calculations of the equilibrium distribution of -- 6 -- an element between dust and gas have been reported by many authors and, most recently, by Lodders (2003; her Table 1). Principal factors influencing the gas phase abundances are the temperature, pressure, and initial composition. For elements providing major species of grains (e.g., Al and corundum -- Al2O3), one may define a condensation temperature (here, T′ C) at which the vapor pressure of the grain equals the partial pressure of the species in the gas. Below this condensation temperature, a species is highly underabundant in the gas. Other elements, particularly trace elements, are absorbed by major species of grains with a degree of absorption which may differ from element to element for a given grain type. This has led to the concept of a '50% condensation temperature' -- the temperature at which 50% of the trace element is in the gas and 50% in grains. We adopt the 50% condensation temperatures given by Lodders (2003) in her Table 8 for gas of solar composition (Her Table 1) at a total pressure of 10−4 bar. We denote the adopted temperatures by the symbol TC. For elements providing the major species of grains, the difference between T ′ C and the cooler TC is less than 40 K for all but two elements - Si and Ca - which we discuss below (Section 4.1). Our assumption is that the effects of dust-gas separation on a stellar composition will be revealed as a correlation between the underabundance of an element and that element's condensation temperature. Obvious approximations may invalidate the assumption: grain formation around the RV Tauri stars may not occur under conditions of thermodynamic equi- librium; the pressures may differ from one formation site to another; the initial composition may differ, as must certainly be the case for the high velocity stars, from the assumed solar mix. Yet, the fact is that, in many cases, the underabundances are remarkably smoothly correlated with TC (and T ′ C). In the following subsections, we present and discuss the compositions of the stars in Table 1 but leave comment on the C, N, and O abundances and heavy (Y to Eu) elements to subsequent sections. Six stars have a composition greatly affected by dust-gas separation (or another process): UY CMa, EQ Cas, HP Lyr, DY Ori, LR Sco, and BZ Sct. For the remainder of the sample, the signature of dust-gas separation is less distinct and may be absent. 3.1. UY Canis Majoris Our abundance analysis (Table 3) shows clearly that UY CMa is a victim of severe dust-gas separation. This is not unexpected because Preston et al. (1963) assigned it their spectroscopic class B, and our earlier analyses of Class B stars found them affected by dust- -- 7 -- gas separation. Elements of the highest TC (e.g., Sc and Ti) are underabundant by a factor of about 200. The deficiencies [X/H] are very well correlated with condensation temperature TC (Figure 2).2 Judged by the abundances of Na, S, and Zn, elements of low condensation temperature, the initial metallicity corresponds to [Fe/H] ≃ −0.4. A possibly slightly higher metallicity is suggested by the O abundance with the implication that even Na, S, and Zn may be slightly depleted. The high radial velocity suggests that UY CMa belongs to the thick disk. We have not adjusted the [X/H] by the small amounts necessary to account for the fact that a thick disk star of [Fe/H] ≃ −0.4 has a non-solar mix of elements, i.e., [X/Y] 6= 0 (Reddy et al. 2003). 3.2. RX Capricorni Inspection of the abundances listed in Table 4 suggests that RX Cap is either a mod- erately metal-poor thick disk star or a rather metal-rich halo star, but possessing a normal composition unaffected by dust-gas separation. The inferred initial iron abundance is [Fe/H] ≃ -0.7 from consideration of S, Zn, Fe and Ni abundances. Within the possible errors and especially considering that some abundances are based on a single line, a majority of the relative abundances [X/Fe] are as expected. We note, for example, the positive [X/Fe] for the α-elements Mg, Si, S, and Ti, and the mild deficiency of Mn. One detects a suggestion of dust-gas separation from the low Sc abundance: [Sc/Fe] = −0.4 when [Sc/Fe] ≃ 0.0 is expected, and just possibly, from Ca which is slightly underabundant relative to expectation for an α-element. 3.3. EQ Cassiopeiae Three spectra of this variable were available. One was rejected because some lines showed distinct doubling and He i 5875 A was in emission. The spectrum from 1999, August 19 was crowded with lines, but unblended lines found with the help of the Arcturus spectrum are symmetric. The atmosphere at this time was quite cool (Teff = 4500 K). A useful spectrum from 2001, August 19 caught the star at a warmer phase (Teff = 5300 K). Although Hα was in emission, unblended lines were symmetric and deemed suitable for an abundance analysis. A radial velocity of −158 ± 1 km s−1 was measured from the 1999 August and 2001 August spectra confirming that the star is of high velocity. The two spectra provide similar results for the elemental abundances. Iron abundances from the two spectra are given in Table 2. 2Usual spectroscopic notation is adopted: [X/Y] = log(X/Y)star -log(X/Y)⊙. -- 8 -- In Table 3, we list the abundances from the 2001 spectrum. Differences in [X/Fe] between the 2001 and 1999 spectra are in the range −0.20 to +0.14 with differences of less than ±0.1 for nine of the 15 elements common to the two analyses. There are what appear to be signatures of the dust-gas separation (Figure 3) but the [X/H] versus TC relation is not as striking and simple as that for UY CMa. Preston et al.'s (1963) assignment of the class B? may reflect a hint of the unusual composition differences between the star and RV B stars like UY CMa. The abundances [S/H] ≃ [Zn/H] ≃ −0.4 point to an initial [Fe/H] of −0.5 to −0.4. The gross underabundances of Ca and Sc ([Ca/H] = −2.0 and [Sc/H] = −3.1) point to dust-gas separation. Yet, among the RV Tauri variables analysed by us in this series of papers, EQ Cas is unique in showing additional anomalies. In particular, the Na abundance is remarkably low: [Na/Fe] = -0.9 where all other RV Tauri variables, even those severely affected by dust-gas separation, show a positive [Na/Fe]. Furthermore, our estimate for [Na/Fe] may be an upper limit because the Na i lines are very weak. The other anomalies are the low values of [Cr/H], [Mn/H], and [Cu/H] and the high value of [Ti/H] relative to other elements of similar TC. These outstanding anomalies are shown by the two independent analyses of the star except that the Cu i lines were not observed in the 1999 spectrum. Rao & Reddy (2005) show that EQ Cas abundances are tightly correlated with an element's first ionisation potential (FIP). This suggests that a mechanism other than dust-gas separation has affected the photospheric abundances. 3.4. DF Cygni Two spectra were acquired but only one had symmetric unblended lines suitable for an abundance analysis (Table 4). Line selection was made by reference to the spectrum of Arcturus, but given the crowded spectrum, some key elements (e.g., Al and S) proved undetectable, and many others are represented by no more than one or two lines. The iron abundance is close to solar. Setting aside elements represented by just one line, the sole apparent anomaly is Sc ([Sc/H] = [Sc/Fe] = -1.0 not 0.0). The Sc underabundance is suggestive of the onset of dust-gas separation but, as for other stars in Table 4, Sc (and Ca, in some stars) is the only indicator of this effect. This isolation of Sc leads one to wonder about other possible explanations, e.g., non-LTE effects. -- 9 -- 3.5. HP Lyr This star came to our attention through Graczyk et al.'s (2002) report of it as 'possibly the hottest RV Tau type object'. These authors estimated the effective temperature to be about 7700 K. Our spectrum corresponds to Teff = 6300 K. The radial velocity of -107 km s−1 indicates that HP Lyr is a high-velocity star. The intrinsic metallicity of the star as assessed from S, Zn, and Na is slightly sub-solar ([Fe/H] ≃ −0.2). We suppose that HP Lyr may belong to the thick disk. The abundances (Table 3 and Figure 4) show the signature of dust-gas separation. The correlation between [X/H] and TC is tight with almost no scatter in excess of that attributable to the observational errors in [X/H]. 3.6. TX Ophiuchi Most elements (Table 4) have the abundance expected of a normal star with [Fe/H] = −1.2, the measured iron abundance. This metallicity and the high radial velocity suggest that the star belongs to the Galactic thick disk. One notes the normal [Na/Fe] and [Zn/Fe] but observes that Ca, Sc, and Ti, elements of high condensation temperature, are apparently underabundant with [Ca/Fe], [Sc/Fe], and [Ti/Fe] by about 0.5, 0.6, and 0.3 dex, respectively, below the expected initial value for a thick disk star of [Fe/H] = −1.2. Aluminum, also of high condensation temperature, however, has its expected initial abundance. A comparison with RX Cap reveals that the abundances scale with the difference in [Fe/H] of 0.4 dex between the two stars. The mean difference in [X/H] in the sense TX Oph minus RX Cap is 0.5 dex from 20 elements with an element-to-element scatter consistent with the measurement errors. This consistency extends to Ca, Ti, and Sc. 3.7. UZ Ophiuchi Two spectra are suitable for an abundance analysis. The spectrum from the 2.7m telescope taken on 2002, June 30 is our primary source. An earlier spectrum from 2001, July 13 taken on the 2.1m with the Sandiford Cassegrain echelle (McCarthy et al. 1995) provided coverage of the interval 4500 A to 5200 A. Since most of the elements were covered in 2002 spectrum and abundances derived at two epochs are in good agreement, we chose to use only 2002 estimates. The radial velocities of −95.1 ± 0.9 km s−1 in 2001 July and −90.2 ± 0.8 km s−1 in 2002 June suggest membership of the Galactic thick disk for UZ Oph. Abundances are given in Table 4. -- 10 -- The composition resembles that of RX Cap, a star of similar metallicity and which we described as essentially possessing its initial composition. Relative to RX Cap, the mean abundance difference between UZ Oph and RX Cap is ∆[X/H] = −0.15 from 13 elements from Na to Zn with no element having a ∆ outside the errors of measurement. The six heavy elements Y, La, Ba, Ce, Nd, Sm and Eu do provide a significantly different result: ∆ = −0.63: UZ Oph is relatively underabundant in Y to Sm but not in Eu relative to RX Cap and the expected initial composition. 3.8. DY Orionis DY Orionis was analysed in Paper II. Our reanalysis uses a new spectrum providing greater wavelength coverage which enabled us to cover more elements relative to the earlier study. Our new results are in good agreement with the earlier results. The star, as noted in Paper II, is severely affected by dust-gas separation (Table 3 and Figure 5). Particularly striking is the sharp onset of depletion for elements with TC above 1100 K: Na is barely depleted with [Na/Fe] of −0.3 but Si, Mn and Cr have [X/H] ≃ −1.9. 3.9. AI Scorpii The abundance analysis (Table 4) implies that AI Sco's composition is that expected of a star with the measured [Fe/H] = -0.7. The apparent anomalies are a mild overabundance of Na and S and an underabundance of Ca, Sc, and Ti. Relative to RX Cap, Na to Zn are slightly overabundant: ∆ = +0.13 from 13 elements. The heavy elements -- Y to Nd -- are underabundant: ∆ = -0.13 from five elements. Europium has a normal abundance. These small differences between AI Sco and RX Cap could be attributed to the inevitable observational errors. 3.10. LR Scorpii This star was analysed previously (Giridhar, Rao, & Lambert 1992) and discussed as a victim of dust-gas separation in Paper V. Our new analysis is summarized in Table 3. Maas (2003) has presented an abundance analysis based on spectra obtained in 2000 and 2001 when the spectroscopically estimated parameters (Teff, log g) were (6250, 0.5) and (5250, 0.0), respectively. His abundances agree well with ours, especially for [X/Fe]. Elements with TC . 1400 K suggest an initial iron abundance [Fe/H] = −0.2 which -- 11 -- with the low radial velocity indicates that LR Sco is a member of the Galactic thin disk. Our abundance analysis (Table 3 and Figure 6) shows that elements with a condensation temperature higher than 1600 K are clearly underabundant for which [X/H] ≃ −0.2 for all elements is expected. Our abundances [X/H] are generally in good agreement with the 1992 results obtained from a 1989 spectrum but for Al, Sc, Ti, and Y our results differ from the 1992 values by −0.4, −0.4, −0.7, and −1.5 dex, respectively. We note that the quartet are the elements with TC of about 1600 K. May one ascribe the differences between our present and previous results to a growth of the depletion of the highest TC elements abundance? According to Lodders & Fegley, the initial condensates of the five elements are Al2O3, CaAl12O19, Sc2O3, CaTiO3, and Y2O3. This mix precludes a substantial depletion of Ca. Calcium with TC = 1505 K gave the same result ([Ca/H] = −0.5) from the 1989 and 2002 spectra. Maas (2003) abundances may be consistent with the suspected lowering of the abundances of Al, Sc, Ti and Y between 1989 and 2002, but a uniform analysis of all available spectra is desirable. Observations of LR Sco should be continued. 3.11. AR Sagittarii The abundance analysis is summarized in Table 4. This high-velocity star shows no convincing evidence for a dust-gas separation. The initial metallicity is identified as [Fe/H] ≃ −1.2. Relative to TX Oph, a star of the same measured [Fe/H], the mean ∆ is 0.0 from 14 elements from Na to Zn, and −0.12 from seven elements from Ba to Eu. In short, AR Sgr and TX Oph are of essentially identical normal composition. 3.12. AZ Sagittarii The star was in a cool phase at the time of observation and the spectrum is crowded with lines. The abundance analysis (Table 4) is based on few lines. The available spectrum did not go shortward of about 4900A and, therefore, the blue and useful Zn i lines were not recorded. Our spectrum gives the radial velocity as −110 km s−1. This velocity identifies AZ Sgr as a high-velocity star. Judged with respect to a normal star of the measured [Fe/H] = −1.6, AZ Sgr's anomalies are a high S abundance and a low Cr abundance: [S/H] is too high by about 0.8 dex and [Cr/H] is too low by about 0.8 dex. The S abundance is based on weak lines and is sensitive to the adopted effective temperature. The Cr abundance appears to be securely based as it -- 12 -- is derived using five neutral Cr lines of moderate strengths. Relative to TX Oph and AR Sgr, stars with a slightly higher measured [Fe/H], the S overabundance is less marked, but the Cr underabundance remains distinctive. Pending acquisition and analysis of additional spectra of AZ Sqr, preferably at a warmer phase, we adopt the conclusion that this star is a metal-poor high-velocity star unaffected by dust-gas separation. The Cr underabundance is unexplained. 3.13. BZ Scuti This star's composition (Table 3 and Figure 7) shows evidence of dust-gas separation. Sodium, S, and Zn abundances show that BZ Sct's initial metallicity was close to solar. These abundances are based on unblended lines in the spectrum of this warm RV Tauri star. Elements with a condensation temperature higher than about 1200 K are depleted by a factor of about eight with an element-to-element scatter comparable to the errors of measurement. There is a hint that elements with TC greater than 1600 K are more severely depleted. The extreme depletion of La is based on a single line. 3.14. V Vulpeculae This star is clearly C-rich. The carbon enrichment derived from C i lines is also signaled by the presence of C2 Swan bands. Apart from carbon and possibly sulfur, the composition of V Vul resembles closely that of the other mildly Fe-poor stars in Table 4. The mean ∆, for example, relative to RX Cap is +0.40 for 11 elements from Na to Zn, and +0.27 for 4 elements from Y to Eu, where the [Fe/H] difference is +0.4 also. 4. The compositions of RV Tauri stars 4.1. Abundance Anomalies and Condensation Temperatures In recent papers on RV Tauri compositions, abundance versus condensation temperature plots have been constructed with values of TC taken from Lodders & Fegley (1998) which are also 50% condensation temperatures. Here, we adopt TC from Lodders (2003) (see Section 3). The effect of a substitution of the 2003 for the 1998 estimates of TC is in most cases a reduction in the scatter of [X/H] about the mean trend with TC for those stars obviously affected by the dust-gas separation. In our present sample, this is certainly the case for UY -- 13 -- CMa, HP Lyr, DY Ori, and LR Sco. The elements for which replacement of the 1998 by the 2003 TC estimates reduces their status as outliers in the [X/H] versus TC plot are notably Si, and Ba. In the case of Si, TC is lowered to 1302 K from 1529 K in 1998. The TC for Ba is raised from 1162 K in 1998 to 1447 K in 2003. For other elements measured by us the change in TC from the 1998 and 2003 estimates differs by less than about 50 K, except for Ca and Cu for which the 2003 values are lower by 130 -- 140 K, too small of a change to affect the appearance of a [X/H] versus TC correlation. It is useful to look behind the TC estimates at the highest temperature condensates expected in equilibrium for solar composition gas. For this exercise, Lodders (2003) provides the details. Our discussion considers the development of condensates as gas is cooled in equilibrium. As long as equilibrium is maintained, the discussion is readily reworded to consider the dissolution of condensates as the gas-dust mixture is heated. Inspection of Figures 2, 4, 6, and 7 show that the four most depleted elements are Al, Sc, Ti, and Y with very similar TC (≃ 1640K). The initial condensates are Al2O3 and CaTiO3 leading at a slightly lower temperature to the major condensates CaAl12O19 (hibonite) and CaTi3O10, Ca3Ti2O7, and Ca4Ti3O10 (three forms of calcium titanate). Calcium titanate may severely deplete gas of Ti but not Ca; the initial Ca/Ti ratio is about 20. Similarly, if hibonite scours Al from the gas, little Ca is removed and the initial ratio Ca/Al ≃ 1 is increased to Ca/Al ≃ 10. Sc and Y are depleted as Sc2O3 and Y2O3 dissolve in hibonite. Calcium and Ba are lost to grains at a slightly lower TC. The minor loss of Ca from the gas to hibonite is enhanced by the condensation of Ca2Al2SiO7 (gehlenite) leading to Ca/Al ∼ 1 in the gas with both Ca and Al depleted for TC ≃ 1500 K. Barium is removed from the gas when BaTiO3 dissolves in calcium titanate. In the 1998 calculations, Ba was identified as dissolving in CaTiO3 (perovskite), but in the 2003 calculations it was considered to dissolve into forms of calcium titanate. It is interesting that although the mode of transfer of Ca and Ba to condensates differs their depletion [X/H] in the gas is generally quite similar for stars severely affected by dust-gas separation. (Rare earths -- La to Eu in our analyses -- dissolve in hibonite and titanate.) Descending the TC scale, we encounter Mg and Si. Magnesium is removed from the gas primarily through Mg2SiO4 (forsterite) with MgAl2O4 (spinel) also contributing. If this were the sole way to remove Mg and Si from the gas, the Si/Mg ratio in the gas would swing from Si/Mg ≃ 1 to ≃ 2. The TC for Si listed by Lodders & Fegley (1998) is the temperature at which gehlenite (Ca2Al2SiO7) forms but, because Ca and Al are at least an order of magnitude less abundant than Si, this condensate removes very little silicon from the gas. At a slightly lower temperature, MgSiO3 (enstatite) condenses and then one expects Si/Mg ∼ 1 and similar depletions of Mg and Si. The TC taken from Lodders (2003) refers -- 14 -- to condensation of enstatite. 4.2. Depletion patterns The stars in this and earlier papers of our series exhibiting dust-gas separation seem to form a family with a single outcast. The main characteristic of the family is an ordering of depletions [X/H] by the condensation temperature TC. In our present sample, the extremes are marked by LR Sco and the pair HP Lyr and DY Ori. LR Sco would be deemed to have a normal composition but for depletions ([X/H] ∼ −1) of Al, Sc, Ti, and Y, the elements of the highest TC. At the other extreme, all elements with TC ≥ 1200 K show a depletion increasing with TC and attaining a level [X/H] ∼ −3 at the highest TC. This is not a single parameter family: see, for example, the differences in [X/H] between the extremes HP Lyr and DY Ori. BZ Sct appears not to show the family's smooth run of [X/H] with TC, but the several outliers (Cu, K, Sm, and La) in Figure 7 are based on a single line. If this is deemed a possible reason for the scatter, one is left with just Mn as showing a slightly discrepant [X/H]. R Scuti was considered quite anomalous by Luck (1981) but, in Paper V, we showed that the underabundant elements, including the heavy elements analysed by Luck, were among those with the highest TC. With due allowance for errors in the abundance determinations, differences in initial abundances as a function of metallicity and population class (i.e., thick versus thin disk), and small familial differences, all analysed stars but one may be said to belong to the same family bounded by LR Sco and HP Lyr-DY Ori3. The sole non-member of the family is EQ Cas where the upper envelope to the run of [X/H] with TC resembles the pattern shown by HP Lyr, but several elements fall well below this envelope: notably, Na, Mn, and Cr. Rao & Reddy (2005) show that the [X/H] for EQ Cas are well correlated an elements's FIP (see below). 4.3. Interstellar Depletions The abundances for stars affected by dust-gas separation correlate quite well with those for the cool interstellar gas towards ζ Oph (Savage & Sembach 1996) (Figure 8). Depletions 3Here, we note a good correspondence between R Sct and LR Sco. It would be useful to extend the analysis of LR Sco to more heavy elements: Paper V gave [X/H] for nine elements between Y and Eu but Table 3 includes just two. -- 15 -- of elements in interstellar gas are attributed to incorporation of the depleted element into and onto grains. The grains may have formed primarily in denser regions of interstellar clouds, but, as the detection and analysis of grains in meteorites demonstrates, some grains were formed in outflows of red giants, stellar ejecta: supernovae and novae shells for example. Grain formation and depletion of atoms in the gas in the interstellar medium cannot be expected to mimic very closely the convolution of grain formation, depletion of gas in the dusty reservoir, separation of gas from dust, accretion of the gas, and mixing of the gas with the stellar atmosphere. Despite these caveats, the close correspondence between the affected RV Tauri stars and the interstellar gas is suggestive of a similarity between the physical processes involved. The differences between the stellar and interstellar depletions may be clues to the physical processes at work. 4.4. The Initial Fe Abundances On the assumption that S and Zn are unaffected by dust-gas depletion, their abundances may be taken to be a star's initial abundances, and, hence, used to provide estimates of the initial abundance of Fe and other elements. Sulfur and Zn are the sampled elements with the lowest TC which are considered to be unaffected by internal nucleosynthesis and mixing. The run of S and Zn abundances with Fe/H is known for unevolved stars (Mishenina et al. 2002; Nissen et al. 2004; Prochaska et al. 2000; Ryde & Lambert 2004). We are here interested in the abundances of thin and thick disk stars for which differences in [S/Fe] and [Zn/Fe] at a given [Fe/H] are small, say, about 0.1 dex with the larger values found for the thick disk stars. In successive figures we show [Zn/Fe] and [S/Fe] versus [Fe/H] (Figure 9 & 10), and [S/Zn] versus [Zn/H] (Figure 11). Stars with a radial velocity of greater than 100 km s−1 are termed 'high velocity' and shown by filled symbols in the figures. In Figure 9, we show [Zn/Fe] versus [Fe/H] for all of the RV Tauri variables considered in this and previous papers in this series. Results for additional RV Tauri variables are taken from Maas et al. (2002) for RU Cen and SX Cen, and Maas (2003) for IRAS 09538- 7622, 16230-3410, 17038-4815, and 17233-4330. We have added a point for QY Sge (Rao, Goswami, & Lambert 2002) which, on circumstantial evidence, was described as a dust- obscured RV Tauri variable. We elected not to add other post-AGB stars discussed by Maas (2003). Although, these omissions are spectroscopic binaries and some show effects of dust- gas separation, available photometry does not show RV Tauri-like light variations. The run of [Zn/Fe] for unevolved stars corresponds to [Zn/Fe] ≃ 0 for the range of [Fe/H] covered in Figure 9. Thick disk stars may have [Zn/Fe] ≃ +0.1. The limit [Zn/Fe] ≃ 0.0 represents the lower boundary to the distribution of the observed -- 16 -- points. (The star with a distinctly negative value of [Zn/Fe] is DF Cyg with a crowded spectrum and an uncertain Zn abundance.) Stars along the lower boundary are those with an abundance pattern indicating little or no dust-gas separation. Several but not all high- velocity stars are near the boundary. The upper boundary to the data points follows the dashed line. This corresponds to approximately [Zn/Fe] = −[Fe/H] i.e., the trajectory of a star with an initial abundance [Fe/H] = 0 and then depleted in Fe to varying degrees but with Zn undepleted. On the assumptions that Fe but not Zn is depleted and that initially [Zn/Fe] = 0, the observed Zn abundance provides the star's initial Fe abundance [Fe/H]0. By inspection, we see from Figure 9 that the [Fe/H]0 runs from about +0.1 to −1.4. This is shown by horizontal dotted line corresponding to [Zn/Fe] ∼ 0. It is shown by the figure that the dust-gas separation is not an either-or effect. The run of [S/Fe] versus [Fe/H] (Figure 10) resembles that in Figure 9 but with differ- ences concerning the upper and lower boundaries. The lower boundary to the distribution of points runs systematically about 0.2 dex above the run of [S/Fe] with [Fe/H] for unevolved stars taken from Nissen et al. (2004). This offset is possibly attributable to systematic errors in one or both analyses. The dashed line corresponds to [S/Fe] = [Fe/H]. Several stars fall above this line. On the assumption that the S abundance is unaffected by dust-gas separation, the upper boundary in Figure 10 implies that the initial iron abundance for some stars is about [Fe/H]0 = +0.4, which seems implausible. A possible alternative explanation is that the S abundances for the RV Tauri stars are overestimated by about 0.2 to 0.4 dex. These differences between Figures 9 and 10 necessarily make an appearance in Figure 11 which shows that the [S/Zn] of the RV Tauri stars fall generally above the expected trend for unevolved stars taken from Nissen et al. In summary, we suggest that the initial [Fe/H] of the RV Tauri and related stars is obtainable from their measured Zn abundance and the assumption that [Zn/Fe] = 0 for normal stars. 4.5. Carbon and Oxygen Abundances Main sequence stars in evolving to red giants develop a deep convective envelope which brings CN-cycled products to the surface and, thus, reduces the surface C abundance and increases the surface N abundance (Iben 1967). Oxygen is expected to be very little affected. The reduction in the C abundance is 0.2 to 0.3 dex (Lambert & Ries 1981). After completion of He-core burning, material from the He-burning shell is dredged to the surface to increase the surface C abundance and also the O abundance. A star at this stage is known as an asymptotic giant branch (AGB) star. The He-burning shell may experience ignition of a -- 17 -- neutron source and synthesis of heavy elements by the the neutron-capture s-process. By examining the C, N, O, and heavy element (e.g., Y and Ba) abundances of the RV Tauri stars and taking due note of the initial abundances and the possible depletion of the surface abundances of these elements, one hopes to specify the evolutionary state of the RV Tauri variables. In the following discussion, [Zn/H] replaces [Fe/H] as the representative of a star's initial composition. The initial C and O abundances as a function of Zn abundance are taken from measurements of unevolved stars (Nissen et al. 2004; Akerman et al. 2004). We adopt the LTE abundances given by Akerman et al. (2004). Their tabulated NLTE abundances for oxygen are about 0.1 to 0.2 dex smaller than the LTE values. Non-LTE abundances were not given for carbon. We assume that the first dredge-up experienced by a red giant reduced the initial C abundance by 0.25 dex, but left the initial O abundance undisturbed. The observed C abundances (Figure 12) lie above the predictions for a red giant by about 0.4 dex with a real star-to-star scatter. Carbon enrichment is suggested. However a similar displacement of observed from initial abundances occurs for [S/Zn]. Since both the C and the S abundances are based on high-excitation atomic lines, the displacements may arise from similar systematic errors. The observed O abundances (Figure 12) follow quite closely the trend of the initial abundances with [Zn/H]. The observed C/O abundance ratios versus [Zn/H] are shown in Figure 12. The lower boundary to the observed ratios follows the predicted trend for first dredge-up red giants. These may be stars which suffered a reduction of surface C in becoming a red giant but did not later experience replenishment of C. Stars with the highest C/O ratio at a fixed [Zn/H] fall about 0.5 dex above the predicted trend. This spread probably represents a real star-to-star spread in C/O. Very few stars are C-rich, i.e., log C/O ≥ 0. In summary, the RV Tauri variables have suffered differing degrees of surface enrichment of carbon. Some stars, especially stars of high-velocity, have the C/O ratio anticipated for fresh red giants, i.e., no obvious C enrichment occurred following the first dredge-up. Other stars have a C/O ratio indicating up to a factor of 10 enrichment over the ratio of a fresh giant. Few stars are carbon rich. There is no clear distinction in Figure 12 between the stars greatly affected by dust-gas separation and those unaffected. 4.6. The s-process Abundances Mature AGB stars are carbon-rich (C/O > 1) with enrichments of s-process products. Cool carbon stars on the AGB are s-process enriched by up to one dex (Abia et al. 2002). -- 18 -- Post-AGB carbon-rich stars are similarly enriched (Reddy et al. 2002; Van Winckel 2003). In the case of the RV Tauri stars unaffected by dust-gas separation, s-process abundances are directly obtainable from the measured abundances with respect to the iron abundance. No obvious enrichments are detected. For stars affected by dust-gas separation, the degree of s-process enrichment must be judged relative to the abundances of other elements of a similarly high TC. Inspection of Figures 2 through 7 shows that there is no detectable overabundance of s-process elements. This is true for all other RV Tauri variables from our series of papers and those analysed by Maas and colleagues. Our earlier suggestion that Ba may be overabundant in AR Pup and DS Aqr deserves reexamination. Barium, the only s-process element examined in AR Pup, was represented by a single strong line (Paper II). A similar situation applies to DS Aqr from Paper IV. New spectra of broad wavelength coverage should be obtained and analysed before DS Aqr and AR Pup are tagged as 's-process enriched'. In summary, the RV Tauri variables are unlikely to be descendants of luminous AGB stars in which s-process products from the He-burning shell have been dredged to the stellar surface. 4.7. Boundary Conditions for Dust-gas Separation Dust-gas separation is not ubiquitous among RV Tauri stars. Inspection of the stars unaffected by the separation shows that their temperature and metallicity domains are both well bounded . These bounds are surely clues to the site of the dust-gas separation and the process by which a star of anomalous surface composition is created. Our sample of RV Tauri variables show a low and high temperature boundary to the stars of anomalous composition. Stars of all metallicities with Teff ≈ 5000 K and cooler are unaffected by dust-gas separation.4 This is shown by Figure 13. Our Teff estimates are spectroscopic determinations from spectra at phases showing well defined symmetric absorption lines. The spectroscopic Teff may differ with phases (at which spectroscopic analysis is pursued) for a given star; for example our three spectra of SS Gem gave Teff estimates of 4750K, 5500K and 6500K (Paper III). Effects of dust-gas separation are seen to the high temperature end of the RV Tauri sample. This boundary is presumably set by the blue edge of the instability strip. Depletion is seen in hotter stars (Van Winckel 2003), 4Among the stars with Teff ≤ 5000 K, DY Aql appears to defy the above condition. Our analysis (Paper III) of DY Aql was based on very few lines; evidence of depletion rests almost exclusively on a lone Sc ii line. -- 19 -- notably the extremely metal-poor star HR 4049 with Teff = 7600 K and with an observed iron abundance [Fe/H] ≃ −4.8. The oxygen abundance suggests an initial [Fe/H] near -0.5 for HR 4049. In paper V, we suggested that the coolest stars were unaffected because the convec- tive envelope was of sufficient mass to dilute accreted gas even were it cleansed of dust. Frankowski (2003) discusses the mass of the convective envelope (MCE) of post-AGB stars by drawing on published numerical calculations. It is clear that MCE increases with decreas- ing effective temperature: for example, MCE at 4000 K, 5000 K, and 6000 K is 0.016M⊙, 0.002M⊙, and 0.001M⊙, respectively. The MCE at 6000 K is some 1000 times the mass of the observable photosphere. Abundance deficiencies for the high TC elements for stars on the hot side of the boundary may approach a thousandfold, and, in such cases, the surface and convective envelope must be composed of almost undiluted accreted gas. Although the increase of MCE to lower temperatures may play a role in suppressing the abundance anoma- lies, it hardly seems to offer the full explanation. Perhaps, cooler variables have stronger winds which impede accretion of gas from a circumbinary disk. The dust-gas separation appears suppressed among stars of a low initial metallicity. Our sample of RV Tauri stars suggests that the limiting initial metallicity is [Fe/H] ≈ −1. More metal-rich stars are affected, but less metal-rich stars do not show the abundance anomalies resulting from dust-gas separation, even if they are hotter than the above Teff boundary. For stars with [Fe/H] & −1, there is no strong evidence that the dust-gas separation efficiency is sensitive to the initial metallicity. In several examples, an affected RV Tauri star currently has a metallicity well below the limit [Fe/H] of −1 for occurrence of a dust-gas separation . This is especially true for hotter stars like HR 4049. Our inference from these observations is that it is the initial and not the current metallicity that is the key factor in establishing dust-gas separation. Hence, the dust-gas reservoir is a region with a metallicity in excess of [Fe/H] of -1. In a reservoir of low metallicity separation of dust and gas is inhibited and then if material is accreted from the reservoir by the star, the surface composition is unchanged. There is a third boundary condition that may exist: effective dust-gas separation and accretion of gas but not dust by a star occurs efficiently only in a binary system. This is spectacularly the case for the small sample of hot post-AGB stars like HR 4049 for which radial velocity monitoring has shown them to be single-lined spectroscopic binaries (Van Winckel et al. 1995). For the RV Tauri variables, detection of orbital variations in the face of pulsational variations is a complexity calling for long campaigns of radial velocity monitoring. Nonetheless, the number of RV Tauri variables known to be spectroscopic binaries is growing as can be seen from Mass (2003). Van Winckel et al. (1998) have adduced indirect evidence -- 20 -- in support of the idea that RV Tauri variables are binaries with a circumbinary disk. Finally, there is the caution suggested by Rao & Reddy (2005) from their consideration of the compositions of EQ Cas and CE Vir that showed that dust-gas separation may not always be the dominant factor behind abundance anomalies. We have mentioned the special abundance anomalies of EQ Cas in Section 3.3. We remarked upon the lack of the usual correlation between [X/H] and TC. Rao & Reddy find that our [X/H] for EQ Cas are anti- correlated with first ionisation potential, FIP of the elements. These authors suggest that the stellar wind preferentially picks up ions over neutral atoms. At the top of the stellar atmosphere, low FIP elements will be present as the ions X+ but atoms of the high FIP elements will remain neutral. The FIP-effect is a minor contributor to the stars for which anomalies exhibit a strong TC correlation (Paper IV). 5. Concluding Remarks Conversion of the boundary conditions into an understanding of how RV Tauri stars with and without abundance anomalies arise remains elusive. Two contrasting scenarios were sketched in Paper V: the stars with anomalous abundances are (1) single stars with dust-gas separation occurring in the stellar wind (the S hypothesis), and/or (2) binary stars with dust-gas separation occurring in a circumbinary disk (the B hypothesis). The boundary condition on Teff is met by both hypotheses, if the deeper convective envelope for cooler stars acts to dilute the accreted gas. The boundary condition on initial metallicity seems more readily met by the B hypothesis; the circumbinary disk is comprised of gas ejected by either the RV Tauri's companion and/or the RV Tauri star before the onset of accretion and alteration of the photospheric abundances. Indeed, mass loss induced by a companion may provide not only the circumbinary gaseous disk but also shifts the primary star off the giant branch and into the instability strip. This boundary condition is not easily satisfied for all stars by the S hypothesis; some RV Tauri stars have metal abundances less than the boundary [Fe/H] ≃ −1. The B hypothesis provides a natural link between the RV Tauri stars and the A-type extremely metal-deficient stars like HR 4049 which are known to be spectroscopic binaries. Binary RV Tauri stars are known. A determined campaign of radial velocity measurements for RV Tauri stars, especially those displaying anomalous abundances, is needed to establish the frequency of binaries. It should be noted too that the sample of HR 4049-like stars is presently very small. A search for additional examples and a demonstration that they are (or are not!) spectroscopic binaries would also be valuable. -- 21 -- The B hypothesis requires gas to be accreted from the circumbinary disk by the star. This accretion, one presumes, competes with the wind from a RV Tauri star. Direct detec- tion of the infalling gas should be sought. On the S hypothesis, it would seem that there must be a circulation of gas up to the dust-gas separation sites and back down to the photo- sphere. Spectroscopic evidence for this circulation may be obtainable from high-resolution spectroscopy over the pulsational period of a star with anomalous abundances. Finally, we note that the sparsely populated region of the HR diagram that is home to RV Tauri variables and other luminous post-AGB stars deserves continued exploration. Van Winckel's (2003) review gives a thorough description and discussion of the compositions of post-AGB stars of which few have 'normal' abundances. Our study of RV Tauri stars began with a chance observation of IW Car. Perhaps, continued but systematic spectroscopic exploration of post-AGB stars will uncover objects that will reveal vital clues to the operation of dust-gas separation in the RV Tauri and other stars. Galactic post-AGB stars offer the best opportunity for detailed study of the stellar photosphere and the circumstellar environment. Although considerably fainter, post-AGB stars in the Magellanic Clouds offer some advantages, e.g., their luminosities are obtainable with fair precision. We thank the referee for a most thorough and constructive review of the paper. This research has been supported in part by the Robert A. Welch Foundation of Houston, Texas. REFERENCES Abia, C.; Domnguez, I., Gallino, R., Busso, M., Masera, S., Straniero, O., de Laverny, P., Plez, B., Isern, J. 2002, ApJ, 579, 817 Akerman, C.J., Carigi, L., Nissen, P.E., Pettini, M., & Asplund, M. 2004, A&A, 414, 913 Busso, M., Gallino, R., Wasserburg, G. J. 1999, ARA&A, 37, 239 Frankowski, A. 2003, A&A, 406, 265 Giridhar, S., Lambert, D. L., & Gonzalez, G. 1998, ApJ, 509, 366 (Paper IV) Giridhar, S., Lambert, D. L., & Gonzalez, G. 2000, ApJ, 531, 521 (Paper V) Giridhar, S., Rao, N. K., & Lambert, D. L. 1992, JA&A, 13, 307 Giridhar, S., Rao, N. K., & Lambert, D. L. 1994, ApJ, 437, 476 (Paper I) Gonzalez, G, Lambert, D. L., & Giridhar, S. 1997a, ApJ, 479, 427 (Paper II) Gonzalez, G, Lambert, D. L., & Giridhar, S. 1997b, ApJ, 481, 452 (Paper III) Gonzalez, G, & Lambert, D. L., 1997, AJ, 114, 341 -- 22 -- Goswami, A., & Prantzos, N. 2000, A&A, 359, 191 Graczyk, D., Mikolajewski, M., Leedjarv, L., Fr¸ackowiak, S.M., Osiwala, J.P., Puss, A., & Tomov, T. 2002, Act Astr, 52, 293 Iben, I. Jr. 1967, ARA&A, 5, 571 Kholopov, P. N., Samus, N. N., Durlevich, O. V., Kazarovets, E. V., Kireeva, N. N., & Tsvetkova, T. M. 1985, General Catalogue of Variable Stars, 4th edn. (Nauka Pub- lishing House: Moscow) Kukarkin, B. V., Parenago, P. P., Yu, N., & Kholopov, P. N. 1958. General Catalogue of Variable Stars, 2nd edn. (Academy of Sciences of USSR: Moscow) Kurucz, R.L. 1993 ATLAS9 Stellar Atmosphere Program and 2km/s grid CD-ROM vol 13 (Cambridge: Smithsonian Astrophys. Obs) Lambert, D.L., & Ries, L.M. 1981, ApJ, 248, 228 Lloyd Evans, T. 1999, in Asymptotic Giant Branch Stars, ed. T. Le Bertre, A. L`ebre, & C. Waelkens, IAU Symp. 191, 453 Lodders, K. 2003, ApJ, 591, 1220 Lodders, K., & Fegley, B. 1998, in The Planetary Scientist's Companion, (Oxford University Press: New York) Luck, R. E. 1981, PASP, 93, 211 Maas, T. 2003, Dissertation, Univ. of Leuven, Leuven Maas, T., Van Winckel, H, & Waelkens, C. 2002, A&A, 386, 504 McCarthy, J.K., Sandiford, B.A., Boyd, D., & Booth, J. 1995, PASP, 105, 881 Mishenina, T.V., Kovtyukh, V.V., Soubiran, C., Travaglio, C. &, Busso. M. 2002, A&A, 396. 189 Nissen, P.E., Chen, Y.Q., Asplund, M., & Pettini, M. 2004. A&A, 415, 993 Preston, G. W., Krzeminski, W., Smak, J., & Williams, J. A. 1963, ApJ,137, 401 Prochaska, J.X., Naumov, S.O., Carney, B.W., McWilliam, A., & Wolfe, A.M. 2000, AJ, 120, 2513 Rao, N.K., Reddy, B.E. 2005, MNRAS (in press) Rao, N.K., Goswami, A., & Lambert, D.L. 2002, MNRAS, 334, 129 Reddy, B.E., Lambert, D.L., Gonzalez, G., & Yong, D. 2002, ApJ, 564, 482 Reddy, B.E., Tomkin, J., Lambert, D.L. & Allende Prieto, C. 2003, MNRAS, 340, 304 -- 23 -- Russell,S.C. 1998, PASA, 15, 189 Ryde, N., & Lambert, D.L. 2004, A&A, 415, 559 Savage, B.D., & Sembach, K.R. 1996, ARA&A, 34, 279 Sneden,C. 1973, Ph.D. Thesis, Univ. of Texas at Austin, Austin Tull, R. G., MacQueen, P. J., Sneden, C., & Lambert, D. L. 1995, PASP, 107, 251 Van Winckel, H. 2003, ARA&A, 41, 391 Van Winckel, H., Waelkens, C., & Waters L. B. F. M. 1995, A&A, 293, L25 Van Winckel, H., Waelkens, C., Waters, L.B.F.M., Molster, F.J., Udry, S., & Bakker, E.J. 1998, A&A, 336, L17 This preprint was prepared with the AAS LATEX macros v5.2. -- 24 -- Table 1. The Program Stars Star Date Rad. Vel. Perioda Spec. groupb Phot. type.a km s−1 (days) UY CMa RX Cap EQ Cas DF Cyg HP Lyr TX Oph UZ Oph DY Ori AI Sco LR Sco AR Sgr AZ Sgr BZ Sct V Vul 1999 Nov 1 1999 Nov 2 1999 Aug 19 2001 Aug 19 1999 Nov 1 2002 Nov 14 1999 Aug 19 2001 Jul 13 2002 Jun 30 2002 Nov 14 2002 Jun 30 2002 Jun 22 1999 Oct 31 2002 Jun 19 1999 Jun 6 1999 Aug 19 +128 -123 -158 -158 -14 -107 -166 -95 -90 -8 -35 -18 -112 -110 +89 -28 114.6 67.9 58.3 58.3 49.8 69.3c 135.0 87.4 87.4 83.4 71.0 104.4 87.9 113.6 75.7 81.1 B A B? B? A · · · A A A B A · · · A? · · · · · · A aData GCVS Kukarkin et al. 1958: Kholopov et al. 1985 bdata from Preston et al. 1963 cData from Graczyk et al. 2002 a · · · a a · · · · · · · · · a a a b · · · a · · · · · · a -- 25 -- Table 2. Stellar Parameters Derived from the Fe-line Analyses Star UT Date Teff , log g, [Fe/H] (km s−1) log ǫ Modela ξb t Fe Ic UY CMa RX Cap EQ Cas DF Cyg HP Lyr TX Oph UZ Oph UZ Oph DY Ori AI Sco LR Sco BZ Sct AR Sgr AZ Sgr V Vul 1999 Nov 1 1999 Nov 02 1999 Aug 19 2001 Aug 19 1999 Nov 01 2002 Nov 14 1999 Aug 19 2001 Jul 13 2002 Jun 30 2002 Nov 14 2002 Jun 30 2002 Jun 22 1999 Jun 6 1999 Oct 31 2002 Jun 19 1999 Aug 19 5500, 0.0, −1.4 5800, 1.0, −0.8 4500, 0.0 , −0.8 5300, 0.7 , −0.8 4800, 1.7, −0.0 6300, 1.0, −1.0 5000, 0.5, −1.2 5000, 0.5, −0.7 4800, 0.0, −0.8 6000, 1.5, −2.0 5300, 0.25, −0.7 6000, 0.50, −0.2 6250, 1.0, −0.8 5300, 0.0, −1.5 4750, 0.5, −1.6 4500, 0.0, −0.4 5.0 4.8 4.6 3.6 3.6 3.2 4.9 3.6 3.6 4.0 3.3 4.2 3.2 5.2 4.2 4.6 6.15 ± 0.17 6.69 ± 0.16 6.68 ± 0.13 6.73 ± 0.15 7.50 ± 0.17 6.48 ± 0.12 6.24 ± 0.17 6.78 ± 0.14 6.70 ± 0.15 5.23 ± 0.14 6.76 ± 0.14 7.29 ± 0.16 6.65 ± 0.12 6.16 ± 0.13 4.90 ± 0.15 7.10 ± 0.11 Fe IIc log ǫ 6.20 ± 0.14 6.67 ± 0.13 6.71 ± 0.12 6.71 ± 0.14 7.36 ± 0.18 6.52 ± 0.10 6.28 ± 0.16 6.80 ± 0.10 6.63 ± 0.14 5.27 ± 0.12 6.82 ± 0.10 7.24 ± 0.11 6.64 ± 0.13 6.09 ± 0.14 4.93 ± 0.13 7.05 ± 0.11 n 11 16 13 9 7 15 17 6 14 12 8 7 16 17 10 10 n 65 80 54 53 22 99 81 68 98 30 38 50 63 62 45 52 aTeff in K, log g in cgs, [Fe/H] in dex. bξt is the microturbulence determined from the Fe I lines clog ǫ is the mean abundance relative to H (with log ǫH = 12.00). The standard deviations of the means, as calculated from the line-to-line scatter, are given. n is the number of considered lines. -- 26 -- Table 3.1. Elemental Abundances UY Cma EQ Cas HP Lyr Species log ǫo ⊙ [X/H] n [X/Fe] [X/H] n [X/Fe] [X/H] n [X/Fe] C I N I O I Na I Mg I Al I Si I Si II S I K I Ca I Sc II Ti I Ti II V I Cr I Cr II Mn I Fe Co I Ni I Cu I Zn I Y II Ba II La II Ce II Pr II Nd II Sm II Eu II 8.39 7.83 8.69 6.30 7.55 6.46 7.54 7.54 7.19 5.11 6.34 3.07 4.92 4.92 4.00 5.65 5.65 5.50 7.47 4.91 6.22 4.26 4.63 2.20 2.18 1.18 1.61 0.78 1.46 0.95 0.52 −0.03 ± 0.19 −0.18 ± 0.09 −0.44 −1.35 ± 0.12 −1.84 −0.97 ± 0.13 −0.95 −0.32 ± 0.12 −1.65 ± 0.22 −2.22 ± 0.15 −2.38 ± 0.13 −1.88 ± 0.11 −1.72 ± 0.09 −1.12 ± 0.18 −1.29 −1.57 ± 0.10 −0.75 ± 0.20 −0.59 ± 0.06 −2.32 ± 0.13 −2.02 ± 0.15 −1.31 5 ... 3 1 4 1 4 1 5 ... 4 5 ... 5 ... 7 4 5 ... 6 2 3 4 3 ... ... ... ... ... 1 +1.26 −0.23 ± 0.19 +0.04 ± 0.04 −1.65 ± 0.13 −0.52 ± 0.19 −0.26 ± 0.10 −0.29 ± 0.23 −0.35 ± 0.15 −2.02 ± 0.11 −3.06 −1.32 ± 0.17 −1.19 ± 0.09 −2.05 ± 0.03 −1.90 ± 0.10 −1.78 ± 0.09 −0.75 −0.59 ± 0.06 −0.79 ± 0.17 −1.04 −0.33 ± 0.17 −1.94 ± 0.14 +1.11 +0.85 −0.06 −0.55 +0.32 +0.34 +0.97 −0.36 −0.93 −1.09 −0.59 −0.43 +0.17 −0.28 +0.54 +0.70 −1.03 −0.73 −0.02 +0.52 +0.79 −0.90 +0.23 +0.49 +0.46 +0.40 −1.27 −2.31 −0.57 −0.44 −1.30 −1.15 −1.03 +0.19 −0.04 −0.29 +0.46 −1.19 3 ... 2 2 3 ... 11 2 3 ... 4 1 2 7 ... 4 3 3 2 20 1 3 5 ... ... ... ... ... ... ... −0.29 ± 0.12 −0.19 ± 0.01 +0.03 ± 0.19 −0.17 ± 0.08 −0.86 ± 0.15 −3.21 ± 0.09 −0.45 ± 0.12 −0.58 ± 0.20 +0.05 ± 0.15 −1.95 ± 0.14 −2.87 −2.97 ± 0.19 −1.24 ± 0.16 −1.31 ± 0.08 −0.74 ± 0.10 −0.98 −1.14 ± 0.12 −0.47 ± 0.02 −0.35 ± 0.07 −2.78 −1.93 ± 0.10 +0.69 +0.79 +1.01 +0.81 +0.12 −2.23 +0.53 +0.40 +1.03 −0.97 −1.89 ... −1.99 −0.26 −0.33 +0.24 −0.16 +0.51 +0.63 −1.80 −0.95 15 3 4 4 4 2 6 2 8 ... 4 1 5 ... 5 8 4 ... 10 2 4 1 2 ... ... ... ... ... ... -- 27 -- Table 3.2. Elemental Abundances DY Ori LR Sco BZ Sct Species log ǫo ⊙ [X/H] n [X/Fe] [X/H] n [X/Fe] [X/H] n [X/Fe] C I N I O I Na I Mg I Al I Si I Si II S I K I Ca I Sc II Ti I Ti II V I Cr I Cr II Mn I Fe Co I Ni I Cu I Zn I Y II Ba II La II Ce II Pr II Nd II Sm II Eu II 8.39 7.83 8.69 6.30 7.55 6.46 7.54 7.54 7.19 5.11 6.34 3.07 4.92 4.92 4.00 5.65 5.65 5.50 7.47 4.91 6.22 4.26 4.63 2.20 2.18 1.18 1.61 0.78 1.46 0.95 0.52 +0.10 ± 0.16 +0.54 ± 0.01 −0.20 ± 0.14 −0.25 ± 0.06 −2.32 ± 0.11 −3.29 −1.44 −1.69 ± 0.03 +0.20 ± 0.06 −2.18 ± 0.14 −2.85 +2.33 ± 0.29 −1.93 ± 0.06 −1.99 ± 0.17 −1.84 ± 0.19 −2.23 −2.28 ± 0.26 −0.17 ± 0.16 −2.56 ± 0.14 −2.05 ± 0.12 16 3 2 3 2 1 1 2 8 · · · 2 1 · · · 6 · · · 2 8 3 · · · 2 · · · 4 3 2 · · · · · · · · · · · · · · · · · · +2.33 +2.77 +2.43 +1.98 +0.09 −1.06 +0.79 +0.54 +2.43 +0.05 −0.62 −0.10 +0.33 +0.24 +0.37 −0.05 ± 0.13 −0.05 ± 0.15 −0.04 ± 0.17 −0.29 ± 0.08 −0.81 ± 0.17 +0.07 ± 0.10 −0.08 +0.15 ± 0.05 −0.51 ± 0.11 −1.30 ± 0.08 −0.96 −0.28 ± 0.05 −0.11 ± 0.03 −0.32 ± 0.05 −0.17 −0.05 −0.46 ± 0.17 +2.06 +0.33 −0.18 −0.32 −1.63 ± 0.16 −0.24 12 · · · 4 4 2 3 10 1 4 · · · 6 4 1 · · · · · · 2 4 2 · · · 9 · · · 1 2 · · · · · · · · · · · · · · · · · · 1 +0.22 +0.12 +0.21 −0.12 −0.64 +0.24 +0.09 +0.32 −0.34 −1.13 −0.79 −0.11 +0.06 −0.15 −0.29 −0.15 −1.46 +0.05 ± 0.12 +0.12 ± 0.01 +0.06 ± 0.15 −0.05 ± 0.05 −0.75 ± 0.02 −0.59 ± 0.01 −0.73 ± 0.15 +0.18 ± 0.15 −0.73 −0.91 ± 0.10 −1.13 ± 0.17 −1.25 ± 0.06 −1.18 ± 0.12 −0.74 ± 0.10 −0.91 ± 0.08 −1.09 ± 0.13 −0.82 −0.75 ± 0.11 −0.75 +0.04 ± 0.05 −1.36 ± 0.08 −0.91 −1.79 −0.92 ± 0.01 −0.07 −0.61 −0.64 20 3 4 4 3 · · · 2 2 5 1 7 5 2 6 · · · 5 9 4 · · · 11 1 3 4 1 1 2 · · · · · · 1 1 +0.87 +0.93 +0.88 +0.77 +0.07 +0.23 +0.09 +1.00 +0.09 −0.09 −0.31 −0.43 −0.36 +0.08 −0.09 −0.27 +0.07 +0.07 +0.94 −0.54 −0.09 −0.97 −0.10 +0.21 +0.18 { 1 { -- 8 2 -- Tabe 4.1. Eeea Ab da e RX Ca DF Cyg TX h UZ h Se ie g (cid:15) [X/℄  [X/Fe℄ [X/℄  [X/Fe℄ [X/℄  [X/Fe℄ [X/℄  [X/Fe℄ (cid:12)  C  8.39 0:82 1 0:04 0:26 1 0:26    0:71  0:12 2 0:08   7.83            8.69 0:15  0:14 2 0:63    0:71  0:10 2 0:51 0:49  0:20 3 0:30 a  6.30 0:47  0:12 2 0:31 0:17  0:18 3 0:17 1:13  0:02 3 0:09 0:65  0:08 4 0:14 g  7.55 0:43  0:05 4 0:35    0:66  0:04 2 0:56 0:45  0:27 2 0:37 A  6.46    1:38  0:12 2 0:16 1:15  0:14 3 0:36 Si  7.54 0:34  0:09 8 0:44 0:11  0:06 5 0:11 0:82  0:10 8 0:40 0:45  0:12 12 0:34 Si  7.54 0:52  0:12 2 0:26    0:71  0:22 2 0:51 0:64  0:20 2 0:15 S  7.19 0:57  0:18 3 0:21    0:63  0:13 3 0:59 0:38  0:15 4 0:41   5.11 0:36 1 0:41    Ca  6.34 0:79  0:14 9 0:01 0:23  0:12 2 0:23 1:43  0:14 14 0:21 1:10  0:19 9 0:31 S  3.07 1:20  0:09 5 0:42 0:96  0:32 2 0:96 1:79  0:11 4 0:57 1:26  0:18 7 0:47 Ti  4.92    0:12  0:16 4 0:12 1:16  0:16 8 0:06 1:07  0:27 7 0:28 Ti  4.92 0:62  0:19 7 0:16 0:19 1 0:19 1:19  0:14 7 0:03 0:93  0:10 8 0:14 V  4.00 0:83 1 0:05 0:24  0:19 2 0:24    1:31 1 0:52 C  5.65 1:00  0:18 12 0:22 0:01  0:13 3 0:01 1:55  0:17 9 0:33 1:38  0:22 5 0:59 C  5.65 1:03  0:17 9 0:25 0:15  0:07 2 0:15 1:49  0:16 7 0:27 1:25  0:16 9 0:46   5.50 1:13  0:11 5 0:35    1:88  0:10 3 0:66 1:48  0:14 5 0:69 Fe  7.47 0:78 0.00 1:22 0:79 C  4.91    0:17  0:0 6 2 0:17 1:21  0:08 2 0:01 i  6.22 0:77  0:12 14 0:01 0:01  0:08 3 0:01 1:32  0:14 16 0:10 0:91  0:17 23 0:12 C  4.26 0:88 1 0:10    1:57 1 0:35 1:11  0:10 2 0:32 Z  4.63 0:63  0:17 3 0:15 0:62 1 0:62 1:23  0:08 3 0:01 0:74  0:18 3 0.05 Y  2.20 1:26  0:26 7 0:78 0:73 1 0:73 1:76  0:10 5 0:51 1:71  0:14 5 0:92 Ba  2.18    1:28  0:08 2 0:06 1:48 1 0:69 a  1.18 0:63  0:05 2 0:25 1:38  0:28 2 0:16 1:68  0:20 2 0:89 Ce  1.61 0:96  0:18 7 0:18 1:54  0:22 8 0:32 1:57  0:22 3 0:78   0.78    1:45 1 0:23 1:60 1 0:81 d  1.46 0:76  0:09 5 0:02 1:26  0:10 6 0:04 1:35  0:15 3 0:56 S  0.95 0:63  0:10 3 0:15 1:11  0:12 5 0:11 1:11  0:15 5 0:32 E  0.52 0:31 1 0:46 0:09 1 0:09 0:65 1 0:57 0:88 1 0:09 { 1 { -- 9 2 -- A S  AR Sg AZ Sg V V Se ie g (cid:15) [X/℄  [X/Fe℄ [X/℄  [X/Fe℄ [X/℄  [X/Fe℄ [X/℄  [X/Fe℄ (cid:12)  Tabe 4.2. Eeea Ab da e C  8.39 0:10  0:24 7 0:79 1:07  0:24 3 0:26 0:56  0:13 2 0:99 0:70  0:22 5 1:09   7.83   8.69 0:26  0:11 2 0:43 0:39  0:08 2 0:94 0:12  0:10 2 1:43 0:02  0:01 2 0:41 a  6.30 0:25  0:11 4 0:44 1:15  0:04 3 0:18 0:12  0:02 2 0:27 g  7.55 0:36  0:17 2 0:33 0:92  0:11 3 0:41 1:15 1 0:40 0:06 1 0:33 A  6.46 0:70 1 0:01 1:73  0:07 4 0:40 0:45  0:02 2 0:06 Si  7.54 0:39  0:14 7 0:30 0:75  0:12 11 0:58 0:80  0:10 9 0:75 0:23  0:13 8 0:16 Si  7.54 0:20  0:16 2 0:49 0:81  0:09 2 0:52 S  7.19 0:08  0:13 4 0:61 0:82  0:06 4 0:51 0:34  0:13 4 1:21 0:57  0:14 3 0:96   5.11 Ca  6.34 0:64  0:08 6 0:05 1:44  0:08 10 0:09 1:80  0:16 10 0:25 0:47  0:08 8 0:08 S  3.07 0:96  0:19 5 0:27 1:41  0:19 5 0:08 1:78  0:15 7 0:23 0:69  0:04 4 0:30 Ti  4.92 0:91  0:18 2 0:22 0:22  0:14 2 0:17 Ti  4.92 0:86  0:20 5 0:17 1:21  0:18 9 0:12 1:57  0:07 2 0:02 0:14  0:09 2 0:25 V  4.00    0:36  0:09 7 0:03 C  5.65 0:61 1 0:08 1:56  0:15 10 0:23 2:32  0:07 5 0:77 0:43  0:16 2 0:04 C  5.65 0:70  0:07 4 0:01 1:52  0:14 6 0:19 0:35  0:14 5 0:04   5.50 0:99  0:02 2 0:30 1:74  0:10 5 0:41 2:08 1 0:53 0:47  0:21 2 0:08 Fe  7.47 0:69 1:33 1:55 0:39 C  4.91 1:77 1 0:22 0:34  0:06 3 0:05 i  6.22 0:78  0:12 6 0:09 1:19  0:16 18 0:14 1:76  0:14 8 0:21 0:53  0:11 14 0:14 C  4.26 0:70 1 0:01 1:41 1 0:08 1:91 1 0:36 Z  4.63 0:61  0:10 3 0:08 1:20  0:05 3 0:13 0:27  0:24 2 0:12 Y  2.20 1:12  0:05 6 0:43 2:15  0:03 3 0:60 0:52  0:19 4 0:13 Ba  2.18 1:62  0:21 2 0:93 1:59 1 0:26 a  1.18 1:06 1 0:37 1:17  0:05 2 0:16 0:52  0:01 3 0:13 Ce  1.61 1:25  0:15 4 0:56 1:58  0:19 7 0:25 0:84  0:13 2 0:45   0.78 1:19 1 0:14 0:49 1 0:10 d  1.46 1:00  0:17 3 0:31 1:36  0:09 15 0:03 1:51  0:14 2 0:04 S  0.95 1:30  0:14 12 0:03 E  0.52 0:12 1 0:57 0:93 1 0:40 1:01 1 0:54 0:22 1 0:17 -- 30 -- RX Cap UY CMa AR Sgr TX Oph V Vul EQ Cas 6100 6120 6140 6160 6180 4 3 2 1 0 Fig. 1. -- Spectra of representative sample are presented in descending order of their tem- peratures (from top to bottom). The temperature of UY CMa is within 200K of RX Cap and AR Sgr but the lines of Ca I and Fe I are considerably weak in UY CMa due to dust-gas separation effect. -- 31 -- Fig. 2. -- Abundance [X/H] versus condensation temperature TC for UY CMa. Elements are identified by their chemical symbol. -- 32 -- Fig. 3. -- Abundance [X/H] versus condensation temperature TC for EQ Cas. Elements are identified by their chemical symbol. -- 33 -- Fig. 4. -- Abundance [X/H] versus condensation temperature TC for HP Lyr. Elements are identified by their chemical symbol. -- 34 -- Fig. 5. -- Abundance [X/H] versus condensation temperature TC for DY Ori. Elements are identified by their chemical symbol. -- 35 -- Fig. 6. -- Abundance [X/H] versus condensation temperature TC for LR Sco. Elements are identified by their chemical symbol. -- 36 -- Fig. 7. -- Abundance [X/H] versus condensation temperature TC for BZ Sct. Elements are identified by their chemical symbol. -- 37 -- Ca Ni Fe Cr Ti C O S Na Zn Cu Si Mn Mg -3 -2 -1 0 0 -1 -2 -3 -4 Fig. 8. -- Abundance [X/H] for UY CMa versus [X/H] for the cool interstellar cloud along the line of sight to ζ Oph. Elements are identified by their chemical symbol. -- 38 -- This series QY Sge Maas Fig. 9. -- Abundance ratio [Zn/Fe] versus abundance [Fe/H] for RV Tauri and related vari- ables. RV Tauri stars from our series of papers are represented by circles. QY Sge (Rao et al. 2002) is shown by the triangle. RV Tauri stars analysed by Maas et al. (2002) and Maas (2003) are shown by squares. Filled circles and squares denote high velocity stars. -- 39 -- This series QY Sge Maas Fig. 10. -- Abundance ratio [S/Fe] versus abundance [Fe/H] for RV Tauri and related vari- ables. RV Tauri stars from our series of papers are represented by circles. QY Sge (Rao et al. 2002) is shown by the triangle. RV Tauri stars analysed by Maas et al. (2002) nd by Maas (2003) are shown by squares. Filled circles and squares denote high velocity stars. The solid line shows the run of [S/Fe] with [Fe/H] for unevolved stars (see text). -- 40 -- This series QY Sge Maas Fig. 11. -- Abundance ratio [S/Zn] versus abundance [Zn/H] for RV Tauri and related variables. RV Tauri stars from our series of papers are represented by circles. QY Sge (Rao et al. 2002) is shown by the triangle. RV Tauri stars analysed by Maas et al. (2002) and Maas (2003) are shown by squares. Filled circles and squares denote high velocity stars. The solid line shows the run of [S/Zn] with [Zn/H] for unevolved stars (see text). -- 41 -- This series QY Sge Maas Fig. 12. -- The abundances [C/Zn] (top panel), [O/Zn] (middle panel), and log C/O (bottom panel) as a function of the zinc abundance [Zn/H] for RV Tauri and related variables. RV Tauri stars from our series of papers are represented by circles. QY Sge (Rao et al. 2002) is shown by the triangle. Stars analysed by Maas et al. (2002) and Maas (2003) are shown by squares. Filled circles and squares denote high velocity stars. The solid line shows the expected runs with [Zn/H] for giants after the first dredge-up (see text). -- 42 -- 2 1 0 -1 4000 5000 6000 7000 Fig. 13. -- Abundance ratio [Zn/Fe] versus Teff for RV Tauri stars. Data points include the present work as well as from our earlier papers. The ratio [Zn/Fe] increases towards hotter temperature indicating increasing severity of the dust-gas separation in hotter stars. to cooler ones. Symbols represent Preston spectroscopic groups A, B and C.
0801.3104
1
0801
2008-01-20T20:06:03
Limits on Planets Around White Dwarf Stars
[ "astro-ph" ]
We present limits on planetary companions to pulsating white dwarf stars. A subset of these stars exhibit extreme stability in the period and phase of some of their pulsation modes; a planet can be detected around such a star by searching for periodic variations in the arrival time of these pulsations. We present limits on companions greater than a few Jupiter masses around a sample of 15 white dwarf stars as part of an on-going survey. One star shows a variation in arrival time consistent with a 2 M_J planet in a 4.5 year orbit. We discuss other possible explanations for the observed signal and conclude that a planet is the most plausible explanation based on the data available.
astro-ph
astro-ph
Limits on Planets Around Pulsating White Dwarf Stars Fergal Mullally1, D. E. Winget2, Steven Degennaro2, Elizabeth Jeffery2, S. E. Thompson3, Dean Chandler4 and S. O. Kepler5 ABSTRACT We present limits on planetary companions to pulsating white dwarf stars. A subset of these stars exhibit extreme stability in the period and phase of some of their pulsation modes; a planet can be detected around such a star by searching for periodic variations in the arrival time of these pulsations. We present limits on companions greater than a few Jupiter masses around a sample of 15 white dwarf stars as part of an on-going survey. One star shows a variation in arrival time consistent with a 2 MJ planet in a 4.5 year orbit. We discuss other possible explanations for the observed signal and conclude that a planet is the most plausible explanation based on the data available. Subject headings: planetary systems -- white dwarfs 1. Introduction All main-sequence stars with mass less than about 8 M⊙ will end their lives as white dwarf stars (WDs). As such, WDs are a fossil record of star formation in the Galaxy from earliest times to just a few hundred million years ago. WDs offer a window into the ultimate fate of planetary systems, including our own solar system, and whether planets can survive the final stages of stellar evolution. The proper- 1Department of Astrophysical Sciences, Princeton University, Princeton, NJ 08544 2Department of Astronomy, University of Texas at Austin, Austin TX 78712 3Department of Physics and Astronomy, University of Delaware, Newark, DE 19716 4Meyer Observatory, Clifton, TX 76530 5Instituto de F´ısica, Universidade Federal do Rio Grande do Sul, 91501-900 Porto-Alegre, RS, Brazil ties of a WD are relatively insensitive to the mass of the progenitor: the mass distribu- tion of isolated WDs is narrowly distributed in a peak 0.1 M⊙ wide around a mean mass of 0.59 M⊙ (Kepler et al. 2007). Surveys of WDs can therefore search a wide range of main se- quence progenitor masses and the low lumi- nosity of a WD means any companion planets can be potentially followed up with current di- rect detection technology in the mid-infrared. Livio & Soker (1984) considered the fate of a planet engulfed in the envelope of a red gi- ant star. They determined that below a cer- tain mass the planet would be evaporated and destroyed, but larger objects would accrete material and spiral in toward the stellar core. They predicted that the end state of these sys- tems would be a tight binary consisting of a white dwarf and a brown dwarf and suggested 1 this mechanism might explain the origin of cataclysmic variable (CV) systems. Planets that are not engulfed, and suffi- ciently far from the stellar surface that tidal drag is small, will drift outwards to conserve angular momentum (as described in Jeans 1924). Duncan & Lissauer (1998) investi- gated the stability of the outer solar sys- tem when the sun undergoes mass loss as a red giant. They found that the system was stable on timescales of at least 10 bil- lion years for reasonable amounts of mass loss, but for larger amounts typical of WDs formed from more massive stars, the planets' orbits became unstable on timescales of . 108 years. Debes & Sigurdsson (2002) looked what would happen if the orbits of two plan- ets became unstable and determined that if orbits crossed, the most likely result was that one planet would be scattered into a shorter period orbit, while the other would be boosted into a longer orbit or ejected from the system. For planets scattered inward, the extreme flux from a newborn white dwarf would strip the outer atmospheric layers. Villaver & Livio (2007) estimated that a 2 MJ planet 1.8 AU from a young white dwarf would lose half its mass in this manner. The picture drawn by this brief survey of theory is a population of planets in long period orbits around WDs, with a number of objects scattered closer to the star, or in very tight binaries. report the detection of a planetary mass com- panion in a 1.7 AU orbit around an extreme horizontal branch sdB star. Pulsating WDs allow the possibility of searching for the presence of companion plan- ets as changes in the observed arrival time of the pulsations. Hydrogen atmosphere (DA) WDs pulsate in an instability strip approx- imately 1200 K wide near 12,000 K and are known as DAVs (or ZZ Ceti stars). The pul- sations are non-radial g-modes with periods of order 100 -- 1500s and amplitudes of a few percent. The pulsation properties of DAVs vary with temperature. Those near the hot end of the strip tend to have a smaller number of shorter period, lower amplitude modes with sinu- soidal lightcurves and are known as hDAVs, while those nearer the red edge show more, larger amplitude modes and asymmetric pulse shapes. The change in pulsation properties is most likely due to increasing depth of the convection layer near the surface of the star (Brickhill 1983, and subsequent articles). Stover et al. (1980) first noted that modes on hDAVs often exhibited an impressive sta- bility in the period and phase of pulsation. Kepler et al. (2005) measured the rate of pe- P , of one mode in the hDAV riod change, G117-B15A to be 3.57(82) × 10−15, while Mukadam et al. (2003) constrained P of a mode in R548 to ≤ 5.5(1.9) × 10−15 and O'Donoghue & Warner (1987) constrained P of L19-2 to < 3.0 × 10−14. Together with pulsars, these objects are the most stable as- trophysical clocks known. Measurements of P require datasets of be- tween 10 and 30 years, but the investment in time yields a suitable scientific reward. Mea- P provides a rare opportunity surement of to directly test models of structure and com- position of the core of a star (Kepler et al. 2005), constrain the current rate of change of the gravitational constant (Benvenuto et al. for Searches sub-stellar companions to WDs have concentrated on exploiting the lower contrast between star and compan- ion, especially in the infrared (e.g. Probst 1983; Burleigh et al. 2002; Farihi et al. 2005; Friedrich et al. 2006; Debes et al. 2006; Mullally et al. 2007). Although a couple of brown dwarf stars have been found with this approach (Zuckerman & Becklin 1992; Farihi & Christopher 2004), as yet no direct detections of planets have been claimed. Silvotti et al. (2007), us- ing the same timing method discussed here, 2 2004), as well as provide useful constraints on the mass of the hypothesized axion or other super-symmetric particles (Isern et al. 1992; C´orsico et al. 2001; Bischoff-Kim et al. 2007) If a planet is in orbit around a star, the star's distance from the Sun will change pe- riodically as it orbits the center of mass of the planetary system. If the star is a stable pulsator like a hDAV, this will cause a peri- odic change in the observed arrival time of the otherwise stable pulsations compared to that expected based on the assumption of a con- stant period. The change in arrival time, τ , is given by apmp sin i M∗c τ = (1) where ap is the semi-major orbital axis of the planet, mp is the planet mass, M∗ is the mass of the white dwarf, c is the speed of light and i is the inclination of the orbit to the line of sight. In common with astrometric methods, the sensitivity increases with the orbital sep- aration, making long period planets easier to detect given data sets with sufficiently long baselines. In 2003 we commenced a pilot survey of a small number of DAVs in the hope of de- tecting the signal of a companion planet. We present here a progress report of the first 3 -- 4 years of observations on 12 objects, as well as presenting limits around 3 more objects based partly on archival data stretching as far back as 1970. For one object we find a signal con- sistent with a planetary companion. Further observations are necessary to confirm the na- ture of this system. For our other objects, we can constrain the presence of planets down to a few Jupiter masses at 5 AU, with more strin- gent limits for stars with archival data. 2. Our Survey Kleinman et al. (2004) and Eisenstein et al. (2006) published a large number (∼ 600) can- 3 didate DAVs with spectra taken by the Sloan Digital Sky Survey (Adelman-McCarthy et al. 2006). A follow up survey by Mukadam et al. (2004) and Mullally et al. (2005) confirmed 46 of these candidates to be pulsators. We selected our targets from these two papers as well as earlier known DAVs published in the literature (see Bergeron et al. 1995; Fontaine et al. 2003). The ideal DAV for this survey would ex- hibit a number of isolated, relatively low am- plitude (0.5 -- 2%) modes. Multiplet, or oth- erwise closely-spaced (. 1 s) modes are diffi- cult to resolve in single-site data and inter- ference between the unresolved modes makes accurately measuring the phase difficult. We selected a sample of 15 stars brighter than 19th mag for long-term study. With the ex- ception of R548, which has a well studied double mode, we chose stars with one iso- lated mode with amplitude & 0.5%. Only one star, SDSS J221458.37−002511.7 has two modes suitable for study. We have monitored this sample of stars for 4 years using the Argos Prime Focus ccd pho- tometer (Nather & Mukadam 2004) on the 2.1m Otto Struve Telescope at McDonald Ob- servatory. We list the observed objects in Ta- ble 1, along with the period and P of the analyzed modes. We observed each object with exposure times of 5-15s for periods of 4-8 hours per night. The exposure times are cho- sen to be very much shorter than the Nyquist frequency of the shortest period mode ob- served on the star (& 100s), and the observing time to sample many consecutive cycles of the pulsation. We reduce our data in the manner de- scribed in Mullally et al. (2005) with one im- provement. Argos suffers from a fluctuating bias level but does not have an overscan re- gion. To account for this we measure the bias from a dead column and subtract this value from both our science and dark frames be- fore flatfielding. This is clearly not ideal, but the best approach to measuring the bias avail- able, given that the level can vary by up to 5 dn/pixel on timescales shorter than the expo- sure time. We perform weighted aperture photometry with a variety of apertures using the IRAF package wphot, choosing the aperture that gives the best signal-to-noise by eye. We di- vide the light curve by a combination of one or more reference stars, remove points affected by cloud, fit a second order polynomial to re- move the long term trend caused by differen- tial extinction, and correct our timings for the motion of the Earth around the barycenter of the solar system using the method of Stumpff (1980), accounting for all UTC leapseconds up to and including January 2006. We combine all the data on a star in a given month for analysis, typically 8-16 hours. We first compare the alias pattern of the peaks in the Fourier transform (FT) with a window function to identify closely spaced modes and multiplets. It is more difficult to measure the phase of closely spaced modes (. 70 µHz) be- cause interference between the unresolved pe- riodicities requires significantly more data to resolve, so we focus only on isolated modes. Having selected a mode for analysis, we first measure the period by fitting a sine curve using the Marquant-Levenberg non-linear least squares technique (Bevington 1969). We attempt to obtain between two and four ac- curate timings on each star per year. As we accumulate data, we re-measure the period using the entire dataset, before measuring the phase of that period in each month's data using a least squares fit. We compare the ob- served phase to that expected based on the assumption of a constant period and plot the result in an O-C diagram. We also check that the amplitude of a mode is stable from month to month. Varying amplitudes are a symp- tom of either unresolved companion periods, 4 or an instability in the pulsation mechanism. None of the modes discussed here displayed any amplitude variability inconsistent with observational error. 3. Results 3.1. GD66 GD66 (WD0517+307) is an 11,980 K, log g=8.05, 0.64 M⊙ hDAV (Bergeron et al. 2004) with a V magnitude of 15.6 (Eggen 1968) corresponding to a distance of about 51pc (Mullally & de Graff 2005). The FT is dominated by a single mode at 302 s, triplets of modes separated by ≈6.4 µHz at 271 and 198 seconds and a handful of other lower am- plitude modes. There are also some combina- tion and harmonic peaks present. A sample FT is shown in Figure 1. The presence of closely spaced peaks at 271 and 198 s makes it more difficult to accurately measure their phase and our analysis concen- trates on the 302 s mode. We show an O-C di- agram for the arrival times of the 302 s mode in Figure 2. The curvature in this diagram is unmistakable. Instabilities in the pulsation modes of DAVs often manifest as variations in the amplitude of pulsation. In Figure 3 we plot the amplitude of the 302s mode as a function of time and find the amplitude varies between 1.1 and 1.2%. We can reproduce vari- ations of this magnitude by small changes in our reduction method, and conclude that the amplitude is stable within our ability to mea- sure it. By fitting a sine curve to the O-C diagram we find a period of 4.52 years and an amplitude of 3.84 s. The data collected to date is also con- sistent with a parabola. O-C diagrams of hDAVs are expected to show parabolic be- havior as the cooling of the star produces a monotonic increase in the period of pulsa- tion (see Kepler et al. 1991). However, based on observations of other DAVs and mod- els of white dwarf interiors (Bradley 1998; Benvenuto et al. 2004) we expect the cooling to cause a Pcool ∼ 10−15. If we fit a parabola to our data we find a P = 1.347(95) × 10−12, three orders of magnitude larger than ex- pected from cooling alone. The tangential motion of the star with respect to the line of sight also causes a parabolic curvature in the O-C diagram. As the star moves linearly in space perpendicular to our line of sight, its distance to us changes parabolically (Shklovskii 1970; Pajdosz 1995). The USNO-B1 catalogue (Monet et al. 2003) quotes a proper motion of 131.6(5.0) mas/yr corresponding to a Ppm of 6.4 × 10−16, again too small to explain the observed data. Apparently periodic signals in O-C dia- grams can be caused by random jitter or drift in the period of the pulsator. A likelihood statistic, L, that a given data set was caused by different combinations of observational er- ror, period jitter and drift can be calculated according to Koen (2006). We expand his methodology to determine the likelihood that the data shown in Figure 2 is the signature of a companion, or the result of stochastic changes in the pulsation period. We first calculate L for a model of the data that seeks to explain the data by invoking jitter or drift in the pe- riod and find values for log L of -15.1673 and -15.0918 respectively. A model including both jitter and drift gives a similar value. Next, we calculate the likelihood that the residuals of the sine fit can be explained by observa- tional error alone, and find a value of log L of -11.5549. This strongly disfavors the hypoth- esis that the observations can be explained by small random changes in the pulsation period. Time series observations of another DAV, G29-38, showed a variation in the phase of one mode over 3 months consistent with an 0.5M⊙ object in an eccentric 109 day orbit, but a change in amplitude the following year made the mode unreliable as an accurate clock (Winget et al. 1990). However, analysis of other modes on the same star failed to repro- duce this behavior (Kleinman et al. 1994) and near-infrared imaging (Kuchner et al. 1998; Debes et al. 2005) did not detect any sub- stellar companions. It is possible that the same internal effect that mimicked a compan- ion to G29-38 is also present in GD66 albeit with a much smaller amplitude and consider- ably longer period. If we assume the curvature is caused by a planet in a circular orbit, the best fit pe- riod is 4.52(21) years and we can use Ke- pler's laws to determine an orbital separa- tion, ap = 2.356(81) AU. The amplitude of the sine curve, τ = 3.84(32) s is related to the semi-major axis of the star's orbit, a∗, by τ c = a∗ sin i where i is the inclination of the orbit to the line of sight and c is the speed of light. The mass of the planet, mp, is equal to (M∗a∗)/ap, where M∗ is the mass of the star. Using these two equations we find an mp sin i of 2.11(14) MJ . From our best fit cir- cular orbit, we predict we will obtain observa- tions spanning an entire orbit in early 2008. 3.2. Other stars GD66 is the only star in our sample that shows strong evidence for a planetary com- panion. However, we can place interesting limits on the presence of planets around the other stars. The fundamental limit on our ability to de- tect planets is set by the scatter in the O-C diagram. Other factors which affect this limit include the timespan and sampling pattern of the data. We perform a Monte Carlo analy- sis to estimate the region of the mass -- orbital separation plane in which planets are actually detectable around each star based on the data available. We randomly choose a planet mass, orbital separation, eccentricity, and other or- bital parameters and calculate the effect this planet would induce on the O-C diagram of a 5 0.59 M⊙ white dwarf. We then sampled this O-C diagram with the same observing pattern and error bars as our actual data for each star and fit the resulting O-C diagram with a sine curve and a parabola. If either the ampli- tude of the sine curve or the curvature of the parabola were measured with 3σ confidence, the hypothetical planet was determined to be detected. We repeated this process 106 times for each star and drew a shaded relief map in- dicating the percentage of the time a planet with a given mass and orbital separation was detected with either technique to 3σ, with dark shades indicating near 100% detection efficiency, and white indicating regions where planets were unlikely to be detected. We show the O-C diagram and the relief map for each star in Figures 3-16. As can be seen in the fig- ures, the annual sampling pattern means that planets with orbital periods of 1 (Earth) year are difficult to detect, resulting in the weak limits for planets at separations of slightly less than 1 AU. 3.3. Notes on Individual Stars G117-B15A. -- Also known as WD0921+354. 30 years of archival data comes from Kepler et al. (2005), although some of the more recent data in that work was taken in conjunction with this project. Where data did not come from our observations we used the O-C value quoted in their Table 1. With this timebase, the curvature caused by the change in period due to cooling becomes evident. This cooling effect is removed from the data before per- forming the Monte Carlo simulation. The lim- its on long period planets placed around this star are among the best constraints placed around any stellar object by any technique. See Figure 4. G185-32. -- Also known as WD1935+279. The point from the early 90's comes from archival data from the Whole Earth Telescope (Xcov8, Castanheira et al. 2004). This extra 6 point gives a long baseline, but the poor cov- erage in our survey reduces the sensitivity. See Figure 5. R548. -- Also known as ZZ Ceti and WD0133- 116. The entire data set comes from Mukadam et al. (2003). See Figure 8. SDSS J011100.63+001807.2. -- At g =18.6th magnitude, this is our faintest target and cor- respondingly has our weakest limits. Also, because it could only be observed under the best conditions the data coverage is quite low. The apparently impressive curvature in this O-C diagram in entirely due to the last data point and should be treated with considerable skepticism. See Figure 10. SDSS J135459.89+010819.3. -- This bright (16.4th magnitude) star has a baseline stretch- ing back to early 2003 and some of the highest accuracy time measurements, and as a result has the best limits on planets for stars with- out archival data. We could have detected a Jupiter mass planet at 5 AU had one been present. See Figure 14. SDSS J221458.37−002511.7. -- This is the only star in the sample for which reasonable O-C diagrams were obtained for two modes. The O-C diagram shown (see Figure 17) is the weighted sum of the O-C values for these two individual modes. 4. Discussion In this pilot study we find a signal con- sistent with a companion planet around one star in a sample of only 15. Although further observations will be necessary to conclusively identify the cause, it augers well for the fu- ture potential of this and other white dwarf planet searches. According to the theoretical arguments discussed in §1, the distribution of planets around WDs can be expected to be weighted toward planets in long period orbits. The population of hot Jupiters around the main sequence progenitors will likely be de- stroyed, while more distant planets will drift outward with stellar mass loss. Evidence of a planet in a relatively short orbit encour- ages us to continue monitoring for planets with greater orbital separations. For the other stars in our sample, we can rule out the pres- ence of planets down to a few Jupiter masses at 5 AU. With more data, we will extend our limits beyond 10 AU into the regime where we expect planets to be most frequent. Our search spans a broad range of progeni- tor masses. To estimate the progenitor masses we compare the spectroscopically measured Teff and log g from Eisenstein et al. (2006) and Bergeron et al. (2004) to WD interior models from Holberg & Bergeron (2006) to obtain a WD mass. We then used the initial fi- nal mass relations (IFMRs) of Williams et al. (2004) and Ferrario et al. (2005) to calcu- late the progenitor mass, taking the weighted mean of the two relations as the best value, and adding the difference between the two methods in quadrature to the uncertainty. We present the masses in Table 2. Although there is still considerable uncertainty in this relation, the results give some indication of the type of star that created the WD. The progenitor masses correspond to a range of spectral type from approximately B6 to F9 (Habets & Heintze 1981), a range that is largely complementary to the radial velocity method. 5. Conclusion We present our results on an on-going sur- vey for planets around 15 pulsating white dwarf stars. Our survey data spans 3 -- 4 years with archival data on some stars stretching back to 1970. We are already sensitive to planets down to a few Jupiter masses at dis- tances of approximately 5 AU. For one star, we observe a curvature in the O-C diagram consistent with a planet in a 4.5 year orbit. Further observations are necessary to span a 7 full orbit of this candidate object. If con- firmed, this will be the first planet discovered around a WD, and together with the planet discovered around an sdB, suggests that plan- ets regularly survive the death of their parent star, and that WDs will be fruitful targets for planet searches. This work was supported by a grant from the NASA Origins program, NAG5-13094 and is performed in part under contract with the Jet Propulsion Laboratory (JPL) funded by NASA through the Michelson Fellowship Pro- gram. JPL is managed for NASA by the Cal- ifornia Institute of Technology. REFERENCES Adelman-McCarthy, J. K., et al. 2006, ApJS, 162, 38 Benvenuto, O. G., Garc´ıa-Berro, E., & Isern J., 2004, Phys. Rev. D, 69, 082002 Bergeron, P., Fontaine, G., Bill`eres, M., Boudreault, S., & Green, E. M. 2004, ApJ, 600, 404 Bergeron, P., Wesemael, F., Lamontagne, R., Fontaine, G., Saffer, R. A., & Allard, N. F. 1995, ApJ, 449, 258 Bevington, P. R. 1969, "Data reduction and er- ror analysis for the physical sciences" (2nd ed.; New York: McGraw-Hill, 1969) Bischoff-Kim, A., Montgomery, M. H., & Winget D. E. , 2007, astro-ph/0711.2041 Bradley, P. A. 1998, ApJS, 116, 307 Brickhill, A. J. 1983, MNRAS, 204, 537 Burleigh, M. R., Clarke, F. J., & Hodgkin, S. T. 2002, MNRAS, 331, L41 C´orsico, A. H., Benvenuto, O. G., Althaus, L. G., Isern, J., & Garc´ıa-Berro, E. 2001, New As- tronomy, 6, 197 Castanheira, B. G., et al. 2004, A&A, 413, 623 Debes, J. H., & Sigurdsson, S. 2002, ApJ, 572, 556 Debes, J. H., Ge, J., & Ftaclas, C. 2006, AJ, 131, 640 Debes, J. H., Sigurdsson, S., & Woodgate, B. E. 2005, ApJ, 633, 1168 Duncan, M. J., & Lissauer, J. J. 1998, Icarus, 134, 303 Eggen, O. J. 1968, ApJS, 16, 97 Eisenstein, D. J., et al. 2006, ApJS, 167, 40 Farihi, J., & Christopher, M. 2004, AJ, 128, 1868 Farihi, J., Becklin, E. E., & Zuckerman, B. 2005, ApJS, 161, 394 Ferrario, L., Wickramasinghe, D., Liebert, J., & Williams, K. A. 2005, MNRAS, 361, 1131 Stover, R. J., Nather, R. E., Robinson, E. L., Hesser, J. E., & Lasker, B. M. 1980, ApJ, 240, 865 Stumpff, P. 1980, A&AS, 41, 1 Villaver, E., & Livio, M. 2007, ApJ, 661, 1192 Williams, K. A., Bolte, M., & Koester, D. 2004, Fontaine, G., Bergeron, P., Bill`eres, M., & ApJ, 615, L49 Charpinet, S. 2003, ApJ, 591, 1184 Friedrich, S., Zinnecker, H., Correia, S., Brandner, W., Burleigh, M., & McCaughrean, M. 2006, astro-ph/0611511 Habets, G. M. H. J., & Heintze, J. R. W. 1981, A&AS, 46, 193 Holberg, J. B., & Bergeron, P. 2006, AJ, 132, 1221 Isern, J., Hernanz, M., & Garc´ıa-Berro, E. 1992, ApJ, 392, L23 Jeans, J. H. 1924, MNRAS, 85, 2 Kepler, S. O., et al. 1991, ApJ, 378, L45 -- -- . 2005, ApJ, 634, 1311 Kepler, S. O., Kleinman, S. J., Nitta, A., Koester, D., Castanheira, B. G., Giovannini, O., Costa, A. F. M., & Althaus, L. 2007, MNRAS, 375, 1315 Kleinman S. J. Harris, H. C., et al. 2004, ApJ, 607, 426 Kleinman, S. J., et al. 1994, ApJ, 436, 875 Koen, C. 2006, MNRAS, 365, 489 Kuchner, M. J., Koresko, C. D., & Brown, M. E. 1998, ApJ, 508, L81 Livio, M., & Soker, N. 1984, MNRAS, 208, 763 Monet, D. G., et al. 2003, AJ, 125, 984 Mukadam, A. S., et al. 2003, ApJ, 594, 961 -- -- . 2004, ApJ, 607, 982 Mullally, F., & de Graff, M. 2005, 14th European Workshop on White Dwarfs, ed. Koester D., & Moehler S., 334, 506 Mullally, F., Kilic, M., Reach, W. T., Kuchner, M. J., von Hippel, T., Burrows, A., & Winget, D. E. 2007, ApJS, 171, 206 Mullally, F., Thompson, S. E., Castanheira, B. G., Winget, D. E., Kepler, S. O., Eisenstein, D. J., Kleinman, S. J., & Nitta, A. 2005, ApJ, 625, 966 Nather, R. E., & Mukadam, A. S. 2004, ApJ, 605, 846 O'Donoghue, D., & Warner, B. 1987, MNRAS, 228, 949 Pajdosz, G. 1995, A&A, 295, L17 Probst, R. G. 1983, ApJS, 53, 335 Shklovskii, I. S. 1970, Soviet Astronomy, 13, 562 Silvotti, R., et al. 2007, Nature, 449, 189 Winget, D. E., et al. 1990, ApJ, 357, 630 Zuckerman, B., & Becklin, E. E. 1992, ApJ, 386, 260 This 2-column preprint was prepared with the AAS LATEX macros v5.2. 8 Table 1 Modes used to construct O-C diagrams Star Period (sec) Amplitude % T0 (bjd) P G117−B15A. . . G185−32 G238−53 GD244 GD66 R548 SDSS J001836.11+003151.1 SDSS J011100.63+001807.2 SDSS J021406.78−082318.4 SDSS J091312.74+403628.8 SDSS J101548.01+030648.4 SDSS J135459.88+010819.3 SDSS J135531.03+545404.5 SDSS J172428.42+583539.0 SDSS J221458.37−002511.7 SDSS J221458.37−002511.7 215.1973888(12) 370.2203552(55) 122.1733598(38) 202.9735113(40) 302.7652959(21) 212.76842927(51) 257.777859(13) 292.9445269(90) 262.277793(11) 172.605159(15) 254.9184503(56) 198.3077098(14) 323.9518703(69) 335.536871(14) 195.1406388(64) 255.1524057(30) 1.9 0.1 0.2 0.4 1.2 0.4 0.6 1.9 0.6 0.3 0.7 0.6 2.2 0.6 0.4 1.3 2442397.9194943(28) −1.07(49) × 10−13 −0.5(1.0) × 10−13 2453589.6557652(39) −5.7(2.4) × 10−13 2453168.6334567(35) 0.2(2.8) × 10−13 2452884.8712580(31) 1.347(95) × 10−12 2452938.8846146(28) 1.2(4.0) × 10−15 2446679.833986 9.4(9.2) × 10−13 2452962.6358455(41) 3.87(43) × 10−12 2452963.7174455(44) −1.5(7.5) × 10−13 2452941.7929412(37) 9.6(9.8) × 10−13 2453024.8275265(47) 7.2(3.6) × 10−13 2453065.6152116(41) −5.3(7.8) × 10−14 2452665.9507137(33) 1.39(47) × 10−12 2453082.8582407(39) 1.23(85) × 10−12 2453139.8477241(37) 6.2(3.6) × 10−13 2452821.8513218(35) 1.7(2.1) × 10−13 2452821.8521749(35) Note. -- T0 is the time of the arbitarily defined zeroth pulse and is given in units of barycentric corrected julian day. Data on R548 comes from Mukadam et al. (2003) who do not provide a value for uncertainty in T0. Except for GD66, we do not claim statistical significance for the measurement of P for any star. 9 Table 2 Stellar Parameters Star Magnitude G117−B15A G185−32 G238−53 GD244 GD66 R548 SDSS J001836.11+003151.1 SDSS J011100.63+001807.2 SDSS J021406.78−082318.4 SDSS J091312.74+403628.8 SDSS J101548.01+030648.4 SDSS J135459.89+010819.3 SDSS J135531.03+545404.5 SDSS J172428.42+583539.0 SDSS J221458.37−002511.7 15.7 13.0 15.5 15.6 15.6 14.2 17.4 18.8 17.9 17.6 15.7 16.4 18.6 17.5 17.9 Teff (K) 11630 12130 11885 11645 11980 11894 11696 11507 11565 11677 11584 11658 11576 11544 11439 log g Initial Mass Final Mass (M⊙) (M⊙) 7.98 8.05 7.91 8.01 8.05 7.97 7.93 8.26 7.92 7.87 8.14 8.01 7.95 7.90 8.33 1.69(51) 2.10(44) 1.36(55) 1.85(49) 2.10(44) 1.65(51) 1.45(54) 3.33(37) 1.40(55) 1.17(60) 2.62(37) 1.85(49) 1.61(61) 1.31(57) 3.75(37) 0.595(29) 0.638(32) 0.562(25) 0.611(32) 0.638(32) 0.591(28) 0.571(25) 0.769(37) 0.566(26) 0.542(25) 0.693(32) 0.611(32) 0.580(49) 0.556(25) 0.814(32) Note. -- Sloan magnitudes are in the g filter, V magnitudes for the other stars are taken from Simbad. The sources of the temperatures, gravities and masses are discussed in the text. 10 f1.ps Fig. 1. -- Sample Fourier transform of GD66 from a single 6 hour run. The larger amplitude modes are labeled with their periods. The peaks at 271 and 198 seconds are composed of triplets of closely spaced modes separated by approximately 6.4 µHz that are not resolved in this data Fig. 2. -- O-C diagram of the 302s mode of GD66. The solid line is a sinusoidal fit to the data. f2.ps 11 f3.ps Fig. 3. -- Amplitude of the 302s mode of GD66 as a function of time. The error bars are the formal errors of a non-linear least squares fit, the systematic error is approximately 0.1%. An unstable amplitude would indicate that the observed phase variations are due to some process internal to the star, however the amplitude is stable within our ability to measure it f4.ps Fig. 4. -- Left: O-C diagram of G117-B15A. Each point represents the phase of pulsation based on a single night of data. The solid line is a parabolic fit to the data. Right: Relief map of the region of parameter space around G117-B15A where we are sensitive to planets. Dark regions indicate a high probability a planet would have been detected. The large white circles indicate the location of Jupiter and Saturn, while the small circles mark the postions of known extra-solar planets. 12 f5.ps Fig. 5. -- Same as Figure 4, except for a different star, G185-32. In this O-C diagram, the points indicate the phase of pulsation measured from the combination of data taken on this star over a month or more as described in the text. Fig. 6. -- G238-53 f6.ps 13 Fig. 7. -- GD244 Fig. 8. -- R548 f7.ps f8 14 Fig. 9. -- SDSS J001836.11+003151.1 Fig. 10. -- SDSS J011100.63+001807.2 f9 f10 15 Fig. 11. -- SDSS J021406.78−082318.4 Fig. 12. -- SDSS J091312.74+403628.8 f11 f12 16 Fig. 13. -- SDSS J101548.01+030648.4 Fig. 14. -- SDSS J135459.88+010819.3 f13 f14 17 Fig. 15. -- SDSS J135531.03+545404.5 Fig. 16. -- SDSS J172428.42+583539.0 f15 f16 18 Fig. 17. -- SDSS J221458.37−002511.7 f17 19
astro-ph/9809046
1
9809
1998-09-04T07:01:24
SBF Distances to Dwarf Elliptical Galaxies in the Sculptor Group
[ "astro-ph" ]
As part of an ongoing search for dwarf elliptical galaxies (dE) in the vicinity of the Local Group, we acquired deep B and R-band images for five dE candidates identified in the Sculptor (Scl) group region. We carried out a surface brightness fluctuation (SBF) analysis on the R-band images to measure the apparent fluctuation magnitude \bar{m}_R for each dE. Using predictions from stellar population synthesis models the galaxy distances were determined. All of these dE candidates turned out to be satellites of Scl group major members. A redshift measurement of the dE candidate ESO294-010 yielded an independent confirmation of its group membership: the [OIII] and H$_\alpha$ emission lines from a small HII region gave a heliocentric velocity of 117(\pm 5) km s-1, in close agreement with the velocity of its parent galaxy NGC 55 (v_\odot=125 km s-1). The precision of the SBF distances (5 to 10%) contributes to delineating the cigar-like distribution of the Scl group members, which extend over distances from 1.7 to 4.4 Mpc and are concentrated in three, possibly four subclumps. The Hubble diagram for nine Scl galaxies, including two of our dEs, exhibits a tight linear velocity--distance relation with a steep slope of 119 km s-1 Mpc-1. The results indicate that gravitational interaction among the Scl group members plays only a minor role in the dynamics of the group. However, the Hubble flow of the entire system appears strongly disturbed by the large masses of our Galaxy and M31 leading to the observed shearing motion. From the distances and velocities of 49 galaxies located in the Local Group and towards the Scl group, we illustrate the continuity of the galaxy distribution which strongly supports the view that the two groups form a single supergalactic structure.
astro-ph
astro-ph
SBF Distances to Dwarf Elliptical Galaxies in the Sculptor Group Mount Stromlo & Siding Spring Observatories, Private Bag, Weston Creek P.O., ACT Helmut Jerjen Astronomical Institute of the University of Basel, Venusstrasse 7, CH-4102 Binningen, 2611, Canberra, Australia and Mount Stromlo & Siding Spring Observatories, Private Bag, Weston Creek P.O., ACT Switzerland Ken C. Freeman 2611, Canberra, Australia and Bruno Binggeli Astronomical Institute of the University of Basel, Venusstrasse 7, CH-4102 Binningen, Switzerland ABSTRACT As part of an ongoing search for dwarf elliptical galaxies (dE) in the vicinity of the Local Group (Jerjen et al. 1998a), we acquired deep B and R-band images for five dE candidates identified on morphological criteria in the Sculptor (Scl) group region. We carried out a surface brightness fluctuation (SBF) analysis on the R-band images to measure the apparent fluctuation magnitude ¯mR for each dE. Using predictions from stellar population synthesis models (Worthey 1994) giving ¯MR values in the narrow range between −1.17 to −1.13, the galaxy distances were determined. All of these dE candidates turned out to be satellites of Scl group major members. A redshift measurement of the dE candidate ESO294-010 yielded an independent confirmation of its group membership: the [O III] and Hα emission lines from a small H II region gave a heliocentric velocity of 117(±5) km s−1, in close agreement with the velocity of its parent galaxy NGC 55 (v⊙ = 125 km s−1). The precision of the SBF distances (5 to 10%) contributes to delineating the cigar-like distribution of the Scl group members, which extend over distances from 1.7 to 4.4 Mpc and are concentrated in three, possibly four subclumps. The Hubble diagram for nine Scl galaxies, including two of our dEs, exhibits a tight linear velocity -- distance relation with a steep slope of 119 km s−1 Mpc−1. The results indicate that gravitational interaction -- 2 -- among the Scl group members plays only a minor role in the dynamics of the group. However, the Hubble flow of the entire system appears strongly disturbed by the large masses of our Galaxy and M31 leading to the observed shearing motion. From the distances and velocities of 49 galaxies located in the Local Group and towards the Scl group, we illustrate the continuity of the galaxy distribution which strongly supports the view that the two groups form a single supergalactic structure. Subject headings: galaxies: clusters: individual (Sculptor Group) -- galaxies: distances and redshifts -- galaxies: dwarf -- galaxies: elliptical -- galaxies: individual (ESO294-010, ESO540-030, ESO540-032, NGC 59) 1. INTRODUCTION It is notoriously difficult to determine distances to dwarf elliptical galaxies (hereafter dEs, subsuming "dwarf spheroidals", cf. Ferguson & Binggeli 1994). Their low gas content rules out HI 21 cm-line observations, and their low surface brightness makes optical spectroscopy feasible only for the very brightest members of the class. Most of the available redshifts of dEs are for galaxies in the Virgo cluster (Binggeli et al. 1993) and the Centaurus cluster (Stein et al. 1997), and reflect peculiar velocities rather than individual distances within the cluster. But in clusters there is no real need for individual distances: dEs are so abundant and strongly clustered (see, e.g., Binggeli et al. 1987 for Virgo, or Thompson & Gregory 1993 for Coma) that they must lie near the mean distance of the cluster. In poor groups and the field, where the distribution of dEs is sparse, individual distance information is absolutely indispensable. (We recall that true "field" (i.e. are apparently very rare: cf. Binggeli et al. 1990). Here the only way to locate a diffuse, dwarf-like system and thus to unveil its physical nature is to estimate its distance. isolated) dEs For nearby dE candidates, it is possible to resolve the stellar population and derive the distance from a deep colour-magnitude diagram (CMD, e.g. Da Costa et al. 1996, Smecker-Hane et al. 1996, Stetson 1997), or from the tip of the red giant branch (TRGB, Lee et al. 1993, Caldwell et al. 1998). These methods are costly and time consuming. With HST, the CMD method is feasible out to D ≈ 4 Mpc, and the TRGB method out to D ≈ 12 Mpc. Alternatively, the surface brightness-magnitude relation of dEs (e.g. Binggeli & Cameron 1993, Jerjen & Binggeli 1997) can be used to estimate the distance of a dE out to perhaps 50 Mpc. However, the uncertainty (1σ) in the distance modulus for a single galaxy is large, typically ± 0.7 mag, corresponding to a 40% distance error. Young & Currie -- 3 -- (1995) claimed that this error can be reduced to ± 0.4 mag (comparable to the accuracy of the Tully-Fisher method) with a similar relation based on the shape of the dE luminosity profile, but this claim appears to be incorrect (Binggeli & Jerjen 1998). Here we show that the well-known surface brightness fluctuation (SBF) method can be used to determine distances to dwarf ellipticals in the intermediate distance range of 1 -- 10 Mpc (and possibly beyond) with an accuracy of a few percent. The SBF method was introduced by Tonry & Schneider (1988, hereafter TS88) to measure distances to bright elliptical (E) galaxies. The method is based on the discrete sampling of a galaxy image with a CCD detector and the analysis of the resulting pixel-to-pixel variance caused by unresolved stars. For a detailed description of the method the reader is referred to TS88 and Jacoby et al.(1992). Application of the SBF method was subsequently extended to bulges of spiral galaxies (Tonry 1991, Luppino & Tonry 1993) and to globular clusters (Ajhar & Tonry 1993). A massive SBF observing programme is in progress to improve the cosmic distance scale (Tonry et al. 1997). The first, and up to now only, application of the method to dwarf ellipticals is due to Bothun et al. (1991) who observed the fluctuations in very large, nearly flat, low-surface brightness galaxies (a special type of dEs) in the Fornax cluster of galaxies. We show here that the SBF method is particularly well suited for dEs at a distance of a few Mpc -- slightly beyond the distance where the galaxies would be resolved into stars. In this paper we present a pilot SBF analysis for five dwarf ellipticals in the nearby Sculptor (Scl) group, also known as the South Polar group. These galaxies have recently been identified in an extensive survey of the southern Scl and Centaurus A (Cen A) groups for faint dE members (Jerjen et al. 1998a: JBF98a). Special care has been taken with the calibration of the fluctuation magnitude ¯MR, because the stellar population of dwarf ellipticals is quite distinct from that of the brighter ellipticals. In fact, even among the local dwarf spheroidals there is a wide range of star formation histories. Nevertheless, we find that ¯MR is sufficiently robust against such variations, at least for stellar systems dominated by old populations, and propose a first estimate of ¯MR for dEs. We will show that the SBF distances derived for the five dEs (between 1.7 and 4.4 Mpc), unambiguously identify these objects as Sculptor group members. Moreover, the dwarfs neatly fall into place with the known substructure of the group. We demonstrate the great potential of the SBF method applied to dEs as a tool to map the supergalactic structure in the local volume (out to the Virgo cluster). With our enlarged Scl sample, we re-analyse the 3D distribution of these galaxies and confirm earlier findings (Tully & Fisher 1987, Binggeli 1989) that the Sculptor complex has a prolate structure and is part of a much larger cloud of galaxies which includes the Local Group (LG). The SBF results for the -- 4 -- dE sample from our Cen A group survey will be published elsewhere (Jerjen et al. 1998b: JFB98b). 2. SAMPLE Following Cot´e et al.'s (1997, hereafter C97) successful search of the Scl group region for faint new dwarf irregular members, we conducted a similar visual search for faint dE candidates using the same plate material. Seven very diffuse objects were identified which had dE-like morphology and which were thus suspected to be dwarf spheroidal members of the Scl group (for the details of this survey, see JBF98a). Subsequent CCD imaging in the B and R bands showed two of these objects to be irregulars or background galaxies. The five remaining galaxies of early-type morphology are the subject of the present SBF analysis. They are listed in Table 1 with their basic photometric properties from JBF98a. The photometric parameters span a wide range in total R magnitude (11.9 < RT < 17.0, column 5), effective radius which contains half of the total light (19′′ < ref f,R < 26′′, column 6), and mean surface brightness within the effective radius (21.0 < hµief f,R < 25.4, column 7). To get the extinction-corrected colour (B − R)0 T (column 10) the foreground reddening E(B − V ) was determined from the dust maps of Schlegel et al. (1998) and converted with the ratio AB : AR : E(B − V ) = 4.315 : 2.673 : 1. Further details about the photometric data can be found in JBF98a. Images of the dwarfs are shown in Figs. 1 and 2, and the classifications given in column 2 of Table 1. There is some morphological variety among early-type dwarfs (cf. the atlas of Virgo cluster dwarfs by Sandage & Binggeli 1984). At the bright end of the luminosity function, there is a variant of the pure dE which was named dS0, because it is often distinguished by a S0-like, two-component structure (see also Binggeli & Cameron 1991). NGC 59 is clearly of this type, ESO294-010 arguably so. At fainter magnitudes, dEs (dwarf spheroidals) are often hard to distinguish from smooth irregulars of type Im V. In fact, there may be a true evolutionary transition between gas-rich irregulars and gas-poor dEs (e.g. Ferguson & Binggeli 1994). Dwarfs with a mixed morphology (appearing too smooth for a plain Im, but too lumpy for a pure dE) were called "intermediate" or "ambiguous" and classified dE/Im (Sandage & Binggeli 1984). A well-known local example is the Phoenix dwarf system (Van de Rydt et al. 1991). Three of our objects seem to fall in this class and are classified accordingly. ESO540-030 and ESO540-032 show a sprinkle of resolved stars in their central regions. These stars may mark the presence of a small population of young stars. ESO294-010, on the other hand, -- 5 -- Fig. 1. -- R-band images of four Sculptor group dwarfs made from the combination of 4 -- 5 exposures with a total integration time between 1800 and 3000 seconds. The areas depicted are 3 arcmin on a side. From left to right and top to bottom the galaxies are: NGC 59, Scl-dE1, ESO540-030, and ESO540-032. -- 6 -- Table 1. Galaxy properties Galaxy Type (1) (2) RA (2000) (3) DEC (2000) (4) NGC 59 Scl-dE1 (SC221) ESO294-010 ESO540-030 ESO540-032 00 15 25.1 −21 26 38 dS0 dE 00 23 51.7 −24 42 18 dS0/Im 00 26 33.4 −41 51 19 dE/Im 00 49 21.1 −18 04 34 00 50 24.5 −19 54 23 dE/Im RT mag (5) 11.90 16.94 14.36 15.54 15.36 ref f,R arcsec (6) 25.9 19.3 20.3 21.5 25.0 hµief f,R (B − R)T mag arcsec−2 (7) 21.0 25.4 22.9 24.2 24.4 mag (8) 1.07 0.79 1.17 0.83 1.08 AB mag (9) 0.09 0.06 0.03 0.10 0.09 (B − R)0 T mag (10) 1.04 0.77 1.16 0.79 1.05 1Name from Cot´e et al. (1997) Fig. 2. -- B and R-band images of ESO294-010 made from the combination of 5 exposures with total integration times of 2500 and 2250 seconds, respectively. The angular frame size is 3 arcmin. The possible H II region is the bright feature 18 arcsec south of the centre of the galaxy. -- 7 -- appears slightly lumpy in the centre (see Fig.2). Subsequent spectroscopy (in Sec.6) showed emission lines, so this is a clear example of a mixed type. But even the smoothest dEs probably contain small amounts of intermediate-age or young stars; this we know from the local dwarfs (cf. Sec.5). Note also that the colours of our objects (with 0.75 < B − R < 1.04, cf. column 10 of Table 1) are rather blue for dEs, whose B − R values range from 0.5 to 2.3 (e.g., Evans et al. 1990). So the morphological deviations from a pure dE are simply the manifestation of an underlying variety in the (unresolved) stellar populations. One of the principal results of the present paper is that this variety does not cause difficulties for the application of the SBF method to dEs, because the fluctuation magnitude ¯MR turns out to be rather insensitive to variations in the stellar contents within the observational constraints (Sec.5). In the following we will use the term "dE" in a loose way, including all of the morphological variants discussed above. 3. Imaging The galaxies were observed in two runs, on September 11 -- 15, 1996, and August 28 -- 30, 1997, using the imager at the Nasmyth B focus of the 2.3m ANU telescope at Siding Spring Observatory. The detector was a Tek 1k×1k thinned CCD with a pixel size of 24µm, yielding a scale of 0.6′′ pixel−1 and a 6.7′ diameter circle field of view. The CCD gain was 1 e−/ADU and the readout noise was 7 e−. We observed in the Kron-Cousins R-band, because our application of the SBF technique differs from previous studies which used I or K; we are working on small, low-surface brightness objects so our data are photon-limited by the sky rather than by the galaxy itself. The sky is significantly fainter in R than in I relative to the RGB stars which produce the surface brightness fluctuations [(R − I)sky=1.5 versus (R − I)RGB=0.8]. Another advantage of the R filter is to avoid fringing which occur with thinned CCDs beyond 7000A. Weather conditions were photometric and the seeing measured between 1.2 − 1.5′′. For each of our galaxies a series of 4 to 5 exposures were taken each randomly offset by ∼ 10′′ and of 450 to 600 sec duration, giving a total integration time of 1800 to 3000 sec. Flat fields were obtained every night from exposures of the morning and evening sky. Standard stars of Graham (1982) were observed throughout each night for the photometric calibration. Processing of the CCD frames with IRAF was carried out in the usual manner. The overscan region provided the bias level, which was subtracted from the frame. Next, each science frame was flattened with a master twilight-sky flat, constructed from a median of 5 -- 9 sky flats taken during the same night. All reduced frames were flat to 0.1%. The sky was modelled by fitting a plane to selected star-free areas well away from the galaxy. While -- 8 -- sky determinations can be quite difficult for giant ellipticals due to their extended halos and galaxy sizes comparable to the field of view, this is simple in our case. Finally, the sky subtracted images of a galaxy were registered and combined to a master frame in order to increase the signal-to-noise ratio and to remove artifacts, bad pixels, and cosmic rays. In Fig.1 we show the processed R-band images for four of our galaxies. B and R-band images of ESO294-010 are shown in Fig.2. 4. SBF ANALYSIS To prepare the master frames for the SBF analysis we essentially followed the procedure described in TS88. First, we employed DAOPHOT (Stetson 1987) routines to identify point sources which would disturb a 2D-fit of the mean galaxy light distribution. Because of the small size of the galaxies and the high galactic latitude, there is only little contamination by foreground stars and background galaxies. Globular clusters are also a minor problem here, because globular clusters are rare in faint dwarf galaxies. However, as mentioned in Sec.2, two of our galaxies, classified as dE/Im, are contaminated with a few bright stars of their own. These also had to be removed. All identified point sources were replaced by a nearby patch of galaxy of the same surface brightness. Isophotes were computed for the cleaned galaxy image by fitting ellipses with variable radius, ellipticity, and position angle. These were used to model the mean galaxy surface brightness distribution, which was then subtracted from the original master frame. To normalise the variance amplitude of the stellar fluctuations across the residual image, we divided the residual image by the square root of the mean galaxy surface brightness distribution. For each galaxy the SBF analysis was carried out on two different square subimages (field 1 and 2) with size between 30 to 60 pixels which were selected within the 27.5 R mag arcsec−2 isophote. The overlap of the fields was kept as small as possible (< 10%) to get independent distance measurements. Special care was further taken to choose subimages with only few (< 10) resolved point sources, galaxies, or other features DAOPHOT formerly had identified. For example, the H II region of ESO294-010 (see Sec.6 for more details) was excised from the analysis. Remaining objects in the subimages were replaced by uncontaminated patches (i) randomly selected from the region outside of the subimage area and (ii) in the same surface brightness range. This patching process was employed as alternative to the masking method described in TS88. The idea was to replace the few disturbed image parts with patches carrying the fluctuation signal. As artificial periodicities and poor signal-to-noise data could be introduced in this way it is crucial to take into -- 9 -- account criteria (i) and (ii). The number of pixels corrected never exceeded 5% of the total subimage area. Experiments showed that, at this percentage level, the patching method is uncritical for the signal we intend to measure. At this stage of the procedure, we have images whose main sources of variance are the stellar fluctuations and a photon shot noise that is nearly uniform over the image (the CCD readout noise is negligible). To disentangle the two components we Fourier-transformed the cleaned subimages and analysed their power spectra (PS), which are shown in Fig. 3. Having calculated the PS of the Point Spread Function (PSF) from well isolated stars on the master frame, the observed PS of the galaxy was modelled as a weighted combination of a PSF-convolved component and a constant white-noise component: PS(galaxy) = P0 · PS(PSF) + P1 (1) The free parameters P0 and P1 (the two weights) were determined by a least-squares fit of Equ.1 to the data. The fit was restricted to wave numbers k > 4 to exclude the range which is affected by the galaxy subtraction. The result of this decomposition for every galaxy field is shown in Fig. 3. The quantity P0 is directly linked to the apparent fluctuation magnitude ¯mR which is the luminosity-weighted average stellar luminosity of the stellar population and is approximately the apparent magnitude of a giant star in the galaxy which contributes to the SBF signal: The system (telescope-filter-detector)-related quantity m1 is the magnitude of a star yielding 1 ADU per second on the CCD, and t is the exposure time of the observation. ¯mR = m1 − 2.5 · log(P0/t) . (2) From theory (TS88), we expect P1 to be the variance of the photon shot noise σph divided by the square root of the number n of exposures: P1 = (1 + s/¯g)/√n , (3) where s is the flux from the sky (hµsky,Ri = 20.8 mag arcsec−2 at new moon) and ¯g is the mean surface brightness of the galaxy on the subimage (in counts). That Equ.3 is approximately fulfilled can easily be verified from Table 2 where all relevant parameters are listed. The sky brightness of individual images varied a few percent due to changes in the airglow intensity during the night. Column 6 gives the average sky value we measured for each series of images. The indicated error corresponds to the uncertainty in the sky flatness of the masterimage. -- 10 -- Fig. 3. -- Power spectra of the Fourier transform for the R subimages analysed. There are two fields for every dwarf, denoted F1 and F2. The spectra are azimuthally averaged. The points are the data. The full lines are best-fitting combinations of a PSF-convolved stellar fluctuation component and a constant photon noise component, as modelled by Equ.1. The separate contributions are shown as broken lines. -- 11 -- The error of a P0 measurement is dominated by the fitting error and the error in the sky value. While the first accounts for 5 -- 9% in all cases, the latter effect depends on ¯g going from a negligible 1% for the fields of NGC 59 to 18% for ESO540-032 where the galaxy counts are few. Other possible sources of errors are the PSF normalisation and the shape variation of the stellar PSF over the CCD area. Using different bright stars as template for the PSF power spectrum we found the derived P0 of a field changed by only 1 -- 3%. Column 7 of Table 2 gives the values of P0 together with the combined error from all discussed sources. Assuming further an accuracy in photometry to ∆m1 = 0.02 mag (column 2) and a random extinction error at the South Galactic Pole of ∆AR = 0.01 mag, we get an overall uncertainty for ¯m0 each galaxy derived from the two independent fluctuation magitude measurements are given in column 2 of Table 3. R of 0.06 to 0.19 mag (column 11). The error-weighted mean of ¯m0 R for 5. ¯M CALIBRATION AND SBF DISTANCES We chose to use R-band observations (instead of the usual I or K) to avoid fringing effects on thinned CCDs and to take advantage of the relatively darker sky. However, the drawback with this filter at present is the missing empirical calibration of ¯M in this photometric band. Most of the SBF applications have focussed on the I-band (e.g. Tonry et al. 1989, 1990; Tonry 1991; Tonry et al. 1997) and K-band (e.g. Luppino & Tonry 1993; Pahre & Mould 1994; Jensen et al. 1996). But even working in the I-band would not help in our case because all galaxies analysed to date are high-surface brightness, giant ellipticals, while our objects are low-surface brightness, dwarf galaxies. The two classes are quite different in almost all of their properties, including stellar composition, although these differences are by no means well understood (e.g., Ferguson & Binggeli 1994). In particular, ¯M strongly depends on the underlying stellar population of a galaxy, and it would be unclear how to adapt empirical results from E galaxies to dEs. To bypass this problem, we constructed a mean ¯MR for dEs by applying stellar population synthesis models to the local, resolved dwarf spheroidals. A recent comparison between theory and empirical results (Tonry et al. 1997) found very good agreement in the I-band when using the model predictions of Worthey (1993a, 1993b, 1994). We therefore used Worthey's on-line program1 to calculate ¯M values for various photometric bands and different stellar mixtures composed of a series of single-burst stellar populations. For this procedure each population requires a set of input parameters: mean metal abundance, age, 1http://astro.sau.edu/∼ worthey/ -- 12 -- Table 2. Parameters of the SBF Analysis Galaxy (1) m1 mag (2) exp time sec (3) seeing arcsec (4) ¯g counts (5) s(∆s) counts (6) P0(∆P0) P1(∆P1) ¯mR(∆ ¯m) counts (7) counts (8) mag (9) NGC 59 F1 F2 Scl-dE1 F1 F2 ESO294-010 F1 F2 ESO540-030 F1 F2 ESO540-032 F1 F2 24.30 5×450 24.31 5×600 24.30 4×450 24.32 5×450 24.27 5×450 1.4 1.5 1.5 1.5 1.3 4166(8) 4456(11) 3572(7) 3769(6) 4146(17) 836 706 101 77 329 234 193 96 123 95 34.9(1.7) 33.3(1.8) 131.1(16.5) 118.8(19.7) 234.3(15.1) 214.0(13.8) 62.0(4.4) 65.8(7.4) 112.9(17.0) 151.0(28.2) 2.7(0.1) 3.5(0.1) 18.5(0.4) 26.5(0.5) 5.9(0.4) 10.7(0.5) 9.7(0.3) 24.5(1.0) 12.8(0.5) 16.3(0.6) 27.08(0.06) 27.12(0.06) 25.96(0.13) 26.08(0.17) 25.00(0.07) 25.11(0.07) 26.47(0.07) 26.41(0.11) 25.77(0.15) 25.46(0.19) AR mag (10) 0.06 0.04 0.02 0.06 0.06 ¯m0 R(∆ ¯m) mag (11) 27.02(0.06) 27.06(0.06) 25.92(0.13) 26.04(0.17) 24.98(0.07) 25.09(0.07) 26.41(0.07) 26.35(0.11) 25.71(0.15) 25.40(0.19) Table 3. Mean observed ¯mR values with resulting distances and absolute magnitudes Galaxy (1) ¯m0 R mag (2) (m − M)0 mag (3) D Mpc (4) M 0 B mag (5) M 0 R mag (6) NGC 59 Scl-dE1 ESO294-010 ESO540-030 ESO540-032 27.04(0.04) 25.96(0.10) 25.04(0.05) 26.39(0.06) 25.59(0.12) 28.21(0.07) 27.13(0.12) 26.17(0.08) 27.52(0.08) 26.72(0.13) 4.39(0.15) −15.30 −16.34 2.67(0.16) −9.50 −10.25 1.71(0.07) −10.67 −11.83 3.19(0.13) −11.22 −12.02 2.21(0.14) −10.39 −11.42 -- 13 -- relative mass weight, and the slope of the IMF. To use the program, we first had to estimate the metal abundance of the dEs. For this we used the observed [Fe/H] -- MV relation of Local Group dEs (Da Costa 1994, 1998). Absolute visual magnitudes for our dwarfs were estimated by combining the extinction-corrected observed B magnitudes, the mean (B − V ) colour of 0.75 (Bothun et al. 1989), and an average distance modulus (m -- M)=27.0 for the Sculptor group (C97). The resulting magnitudes, in the range −14.8 < MV < −10.1, were converted into abundances according to the relation [Fe/H]= −0.15·MV − 3.45 (derived from Fig.7 of Da Costa 1998). As expected, our dEs are all metal poor systems, with metallicities in the range −1.9 <[Fe/H]< −1.2. As emphasised by Da Costa (1997), dEs are not single-burst populations like a globular cluster. Rather, a diverse and complex set of star formation histories (SFH) is observed among the local dwarf spheroidals. Their stellar populations range from old (Ursa Minor) and mainly old (e.g. Tucana, Leo II) through intermediate-age episodic (e.g. Carina, Leo I) to intermediate-age continuous (e.g. Fornax). On the other hand, Phoenix and LSG3 are classified as dE/Im, because they shows similarities to both dwarf spheroidals and dwarf irregulars. These systems are dominated by an old metal-poor population with no evidence for major star formation activities after the initial episode 8 -- 10 Gyr ago. However, both systems have a minor population of young stars, with ages of about 150 Myr, which makes these galaxies resemble dwarf irregulars. This situation would make us suspect that ¯MR for these galaxies is also diverse. Fortunately, this is not the case. We will show that ¯MR does not depend sensitively on the star formation history within the observed range. This issue will be discussed further in JFB98b. We used Worthey's program to calculate ¯MR values for a metal poor old (age > 8.5 Gyr) population with three different types of SFHs (single burst, episodic, continuous). Seven SFHs were modelled with a series of star formation bursts, as given in Table 4. The bursts were taken to occur at equal intervals of time between the first and the last burst (columns 2 -- 4 of Table 4). The adopted initial metal abundance was −1.9 or −2.0. The enrichment per burst was taken to be constant (column 6), where the value of this constant was chosen such that the present-day abundances of the model spanned the range of abundances derived above for our galaxies. A Salpeter IMF was adopted throughout. The relative mass weight per burst was taken to be constant. The resulting ¯MR values are listed in column 9. The modelling reveals ¯MR to be essentially insensitive to the changes in SFH for these populations older than 8 Gyr. All ¯M magnitudes are found in the narrow interval −1.207 < ¯MR < −1.136 with a median value ¯MR = −1.165. This value can be expected to -- 14 -- be a good calibration constant for metal poor old populations as observed in Ursa Minor, Tucana, or NGC 205. If we assume that the morphological similarities between these LG dwarfs and Scl-dE1 and NGC 59 indicate similar star formation histories, then the median value from the models can be adopted for our two sample galaxies. The remaining three Scl dwarfs show some indication of a small population of young stars as found in Phoenix. To estimate ¯MR for these dwarfs, we analysed the old populations (column 1, Table 4) but polluted with a 5 and 10 percent (in mass) metal-rich intermediate and young population. This second component was introduced in two different ways: (1) a constant star formation rate between 1 and 8 Gyr simulated with a series of bursts every 1 Gyr, (2) an episodic component with two bursts at 3 and 6 Gyr. Each burst got equal mass weight and a metallicity of [Fe/H]= −0.225 (this is the lowest metallicity allowed in Worthey's program for bursts younger than 8 Gyr). As in the pure "old" cases, ¯MR is insensitive to the underlying star formation history of the pollution, i.e. continuous or episodical. The important factor which changes ¯MR (see Fig.4) is the mass ratio Myoung/Mold of the two subpopulations. Applying equal weight to the results of all seven considered SFHs yields an average empirical relation ¯MR = −1.165 + 0.70 · (Myoung/Mold). For ESO294-010, ESO540-030, and ESO540-032, we estimate the contribution of the young stars to the total integrated light to be less than 3%. The typical mass-to-light ratio for the generated synthetic stellar systems is about 1.5 in the R-band, so we consider an old population with a young star pollution at the 5 percent level as a fair approximation for these three dwarfs. To determine their distances, we will use the mean value from our models ¯MR = −1.13. We are very grateful for the use of Worthey's public program which enabled the work on ¯MR described above. We note however that star formation histories of dEs like Fornax or Carina could not be explored with this program. Both dwarfs produced a significant fraction of their metal poor ([Fe/H]< −1.0) intermediate population between 4 and 8 Gyr (Da Costa 1997). This part of the parameter space, i.e. young (< 8 Gyr) and metal poor ([Fe/H]< −0.225), is not accepted by the model program at present. As an independent, zero-order check on the intrinsic scatter of ¯MR for an old metal poor stellar population, we explored the ¯M -- [Fe/H] relation for globular clusters (GC) in V and I (Ajhar & Tonry 1994) for the relevant metallicity range. Our primary interest here is not with the absolute values but the observed intrinsic dispersions. Using all 11 GCs in their sample with −1.9 <[Fe/H]< −1.2, we derived a mean scatter of σV = 0.05 and σI = 0.07, respectively. The exclusion of the anomalous cluster ω Cen reduces the scatter in the I band to 0.06. Assuming that the results for the R band are comparable, we conclude that (1) theory and observations predict a very small variation of ¯M in this metallicity -- 15 -- Table 4. Synthetic star formation histories of pure old populations and ¯MR values. Initial [Fe/H] ∆[Fe/H] Final [Fe/H] Weight SFH (1) Start (Gyr) (2) Step (Gyr) (3) End (Gyr) (4) Burst13 Burst11 Burst9 Episodical-1.9 Episodical-1.2 Continuous-1.9 Continuous-1.2 13 11 9 13 13 13 13 0 0 0 1.5 1.5 0.5 0.5 13 11 9 8.5 8.5 8.5 8.5 −1.9 −1.9 −1.9 −2.0 −2.0 −2.0 −2.0 (5) (6) 0 0 0 0.033 0.267 0.011 0.089 (7) −1.9 −1.9 −1.9 −1.9 −1.2 −1.9 −1.2 ¯MR (mag) (9) (%) (8) 100 −1.136 100 −1.165 100 −1.207 25 −1.179 25 −1.138 10 −1.177 10 −1.142 Fig. 4. -- The model ¯MR values for a mainly old (>8.5 Gyr) stellar population show a systematic trend to fainter magnitudes with an increasing portion of young stars. Thereby the slope is independent whether the star formation history of the old population is a single burst event (filled circle), continuous or episodical (open symbols) as listed in column 9, Table 4. The solid line indicates the average linear relation between the mass ratio and ¯MR for our models. -- 16 -- range, (2) the empirical scatter depends only weakly on the photometric passband. For the following discussion, we will adopt ¯MR = −1.17 (model) for our "dE" classified dwarfs and ¯MR = −1.13 (model) for the intermediate type "dE/Im". In both cases an error of ∆ ¯MR = 0.06 (from GCs) is adopted. We regard this as a preliminary answer in the search for the empirical calibration of the SBF distance indicator for dEs. Interestingly, the results from the models are right between the average values of ¯MI and ¯MV for metal poor GCs with −2.03 and −0.33, respectively (Ajhar & Tonry 1994). The distances to our galaxies can now be determined in the usual manner: log(D[Mpc]) = 0.2 · ( ¯m0 R − ¯MR − AR − 25). (4) The true (extinction-corrected) distance moduli, the distances in Mpc, and the resulting absolute total magnitudes in R and B for the five Scl dwarfs are listed in Table 3. Also given there are the uncertainties in these quantities. We recall here the detailed origin of these errors: (∆P0/P0) = 0.06 − 0.18 (fitting, sky, PSF variation), ∆m1 = 0.02 mag (calibration), ∆ ¯MR = 0.06 mag (modelling), and ∆AR = 0.01 mag. The total SBF distance error for a single measurement is 5 to 10%. The new SBF distances and their relevance for the 3D-structure of the Sculptor group are discussed in Sec.7 below. First we report on a spectroscopic confirmation of the distance for one dwarf. 6. SPECTROSCOPY For an independent proof of group membership for at least one of our dEs, we selected our second brightest galaxy ESO294-010 for a spectroscopic follow-up. This galaxy was recently part of a study on the star formation histories of Sculptor group dwarf galaxies (Miller 1996). Miller found that ESO294-010 is a typical early-type dwarf with no evidence for ongoing massive star formation. No H II regions were discovered and the total Hα luminosity is very low. Nevertheless, with a sufficiently long integration time we hoped to detect Balmer absorption lines in the spectrum which then could be used to measure a redshift. This galaxy is even more interesting because C97 reported a 2.5σ detection in H I at a redshift of 4450 km s−1 but we measured the galaxy to be only 2.4 times more distant than M31. Long-slit spectroscopy was carried out for ESO294-010 in September 1997. We used the double-beam spectrograph (DBS) at the Nasmyth A focus of the 2.3m ANU telescope to observe a blue and a red spectrum simultaneously. The detectors were two SITe 1752×532 -- 17 -- thinned CCDs with a pixel size of 15µm, and a scale of 0.9′′ pixel−1 across the dispersion. The wavelength scales were 1.1A pixel−1 in the blue and 0.55A pixel−1 in the red. Beam splitter and grating angles were chosen to cover the wavelength ranges 3500 -- 5500A and 6000 -- 7000A, respectively. The slit was positioned at the galaxy centre and aligned along the major axis. The slit length was 6.7 arcmin with a width of 2′′. A series of four 2000 sec exposures were taken. After each science exposure a Ne-Ar lamp was observed for the wavelength calibration. IRAF procedures were used for the spectral reduction. The galaxy spectra were individually wavelength-calibrated and sky subtracted, and each set of blue and red spectra were combined to improve the signal-to-noise ratio. Fig. 5 shows slightly smoothed versions of the important parts of the spectra. No significant Balmer absorption lines are visible. However, narrow emission lines at [O III] λ5007A and Hα λ6563A are prominent with 5 and 10σ significance relative to the adjacent continuum rms variations (the two apparent [S II] lines λ6716A and λ6731A are not significant). In view of Miller's (1996) results mentioned above, this finding was unexpected, even though modest amounts of dust and gas in dEs are not uncommon, e.g. in NGC 185 and NGC 205 (Hodge 1973, Young & Lo 1997 and references therein), or NGC 59 (C97). In fact, there is a hint of lumpiness in the central part of ESO294-010, clearly seen in Fig. 2, which made us classify this galaxy a posteriori as an intermediate-type dS0/Im (see Sec.2). Tracking down the source of the emission, we noticed a circular patch 7 arcsec or 0.06 kpc across, located 18 arcsec south of the galaxy centre (Fig. 2). This patch coincides in position and extent with the position and width of the two emission lines along the spatial axis of the spectra, and is probably a small H II region. The agreement of the velocities derived from the individual lines is excellent, with v⊙ = 117(±0.8)km s−1. We suggest that 5km s−1 is a more realistic estimate of the velocity error. This low velocity is in accord with our low SBF distance for that galaxy, and it will be shown below that it fits nicely into the velocity pattern of other Scl group members. 7. 3D-STRUCTURE OF THE SCULPTOR GROUP The most important result of our SBF analysis is that all five objects fall in the distance range of previously known Scl group members, thus confirming their suspected membership as group dwarfs. Moreover, the new dE members help to clarify our view of the complex 3D-structure of the group. As it will turn out in the following, the notion of a Scl "cloud" of galaxies would be a better description of this supergalactic structure. -- 18 -- Fig. 5. -- Smoothed versions of the blue (top) and red (bottom) spectrum of ESO294-010 with a total integration time of 8,000 sec. The centrally located [O III] and Hα emission lines are the only significant features in the spectra. -- 19 -- The main members of the Scl group are the six late-type (Sc, Sd, Sm) spirals NGC 45, 55, 247, 253, 300, and 7793. Individual distances to these galaxies, determined from various methods, are listed in Table 5, column 2. The rms uncertainties of the distances are given in parentheses. Most data are taken from Puche & Carignan (1988, Table II). Where more recent measurements are available, we have calculated a mean value from the references given in column 3. One Scl dwarf, the irregular SDIG, which already had a known distance, is also included here. To this list we can now add our 5 Scl early-type dwarfs with their newly determined SBF distances. Velocities (where available) are also given in Table 5. Heliocentric velocities (column 4) are taken from C97, Da Costa et al. (1991), and Lauberts & Valentijn (1989) for all but one galaxy (ESO294-010) for which the velocity was established in the present study (see Sec.6). These velocities were transformed into galactocentric velocities (column 5) and velocities relative to the barycentre of the LG (column 6), applying the apex vectors given by de Vaucouleurs et al. (1991) and Karachentsev & Macharov (1996), respectively. Fig. 6 is a pseudo-3D plot of the distribution of the 12 Scl galaxies with known distances listed in Table 5: it shows the sky distribution of these galaxies, with the distance of every galaxy indicated by the symbol size (the distances and velocities are also given as numbers). We first note the large depth spanned by the galaxies: distances range from 1.7 Mpc (NGC 55) to 4.4 Mpc (NGC 45). In fact, Puche & Carignan (1988) excluded the distant NGC 45 from their study because they insisted on a gravitationally bound Scl group with a suitably low mass-to-light ratio. Here we relax any dynamical requirements because, judged from the distribution of galaxies, such a cut seems artificial. This will become clearer below. We further note the close match in distance of ESO294-010 with NGC 55 at the near end (1.7 Mpc), and of NGC 59 with NGC 45 at the far end (4.4 Mpc) of the Scl complex (despite the large distance error for NGC 45). It seems very likely that the two dwarfs are bound companions to the major members NGC 55 and 45. Both pairs are separated by less than 200 kpc in projection and have a radial velocity difference of less than 100 km s−1. ESO540-030, ESO540-032, and Scl-dE1, on the other hand, seem to be associated with NGC 247 and 253 at an intermediate distance of ≈ 2.5 Mpc. ESO540-032 could be bound to NGC 247. The other two dwarfs are too far off for a single companionship to either giant, but a bound substructure centred on NGC 247 and 253 is strongly suggested. The close spatial coincidence of our five dwarf objects with the substructure of the Scl group traced out by its main members demonstrates two things: (1) The SBF method to determine distances to early-type dwarfs is reliable and accurate to the claimed level. (2) The trend of "field" dEs to be satellites of giant galaxies is confirmed, in accord with the morphology-density relation for dwarf galaxies (Binggeli et al. 1990). The only isolated -- 20 -- Table 5. Distances and velocities of galaxies in the Sculptor group region Name D(∆D) Mpc (2) (1) NGC 55 ESO294-010 NGC 300 ESO540-032 NGC 247 Scl-dE1 SDIG NGC 253 ESO540-030 NGC 7793 NGC 45 NGC 59 Ref. (3) v⊙(∆v) km s−1 (4) vGSR km s−1 (5) vLG km s−1 (6) Ref. (7) this study this study 1.66(0.20) PC88 1.71(0.07) 2.10(0.10) Fr92 2.21(0.14) 2.48(0.15) T87, PC88, F98 2.67(0.16) this study 2.63(0.80) L77, H97 2.77(0.13) PC88, F98 3.19(0.13) this study 3.27(0.08) T87, PC88 4.35(1.40) PC88 4.39(0.15) this study 129(3) 117(5) 144(1) 156(4) 229(10) 245(5) 227(2) 463(3) 362(10) 98 71 101 173 217 246 226 488 392 111 DC91 80 114 ESO-LV this study 211 DC91 238 C97 278 DC91 250 ESO-LV 525 DC91 432 C97 References. -- C97: Cot´e et al. 1997; DC91: Da Costa et al. 1991; ESO-LV: Lauberts & Valentijn 1989; F98: Federspiel 1998 (NGC 247 (m -- M)0=27.17, NGC 253: (m -- M)0=27.36); Fr92: Freedman et al. 1992; H97: Heisler et al. 1997; L77: Laustsen et al. 1977; PC88: Puche & Carignan 1988, and references therein; T87: Tammann 1987. -- 21 -- ESO540-030 (3.2) ESO540-032 (2.2) - 20 o N59 (4.4 / 392) N45 (4.4 / 488) N247 (2.5 / 173) N253 (2.8 / 246) Scl-dE1 (2.7) - 30 o ) 0 0 0 2 ( C E D - 40 o N7793 (3.3 / 226) N300 (2.1 / 101) SDIG (2.6 / 217) N55 (1.7 / 98) ESO294-010 (1.7 / 72) 01h 20m 00h 40m RA (2000) 00h 00m Fig. 6. -- Sky distribution of all major Sculptor group members (triangles) plus SDIG (box) and the dEs (circles) with known distances. Names, distances (in Mpc), and galactocentric velocities (if available) are indicated. The symbol size is inversely proportional to the galaxy distance to simulate a depth effect. -- 22 -- pure (old-population) dE known so far is the Tucana system in the LG (Da Costa 1994, see also Fig 9. below). The only galaxy classified here as a pure dE, Scl-dE1, may also be fairly isolated among our dwarfs. So far we have not considered the 14 (!) additional dwarf irregular Scl group members discovered by C97, because of a lack of distance information. However, we can include them in our analysis by using their observed radial velocities as distance indicators via the tight velocity-distance relation which will be discussed below. Dwarf irregulars are known (and below also shown) to avoid high-density regions (cf. again Binggeli et al. 1990) and therefore to possess fairly unperturbed velocities. Fig. 7 is a pseudo-3D plot of the Scl group, like Fig. 6 but now with the galactocentric velocity as third dimension. All C97 irregulars are included here, raising the number of Scl galaxies to 23. We recognize again the NGC 55 subclump on the near side with ESO294-010, and possibly also SC18 as bound companions. NGC 300 and SDIG are slightly more distant. Unfortunately, the three dEs associated with NGC 247 and 253 (Fig. 5) have no velocity and hence are lost here in Fig. 7. DDO006 and 226 with 300 km s−1< vGSR <400 km s−1 must lie behind NGC 247 (173 km s−1). Likewise, ESO473-024 (571 km s−1) is probably behind NGC 45/59. On the other hand, a possible pair of irregulars appears in the SE (lower left) corner with NGC 625 (333 km s−1) and ESO245-005 (314 km s−1). There are some very nearby dwarf irregulars, such as SC24 (61 km s−1) and SC2 (98 km s−1), but also some very distant ones, such as ESO348-009 (626 km s−1) and ESO347-017 (674 km s−1). Overall, the impression is that the dwarf irregulars are not strongly associated with the main Scl group members but are scattered over the entire volume in the form of a "cloud". This is of course again in accord with the morphology-density relation for dwarf galaxies (Binggeli et al. 1990). The whole Scl group seems to have the shape of a cigar of length 3 Mpc and thickness 1 Mpc, which we see pole-on because the LG lies near its end. A much better view of this can be gained if the cigar is turned to the side by plotting the distances linearly. For this we have first to transform velocities to distances via a Hubble diagram. In Fig. 8 we show the Hubble diagram for the total of nine galaxies with known distances and velocities, to which we have contributed ESO294-010 and NGC 59. The velocities (vGSR) are relative to the centre of our Galaxy. The major Scl group members show a tight linear relation between velocity and distance. Such a trend was already seen by Puche & Carignan (1988) and is now confirmed and amplified by the two new data point from the dEs (filled squares). A maximum likelihood linear fit to the data yields: vGSR[km s−1] = 119(±7) · D[Mpc] − 136(±14) , (5) where vGSR is taken to be the independent variable and the distance errors are given in Table 5. -- 23 -- - 20 o - 30 o ) 0 0 0 2 ( C E D - 40 o - 50 o N247 (173) DDO006 (318) DDO226 (385) ESO473-024 (571) N253 (246) N59 (392) N45 (488) SC24 (61) AM0106-382 (594) N300 (101) N625 (333) UGCA442 (286) N7793 (226) SDIG (217) N55 (98) SC2 (98) ESO347-017 (674) ESO348-009 (626) SC18 (118) ESO294-010 (72) ESO293-035 (74) ESO245-005 (314) SC42 (70) 02h 40m 02h 00m 01h 20m 00h 40m 00h 00m 23h 20m RA(2000) Fig. 7. -- Sky distribution of all major group members (triangles), dEs (circles), and dwarf irregulars (boxes) with known velocities. Names and galactocentric velocities are indicated. Symbol size is inversely proportional to the velocity. Also shown is the survey boundary of the dwarf galaxy searches of Cot´e et al. (1997) and JBF98a. -- 24 -- The mean scatter of the residuals about the line is remarkably small, with σD = 0.34 Mpc (0.23 Mpc without NGC 45) and σvGSR = 40 km s−1(27 km s−1without NGC 45). Large peculiar velocities due to internal group dynamics are clearly absent. This strongly suggests that our Scl galaxies form a "cloud" (e.g. de Vaucouleurs 1975, Tully 1982, Tully & Fisher 1987), meaning a supergalactic structure of higher-than-average galaxy density which, in contrast to a "group", is gravitationally unbound. However, bound substructures, e.g. defined by one giant galaxy plus a swarm of dwarf companions, are often found embedded in such a cloud. This also seems to be the case here. The slope of the Hubble relation of Equ.5 is very steep (119 km s−1 Mpc−1). A plausible reason for this deviation from the global Hubble law (H0 ≈ 60± 10 km s−1 Mpc−1) is the gravitational influence of the LG. The deceleration of the local expansion field (for D < 4 Mpc) by the LG was nicely recovered by Sandage (1986, 1987) who used his local galaxy data to put constraints on the mass of the LG. However, an attempt to fit one of Sandage's template curves for the local Hubble diagram (1986, his Fig. 2) to our data (Fig. 8) proved unsuccessful. At present we do not understand the dynamics underlying the Scl cloud velocity field, and a dynamical analysis is beyond the scope of this paper. Because the velocity-distance relation of Equ.5 is so tight, we can use it empirically to obtain approximate distances to those Scl galaxies which have only velocities, i.e. the dwarf irregulars of C97. Based on these distances we show, in Fig. 9, the distribution of all Scl cloud members in projection on to the supergalactic plane (SGP). This provides us with a side view of the cigar-shaped Scl cloud, as it was described above. The scatter of the third dimension here, the height above/below the SGP, is quite small (σz = 0.53 Mpc). We have also included LG "members" and galaxies at the outer fringes of the LG in this plot (data taken from van den Bergh 1994), to demonstrate that the Scl cloud and the LG may actually form one structure. This conjecture was raised before by Binggeli (1989). There is a marked asymmetry in the distribution of galaxies within a distance of 1.5 Mpc. Aside from the close companions to our Galaxy and M31, most LG dwarf members and suspected members (IC1316, WLM, NGC 6822, Phoenix, Tucana, SagDIG, DDO210, Pegasus, IC5152, ...) form a southern extension of the LG -- A bridge across to the Scl cloud (see also Fig.2 in Binggeli 1989). The whole structure can be described as a prolate (6 × 2 Mpc) cloud of (unbound) dwarf irregulars, in which a number of denser groups are embedded, each containing one or more massive galaxies plus a swarm of (bound) dwarf companions which tend to be of early type. Certainly our Galaxy and M31 with their satellites form such condensations -- each alone, and together as the LG. In the Scl region we note again the three subclumps mentioned before: centred on NGC 55 (plus NGC 300?), NGC 247/253, and NGC 45. NGC -- 25 -- 7793 may define a fourth subclump or may be member of the first (Davidge 1998). We note again the position of our five early-type dwarfs in this map: they tend to mark the condensations. The dynamics of these Scl subsystems with respect to each other, and with respect to the LG in particular, is not clear at present. We note that the major Scl group members are much less massive than M31 or our Galaxy, and that M31 and the Galaxy lie at the extreme end of the prolate cloud, but we do not know if this is physically significant. Finally, it is worth mentioning that the Scl-LG cloud is part of a yet larger structure -- a prolate cloud that stretches out to the Coma I cluster at D ≈ 15 Mpc, in the direction of the Virgo cluster. The feature is indeed called the "Coma-Sculptor cloud" (Tully & Fisher 1987, see plate 14). 8. SUMMARY AND CONCLUSIONS We have shown that the Surface Brightness Fluctuation method can be successfully used as a distance indicator for low surface brightness dwarf elliptical galaxies as faint as MB = −9.5. The application to the dwarf ellipticals rather than normal ellipticals has two advantages: (1) there are many dEs in the LG and its vicinity, and independent distances to these nearby dEs can be used to calibrate the absolute fluctuation magnitude ¯MR, (2) Through HST studies, we have detailed information about the variety of stellar content and star formation history of this galaxy type. However, most of them are dominated by an old metal poor stellar population. In this situation, first results from stellar population synthesis models suggest that ¯MR is relatively insensitive to the SFH. The internal fluctuation error for our dEs is 0.04-0.12 mag which includes the results from comparison of two different fields of the same galaxy. A small systematic uncertainty for ¯MR comes from the unknown stellar content of the galaxy. For example, ¯MR gets 0.12 mag fainter when going from a pure old population (all stars older than 8 Gyr) to an old population contaminated by 10% of young population. This trend was taken into account here by applying appropriate calibration constants to the different morphological galaxy types. The overall distance error we derive is between 4 and 7%. Although a more detailed analysis is required to follow up the population issue, a very satisfactory hint at the reliability of the theoretical values comes from the fact that all dEs turned out to be close companions of Scl group members for which independent distance measurements exist. These parent galaxies cover the full 3 Mpc depth of the Scl group. Based on the results for our five dEs, we further reviewed the kinematics and the -- 26 -- 3D-distribution of the Scl group. Surprisingly, the Hubble flow at the group shows no evidence for disturbance due to gravitational interaction between the main group members but is possibly decelerated by the tidal force of the nearby, massive LG leading to a large local Hubble constant of 119km s−1 Mpc−1. Taking advantage of the tight empirical velocity-distance relation we determined distances to dwarf irregulars in the Scl region and compiled distances from the literature for galaxies in the local volume to illustrate that the LG and Scl group are not isolated in space but are part of a large prolate cloud (6 × 2 × 1 Mpc3 in size) of dwarf irregulars well defined within the supergalactic plane. A number of groups are embedded in this structure, each containing one or more massive galaxies accompanied by a population of bound dwarf satellites which tend to be of early type. Finally it is interesting to note that four of our dE galaxies were already known prior this study. It is their membership of the Scl group which was uncovered here. This is a clear signal that the search for nearby dwarf elliptical galaxies should not only be focussed on the discovery of new, very low surface brightness galaxies but also on a careful follow-up of already known galaxies which appear nearby from their morphology. For such follow-up work, the SBF method is a powerful tool for measuring the distance of a galaxy. For the future, this opens up the possibility to measure accurate distances to all known dEs in the distance range between 1 and 10 Mpc without the expensive requirement of resolving the stellar system into stars. This will allow us to establish background-free dE samples, to explore the galaxy content of medium and low-density regions of the nearby universe, and to study the physical properties of dEs in greater details. Furthermore, the existence (or non-existence) of isolated early-type dwarfs can be established as a constraint on the formation and evolution of galaxies. We are grateful to John Tonry and Alan Dressler for very helpful advice, to the referee who made some important suggestions, and to Sylvie Beaulieu for doing test observations of NGC 5128 which were used to establish the SBF method on low-surface brightness regions. Two of us (HJ and BB) thank the Swiss National Science Foundation for financial support. HJ thanks further the Freiwillige Akademische Gesellschaft der Universitat Basel and the Janggen-Pohn Stiftung for their partial financial support. This project made use of the NASA/IPAC Extragalactic Database (NED), which is operated by the Jet Propulsion Laboratory, Caltech, under contract with the National Aeronautics and Space Administration. -- 27 -- ] 1 - s m k [ R S G V 500 400 300 200 100 0 1 NGC55 NGC253 SDIG NGC247 NGC300 ESO294-010 2 3 D [Mpc] NGC45 NGC59 NGC7793 4 5 Fig. 8. -- Hubble diagram for the 9 Scl group galaxies with known velocities and distances (from Table 5). The two filled squares are the new entries, ESO294-010 and NGC 59. All points are shown with ±1σ distance error bars. Velocity errors are small for all galaxies (< 5%) and not indicated here. The solid line is the best linear ML fit to the data. -- 28 -- 2 ] c p M [ D 0 48 49 -2 46 47 -6 43 42 38 45 44 39 41 40 33 32 36 37 34 28 30 35 31 22 25 26 23 20 21 24 14 19 27 7 5 17 10 9 8 1 6 3 4 12 2 11 13 16 18 15 29 2 -4 -2 D [Mpc] 0 Fig. 9. -- A map of the Local Group and its extension towards the Sculptor group region out to a distance of 6 Mpc, projected onto the supergalactic plane. Filled symbols are galaxies with measured distances. Circles: dwarf elliptical galaxies. The five new dE galaxies are #25, 32, 34, 38 and 45. Triangles: spiral galaxies. The two larger triangles indicate the positions of the Milky Way (at the origin) and M31 with their systems of close dwarf companions. Boxes: dwarf irregular galaxies. Open boxes are the dwarf irregulars detected by Cot´e et al. (1997) with distances reckoned from their velocities through the Hubble diagram (Fig. 8). The dashed lines show a zoom-in of the aggregate at NGC 55. The names of the galaxies are, 1: MW & companions, 2: LeoI & II, 3: NGC 6822, 4: Phoenix, 5: DDO210, 6: SagDIG, 7: M31 & companions, 8: IC1613, 9: LSG3, 10: M33, 11: NGC 3109, 12: Tucana, 13: Antlia, 14: WLM, 15: Sex A, 16: Sex B, 17: IC10, 18: GR8, 19: ESO407-018, 20: IC5152, 21: Pegasus, 22: NGC 55, 23: SC42, 24: SC24, 25: ESO294-010, 26: ESO293-035, 27: UGC-A86, 28: SC2, 29: Leo A, 30: NGC 300, 31: SC18, 32: ESO540-032, 33: NGC 247, 34: Scl-dE1, 35: SDIG, 36: NGC 253, 37: NGC 7793, 38: ESO540-030, 39: UGCA442, 40: ESO245-005, 41: NGC 625, 42: DDO006, 43: DDO226, 44: NGC 45, 45: NGC 59, 46: ESO473-024, 47: AM0106-382, 48: ESO348-009, 49: ESO347-017.
astro-ph/0104353
1
0104
2001-04-21T18:31:43
Nonthermal X-Ray Emission from G266.2-1.2 (RX J0852.0-4622)
[ "astro-ph" ]
The newly discovered supernova remnant G266.2-1.2 (RX J0852.0-4622), along the line of sight to the Vela SNR, was observed with ASCA for 120 ks. We find that the X-ray spectrum is featureless, and well described by a power law, extending to three the class of shell-type SNRs dominated by nonthermal X-ray emission. Although the presence of the Vela SNR compromises our ability to accurately determine the column density, the GIS data appear to indicate absorption considerably in excess of that for Vela itself, indicating that G266.2-1.2 may be several times more distant. An unresolved central source may be an associated neutron star, though difficulties with this interpretation persist.
astro-ph
astro-ph
Nonthermal X-Ray Emission from G266.2−1.2 (RX J0852.0−4622) P. Slane1, J. P. Hughes2, R. J. Edgar1, P. P. Plucinsky1, E. Miyata3, and B. Aschenbach4 1Harvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138 2Rutgers University, Piscataway, NJ 08854-8019 3Osaka University, Osaka 560-0043 JAPAN 4Max-Planck-Institut fur extraterrestrische Physik, Garching, Germany Abstract. The newly discovered supernova remnant G266.2−1.2 (RX J0852.0 -- 4622), along the line of sight to the Vela SNR, was observed with ASCA for 120 ks. We find that the X-ray spectrum is featureless, and well described by a power law, extending to three the class of shell-type SNRs dominated by nonthermal X-ray emission. Although the presence of the Vela SNR compromises our ability to accurately determine the column density, the GIS data appear to indicate absorption considerably in excess of that for Vela itself, indicating that G266.2-1.2 may be several times more distant. An unresolved central source may be an associated neutron star, though difficulties with this interpretation persist. INTRODUCTION G266.2−1.2 was discovered by Aschenbach [1] using data from the ROSAT All- Sky Survey. Situated along the line of sight to the Vela SNR, the emission stands out above the soft thermal emission from Vela only at energies above ∼ 1 keV. Iyudin et al. [2] reported the detection of 44Ti from the source GRO J0852-4642, which was tentatively associated with the SNR. If correct, the very short 44Ti lifetime (τ ≈ 90 y) would imply a very young SNR, and the large angular size would require that the remnant be very nearby as well. Estimates based on the X-ray diameter and γ-ray flux of 44Ti indicate an age of ∼ 680 y and a distance of ∼ 200 pc [3]. The hard X-ray spectrum, if interpreted as high temperature emission from a fast shock, would seem to support this scenario. However, as we summarize here (see [4] for a detailed discussion), the X-ray emission is not from hot, shock-heated gas; it is nonthermal. Further, reanalysis of the COMPTEL data finds that the 44Ti detection is only significant at the 2 − 4σ confidence level [5]. In the absence of such emission, and given that the X-ray emission is nonthermal, the nearby distance and young age may need to be reexamined. FIGURE 1. Left: ASCA GIS image of G266.2 -- 1.2 (E = 0.7 − 10 keV). The image consists of a mosaic of 7 individual fields. Contours represent the outline of the Vela SNR as seen in ROSAT survey data with the PSPC. Right: ASCA spectra from both GIS detectors for regions of G266.2 -- 1.2. The featureless spectrum is well described by a power law. Excess flux at low energies is presumably associated with soft thermal emission from the Vela SNR. OBSERVATIONS AND ANALYSIS We have carried out ASCA observations of G266.2−1.2 using 7 pointings, each of ∼ 17 ks duration. The resulting image from the ASCA GIS is illustrated in Fig. 1 [6], where we also present GIS spectra from three regions along the rim of the remnant: 1) the bright NW rim; 2) the NE rim; and 3) the W rim. Unlike the line-dominated thermal emission typical of a young SNR, the spectra are featureless and well described by a power law of index ∼ 2.6. G266.2−1.2 thus joins SN 1006 [7] and G347.3−0.5 [8,9] as a shell-type SNR dominated by nonthermal emission in X-rays. Weaker nonthermal components are observed for other SNRs as well, indicating the shock-acceleration of particles to energies of order 10 − 100 TeV. The best-fit spectral parameters for G266.2−1.2 yield NH ∼ (1 − 4) × 1021 cm−2. We include a soft thermal component with the column density fixed at a low value appropriate for Vela [10]. The column density for the power law component is significantly higher than that for Vela. While simple scaling of the column density to estimate the distance to G266.2−1.2 is clearly rather uncertain, it would appear that the remnant is at least several times more distant than Vela; improved column density measurements are of considerable importance. CO data [11] reveal a concentration of giant molecular clouds -- the Vela Molecular Ridge -- at a distance of ∼ 1 − 2 kpc in the direction of Vela. The column density through the ridge is in excess of 1022 cm−2, with the NE rim of G266.2−1.2 (nearest the Galactic Plane) falling along the steeply increasing column density region while the western rim lies along a line of much lower NH (Fig. 2); the CO column density varies by more than a factor of 6 between these regions. The lack of a strong variation in NH across G266.2−1.2 indicates that the remnant cannot be more FIGURE 2. Left: CO emission (VLSR = −5−+20 km s−1) in the direction of G266.2−1.2. Con- tours are GIS data. Note the Galactic coordinates. Right: ASCA SIS spectrum from Vela "Bullet D" showing thermal nature of emission, in contrast to nonthermal emission from G266.2−1.2. distant than the Ridge. On the other hand, if NH is much larger than for the Vela SNR, as the GIS data reported here indicate, then G266.2−1.2 must be as distant as possible consistent with being in front of most of the Molecular Ridge gas. In a recent study of optical emission in the vicinity of G266.2−1.2, Redman et al. [12] discuss a filamentary nebula lying directly along the edge of the so-called Vela "bullet D" region [13], just outside the eastern edge of Fig. 1 (in the direction of the extended contour), and argue that both structures represent a breakout region from G266.2−1.2. However, the ASCA spectrum of this region (Fig. 2) is clearly thermal, with large overabundances of O, Mg, and Si -- quite in contrast to the nonthermal spectrum observed for G266.2−1.2. We conclude that an association between this region and G266.2−1.2 is unlikely. The central region of the remnant contains a compact source, which we desig- nate as AX J0851.9-4617.4, surrounded by diffuse emission that extends toward the northwest. The diffuse emission is well described by a power law of spectral index ∼ 2.0, again accompanied by a soft thermal component that we associate with Vela. Given the nonthermal nature of the shell emission, however, it is quite possible that the diffuse central emission is associated with emission from the shell projected along the central line of sight. We note that the spectrum of this central emission appears harder than that from the rest of the remnant, perhaps suggesting a plerionic nature, but more sensitive observations are required to clarify this. The spectrum of the compact source is rather sparse and can be adequately described by a variety of models. A power-law fit yields a spectral index that is rather steep compared with known pulsars. The luminosity, on the other hand, is quite reasonable for a young neutron star if the distance is indeed several kpc, and the observed column density is compatible with that for the SNR shell. A blackbody model leads to an inferred surface temperature of ∼ 0.5 keV with an emitting region only ∼ 200dkpc m in radius which could be indicative of emission from compact polar cap regions of a neutron star, although this region is somewhat large for such a scenario. It is possible that the source is associated with a Be star within the position error circle [1], although the X-ray and optical properties appear inconsistent for such an interpretation [4]. Recent SAX observations [14] have revealed the presence of another source in the central region of G266.2−1.2. This source, designated SAX J0852.0-4615, has a harder spectrum than AX J0851.9-4617.4, but is considerably fainter. If at a distance of ∼ 3 kpc, its luminosity would be similar to the Vela pulsar, but this would place the source beyond the Vela Molecular Ridge. Further observations are required to investigate these central sources more completely. CONCLUSIONS The ASCA observations of G266.2−1.2 reveal that the soft X-ray emission from this SNR is dominated by nonthermal processes. This brings to three the number of SNRs in this class, and provides additional evidence for shock acceleration of cosmic rays in SNRs. Because of the bright and spatially varying background caused by the Vela SNR, limits on the thermal emission from G266.2−1.2 are difficult to establish. However, the ASCA data suggest a larger column density for this remnant than for Vela, indicating that G266.2−1.2 is at a larger distance, and perhaps associated with the star formation region located at a distance of 1 − 2 kpc. The point source AX J0851.9-4617.4 could represent an associated neutron star, with the diffuse emission around the source being an associated synchrotron nebula, albeit a very faint one. Recent SAX studies reveal another possible candidate. Higher resolution X-ray observations of these sources are of considerable importance. Acknowledgments. This work was supported in part by NASA through contract NAS8-39073 and grant NAG5-9106. REFERENCES 1. Aschenbach, B. 1998, Nature 396, 141 2. Iyudin, A. F. et al. 1998, Nature 326, 142 3. Aschenbach, B., Iyudin, A. F., & Schonfelder, V. 1999, A&A 350, 997 4. Slane, P. et al. 2001 -- to appear in ApJ (see astro-ph/0010510) 5. Schonfelder, V. et al. 2000, 5th Compton Symposium, Portsmouth, AIP Conf. Proc. 510, p. 54, eds. M.L. McConnell and J.M. Ryan 6. Tsunemi, H. et al. 2000, PASJ, in press (astro-ph/0005452) 7. Koyama, K. et al. 1995, Nature, 378, 255 8. Koyama, K. et al. 1997, PASJ 49, L7 9. Slane, P. et al. 1999, ApJ, 525, 357 10. Bocchino, F., Maggio, A., & Sciortino, S. 1999, A&A 342, 839 11. May, J., Murphy, D. C., & Thaddeus, P. 1988, A&ASS 73, 51 12. Redman, M. P., et al. 2000, ApJ, 543, L153 13. Aschenbach, B., Egger, R., Truemper, J. 1995, Nature, 373, 587 14. Mereghetti, S. 2001, ApJ - accepted (see astroph/0011554)
astro-ph/0210548
1
0210
2002-10-24T16:58:59
The largest black holes and the most luminous galaxies
[ "astro-ph" ]
The empirical relationship between the broad line region size and the source luminosity in active galactic nuclei (AGNs) is used to obtain black holes (BH) masses for a large number of quasars in three samples. The largests BH masses found exceed 10^{10} Msun and are correlated, almost linearly, with the source luminosity. Such BH masses, when converted to galactic bulge mass and luminosity, indicate masses in excess of 10^{13} Msun and sigma(*) in excess of 700 km/sec. Such massive galaxies have never been observed. The largest BHs reside, almost exclusively, in high redshift quasars. This, and the deduced BH masses, suggest that several scenarios of BH and galaxy formation are inconsistent with the observations. Either the observed size-L relationship in low luminosity AGNs does not extend to very high luminosity or else the M(BH)-M_B(bulge)-sigma(*) correlations observed in the local universe do not reflect the relations of those quantities at the epoch of galaxy formation.
astro-ph
astro-ph
The largest black holes and the most luminous galaxies Hagai Netzer 1 ABSTRACT The empirical relationship between the broad line region size and the source luminosity in active galactic nuclei (AGNs) is used to obtain black holes (BH) masses for a large number of quasars in three samples. The largests BH masses found exceed 1010 M⊙ and are correlated, almost linearly, with the source lumi- nosity. Such BH masses, when converted to galactic bulge mass and luminosity, indicate masses in excess of 1013 M⊙ and σ∗ in excess of 700 km/sec. Such massive galaxies have never been observed. The largest BHs reside, almost exclusively, in high redshift quasars. This, and the deduced BH masses, suggest that several scenarios of BH and galaxy formation are inconsistent with the observations. Ei- ther the observed size-L relationship in low luminosity AGNs does not extend to very high luminosity or else the MBH − Mbulge − σ∗ correlations observed in the local universe do not reflect the relations of those quantities at the epoch of galaxy formation. Subject headings: galaxies: high redshift -- quasars: general black hole physics -- galaxies: active -- galaxies: nuclei -- 1. Introduction Recent progress in reverberation mapping of active galacic nuclei (AGNs) allowed the first meaningful correlation between the broad line region (BLR) size (RBLR) and the black hole (BH) mass in more than 30 objects. This provided a simple way to calculate BH masses for a large number of sources and resulted in a flood of papers on this topic. Some papers (e.g. Vestergaard 2002, hereafter V02; McLure and Jarvis 2002) investigated, in great detail, the wavelength dependence of the RBLR − L − M relationship and provided useful ways for adopting the method to other wavelength bands. This opens the way for the study of BH masses in large samples of high luminosity high-z quasars. 1School of Physics and Astronomy, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel-Aviv University, Tel-Aviv 69978, Israel. -- 2 -- All the new BH mass estimates are based on a single relationship obtained for a single sample of 34 AGNs for which BLR sizes are available from decade long reverberation mapping campaigns. More than half the sample was observed at the Wise observatory over a period of about 12 years (Kaspi et al. 2000, hereafter K00). Other objects have been monitored in other observatories and in several "AGN watch" campaigns (Netzer & Peterson 1997; Peterson 2001). The main findings are a significant RBLR −λLλ(5100) relationship (Lλ(5100) is the monochromatic luminosity at 5100A) and the confirmation that the BLR gas is in virial motion (e.g. Peterson and Wandel 2000). These, plus the (model dependent) conversion of the observed full-width at half maximum (FWHM) of various emission lines into 3-D gas velocities, are sufficient to derive the mass of the central BH. This letter discusses the mass of the largest BH in the universe; those found in the centers of the most luminous quasars. It follows the works of Laor (1998; 2001), McLure & Jarvis (2002), Woo & Urry (2002) and others who used such methods for obtaining BH masses beyond the original K00 sample. The paper addresses also the Shields et al. (2002) new results and extends the mass estimates to much larger quasar samples. Section 2 presents new mass calculations for a large number of sources and §3 illustrates the new correlations found. Section 4 discusses the new results in light of the available information on the largest, most luminous elliptical galaxies and the epochs of quasars and galaxy formation. 2. The largest BH 2.1. BH Mass measurements New mass estimates have been obtained for a large number of AGNs using the RBLR − L relationship obtained from the K00 sample; the only sample available for such calibration. This relationship is given, schematically, by which results in the following mass estimate: RBLR = c1Lγ λ MBH = c2Lγ λ[F W HM]2 . (1) (2) Here c1 and c2 are constants that include the flux normalization and various assumptions about the velocity field in the BLR. The slope γ is derived from the reverberation campaigns results and is in the range 0.5-0.7 (see below). The expression in eqn. 2 can be used to derived "single epoch" masses that combine the constants γ and c2 with observed FWHM of certain emission lines in individual objects. The method has been described in various papers -- 3 -- including K00, V02, and McLure & Jarvis (2002). Its more useful applications are based on measured λLλ(5100) and FWHM(Hβ) for low redshift sources (the quantities used by K00) and the combination of λLλ(1350) and FWHM(C iv λ1549) for high redshift objects. V02 has looked into the inter-calibration of the two and supplied the expressions that are used in this work except for a small correction in the value of c2 that was introduced to adjust her constants to the cosmology assumed here: H0 = 70 km s−1 Mpc−1, Ωm = 0.3 and ΩΛ = 0.7. McLure & Jarvis (2002) provided similar expressions for MgII λ2798 which are not used in this work. 2.2. The sample Three AGN samples have been used in this work: 1) the LBQS sample (Forster et al. 2000 and references therein), 2) A sample of 104 high redshift high luminosity quasars with ground-based spectrophotometry (Lλ(1350)) and good FWHM(C iv λ1549) measurements, and 3) the new Lλ(5100) and FWHM(Hβ) listed by Shields et al (2002). Many of the sources in the second sample are UM quasars and the raw data can be found in MacAlpine and Feldman 1982, Baldwin, Wampler & Gaskell (1987) and Baldwin (1977). Forster et al. (2000) supplied monochromatic luminosities and FWHMs for many emis- sion lines in about 1000 LBQS quasars. Since many of the sources have been observed through relatively small aperture, and under poor weather conditions, it was decided to use the Bj magnitudes that are much more accurate. This follows Green et al. (2001) who studied the Baldwin relationship in this sample and obtained monochromatic luminosities using the same method. All fluxes have been corrected for galactic reddening using the Green et al. procedure. A major assumption here, and in Green et al. (2001), is that the observed continuum can be described by a single Lν ∝ ν −α power-law with α = 0.5. This approximation neglects the possible dependence of α on source luminosity which may affect the L − M relationship (see §4). Forster et al. (2001) provided several different measure- ments of FWHM(C iv λ1549) with and without the narrow line component. The "single" component fit was used and the "broad only" fit was checked to verify that the results are not sensitive to this choice. A handful of sources with FWHM(Hβ)< 1, 000 km s−1 or with FWHM(C iv λ1549)> 20, 000 km s−1 were removed from the sample since those were considered unreliable or affected too much by the narrow emission line. As for the second C iv λ1549 sample, no galactic reddening was applied and the same assumption about Lν was used. In this case there is no significant dependence on α since the original papers quote the observed flux at around rest wavelength of 1450A. The above samples are optically selected and suffer from various selection effects. This -- 4 -- is of no real consequence to the main goal of the paper which is to derive the mass of the largest known BHs. It may affect, however, the derived M − L correlations (§4). 3. The L − M relationship for high luminosity AGNs BH masses have been calculated using equation 2 and the normalizations derived by K00 and V02 adjusted to the cosmology chosen here. The determination of the slope γ is crucial for the present work and will be discussed prior to presentation of the new results. We start from the original K00 sample to which we apply two statistical methods for finding γ: the Akritas & Bershady (1996) BCES estimator (for which we only consider the BCES bisector) and the fitexy method described in Press et al. (1992). The merits of the different methods have been discussed, extensively, in several papers and will not be repeated here. Our experience shows that that the differences between the slopes obtained by the different methods are larger than the formal uncertainties on the slopes of each method. The K00 sample adjusted to the new cosmology gives γ = 0.58 ± 0.12 for the BCES bisector estimator and γ = 0.68 ± 0.03 for the fitexy method. The two are formally consistent with each other and γ(BCES) was adopted here. Since the purpose of this work is to extrapolate to very large L, we also experimented with removing the lowest luminosity objects from the sample. Removing the three objects with λLλ(5100) < 1043 ergs s−1 resulted in γ(BCES) = 0.71 ± 0.21 and γ(f itexy) = 0.69 ± 0.03. Removing the seven objects with λLλ(5100) < 1043.7 ergs s−1 resulted in γ(BCES) = 0.58 ± 0.19 and γ(f itexy) = 0.74 ± 0.04. All these results suggest that the two methods are consistent with each other and the slope cannot be determined to an accuracy better than about 0.15. The value adopted for illustrating the results of this work is the smaller one found for the entire K00, γ = 0.58. The implications for the case of larger or smaller γ are discussed in §4. Shields et al. (2002) suggested the use of the "physically motivated" value of γ = 0.5. The strongest argument for using this value is the suggestion by Netzer & Laor (1993) that the outer boundary of the BLR is determined by the dust sublimation radius which is similar to the measure RBLR to within a factor ∼ 2. There are several problems in applying this idea to the present mass determination. First, the "reverberation radius" is determined by the responsivity of Hβ to changes in the ionizing luminosity, Lion, which is smaller than the bolometric luminosity that determines the dust sublimation radius. In addition, γ = 0.5 means the same BLR ionization parameter for low luminosity Seyferts and the highest luminosity quasars. This has never been shown to be the case in large QSO -- 5 -- samples. Thus, more work is required to justify this theoretical value of γ. The masses computed with the γ = 0.58 slope are presented in Fig. 1. The diagram contains data for 505 QSOs with C iv λ1549 measurements and 219 source with Hβ mea- suremets. The luminosity range is roughly λLλ(1350) = 1044−47.5 ergs s−1. Also shown is the best regression line (see below) and the mass range of ±σM around the median calcu- lated in luminosity bins of 0.3 dex. The largest BHs are found in sources with z > 2 with MBH ≃ 1010.2 M⊙ (15 with mass exceeding 1010 M⊙). Using γ = 0.68 (the slope found with the f itexy method), raise this number to about 1010.4 M⊙ (62 with mass exceeding 1010 M⊙). Hβ Civ 1549 ) Θ M ( s s a m H B 1010 109 108 1045 1046 λLλ(1350 A) 1047 Fig. 1. -- Black hole mass as a function of λLλ(1350) for the quasar samples described in the text assuming γ = 0.58. Open symbols represent mass obtained from the Hβ line and full symbols masses obtained using C iv λ1549. The two C iv λ1549 samples completely overlap in properties and were not given different symbols. The dashed lines represent the ±σM range around the median and the straight line the M ∝ L0.8 relationship. -- 6 -- Fig. 2 shows MBH as a function of redshift for the same sample under the same assump- tions. ) Θ M ( s s a m H B 1011 1010 109 108 107 106 0 1 2 z 3 Fig. 2. -- MBH vs. redshift for the sample in figure 1 with the same symbols. The data in Fig. 1 suggest a simple linear dependence of the form MBH ∝ Lβ. This has been tested by performing a linear regression analysis using the same two methods described earlier. The procedure used for calculating the errors is the following: For L1350, the assumption is of a constant error of 0.15 dex representing the measurement uncertainty, the extrapolation in wavelength and the typical range in luminosity due to continuum variability. This number does not affect the resulting slope β in any significant way. As for the mass, this was done using standard error propagation combining all errors due to the uncertainties in L and in FWHM (line width uncertainties are given in Forster et al. 2000). The combined error for this case is typically 0.15-0.25 dex. No uncertainties are listed for FWHM(C iv λ1549) in the second quasar sample and for FWHM(Hβ) in Shields et al. (2002). A uniform error of 0.2 dex in MBH was assumed in those cases. The errors are relatively large and are expressed -- 7 -- in logarithmic form (i.e. 0.5(log(x+dx)−log(x−dx), see Lyons, 1991). Table 1 lists several slopes obtained by the two methods for our standard case of γ = 0.58 and for γ = 0.68, the slope obtained by the f itexy method. Given the various biases and unknowns, it is reasonable to assume that the real uncertainty in β is at least as large as ±0.15. With this uncertainty, the slopes of the C iv λ1549 sample and the entire sample are barely consistent with each other and the slopes of the Hβ sample and the entire sample are indistinguishable. The scatter in slope is probably due to the very different luminosity range of the C iv λ1549 and the Hβ samples. A second approach that was tries assumed a uniform uncertainty in MBH of 0.3 dex for all objects. This gave very similar results. The overall conclusion is that for luminous AGNs, MBH ∝ L0.9±0.15. The MBH − L correlation found here is very different from the one found in K00. The reason is probably the incompleteness of the small K00 sample which resulted in a biased sampling of FWHM(Hβ) vs. λLλ(5100) not representing the parent population. Indeed, K00 found FWHM(Hβ)∝ L−0.27 while in the samples under study the correlation is much flatter. The FWHM -- luminosity dependence in various samples will be addressed in a separate paper (Corbett et al. 2003). The tight MBH − L relationship enables the study of the Eddington ratio, L/LEdd, in these samples. The observed M − L relationship suggests a very weak, if any dependence of L/LEdd on L or on MBH. This impression is confirmed by a formal statistical analysis. Since the results are marginal, they will not be presented here. Another important issue is the mean L/LEdd. This depends on the distribution in this property as well as on the exact conversion from λLλ to bolometric luminosity and the value of γ. Assuming first γ = 0.58 and LEdd = 9λLλ(5100), as in K00, gives a median L/LEdd of 0.53. The composite spectra published recently by Telfer et al. (2002) suggest a different conversion with LEdd ≃ 5λLλ(5100). This translates to a median of 0.28. The above values are transformed to 0.33 and 0.18 for the case of γ = 0.68. In both cases the distribution wis wide, covering about a factor 10 in LEdd. Thus, the choice of γ = 0.58 results in a large mean L/LEdd and a large number of sources with super Eddington luminosities. As explained by Woo and Urry (2002), the implications to the derived MBH − L relation are very importance (see §4). 4. Discussion: the largest BHs and the most luminous galaxies The new results presented here suggest that the largest BHs are situated in the most luminous quasars that are, typically, the highest redshift sources. At the extreme end of the distribution we find BH masses of order 5 × 1010 M⊙ if γ = 0.7, and 1.1 × 1010 M⊙ if γ = 0.5. -- 8 -- This is greater than obtained so far in large samples. The recent work by Shields et al. (2002) aimed at the calibration of the the [O iii] λ5007 line width as a bulge mass estimator. The method is based on the close agreement between FWHM([O iii] λ5007) and the stellar velocity dispersion σ∗ at low luminosity and the K00 mass estimates at higher luminosity. Using this method and γ = 0.5 (their Table 2) they find one object with MBH exceeding 1010 M⊙ and several others approaching this mass. As shown in Fig. 1, the C iv λ1549 samples includes many more sources with such large masses. Before addressing the cosmological consequences we note the various factors influencing the M − L relationship and likely reasons for overestimating MBH. 1. The K00 sample covers a limited luminosity range and all mass estimates corresponding to λLλ(1350) > 1046 are necessarily obtained by extrapolation. Since this is the only sample available so far, there is no independent way to verify the largest masses until successful reverberation mappings are obtained for higher luminosity AGN. Moreover, as explained in §2, the slope of the RBLR-L relationship is uncertain. The slope chosen here (γ = 0.58) is close to the middle of the range. Its increase to 0.7 will increase the mass at the high luminosity end by a factor of about 2.5. 2. The largest new mass estimates are based on the measured λLλ(1350) which is scaled to the K00 luminosity assuming the same spectral energy distribution (SED) for high and low luminosity AGNs. This assumption has never been tested in large quasar samples. The data for such test are already available (Telfer et al. 2002) but the results are not yet known. Intrinsic reddening, in the quasar host galaxy, is another potential complication related to the inter-calibration of optical and UV luminosities. 3. The FWHM(C iv λ1549) may not reflect the virial motion of the BLR gas in high luminosity quasars. 4. The samples used here suffer from various selection effects. This influence only slightly the largest derived masses but can affect much more the L − MBH correlation. For example, magnitude limited samples may not include the less luminous quasars, those with the smallest L/LEdd. This results in a false impression of a very strong M − L correlation. Woo and Urry (2002) investigated this idea in great detail and concluded that all strong M − L correlations obtained so far suffer from such a selection effect. The main conclusion of this work is that the largest BH masses are found in the highest luminosity quasars. The masses of such BHs can reach the extreme values of 1010.3−10.6 M⊙, depending on the value of γ. Using recent conversions to host galaxy properties one finds -- 9 -- Mbulge ∼ 1013.1−13.4 M⊙ (Kormendy & Gebhardt 2001), MB,bulge ∼ −25 mag (Kormendy 2001) and σ∗ exceeding 800 km/sec (Tremaine et al. 2002). Such galaxies have never been observed and are not predicted to exist by standard galaxy formation theories. In principle, this is still consistent with the observations since the sources with the largest MBH are the most luminous ones and will completely out-shine any host galaxy. Thus, there is no direct way to rule out the existence of such galaxies. However, the theoretical implication are in conflict with recent ideas that the largest galaxies attain their mass through a series of mergers, a process that operates continuously to redshift 2 or smaller. A similar difficulty is found for the BH growth since those same theories (e.g. Haehnelt and Kauffman 2000; Yu and Tremaine 2002) assume that galactic nuclei BHs increase their mass up to redshifts smaller than 2 by the same series of mergers (or, perhaps, only through large mergers). Thus, the largest BHs are predicted to be associated with the most massive galaxies at z < 2, in conflict with the data in Fig. 2. It is clear that active BH with MBH > 1010 M⊙ are not found in the local universe. It is also clear that dormant BHs of this mass, or the galaxies with extreme properties that are supposed to host such BHs, have never been found. The whereabout of the huge BH formed at z ≃ 3 is thus unknown. A more plausible suggestion is that some or all the conversion factors used to obtain the galactic mass, magnitude and σ∗ from the BH mass, that are based on measurements in the local universe, cannot be extrapolated to high luminosity high redshift objects. Perhaps they are only valid at z < 2, after galaxies and nuclear BHs have accumulated most of their mass. If correct, this would mean that some "normal looking" galaxies contain extremely massive BHs. A similar suggestion by Laor (2001) involves a dependence of MBH/Mbulge on the BH mass or the absolute magnitude of the host galaxy. To conclude, either the measurements of BH masses presented here for the most lumi- nous quasars are grossly overestimated, because of the reasons described above, or else the relationships between BH masses and various properties of their host galaxies at high z are very different from those measured in the local universe. A second conclusion, which is less certain because of various selection effects, is that for AGNs, MBH ∝ L0.9±0.15. The work described in this paper is based primarily on a decade-long AGN monitoring I am grateful to many of my colleagues and students project at the Wise observatory. that helped in making this into a very successful project. Special thanks go to Dan Maoz, Shai Kaspi and Ohad Shemmer who led various parts of the project and without whom it would have been impossible to bring it to completion. Useful discussions with Ari Laor are gratefully acknowledged. This work is supported by the Israel Science Foundation grant 545/00. -- 10 -- REFERENCES Baldwin, J.A., 1977, ApJ 214, 679 Baldwin, J.A., Wampler, E.J., & Gaskell, C.M., 1987, ApJ 338, 630 Ferrarese, L., Pogge, R.E., Peterson, B.M., Merritt, D., Wandel, A., & Joseph, C.L., 2001, ApJ 555, L79 Forster, K., Green, P.J., Aldcroft, T.L., Vestergaard, M., Foltz, C.B., & Hewett, P.C., 2001, ApJS 134, 35 Green, P.J., Forster, K., & Kuraszkiewicz, J., 2001, ApJ 556, 727 Haehnelt, M., & Kauffman, G., 2000, MNRAS 318, l35 Kaspi, S., Smith, P.S., Netzer, H., Maoz, D., Jannuzi, B.T., & Giveon, U., 2000, ApJ 533, 631 (K00) Kormendy, J., & Gebhardt, K, 2001, in The 20th Texas Symposium on Relativistic Astro- physics, ed. H. Martel & J.C. Wheeler (AIP) , 363 Kormendy, J., 2001, RevMexAA conference series, 10, 69 Laor, A., 1998, ApJ 505, L83 Laor, A., 2001, ApJ 553, L677 Lyons, L., A Practical Guide to Data Analysis for Physical Science Students, 1991, Cam- bridge University Press MacAlpine, G., & Feldman, 1982, ApJ 261, 412 Maoz, D., 2002 (astro-ph/0207295) McLure, R.J., & Jarvis, M.J., 2002 (MNRAS in press, astro-ph0204473) Netzer, H., & Laor, A., 1993, ApJ 404, L51 Netzer, H., & Peterson, B.M., 1997, in Astronomical Time Series, Kluwer, (Maoz, Sternberg & Leibowitz eds.) Peterson, B.M., 2001, in The Starburst-AGN connection, World Scientific (I. Aretxaga, D. Kunth & R. Mujica eds) -- 11 -- Peterson, M.M., & Wandel, A., 2000 ApJ 540, L13 Peterson, B.M., et al. 2000, ApJ 542, 161 Shields, G.A., Gebhardt, K., Salviander, S., Wills, B.J., Bingrong, X., Brotherton, M.S., Yuan, J., & Dietrich, M., ApJ (in press astro-ph/0210050) Telfer, R.C., Zheng, W., Kriss, G.A., & Davidsen, A.F., 2002, ApJ 565, 773 Tremaine, S., et al., 2002, ApJ 547, 740 Yu, Q., & Tremaine, S., 2002, MNRAS 335, 965 Vestergaard, M, 2002, ApJ 571, 733 (V02) Woo, J.H., & Urry, C.M., 2002 (ApJ, in press; astro-ph/0207249) This preprint was prepared with the AAS LATEX macros v5.0. -- 12 -- Table 1. Regression analysis results for MBH = aLβ 1350 method γ sample β a BCES Bisector f itexy BCES Bisector f itexy BCES Bisector f itexy BCES Bisector f itexy 0.58 0.58 0.58 0.58 0.58 0.58 0.68 0.68 all C iv λ1549 all C iv λ1549 all Hβ all Hβ all objects all objects all objects all objects 1.12±0.05 1.13±0.04 0.83±0.03 0.80±0.02 0.80±0.02 0.78±0.01 0.90±0.02 0.89±0.01 -42.9 -43.1 -29.2 -27.6 -27.7 -26.9 -32.2 -31.9
0801.1272
1
0801
2008-01-08T16:40:14
Spurious source generation in mapping from noisy phase-self-calibrated data
[ "astro-ph" ]
Phase self-calibration (or selfcal) is an algorithm often used in the calibration of interferometric observations in astronomy. Although a powerful tool, this algorithm presents strong limitations when applied to data with a low signal-to-noise ratio. We analyze the artifacts that the phase selfcal algorithm produces when applied to extremely noisy data. We show how the phase selfcal may generate a spurious source in the sky from a distribution of completely random visibilities. This spurious source is indistinguishable from a real one. We numerically and analytically compute the relationship between the maximum spurious flux density generated by selfcal from noise and the particulars of the interferometric observations. Finally, we present two simple tests that can be applied to interferometric data for checking whether a source detection is real or whether the source is an artifact of the phase self-calibration algorithm.
astro-ph
astro-ph
Astronomy&Astrophysicsmanuscript no. 2007-8690 October 28, 2018 c(cid:13) ESO 2018 Spurious source generation in mapping from noisy phase-self-calibrated data I. Mart´ı-Vidal1 and J. M. Marcaide1 Dpt. Astronomia i Astrof´ısica, Universitat de Val`encia, C/Dr. Moliner 50, 46100 Burjassot (Valencia), SPAIN e-mail: [email protected] Accepted on 12 December 2007 ABSTRACT Phase self-calibration (or selfcal) is an algorithm often used in the calibration of interferometric observations in astronomy. Although a powerful tool, this algorithm presents strong limitations when applied to data with a low signal-to-noise ratio. We analyze the artifacts that the phase selfcal algorithm produces when applied to extremely noisy data. We show how the phase selfcal may generate a spurious source in the sky from a distribution of completely random visibilities. This spurious source is indistinguishable from a real one. We numerically and analytically compute the relationship between the maximum spurious flux density generated by selfcal from noise and the particulars of the interferometric observations. Finally, we present two simple tests that can be applied to interferometric data for checking whether a source detection is real or whether the source is an artifact of the phase self-calibration algorithm. Key words. Techniques: interferometric -- Methods: data analysis -- Techniques: image processing 1. Introduction Phase self-calibration (or selfcal) is an algorithm often used in the calibration of radio astronomical data. It was intro- duced by Readhead & Wilkinson (1978) and Cotton (1979), and it has been essential for the success of Very Long Baseline Interferometry (VLBI) imaging. Also, the antenna- based calibrations obtained from the Global Fringe Fitting al- gorithm (Schwab & Cotton 1983) are equivalent to a phase self-calibration. The phase selfcal will also be an algorithm widely used with future interferometric instruments, such as the Atacama Large Millimeter Array (ALMA) or the Square Kilometre Array (SKA), now under construction or planned. Optical interferometric observations (like those in the Very Large Telescope Interferometry, VLTI) will also eventually ben- efit from some form of selfcal, although closure phases and am- plitudes are measured in optical interferometry in a very differ- ent way than in radio. Thus, the statistical analysis presented here may need some substantial changes to rigorously describe the probability of false detections by optical interferometers. Given that part of the interferometric observations obtained from all those instruments may come from very faint sources, it is important to take into account the undesired and uncontrol- lable effects that the instrumentation and/or the calibration and analysis algorithms applied to the data could introduce in the in- terferometric observations. A deep study of all our analysis tools and their effects on noisy data is essential for discerning the re- liability of detections of very faint sources. Some discoveries made by pushing the interferometric instruments to their sensi- tivity limits could turn out to be the result of artifacts produced by the analysis tools. The main limitations of the phase self-calibration algorithm have been analyzed in many publications (e.g., Linfield 1986, Wilkinson et al. 1988). It is well known that an unwise use of Send offprint requests to: I. Mart´ı-Vidal selfcal can lead to imperfect images, even to the generation of spurious source components, elimination of real components, and deformation of the structure of extended sources. In this pa- per, we focus on the effects that phase self-calibration produces when applied to pure noise. We show that selfcal can generate a spurious source from pure noise, with a relatively high flux den- sity compared to the rms of the visibility amplitudes. We ana- lytically and numerically study how the recoverable flux density of such a spurious source depends on the details of the observa- tions (the sensitivity of the interferometer, the number of anten- nas, and the averaging time of the selfcal solutions). Finally, we study the effects of phase self-calibration applied to the visibili- ties resulting from observations of real faint sources, instead of pure noise. We present two simple tests that can be applied to real data in order to check whether a given faint source is real or not, and apply these tests to real data, corresponding to VLBI observations of the radio supernova SN 2004et (Mart´ı-Vidal et al. 2007). 2. Basics of phase self-calibration The basics of phase self-calibration can be found in many publi- cations (e.g., Readhead & Wilkinson 1978, Schwab 1980). Here, we explain the essence of this algorithm in a few lines. Let us suppose that we have made an interferometric observation us- ing a set of N antennas. We obtain one visibility, Vi j, for each baseline, that is, for each pair of antennas (i, j). Let us call φi j the phase of the visibility Vi j. Any atmospheric or instrumental effect on the optical path of the signals that arrived to antennas i and j will affect the phase φi j in the way: , j l φi j = φstr i j + φatm i − φatm where φatm represents all the undesired (i.e., atmospheric and instrumental) contributions to the optical path of the signal re- ceived by the antenna l and φstr the contribution to the phase that i j (1) 2 I. Mart´ı-Vidal et al.: Spurious source generation from noisy self-calibrated data. comes purely from the structure of the observed source, that is, the so called source structure phase. It can be easily shown that the quantities known as closure phases, and defined as (Jennison 1958, Rogers et al. 1974): l (2) Ci jk = φi j + φik − φ jk are independent of φatm . That is, the closure phases Ci jk are only defined by the structure of the observed source. Thus, they are in- dependent of any atmospheric or instrumental contribution that may affect the signals received by the antennas of the interfer- ometer. The phase self-calibration algorithm takes advantage of the closure phases to estimate the undesired antenna-dependent contributions φatm . In short, the phase selfcal finds which set of (called phase gains) generate antenna-dependent quantities φ the set of phases: gain l l φ sel f i j = φi j − φ gain i + φ gain j , (3) sel f i j and φ sel f i j = φstr are the phases that better represent the source struc- where φ ture given by the closure phases Ci jk. In the ideal case, φ φatm l (for l = i, j) are obtained is called hybrid mapping, and its ex- planation can be found in many publications (e.g., Cornwell & Wilkinson 1981). Here, suffice to say that the phase gains of the antennas are obtained from a least-square fit of the raw visibili- ties to the source model obtained from the mapping. i j . The process from which the values φ = gain l gain l The hybrid mapping is an iterative process from which the structure of the source model is refined step by step. Often, the model used in the first iteration of hybrid mapping is a point source located at the center of the map (obviously, the flux den- sity of this point source will not affect the phase calibration). The successive steps of hybrid mapping and selfcal correct this point source model until the structure that better represents all the closure phases is obtained. 3. Probability distribution of the visibilities due to pure noise When an interferometer observes a given source with a flux den- sity well below the sensitivity limit of its baselines (that is, when the interferometric data contain only noise), both the real and imaginary parts of the resulting visibilities follow Gaussian dis- tributions, centered at the origin. The amplitudes and phases of the visibilities follow distributions different from Gaussian. It can be shown that, for each baseline, the probability distribution of the phases is uniform between −π and π, and that instead the distribution of the amplitudes is given by: g(A) = A σ2 i j A2 2σ2 i j − exp  where g(A) is the probability density of the amplitudes and σi j is the width of the Gaussian distributions of the real and imaginary parts of the visibilities of the baseline (i, j). The width σi j is related to the thermal noise of the baseline (i, j). We assume, for simplicity, that all the baselines of the interferometer have the same value of σ. It can be shown that the rms of the visibility amplitudes of a pure noise signal is ρ = √2 σ and that the mean amplitude, < A >, is √π/2 σ, different from zero. 4. Probability of generating a spurious source from pure noise Given that the real part of a visibility with phase in the range (−π/2, π/2) is positive, all the visibilities with phases in that range bring a positive mean flux density to the map. We call phase close to zero to a phase in the range (−π/2, π/2) and phase distant from zero to a phase outside that range. The distribution of closure phases is uniform between −π and π, as it is also the case for the distribution of phases. This means that there is a subset of closure phases that by chance are close to zero, being compatible with a point source. However, there are also closure phases distant from zero, which are totally incom- patible with a compact source. If self-calibration is not applied to the data, then the uniform distribution of phases (and closure phases) will result in a noisy map with no source defined in it. But if a single iteration of phase self-calibration is applied, there is a selection process of the closure phases in the calibration, which may generate a spurious source with a flux density com- parable to the rms of the amplitudes (which can be much higher than the rms of the image), as we show below. For each scan, the effects of the least-square fit described in Sect. 2 can be understood in the following way: selfcal searches for the visibilities corresponding to the antennas most commonly appearing in the closure phases that are close to zero (which usually correspond to phases that can be modelled with a point source). Then, selfcal minimizes the phases of such visibilities by calibrating those antennas, leaving all the other visibilities with the phases dispersed between −π and π. In other words, the phases of the visibilities with large closure phases contribute to increase the value of the χ2 at the minimum, but the position of such minimum only depends on the visibilities with phases that can be modelled with a point source. That is, all the visi- bilities susceptible of producing a compact source are calibrated and their phases concentrate around zero. All the other visibili- ties tend to have their phases uniformly distributed between −π and π, thus generating, after the Fourier inversion, a null mean flux density in the map. Thus, selfcal always produces a positive bias in the mean flux density of the map, given that selfcal only acts, effectively, on the phases that can be approached to zero, because their corresponding closure phases are close to zero. One might think that it is very difficult for an antenna to be involved in a large number of closure phases close to zero, given that the distribution of phases is uniform. One might think that, in average, a given antenna is involved in the same number of closure phases that are close to zero than in the closure phases that are distant from zero. In such case, it would be impossible for selfcal to select which antenna should be calibrated, given that all the antennas participating in each scan would have the same chances for being calibrated. But this is only true in aver- age. In the distributions of all the interferometric observations, there are statistical fluctuations, which are always used by self- cal for the generation of a spurious point source. For a given antenna i, the probability of finding n closure phases (in which that antenna is involved) close to zero is: (5) where N′ = (N − 1)(N − 2)/2 is the total number of closure phases in which antenna i is involed. (Notice that even though all closure phases are not independent, the closure phases with one common antenna are.) Thus, there is a finite probability of finding an antenna that appears in more than N′/2 closure phases close to zero. In such cases, the phases of the baselines in which (4) Pi(n) = 1 2N′ (N′)! n!(N′ − n)! I. Mart´ı-Vidal et al.: Spurious source generation from noisy self-calibrated data. 3 the antenna i appears will be minimized with success, generating a positive mean flux density in the map. Actually, even the cases in which n < N′/2 can also be used by selfcal for the generation of a spurious point source. In such cases, the closure phases in which antenna i appears will tend to gather around −π and π, meaning that there are other antennas that will appear in a large number of closure phases close to zero (i.e., the antennas be- longing to the closure phases distant from zero in which antenna i appears). In short, there are always statistical fluctuations in the pure noise distribution of phases that can be used by selfcal to produce (by a selection process of the antennas most commonly being involved in the closure phases close to zero) a point source with a spurious source flux density. It must be said that Global Fringe Fitting, when applied to noisy data, can also generate a spurious source from pure noise, given that this algorithm also finds antenna-based calibrations for adjusting the interferometric fringes of all the baselines in each scan. The spurious source is generated as long as the min- imum SNR of the fringes to be considered in the fit is set to a small value (lower than 2 or 3)1. In those cases, the Fringe Fitting would work on correlation peaks (fringes) produced, in many cases, by spurious noise fluctuations. Then, by the same reasons given above, there would be a relatively high probability of generating a spurious point source from pure noise. 5. Dependence of the flux density of the spurious source on the characteristics of the observations In this section we consider how the flux density of the spurious source generated by selfcal depends on the parameters defining a set of observations. The parameters that we consider are the sen- sitivity of the array (for simplicity, we assume the same sensitiv- ity for all the baselines of the interferometric array), the number of antennas of the interferometer, and the averaging time of the selfcal solutions. For the case of the Global Fringe Fitting algo- rithm, the averaging time of the solutions is equal to the duration of the scans. 5.1. Numericalstudy We generated a set of synthetic interferometric data with the pro- gram fake of the Caltech Package (Pearson 1991). We generated 6 hours of observations using a set of 20 antennas. All these antennas had the same diameters (25 m) and the same system temperatures (60 K). The correlator integration time was set to 2 seconds. The source model used by fake consisted on a single point source with a flux density of 1 nJy (of course, completely undetectable by the interferometer). The data generated this way thus contain 6 hours of pure noise observations made at 20 iden- tical antennas under identical conditions. The mean of all the visibility amplitudes is 106 mJy. We used the program difmap (Shepherd et al. 1995) for hy- brid mapping. We applied the natural weighting scheme to the visibilities, for sensitivity optimization, and applied an initial selfcal using a centered point source. The hybrid mapping steps were repeated until the χ2 of the fit of selfcal arrived to conver- gence. Then, we deconvolved the resulting point source using the CLEAN algorithm to see how much flux density was gener- ated by selfcal. This process was repeated for different numbers of antennas and for different averaging times of the selfcal so- lutions. The spurious point source flux densities obtained in all these cases are shown in Fig. 1 as filled circles. 5.2. Analyticalstudy When the interferometer observes only noise, the sensitivity of the baselines defines the value of the standard deviation, σ, of the Gaussian distributions of the real and imaginary parts of the visibilities. As the sensitivity increases, the thermal noise of the baselines decreases, decreasing also the value of σ. Given that selfcal has to do only with phases and leaves the amplitudes un- altered, the recoverable flux density of the spurious source de- pends linearly on the rms of the amplitudes, which in turn also depends linearly on the value of σ. Thus, an increase in the sen- sitivity of the interferometer decreases the amount of spurious flux density recoverable from the data using selfcal. The con- stant factor in the ratio between the flux density recovered and the rms of the visibility amplitudes depends on the method used to estimate the flux density (i. e. different deconvolution algo- rithms or modelfitting to the visibilities). In our case, from the numerical study described in the previous section, we determine it to be 0.907 ± 0.002. The averaging time of the selfcal solutions also has an effect on the spurious source flux density. If we have one observation every t0 seconds (usually, t0 is 2 seconds) and find only one so- lution of selfcal every t seconds, with t > t0, then the spurious source flux density decreases. Finding a single solution of selfcal every t seconds is equivalent to averaging all the observations in bins of t seconds and, afterwards, self-calibrating the resulting visibilities. When we average the visibilities in blocks of t sec- onds, we are averaging separately the real and imaginary parts of the visibilities, which follow Gaussian distributions. The ef- fect of this average is that the standard deviations of the resulting distributions decrease by a factor √t/t0, because of the Central Limit Theorem. The dependence of the spurious source flux density on the number of antennas is more difficult to find. There are lots of possible combinations of phases and baselines that help selfcal to generate a spurious source, and each one of these combina- tions has a different weight in the final spurious flux density. We can use a simplified phenomenological model to find out the re- covered flux density as a function of the number of antennas. In principle, the recovered flux density depends directly on how well the point model fits the data. A good indicator of the ad- justability of the data, with the phases randomly distributed, is, for each scan, equal to the number of visibility phases divided by the number of phase gains to fit. That is, the number of phases per parameter. When the number of phases per parameter in- creases, one single parameter must account for the minimization of more phases, and the quality of the fit decreases. In our case, the number of parameters is equal to N − 1, given that one an- tenna (the reference antenna) has a null phase gain by definition. Thus, the number of phases per gain to be fitted is: #phases #gains = N(N − 1) 2 1 N − 1 = N 2 (6) 1 Even though a SNR of 2−3 is not generally used in the calibration of typical observations (being SNR cutoffs of ∼5 more common), low SNR cutoffs may be applied with very small delay/rate windows, say, after a phase-reference pre-calibration. We can now look for a model of the dependence of the re- covered spurious source flux density on the number of antennas using the quantity N/2 as variable. From the numerical simula- tions described in the previous section, we have found that the 4 I. Mart´ı-Vidal et al.: Spurious source generation from noisy self-calibrated data. Fsp rms 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 t = 2s t = 6s t = 30s 3 5 7 9 13 15 17 19 11 N Fig. 1. Flux densities of an spurious point source, Fsp, in units of amplitude rms, ρ, recovered from hybrid mapping using pure noise synthetic visibilities, as a function of the number of anten- nas (N) and the averaging time of the selfcal solutions (t). The lines correspond to our model (Eq. 7) and the dots to the numer- ical simulations. spurious source flux density is fitted very well using the simple model Fsp ∝ (N/2)γ, with γ = −0.413 ± 0.001. Taking these considerations into account, the recoverable flux density using selfcal on a set of randomly distributed data can be written as: (7) 2 (cid:19)−0.413 t (cid:18) N Fsp = 0.907ρrt0 where Fsp is the spurious source flux density that can be gener- ated by selfcal, ρ is the root-mean-square (rms) of the visibility amplitudes, t0 is the averaging time used in the correlator (typi- cally, t0 = 2 seconds), t is the averaging time of the selfcal solu- tions, and N is the number of antennas of the interferometer. We note that the duration of the whole set of observations does not affect Fsp, since this flux density depends on the ratio between the number of phases close to zero and the number of phases distant from zero, but does not depend on the total amount of visibilities used in the Fourier inversion. This model is shown in Fig. 1. 5.3. Theuseofselfcalinspecificsituations Equation 7 gives an estimate of the contribution of the artifacts of selfcal to the flux density of a source obtained by calibrating the antennas with the hybrid mapping technique. For cases of high SNR data, such contribution to the total flux density of the sources is negligibly small. However, when the flux density of a source is comparable to the rms of the visibility amplitudes, care must be exercised with the use of selfcal or the Global Fringe Fitting algorithm. We note that in the worst situation for the use of selfcal (i.e., 3 antennas and t = t0) the amount of spurious source flux den- sity is as large as 76% of the rms of the visibility amplitudes. For 10 antennas (the case of the VLBA) the recoverable flux density decreases to 46% of the rms (and can be lower if we set t > t0). For interferometers with a large number of antennas, the amount of spurious source flux density is, of course, smaller. For exam- ple, if we extrapolate the results shown in Fig. 1 to 50 antennas (the case of ALMA), the spurious source flux density generated by selfcal would be 24% of the rms of the visibility amplitudes, using t = t0. All these results assume the same sensitivity for all the base- lines. In real cases, each baseline has its own sensitivity, with the longest baselines noisier than the shortest ones. The use of data from all the baselines in the fit can worsen the situation. A good alternative for avoiding the spurious source generated by selfcal or, at least, to make its flux density smaller is to flag or downweight the longest baselines in the fit and/or to increase the statistical weight of the data coming from the most sensitive an- tennas of the array. Nevertheless, even doing so, the statistical fluctuations of the closure phases will always tend to make, after the use of selfcal, a spurious source with a considerably large flux density. A better way to calibrate faint source data is using the phase- reference technique (e.g., Beasley & Conway 1995). When us- ing this technique, scans of a strong (reference) source are in- troduced between the scans of the faint (target) source. Then, the antenna gains are determined from the observations of the strong source and then interpolated to the scans of the faint (tar- get) source. This technique is rather insensitive to the artifacts of selfcal and the probability of generating a fake signal from noise is practically zero. This is so, because the calibration of the target source comes from the analysis of data coming from another source (the reference source). Therefore, the noise in the data of the faint source does not affect the antenna calibrations. However, it is common to use the antenna gains determined from the phase reference as an a priori calibration, performing then a Global Fringe Fitting on the target source data using small search windows (based on the calibration from the reference source data) or applying self-calibration to the target visibilities in order to improve the dynamic range of the final image. In some cases, this might be malpractice, because the probability of generating a spurious source flux density from noise appears anew with full strengh, wasting all the benefits of the phase referencing. 6. Tests of the reliability of a source detection from noisy data In this section we present two simple tests that can be per- formed on real data in order to check the reliability of a possible source detection, or to check if part of the flux density of a de- tected source may come from artifacts of selfcal. These tests are only meaningful when they are applied to extremely noisy data. However, the application of these tests to high SNR data can still give us an idea of the possible contribution of the artifacts of selfcal to the flux density recovered from a source. We assume, in all our discussion, that the detected source is compact enough to be considered point-like, without any impor- tant loss of precision. As in the previous sections, to simplify the expressions we also assume that all the baselines of the interfer- ometer have the same sensitivity. 6.1. Testbasedontheaveragingtimeoftheselfcalsolutions The dependence of the spurious source flux density on the aver- aging time of the selfcal solutions is determined by the depen- dence of the standard deviation σ on the averaging time. That dependence translates into equation 7. However, the flux density of a real source in the data is independent of the averaging time of the selfcal solutions. We can use this condition to estimate the flux density of a (possibly real) source detected under crit- ical circumstances. If we apply phase self-calibrations to a real data set for different averaging times, t, of the selfcal solutions, I. Mart´ı-Vidal et al.: Spurious source generation from noisy self-calibrated data. 5 the flux densities recovered after each self-calibration, Fself, are given by the formula: Fself = Fsp + Freal = K √t + Freal (8) where K (related to the specifics of the interferometer) and Freal (an estimate of the flux density of the real source) are parameters to be fitted. 6.2. Testbasedonthedistributionofclosurephases In the case that the signal of a real source is included in the vis- ibilities, the probability distributions of the real and imaginary parts are still Gaussian, but the mean value of the real part of the visibilities (if such visibilities are well calibrated) will be equal to the flux density of the real source. Then, it can be easily shown that the probability distribution of the resulting visibility amplitudes and phases is: g(A, φ) = A σ2 exp − (A + F)2 2σ2 ! exp(cid:18) AF cos φ σ2 (cid:19) (9) where g(A, φ) is the probability density of amplitudes (variable A) and phases (variable φ), F is the flux density of the real source, and σ is the width of the distributions of the real and imaginary parts of the visibilities. Equation 9 turns into equation 4 for F = 0. When F is dif- ferent from zero, the distribution of phases is not uniform. As F increases, the phases gather around zero in a Gaussian-like man- ner. The distribution of amplitudes also changes, increasing the ratio between the rms of the visibility amplitudes and σ. How could the information provided by equation 9 be used to check the reliability of a possible source detection? The clo- sure phases are robust quantities that can be used to check the reliability of a source detection. The closure phases are sensitive to F and are not affected by the phase self-calibration. The dis- tribution of closure phases tends to gather around zero if there is signal of a real source in the data (and particularly so if the source has no structure) and is uniformly distributed if there is only noise in the data. From the definition of closure phase (see equation 2) we conclude that, if the visibility phases are well calibrated, the probability distribution of closure phases is equal to: c(β) = Z π −πZ π −π p(φ1)p(φ2)p(β − φ1 − φ2)dφ1dφ2 (10) where c(β) is the probability density of the closure phase, β, and p(φ) = Z ∞ 0 g(A, φ)dA (11) Let us ellaborate on this: for the case of a perfect calibration, the phases of data from a given baseline (i, j) are independent of the phases from any other baseline, given that all the contri- butions to φi j come from noise, which is uncorrelated between the different baselines. Thus, the probability distribution func- tion of any linear combination of visibility phases (as it is the case of the closure phase) is equal to the product of the proba- bility distributions of the visibility phases. However, for the case of a non-perfect calibration, in which selfcal has introduced a spurious source flux density in the data, the distributions of the phases from the different baselines will no longer be independent and equation 10 will not quite apply. In such cases, the phase distributions, p(φ), will be more peaked around zero, but corre- lations will appear among the phases of the different baselines. Therefore, the probability distribution function of a linear com- bination of phases will not be equal to a simple product of p(φ). Nonetheless, the correlations between baselines that selfcal in- troduces in the data keep the distribution of closure phases, c(β), unaltered. That is, even after a selfcal iteration has generated a spurious source in the map, the distribution of closure phases is still equal to c(β), as computed from the distribution of the phases, p(φ), corresponding to perfectly calibrated data. This invariance property of the closure phase distribution, c(β), can be used to check the reliability of a possible source de- tection. In Fig. 2, the theoretical flux density of a real source (in units of the rms of the amplitudes, ρ) is shown as a function of the mean absolute value of the closure phases and as a function of the mean cosine of the closure phases. Both quantities are di- rectly related to the deviation of the closure phase distribution from a uniform distribution. If the closure phases are uniformly distributed between −π and π, the average of their absolute val- ues will be equal to π/2 and the mean cosine will be equal to zero. Any concentration of closure phases around zero (i.e., the existence of any real source in the data) will result in a decrease of the first value and an increase of the second. If a faint source is detected and the reliability of the detec- tion needs to be checked, the average of the absolute values of the closure phases of the observations (or the average of their cosines) can be computed and compared in Fig. 2 to determine how concentrated around zero the closure phases are. This way, it can be seen whether the flux density of our detection is con- sistent with a real source or not. 6.2.1. Comments on the closure-phases test At this stage, it must be said that, even though the self-calibration does not affect the closure phases, a time average of a set of self- calibrated visibilities will change the closure phase distribution. That is, if selfcal generates a spurious source from noise and the visibilities are averaged in time bins of t seconds, with t > t0, the closure phase distribution changes and the resulting closure phases concentrate around zero, creating the effect of a source that is completely indistinguishable from a real one using any test. For very small values of the flux density F, this test is not as good as the first one. As we can see in Fig. 2, the closure phases do not dramatically change their distribution for flux densities in the range between 0 and ∼15% of the rms of the visibility amplitudes. For tentative detections under critical circumstances in that range, this test could lead us to the wrong conclusion about the reliability of a source detection. Looking at it from a different viewpoint, Fig. 2 provides an interesting lesson: for a given dataset, there can exist a real faint source (appearing in a map with a dynamic range of 6 or more) even if the distribution of closure phases is uniform, that is, even if the closure phase distribution is noise-like. Thus, a conclusion on the reliability of a source detection based only on the closure phase distribution being extremely noisy, is not definitive. Another thing worth-noticing is that this test assumes the same sensitivity for all the antennas and a source compact enough for generating closure phases close to zero even for the longest baselines (which may not be the case, specially for sources with a very low flux density per unit beam). These as- sumptions impose limitations to the use of this test. However, it could still be applied by restricting its use to the shortest base- lines with similar antennas. For an array with a large antenna, 6 I. Mart´ı-Vidal et al.: Spurious source generation from noisy self-calibrated data. 0.4 0.3 0.2 F ρ 0.1 0 0 0.01 0.02 0.03 π/ 2 − < C > 0.04 0.05 0.06 0 0.01 0.02 0.03 0.04 0.05 < cos(C) > Fig. 2. Theoretical dependence of the flux density of a real observed source (in units of the rms of the visibility amplitudes), as a function of "π/2 minus the mean absolute values of the closure phases" (left side) and as a function of the mean cosine of the closure phases (right side). These theoretical behaviors have been obtained from equation 10. the closure phases in which that antenna appears could still be used (for sensitivity optimization in the closure phase distribu- tion), but then the flux density estimated from the amplitude rms might have a bias produced by the very different rms in different baselines. We must also note that in cases of high SNR, the rms of the visibility amplitudes is no longer related only to the thermal noise of the baselines (the flux density of the source affects the value of the rms), and the fraction F/ρ shown in Fig. 2 should be accordingly corrected. For cases of high SNR, the quantity that will substitute ρ in the fraction F/ρ shown in Fig. 2 is pρ2 − F2. 6.3. Applicationtorealcases Real data do not obey the simplifying assumptions that we have used in the earlier sections. The baselines of a real interferome- ter have different sensitivities, which also vary in time. Thus, in order to check the reliability of a source detection from real data we must search a subset of observations in which the sensitivity of the antennas is approximately constant. Moreover, we must only work with the subset of most sensitive antennas of the in- terferometer. If there is one antenna in our interferometer that is clearly more sensitive than the others, we should compute only the average of the closure phases in which this more sensitive antenna appears, in order to insure the possible signature of the source in the closure phase distribution. In what follows, we ap- ply our reliability criteria to real data corresponding to the radio supernova SN 2004et (Mart´ı-Vidal et al. 2007). 6.3.1. Supernova SN 2004et We observed this supernova on 20 February 2005. From all data reported in Mart´ı-Vidal et al. (2007), we have chosen the fol- lowing subset of antennas: Brewster, Fort Davis, Green Bank, Hancock, Kitt Peak, and Owens Valley. We have only computed the closure phases in which the antenna Green Bank appears, and we have used data only from 14 hr to 20 hr (UT). These choices are based on the quality of the data for our purposes (i.e., the sta- bility of the antenna sensitivities, which we assume proportional to the system temperatures registered for each station). First test: we self-calibrated the SN 2004et data using dif- ferent averaging times, ranging from 2 to 120 seconds (roughly, the duration of one scan). The fit of the flux densities recov- ered from the SN 2004et data as a function of the averaging time of the selfcal solutions, equation 8, results in a value of Freal = 0.90 ± 0.13 mJy. This value is clearly higher than zero, indicating that there is a real signal in the data. This value is also close to the flux density of SN 2004et reported by Mart´ı-Vidal et al., 0.87 ± 0.03 mJy, recovered from phase referenced data with a deconvolution using CLEAN. Second test: even though we know that it is not quite appro- priate, we have also performed this test on the SN 2004et data. The flux density of SN 2004et is too low compared to the rms of the visibility amplitudes (∼20 mJy) for obtaining a good result with the test of the closure phase distribution. The average value of the cosines of all the closure phases considered in these obser- vations is equal to 0.005 ± 0.002. This average value is slightly higher than 0. The average of the absolute values of the closure phases is 0.0065 ± 0.0020. From Fig. 2 we estimate a flux den- sity of the supernova of 0.18±0.08 times the rms of the visibility amplitudes, ρ, used in our computations, which, as said above, is ∼ 20 mJy. Hence, the estimated flux density of the (real) source in the data is, then, 3.8±1.6 mJy. This value is too high, but com- patible (at a 2 -- sigma level) with the flux density estimated from the other reliability test. As expected, also the uncertainty of the flux density estimated is much higher in this case. Thus, the reliability tests for this source are successful. We must note that Mart´ı-Vidal et al. calibrated the data of SN 2004et using the phase-reference technique, in which they interpolated the antenna gains obtained from the observations of the source J2022+614 to the scans of the supernova. These authors did not refine afterwards such a calibration applying a Global Fringe Fitting to the supernova data. This procedure assured a reliable detection of the supernova. These authors did not apply any other calibration (selfcal) to the phased-referenced supernova data, in order to avoid any possible artifact introduced by the use of self- cal. 7. Conclusions We have analyzed the consequence of the phase-self-calibration algorithm when it is applied to extremely noisy data. We have studied how this algorithm and the statistical fluctuations of the visibility phases can create a spurious source from pure noise. The flux density of the spurious source can be a considerable fraction of the rms of the visibility amplitudes. The applica- tion of other other antenna-based calibration algorithms (like the Global Fringe Fitting) to noisy data can have similar conse- I. Mart´ı-Vidal et al.: Spurious source generation from noisy self-calibrated data. 7 quences to those of selfcal if the SNR cutoff of the gain solutions is set to small values. We have considered numerical and analytic studies to show how the flux density of a spurious source created by selfcal de- pends on the number of antennas, the sensitivity of the array, and the averaging time of the selfcal solutions. We have also pre- sented two simple tests that can be applied to real data in order to check if the detection of a faint source could be the result of the application of an antenna-based calibration algorithm to noisy data. These tests basically relate the averaging time of the selfcal solutions and the characteristics of the closure phase distribution to the flux density of a compact source possibly present in the data. To show a practical case, we have applied these tests to a set of real VLBI observations of supernova SN 2004et and found good agreement between the flux density recovered by CLEAN from the (phase-referenced) visibilities of this supernova (Mart´ı- Vidal et al. 2007) and the flux density estimate provided by our reliability tests. Acknowledgements. This work has been partially funded by Grants AYA2004- 22045-E and AYA2005-08561-C03-03 of the Spanish DGICYT. We agree the anonymous referee for his/her very helpful comments, corrections, and sugges- tions. References Beasley A.J. & Conway J.E., 1995, ASP Conf. Ser. 82: Very Long Baseline Interferometry and the VLBA, 82, 328 Cornwell T.J. & Wilkinson P.N., 1981, MNRAS, 196, 1067 Cotton W. D., 1979, AJ, 84, 1122 Jennison R. C., 1958, MNRAS, 118, 276 Linfield R.P., 1986, AJ, 92, 213 Mart´ı-Vidal I., Marcaide J.M., Alberdi A., et al., 2007, A&A, 470, 1071 Pearson T.J., 1991, BAAS, 23, 991 Readhead A.C.S. & Wilkinson P.N., 1978, ApJ, 223, 25 Rogers A.E.E., Hinteregger H.F., Whitney A.R., et al., 1974, ApJ, 193, 293 Shepherd M.C., Pearson T.J., & Taylor G.B., 1995, BAAS, 26, 987 Schwab F.R., 1980, Proc. Soc. Photo-Opt. Instrum. Eng., 231, 18 Schwab F.R., Cotton W. D., 1983, AJ, 88, 688 Wilkinson P.N., Conway J., & Biretta J., 1988, IAUS, 129, 509
0806.3494
2
0806
2008-06-25T04:45:28
A Super-high Angular Resolution Principle for Coded-mask X-ray Imaging Beyond the Diffraction Limit of Single Pinhole
[ "astro-ph" ]
High angular resolution X-ray imaging is always demanded by astrophysics and solar physics, which can be realized by coded-mask imaging with very long mask-detector distance in principle. Previously the diffraction-interference effect has been thought to degrade coded-mask imaging performance dramatically at low energy end with very long mask-detector distance. In this work the diffraction-interference effect is described with numerical calculations, and the diffraction-interference cross correlation reconstruction method (DICC) is developed in order to overcome the imaging performance degradation. Based on the DICC, a super-high angular resolution principle (SHARP) for coded-mask X-ray imaging is proposed. The feasibility of coded mask imaging beyond the diffraction limit of single pinhole is demonstrated with simulations. With the specification that the mask element size of 50* 50 square micrometers and the mask-detector distance of 50 m, the achieved angular resolution is 0.32 arcsec above about 10 keV, and 0.36 arcsec at 1.24 keV where diffraction can not be neglected. The on-axis source location accuracy is better than 0.02 arcsec. Potential applications for solar observations and wide-field X-ray monitors are also shortly discussed.
astro-ph
astro-ph
Chinese Journal of Astronomy and Astrophysics manuscript no. (LATEX: fix-final.tex; printed on April 15, 2019; 7:33) A Super-high Angular Resolution Principle for Coded-mask X-ray Imaging Beyond the Diffraction Limit of Single Pinhole Chen Zhang1 and Shuang Nan Zhang2 , 3 1 Department of Engineering Physics & Center for Astrophysics, Tsinghua University, Beijing 100084, P. R. China; 2 Key Laboratory of Particle Astrophysics, Institute of High Energy Physics, the Chinese Academy of Sciences, Beijing 100049, P. R. China; 3 Department of Physics & Center for Astrophysics, Tsinghua University, Beijing 100084, P. R. China; Received 2007 month day; accepted 2007 month day Abstract High angular resolution X-ray imaging is always demanded by astrophysics and solar physics, which can be realized by coded-mask imag- ing with very long mask-detector distance in principle. Previously the diffraction-interference effect has been thought to degrade coded-mask imag- ing performance dramatically at low energy end with very long mask- detector distance. In this work the diffraction-interference effect is described with numerical calculations, and the diffraction-interference cross correla- tion reconstruction method (DICC) is developed in order to overcome the imaging performance degradation. Based on the DICC, a super-high angular resolution principle (SHARP) for coded-mask X-ray imaging is proposed. The feasibility of coded mask imaging beyond the diffraction limit of single pinhole is demonstrated with simulations. With the specification that the mask element size of 50 × 50 µm2 and the mask-detector distance of 50 m, the achieved angular resolution is 0.32 arcsec above about 10 keV, and 0.36 arcsec at 1.24 keV (λ = 1 nm) where diffraction can not be neglected. The on-axis source location accuracy is better than 0.02 arcsec. Potential appli- cations for solar observations and wide-field X-ray monitors are also shortly discussed. 2 C. Zhang & S. N. Zhang Key words: instrumentation: high angular resolution -- techniques: image processing -- telescopes 1 INTRODUCTION High angular resolution X-ray imaging is always demanded by astrophysics and solar physics, for example to study black holes near event horizon, relativistic jets of super mas- sive black holes, as well as solar flares and coronal activities. So far the best imaging tech- nology for X-ray observation is realized by grazing incidence reflection (Aschenbach 1985), which provides a very good angular resolution, for example down to 0.5 arcsec as in the case of the Chandra X-ray Observatory (Weisskopf et al. 2000). The diffraction limit of Chandra is about 8 miliarcsec at 6 keV; however slope errors and other surface irregu- larities that affect the angle of reflection prevent the angular resolution better than 0.5 arcsec. Further more, the grazing incidence reflection is limited by the working energy band, which hardly exceeds tens of keV. A possible way to improve the angular resolution beyond Chandra's requires a different technology, for example the diffractive-refractive X-ray optics proposed by Gorenstein (2007) or the Fourier-Transform imaging by Prince et al. (1988). As an alternative, a super-high angular resolution principle (SHARP) for a coded-mask X-ray imaging telescope concept is proposed here. The coded mask imaging technology, which has been reviewed by Zand (1996), is widely applied for X-ray observations, for example the INTEGRAL mission (Winkler et al. 2003) of ESA and the SWIFT mission (Gehrels 2004) of NASA. Both missions have angular resolution in tens of arcmin, since the distances between the masks and the detectors are limited by the dimensions of the satellites. In principle super-high angular resolution better than Chandra can be achieved by a coded-mask telescope with sub- millimeter size of mask pinholes and long mask-detector distances, which can be tens, even hundreds of meters with the formation-flying technology (for example proposed for XEUS (ESA 2001) of ESA) or the mast technology (for example applied in Polar mission of NASA). The X-ray diffraction effect in a coded-mask system has been commonly thought to be negligible due to high photon energies in hard X-ray and gamma-ray range as well as relatively small mask-detector distance. Whereas for coded-mask telescope with a super high angular resolution in the milli-arc-second range, the X-ray diffraction effect could not be neglected, especially at low energy end (in the range of several keV), which might degrade the imaging performance remarkably, as pointed out by Prince et al. (1988) and Skinner (2004). In this work, we study the diffraction effect in SHARP with numerical computations, and propose a diffraction-interference cross correlation reconstruction method and demonstrate the feasibility of coded mask imaging beyond ⋆ E-mail: [email protected] SHARP - A super-high angular resolution principle 3 the diffraction limit of single pinhole. A potential application for solar observations is also shortly discussed. 2 THE CODED-MASK PRINCIPLE AND DIFFRACTION-INTERFERENCE RECONSTRUCTION 2.1 The coded-mask principle and SHARP In the limit of geometrical optics, i.e., the diffraction effect is not considered, the ba- sic concept of coded-mask imaging is shown in Fig. 1. The coded-mask camera has a mask on top of a position-sensitive detector with a mask-detector distance D. Two point sources project the mask pattern onto the detector plane. The shift and the strength of projections encode the position and the flux of sources separately. The detection of the X-ray flux can be described with Equ. 1 (Fenimore et al. 1978), P = O ∗ M + Nnoise, (1) where O is the X-ray flux spatial distribution, M is the encoding pattern of mask and ∗ means cross correlation. For the commonly applied cross correlation method, a matrix G is used for reconstruction, as shown in Equ. 2, O′ = P ∗ G = O ∗ (M ∗ G) + Nnoise ∗ G, (2) where O′ is the estimation of X-ray flux spatial distribution, if M ∗ G is a δ function. The encoding pattern M is often chosen so that its auto-correlation function is a δ function. Therefore G = M is commonly applied. The width of the Point Spread Function (PSF) obtained with Equ. 2 is defined as ∆i = dm D (dm is the size of the mask element). The cyclic optimal coded-mask configuration (Zand 1996) is also employed here. The coded mask or the mask pattern consisted of four basic patterns is about four times the size of the detector, and one basic pattern cyclically extends about twice along the width and length directions of the mask. In the next sections, simulations are done with the following assumptions: the basic pattern is totally random; the detector pixel size is equal to the mask element size, both of which are in square shape; the mask and the detector are assumed to be idealized, i.e., the mask has no thickness, the opaque mask elements block the X-ray completely and one incident event can only be recorded in one detector pixel. In this work, dm = 50 µm and D = 50 m are chosen to be the basic parameters for the simulations of SHARP, which means ∆i = 0.2 arcsec in sub-arcsec range. Since the simulations employ two dimensional FFT (please refer to Sec. 2.2) which requires tremendous computing resources, the basic pattern simulated contains only 200 × 200 elements, i.e., the reconstructed image contains the same pixels and each pixel is 0.2 × 0.2 arcsec2. The fully coded Field-Of-View (FOV) is 40 × 40 arcsec2. 4 C. Zhang & S. N. Zhang Fig. 1 The basic concept of coded-mask imaging, in the limit of geometrical optics, i.e., the diffraction effect is negligible. Two point sources illuminate a position-sensitive detector plane through a mask made of many square pinholes. The detector plane thus records two projections of the mask pattern. The shift and the strength of projections encode the position and the flux of the sources separately. The PSF width is defined as ∆i = dm D . 2.2 The diffraction-interference cross correlation method As shown in Fig. 2, A is the mask modulation function, which is 1 or 0 with idealized mask. r is a vector from point (x0, y0) on the mask to the (x, y) on the detector. k is the wave vector. Since there are many pinholes in the mask, both diffraction and interference will take place on the detector plane. The diffraction-interference is described with Fresnel- Huygens principle (Lindsey 1978) as, ER(x, y) = C Z ∞ ∞ Z ∞ ∞ Ei(x0, y0)A(x0, y0) eikr r dx0dy0, (3) where Ei is the amplitude distribution of incident photons, which is described with the plane wave in our case as Ei = Ai ∗ eik·r. The flux distribution is proportional to E2 i . C = −i/λ is constant. In our case the dimension of the mask is far less than the mask- detector distance D, we have r ≃ D + x2+x2 ER(x, y) = C2 Z ∞ ∞ Z ∞ 2D (x2+y2), λD eikDe ik 2D (x2 0). Equ. 4 is the Fourier transform approximation of Fresnel- M e−2iπ(fxx0+fy y0)dx0dy0, fy = y/λD, M = . Therefore Equ. 3 becomes fx = x/λD, Ei(x0, y0)A(x0, y0)e where C2 = −i ik 0+y2 0+y2+y2 2D 0 −2xx0−2yy0 ∞ (4) Huygens principle, which can be simulated with an FFT algorithm. The conditions for the validity of this approximation are discussed by Lindsey (1978), and are satisfied in our simulations. The X-ray diffraction can not be neglected at the low energy end as in Fig. 3, which shows the simulated flux distribution on the detector plane with a 50 µm square pinhole in front of it (50 m away) for 1.24 keV photons (λ = 1 nm). The flux spreads much larger than 50 µm (each pixel in Fig. 3 is 50 × 50 µm2) due to the diffraction effect. SHARP - A super-high angular resolution principle 5 Fig. 2 Monochromatic X-ray with wave vector k illuminating onto a mask in the x0-y0 plane produces a diffraction pattern in the x-y plane (Lindsey 1978). If the reconstruction matrix G in the cross correlation method is still chosen to be equal to M (so-called geometrical reconstruction matrix, which is indicated as GO) at the low energy end, the diffraction effect would degrade the reconstructed image remarkably. Fig. 4a shows the reconstructed image of a monochromatic on-axis point source (λ = 1 nm) obtained by geometrical matrix. The reconstructed image is normalized so that the total integral flux is one photon, and all the reconstructed images in this work are normalized in the same way if not specified. The point source is hardly identified with the Fresnel diffraction stripes clearly observed. Since the X-ray probability wave through different mask open elements interferes with each other, the mask spatial information should be encoded into the diffraction-interference pattern on the detector. We then choose the reconstruction matrix G at certain energy as the normal incident diffraction- interference pattern (so-called diffraction-interference matrix, which is indicated as GDI ) on the detector plane. Fig. 4b shows the reconstructed image obtained with GDI of the same source in Fig. 4a, and the point source can be clearly identified although slight Fresnel stripes still exist. This new reconstruction method is called diffraction-interference cross correlation method (DICC), which can be described with Equ. 5, O′ = P ∗ GDI = O ∗ (M ∗ GDI ) + Nnoise ∗ GDI . (5) GDI can be obtained by numerical simulation as shown in Equ. 4 or from actual mea- surements. The angular resolution of DICC is beyond the single pinhole (the single mask open element) diffraction limit as indicated in Fig. 4. 6 C. Zhang & S. N. Zhang Fig. 3 The simulated flux distribution on the detector plane for 1.24 keV pho- tons (λ = 1 nm) with a 50 µm square pinhole in front of it (50 m away). The flux spreads much larger than 50 µm caused by diffraction as one pixel is 50×50 µm2. Therefore this diffraction effect must be included when reconstructing the image of the incoming X-ray with cross-correlation effect. Fig. 4 The reconstructed images of a monochromatic (λ = 1 nm) on-axis point source: (a) the cross-correlation with the GO, the diffraction degrades the reconstruction dramatically, because the diffraction pattern shown in Fig. 3 has not been taken into consideration; (b) with DICC, the point source can be clearly identified although slight Fresnel diffraction stripes still exist. The DICC can thus largely overcome the degradation of the imaging performance due to the diffraction of each pinhole. 3 THE SIMULATION RESULTS We have simulated the reconstructed images at several energies from 1.24 keV (λ = 1 nm) to 6.2 keV (λ = 0.2 nm), where DICC is valid. To illustrate the main properties of SHARP - A super-high angular resolution principle 7 this method, we mainly discuss the imaging performance of DICC around 1 nm, which is also compared with that in the limit of geometrical optics. 3.1 Angular resolution We calculate the reconstructed image of two point sources with the same flux (Poisson fluctuations are not considered here). The angular resolution is defined as the minimum angular distance between these two sources, which can be separated by the 50% contour line of the maximum flux on the reconstructed image. Fig. 5a shows the reconstructed images in the limit of geometrical optics for two point sources with angular distance 1.5∆i = 0.3 arcsec; the two sources can not be separated. However in Fig 5b two sources with angular distance 1.6∆i = 0.32 arcsec can just be separated. Therefore the angular resolution is 0.32 arcsec in the limit of geometrical optics, i.e., at high energy end (above about 10 keV) with diffraction negligible. With the same procedure, the angular resolu- tion of DICC for λ = 1 nm photons is obtained as 0.36 arcsec, i.e., the two sources can not be separated in Fig. 5c, but can be separated in Fig. 5d. This means the diffraction effect degrades the angular resolution by only about 12.5%, a significant improved com- pared to that in Fig. 3. We also calculated the reconstructed images with DICC for λ = 1 nm point sources at different locations inside the FOV, but no observable reconstruction degradation is found compared with the on-axis source case. 3.2 Source location accuracy The reconstructed images for an on-axis point source is simulated 10000 times with 1000 photons recorded (Poisson fluctuations are considered). The source location of each simulation is calculated by fitting the reconstructed image to the PSF. The calculated source location distribution with DICC for λ = 1 nm is shown in Fig. 6a, which is concentrated within 0.1∆i = 0.02 arcsec. The same simulation is also done in the limit of geometrical optics, which is shown in Fig. 6b. Therefore the source location accuracy is better than 0.1∆i = 0.02 arcsec for 1000 detected photons. 3.3 System constraints Since the diffraction pattern on the detector plane is energy dependent, the diffraction matrix G is also energy dependent. In practice the detector can not identify two photons with an energy difference within the detector's spectroscopy resolution, i.e., in practice photons within a small energy band must share a common diffraction matrix. Fig. 7 shows the reconstructed images of several on-axis point sources with wavelengths around 1 nm, which all apply the diffraction matrix of 1 nm photons (marked as G1). The point source can be clearly identified. Obviously the imaging qualities are degraded by applying the 8 C. Zhang & S. N. Zhang Fig. 5 The angular resolution of the simulated system: (a) and (b) are in the limit of geometrical optics; (c) and (d) are obtained with DICC for λ = 1 nm sources. In (a) and (c) two sources can not be separated. However in (b) and (d) two sources can be separated. The angular resolution is 0.32 arcsec in the limit of geometrical optics and degraded to 0.36 arcsec for λ = 1 nm sources. Table 1 The Degradation of the Angular Resolution with Shared Reconstruction Matrix Wavelength (nm) 0.95 0.97 0.99 1 1.01 1.03 1.05 Angular resolution (arcsec) 0.39 0.38 0.37 0.36 0.37 0.39 0.41 G1 instead of their own diffraction matrixes. Tab. 1 gives the angular resolution at several wavelengths around 1 nm by applying G1 as reconstruction matrix instead of their own. For the 1.05 nm on-axis source, the angular resolution is degraded to 2.05∆i = 0.41 arcsec by applying G1. Based on the detector energy resolution, the system constraint is considered. Assuming the detector pixel size dd, the difference δd of the main diffraction stripe diameters of photons with wavelength difference δλ is δd = δλ ∗ D dm . (6) SHARP - A super-high angular resolution principle 9 Fig. 6 The calculated source location distribution of the reconstructed images for an on-axis point source, which is simulated 10000 times with 1000 photons recorded: (a) with DICC for λ = 1 nm photons; (b) in the limit of geometrical optics. The source location accuracy is better than 0.1∆i = 0.02 arcsec. If the spectroscopy resolution of the detector is δλ, spatially the detector does not need to distinguish δd for photons with wavelength difference δλ, δd ≤ f ∗ dd. (7) Here f ≥ 1 is a constant and depends upon the mask open fraction. For a 50% open mask, our simulations indicates that 1 ≤ f ≤ 2. For a mask fraction much less than 50%, f > 2 is possible. Then with known basic parameters, i.e., dm and D, the requirement for detector spectroscopy resolution is In practice dm and D are chosen to match the detector performance, which means δλ ≤ f ∗ ∆i ∗ dd. or dm ≥ δλ f ∗ ∆i , D = dm ∆i ≤ dd ∆i ≤ δλ (∆i)2f . (8) (9) (10) Therefore the larger the f is, the more compact the system can be. Here f = 1 is chosen for convenience, with which δλ = 0.05 nm, dm = 50 µm and D = 50 m satisfies the system constraint. The spectroscopy resolution 0.05 nm requires about 120 eV energy resolution @λ = 1 nm, which can be provided by modern silicon imaging detectors like CCD or DEPFET (Struder 2000). 4 DISCUSSION AND CONCLUSION By applying DICC, the system angular resolution is no more limited by single pinhole diffraction limit at low energy end, but limited by detector spectroscopy performance. 10 C. Zhang & S. N. Zhang Fig. 7 The reconstructed images of (a) 0.9 nm (b) 0.95 nm (c) 1.05 nm and (d) 1.1 nm photons, which all apply the same diffraction-interference matrix G1 (the diffraction-interference matrix of 1 nm photons). Tab. 2 shows a brief comparison of SHARP with several X-ray astronomy missions. The advantage of SHARP is quite obvious: excellent angular resolution and easily fabricated mask which does not require high accuracy mechanical fabrication of X-ray grazing lens. However the detector effective area of a coded-mask telescope is smaller than the its actual area due to the encoding pattern. Further more each detector pixel receives background photons from all over the FOV. Therefore the sensitivity is limited by its large background and small effective area compared with that in the grazing incidence reflection case. Current detector technology can provide even better energy resolution than 0.05 nm @1 nm, which means the angular resolution discussed above can be improved further by increasing the mask-detector distance D as indicated in Equ. 10 or by decreasing the mask element size as indicated in Equ. 9. For example for a detector with 0.002 nm spectroscopy resolution (5 eV @ 1 keV), dm = 40 µm and D = 800 m satisfies the system constraint with ∆i = 0.01 arcsec. A potential application of SHARP will be the solar observation at 1-100 keV en- ergy band with sub-arcsec angular resolution. The main objectives may be the coronal SHARP - A super-high angular resolution principle 11 Table 2 The Comparison Between SHARP and Several Missions Mission SHARP Chandra Integrala RHESSI Hinode (Solar B)b Imaging Coded- Focusing Coded- Rotation Focusing technol- mask mask Modulation ogy & Fourier Transform Energy 1-100 keV Up to 10 15 keV-10 3-150 About to 6 band Energy 133 keV ACIS: MeV keVc keV 9 % @ 100 < 1 keVf < 280 eV resolution [email protected] about 148 keV @5.9 keVg keVd [email protected] keVe Angular 0.32-0.36 0.5 arcsech 12 arcmin 2.26 arc- 2 arcsec @ resolution arcsec seci 0.523keVj a IBIS onboard b XRT onboard c The upper layer d Take DEPFET as focal plane detector (Treis et al. 2006) e CCD S3 on board f Below 100 keV g Calculated based on the ENC < 30 el. h On-axis point source i The collimator # 1 j 68% source flux within 2 arcsec mass ejection (CME) and solar flares, including fine structures and evolutions of the solar flares, nonlinear solar flare dynamics, solar particle acceleration mechanism et al. Therefore applications of SHARP are foreseen to make significant progress on the study of solar high energy explosive events and the space weather forecast model. Another po- tential application is to make wide-field X-ray monitors with SHARP; its sub-arc angular resolution may allow the counterparts of gamma-ray bursts, X-ray bursts, black hole and neutron star transients to be identified without the requirement of subsequent follow-up observations of focusing X-ray telescopes. Acknowledgements We thank C. Fang, W. Q. Gan, J. Y. Hu, Z. G. Dai, X. D. Li, Y. F. Huang and X. Y. Wang for many interesting discussions and suggestions on SHARP and its potential applications. We are grateful to the anonymous referee for providing valuable comments and suggestions promptly. SNZ acknowledges partial funding support by the Ministry of Education of China, Directional Research Project of the Chinese Academy of Sciences under project No. KJCX2-YW-T03 and by the National Natural Science Foundation of China under grant Nos. 10521001, 10733010, 10725313 and 10327301. 12 References C. Zhang & S. N. Zhang Aschenbach B., 1985, Reports on Progress in Physics, 48 (5): 579-629 Fenimore E. E., Cannon T. M., 1978, Applied Optics, 17: 337-347 Gehrels N., 2004, New Astronomy Reviews, 48, 431-435 Gorenstein P., 2007, Advances in Space Research, 40 (8): 1276-1280 Lindsey C. A., 1978, J. Opt. Soc. Am., 68 (12): 1708 Prince T. A., Hurford G. J., Hudson H. S., Crannell C. J., 1988, Solar Physics, 118: 269-290 Skinner G. K., 2004, New Astronomy Reviews, 48: 205-208 Struder L., 2000, Nucl. Instr. and Meth. A, 454 (1): 73-113 The XEUS Telescope Working Group, 2001, X-ray evolving universe spectroscopy-the XEUS telescope, ESA SP-1253 Treis J., Fischer P., Halker O. et al., 2006, Nucl. Instr. and Meth. A, 568 (1): 191-200 Weisskopf M. C., Tananbaum H. D., Van Speybroeck L. P., O'Dell S. L., 2000, In: J. Truemper, B. Aschenbach, eds., Proceedings of SPIE, 4012, p. 2 Winkler C., Courvoisier T. J.-L., Cocco G. D. et al., 2003, A&A, 411, L1 Zand aperture Coded camera imaging J., 1992, 1996, concept [EB/OL], http://lheawww.gsfc.nasa.gov/docs/cai/coded.html This manuscript was prepared with the ChJAA LATEX macro v1.0.
astro-ph/0304496
1
0304
2003-04-28T12:21:57
Dynamics of magnetic flux tubes in close binary stars I. Equilibrium and stability properties
[ "astro-ph" ]
Surface reconstructions of active close binary stars based on photometric and spectroscopic observations reveal non-uniform starspot distributions, which indicate the existence of preferred spot longitudes (with respect to the companion star). We consider the equilibrium and linear stability of toroidal magnetic flux tubes in close binaries to examine whether tidal effects are capable to initiate the formation of rising flux loops at preferred longitudes near the bottom of the stellar convection zone. The tidal force and the deviation of the stellar structure from spherical symmetry are treated in lowest-order perturbation theory assuming synchronised close binaries with orbital periods of a few days. The frequency, growth time, and spatial structure of linear eigenmodes are determined by a stability analysis. We find that, despite their small magnitude, tidal effects can lead to a considerable longitudinal asymmetry in the formation probability of flux loops, since the breaking of the axial symmetry due to the presence of the companion star is reinforced by the sensitive dependence of the stability properties on the stellar stratification and by resonance effects. The orientation of preferred longitudes of loop formation depends on the equilibrium configuration and the wave number of the dominating eigenmode. The change of the growth times of unstable modes with respect to the case of a single star is very small.
astro-ph
astro-ph
Astronomy&Astrophysicsmanuscript no. ms3669 (DOI: will be inserted by hand later) October 25, 2018 3 0 0 2 r p A 8 2 1 v 6 9 4 4 0 3 0 / h p - o r t s a : v i X r a Dynamics of magnetic flux tubes in close binary stars I. Equilibrium and stability properties V. Holzwarth1,2 and M. Schussler1 1 Max-PlanckInstitut fur Aeronomie, Max-Planck-Str. 2, 37191 Katlenburg-Lindau, Germany e-mail: [email protected] 2 School of Physics and Astronomy, University of St. Andrews, North Haugh, St. Andrews KY16 9SS, UK e-mail: [email protected] Received gestern; accepted morgen Abstract. Surface reconstructions of active close binary stars based on photometric and spectroscopic observations reveal non- uniform starspot distributions, which indicate the existence of preferred spot longitudes (with respect to the companion star). We consider the equilibrium and linear stability of toroidal magnetic flux tubes in close binaries to examine whether tidal effects are capable to initiate the formation of rising flux loops at preferred longitudes near the bottom of the stellar convection zone. The tidal force and the deviation of the stellar structure from spherical symmetry are treated in lowest-order perturbation theory assuming synchronised close binaries with orbital periods of a few days. The frequency, growth time, and spatial structure of linear eigenmodes are determined by a stability analysis. We find that, despite their small magnitude, tidal effects can lead to a considerable longitudinal asymmetry in the formation probability of flux loops, since the breaking of the axial symmetry due to the presence of the companion star is reinforced by the sensitive dependence of the stability properties on the stellar stratification and by resonance effects. The orientation of preferred longitudes of loop formation depends on the equilibrium configuration and the wave number of the dominating eigenmode. The change of the growth times of unstable modes with respect to the case of a single star is very small. Key words. binaries: close -- stars: magnetic fields -- stars: spots -- stars: activity -- stars: imaging 1. Introduction Owing to their rapid rotation and deep convection zone, evolved (sub-)giants and cool main sequence components in synchronised close binaries like RS CVn and BY Dra systems typically show high levels of magnetic activity (Strassmeier et al. 1993), e.g., in the form of large starspots and enhanced chromospheric and coronal emission in the UV and X-ray spectral ranges. Analyses of light curves with in- version techniques allow to derive the characteristic proper- ties of spots like their number, position, size, and temperature (Eaton & Hall 1979; Rodon`o et al. 1986; Budding & Zeilik 1987). Utilising rotation-modulated, asymmetric spectral line profiles of spectro- and spectropolarimetric observations, in- version techniques like the Doppler Imaging (Vogt & Penrod 1983; Vogt et al. 1987) and Zeeman-Doppler Imaging (Semel 1989; Donati et al. 1992, 1999) reveal information about tem- perature and magnetic field inhomogeneities on the stellar sur- face, which are interpreted as cool spots caused by emerg- ing magnetic fields. Recent observations are complemented by photometric data which cover up to several decades and thus enable the determination of cyclic variations (Rodon`o et al. 1995, 2000). Surface reconstructions of close, fast-rotating binaries show extended starspots which cover large fractions of the stellar surface (O'Neal et al. 1998; Lanza et al. 2001) and fre- quently occur, in contrast to the case of the Sun, also at inter- mediate, high and even polar latitudes. Non-uniform spot dis- tributions in longitude indicate the existence of preferred longi- tudes (PL), where spots occur more frequent or last longer than at other longitudes (Henry et al. 1995; Jetsu 1996). Preferred longitudes are often found to be separated by about 180◦, i.e., in opposite active quadrants of the hemisphere, but their ori- entation with respect to the companion star is not unique. In short-period systems with rotation periods . 1 d active regions tend to prefer the quadrature longitudes (Zeilik et al. 1989, 1990a,b, 1994; Ol´ah et al. 1994; Heckert et al. 1998), i.e., the longitudes perpendicular to the line connecting both stellar cen- tres. In systems with longer periods the situation is less clear: several systems show PL at the substellar point, its antipode or other fixed longitudes with respect to the companion star (e.g., Lanza et al. 1998, 2001; Ol´ah et al. 2002), whereas other stars reveal a longitudinal migration of the spotted regions with respect to the companion star, which is usually ascribed to differential rotation (Strassmeier 1994; Rodon`o et al. 1995, 2000). Considering the dependence of PL on the orbital period, 2 Holzwarth & Schussler: Flux tubes in close binaries I. Heckert & Ordway (1995) find a transition from fixed PL for orbital periods less that one day to migrating and eventually no PL at all for more separated systems with periods exceed- ing about 10 days, leading to the suggestion that tidal effects might influence the spot distribution. However, observations of close binaries with giant components show that spot clusters at migrating PL may also occur in systems with longer orbital pe- riods (Berdyugina & Tuominen 1998; Berdyugina et al. 1999), giving rise to the assumption that the formation and orientation of clusters are not exclusively dependent on the binary separa- tion alone. In this and a subsequent paper we study the influence of tidal effects on the dynamics and evolution of magnetic flux tubes in active binary components, assuming -- in analogy to the case of the Sun -- that starspots are formed by erupting flux tubes which originate from the bottom of the stellar convec- tion zone. The present work delves into the equilibrium and linear stability properties of flux tubes stored in the convective overshoot region below the convection zone proper in order to examine whether the influence of the companion star is able to trigger rising loops at preferred longitudes which may penetrate to the convection zone above. The decomposition of small per- turbations of an equilibrium flux tube in terms of eigenmodes and the determination of the corresponding eigenfrequencies are accomplished by a linear stability analysis which extends earlier treatments by Ferriz-Mas & Schussler (1993, 1995) and Holzwarth & Schussler (2000). In contrast to rotating single stars, in binary stars the existence of the companion star breaks the rotational symmetry. Consequently, the dynamics of flux tubes is not only subject to the tidal force but is additionally affected by the deviation of the stellar structure from spherical symmetry. The linear stability analysis deals with the response of equilibrium flux tubes to small perturbations inside the over- shoot region, whereas the growing displacements of unstable loops have to be followed by non-linear numerical simulations. The evolution of flux loops rising through the convection zone and the resulting surface distribution upon their eruption is de- scribed in a following paper. In Sect. 2 we briefly summarise the underlying (solar) ac- tivity model based on erupting magnetic flux tubes and outline the binary model. Sect. 3 describes the equilibrium properties and Sect. 4 gives the results of the linear stability analysis of toroidal flux tubes. Sect. 5 contains a discussion of the results and Sect. 6 gives our conclusions. penetrate from the convection zone above. While Hale's po- larity rules and the East-West-orientation of bipolar sunspot groups suggest equilibrium flux rings parallel to the equato- rial plane, overshooting gas motions cause perturbations which can trigger the onset of an undulatory (Parker-type) instabil- ity (Spruit & van Ballegooijen 1982; Ferriz-Mas & Schussler 1993, 1995). If a tube segment is lifted upward, a net mass downflow from the crest increases the density contrast with re- spect to the environment. Once a critical magnetic field strength is exceeded the resulting buoyancy surpasses the opposing magnetic curvature force, the loop continues to rise through the convection zone, and eventually erupts at the stellar sur- face (Parker 1955; Moreno-Insertis 1986). Upon emergence it causes the various signatures of magnetic activity like active re- gions, dark spot groups, bright faculae, chromospheric activity, X-ray bright coronal loops, and flares. The buoyancy-driven mechanism implies a crucial dependence on the stellar strat- ification, viz. the superadiabaticity, δ = ∇ − ∇ad. In the case of the Sun, magnetic flux tubes with field strengths of about 105 G are required to obtain theoretical results which are in accordance with observational constrains like flux emergence latitudes, tilt angles, and asymmetries of bipolar spot groups (Moreno-Insertis 1992; D'Silva & Choudhuri 1993; Fan et al. 1994; Schussler et al. 1994; Caligari et al. 1995, 1998). The quantitative elaboration of the applied flux tube model is also consistent with the emergence of polar spots on rapidly rotat- ing young stars (Granzer et al. 2000) and with the conspicu- ous decline of coronal X-ray emission in giant stars across the 'coronal dividing line' (Holzwarth & Schussler 2001). We consider the magnetic flux in the convection zone in the form of isolated magnetic flux tubes, which are embedded in a field-free environment. The stellar convection zone is treated as an ideal plasma with vanishing viscosity and infinite con- ductivity, implying the concept of frozen-in magnetic flux. All calculations are carried out in the framework of the 'thin-flux- tube approximation' (Roberts & Webb 1978; Spruit 1981): the diameter of a circular flux tube is assumed to be small with re- spect to the other relevant length scales of the problem like the pressure scale height, radius of curvature, or the wave length of a perturbation along the tube. As a consequence of very short signal travel times it remains in lateral (total) pressure equilib- rium with its environment. 2. Model assumptions 2.1. The'fluxtubeparadigm' 2.2. Binarymodel It is widely accepted that sunspots and bipolar spot groups are caused by erupting magnetic flux tubes which origi- nate from the base of the solar convection zone. The mag- netic field is believed to be amplified in the rotational shear layer (tachocline) near the base of the convection zone and stored in the form of toroidal flux tubes in the convective overshoot region (Moreno-Insertis et al. 1992; Schussler et al. 1994). This stably stratified subadiabatic layer at the inter- face to the radiative core is generated by gas motions which Observations indicate that the orbital motion of binary sys- tems with periods T . 10 d is effectively synchronised with stellar rotation and typically exhibit very small eccentricity (Duquennoy & Mayor 1991; Strassmeier et al. 1993). Hence, we consider a close, detached, synchronised and circularised binary system, for which the orbital and stellar rotation periods are equal and the spin axes of the components are perpendicu- lar to the orbital plane. The active star with mass M⋆ and the companion star with mass Mco = qM⋆ move on circular orbits Holzwarth & Schussler: Flux tubes in close binaries I. with a period of a few days. According to Kepler's third law, they are separated by the constant distance 3 ey ze Ω ✶M CM a Mco e x r λ ϕ f (r, φ, λ) = f (Ψ⋆ + Ψtide) ≈ f (Ψ⋆) + Ψtide 3. Stationary equilibrium R⊙! ≃ 4.21 (1 + q)1/3 M⋆ a M⊙!1/3(cid:18) T d(cid:19)2/3 . (1) Our reference binary system consists of two solar-type stars (M⋆ = Mco = 1 M⊙, i.e. stellar mass ratio q = 1) with a period of T = 2 d. The active star is described by a perturbed sin- gle star (here, standard solar) model, whereas the companion star is considered as an idealised point mass. In the case of the Sun, the differential rotation profile affects the stability proper- ties only slightly (e.g., Ferriz-Mas & Schussler 1993). Owing to tidal interactions and synchronisation, the differential rota- tion in close binary stars is much weaker than in the solar case (Donahue 1993; Rudiger et al. 1998), so that we assume solid- body rotation. For systems with orbital periods of a few days, tidal effects are sufficiently small to be treated in lowest-order perturbation theory. Using spherical coordinates in a co-rotating frame of reference (radius r, latitude λ, azimuth φ, with the companion star in the direction φ = 0), the effective potential, Ψeff = Ψ⋆ + Ψtide, is the sum of the spherical gravitational potential of the star, Ψ⋆(r), and the potential of the tidal perturbation1, Ψtide(r, φ, λ) = −ǫ3 GM⋆ 2r 5 2 " 1 + q! cos2 λ − q q cos2 λ cos 2φ# + O(cid:16)ǫ4(cid:17) , + 3 2 where ǫ = r/a is the expansion parameter and G the gravita- tional constant. The potential Ψtide gives rise to the tidal accel- eration gtide(r, φ, λ) = ǫ3g⋆(r)(cid:2)er + 3q (er · ea) ea − (1 + q) (er · eΩ) eΩ(cid:3) + O(cid:16)ǫ4(cid:17) , (3) where er, eΩ, and ea are unit vectors according to the directions indicated in Fig. 1. Following Eq. (3), the component of the tidal acceleration in azimuthal direction eφ, for example, is 3 2 (4) (cid:16)gtide · eφ(cid:17) = − g⋆(r)ǫ3q cos λ sin 2φ + O(cid:16)ǫ4(cid:17) . Our model includes the deviation of the stellar structure from spherical symmetry arising from the tidal acceleration gtide. For a not to strongly deformed star, we suppose that the stellar structure is still described appropriately by functions of the effective potential, Ψeff, only, i.e. f (r, φ, λ) = f (Ψeff), where f stands for any stellar quantity (like gas pressure, density,. . .). This permits an analytical expression for the tidal deformation, namely (2) = f0 1 + = f0 1 + ¯r r H f r H f Ψtider GM⋆! + r r H f d f dΨ(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)r cos 2φ! , 1 Equations (2) and (3) also contain the centrifugal potential and acceleration, respectively, although these contributions do not strictly belong to tidal interactions but occur in rotating single stars as well. Fig. 1. Geometry of the binary system. The origin of the co- rotating frame of reference lies in the centre of mass, CM. The stability analysis is carried out in spherical coordinates (r, φ, λ) with respect to the centre of the active star, M⋆. with the dimensionless coefficients ¯r = ǫ3 1 q! cos2 λ − q# , 2" 1 + 5 2 r = ǫ3 3 4 q cos2 λ , (6) (7) and the scale height H f = −(d ln f /dr)−1. Equation (5) includes the rotational flattening and the tidal (ellipsoidal) deformation, with tidal bulges centred on the line connecting the two stars. The deviation of a quantity, ∆ f = f (r, φ, λ) − f0(r), from the single star stratification, f0, depends on its local scale height, H f . Both values, f0 and H f , are taken from the spherical model at radius r. The quality of the approximation can be estimated by considering the ratio between the geometrical deformation, ≈ (8) Ψtide Ψ⋆ = ¯r + r cos 2φ + O(ǫ4) , δr r and the smallest scale height H f of all quantities f . Since in the overshoot region, the supposed storage location of mag- netic flux, the scale height of the superadiabaticity, Hδ, is con- siderably smaller than of any other quantity (see Tab. 1), the tidal influence on the stability properties of flux tubes is mainly controlled by the variation of δ (Holzwarth & Schussler 2000). For our reference system (T = 2 d, M⋆ = M⊙ and q = 1), Eq. (1) yields a separation a ∼ 8 R⊙ and ǫ3 . 10−3. The tidal effects are of small magnitude only, but they imply a significant qualitative difference, since both Eqs. (2) and (3) exhibit a pe- riodic azimuthal dependence, which breaks the axial symmetry present in the single star problem. (5) We consider magnetic flux rings in mechanical stationary equi- librium, which are embedded in the overshoot region and par- allel to the equatorial plane. The tubes are non-buoyant and in lateral pressure equilibrium with the environment. The inward directed magnetic curvature force is compensated by the out- ward directed Coriolis force, which arises from a prograde flow inside the tube with a velocity v in excess of the stellar solid- body rotation. The equilibrium tube forms a torus with constant 4 Holzwarth & Schussler: Flux tubes in close binaries I. Q1 ϕ toroidal flux tube toward the V 1 V2 companion star deformed overshoot region Q2 Fig. 2. Sketch of a toroidal equilibrium flux tube embedded in the tidally deformed overshoot region. The flux ring cuts through different equipotential surfaces, which implies a pe- riodic variation of the environment along the tube: the values of the stratification at the points on the line of centres, V1,2, correspond to deeper layers than those at the quadrature points, Q1,2. radius of curvature, R0 = r0 cos λ0, where r0 and λ0 are the ra- dial and latitudinal equilibrium positions, respectively. Since the flux ring is situated in a tidally deformed star (Fig. 2), the internal density, ρ (= ρe, where ρe is the density of the environ- ment), depends, according to the approximation in Eq. (5), on the azimuth as ρ(φ) ρ0 = 1 + ¯r0 r0 Hρ0 + r0 r0 Hρ0 cos 2φ + O(ǫ4) . (9) This entails a periodic variation of the magnetic field strength, (10) (11) B(φ) B0 = 1 − M2 α0 ρ0 (cid:17) − M2 (cid:16) ρ(φ) α0 ρ(φ) ρ0 ! , and internal flow velocity, v(φ) v0 = 1 − M2 α0 (cid:16) ρ(φ) ρ0 (cid:17) − M2 α0 , where Mα0 = v0/ca0 is the Alfv´enic Mach number, i.e., the flow velocity in units of the Alfv´en velocity, ca0 = B0/p4πρ0. Quantities with index '0' refer to a comparable axisymmetric equilibrium flux tube in a slowly rotating single star, where the tidal and centrifugal acceleration as well as the deformation of the stellar structure are negligible (e.g., Moreno-Insertis et al. 1992). For flux tubes in mechanical equilibrium, these quanti- ties obey the relation s1 + ca0 R0Ω!2 v0 = R0Ω > 0 . − 1 (12) Table 1. Parameters at equilibrium depth r0 of the reference configuration equilibrium depth gas pressure pressure scale height density density scale height superadiabaticity superad. scale height rotation binary separation expansion parameter deformation parameters (at equator, λ = 0) r0 = 5.07 · 1010 cm = 0.73 R⊙ pe0 = 4.31 · 1013 dyn/cm2 Hp0 = 5.52 · 109 cm ρe0 = 0.15 g/cm3 Hρ0 = 9.21 · 109 cm δ0 = −9.77 · 10−7 Hδ0 = 4.43 · 108 cm T = 2 d a = 8.41 R⊙ ǫ3 = 6.53 · 10−4 ¯r = 8.16 · 10−4 r = 4.90 · 10−4 In the following, we consider flux rings in the middle of the overshoot region. Table 1 gives values of the characteristic pa- rameters of this reference configuration for a model of a 1 M⊙ star. 4. Linear stability analysis 4.1. Determinationandinterpretationofeigenmodes Once all quantities of the external stratification are determined by the equilibrium depth r0, the stability properties of the flux ring depend only on its latitude, λ0, and magnetic field strength, B0. Lagrangian displacements are decomposed in eigenmodes, Ξ(φ, t) = ξ(φ) exp (iωt) , (13) characterised by the eigenfrequency ω and the eigenfunction ξ(φ) = (ξt, ξn, ξb)T (in the co-moving trihedron with the local tangent, normal, and binormal unit vectors, [t, n, b], respec- tively). For unstable eigenmodes with increasing amplitudes τ = 1/ℑ(ω) provides the characteristic growth time. The dis- placement vector ξ is determined by the linearised equations of motion, ξ′′ +(cid:16)Mφ + iωMφt(cid:17) ξ′ +(cid:16)Mξ + iωMt − ω2Mtt(cid:17) ξ = 0 , where primes denote derivatives with respect to the azimuthal coordinate, φ (see Appendix A). Taking the tidal force, Eq. (3), and the azimuthal variation of the equilibrium quantities, Eqs. (9) -- (11), into account, the coefficient matrices depend periodically on φ: (14) M(φ) = M(φ + π) = M0 + ǫ3Mc cos 2φ + ǫ3Ms sin 2φ + O(ǫ4) . (15) Closed flux tubes require ξ(φ + 2π) ≡ ξ(φ) and ξ′(φ + 2π) ≡ ξ′(φ), suggesting the ansatz ξ(φ) = ξ(φ) exp (imφ) (16) exp (imφ) . (17) ∞ =  Xk=−∞ ξk exp (ikφ) In analogy to the single-star problem, a phase factor with the azimuthal wave number m has been separated. Due to the φ- dependent contributions in Eq. (15), the substitution of Eq. (17) into Eq. (14) yields the 3-term recursion formula Lm+k ξk−2 + Cm+k ξk + Rm+k ξk+2 = 0 (18) ∀k . , Holzwarth & Schussler: Flux tubes in close binaries I. 5 k=-2 m=2 k=0 k=2 n= -2 -1 0 1 2 3 4 5 k=-2 k=0 m=1 k=2 Fig. 3. An eigenmode in the binary problem is represented by a spectrum of coupled wave modes with wave numbers n = m ± k, k = 0, 2, 4, . . . Due to the weak coupling (∝ ǫ3) the eigenmode can still be characterised by the wave number m of its dominating constituent. The contributions of wave modes with k , 0 lead to non-axisymmetric modifications, whereas wave modes with k ≥ 4 (indicated by dots) are negligible. In contrast to the single-star problem, where an eigenmode cor- responds to a single azimuthal wave number m, i.e., ξ0 = const. and ξk = 0 for k , 0, an eigenmode here consists of a spectrum of coupled wave modes n = m+k with k = 0, ±2, ±4, . . ., whose amplitudes ξk are recursively related to each other through Eq. (18). This relation is illustrated in Fig. 3. Since the cou- pling is weak [∝ O(ǫ3)], the amplitudes decrease quite fast with increasing k, so that more distant wave modes (here, with k ≥ 4) do practically not contribute to the eigenmode. In most cases the amplitude ξ0 is much larger than ξ±2, so that the re- sulting eigenmode can still be characterised by the wave num- ber m. The linear eigenfunctions characterise the structure of small displacements of flux tubes in the overshoot region prior to their rapid, non-linear rise through the convection zone. Owing to the stellar stratification, the most important aspect of an eigenfunction is its component in the radial direction, ξr = ( ξ · er): the larger the radial displacements, the larger are the changes in the environment of the corresponding tube segments, which become located in either more stable or more unstable layers. Assuming that larger radial displacements to- ward more unstable layers favour the penetration of a growing flux loop to the convection zone, we use the envelope, ξr(φ), of the radial component of the (normalised) eigenfunction as a measure for the onset of rapidly rising flux loops at azimuth φ. Since for eigenmodes with ℑ(ω) < 0 the magnitude of the radial envelope increases exponentially in time, Ξr(φ, t) = ξr(φ) exp(cid:2)−ℑ(ω)t(cid:3) , an unstable flux tube in the overshoot region will eventually enter the superadiabatical region. Its actual displacement de- pends, however, also on the azimuthal wave number m, the fre- quency ℜ(ω), and the phase relation (19) arg Ξr(φ, t) = mφ + arg ξr(φ) + ℜ(ω)t , (20) corresponding to an azimuthal propagation of the tube's crest along the envelope, Eq. (19), with the phase velocity vp(φ) = − ℜ(ω)R0 m 1 + 1 m d dφ arg ξr(φ)!−1 . (21) Fig. 4. Relative deviation of the growth time, (τbin − τsin)/τsin, in parameter domains which have been unstable in the rotating single-star problem, i.e., τsin , 0. The radial envelope thus represents a kind of 'statistical weight', which is applicable either to a large ensemble of comparable flux tubes with similar equilibrium properties and phase shifts evenly distributed in the range 0 ≤ arg Ξr < 2π, or to an individual flux tube with a growth time much longer than the wave frequency, τ ≫ 2π/ℜ(ω). If an eigenmode is dom- inated by the wave mode m and only slightly modified by the adjacent wave modes m ± 2, i.e., ξr,±2 ≪ ξr,0 := 1, the radial envelope is approximately given by (cid:12)(cid:12)(cid:12) ξr(φ)(cid:12)(cid:12)(cid:12) = (cid:12)(cid:12)(cid:12) ξr,0 + ξr,2ei2φ + ξr,−2e−i2φ + . . .(cid:12)(cid:12)(cid:12) ξr,2(cid:12)(cid:12)(cid:12) cos (2φ + α2) +(cid:12)(cid:12)(cid:12) ≃ 1 +(cid:12)(cid:12)(cid:12) + O( ξ2 r,±2) , ξr,−2(cid:12)(cid:12)(cid:12) cos (2φ − α−2) (22) with α±2 = arg ξr,±2. The eigenfunctions thus inherit the prop- erties of the underlying problem and exhibit a π-periodicity in azimuthal direction. Assuming that the shape of ξr(φ) repre- sents a measure for the onset of rising loops, Eq. (22) implies the existence of preferred longitudes at the maxima of ξr. 4.2. Growthtimesandeigenfrequencies The influence of the companion star modifies the stability prop- erties of flux tubes by breaking the axial symmetry. The result- ing quantitative difference in growth times τ with respect to a rotating single star is very small; as illustrated in Fig. 4, the rel- ative deviations in all cases are considerably smaller than 1%. Qualitative differences occur, however, for equilibrium config- urations which had been stable in the single-star problem. In previously stable parameter domains new types of instabilities with very long growth times lead to an 'instability background' consisting of extended 'plateaus' which are pervaded by nar- row 'ridges' (see Fig. 5). The 'plateaus' are due to the tidal force and the azimuthal variation of the density along the equilibrium flux ring. Using the lateral (total) pressure equilibrium and the variation of the magnetic field strength, Eq. (10), it can be shown that the in- ternal specific entropy, S , is also a function of φ and varies 6 Holzwarth & Schussler: Flux tubes in close binaries I. Fig. 5. Growth times of the 'instability background' due to tidal effects. In addition to the Parker-type instabilities there are localised ridges with τ & 103 . . . 105 d traversing extended plateaus with τ > 105 d. Because of the limited resolution of the parameter grid in the figure the very narrow ridge at λ ≃ 1◦ appears to be discontinuous. approximately as 1 cp ∂S ∂φ ∼ 1 β d ln pe dφ ∼ ǫ3 β r Hp,e , (23) where β = 8πp/B2 is the ratio between the gas pressure and the magnetic pressure, and Hp,e is the local pressure scale height. Although we have r/Hp,e ≃ 9, the azimuthal variation is nev- ertheless small compared to the radial variation of the stellar stratification, since the relevant field strengths considered here lead to values in the regime β ∼ 104 . . . 106 ≫ 1. Due to its velocity excess, v, a gas element inside the tube would have to change its entropy to preserve the equilibrium conditions. But since there is no mechanism available to accomplish this entropy change, the azimuthal variation leads to a loss of the mechanical equilibrium and to very slow monotonic displace- ments of the flux tube out of its toroidal configuration. For fast- rotating stars, the internal flow velocity v is small and thus the growth times of plateau instabilities are typically several 104 days (up to τ ∼ 107 d). The effects of the azimuthal density variation and tidal force along the tube are negligible because the stability properties are dominated by the azimuthal varia- tion of the superadiabaticity, δ(φ) = δ(r0) 1 + δr Hδ! , (24) owing to its relative small scale height in the overshoot region (with Hδ/Hp ≃ 0.08, see Tab. 1). If all azimuthal variations other than that of δ are omitted in the stability analysis, the plateau instabilities vanish and only the ridge instabilities re- main. The 'ridge' instabilities with typical growth times of sev- eral thousand days owe their existence to a resonant coupling of wave modes with different wave numbers. As an example, the top panel of Fig. 6 shows the growth times of both Parker- type and background instabilities along a cut at λ = 1.64◦ in Fig. 6. Growth times τ (top) of Parker-type, ridge- and plateau- instabilities at λ = 1.64◦ (see Fig. 5). The ridge instabilities (grey shading) are due to resonant interactions of adjacent wave modes with m = 1 and 3 (middle) and m = 0 and 2 (bottom), respectively, when the frequencies ℜ(ω) are similar. Dotted lines show the frequencies of the uncoupled modes in the corre- sponding single-star problem. Dashed and dotted-dashed lines indicate unstable modes with ℑ(ω) , 0. Fig. 5. The panels below show the frequencies ℜ(ω) of the wave modes involved. In the single-star problem (dotted lines, dashed-dotted if unstable) the wave modes corresponding to different values of the azimuthal wave number, m, are indepen- dent. But the eigenmodes in the binary problem consist of a spectrum of coupled wave modes (see Fig. 3) with wave num- bers n = m, m ± 2, m ± 4, . . .: modes of wave number m with frequency ℜ(ω) stimulate excitations of the wave modes m ±2, which, supposed ℜ(ω) is close to the eigenfrequency of the ad- jacent wave mode, results in a resonant amplification and even- tually in the onset of an instability (solid lines in Fig. 6, dashed if unstable). In the following, we dismiss instabilities with very long growth times τ > 104 d (≈ 27 years), since they are not ex- pected to play any role for the eruption of magnetic flux tubes in active stars. Holzwarth & Schussler: Flux tubes in close binaries I. 7 Fig. 7. Azimuth φmax of the radial envelope maxima. Due to the azimuthal π-periodicity only the maximum in the interval 0 ≤ φmax < π is shown, i.e., between the points V1 and V2 on the line of centres implying the second maximum at φmax + π. The dashed framed region is the instability domain of eigenmodes with a dominating wave number m = 2. Fig. 8. Relative peak-to-peak variation, ∆ ξr, of the radial en- velope. For ridge instabilities at low latitudes, ∆ ξr is close to unity, and even in the instability island at λ ∼ 15 . . . 40◦ and low field strengths the resonant wave mode coupling leads to considerable values of ∆ ξr. 4.3. Eigenfunctions The eigenfunctions, ξ, inherit the property of the underlying problem and exhibit a π-periodicity in azimuthal direction, im- plying a φ-dependent probability that a loop penetrates to the superadiabatic part of the convection zone and rises rapidly to- ward the surface. The preferred longitudes, φmax (and φmax + π), of loop penetration are determined by the maxima of the ra- dial envelope, ξrmax = ξr(φmax). They are shown in Fig. 7 as a function of the equilibrium latitude λ0 and field strength B0. Background instabilities with growth times exceeding 104 d have been neglected, leaving behind only a fraction of a ridge at low latitudes and a plateau at λ & 60◦ with τ & 3 · 103 d. This high-latitude feature is caused by the slow loss of equilib- rium due to the internal flow described above, which exhibits only small displacements in radial direction (here, ξr/ ξφ ∼ O(10−2)). In contrast to the case of stable eigenfunctions, for which the envelope maxima are located exactly at the 'symme- try points', Q1,2 or V1,2, the orientation of ξrmax of unstable eigenfunctions show a phase shift with respect to these points. For instabilities at equatorial and low latitudes dominated by the wave number m = 1, the maxima are located in the vicin- ity of V1,2, i.e., around the line connecting both stellar centres, whereas for modes with m = 2 they are distributed in a broad interval around the quadrature points Q1,2; preferred longitudes of loop penetration thus depend on the equilibrium parameters (B0, λ0) as well as on the dominating wave mode. The radial envelope, ξr, is used as a measure for the prob- ability of loop penetration into the convection zone, subject to the normalisation ξrmax := 1. The larger the relative peak-to- peak variation, ∆ ξr := ξrmax − ξrmin ξrmax , (25) the larger is the possibility for the transition of a loop from the overshoot region to the convection zone around φmax. Figure 8 shows ∆ ξr as a function of λ0 and B0. For B & 105 G the peak-to-peak variation is . 5% at intermediate latitudes and up to about 20% near the equator. Typically ∆ ξr decreases with increasing field strengths since the stability properties strongly depend on βδ ∝ δ/B2. High field strengths reduce the effect of the azimuthal δ-variation, so that magnetic flux tubes with smaller field strengths are, in general, more susceptible to tidal effects. Particularly for ridge instabilities and instability islands at low field strengths, ∆ ξr is significant and can even reach values close to unity, so that radial displacements are almost suppressed near the envelope minima. The approximation used in Eq. (22) is not applicable here, since the dominance of the wave mode ξr,0 is not always given. Ridge instabilities are due to resonant interactions of adjacent wave modes (see Sec. 4.2) with comparable amplitudes while the amplitudes of all other modes are small. The case ξr,0, ξr,2 ≫ ξr,k, ∀k , 0, 2, for ex- ample, yields the radial envelope ξr2 ∝ 1 + 2 ξr,0 ξr,2 ξr,02 + ξr,22 cos (2φ + α2 − α0) , (26) with α0/2 = arg ξr,0/2. For ξr,0 ≈ ξr,2, the radial envelope drops to very small values at longitudes which are determined by the phase factors α0 and α2. The low magnetic field strengths cor- responding to these particular instabilities imply, however, long growth times which are probably not relevant for the evolution of rising flux loops in stars. 4.4. Parameterdependence We examine the dependence of the stability properties on tidal effects in various binary systems which are characterised by the orbital period, T , the mass ratio, q, and the mass of the primary star, M⋆. Equations (3) and (5) show that the magnitude of the azimuthal variation caused by tidal effects is governed by the 8 Holzwarth & Schussler: Flux tubes in close binaries I. Fig. 9. Comparison of growth times, τ, for systems with var- ious orbital periods, T . The isolines mark the growth times τ = 1, 5, 10, 50, 100 (thick line), 500, 1000 (dashed line), and 5000 d. Instabilities with τ ≥ 104 d are omitted. term ǫ3q = 4π2 G q r3 1 + q M⋆T 2 . (27) ≃ 10−2 q 1 + q! r R⊙!3 M⋆ M⊙!−1(cid:18) T d(cid:19)−2 Here, the solar-type model is retained, which fixes M⋆ and r0. The variation of Eq. (27) due to a different value of r0 is negli- gible since the equilibrium radius is restricted to the rather thin overshoot layer, which is prescribed by the stellar model. First we consider binary systems with orbital periods in the range T = 0.75 . . . 10 d, mass ratio q = 1, and, follow- ing Eq. (1), separations a ≃ 4.4 . . . 25 R⊙. Since the expansion parameter covers the range ǫ3 = 4.5 · 10−3 . . . 2.5 · 10−5, the analytical approximation for the stellar stratification, Eq. (5), is well applicable even for the shortest period. A comparison of growth times for different T is shown in Fig. 9. A smaller or- bital period has a stabilising effect, which is indicated by the shifting of corresponding isolines of growth times to higher field strengths. This effect is reinforced by the rotational flat- tening of the star since flux rings at a given, constant equilib- rium depth, r0, are embedded in slightly deeper and more sub- adiabatic layers of the overshoot region. The tidal effects on τ are nevertheless small and primarily affect the background instabilities, leading to shifts of ridges and plateaus. With de- creasing system period, the high-latitude plateau grows toward smaller field strengths since the azimuthal variation of the den- sity and tidal force along the flux ring become larger; however, the growth times remain very large. The orientations of the ra- dial envelope maxima do not show a significant dependence on orbital period and are well represented by the reference case with T = 2 d (Fig. 7). The relative peak-to-peak variations, ∆ ξr, are shown in Fig. 10. ∆ ξr increases significantly with Fig. 10. Relative peak-to-peak variation, ∆ ξr, of the radial en- velope for different system periods, T . decreasing orbital period, which results in strongly preferred longitudes of loop penetrations to the convection zone in short- period systems. The tidal effects are very effective for flux rings with small field strengths, particularly for those which are li- able to ridge instabilities with ∆ ξr close to unity. For Parker- type instabilities with short growth times, the relative peak-to- peak variation remains much smaller. Figure 10 also shows that the peak-to-peak variation strongly decreases with increasing system period for T & 3 d and eventually becomes insignifi- cant. Considering a system with period T = 2 d, we find that a variation of the mass ratio in the range q = 0.1 . . . 10, has only marginal influence on the stability properties. In the do- main of Parker-type instabilities, the relative changes of the growth time and of the peak-to-peak variation with respect to the reference case are well below 0.1% and 10%, respectively. Furthermore, the orientations of the envelope maxima exhibit hardly any dependence on the mass ratio. In summary, the growth times show only a weak depen- dence on the binary parameters, except for the instability back- ground and the rotational stabilisation of flux tubes at short sys- tem periods. The latter effect is mainly due to the conservation of angular momentum and not innate to the tidal effects. The relative peak-to-peak variation of the radial envelope exhibits a considerable dependence on the orbital period T , while the de- pendence on the mass ratio q is almost negligible. Finally, the orientations of the envelope maxima are largely unaffected by variations of the binary parameters. 5. Discussion Tidal effects and high rotation rates in close binaries cause characteristic changes of both the equilibrium and the stability properties of magnetic flux tubes stored in the convective over- shoot region. Owing to tidal effects, a flux ring in stationary Holzwarth & Schussler: Flux tubes in close binaries I. 9 equilibrium experiences periodic variations of its environment in azimuthal direction, which modify its characteristic eigen- modes. To first order, the azimuthal variation of a given quan- tity depends on its local scale height. Since the stability proper- ties are dominated by the superadiabaticity, which is the most strongly depth-dependent quantity in the overshoot region, tidal effects are mediated mainly by the azimuthal variation of δ. The growth times of Parker-type instabilities with τ . 1000 d are only marginally affected by the presence of the com- panion star, whereas for instabilities with τ & 1000 d the influ- ence of tidal effects is considerably stronger. This applies par- ticularly to the instability background, since ridge and plateau instabilities owe their existence to the binary character and are most susceptible to changes of the tidal effects. However, since the eigenfrequency is a global parameter describing the whole flux tube and thus representing an average over the properties of all tube segments, the dependence of ω on the binary pa- rameters is dominated by the axially symmetric influence of stellar rotation and not so much by the non-axially symmetric tidal interactions: for axially symmetric changes of the equi- librium configuration, e.g., those due to the rotational flatten- ing, the changes are identical for each tube segment and thus may add up to a considerable deviation of ω, whereas for non- axial, periodic changes along the flux ring, like those due to tidal effects, the changes of individual tube segments mutually cancel and eventually average to only minor deviations of ω. The growth times of background instabilities are typically sev- eral thousand days and more and therefore very long compared to the system period as well as to the growth times which are relevant for rising flux tubes in the Sun. Whether these insta- bilities are important for loop formation at preferred longitudes seems questionable, although they possess the most conspicu- ously asymmetric eigenfunctions. In contrast to the modest effect on the eigenfrequencies and growth times, tidal effects can significantly alter the azimuthal structure of the eigenfunctions since the periodic variations along the tube affect the displacement of each segment indi- vidually. The resulting eigenfunction, ξ(φ), characterised by its radial envelope ξr, consists of a superposition of coupled wave modes and inherits the π-periodicity of the azimuthal variation of the equilibrium properties. In contrast to stable eigenmodes, whose envelope maxima are located either at the line of centres or at the quadrature points, the maxima of unstable eigenmodes typically exhibit a phase offset with respect to these points; this result is schematically illustrated in Fig. 11. The orientations of the maxima are not unique but depend on the equilibrium configuration, predominantly on its latitude and on the wave number of the dominating wave mode. Considering Fig. 2, one might expect that the quadrature points would be favoured for rapid loop growth, since they are located in less stable layers of the overshoot region and nearest to the superadiabatic con- vection zone. However, the Parker-type instability is driven by a downflow from the loop summit, which increases the density contrast with respect to the environment. Following the equa- tion of continuity, a net downflow implies an asymmetry of mass flux at the loop summit, i.e., the amount of plasma flowing down in the preceding slope of the loop outbalances the mass flux coming up in the following loop slope. The azimuthal vari- Fig. 11. Schematic illustration of flux tubes in a single star (left) and in a binary star (middle and right). For the latter, the direction to the companion star is indicated by the central ar- row. The dashed line indicates the equilibrium flux ring, the grey line an actual displacement of the perturbed flux tube (here for a wave mode with m = 1), and the solid black lines the enve- lope, ξr, of the eigenfunction. In the middle panels, two stable modes are shown with the maxima of their radial envelope lo- cated either at the quadrature points (upper panel) or at the line of centres (lower panel); unstable modes in binaries typically exhibit phase offsets with respect to these points (right). ation of the internal equilibrium density and velocity, Eqs. (9) and (11), enhance the asymmetry in both slopes and thereby favour maximal displacements at intermediate longitudes. In contrast, the extrema of stable eigenmodes prefer either the quadrature points or the directions along the line connecting both stellar centres since at these 'symmetry points' (concern- ing the internal equilibrium density and velocity structure) the downflow and thus buoyancy effects can be minimised. The peak-to-peak variation, ∆ ξr, serves as a measure for the probability of loop penetration into the convection zone, based on the assumption that the loop segments lifted furthest toward the superadiabatic region are the most likely to leave the overshoot region. The radial envelope determines the dis- placement of the perturbation wave and thus the formation of rapidly growing loops inside the convection zone is favoured around its maxima at φmax and φmax + π. In the case of close binaries with orbital periods of a few days, the peak-to-peak variation of fast growing Parker-type instabilities is of the or- der of several percent, up to about 20%. In the case of ridge instabilities, radial displacements are strongly suppressed at some longitudes, giving rise to distinct preferred longitudes in the vicinity of φmax and φmax + π. Although it can not be excluded that ridge instabilities contribute to the non-uniform spot distribution, their importance for the formation of pre- ferred longitudes at the stellar surface is questionable since the relevant equilibrium configurations are restricted to rather small parameter domains (B0, λ0), where instabilities have large growth times. Based on the parameter study in Sect. 4.4, we would predict that preferred longitudes appear (in bina- ries with solar-type components) for orbital periods T . 5 d, 10 Holzwarth & Schussler: Flux tubes in close binaries I. are expressed in terms of the displacement vector ξ(φ) (see, e.g., Ferriz-Mas & Schussler 1993). The substitution of the cor- responding expressions in the equation of motion followed by a linearision with respect to ξ yields Eq. (14), with the coefficient matrices3 which is in agreement with the results of Heckert & Ordway (1995). However, in some binaries with longer periods (consist- ing of giant components) preferred longitudes have also been observed (Berdyugina & Tuominen 1998). The linear stability analysis establishes the existence of preferred longitudes at the bottom of the convection zone. However, the determination of the actual longitude where an individual loop enters the convection zone is not possible since our procedure only allows statistical predictions. The tendency to constitute preferred longitudes is instead discernible for a large ensemble of similar flux tubes, supposed that the environ- ment conditions in the overshoot region, e.g., the amount and distribution of overshooting convective motions, are uniform and unaltered for a sufficiently long time. Preferred longitudes at the bottom of the convection zone are not necessarily sim- ply related to the observed preferred longitudes of starspots at stellar surfaces, because the evolution of rising loops through the convection zone is still subject to tidal effects2. The pre- ferred longitudes of flux eruption at the stellar surface based on the flux tube model described above have to be determined by non-linear numerical calculations. Results of such calculations are presented in the subsequent paper II. 6. Conclusion The equilibrium and stability properties of toroidal magnetic flux tubes in the overshoot region of an active component of a close binary system are considerably affected by tidal effects. The breaking of the axial symmetry due to the presence of the companion star leads to azimuthal variations along the tube of the equilibrium configuration and of the probability of the pen- etration of unstable loops to the convection zone. This demon- strates the existence of preferred longitudes at the bottom of the convection zone, which are well pronounced for orbital periods of a few days but strongly fade with increasing period. The ori- entations of the preferred longitudes depend on the equilibrium latitude, the field strength, and the dominating wave mode of the unstable eigenmode. Despite their rather small magnitude, tidal effects are nevertheless capable of influencing the dynam- ics of magnetic flux tubes considerably because they are en- hanced by the sensitive dependence of the stability properties on the superadiabaticity of the local environment and by the resonant coupling between interacting wave modes. Acknowledgements. Volkmar Holzwarth thanks Prof. S. K. Solanki and Prof. F. Moreno-Insertis for valuable discussions and the Max-Planck-Institut fur Aeronomie in Katlenburg-Lindau and the Kiepenheuer -- Institut fur Sonnenphysik in Freiburg/Brsg. for financial support during the accomplishment of this work. Appendix A: Numerical treatment The perturbations of all physical and geometrical quantities (magnetic field, flow velocity, density, position, curvature,. . . ) arising from the small perturbations of an equilibrium flux ring 2 If the tube evolution would be linear throughout the convection zone, the maxima of the radial envelope would, in fact, also determine preferred longitudes of flux eruption at the surface. Mφt = − Mtt = − E 1 Ω 1 4 E 1 − M2 α M2 α 1 Ω2 gn g∗ 1 + M2 α Y 1 + F3 1 + F2 Mt = 1 2 1 Ω gb g∗ Z Mξ = F1 gt gn gb 0 0 0 where the constant matrices M2 α t −gt 0 g′ gt 2gn gb − 1! Y + F1 Mφ = F1 g∗  0 gb 0  g∗  0 0 0  0 0 1 , Y = 0 0 0 (cid:0)1 − M2 α(cid:1) 1 h(1 − ∇) M2 α +(cid:16)1 − M2 α(cid:17) ∆(cid:16)1 − M2 γ 1 2! and the abbreviations 1 γ E = 0 1 0 −1 0 0 F2 = −F 2 1 0 0 0 1 0 F3 = F 2 1 ∆ = βδ − γ2 2 2 − γ F1 = ρe pe g∗ κ 1 1 (A.1) (A.2) (A.3) (A.4) (A.5) , (A.6) (A.7) (A.8) (A.9) (A.10) 1 g2 ∗  gtgt gtgn gtgb gngt gngn gngb gbgt gbgn gbgb  , , Z = 0 0 1 0 0 0 −1 0 0 gt g∗ α(cid:17) ∇i γ have been used. The components gx = (geff · x) , x ∈ [t, n, b] of the effective gravitational acceleration geff are expressed in the co-moving trihedron. Owing to the azimuthal variation of the equilibrium quan- tities induced by the presence of the companion star, the coef- ficient matrices (A.2) -- (A.5) can be expressed in the form of Eq. (15) and comprise small contributions which are, in low- est order, π-periodic in the longitude φ. Following the ansatz in Eq. (17), these contributions cause the coupling of wave modes ξn with wave numbers n = m, m ± 2, m ± 4, . . ., which is ex- pressed either by the 3-term recursion formula in Eq. (18) or by the Hill-type algebraic system  Lm−2K Cm−2K Rm−2K . . . . . . . . . O Lm Cm Rm . . . . . . . . . O Lm+2K Cm+2K Rm+2K {z } 3 In the case β ≫ 1. H(ω) = 0 , (A.11) ξ−2K ... ξ0 ... ξ2K    Holzwarth & Schussler: Flux tubes in close binaries I. 11 with K → ∞ and the block matrices Ln = Rn = and 1 1 2h(n − 2)(cid:16)iMφ,c + Mφ,s − ω(cid:16)Mφt,c − iMφt,s(cid:17)(cid:17) +Mξ,c − iMξ,s + ω(cid:0)iMt,c + Mt,s(cid:1) −ω2(cid:0)Mtt,c − iMtt,s(cid:1)i , 2h(n + 2)(cid:16)iMφ,c − Mφ,s − ω(cid:16)Mφt,c + iMφt,s(cid:17)(cid:17) +Mξ,c + iMξ,s + ω(cid:0)iMt,c − Mt,s(cid:1) −ω2(cid:0)Mtt,c + iMtt,s(cid:1)i , Cn = −n2E − nωMφt,0 − ω2Mtt,0 + Mξ,0 +i(cid:16)nMφ,0 + ωMt,0(cid:17) . The eigenfrequencies ω are determined by the roots of the dispersion relation, det(cid:2)H(ω)(cid:3) = 0. A treatment of the case K → ∞ is not practical, but since the coupling between ad- jacent wave modes is small [L, R ∼ O(ǫ3)], the wave mode spectrum can be truncated at a finite number of constituents, here K = 3. A verification of the eigenfrequency and the deter- mination of the eigenfunctions ξ(φ) is based on the following approach. Let the independent solutions x(i)(φ), i = 1, . . . , n of a system x′ = A(φ)x with A(φ + π) = A(φ) (A.15) represent the fundamental matrix X(φ) = (cid:0)x(1), . . . , x(n)(cid:1) with det[X(φ)] , 0. Furthermore, we have X(φ + 2π) = X(φ)M, where M is the constant transition matrix with det(M) , 0. Since an arbitrary solution of Eq. (A.15) can be represented by a constant vector c in the form x(φ) = X(φ)c, it follows that x(φ + 2π) = X(φ + 2π)c = X(φ)Mc. Periodic solutions with x(φ + 2π) ≡ x(φ) thus require Mc = c and an eigenvalue of the transition matrix M equal to unity. If the fundamental matrix fulfils the initial condition X(0) = E, the transition matrix is calculated by integrating over a period of 2π, i.e., X(2π) = M. In our case, the transition matrix M and its eigenvalues depend on the eigenfrequency, ω. If an eigenfrequency determined by the dispersion relation above does not entail Mc = c, ω is iteratively improved and the eigenfunction ξ(φ) determined. References Berdyugina, S. V., Berdyugin, A. V., Ilyin, I., & Tuominen, I. 1999, A&A, 350, 626 Berdyugina, S. V. & Tuominen, I. 1998, A&A, 336, L25 Budding, E. & Zeilik, M. 1987, ApJ, 319, 827 Caligari, P., Moreno-Insertis, F., & Schussler, M. 1995, ApJ, 441, 886 (A.12) (A.13) (A.14) D'Silva, S. & Choudhuri, A. R. 1993, A&A, 272, 621 Duquennoy, A. & Mayor, M. 1991, A&A, 248, 485 Eaton, J. A. & Hall, D. S. 1979, ApJ, 227, 907 Fan, Y., Fisher, G. H., & McClymont, A. N. 1994, ApJ, 436, 907 Ferriz-Mas, A. & Schussler, M. 1993, Geophys. Astrophys. Fluid Dyn., 72, 209 -- . 1995, Geophys. Astrophys. Fluid Dyn., 81, 233 Granzer, T., Schussler, M., Caligari, P., & Strassmeier, K. G. 2000, A&A, 355, 1087 Heckert, P. A., Maloney, G. V., Stewart, M. C., et al. 1998, Astron. J., 115, 1145 Heckert, P. A. & Ordway, J. I. 1995, Astron. J., 109, 2169 Henry, G. W., Eaton, J. A., Hamer, J., & Hall, D. S. 1995, ApJS, 97, 513 Holzwarth, V. & Schussler, M. 2001, A&A, 377, 251 Holzwarth, V. & Schussler, M. 2000, Astron. Nachr., 321, 175 Jetsu, L. 1996, A&A, 314, 153 Lanza, A. F., Catalano, S., Cutispoto, G., Pagano, I., & Rodon`o, M. 1998, A&A, 332, 541 Lanza, A. F., Rodon`o, M., Mazzola, L., & Messina, S. 2001, A&A, 376, 1011 Moreno-Insertis, F. 1986, A&A, 166, 291 Moreno-Insertis, F. 1992, in NATO ASIC Proc. 375: Sunspots. Theory and Observations, ed. J. H. Thomas & N. O. Weiss (Kluwer Acad. Publ., Dordrecht), 385 -- 410 Moreno-Insertis, F., Schussler, M., & Ferriz-Mas, A. 1992, A&A, 264, 686 Ol´ah, K., Strassmeier, K. G., & Weber, M. 2002, A&A, 389, 202 Ol´ah, K., Budding, E., Kim, H.-L., & Etzel, P. B. 1994, A&A, 291, 110 O'Neal, D., Saar, S. M., & Neff, J. E. 1998, ApJL, 501, L73 Parker, E. N. 1955, ApJ, 121, 491 Roberts, B. & Webb, A. R. 1978, Sol. Phys., 56, 5 Rodon`o, M., Cutispoto, G., Pazzani, V., et al. 1986, A&A, 165, 135 Rodon`o, M., Lanza, A. F., & Catalano, S. 1995, A&A, 301, 75 Rodon`o, M., Messina, S., Lanza, A. F., Cutispoto, G., & Teriaca, L. 2000, A&A, 358, 624 Rudiger, G., von Rekowski, B., Donahue, R. A., & Baliunas, S. L. 1998, ApJ, 494, 691 Schussler, M., Caligari, P., Ferriz-Mas, A., & Moreno-Insertis, F. 1994, A&A, 281, L69 Semel, M. 1989, A&A, 225, 456 Spruit, H. C. 1981, A&A, 102, 129 Spruit, H. C. & van Ballegooijen, A. A. 1982, A&A, 106, 58 Strassmeier, K. G. 1994, A&A, 281, 395 Strassmeier, K. G., Hall, D. S., Fekel, F. C., & Scheck, M. 1993, A&AS, 100, 173 Caligari, P., Schussler, M., & Moreno-Insertis, F. 1998, ApJ, Vogt, S. S. & Penrod, G. D. 1983, Publ. Astron. Soc. Pac., 95, 502, 481 565 Donahue, R. A. 1993, PhD thesis, New Mexico State University Vogt, S. S., Penrod, G. D., & Hatzes, A. P. 1987, ApJ, 321, 496 Zeilik, M., Cox, D. A., de Blasi, C., Rhodes, M., & Budding, Donati, J.-F., Brown, S. F., Semel, M., et al. 1992, A&A, 265, E. 1989, ApJ, 345, 991 682 Zeilik, M., Cox, D. A., Ledlow, M. J., et al. 1990a, ApJ, 363, Donati, J.-F., Collier Cameron, A., Hussain, G. A. J., & Semel, 647 M. 1999, Mon. Not. Royal Astron. Soc., 302, 437 Zeilik, M., Gordon, S., Jaderlund, E., et al. 1994, ApJ, 421, 303 12 Holzwarth & Schussler: Flux tubes in close binaries I. Zeilik, M., Ledlow, M., Rhodes, M., Arevalo, M. J., & Budding, E. 1990b, ApJ, 354, 352
astro-ph/0203496
1
0203
2002-03-27T21:04:44
Powerful Water Masers in Active Galactic Nuclei
[ "astro-ph" ]
Luminous water maser emission in the 6_(16)-5_(23) line at 22 GHz has been detected from two dozen galaxies. In all cases the emission is confined to the nucleus and has been found only in AGN, in particular, in Type 2 Seyferts and LINERs. I argue that most of the observed megamaser sources are powered by X-ray irradiation of dense gas by the central engine. After briefly reviewing the physics of these X-Ray Dissociation Regions, I discuss in detail the observations of the maser disk in NGC 4258, its implications, and compare alternative models for the maser emission. I then discuss the observations of the other sources that have been imaged with VLBI to date, and how they do or do not fit into the framework of a thin, rotating disk, as in NGC 4258. Finally, I briefly discuss future prospects, especially the possibility of detecting other water maser transitions.
astro-ph
astro-ph
Powerful Water Masers in Active Galactic Nuclei Philip R. Maloney CASA, University of Colorado, Boulder, CO, 80309-0389 [email protected] Abstract Luminous water maser emission in the 616 − 523 line at 22 GHz has been detected from two dozen galaxies. In all cases the emission is confined to the nucleus and has been found only in AGN, in particular, in Type 2 Seyferts and LINERs. I argue that most of the observed megamaser sources are powered by X-ray irradiation of dense gas by the central engine. After briefly reviewing the physics of these X-Ray Dissociation Regions, I discuss in detail the observations of the maser disk in NGC 4258, its implications, and compare alternative mod- els for the maser emission. I then discuss the observations of the other sources that have been imaged with VLBI to date, and how they do or do not fit into the framework of a thin, rotating disk, as in NGC 4258. Finally, I briefly dis- cuss future prospects, especially the possibility of detecting other water maser transitions. Keywords: galaxies: Seyfert -- masers -- molecular lines -- radio lines: galaxies -- accretion, accretion disks 1 Extragalactic Water Masers Here I briefly review the discovery of powerful water maser sources in other galaxies; more detailed reviews, by active participants in the observation and interpretation of H2O megamasers using VLBI, are provided in Moran et al. (1995) and Greenhill (2001). 1.1 Water Maser Emission in Galactic Sources Maser emission in rotational lines of water -- usually the 616 → 523 transition at 22 GHz (1.35 cm) has been observed from Galactic sources for more than three decades (Cheung et al. 1969), from star-forming regions and late-type stars. The population inversion necessary to produce maser emission in this transition can be produced simply by collisions (e.g., shocks), due to the interactions between the rotational "ladders" of the H2O molecule (de Jong 1973; see Elitzur 1992 for a detailed recent discussion). The observed luminosities are typically LH2O ∼ 10−3 L⊙, although the most luminous 1 Galactic source, W49, occasionally reaches LH2O ∼ 1 L⊙ (Genzel & Downes 1979)1. Since the levels involved in the 22 GHz transition lie at energies E/k ≈ 600 K above the ground state, the gas temperature must be at least a few hundred K in order to excite the emission. 1.2 Extragalactic Sources The first water masers to be detected in other galaxies (e.g., M33: Churchwell et al. 1977) resembled W49, with LH2O ∼ 1 L⊙, and were typically found in galactic disks, in regions of star formation. These could be easily understood as extragalactic analogues of the Galactic water maser sources. However, a new and unexpected type of extragalactic H2O maser was discovered by dos Santos & Lepine (1979) in the edge- on spiral NGC 4945. With an isotropic luminosity LH2O ∼ 100 L⊙, this source -- which evidently arose in the galactic nucleus -- was five orders of magnitude more luminous than typical Galactic H2O masers, and was classed as a water "megamaser" (L ∼ 106 × LH2O(Galactic)). Four more megamasers were found over the next 5 years, including Circinus (Gardner & Whiteoak 1982), NGC 3079 (Henkel et al. 1984a,b) and NGC 1068 and NGC 4258 (Claussen, Heiligman, & Lo 1984). Typically the emission consisted of one to a few components with velocity widths δV ∼ few km s−1, spread over a total velocity range ∆V ∼ 100 km s−1 about the systemic velocity of the galaxy Vsys; an example (NGC 4945) is shown in Figure 1. Additional surveys over the next several years, mostly in late-type and infrared-luminous galaxies, failed to detect any new sources (e.g., Claussen & Lo 1986; Whiteoak & Gardner 1986). However, all 5 of the known megamasers shared common features, as first noted by Claussen et al. (1984): • They all were found in host galaxies with evidence of nuclear activity: three were classified as Seyferts or LINERs, and three showed unusual, extended radio emission or strong nuclear radio continuum emission; • The maser emission was centered on the nucleus; • In the cases of NGC 1068 and NGC 4258, the spatial diameter of the emitting regions were constrained by interferometric observations to be d < 3.5 and 1.3 pc, respectively (Claussen & Lo 1986). Taking a cue from these features of the galaxies containing H2O megamasers, Braatz, Wilson, & Henkel (1996, 1997), undertook a systematic survey of AGN for H2O megamasers. They observed 354 active galaxies (both Seyferts and LINERs). 1Maser luminosities are traditionally given as the isotropic value 4πD2F , where D is the source distance and F is the line flux. Since maser emission is highly anisotropic, this can overestimate the true luminosity of an observed maser feature. However, precisely because of the anisotropy, there are almost certainly many maser features that are not beamed in our direction, and so the isotropic luminosity based on the observed features may provide a reasonable estimate of the actual maser luminosity. 2 ) y J ( y t i s n e D x u F l 8 6 4 2 0 300 400 500 600 700 -1 800 Heliocentric Velocity (km s ) Figure 1: The spectrum of H2O maser emission from NGC 4945, showing the spread of emission with respect to the systemic velocity (indicated by the vertical bar). From Greenhill et al. 1997a. Ten new megamasers were found, raising the total to 16. An examination of the statis- tics in both distance-limited (cz < 7000 km s−1) and sensitivity-limited (LH2O > 2 L⊙) samples leads to a remarkable result: In the distance-limited sample: • 7% (10/141) Sy 2 detected • 7.5% (5/67) LINERs detected • 0% (0/57) Sy 1 detected For the sensitivity-limited sample, these percentages become: • 14% (10/73) Sy 2 detected • 10% (5/52) LINERs detected • 0% (0/30) Sy 1 detected The fraction of H2O megamasers in AGN is not small, provided that Type 1 Seyferts are excluded. Hence either: (1) Sy 1 don't possess nuclear molecular gas in which the physical conditions are suitable for masing, or (2) the megamasers in Sy 1 galaxies are systematically beamed away from us. 3 Since the Braatz et al. survey, eight more megamasers have been discovered2, bring- ing the current total to 24 (see the list in Moran, Greenhill, & Herrnstein 1999, plus Mrk 348 [Falcke et al. 2000b] and NGC 6240 [Hagiwara, Diamond, & Miyoshi 2002; Nakai, Sato, & Yamauchi 2002]). All of the sources are in galaxies classified as either Sy 2 (14/24) or LINERs (10/24). In all cases the emission is confined to the nucleus (although in the case of the merger NGC 6240, the location of the nucleus is somewhat problematic). The survey reported by Greenhill et al. (2002) detected only one new source out of 131 galaxies searched; however, only about 50 of these sources are classed as Sy 2 or LINERs. Nine megamaser sources have now been imaged using VLBI; the results will be discussed in §4 and 5. 2 The origin of water megamasers As mentioned above, the 616 → 523 transition of H2O at 22 GHz can be driven to mase by collisions, provided that the densities and temperatures are high enough >∼ 107 cm−3, T >∼ 400 K, respectively) to excite the levels involved, and that the (nH2 <∼ 1011 cm−3, levels deviate from thermodynamic equilibrium, which implies that nH2 and that the gas kinetic temperature is substantially higher than the temperature of the radiation field in the mid- to far-infrared, where the pure rotational transitions connecting the levels occur. In the Galactic sources of water maser emission, namely, star-forming regions and the atmospheres of late-type stars, there is little doubt that the necessary conditions are produced by shock waves. However, the unique associa- tion between water megamasers and active galactic nuclei suggests that there is another mechanism at work in many if not all of these sources. As I will show below, the best explanation for the majority of the sources is likely to be maser emission arising in dense gas that is being irradiated by X-rays from the AGN. Before discussing the maser emission, I will first digress briefly to discuss the characteristics of these X-Ray Dissociation Regions. 2.1 X-Ray Dissociation Regions Consider a power-law spectrum incident on a layer of dense gas: after spectral filtering through absorption in the gas, only X-rays, with energies E above a few hundred eV, are left. This is very significant for active galactic nuclei, because for typical AGN spectra, approximately equal amounts of energy per decade are available. Hence the hard X- ray (E >∼ 1 keV) luminosity Lx is roughly one-tenth of the bolometric luminosity Lbol. These hard X-rays can have a profound influence on the physical state of the ISM in a galaxy with an active nucleus, since photons with E > ∼ 1 kev have small cross-sections for absorption, and hence large mean free paths, but large E/photon, so that the local energy input rates when the photons are absorbed are substantial. Much of the locally deposited energy goes into heating the gas; the remainder goes into ionization and 2A database of galaxies that have been searched for 22 GHz water maser emission is maintained by Jim Braatz, and is available at http://www.gb.nrao.edu/∼jbraatz/H2O-list.html. 4 excitation (Maloney, Hollenbach, & Tielens 1996). For a typical (fν ∝ ν −α, with α ∼ 1) AGN X-ray spectrum, the ionization and heating rates fall with the column density N only as ∼ N −1. This is in marked contrast to the photodissociation regions (PDRs) produced by FUV photons (Tielens & Hollenbach 1985), in which the heating rates decline exponentially with column density since the absorption is produced by dust. This means that the column density in an XDR can be large, and therefore that the volume around a luminous AGN occupied by the XDR (i.e., where X-rays dominate the ionization and heating rates) can be substantial. For typical AGN spectra, the ionization and heating rates are always dominated by the lowest energy photons that have yet to be significantly attenuated. The physical state of the gas is described to lowest order by an effective X-ray ionization parameter ξeff, related to the usual ratio of X-ray flux to gas density, but including a scaling with column density that accounts for the optical thickness of the gas to the incident radiation. This scaling depends weakly on the spectral index but is generally close to N −1, as noted above (see Maloney, Hollenbach, & Tielens 1996 for details). 2.2 Water Maser Emission from Dense X-Ray Irradiated Gas in AGN High-pressure XDRs prove to be ideal locations for the production of luminous water maser emission (Neufeld, Maloney, & Conger 1994, hereafter NMC). Provided the gas temperature is above ∼ 400 K but is not so high (T ∼ 4000 K) that H2 is destroyed, the reaction network O + H2 → OH + H OH + H2 → H2O + H forms water very efficiently. This is also the approximate temperature criterion for collisionally exciting the 22 GHz transition, as discussed earlier. An X-ray-irradiated layer of dense gas generally undergoes a phase transition from warm, atomic gas to cooler, molecular gas. An example is shown in Figure 2. The temperature plummets from T ∼ 6000 K to T ∼ 1000 K at the transition, while the H2 abundance climbs from ∼ a few ×10−5 to ∼ ten percent. The water abundance relative to hydrogen <∼ 10−12) to ∼ a few ×10−6, and then climbs to jumps from negligible values (xH2O ∼ 10−4 before starting to decline as the temperature drops with increasing column density. (The initial rise in temperature following the transition is due to the effects of radiative trapping, which acts to reduce the effective cooling rate per unit volume as the optical depths in important cooling transitions increase; this effect is eventually outweighed by the continued decline in the gas heating rate with increasing N.) The physical conditions just past the phase transition are ideal for producing water maser emission over a very broad range of pressures, P/k ≈ 1010 − 1013 cm−3 K. Figure 3 (from NMC) shows a more detailed look at the temperature and water abundance as a function of depth (rather than column density) into an X-ray-irradiated slab. Also shown is the maser volume emissivity in saturation, Φsat. The maser volume emissivity 5 Figure 2: Thermal and chemical structure of an isobaric XDR. This model assumes an X-ray source with a hard X-ray (1 -- 100 keV) luminosity of 1043 erg s−1 and an X-ray spectral index α = 0.7 at a distance of 1 pc. The pressure has been fixed at P/k = 1011 cm−3 K. The gas temperature, electron fraction, and the abundances of several important species (relative to the total hydrogen density) are plotted as a function of column density into the XDR (bottom axis) and X-ray ionization parameter ξeff (top axis). 6 is simply the rate at which maser photons (in a particular transition) are generated per unit volume; this rate can never be larger than the rate when the maser is fully saturated (i.e., when the maser radiation itself controls the population inversion responsible for maser action; see for example Neufeld & Melnick 1991; Elitzur 1992). To the right of the vertical dotted line in this panel, the maser action is quenched by radiative trapping in the non-masing rotational transitions. However, this calculation ignored the role of dust in the radiative transfer. As pointed out by Collison & Watson (1995; see also Neufeld 2000), the presence of dust can be very important for these X-ray- powered masers, because dust absorption sets a lower limit to the escape probability for the far-infrared rotational transitions, and thus acts to inhibit quenching of the maser emission, provided that the dust is substantially cooler than the gas; otherwise the radiation from the dust would be important. This is unique to these XDR masers, because only there can the column densities become large enough that the far-infrared dust optical depths become significant. The importance of dust depends on the pressure in the XDR, and is negligible for P/k >∼ 1012 cm−3 K. This is because with increasing pressure the location of the transition to the molecular regime moves to smaller column density; since the column density at the transition sets the column density scale on which the ionization and heating rates decline, with increasing P the XDR becomes more compressed in column density, so the importance of dust in the radiative transfer diminishes. As shown by NMC, the resulting water maser luminosities from XDRs can be substantial, reaching LH2O ∼ 10s to 100s L⊙ per pc2 of irradiated area. Since this is the luminosity range seen in the megamaser galaxies, with the exception of the so-called "gigamaser" source (Koekemoer et al. 1995), which is an order of magnitude more luminous, it is evident that XDRs can explain the association of H2O megamasers with AGN, provided that areas of only ∼ 1 pc2 are being irradiated by hard X-rays from the AGN. Since the obscuring "tori" inferred to be present in Type 2 Seyferts (see Antonucci 1993 for a review) were also inferred to be ∼pc-scale objects, they naturally provided the requisite surface areas to explain the megamaser luminosities, without appeal to any special geometries. 3 Imaging 1: NGC 4258 Further progress in understanding H2O megamasers would depend on imaging the maser emission. With the advent of the Very Long Baseline Array (VLBA), it became possible to obtain very high resolution interferometric images of the 22 GHz emission in sufficiently bright megamaser sources. The first such galaxy to be observed was the nearby (D ≈ 7 Mpc), LINER spiral NGC 4258, which has a 22 GHz line luminosity LH2O ≈ 85 L⊙. Most strikingly, NGC 4258 was already known from single-dish obser- vations (Nakai et al. 1993) to show an enormous velocity spread in its maser emission, with ∆V up to ±1000 km s−1 with respect to the systemic velocity. Combined with the already established concentration of the emission to the nucleus from earlier VLA observations (Claussen & Lo 1986), this strongly suggested that material very deep in a potential well -- such as gas in proximity to a massive black hole -- could be involved 7 Figure 3: As Figure 2, except plotted as a function of depth rather than column density. Fx denotes the incident hard X-ray flux. In addition to temperature and water abundance, the bottom panel shows the maser emissivity Φsat (photons cm−3 sec−1); see the text for definition. From NMC. 8 in the emission. NGC 4258 was first observed using VLBI by Greenhill et al. (1995). Only the systemic emission was imaged; nevertheless, these observations already revealed some of the key features discussed below. Maser emission over the full velocity range was imaged using the VLBA by Miyoshi et al. (1995). The results, summarized in Figure 4 (from Bragg et al. 2000), are quite remarkable. Among the notable features: • The maser emission does not arise from the inner face of an X-ray-irradiated torus; instead, it is produced in a thin, warped disk or annulus, with inner and outer edges at 0.14 and 0.28 pc. This distance scale corresponds to an inner edge at 38,000 Schwarzschild radii for the mass of the central black hole (see below). The disk is extremely thin: the maser emission is unresolved in the vertical direction, implying that the thickness of the masing layer is less than 1015 cm (Moran et al. 1995). • The rotation curve is Keplerian to better than 1%. This places stringent limits on the nature of the central mass concentration (Maoz 1995): the central mass M = 3.9 × 107 M⊙ within 0.14 pc. This implies that the central mass density ρ > 3.4 ×109 M⊙. Combined with the limits on deviation from Keplerian motion, this rules out a stellar cluster as the central object. If the cluster were composed of stars or stellar remnants with masses above M ∼ 0.03 M⊙, the evaporation timescale tevap ≪ tH , the Hubble time. For a cluster made up of brown dwarfs or other low mass objects with M < 0.03 M⊙, the collisional timescale tcoll ≪ tH. In addition, a stellar density cusp around a low-mass black hole is also ruled out. Hence NGC 4258 provides compelling evidence for the existence of a supermassive black hole in a galactic nucleus. • Since we know the mass of the central object and the associated X-ray lumi- nosity (Makishima et al. 1994) quite accurately, we can make the first reliable determination of the fractional Eddington luminosity in an AGN, subject only to uncertainty in the hard X-ray to bolometric luminosity ratio. This turns out to be quite small, with L/LEdd ∼ 10−4. • Observations of centripetal acceleration (a ≈ 9 km s−1 yr−1: Haschick, Baan, & Peng 1994; Greenhill et al. 1995a; Nakai et al. 1995) or proper motion (µ ≈ 32 µas yr−1: Herrnstein 1997; Herrnstein et al. 1999) of systemic velocity features allow a purely geometric determination of the distance to NGC 4258 (D = 7.3±0.3 Mpc); both methods give the same answer. The tight limit on the distance quoted here comes from the data set analyzed in the last two of these references, which contains multiple epochs of observation from 1994 to 1997. The uncertainties are contributed nearly equally by the statistical errors (i.e., the precision with which a and µ can be determined, which is limited mainly by the difficulty in identifying and tracking individual features) and the uncertainties in the disk model (chiefly the projected disk angular velocity at the radius of the systemic features). • Observations of centripetal acceleration of the systemic and high-velocity fea- tures (the latter having a <∼ 1 km s−1 yr−1: Bragg et al. 2000) confirm that the 9 geometry inferred for the masers (shown in Figure 4) is correct; the scatter of the high-velocity features about the disk midline, as constrained by the upper limits on a, is only a few degrees, while the systemic velocity features lie within a very narrow range of radii (approximately 0.01 pc spread about a mean ra- dius R = 0.14 pc). This latter conclusion is supported by the essentially linear variation of velocity with impact parameter b for the systemic velocity features (with a gradient dv/db ≃ 0.26 km s−1 µas−1); this behavior is expected as the projection of orbital velocity onto our line of sight changes with impact parame- ter (Miyoshi et al. 1995). Calculation of the implied radius for the enclosed mass places the systemic velocity features at the same mean radius as inferred from the centripetal accelerations. • Compact continuum emission has been detected at 22 GHz (Herrnstein et al. 1997). The peak of the emission is located about 0.015 pc north of the dynamical center of the disk. Herrnstein et al. suggest that this is the base of the northern jet of the symmetric pair seen on parsec to kiloparsec scales in the radio and X- ray, and detect a much weaker southern component, at somewhat greater distance from the disk center, which they argue is attenuated by free-free absorption in the warm atomic layer on that side of the disk. Assuming the VLBI jet is intrinsically symmetrical, the failure to detect the southern component at the same offset (0.5 mas) as the peak in the northern jet suggests that the free-free optical depth is >∼ 2. Herrnstein et al. argue that the systemic velocity masers are amplifying the southern jet emission, an argument that is supported by the correlation of the maser flux density with the continuum flux density of the northern jet emission; this in turn implies that the jet is symmetric, as assumed. • By searching for Zeeman-splitting-induced circular polarization of the 22 GHz water maser emission, Herrnstein et al. (1998) derived an upper limit of 300 mG to the magnetic field strength in the disk at a radial distance of 0.2 pc. This upper limit is about four times larger than the value required for an equilibrium disk supported by magnetic pressure. However, it is still significant because one cannot rule out a priori the possibility that the masers trace a thin layer in a much thicker accretion disk. This upper limit to B can also be used to place an upper limit to the mass accretion rate through the disk, under the assumption of equipartition; unfortunately, this method at present is not restrictive enough to discriminate between competing models for the disk (see §3.2). • The angle over which the maser emission from NGC 4258, which is beamed in the plane of the disk, is detectable can be determined from the velocity extent of the emission near the systemic velocity, compared to the rotation velocity. This gives a beaming angle of 8◦ (Miyoshi et al. 1995). One can also estimate the observable solid angle from the magnitude of the disk warp, which has an angular thickness much greater than the thickness of the maser layer, and gives a "warp angle" of approximately 11◦. Using either number, the observed solid angle is consistent with all Seyferts and LINERS containing such maser disks, 10 but only those in which our line of sight intercepts the (nearly edge-on) disk are detectable. This suggests that the masing regions are either identical with, or physically associated with, the obscuring material (the TORUS3) around the AGN. 3.1 Deriving the Mass Accretion Rate The maser emission from NGC 4258 arises in what is quite clearly the accretion disk around the central black hole. It is perhaps remarkable that the first direct detection of an accretion disk around a supermassive black hole was obtained in a molecular transition, using ground-based telescopes. We can also use the H2O maser emission to infer the mass accretion rate through the disk (Neufeld & Maloney 1995; NM95). First, however, we must return briefly to the results of NMC. Figure 5 shows a phase diagram for dense, X-ray-irradiated gas. The abundance of water and the gas temperature are plotted as a function of an X-ray ionization P11), where the incident unattenuated X-ray flux parameter, here written as F5/(N 0.9 24 Fx = 105F5 erg cm−2 s−1, Natt = 1024N24 cm−2 is the attenuating column density to the source of X-rays, and the gas pressure P = 1011k P11 dyne cm−2. (This diagram was actually calculated by varying Fx, but the results will scale approximately with Natt and P as indicated.) Note that there are actually two main branches, corresponding to the warm, atomic phase and the cooler, molecular phase. Over a range of ioniza- tion parameter, both equilibria are present, allowing for the possibility of a two-phase medium with the warm atomic and cooler molecular phases coexisting. The important point is that the upper branch in the temperature plot (the warm atomic phase) does not exist for ionization parameters smaller than a critical value of about 6.5. This means that for a fixed flux and shielding column, there is a maximum pressure for which the warm atomic phase can exist; at pressures higher than this, the gas must be in the molecular phase. To apply this to NGC 4258, we need a model of the accretion disk. Since the masers appear to arise in a thin disk, model it as one: a Shakura-Sunyaev α−disk (Shakura & Sunyaev 1973), but with a warp (which is actually crucial, as shown below). In this model the kinematic viscosity ν is assumed to be proportional to the product of the sound speed cs and the disk scale height H, with the constant of proportionality des- ignated α (see Frank, King & Raine 1992 for a review of the theory of accretion disks). The actual source of viscosity has been a mystery for many years but is now generally believed to be provided by the magnetorotational instability (Balbus & Hawley 1991). In the obliquely illuminated region between 0.14 and 0.28 parsecs (the masing annulus), X- ray heating dominates over viscous heating by about an order of magnitude, and the disk is nearly isothermal. We therefore model the disk using the α prescription for viscosity, but the temperature is set by the X-ray heating rather than viscous heating Because of the warp, portions of the disk see the central X-ray source. 3Thick Obscuration Required by Unified Schemes: see Conway 1999. 11 Figure 4: The maser disk in NGC 4248, as inferred from VLBA observations. The top panel shows a typical spectrum of this system, with the direction of velocity drift of the features indicated. The second panel shows the projection of the maser spots on the plane of the sky. The third shows the velocities of the spots as a function of impact parameter; the solid line is a Keplerian rotation curve. The bottom panel shows the inferred structure of the system. From Bragg et al. 2000. 12 as in a standard α−disk. Ignoring self-gravity, the disk scale-height is H = cs Vorb R = 2.6 × 1014 T 1/2 3 R3/2 0.1 M 1/2 8 cm (1) where the disk temperature T = 103T3 K, the disk radius is R = 0.1R0.1 pc, and the mass of the central black hole Mbh = 108M8 M⊙. Hence the disk is expected to be extremely thin, as observed. The disk midplane pressure (P/k) is P = 1.3 × 1011 M8 M−5 αR3 0.1 cm−3 K. (2) where the mass accretion rate through the disk M = 10−5 M−5 M⊙ yr−1. If we now take the critical pressure derived from the results of NMC, with the flux expressed in terms of the luminosity of the central X-ray source and the disk radius, scaled to appropriate values for NGC 4258, we have that the critical pressure for molecule formation is Pcr ≈ 3.3 × 1010 L41 R2 0.1N 0.9 24 cm−3 K. Equating these two expressions gives the critical radius for molecule formation: Rcr = 0.04 ( M−5/α)0.81M 0.62 8 L0.43 41 µ0.38 pc (3) (4) where µ corrects for oblique illumination at an angle cos−1 µ. In writing this expression we have also assumed that the disk itself provides all of the shielding column density, although violation of this assumption could only change M by at most a factor of two. Notice that from equation (2), the disk midplane pressure falls with radius as R−3, whereas the X-ray flux from the AGN will decrease as R−2. This means that the X-ray ionization parameter actually increases outwards, which means the transition from a molecular to an atomic disk should be identified with the outer edge of the masing annulus. Plugging in the values appropriate for NGC 4258 (M8 = 0.39, L41 = 0.4, and µ ≈ 0.25, and R = 0.28 pc), and solving equation (4) for the mass accretion rate, we get M α ≈ 7 × 10−5 M⊙ yr−1. (5) With the inferred mass accretion rate, the mass of the masing disk is only ∼ 103 M⊙, and the neglect of the disk self-gravity is justified. Why is there an inner edge to the masing region? In NM95, it was speculated that this is because the disk actually flattens out interior to 0.14 pc, so that it is no longer illuminated by the X-ray source. In this case the temperature will be determined by viscous heating, and this will in general be too low to produce abundant water or 13 Figure 5: Phase diagram for dense X-ray irradiated gas. The water abundance relative to hydrogen and the gas temperature are plotted as a function of the X-ray ionization P11), where the incident X-ray flux Fx = 105F5 erg cm−2 s−1, Natt = parameter F5/(N 0.9 24 1024N24 cm−2 is the attenuating column density (i.e., the column density of gas that is neutral enough to absorb X-rays at the energies of interest) to the source of X-rays, and the gas pressure P = 1011k P11 dyne cm−2. The upper and lower branches correspond to the warm atomic and cooler molecular phases, respectively, in the bottom panel, and are reversed in the upper panel. The existence of the intermediate temperature phase is very sensitive to radiative trapping of cooling lines, and it probably does not occur in real tori or disks around AGN. Over the range of ionization parameter in which both branches exist, a two-phase medium is possible. 14 maser action (see further discussion on this point below)4. Such a disk geometry is expected in the case of radiation-driven warping (Pringle 1996; Maloney, Begelman, & Pringle 1996) since in this case the warp grows from the outside inwards. Whatever the mechanism producing the warp, its presence is crucial in the X-ray-powered maser model, since it is the warp that allows the central source to irradiate the disk and produce the maser emission. The physical conditions in the disk inferred in this model are shown in Figure 6. Beyond the critical radius at 0.28 pc, the disk is forced by the X-ray irradiation into the warm atomic state; the temperature is about 8000 K, and the scale height is about H ∼ 5 × 1015 cm. Molecular abundances are negligible. At the critical radius, the pressure is just high enough at the midplane to force the gas into the molecular phase. With decreasing radius the thickness of the molecular zone increases as a larger fraction of the disk is above the critical pressure; the temperature in the molecular zone will rise with increasing z−height above the midplane. Interior to 0.14 pc, where the warp flattens out, the disk temperature drops below 100 K, and masing ceases. If we assume that Lx/Lbol ∼ 0.1, as is typical for AGN, the derived mass accretion rate implies that rest-mass energy is converted to radiation with a radiative efficiency ǫ ∼ 0.1/α (6) (assuming steady accretion over the time it takes material to reach the black hole from 0.28 pc; the validity of this assumption is rather uncertain, as discussed briefly below). This in turn indicates that the low fractional Eddington luminosity of NGC 4258 is due to the low mass accretion rate onto the central black hole, and not due to low radiative efficiency of the accreting material. Also, although we do not know a priori where our line of sight to the X-ray source intercepts the disk, the column densities inferred for the disk strongly suggest that our sightline passes through the disk just at the edge of the molecular zone (NM95). 3.2 Alternative models The model of Neufeld & Maloney (1995), as outlined above, is essentially the "what- you-see-is-what-you-get" model: since the maser emission appears to arise in a thin, warped, disk, assume that this is in fact the case, with the warping of the disk allowing these regions of the disk to see the central X-ray source, which then powers the maser emission by a well-understood process (NMC). Unsurprisingly, in view of the uniquely detailed information on an accretion disk around a supermassive black hole provided by the maser observations, a number of additional models have been proposed. These will be briefly discussed here. Maoz & McKee (1998) (MM98) proposed that rather than X-ray heating, the maser 4Although maser action in the 22 GHz line requires that the gas density be less than n ∼ 1010 cm−3, it is unlikely that the inner edge is produced by the increase of the density with decreasing radius. As the midplane density rises, the masing layer will simply move to a greater height above the disk. See also the discussion in §3.2 15 FLAT, COLD INNER DISK (too cold to mase) T<100K Molecular Gas 0.28pc 0.14pc Redshifted Maser Features Maser Features Close to the Systemic Velocity Blueshifted Maser Features HOT, ATOMIC OUTER DISK (too hot for molecules) T~8000K Atomic Gas ~10 cm16 X-rays WARPED, MASING ANNULUS (obliquely illuminated) X-rays T~8000K Atomic Gas T=200-1000K Molecular Gas Figure 6: Structure of the accretion disk in NGC 4258, as inferred from observations of water maser emission. The disk temperature within the masing region is set by the X- ray heating rate; the outer edge is determined by the disk pressure (where it undergoes a phase transition from partly molecular to completely atomic) and the inner edge results from the flattening of the disk warp, so that it is no longer irradiated by the central source. From Neufeld & Maloney (1995). 16 emission in NGC 4258 is due to shocks in spiral density waves in a self-gravitating disk5. The main motivation for this was the hint of "regular" spacing in radius and velocity of the high-velocity features in the disk, as seen in the data of Miyoshi et al. (1995) (see Figures 4b and c). Although this model contains some ad hoc assumptions (e.g., it is assumed that conditions in the disk are such that water maser action will occur when a density wave -- which by fiat produces a shock of the required speed -- passes through the disk between 0.14 and 0.28 pc, and nowhere else; also, all of the predicted maser luminosities are scaled to a disk thickness of 1016 cm, an order of magnitude larger than the observed upper limit to the dimensions of the maser spots), it did make very definite predictions about the pattern of accelerations that should be seen as a function of impact parameter. This model also explained the relative weakness of the blue-shifted high-velocity features relative to the red-shifted features as a consequence of absorption by non-masing water. Indeed, Maoz & McKee made a generic prediction that all such disks should exhibit the same blue/red asymmetries, since this is an intrinsic feature of the spiral shock model. The mass accretion rate in this model is approximately two orders of magnitude larger than the rate inferred by NM95. The predicted acceleration pattern was tested by Bragg et al. (2001), who found no evidence to support the MM98 model, for any chosen pitch angle of the spiral pattern. The key point is that in the MM98 model, all of the blue-shifted features should exhibit positive accelerations and the red-shifted negative, but in fact positive and negative v are seen for both sides. Furthermore, both NGC 3079 (Nakai et al. 1995) and NGC 5793 (Hagiwara et al. 1997) have exhibited blue-shifted emission that is stronger than the red-shifted emission (always, in the case of NGC 3079), while the high-velocity emission features in IC 2560 are nearly symmetric in intensity (Ishihara et al. 2001). Hence at present there is no reason to suppose that this model applies to NGC 4258, or any other megamaser source. Desch, Wallin, & Watson (1998) proposed that rather than X-ray irradiation, the maser emission in NGC 4258 is powered by viscous dissipation within the accretion disk itself. This requires that the energy generated by viscous dissipation is deposited in a manner which is not proportional to the local density. The viscous heating rate per unit disk face area is D(R) = 3GM M 8πR3 while the disk surface density Σ(R) = M 3πν (7) (8) (9) where the radial dependence of Σ is contained entirely in ν. Both equations (7) and (8) assume a steady-state disk far from the inner edge. Doubling equation (7) to get the total heating rate through the disk, the ratio of viscous heating rate to surface density is 2D(R) Σ(R) = 9 4 GMν R3 5Note that in any scenario in which the maser emission arises in a disk, we must always view it close to edge-on if we are to see the maser emission, which will be strongly beamed in the disk plane. 17 M . This is also the vertically-averaged heating rate per particle, which is independent of which, again assuming the α−prescription for viscosity and a thin, Keplerian disk, as above, can be written as ¯Γ(R) ¯nH2(R) = 2.1 × 10−24 αM 1/2 R3/2 pc 8 c2 1 erg s−1 (10) where ¯Γ(R) is the vertically-averaged heating rate per unit volume and ¯nH2 is the similarly averaged number density of molecular hydrogen; c1 is the gas sound speed in units of 1 km s−1 (appropriate for molecular gas at a few hundred K). Note that this value of the viscous heating rate per particle is independent of how ¯Γ and ¯nH2 are calculated from D and NH2, respectively, provided only that they are calculated the same way. Plugging in the numbers appropriate to NGC 4258, ¯Γ ¯nH2 = 3.8 × 10−23α erg s−1. (11) For α <∼ 1, this is approximately two orders of magnitude smaller than the heating rate per particle needed to power H2O maser action in the disk (Desch et al. 1998). Hence it is necessary to assume that the viscous energy is deposited in a different fashion with height z than the gas density ρ(z), a not implausible assumption. The simplest assumption (and one that is in reasonable agreement with the results of Stone et al. 1996, who explicitly calculated the energy deposition as a function of height in three-dimensional magnetohydrodynamic simulations of the magnetorotational in- stability) is to assume that the viscous heating rate per unit volume is independent of z. In this case the heating rate per particle simply increases with height as a Gaussian with scale height equal to the gas scale height H; the ratio of the heating rate per particle to the vertically-averaged heating rate per particle derived above (equation 9 or 10) is Γ/n(z) ¯Γ/¯n ≃ exp(z2/2H 2) (12) which is (not surprisingly) independent of everything except the height above the disk M ; it is just the ratio midplane in scale heights. In particular, this is independent of of the midplane density to the density at height z. To maintain high enough temperatures to power the maser emission, we require a local heating rate which exceeds the vertically-averaged rate by a factor of about 100/α. Thus it is necessary that exp(z2/2H 2) ≈ 100/α =⇒ z H >∼ 3, (13) even for α ≈ 1, i.e., the density must drop to 1/100 of the midplane value. The only constraints on M are then: (a) The densities three or more scale heights above the plane must be high enough to excite the masing transitions; 18 (b) The column density of masing gas must be large enough to be detectable. This can be made to work for NGC 4258, for values of M /α comparable to (or larger than) the NM95 rate, in the sense that the above conditions can be satisfied. While the outer edge can be understood as the result of either the gas density or column density declining below the threshold for maser emission, the inner radial cutoff is unexplained, since the warping of the disk is irrelevant, and the masing layer should simply move higher into the disk atmosphere with decreasing radius. However, it appears very unlikely that this mechanism can apply to the other masing disk systems that have been observed (see section 4). In general, the factor by which the local viscous heating rate (in the masing region) has to exceed the vertically-averaged rate is Γ/n(z) ¯Γ/¯n ∼ 1300 R3/2 pc αM 1/2 8 . (14) For NGC 1068, for example, with Rpc ∼ 0.65, M8 ∼ 0.15, this is a factor of ∼ 1800/α. For the other systems that appear to be disk-like (Moran et al. 1999; §4) the spatial scales are also typically rmaser ∼ 1 pc, while the masses of the central objects are five to ten times smaller than in NGC 4258 or NGC 1068. This leads to prohibitively large M /α, e.g., in excess of 2 M⊙ yr−1 for the maser disk in NGC 1068. This is values of inconsistent with the mass and luminosity of the central source in NGC 1068 (which, radiating at nearly Eddington luminosity, is most definitely not an ADAF: see below), unless α ∼ 10−2. However, since the value by which the local heating rate must exceed the vertically-averaged value itself scales as α−1 (equation 14), the discrepancy cannot M /α required to satisfy the density or column be solved this way, since the values of density requirements will simply scale up proportionately as α is decreased. Hence I am rather skeptical about the viability of this mechanism as a general explanation for water maser emission in AGN6. Kartje, Konigl, & Elitzur (1999) proposed a variant X-ray powered accretion disk model, which differs from NM95 in (a) being specifically clumpy (although the NM95 model is referred to throughout as the "homogeneous disk model", as though this is a requirement, the same arguments carry over to a clumpy disk, the only difference being that the critical pressure refers to the clumps containing the bulk of the mass at a given radius); (b) has the clumps in NGC 4258 occur in a disk-driven hydromagnetic 6There is an additional aspect of this model that is rather problematic for high accretion rates. As noted earlier, a fundamental requirement for maser action is departure from thermodynamic equi- librium, in particular, that the gas temperature exceed the temperature of the radiation field in the mid- to far-IR. Desch et al. calculated the temperatures in their model under the assumption that the dust temperature is determined by the continuum emission of the central source incident on the disk, while the gas temperature is determined by viscous heating, and conclude that Tgas can exceed Tdust by a large enough factor to allow maser emission. This may be plausible for the NM95 mass accretion rate, since in this case the disk is not very optically thick (at least in the masing region) in a Rosseland mean sense, and the local viscous heating is substantially smaller than the incident continuum. For accretion rates that are much larger than this, however, neither of these statements are true, and it is difficult to see why the dust and gas temperatures will not be well coupled in a disk heated dominantly by viscous heating. 19 wind: the disk itself is actually flat, not warped; and (c) has a mass accretion rate that is two orders of magnitude higher than that derived by NM95, similar to MM98. One might perhaps be somewhat surprised that the observed disk should appear so thin, warped, and Keplerian if the emission actually arises in clumps uplifted off the flat disk in a wind. Since the NGC 4258 portion of this model has parameters that have of course been chosen to match the observations, perhaps its most direct prediction is that the mass accretion rate is much larger than inferred by NM95, exceeding M ≈ 10−2 M⊙yr−1. This is also true of the MM98 model, since accretion rates much higher than the NM95 rate are required to make the disk self-gravitating, and high mass accretion rates were also favored by Desch et al. (1998). Since the luminosity of the nucleus of NGC 4258 is well determined observationally (Makishima et al. 1994), all of these high mass accretion rate models require that the inner portion of the accretion disk (assuming steady accretion; I will return to this point shortly) radiates very inefficiently. Hence these papers have all appealed to "ADAF" models (for Advection-Dominated Accretion Flow; this is really just a renaming of what used to be referred to as "ion tori" [See Svensson 1999]) for accretion, in which the inner region of the flow is very hot (with the ions at close to the virial temperature), but radiates very inefficiently, as the electrons are not well coupled to the protons (so that Te ≪ Ti) (Ichimaru 1977; Rees et al. 1982; Phinney 1983; the modern incarnations begin with Narayan & Yi 1994 and Abramowicz et al. 1995, with literally dozens of follow-up papers). These flows exist only for values M only the standard thin, "cold" (i.e., of T ≪ Tvir disk solutions exist. M below a critical threshold; for higher Indeed, given the extremely low fractional Eddington luminosity of NGC 4258, there were immediate attempts to fit it into the ADAF framework (Lasota et al. 1996), since if its accretion is not described by an ADAF, there is really no reason to assume that ADAF models apply to any AGN. (As noted above, the fractional Eddington luminosity for NGC 1068 is so high -- nearly unity -- that there is no question that it is not described by an ADAF solution.) The success of the ADAF model lies in the fits to the "spectrum" of NGC 4258 (for an example, see Figure 1 of Lasota et al. 1996). M is in fact much larger than the NM95 rate, we must conclude that the disk does If not see the central X-ray source beyond 0.28 pc, since otherwise maser emission would arise further out, since there is no longer a transition to an atomic disk at this radius, as Rcr will also be much larger (see equation [4]). At present the ADAF models for NGC 4258 have been confined to an increasingly small portion of parameter space; the absence of 22 GHz continuum emission associated with the ADAF (Herrnstein et al. 1998) and the detection of X-ray variability on timescales of ∼ 4000 − 40, 000 seconds (Fiore et al. 2001) indicate that the transition to an ADAF must occur on scales of order 100 Schwarzschild radii or less. The most definitive test for the presence of cold material close to the central black hole would be the detection of a broad Fe Kα line; there has not yet been an observation of sufficient sensitivity (Reynolds, Nowak, and Maloney 2000; Fiore et al. 2001) to either detect such a line or rule one out at an interesting level. One important point that has been raised in connection with the mass accretion 20 rate and the radiative efficiency is the assumption of steady-state accretion (Gammie, Blandford, & Narayan 1999). As noted earlier, the efficiency inferred by NM95 is only valid if accretion has been steady over the time it takes material to migrate from the radius of the maser disk down to the central source. However, if the disk flattens out at the inner radius of the maser disk, the viscous timescale increases by a large factor due to the drop in disk temperature, and exceeds 109 years. It is very unlikely that accretion has been steady on this timescale. However, viscous heating at this radius will only heat the disk to T < 100 K (NM95). This means that the disk will be marginally self-gravitating: the Toomre Q parameter Q = 3c3 sα G M ≃ 4.0 × 10−3T 3/2 3 α M (15) is < 1.8 for the NM95 mass accretion rate, at these low temperatures. It therefore seems plausible that the mass flow through the disk at these radii will be determined by gravitational instabilities rather than viscosity, and so the magnitude of the viscous timescale may be irrelevant. 4 Imaging 2: Is NGC 4258 Unique? NGC 4258 was the first of the megamaser sources to be imaged with VLBI. The remarkable results obtained from this experiment generated a great deal of interest in imaging of the other H2O megamaser sources. With existing instrumentation it is difficult to obtain VLBI observations of masers that are weaker than about 0.5 Jy because of the necessity of detecting the maser within the coherence time of the interferometer (Moran et al. 1999). By using a maser feature as phase reference, and referring the other maser features to it, it is possible to extend the coherence time, and thereby observe weaker features, but it is still a difficult endeavour (J. Moran, pers. comm.) Hence not all of the known sources can be imaged at present. To date, nine megamasers have been imaged with VLBI. Of these nine, four show strong evidence for disk structure and two show probable evidence for disks from the spatial and velocity distribution of the emission; one, for which only the systemic emission has yet been imaged, shows kinematic evidence for a disk and may be the most similar to NGC 4258 of any source yet observed. The other two sources appear to be very different in origin. Table 1 (adapted from Moran et al. 1999) summarizes the results. None of the disk sources are as well-defined as in NGC 4258, nor do any of them exhibit Keplerian rotation curves. This latter feature, at least, is not unexpected, at least in the X-ray powered maser model, as I discuss below. First I will review some of the more interesting results from the imaging studies. 4.1 NGC 1068 The archetypal Seyfert 2 galaxy NGC 1068 was the second H2O megamaser source to be imaged with VLBI. Single-dish observations had already shown that maser emission extended up to about ±300 km s−1 away from the systemic velocity; unlike NGC 4258, 21 the systemic emission is not the strongest. Re-analysis of archival VLA observations by Gallimore et al. (1996) demonstrated that, in addition to the nuclear masers (i.e., the masers associated with the radio continuum emission believed to be coincident with the nucleus), a second group of fainter maser features was located about 0.3′′ (30 pc) away, at the same position angle as the radio jet. The initial VLBA observations (Greenhill et al. 1996) covered only the red and systemic velocities, and revealed a linear structure extended over about 1 parsec, oriented at 45◦ to the radio jet axis. Velocity gradients were suggestive of a rotating structure. Greenhill et al. suggested that this was the upper surface of an irradiated torus, and predicted that the blueshifted emission, when imaged, would form the mirror image to the redshifted. The lower (southern) half, which was missing, might be hidden by free-free absorption. However, when the blueshifted emission was imaged, it appeared as a continuation of the redshifted emission, so that the maser emission forms one linear structure that is highly inclined to the radio jet axis. This can be fitted by a thin, rotating disk, with rotation velocities up to 330 km s−1 (Greenhill & Gwinn 1997). The fall-off of rotation velocity with radius is slower than Keplerian, however, declining approximately as r−0.35. This could be due to either the mass of the disk itself, or to a central stellar cluster. There is a substantial spread of the velocities about the rotation curve, perhaps indicating a velocity dispersion of some tens of km s−1. The scale of the masing region is larger than in NGC 4258, with inner and outer edges at approximately 0.65 and 1.1 pc, respectively. The enclosed mass is about M = 1.5 × 107 M⊙; nearly all of this must be contributed by the black hole. The orientation of the disk with respect to the radio jet axis appears peculiar. However, Gallimore, Baum, & O'Dea (1997) imaged what appears to be a thermal source, with brightness temperatures Tb ∼ 105 − 106 K between 5 and 8 GHz, that lies interior to the maser disk and is elongated perpendicular to the jet axis. This appears to be emission from the disk rather than the jet. If this is the case, then interpreting the thermal source and masing disk as one continuous structure would imply that the masing region in fact traces a severely warped disk, with a much greater inclination warp than seen in NGC 4258. Support for a disk geometry, warped or not, is also provided by the absence of accelerations of the high-velocity features: Gallimore et al. (2001) derive upper limits for the redshifted masers of < 0.01 km s−1 yr−1 from fifteen years of monitoring data with the Effelsberg 100m telescope. Gallimore et al. also argue that the nuclear masers vary coherently on timescales of months to years, which is much shorter than the dynamical timescale, and that therefore they must be responding to variations of the central power source. Between October and November of 1997 there was a simultaneous flare of blueshifted and red- shifted maser features, which can be naturally understood as reverberation in a rotating disk. Neufeld (2000) has shown that the masers can respond to variations in the cen- tral source luminosity in two ways: increasing as the X-ray luminosity increases, which raises the maser emissivity, or decreasing as the bolometric luminosity increases, which decreases the difference between the dust and gas temperatures and therefore reduces the maser output. Either mechanism can explain the variation reported by Gallimore et al., provided that the density in the masing regions is at least nH2 ∼ 108 cm−3, which 22 ) s a m ( t e s f f O h t u o S - h t r o N 10 665 0 648…719 -10 -20 -30 557 414 460 451 507 10 0 -10 -20 -30 East-West Offset (mas) Figure 7: The spatial distribution of H2O maser emission from NGC 4945. The labels indicate the velocities of the emission, which is also shown by the grayscale (lighter shading corresponds to bluer velocity). The arrow marks the position of the maser feature closest to the systemic velocity of the galaxy (Vsys = 561 km s−1). At the distance of NGC 4945, 1 mas = 1.9 × 10−2 pc. From Greenhill et al. 1997a. is required to obtain a variability timescale as short as seen for the maser flare. Since, unfortunately, we do not have a direct view of the X-ray source in this galaxy, but see it only in reflection (as is also true for NGC 3079; see §4.3), it is not clear which mechanism is actually at work. As noted above, NGC 1068 also possesses fainter maser emission, located approx- imately 30 pc away from the nucleus along the radio jet axis (Gallimore et al. 1996). Given their location, these masers may arise in a shock that arises at the interface between the radio jet and a near-nuclear molecular cloud. In support of this interpre- tation, the jet appears to be diverted by its interaction with the ISM at this location. 23 4.2 NGC 4945 The first H2O megamaser to be discovered (dos Santos & Lepine 1979), NGC 4945 contains a luminous (Lx ≈ 6 × 1042 erg s−1) nuclear hard X-ray (1 -- 100 keV) source (Iwasawa et al. 1993; Done, Madejski, & Smith 1996; Madejski et al. 2000), which is heavily obscured behind a column NH ≈ 5 × 1024 cm−2, so that the direct X-ray emission is only observable above E ≈ 20 keV. In fact, the obscuring column is so high that a proper treatment of the radiative transfer to take the substantial Thomson optical depth into account is necessary; this raises the hard X-ray luminosity by about an order of magnitude over simple obscuring column density models: see Madejski et al. 2000. The X-ray emission is also highly variable, with tvar ∼ 1 day, implying that the sizescale of the emitting region is no more than r ∼ 1015 cm, and that the absorbing material cannot occupy a large solid angle around the disk (Madejski et al. 2000). The nuclear far-infrared luminosity is approximately an order of magnitude larger than the hard X-ray luminosity, and arises nearly entirely from a region no more than 225 pc by 170 pc in size (Brock et al. 1988). NGC 4945 lies in the southern hemisphere (δ = −49◦), and it is therefore very dif- ficult to observe with the VLBA. Nevertheless, Greenhill, Moran, & Herrnstein (1997) succeeded in observing the maser emission using two of the southernmost VLBA an- tennas. A typical spectrum of the emission (which is time variable) is shown in Figure 1. The high velocity emission brackets the systemic emission, with velocities up to about ±150 km s−1 away from systemic; as in NGC 4258, the redshifted emission is much stronger than the blueshifted, but as in NGC 1068, the redshifted emission is also much stronger than the systemic. The spatial distribution is shown in Figure 7; note that not all of the detected emission, particularly near the systemic and blueshifted velocities, could be mapped. The structure is quite linear, tracing a shallow S shape on the sky, and extends over about 40 mas (0.8 pc). The position-velocity diagram (Figure 8) bears some similarity to that of NGC 4258; however, the redshifted emission does not mirror the blueshifted emission, and the fall-off of velocity with impact parameter on the blueshifted side appears to be faster than Keplerian, which would imply that these features do not all lie in the plane of the sky. If one, nevertheless, interprets the emission in terms of an edge-on, rotating disk, the implied mass is M ∼ 1.4 × 106 M⊙, and the central mass density is at least ρ ∼ 2 × 107 M⊙ pc−3. With this central mass and the observed luminosity, NGC 4945 must be radiating in excess of 10% of the Eddington limit. 4.3 NGC 3079 NGC 3079 is a relatively nearby (D = 16 Mpc), nearly edge-on spiral galaxy. It has a fairly high far-infrared luminosity (LIR ≈ 3 × 1010 L⊙), a LINER-type spectrum (Heckman 1980) and a spectacular nuclear outflow (see Cecil et al. 2001, and references therein), with prominent radio lobes (Duric & Sequist 1988) and loops of Hα emission (Ford et al. 1986) extending a kpc or more away from the nucleus along the minor axis. NGC 3079 also has a prominent ring of molecular material, seen in CO emission, extending from approximately 250 to 750 pc, and a very compact, luminous CO peak 24 ) 1 - s m k ( l e h V 700 600 500 400 20 10 0 -10 -20 Impact Parameter (mas) Figure 8: The velocities of the mapped maser features in NGC 4945 as a function of impact parameter. The positions are measured relative to the feature that is closest in velocity to the systemic velocity (dashed line). The error bars on position are 3σ. From Greenhill et al. 1997. 25 centered on the nucleus (Sofue et al. 2001, and references therein). NGC 3079 possesses one of the most luminous H2O megamasers known (Henkel et al. 1984a,b; Haschick & Baan 1985). The maser emission is almost entirely concentrated to velocities blueward of the systemic velocity, and it is strongly variable in flux density and line width, the flux density having varied by about a factor of eight over ten years, while the linewidths have fluctuated by up to a factor of two (Baan & Haschick 1996). The maser emission from NGC 3079 was observed using the VLBA by Trotter et al. (1998). The observed distribution of maser emission on the sky, color-coded according to velocity, is shown in Figure 9. The two dashed lines show the major axis of the molecular disk discussed above (the nearly vertical line) and the axis of the nuclear jet. The emission arises in compact clumps, with sizes <∼ 0.02 pc; most of the emission arises in a region less than 0.2 pc in diameter, spread over a velocity range of about 130 km s−1. The size of the circles representing the maser emission are proportional to the logarithm of the flux density, and they are labeled with the velocity of the approximate flux density peak. The inset in Figure 9 (from Trotter et al.) shows an expanded version of the concentrated emission; this region includes the maser peak, at vLSR = 957 km s−1. Also sketched in Figure 9 are the positions and approximate sizes of 22 GHz continuum features. It is evident from Figure 9 that the maser emission from NGC 3079 differs markedly from the distribution seen in NGC 4258. The maser spots are approximately aligned with the larger-scale molecular disk, and their measured Doppler velocities are consis- tent with rotation in the same sense as the molecular disk. However, there is evidently a large nonrotational component to the velocity field: as noted above, the spread in velocities within the compact grouping of maser spots covers the entire range of blueshifted emission. This may indicate large turbulent velocities in the disk. The 22 GHz continuum emission is dominated by an unresolved ( < ∼ 0.1 pc) source (labeled B in Figure 9) that lies approximately 0.5 pc to the west of the maser peak. There is no maser emission associated with this (or any other) continuum peak; this argues against the suggestion that the unusually luminous maser emission in NGC 3079 is the result of beamed, unsaturated amplification of compact continuum emission by foreground molecular gas. A sketch showing the model of the emission suggested by Trotter et al. (1998) is shown in Figure 10; this must be regarded as highly speculative (see below). As with NGC 4945, the emission has been interpreted within the framework established by NGC 4258, i.e., as a rotating disk, seen close to edge-on. The disk thickness is unknown. Trotter et al. argued that the maser emission could not be powered by X-ray irradiation, as the observed X-ray luminosity of NGC 3079 was only Lx ∼ 1040 erg s−1 in soft X- rays, and there was no indication of a nuclear hard X-ray source. However, BeppoSax observations (Iyomoto et al. 2001) show that, as in the case of NGC 4945, there is a luminous (Lx ∼ 1042 − 1043 erg s−1) hard X-ray source buried beneath Compton-thick absorption, thus eliminating this argument. The overall distribution of velocities is consistent with the presence of a binding mass M ∼ 106 M⊙ within 0.5 pc. If this is indicative of the central mass, the hard X-ray luminosity indicates that the AGN contributes a large fraction of the far-infrared luminosity of the nucleus of NGC 3079, 26 Figure 9: The spatial and velocity distribution of the maser emission in NGC 3079, as observed with the VLBA, from Trotter et al. (1998). The spots marking the maser features are sized according to the logarithm of the flux density and color-coded ac- cording to velocity; the labels show the velocity of the approximate flux density peak. Also sketched in are the approximate positions and sizes of 22 GHz radio continuum features, and the axes of the large-scale molecular disk and the nuclear radio jet. 27 Figure 10: Model of the central pc of NGC 3079, based on VLBA observations (from Trotter et al. 1998). The maser spots are distributed in a disk that is close to edge-on; the disk center is assumed to be near where the major axis of the maser emission and the jet axis intersect. and it is radiating at 1 − 10% of the Eddington limit. Sawada-Satoh et al. (2000) argue that this interpretation is not correct, as they identify the bright continuum source (labeled "B" in Figure 9) as the location of the nucleus. This identification is based on the apparent fixed position of component B with respect to the strongest maser feature based on several years of observations, unlike components A and C, which they argue are probably jet features. Sawada- Satoh et al. suggest that the maser emission arises in a geometrically thick structure whose rotation axis passes through component B; this axis is inclined by about 110◦ to the large-scale rotation axis of the galaxy. 28 4.4 Circinus The Circinus galaxy is a nearby (D = 4 Mpc) Seyfert 2 galaxy in the Southern hemisphere, and was one of the first megamaser sources to be discovered (Gardner & Whiteoak 1982). It bears interesting similarities to several of the galaxies discussed above, including the presence of a luminous (Lx ∼ 1042 erg s−1) hard X-ray source hidden behind Compton-thick obscuration (Matt et al. 1999), a nuclear outflow along the minor axis (seen in bipolar radio lobes: Elmouttie et al. 1998), a broad (opening angle ∼ 90◦) ionization cone (Marconi et al. 1994; Veilleux & Bland-Hawthorn 1997) and a nuclear (∼ several hundred pc) molecular ring seen in CO emission (Curran et al. 1998). The maser spectrum shows emission out to approximately ±200 km s−1 with respect to the systemic velocity; the red-shifted emission is stronger than the blue, and both are substantially stronger than the systemic (Nakai et al. 1995; Braatz, Wilson, & Henkel 1996; Greenhill et al. 2001). The maser emission is highly variable, and the spectrum alters substantially on timescales of order one month (Whiteoak & Gardner 1986). Variability has been seen on timescales as short as a few minutes (Greenhill et al. 1997b), much more rapidly than any other source. This rapid variability is almost certainly the result of interstellar scintillation; Circinus lies at low Galactic latitude, b = −3.8◦. VLBI observations of the H2O maser emission from Circinus have been reported by Greenhill (2001b) and Greenhill et al. (2001). The emission was observed using the Australia Telescope Long Baseline Array in 1997 and 1998. The distribution on the sky (with gray scale indicating velocity) is shown in Figure 11. Although this appears somewhat confusing, when combined with the velocity information (Figure 12) a coherent picture emerges. At each observing epoch, the maser emission consists of three components: a thin, shallow S shape on the sky (similar to NGC 4945), comprised of highly redshifted and blueshifted emission; emission at close to the systemic velocity, which lies between the two arcs of high-velocity emission; and modestly Doppler-shifted (up to ∼ 100 km s−1) emission whose spatial distribution is correlated with the sign of the velocity shift: blueshifted emission on the southeast side, redshifted on the northwest side. The high-velocity and near-systemic velocity emission can be fitted by a thin (thick- ness < 2 mas = 0.04 pc), warped disk, viewed edge-on, with an outer radius of ap- proximately 0.4 pc (20 mas) and an inner edge at ∼ 0.1 pc. The orbital velocity (assuming circular motion) at the inner edge is 237 km s−1, giving an enclosed mass M ≃ 1.3 × 106 M⊙, and a minimum mass density ρ ∼ 3 × 108 M⊙ pc−3. With this central mass, the observed X-ray luminosity implies a fractional Eddington luminosity of order 10%. The high-velocity emission on the redshifted side can be fitted with a peak velocity that declines with impact parameter b approximately as b−0.4±0.05. The fact that the rotation curve is shallower than Keplerian could be due to the mass of the disk; this would imply a disk mass ∼ 105 M⊙ between 0.1 and 0.4 pc. Alternatively, this could be due to the contribution of a central stellar cluster. Lending additional support to this model is the fact that a single position-velocity gradient connects the highest velocity red- and blue-shifted emission, and the near-systemic velocity emis- 29 60 40 20 0 -20 ) s a m ( t e s f f O h t u o S - h t r o N June 1997 July 1997 > June 1998 60 40 20 0 -20 East-West Offset (mas) 300 400 500 600 Heliocentric Radio Velocity (km s -1) Figure 11: The spatial distribution of H2O maser emission from the Circinus galaxy. The observations were actually taken at three separate epochs; the synthesized beam sizes are plotted at lower left. The velocity of the emission is indicated by the gray scale. At the distance of Circinus, 1 mas = 1.9 × 10−2 pc. The systemic velocity Vsys ≃ 440 km s−1. From Greenhill et al. (2001). 30 High-velocity emission Outflow 600 500 400 300 Outflow Low-velocity emission High-velocity emission ) 1 - l s m k ( y t i c o e V o d a R c i r t n e c o i i l e H 40 -40 Impact Parameter with respect to Black Hole (mas) -20 20 0 Figure 12: Position-velocity diagram for the Circinus maser emission. Labels indicate features of the model discussed in the text. The rotation curve declines with impact parameter b approximately as b−0.4±0.05. From Greenhill (2001b). 31 Figure 13: Model of the warped accretion disk and outflow structure in the Circi- nus galaxy, superimposed on the spatial distribution of the emission. From Greenhill (2001b). sion; this is also seen in NGC 4258, and is expected if all of the near-systemic emission arises at the inner radius of the disk (although this emission is relatively much weaker than in NGC 4258). What does the intermediate-velocity gas represent? Greenhill (2001b) and Green- hill et al. (2001) interpret this as a fairly broad-angle outflow (see Figure 13). The systematic redshifting and blueshifting with position on the sky indicate that the out- flow is tipped with respect to our line of sight. The high-velocity emission ceases where it would be intersected by the outflow, which may indicate that the disk is broken up by the interaction. Interestingly, all of the outflow maser emission arises in regions where the view of the central source would not be shadowed by the disk, which suggests that the maser emission is due to irradiation of the outflow by hard X-rays from the central source. Furthermore, the edges of the shadowed zone coincide very well with the boundaries of the kiloparsec-scale ionization cone seen to the west of the nucleus. However, truncation of the disk also implies that the central black hole will be starved by exhaustion of the disk mass in no more than ∼ 107 years (Greenhill 2001b). 4.5 IC 2560 IC 2560 is a southern (δ ≈ −33◦) barred spiral in the Antlia cluster, at a distance of 26 Mpc. It is classified as a Seyfert 2 on the basis of optical spectra (Fairall 1986). H2O maser emission was first detected by Braatz et al. (1996); the emission velocity was close to the systemic velocity of the galaxy. The maser emission was reobserved by Ishihara et al. (2001), using both single-dish 32 Figure 14: Spectrum of the H2O maser emission from IC 2560 obtained with the NRO 45m telescope, averaged over all epochs. The arrow marks the systemic velocity of the galaxy. From Ishihara et al. (2001). telescopes (the NRO 45m and the Parkes 64m dishes) and the VLBA. Multiple epochs were obtained, with observations spanning the period 1996 -- 2000. Most of the flux is contained in the emission around the systemic velocity, which consists of a number of narrow individual features blended together over a band ∆V ≈ 60 km s−1 wide. (This was the only emission detected by Braatz et al.) The peak flux density is variable and has reached a maximum of about twice the value at the time of the Braatz et al. observation. Most interestingly, however, is the detection of additional, high-velocity emission, extending to approximately 400 km s−1 away from Vsys to both the red and the blue (see Figure 14). The redshifted emission is only slightly stronger than the blueshifted; the brightest high-velocity emission features are about 12 − 14% of the peak flux density. The spectrum of IC 2560 bears remarkable similarity to that of NGC 4258 (com- pare Figure 14 with the top panel of Figure 4), in that there are weaker, high-velocity features spread over a substantial range to either side of the systemic emission, al- though the velocities are not as high in IC 2560, and the high-velocity emission is more symmetrically distributed than in NGC 4258. This is the only megamaser observed to date other than NGC 4258 in which the systemic velocity emission is much stronger than the high-velocity emission. This is not the only similarity, however. Ishihara et al. claim that over the course of the monitoring period (4.4 years using the NRO 45m and 1.1 year with the Parkes 64m), the systemic velocity features showed a secular drift to redward, at a rate v ≃ 2.6 km s−1yr−1. In NGC 4258 this drift (at v ≃ 9 km s−1yr−1) is interpreted as centripetal acceleration, as discussed in §3. A similar velocity drift has been seen in NGC 2639 (Wilson et al. 1995), with v ≃ 6.6 km s−1yr−1. One high- velocity feature was also monitored over a period of about a year and showed no secular acceleration, with an upper limit v <∼ 0.5 km s−1yr−1. Given the spectral similarities, one might expect imaging of the maser emission to be equally intriguing. Unfortunately, only the systemic velocity emission has been observed with VLBI so far, as the high-velocity emission had not been detected at 33 the time of the VLBA observation. Furthermore, the high-velocity emission is faint, with flux densities ∼ 0.050.1 Jy, so that imaging this emission will be very difficult. Nevertheless, the results are indeed interesting. Figure 15 (from Ishihara et al. 2001) shows the spatial distribution of the emission and position-velocity cuts in both right ascension and declination. The emission is confined to a region approximately 0.1 mas by 0.2 mas in RA and Dec (0.01 × 0.02 pc at 26 Mpc); note that the position errors in declination are much larger than those in RA. This is similar to the spatial extent of the systemic velocity emission in NGC 4258. Furthermore, the position- velocity diagram in RA shows a systematic, linear trend, as in NGC 4258, with a magnitude dv/db ≃ 0.85 ± 0.27 km s−1 µas−1. (No trend is seen in the p-v diagram in Dec, which may be due to the large position errors or to the orientation.) There is also an unresolved continuum source detected at 22 GHz, with a very high brightness ∼ 3 ×1011 K), which is coincident with the grouping of maser emission > temperature (Tb within the errors. The maser emission in IC 2560 is certainly suggestive, although the rotation curve has not yet been obtained. When placed in the framework of an edge-on, Keplerian disk, the observations imply a disk with inner and outer radii of 0.07 and 0.26 pc, respectively, in orbit around a central mass of M = 2.8 × 106 M⊙, about an order of magnitude smaller than that for NGC 4258 or NGC 1068, but comparable to the central masses inferred for NGC 4945, Circinus, and NGC 30797. It must be admitted that the distribution of systemic emission on the sky does not appear disk-like. However, it must also be remembered that IC 2560 is about 3.5 times further away than NGC 4258, so that (along with the smaller inner radius) the angular scale of the emission is smaller by about a factor of six. Furthermore, the emission is several times weaker than in NGC 4258, so that the positional error bars are quite large, especially in declination, where they are 50 − 60 µas. With a maximum rotation velocity of 418 km s−1, the velocity drift of the systemic velocity features places them at 0.07 pc, which Ishihara et al. identify with the inner radius of the disk. If this is correct, the lower limit to the central mass density is ρ >∼ 2.1 × 109 M⊙ pc−3, second only to NGC 4258 among the maser disks observed to date. As in NGC 4258, this location for the systemic velocity features is also supported by the gradient seen in the position-velocity diagram: for a given enclosed mass M, the radius implied for a measured dv/db (where v is the projected velocity and b is the impact parameter), for a Keplerian rotation curve, is R = "(GM)1/2 dv/db #2/3 = 4.66 × 10−3 M 1/3 6 d2/3 Mpc (dv/db)2/3 pc (16) where M6 is the mass of the central black hole in units of 106 M⊙, dMpc is the distance to the host galaxy in Mpc, and the velocity gradient dv/db is in km s−1 µas−1. For IC 2560, equation (16) yields R = 0.064 ± 0.02 pc, in excellent agreement with the value estimated from v. This agreement also implies that the disk major axis is aligned 7With this mass, the inner radius of the maser disk is at R ≃ 2.6 × 105 Schwarzschild radii, while the outer edge lies at nearly 106Rs. 34 Figure 15: Spatial and velocity distribution of the systemic-velocity maser emission from IC 2560. (a) shows the position of the maser emission on the sky. (b) and (c) show position-velocity diagrams for cuts in right ascension and declination, respectively. The RA p-v cut shows clear evidence for a systematic, linear gradient of velocity with impact parameter. From Ishihara et al. (2001). 35 within ∼ 20◦ of east-west. The velocity extent of the systemic emission implies (in the Keplerian disk framework) a very similar beaming angle to NGC 4258, about 9◦. There is one additional similarity between IC 2560 and NGC 1068. Ishihara et al. (2001) analyzed the X-ray emission from IC 2560, which was observed by ASCA in 1996 for approximately 42 ksec with the GIS and SIS spectrometers. Although the S/N is poor, Ishihara et al. find evidence for a heavily-absorbed (NH ∼ 3 × 1023 cm−2) hard X-ray source, with a 2 -- 10 keV X-ray luminosity Lx ∼ 1041 erg s−1. It could be substantially lower than this; a much higher S/N spectrum is needed. Assuming that this represents 10% of the bolometric luminosity, the fractional Eddington luminosity is of order 10−3, which classes IC 2560 with NGC 4258 as having very low L/LEdd, as opposed to the other possible maser disk systems with L/LEdd ∼ 0.1 or greater. If we make the assumption that the maser emission is distributed in a thin, Keplerian disk that is irradiated by the central X-ray source, as in NGC 4258, and apply the theory of NMC to identify the outer edge of the disk with the molecular-to-atomic phase transi- tion (equation 4), we obtain a mass accretion rate M/α ∼ 8 × 10−4 M⊙yr−1, and with the above estimate of the luminosity, a radiative efficiency ǫ ∼ 0.02/α, indicating that the luminosity of IC 2560 is low simply because, as in NGC 4258, the mass accretion rate is low. Further observations of this system with VLBI are eagerly awaited. If the geometry is confirmed to be similar to the NGC 4258 disk, it will be possible to obtain another purely geometric measurement of the distance to the maser disk, but for a galaxy which is nearly four times further away than NGC 4258, which would be very important for determination of Ho (as the distance determination for NGC 4258 already has been, since it allows one to simply skip over all of the argument about the distance modulus to the LMC: Newman et al. 2001). Of the sources for which available VLBI imaging data provide evidence for maser disks, only the emission from IC 2560 appears to bear a marked similarity to the NGC 4258 disk (although until the recently discovered high-velocity emission from IC 2560 is imaged, the nature of the rotation curve will remain unknown). Is this surprising? Not necessarily, at least in one aspect. Consider a galaxy containing a supermassive black hole in its nucleus. The black hole dominates the potential within a radius RBH ≈ 2 M7 σ2 150 pc (17) where the velocity dispersion characterizing the depth of the galactic potential well (not including the contribution from the black hole) is σ = 150σ150 km s−1. Using the X-ray powered maser model of NMC and NM95, the critical radius for molecule formation can be written in terms of the fractional Eddington luminosity (cf. equation 4) as Rcr = 1.5 (L/LEdd)0.38M7 (αǫ0.1)0.81µ0.38 pc (18) where the radiative efficiency ǫ = 0.1ǫ0.1. In order for a maser disk to exhibit a 36 Keplerian rotation curve, it must satisfy Rcr ≪ RBH , or L LEdd ≪ 0.5 (αǫ0.1)2.1(µ/0.25) σ5.3 150 . (19) Hence only AGN with very sub-Eddington luminosities will have maser disks in the Keplerian regime. Since, as noted above, IC 2560 appears to fulfill this requirement, with L/LEdd ∼ 10−3, it will be very interesting to see what VLBI reveals about the structure of the high-velocity emission. 5 Imaging 3: The Non-Disk Sources Nearly all of the H2O megamasers detected so far are in disk-dominated galaxies. There are two exceptions to this, which have interesting similarities to one another and also to a third which has yet to be fully imaged. The three galaxies (the first two of which are listed in Table 1) are NGC 1052 (an elliptical galaxy with a LINER nucleus), IRAS 22265, also known as TXFS 2226-184, originally classified as an elliptical or S0 but now definitely known to be a disk galaxy (Falcke et al. 2000a), and Mrk 348, classed as an S0. A fourth object, IRAS F01063-8034, which is definitely a disk galaxy, has a recently discovered maser (Greenhill et al. 2002) which also appears to be similar, and will be discussed briefly below. These four objects stand out from the other megamasers on the basis of their line profiles. As noted in §1, the typical line profile for megamasers consists of a number of narrow components, with widths of a few km s−1, spread over a much broader velocity range, ∆V ∼ 102 − 103 km s−1. The four galaxies discussed in this section have completely different profiles, consisting of single, broad (∆V ≈ 90 − 130 km s−1, except for IRAS F01063, for which ∆V ≈ 40 − 50 km s−1), more or less Gaussian lines. In the cases of NGC 1052 and Mrk 348, the maser emission is offset by approximately 150 km s−1 with respect to the systemic velocity of the galaxy; no such offset has been reported for the IRAS 22265 emission and the IRAS F01063 emission is at the same velocity as the galaxy. The spectral differences alone suggest that these masers differ in a fundamental way from the other megamaser sources. Imaging with VLBI appears to confirm this, at least for the first two mentioned. Claussen et al. (1998) presented VLBI observations of NGC 1052 in both 22 GHz continuum and the 616 − 523 water line. The continuum data show an elongated structure composed of at least seven components, with a gap near the center. The maser emission occurs in two moderately distinct groups, separated by about 0.02 pc, with a total extent of about 0.06 pc. Interestingly, the maser emission appears to lie along the axis of the "jet" seen in the continuum emission. There is also a clear velocity gradient in the maser emission, also in the direction parallel to the jet axis. The masers may be excited by shocks driven by the interaction of the jet with molecular clouds, as may be seen in the "jet masers" of NGC 1068 (Gallimore et al. 2001). Alternatively, given the superimposition of the masers on the continuum emission, the masers may arise in foreground molecular clouds that are amplifying the 37 Figure 16: Spectrum of the maser emission from Mrk 348, at two different epochs. The maser clearly rose in intensity by a factor of three between late 1997 and early 2000. Also note the redshift of the emission with respect to the systemic velocity. From Falcke et al. (2000b). continuum emission (although, given the very small angular and spatial scales involved, the absence of maser emission from the vicinity of the other continuum components is a little surprising in this interpretation). The latter scenario can be tested by searching for correlated variation of the continuum and maser emission (Claussen et al.), since for this model to work the maser emission must be unsaturated. The velocity centroid of the maser emission shifted by 45 km s−1 between two observations spaced five months apart (Braatz et al. 1996), which argues in favor of the jet-excited model. IRAS 22265 has been imaged with VLBI but the results have not yet been published (Greenhill, Moran, & Henkel 2002, in preparation). In Moran et al. (1999), from which Table 1 has been adapted, the distribution of maser emission is described as "messy", with a spatial extent of about 2.4 pc (note that their Table 3 contains a minor typographical error: the spatial extent of the emission in NGC 1052 is given as 0.6 pc rather than 0.06 pc). Greenhill et al. (2002) note that the emission from this source has been resolved into at least four components, spread over roughly a parsec, with individual line widths of tens of kilometers per second. The third source with a single, broad maser line, Mrk 348, is a Seyfert 2 galaxy, with broad emission lines in polarized light. It is unusual among Seyfert galaxies in having a very bright and variable radio nucleus. Among the megamaser sources, in addition to its line profile, it is notable for having flared by about a factor of three in its 22 GHz emission (Falcke et al. 2000b), sometime between the end of 1997 and early 38 2000 (see Figure 16). It was observed but not detected in the Braatz et al. survey; analysis of archival observations by Falcke et al. show that it would have been seen only at the 3σ level at the time of the Braatz et al. observations. Furthermore, Falcke et al. present evidence that the radio continuum emission in Mrk 348 also rose between late 1997 and early 2000 by about the same factor as the maser emission. (Between early 1997 and late 1998, the flux density at 15 GHz was observed to increase by nearly a factor of 5 by Ulvestad et al. 1999.) If the variations of continuum and line are in fact correlated, this would indicate that the maser emission in Mrk 348 is unsaturated. The upper limit to the response time lag between the radio flare and the maser flare (about two years) sets an upper limit to the distance of the masers from the nucleus of about 0.6 pc. A recent VLBI observation with MERLIN (Xanthopoulos & Richards 2001) of the red half of the line observed by Falcke et al. confirms that maser emission arises within ∼ 0.8 pc of one of the continuum peaks; the peak flux density is about three times higher than when observed by Falcke et al. (2000b), so that the maser flux density has increased by about an order of magnitude since late 1997. The fourth source, IRAS F01063-8034, is a moderately distant (57 Mpc) southern edge-on disk galaxy, classified as Sa, which shows no real sign in the optical, infrared, or radio of possessing an active nucleus. Greenhill et al. (2002) discovered luminous (LH2O ∼ 450 L⊙ maser emission from this galaxy. The maser emission consists of a single Gaussian line, with a FWHM ∆V ≈ 54 km s−1. As in NGC 1052, the maser emission exhibited a shift in velocity, with the centroid jumping by 38 km s−1 within thirteen days. As with NGC 1052, such velocity shifts are difficult to understand except in the context of jet-excited models. In the case of IRAS F01063, this offers evidence for the presence of a previously unsuspected AGN in this galaxy. The similarities among these four sources suggest that a common physical process is at work. The broad, smooth emission profiles may indicate that the maser action in all four of these galaxies is unsaturated, which would agree with the superimposition of the maser emission on the radio continuum source in NGC 1052, and with a correlated continuum/maser flare in Mrk 348. The spatial distribution of the emission, which is less cohesive than seen in most of the other megamasers imaged to date, may indicate a different origin than X-ray pumping, such as shocks, although this is not yet clear: both NGC 1052 and Mrk 348 are known to possess luminous (Lx ∼ 1042 −1043 erg s−1), heavily obscured (NH ∼ 1023 cm−2), nuclear hard X-ray sources (Weaver et al. 1999; Warwick et al. 1989). Certainly the velocity shifts seen in both NGC 1052 and IRAS F01063-8034 argue for a shock origin for the masers, perhaps due to jet/gas interactions. 6 The Future Future, deeper surveys for H2O megamasers can be expected to turn up many more sources in Seyfert 2 and LINER galaxies, given the known detection fractions (Braatz et al. 1996, 1997). Unfortunately, with existing instruments very few of these addi- tional sources will be bright enough to image with VLBI; with the VLBA it is difficult (although not impossible, as discussed earlier) to observe sources that are fainter than about 0.5 Jy. The advent of the Atacama Large Millimeter Array (ALMA) will offer 39 a very large increase in sensitivity over the VLA. However, ALMA will not be able to observe the 22 GHz line, as the lower frequency limit will be 30 GHz, and, in any case, would not have the spatial resolution at this frequency (< 1 mas) required to image the nuclear maser emission in distant galaxies. The proposed Square Kilometer Array (SKA) will have a huge collecting area, and consequently enormous sensitivity at 22 GHz; hence SKA will be able to detect very faint, distant megamaser sources. However, it is not certain that SKA will be designed with the capability of performing high-resolution observations, in which case it will be superb for maser surveys but will not be able to do the follow-up imaging work. The Green Bank Telescope (GBT) will of course not have the necessary spatial resolution for imaging, but will be a superb instrument for monitoring studies. However, imaging in the 22 GHz line is not the only method by which our un- derstanding of H2O megamasers can be improved. The only transition that has been observed so far from these sources is the 616 − 523 line at 22 GHz. However, a number of other water lines are predicted to mase, and several other masing transitions have been detected in Galactic sources: the 313 − 220 line at 183 GHz (Waters et al. 1980; Cernicharo et al. 1990); the 1029 − 936 line at 321 GHz (Menten, Melnick, & Phillips 1990); and the 515 − 422 line at 325 GHz (Menten et al. 1990). In all of these sources the 22 GHz line is seen as well, and it is very likely that the maser radiation is arising in the same gas (e.g., the velocity range of emission is the same for all the transitions, except where detection is limited by low signal-to-noise; the line ratios seen in these sources, all of which are single-dish observations, are in reasonable agreement with model predictions; see Menten et al. 1990). All of these transitions are substantially weaker (both observationally and theoretically) than the 22 GHz line. However, the very large increase in sensitivity provided by ALMA will make it possible to observe the 321 and 325 GHz (and, under very favorable atmospheric conditions, the 183 GHz transition) from the ground. At 325 GHz, ALMA will have spatial resolution of 20 mas (assuming the planned 10 km baseline). As discussed in Neufeld & Melnick (1991), observations of these transitions (and others that fall within the atmospheric windows) will make it possible to use masers not only as a probe of dynamics, but also of the physical conditions within the gas. Since these transitions arise at different energies above the ground and have different Einstein A values, the ratios of different masing transitions will depend on the density and temperature. Furthermore, by observing maser lines at different frequencies, it will be possible to determine whether free-free absorption is affecting the emission. For example, Herrnstein et al. (1996) proposed that the asymmetry between the redshifted and blueshifted high-velocity emission in NGC 4258 is not inherent, but is a conse- quence of the geometry (see Figure 6): on the redshifted side, it is the "underside" (from our perspective) of the disk that is irradiated, and so the warm, atomic zone lies behind the masing region, whereas on the blueshifted side they are reversed, and the maser emission is attenuated by free-free absorption within the weakly ionized atomic zone. This hypothesis can be tested by observing the disk at frequencies much higher than 22 GHz, where the optical depth to free-free emission becomes negligibly small. By observing density-sensitive line ratios, the actual densities in the masing zone can 40 be estimated, and compared with model predictions, such as the NM95 model. David Neufeld and I are currently at work on improved models of X-ray powered masers, to provide predictions of the line strengths and ratios for future observations. Acknowledgments My research on masers is supported by the National Science Foundation under grant AST 99-00871. This work was also supported by NASA through HST grant AR- 08747.02-A. I wish to thank James Moran and a second, anonymous referee for their detailed and extremely helpful comments on a draft of this review, and Lincoln Green- hill and Jim Braatz for providing useful information. I am grateful to Brad Gibson and the ASA meeting organizers for the invitation to present the talk on which this article is based, and to Brad and Kristine and Joss and Sue for their hospitality in Melbourne and Sydney, respectively. Thanks to Joe for the company and all the driving in the 4WD monster and to Steve for just about everything else, especially the sushi and beer. Above all, I wish to thank Michelle Storey for her patience in dealing with this extremely late manuscript. References Abramowicz, M.A., Chen, X., Kato, S., Lasota, J.-P., & Regev, O. 1995, ApJ, 438, L37 Antonucci, R. 1993, ARAA, 31 473 Baan, W.A., & Haschick, A.D. 1996, ApJ, 473, 269 Balbus, S.A., & Hawley, J.F. 1991, ApJ, 376, 214 Braatz, J.A., Wilson, A.S., & Henkel, C. 1997, ApJS, 110, 321 Braatz, J.A., Wilson, A.S., & Henkel, C. 1996, ApJS, 106, 51 Bragg, A.E., Greenhill, L.J., Moran, J.M., & Henkel, C. 2000, ApJ, 535, 73 Brock, D., Joy, M., Lester, D.F., Harvey, P.M., & Ellis, H.B., Jr. 1988, ApJ, 329, 208 Cecil, G., Bland-Hawthorn, J., Veilleux, S., & Filippenko, A.V. 2001, ApJ, 555, 338 Cernicharo, J., Thum, C., Hein, H., John, D., Garcia, P., & Mattioco, F. 1990, A&A, 231, L15 Cheung, A.C., Rank, D.M., Townes, C.H., Thornton, D.D., & Welch, W.J. 1969, Nature, 221, 626 Churchwell, E., Witzel, A., Huchtmeier, W., Pauliny-Toth, I., Roland, J., & Suben, W. 1977, A&A 54, 969 Claussen, M.J., & Lo, K.-Y. 1986, ApJ, 308, 592 Claussen, M.J., Heiligman, G.M., & Lo, K.-Y. 1984, Nature, 310, 298 Claussen, M.J., Diamond, P.J., Braatz, J.A., Wilson, A.S., & Henkel, C. 1998, ApJ, 500, L129 Collison, A.J., & Watson, W.D. 1995, ApJ 452, L103 Conway, J.E. 1999, in Highly Redshifted Radio Lines, Ed. C.L. Carilli, S.J.E. Radford, K.M. Menten, & G.I. Langston, (San Francisco: Astronomical Society of the Pacific), 41 259 Curran, S.J., Johansson, L.E.B., Rydbeck, G., & Booth, R.S. 1998, A&A, 338, 863 de Jong, T. 1973, A&A, 26, 297 Desch, S.J., Wallin, B.K., & Watson, W.D. 1998, ApJ, 496, 775 Done, C., Madejski, G.M., & Smith, D.A. 1996, ApJ, 463, L63 dos Santos, P.M., & Lepine, J.R.D. 1979, Nature, 278, 34 Duric, N., & Seaquist, E.R. 1988, ApJ, 326, 574 Elitzur, M. 1992, Astronomical Masers, (Dordrecht: Kluwer), 262 Elmouttie, M., Haynes, R.F., Jones, K.L., Sadler, E.M., & Ehle, M. 1998, MNRAS, 297, 1202 Fairall, A.P. 1986, MNRAS, 218, 453 Falcke, H., Wilson, A.S., Henkel, C., Brunthaler, A., & Braatz, J.A. 2000a, ApJ, 530, L13 Falcke, H., Henkel, C., Peck, A.B., Hagiwara, Y., Almudena Prieto, M., & Gallimore, J.F. 2000b, A&A, 358, L17 Fiore, F., Pellegrini, S., Matt, G., Antonelli, L.A., Comastri, A., della Ceca, R., Gial- longo, E., Mathur, S., Molendi, S., Siemiginowska, A., Trinchieri, G., & Wilkes, B. 2001, ApJ, 556, 150 Ford, H.C., Dahari, O., Jacoby, G.H., Crane, P.C., & Ciardullo, R. 1986, ApJ, 311, L7 Frank, J., King, A., & Raine, D. 1992, Accretion Power in Astrophysics (2nd ed.) (Cambridge: Cambridge University Press) Gallimore, J.F., Baum, S.A., & O'Dea, C.P. 1997, Nature, 388, 852 Gallimore, J.F., Baum, S.A., O'Dea, C.P., Brinks, E., & Pedlar, A. 1996, ApJ, 462, 740 Gallimore, J.F., Henkel, C., Baum, S.A., Glass, I.S., Claussen, M.J., Prieto, M.A., & Von Kap-herr, A. 2001, 556, 694 Gammie, C.F., Narayan, R., & Blandford, R. 1999, ApJ, 516, 177 Gardner, F.F., & Whiteoak, J.B. 1982, MNRAS, 201, P13 Genzel, R., & Downes, D. 1979, A&A 72, 234 Greenhill, L.J. 2001a, in Cosmic Masers, ed. V. Migenes, (San Francisco: Astronomical Society of the Pacific), in press. Greenhill, L.J. 2001b, in Proceedings of the 5th EVN Symposium, ed. J. Conway, A. Polatidis, R. Booth, Onsala Space Observatory, Chalmers Technical University, Gothenburg, Sweden, in press. Greenhill, L.J., & Gwinn, C.R. 1997, Ap&SS, 248, 261 Greenhill, L.J., Gwinn, C.R., Antonucci, R., & Barvainis, R. 1996, ApJ, 472, L21 Greenhill, L.J., Moran, J.M., & Herrnstein, J.R. 1997a, ApJ, 481, L23 Greenhill, L.J., Ellingsen, S.P., Norris, R.P., Gough, R.G., Sinclair, M.W., Moran, J.M., & Mushotzky, R. 1997b, ApJ, 474, L103 Greenhill, L.J., Henkel, C., Becker, R., Wilson, T.L., & Wouterloot, J.G.A. 1995a, A&A, 304, 21 Greenhill, L.J., Jiang, D.R., Moran, J.M., Reid, M.J., Lo, K.-Y., & Claussen, M.J. 1995b, ApJ, 440, 619 Greenhill, L.J., Ellingsen, S.P., Norris, R.P., Gough, R.G., Sinclair, M.W., Moran, 42 J.M., & Mushotzky, R. 1997, ApJ, 474, L103 Greenhill, L.J., Moran, J.M., Booth, R.S., Ellingsen, S.P., McCulloch, P.M., Jauncey, D.L., Norris, R.P., Reynolds, J.P., Tzioumis, A.K., & Herrnstein, J.R. 2001, in Galaxies and their Constituents at the Highest Angular Resolutions, ed. R. Schilizzi, S. Vogel, F. Paresce, & M. Elvis (San Francisco: Astronomical Society of the Pacific), in press. Greenhill, L.J., Moran, J.M., & Henkel, C. 2002, in preparation. Hagiwara, Y., Diamond, P.J., & Miyoshi, M. 2002, A&A, 383, 65 Hagiwara, Y., Kohno, K., Kawabe, R., & Nakai, N. 1997, PASJ, 49, 171 Haschick, A.D., & Baan, W.A. 1985, Nature, 314, 144 Haschick, A.D., Baan, W.A., & Peng, E.W. 1994, ApJ, 437, L35 Heckman, T.M. 1980, A&A, 87, 152 Henkel, C., Gusten, R., Thum, C., & Downes, D. 1984a, IAU Circ. 3983 Henkel, C., Gusten, R., Wilson, T.L., Biermann, P., Downes, D., & Thum, C., 1984b, A&A, 141, L1 Herrnstein, J.R. 1997, Ph.D. Thesis, Harvard University Herrnstein, J.R., Greenhill, L.J., & Moran, J.M. 1996, ApJ, 468, L17 Herrnstein, J.R., Moran, J.M., Greenhill, L.J., Blackman, E.G., & Diamond, P.J. 1998, ApJ, 508, 243 Herrnstein, J.R., Greenhill, L.J., Moran, J.M., Diamond, P.J., Inoue, M., Nakai, N., & Miyoshi, M. 1998, ApJ, 497, L69 Herrnstein, J.R., Moran, J.M., Greenhill, L.J., Diamond, P.J., Miyoshi, M., Nakai, N., & Inoue, M. 1997, ApJ, 475, L17 Herrnstein, J.R., Moran, J.M., Greenhill, L.J., Diamond, P.J., Inoue, M., Nakai, N., Miyoshi, M., Henkel, C., & Riess, A. 1999, Nature, 400, 539 Ichimaru, S. 1977, ApJ, 214, 840 Ishihara, Y., Nakai, N., Iyomoto, N., Makishima, K., Diamond, P., & Hall, P. 2001, PASJ, 53, 215 Iyomoto, N., Fukazawa, Y., Nakai, N., & Ishihara, Y. 2001, ApJ, 561, L69 Iwasawa, K., Koyama, K., Awaki, H., Kunieda, H., Makishima, K., Tsuru, T., Ohashi, T., & Nakai, N. 1993, ApJ, 409, 155 Kartje, J.F., Konigl, A., & Elitzur, M. 1999, ApJ, 513, 180 Koekemoer, A.M., Henkel, C., Greenhill, L.J., Dey, A., van Breugel, W., Codella, C., & Antonucci, R. 1995, Nature, 378, 697 Lasota, J.-P., Abramowicz, M.A., Chen, X., Krolik, J., Narayan, R., & Yi, I. 1996, ApJ, 462, 142 Madejski, G.M., Zycki, P., Done, C., Valinia, A., Blanco, P., Rothschild, R., & Turek, B. 2000, ApJ, 535, L87 Makishima, K., Fujimoto, R., Ishisaki, Y., Kii, T., Loewenstein, M., Mushotzky, R., Serlemitsos, P., Sonobe, T., Tashiro, M., & Yaqoob, T. 1994, PASJ, 46, L77 Maloney, P.R., Begelman, M.C., & Pringle, J.E. 1996, ApJ, 472, 582 Maloney, P.R., Hollenbach, D.J., & Tielens, A.G.G.M. 1996, ApJ 466, 561 Maoz, E. 1995, ApJ, 447, L91 Maoz, E., & McKee, C. F. 1998, ApJ, 494, 218 43 Marconi, A., Moorwood, A.F.M., Origlia, L., & Oliva, E. 1994, Messenger, 78, 20 Matt, G., Fiore, F., Perola, G.C., Piro, L., Fink, H.H., Grandi, P., Matsuoka, M., Oliva, E., & Salvati, M. 1996, MNRAS, 281, L69 Menten, K.M., Melnick, G.J., & Phillips, T.G. 1990, ApJ, 350, L41 Menten, K.M., Melnick, G.J., Phillips, T.G., & Neufeld, D.A. 1990, ApJ, 363, L27 Miyoshi, M., Moran, J., Herrnstein, J., Greenhill, L., Nakai, N., Diamond, P., & Inoue, M. 1995, Nature, 373, 127 Moran, J.M., Greenhill, L.J., & Herrnstein, J.R. 1999, J. Astrophys. Astron., 20, 165 Moran, J., Greenhill, L., Herrnstein, J., Diamond, P., Miyoshi, M., Nakai, N., & Inoue, M. 1995, in Quasars and AGN: High Resolution Imaging, Proc. Nat. Acad. Sci., 92, 11427 Nakai, N., Inoue, M., & Miyoshi, M. 1993, Nature, 361, 45 Nakai, N., Inoue, M., Miyazawa, K., & Hall, P. 1995, PASJ 47, 771 Nakai, N., Sato, N., & Yamauchi, A. 2002, PASJ, 54, in press. Narayan, R., & Yi, I. 1994, ApJ, 428, L13 Neufeld, D.A. 2000, ApJ, 542, L99 Neufeld, D.A., & Maloney, P.R. 1995, ApJ, 447, L19 Neufeld, D.A., Maloney, P.R., & Conger, S. 1994, ApJ, 436, L127 Neufeld, D.A., & Melnick, G.J. 1991, ApJ, 368, 215 Newman, J.A., Ferrarese, L., Stetson, P.B., Maoz, E., Zepf, S.E., Davis, M., Freedman, W.L., & Madore, B.F. 2001, ApJ, 553, 562 Phinney, E.S. 1983, Ph.D. Thesis, University of Cambridge Pringle, J.E. 1996, MNRAS, 281, 357 Rees, M.J., Begelman, M.C., Blandford, R.D., & Phinney, E.S. 1982, Nature, 295, 17 Reynolds, C.S., Nowak, M.A., & Maloney, P.R. 2000, ApJ, 540, 143 Sawada-Satoh, S., Inoue, M., Shibata, K.M., Kameno, S., Migenes, V., Nakai, N., & Diamond, P.J. 2000, PASJ, 52, 421 Shakura, N.I., & Sunyaev, R.A. 1973 A&A 24, 337 Stone, J.M., Hawley, J.F., Gammie, C.F., & Balbus, S.A. 1996, ApJ, 463, 656 Svensson, R. 1999, in Theory of Black Hole Accretion Discs, ed. M.A. Abramowicz, G. Bjornsson, and J.E. Pringle, (Cambridge: Cambridge University Press), 289 Tielens, A.G.G.M., & Hollenbach, D.J. 1985, ApJ, 291, 722 Trotter, A.S., Greenhill, L.J., Moran, J.M., Reid, M.J., Irwin, J.A., & Lo, K.-Y. 1998, ApJ, 495, 470 Ulvestad J.S., Wrobel J.M., Roy A.L., Wilson, A.S., Falcke, H., & Krichbaum, T.P. 1999, ApJ, 517, L81 Veilleux, S., & Bland-Hawthorn, J., 1997 ApJ, 479, L105 Warwick, R.S., Koyama, K., Inoue, H., Takano, S., Awaki, H., & Hoshi, R. 1989, PASJ, 41, 709 Waters, J.W., Kakar, R.K., Kuiper, T.B.H., Roscoe, H.K., Swanson, P.N., Rodriguez Kuiper, E.N., Kerr, A.R., Thaddeus, P., & Gustincic, J.J. 1980, ApJ, 235, 57 Weaver, K.A., Wilson, A.S., Henkel, C., & Braatz, J.A. 1999, ApJ, 520, 130 Whiteoak, J.B. & Gardner, F.F., 1986, MNRAS, 222, 513 Xanthopoulos, E., & Richards, A.M.S. 2001, MNRAS, 326, L37 44 TABLE 1a Water Megamasers Observed with VLBI Masers With Disk Structure Galaxy NGC 4258 NGC 1068 Circinus NGC 4945 NGC 1386 NGC 3079 IC 2560b D Mpc 7 15 4 4 12 16 26 vφ km/s 1100 330 230 150 100 150 420 Ri/Ro pc 0.13/0.26 0.6/1.2 0.08/0.8 0.2/0.4 -- /0.7 -- /1.0 0.07/0.26 M ρ Lx 106 M⊙ 107 M⊙/pc3 1042 erg/s Reference 39 17 1 1 2 1 3 400 3 40 2 4 0.2 210 0.04 40 40 6 0.02 1 -- 10 0.1 M95 GG97 G01 G97 B99 T98, S00 I01 Masers Without Obvious Disk Structure Galaxy IRAS 22265 (S0) NGC 1052 (E4) D Mpc 100 20 v0 km/s 7570 1490 ∆v km/s 150 100 ∆R pc 2.4 0.06 Comment Reference messy "jet" G02 C98 aAdapted and updated from Moran, Greenhill, & Herrnstein (1999) D = distance, v0 = systemic velocity, ∆v = velocity range, ∆R = linear extent, vφ = rotational velocity, Ri/Ro = inner/outer radius of disk, M = central mass, ρ = central mass density, Lx = X-ray luminosity. bOnly the systemic velocity emission in IC 2560 has been imaged; it is included here because of kinematic evidence for a rotating disk (Ishihara et al. 2001). See discussion in text. References: G02 = Greenhill et al. 2002, C98 = Claussen et al. 1998, I01 = Ishihara et al. 2001, M95 = Miyoshi et al. 1995, GG97 = Greenhill & Gwinn 1997, G01 =Greenhill et al. 2001, G97 = Greenhill et al. 1997, B99 = Braatz et al. 1999, T98 = Trotter et al. 1998, S00 = Sawada-Satoh et al. 2000 45
astro-ph/0509185
1
0509
2005-09-07T18:00:28
Chandra and Hubble Study of a New Transient X-ray Source in M31
[ "astro-ph" ]
We present X-ray and optical observations of a new transient X-ray source in M31 first detected 23-May-2004 at R.A.=00:43:09.940 +/- 0.65'', Dec.=41:23:32.49 +/- 0.66''. The X-ray lightcurve shows two peaks separated by several months, reminiscent of many Galactic X-ray novae. The location and X-ray spectrum of the source suggest it is a low mass X-ray binary (LMXB). Follow-up HST ACS observations of the location both during and after the outburst provide a high-confidence detection of variability for one star within the X-ray position error ellipse. This star has $\Delta$B ~ 1 mag, and there is only a ~1% chance of finding such a variable in the error ellipse. We consider this star a good candidate for the optical counterpart of the X-ray source. The luminosity of this candidate provides a prediction for the orbital period of the system of 2.3$^{+3.7}_{-1.2}$ days.
astro-ph
astro-ph
Chandra and Hubble Study of a New Transient X-ray Source in M31 Benjamin F. Williams1, Michael R. Garcia1, Jeffrey E. McClintock1, Frank A. Primini1, and Stephen S. Murray1 ABSTRACT We present X-ray and optical observations of a new transient X-ray source in M31 first detected 23-May-2004 at R.A.=00:43:09.940 ±0.65′′, Dec.=41:23:32.49 ±0.66′′. The X-ray lightcurve shows two peaks separated by several months, reminiscent of many Galactic X-ray novae. The location and X-ray spectrum of the source suggest it is a low mass X-ray binary (LMXB). Follow-up HST ACS observations of the location both during and after the outburst provide a high- confidence detection of variability for one star within the X-ray position error ellipse. This star has ∆B ≈ 1 mag, and there is only a ≈1% chance of finding such a variable in the error ellipse. We consider this star a good candidate for the optical counterpart of the X-ray source. The luminosity of this candidate provides a prediction for the orbital period of the system of 2.3+3.7 −1.2 days. Subject headings: X-rays: binaries -- galaxies: close -- X-rays: stars individual (M31) -- binaries: 1. Introduction X-ray and optical investigations of bright transient Galactic low-mass X-ray binaries (LMXBs), known as X-ray novae (XRNe), have shown these sources to be some of the best stellar-mass black hole candidates known (McClintock & Remillard 2004, and references therein). During outburst, some of these objects are bright enough in X-rays ( > ∼ 1038 erg s−1) to be easily detected in nearby galaxies, and each new X-ray transient source found is a potential gem for studies of black hole and accretion physics. Over the past several years, we have undertaken a monitoring program of the nearby spiral galaxy M31 to search for new transient X-ray sources with Chandra and HST. Through 1Harvard-Smithsonian Center for Astrophysics, 60 Garden 02138; [email protected]; [email protected] [email protected]; [email protected]; Street, Cambridge, MA [email protected]; -- 2 -- this program, and the searches done by other groups using XMM-Newton, dozens of new transient sources have been found (e.g. Williams et al. 2004; Di Stefano et al. 2004; Kong et al. 2002; Trudolyubov et al. 2001; Osborne et al. 2001; Williams et al. in preparation). While the discovery of new XRN candidates is exciting, most of the X-ray transient sources detected in M31 so far have no known optical counterparts, making their classification and orbital parameters difficult to constrain. Our coordinated HST program has provided several of the first optical counterpart candidates for these transient X-ray sources, likely to be XRNe (Williams et al. 2005a,b,c, 2004). Knowledge of the optical counterparts begins the process of constraining the orbital periods of the binaries. As more X-ray/optical systems are found, we will be able to study the orbital period distribution for XRNe in M31. This distribution is one of the fundamental observable parameters for this class of objects that any viable model of their evolution must be able to reproduce. This paper describes our discovery and follow-up study of a new transient X-ray source in M31 detected by Chandra in observations spanning from 23-May-2004 through 04-Oct- 2004. In addition, we describe our efforts to determine the optical counterpart of the source using HST. Section 2 describes the data obtained and the methods used to process them. Section 3 provides our results. Section 4 discusses the implications of the results, including a prediction of the orbital period of the LMXB, and §5 gives our conclusions. 2. Data 2.1. X-ray We obtained six observations with the Chandra ACIS-I between December 2003 and December 2004 relevant to this study. The observation identification numbers, dates, point- ings, roll angles, and exposure times of these observations are given in Table 1. Although the observations were taken for 5 ks each, the effective exposure was reduced by ∼20% because the data were taken in "alternating readout mode" in order to avoid pileup for any bright transient source. These observations were all reduced using the software package CIAO v3.1 with the CALDB v2.28. We created exposure maps for the images using the task merge all,1 and we found and measured positions and fluxes of the sources in the images using the CIAO task 1http://cxc.harvard.edu/ciao/ahelp/merge all.html -- 3 -- wavdetect.2 Each data set detected sources down to (0.3 -- 10 keV) fluxes of ∼6×10−6 photons cm−2 s−1 or 0.3 -- 10 keV luminosities of ∼1036 erg s−1 for a typical X-ray binary system in M31. We aligned the coordinate system of the X-ray images with the optical images of the Local Group Survey (LGS; Massey et al. 2001). The positions of X-ray sources with known globular cluster counterparts were aligned with the globular cluster centers in the LGS images using the IRAF3 task ccmap. The errors of this alignment were typically ∼0.1′′. The precise alignment errors for the observations used to determine the X-ray error circle are given in Table 3. We checked the positions of all of the sources we detected in each observation with those of previously cataloged X-ray sources in M31 and with the Simbad4 database. Herein we focus on one bright source that did not appear in any previous catalogs. This position also had no cataloged source in Simbad within 10′′. We name this source CXOM31 J004309.9+412332, according to the IAU-approved naming convention described in Kong et al. (2002) and r3- 127 according to the short naming scheme described in Williams et al. (2004). This position is s 4.8′ east and 7.4′ north of the M31 nucleus, on the outskirts of the M31 bulge. We measured the position errors for r3-127 using the IRAF task imcentroid. These errors were affected by the size of the Chandra point spread function 9′ off-axis and were ∼1′′ for the brightest detections. The precise values for these errors are provided in Table 3. Upper-limits to the X-ray flux at this position in observations 4680 and 4723 were measured by determining the flux necessary to produce a detection 3-σ above the background flux. Finally, we extracted the X-ray spectrum of r3-127 from observations 4682 and 4721 using the CIAO task psextract,5 binning the spectrum in energy so that each bin contained > ∼ 10 counts. We then fit each spectrum using CIAO 3.1/Sherpa, trying both a power-law model with absorption and a disk blackbody model with absorption. The spectrum did not contain enough counts to constrain the foreground absorption. We therefore fixed the absorption to the typical Galactic foreground value (6×1020 cm−2). This value allowed good fits for both model types. Results are given in Table 4 and discussed in § 3. 2http://cxc.harvard.edu/ciao3.0/download/doc/detect html manual/Manual.html 3IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Associa- tion of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation. 4http://simbad.u-strasbg.fr/ 5http://cxc.harvard.edu/ciao/ahelp/psextract.html -- 4 -- 2.2. Optical We obtained two sets of observations of the position of r3-127 using the HST ACS. Each data set was obtained in a single orbit and was observed through the F435W (B equivalent) filter using the standard 4-point box dither pattern to allow for optimal spatial resolution. Each data set was observed with a total exposure time of 2200 seconds. The first of these data sets was obtained 02-Nov-2004, while the X-ray source was likely still active. The second was obtained 01-Jan-2005, after the source had faded below our Chandra detection limit (∼1036 erg s−1). Each data set was combined into a final, high-resolution photometric image using the PyRAF6 task multidrizzle,7 which has been optimized to process ACS imaging data. The task removes cosmic ray events and geometric distortions, and it drizzles the dithered frames together into a final, high-resolution, photometric image. We aligned the final HST images to the LGS coordinate system with ccmap using stars common to both data sets. The stars positions in both the LGS and ACS images were determined with imcentroid. The resulting alignment had rms errors of ≤0.03′′ (less than 1 ACS pixel). The ACS images, independently aligned with the LGS coordinate system, are shown in Figure 1. Resolved stellar photometry was then performed on the relevant sections of the final images with DAOPHOT II and ALLSTAR (Stetson et al. 1990). The count rates measured from our images were converted to VEGA magnitudes using the conversion techniques provided in the ACS Data Handbook8. 3. Results 3.1. X-ray Source r3-127 was clearly detected 4 times by our monitoring program. These observa- tions provided a lightcurve, precise positional constraints, and X-ray spectral measurements. The X-ray lightcurve of r3-127 is shown in Figure 2, and the measured fluxes and hard- ness ratios are provided in Table 2. Observations before the first detection allowed reliable 6PyRAF is a product of the Space Telescope Science Institute, which is operated by AURA for NASA. 7multidrizzle is a product of the Space Telescope Science Institute, which is operated by AURA for NASA. http://stsdas.stsci.edu/pydrizzle/multidrizzle 8http://www.stsci.edu/hst/acs/documents/handbooks/DataHandbookv2/ACS longdhbcover.html -- 5 -- upper limits to the X-ray flux from that position to be measured. Our final observation yielded a marginal detection with a signal-to-noise of 2. We treated this measurement as a detection keeping in mind that if it was spurious the high end of the error range gives the 3σ upper-limit of the flux during the final observation (2004-December-05). The lightcurve is complex, including a double-peak, as has been observed for several Galactic XRNe (see lightcurves in McClintock & Remillard 2004). There was a factor of 3 drop in flux followed by a factor of 4 recovery between the May and October observations. The brightest detection was the second peak, with a flux of (7.4±0.8) × 10−5 photons cm−2 s−1. The source faded by at least a factor of eight in the 55 days after the second peak. The 3σ flux upper limit for the 2004-December-05 observation (9×10−6 ph cm−2 s−1) yields an e-folding decay time after the second peak of <1 month. If the 2σ detection of (6±3)×10−6 ph cm−2 s−1 in the 2004-December-05 observation is real, then the e-folding decay time was ∼22 days. This decay was faster than that observed after the first peak, where the flux decayed by a factor of 2.8 in 55 days, exhibiting an e-folding decay time of ∼54 days. Although r3-127 was 9′ off-axis, the Chandra images provided a precise source position. We measured the position in the two brightest detections. Table 3 shows the significant sources of error in this measurement. These were the alignment errors, determined by ccmap, and the position errors, determined by imcentroid. We added these errors in quadrature for each measurement. Then we took the weighted mean R.A. and Dec. to obtain our final position and position error of the source in the LGS coordinate system, which we used to plot the 1σ error ellipses shown in Figure 1. The final X-ray position and errors were R.A = 0:43:09.940 ± 0.65′′ and Dec. = 41:23:32.49 ± 0.66′′. At the distance of M31 (780 kpc), the semi-major axis of the error ellipse is 2.5 pc. The two brightest detections contained sufficient counts to perform fits to the X-ray spectrum of r3-127. The results of the fits are given in Table 4. They show that, while the spectrum was well-fitted by both the absorbed disk blackbody and the absorbed power-law models, it was better fitted by the absorbed disk blackbody with an absorption-corrected 0.3 -- 7 keV luminosity of ∼1.1×1037 erg s−1 at the time of the brightest detection. The spectrum was soft, and it may have become softer during the decay. This softening can be seen in the hardness ratios, as HR1 declined from 0.6±0.2 at the first detection to 0.0±0.4 during the decay. There is also a hint that the second peak may have been softer than the first in both the hardness ratios and the spectral fits (see Tables 2 and 4). These soft spectra are typical of Galactic LMXBs, especially those systems that contain a black hole primary (e.g. Tanaka & Lewin 1995; Church & Baluci´nska-Church 2001; McClintock & Remillard 2004). Concisely, the X-ray lightcurve of r3-127 was complex, having at least two peaks. The -- 6 -- position was measured from the Chandra images with 0.7′′ precision, and the spectral fits provided an estimate of the peak absorption-corrected 0.3 -- 7 keV luminosity reached on 04- Oct-2004 of ∼1.1×1037 erg s−1. 3.2. Optical The two HST ACS images of the position of r3-127 are shown in Figure 1 adjacent to the most contemporaneous Chandra ACIS I images available. These data reveal one bright variable source in the southwest part of the error ellipse. By analyzing the completeness of our photometry and carefully examining all potential variable stars, we were able to justify focusing on this bright variable as the most likely optical counterpart of r3-127. The completeness of our photometry in the two epochs was determined by comparing the DAOPHOT results from the two epochs of HST data. Figure 3 shows the results of a completeness analysis of the ACS data in the region within 3′′ of the X-ray position of r3-127. The solid histogram shows the percentage of stars that were detected in the first epoch and not in the second epoch, as a function of magnitude. The dotted histogram shows the percentage of stars that were detected in the second epoch and not in the first epoch, as a function of magnitude. The completeness begins to fall off at B ∼ 26.5, so that our confidence in the photometry also begins to decrease at this magnitude. The completeness falls below 50% at B = 27.8. We therefore took B = 27.8 to be our limiting magnitude, assigning B > 27.8 as the upper-limit to all non-detections. A search for variable stars in the error ellipse of r3-127 was then performed. There were 10 stars detected inside the error ellipse by our DAOPHOT analysis of the first epoch that were more than 4σ brighter than B = 27.8 and changed in brightness by >4σ by the second epoch. Seven of the variable candidates were fainter than B = 26.9 during the first epoch, and they were not detected in the second epoch, likely due to completeness. We therefore removed these 7 stars from the pool of variables and possible counterparts. Two of the other three candidate variable stars had B > 26.5 in at least one epoch. At these faint magnitudes the photometry, like the completeness, was likely significantly affected by crowding issues (see Figure 3). Such faint stars are more likely to be blended with stars of similar brightness, which could change the photometry by ∼0.7 mag. One of these two variable candidates, marked with the triangles in Figure 1, was B = 26.80 ± 0.09 in the first epoch and B = 26.20 ± 0.10 in the second. It showed only a 4.5σ increase in brightness between epochs with ∆B = 0.60 ± 0.13 mag. As the variability was not highly significant and the photometry likely affected by crowding, we removed this star -- 7 -- from the pool of possible counterparts. The other faint variable candidate, marked with the boxes in Figure 1, showed a 4.9σ drop in brightness between epochs as measured by DAOPHOT. The star was B = 26.47±0.08 in the first epoch and B = 27.09 ±0.09 in the second, making it the brightest star that varied in concert with the X-ray flux. However, the star only faded by ∆B = 0.62 ± 0.13 mag, and its photometry could be significantly affected by crowding. We therefore removed it from the pool of possible counterparts. The only optical variable candidate left in the r3-127 error ellipse, marked with the arrow in Figure 1, is clearly seen by visual inspection of the two images. This candidate was B = 26.27 ± 0.07 in the first epoch and B = 25.29 ± 0.03 in the second epoch. This star exhibited a change in brightness of 0.98±0.07 mag. It varied with 13σ significance, clearly separating itself from the other candidate variables, which all showed variations of <5σ. The large difference in significance distinguished this variable from the other candidates, making it our strongest candidate counterpart for r3-127. Succinctly, three stars inside the r3-127 error ellipse had statistically significant (>4σ) brightness changes not easily attributable to completeness limitations. Two of these stars were removed as counterpart candidates of r3-127 because they had comparable brightness changes only slightly above threshold (4.5 -- 4.9 σ) and their photometry was likely significantly affected by crowding. The only remaining candidate (shown by the arrow in Figure 1) showed variability of much higher significance than any other star in the error ellipse. This candidate is therefore our strongest counterpart candidate for r3-127 and the only candidate that will be considered for the remainder of the paper. 4. Discussion Since most Galactic bright transient X-ray sources are either HMXBs or LMXBs, it is likely that r3-127 is one of these types of sources. Our X-ray and optical data suggest that it is an LMXB. If r3-127 is an LMXB and our candidate counterpart is correct, our data can be used to predict the orbital period of the binary. The brightest stars in the error ellipse of r3-127 have B = 25.0. Assuming a distance modulus of M31 of 24.47 (780 kpc; Williams 2003) and typical foreground extinction toward M31 of AB = 0.4, these stars have MB ∼ 0.1. This absolute magnitude is fainter than high- mass O and B stars which are the typical secondaries of high-mass X-ray binaries (HMXBs). Therefore both the faintness of any potential optical counterpart and the X-ray spectrum (see § 3) suggest that r3-127 is an LMXB. -- 8 -- The candidate counterpart was bright when our X-ray data suggest the source was faint, which is unexpected but does not rule the star out as the optical counterpart of r3-127. Such high-amplitude variables at these magnitudes in 2-epoch ACS photometry of M31 have a density of ∼27 arcmin−2 (Williams 2005). As our error circle covers only 3.8×10−4 arcmin2, there is only a ∼1% probability of such a variable randomly falling in our error circle. In addition, the lightcurve of r3-127 (see Figure 2) did not exhibit a simple monotonic decay over one to several months. At the very least, r3-127 exhibited a double-peak remi- niscent of XTE 1550-564 (Jain et al. 2001b). The low sample rate of our lightcurve and the differing decay times of the peaks highlight the possibility that r3-127 was exhibiting fast flaring, as has been seen in some Galactic transient events such as, for example, GRO 1655-40 (see Fig. 1 of Remillard et al. 1999) and GRS 1915+105 (see Fig. 1 of Rau et al. 2003). If the X-ray lightcurve was complex, it is not so surprising that the candidate counterpart did not vary as expected. For example, in XRN XTE J1550-564 the optical "reflare" was brighter than the optical flux during the initial X-ray outburst by more than half a magnitude (Jain et al. 2001b). Another flaring event in this source exhibited a secondary peak in the optical that was not observed in X-rays at all (Jain et al. 2001a). Source r3-127 could be a good candidate for such a reflare as the observed 0.3 -- 7 keV luminosity of the XRN (1.1×1037 erg s−1) is only ∼10% of the Eddington luminosity of a neutron star and only ∼1% of the Eddington luminosity of a typical stellar-mass black hole. These numbers suggest that only a small fraction of the mass of the disk was consumed by the outburst, allowing the potential for another accretion event that could have resulted in a high optical flux in the second epoch of ACS data. All of these arguments point to our candidate, the most clearly variable star in the r3-127 error ellipse, as the most likely optical counterpart to r3-127. With a strong candidate counterpart, we can predict the orbital period of r3-127. There is an empirical relation between the X-ray luminosity, optical luminosity, and orbital period of Galactic LMXBs during outburst (van Paradijs & McClintock 1994). The relation suggests that outbursts that are fainter in the optical have smaller accretion disks due to closer binary separation. Therefore the fainter the optical counterpart, the shorter the orbital period of the binary. More recently studied LMXBs have also been found to fit this relation. Williams et al. (2005a) applied the relation to the brightest observed X-ray and optical luminosities in a single outburst for several recently-discovered transient LMXBs. For example, the brightest optical and X-ray luminosities observed for 4U 1543-47 during the 1983 outburst were applied to the relation regardless of the specific timing of the observations. In most cases, the sources -- 9 -- followed the relation. Furthermore, (Williams et al. 2005c) performed detailed checks of the application of the relation when applied to optical and X-ray observations separated by weeks. For example, the photometry for A0620-00 from Esin et al. (2000) and for XTE J1550- 564 from Jain et al. (2001b) provided period predictions that were correct within the (rather substantial) errors. The X-ray and optical luminosities of r3-127 therefore provide a rough prediction of the orbital period of the binary system. The apparent complexity of the X-ray and optical lightcurves of r3-127 adds uncertainty to our application of the van Paradijs & McClintock (1994) relation, as we do not have precisely simultaneous X-ray and optical observations. On the other hand, the uncertainty introduced by the non-simultaneous nature of our measurements is small compared to the dynamic range of the relation. The relation holds over 8 optical magnitudes and 3 orders of magnitude in X-ray luminosity, and the optical variability between observations of the counterpart candidate is only 1 mag. Additionally, the intensity of both peaks in X-ray flux overlapped at the 1σ level. Therefore, even considering the complexity of the lightcurve and possibility of fast flaring, the brightest observed optical and X-ray luminosities can provide a reasonable orbital period prediction. Applying the van Paradijs & McClintock (1994) relation requires measurements of the optical and X-ray luminosity during outburst. Assuming the same foreground absorption we used for the X-ray spectral fits, and applying the conversion of Predehl & Schmitt (1995), the optical extinction toward r3-127 is AB = 0.44. In addition, the mean B−V color of LMXBs in the Liu et al. (2001) catalog is -0.09±0.14. Correcting for the extinction, color, and distance to r3-127, the brightest measured MV for the counterpart candidate (B = 25.29 ± 0.03), was MV = 0.47 ± 0.14. The brightest observed 0.3 -- 7 keV luminosity of r3-127, according to our best-fitting X-ray spectral model, was LX=1.1×1037 erg s−1. These optical and X-ray luminosities yield an orbital period prediction of P = 2.3+3.7 −1.2 days including the errors of the relation. The final errors change to P = 2.3+0.6 −0.4 days if the errors in the empirical relation are not taken into account. Succinctly, the data show that r3-127 is not likely an HMXB, and therefore it is likely an LMXB. In addition, the complex X-ray lightcurve makes the possibility of a complex optical lightcurve more likely, suggesting the bright variable star could be the optical coun- terpart even though it did not fade between observations. Considering the complex nature of the lightcurve, the highest X-ray and optical luminosity measurements provide our best prediction for the orbital period of the system: P = 2.3+3.7 −1.2 days. This range is consistent with the periods of many Galactic LMXB transient systems. -- 10 -- 5. Conclusions We have discovered a new bright transient X-ray source in M31 at R.A.=00:43:09.940 ±0.65′′, Dec.=41:23:32.49 ±0.66′′. We have named the source CXOM31 J004309.9+412332 and given it the shorter name of r3-127. This source was active for at least 5 months during 2004, but it has not been seen before or since, even in surveys with limiting fluxes nearly a factor of 100 fainter than our brightest detection of r3-127. The observed X-ray lightcurve was double-peaked, hinting that this source may be of a similar nature to some Galactic XRNe with complex lightcurves like XTE J1550-564 or GRO J1655-40 (McClintock & Remillard 2004). The X-ray spectrum of the source was soft, as is typical for LMXBs. It was well-fitted by both absorbed power-law and absorbed disk blackbody models, although the absorbed disk blackbody fits were somewhat better. The spectrum appeared marginally softer during the second half of the 5-month outburst. The best fit model had an inner disk temperature of 0.8 keV and an absorption-corrected 0.3 -- 7 keV peak luminosity of 1.1×1037 erg s−1 on 04-Oct-2004. Coordinated HST ACS observations of the position of r3-127 revealed no stars brighter that MB ∼ 0.1, confirming r3-127 is not an HMXB. The observations also detected one variable star at high confidence within the error ellipse of the X-ray outburst. This star is the strongest candidate counterpart of the X-ray transient source. The counterpart candidate brightened when our data suggest the X-ray source was faint, but the complexity of the lightcurve suggests such optical variations are reasonable. The highest optical and X-ray luminosity measurements yield a prediction for the orbital period of r3-127 of 2.3+3.7 −1.2 days. This predicted period range is large, and it includes the periods of many Galactic LMXB transient systems. The location and outburst date of r3-127, along with those of the other transient sources followed with HST in 2004 (s1-86, r2-70, and r2-71; Williams et al. 2005a,b,c, respectively), are consistent with the spatial distribution and rate of transients measured in Williams et al. (2004). We continue to find < ∼ 1 new transient source each month, and about half of them are in the bulge (within ∼7′ of the nucleus). None of these recent events has appeared in a cluster or shown a high-mass secondary, suggesting that the outbursts are not associated with star formation. Support for this work was provided by NASA through grant number GO-9087 from the Space Telescope Science Institute and through grant number GO-3103X from the Chandra X-Ray Center. MRG acknowledges support from NASA LTSA grant NAG5-10889. JEM acknowledges support from NASA ADP grant NNG-05GB31G. -- 11 -- REFERENCES Church, M. J., & Baluci´nska-Church, M. 2001, A&A, 369, 915 Di Stefano, R., et al. 2004, ApJ, 610, 247 Esin, A. A., Kuulkers, E., McClintock, J. E., & Narayan, R. 2000, ApJ, 532, 1069 Jain, R. K., Bailyn, C. D., Orosz, J. A., McClintock, J. E., & Remillard, R. A. 2001a, ApJ, 554, L181 Jain, R. K., Bailyn, C. D., Orosz, J. A., McClintock, J. E., Sobczak, G. J., & Remillard, R. A. 2001b, ApJ, 546, 1086 Kong, A. K. H., Garcia, M. R., Primini, F. A., Murray, S. S., Di Stefano, R., & McClintock, J. E. 2002, ApJ, 577, 738 Liu, Q. Z., van Paradijs, J., & van den Heuvel, E. P. J. 2001, A&A, 368, 1021 Massey, P., Hodge, P. W., Holmes, S., Jacoby, G., King, N. L., Olsen, K., Saha, A., & Smith, C. 2001, in American Astronomical Society Meeting, Vol. 199, 13005 McClintock, J. E., & Remillard, R. A. 2004, in Compact Stellar X-ray Sources (astro- ph/0306213) Osborne, J. P., et al. 2001, A&A, 378, 800 Predehl, P., & Schmitt, J. H. M. M. 1995, A&A, 293, 889 Rau, A., Greiner, J., & McCollough, M. L. 2003, ApJ, 590, L37 Remillard, R. A., Morgan, E. H., McClintock, J. E., Bailyn, C. D., & Orosz, J. A. 1999, ApJ, 522, 397 Stetson, P. B., Davis, L. E., & Crabtree, D. R. 1990, in ASP Conf. Ser. 8: CCDs in astronomy, 289 Tanaka, Y., & Lewin, W. H. G. 1995, in X-ray binaries (Cambridge Astrophysics Series, Cambridge, MA: Cambridge University Press, -- c1995, edited by Lewin, Walter H.G.; Van Paradijs, Jan; Van den Heuvel, Edward P.J.) Trudolyubov, S. P., Borozdin, K. N., & Priedhorsky, W. C. 2001, ApJ, 563, L119 van Paradijs, J., & McClintock, J. E. 1994, A&A, 290, 133 -- 12 -- Williams, B. F. 2003, MNRAS, 340, 143 Williams, B. F. 2005, AJ, submitted Williams, B. F., Garcia, M. R., Kong, A. K. H., Primini, F. A., King, A. R., Di Stefano, R., & Murray, S. S. 2004, ApJ, 609, 735 Williams, B. F., Garcia, M. R., Kong, A. K. H., Primini, F. A., & Murray, S. S. 2005a, ApJ, accepted Williams, B. F., Garcia, M. R., McClintock, J. E., Kong, A. K. H., Primini, F. A., & Murray, S. S. 2005b, ApJ, 628, 382 Williams, B. F., Garcia, M. R., Primini, F. A., McClintock, J. E., & Murray, S. S. 2005c, ApJ, accepted This preprint was prepared with the AAS LATEX macros v5.2. Table 1. Chandra ACIS-I observations ObsID Date R.A. (J2000) Dec. (J2000) Roll (deg.) Exp. (ks) 4680 4682 4719 4720 4721 4723 27-Dec-2003 23-May-2004 17-Jul-2004 02-Sep-2004 04-Oct-2004 05-Dec-2004 00 42 44.4 00 42 44.4 00 42 44.3 00 42 44.3 00 42 44.3 00 42 50.0 41 16 08.3 41 16 08.3 41 16 08.4 41 16 08.4 41 16 08.4 41 17 15.0 285.12 79.99 116.83 144.80 180.55 269.81 4.2 3.9 4.1 4.1 4.1 4.0 -- 13 -- Table 2. Chandra ACIS-I detections of r3-127 Date Counts Fluxa HR1b HR2c 23-May-2004 17-Jul-2004 02-Sep-2004 04-Oct-2004 05-Dec-2004 60 22 14 82 6 5.7±0.8 2.1±0.5 1.8±0.5 7.4±0.8 0.6±0.3 0.61±0.20 0.60±0.28 0.01±0.40 0.35±0.14 0.31±0.43 0.24±0.27 0.21±0.34 -0.20±0.48 -0.14±0.18 -1.1±1.3 aThe exposure corrected 0.3 -- 10 keV flux in units of 10−5 photons cm−2 s−1 bHardness ratio calculated by taking the ratio of M-S/M+S, where S is the number of counts from 0.3 -- 1 keV and M is the number of counts from 1 -- 2 keV. cHardness ratio calculated by taking the ratio of H-S/H+S, where S is the number of counts from 0.3 -- 1 keV and H is the number of counts from 2 -- 7 keV. Table 3. X-ray position determination of r3-127 ObsID R.A. (J2000) σAL a (′′) σpos b (′′) σtot c (′′) Dec. (J2000) σAL (′′) σpos (′′) σtot (′′) 4682 4721 Mean 0:43:10.067 0:43:09.899 0:43:09.940 0.09 0.14 · · · 1.31 0.74 · · · 1.31 0.75 0.65 41:23:32.73 41:23:32.27 41:23:32.49 0.12 0.11 · · · 0.95 0.91 · · · 0.96 0.92 0.66 aThe alignment error between the X-ray coordinate system and the optical coordinate system from the LGS, determined by ccmap. bThe position error for the source determined by imcentroid. cThe total error for the position, determined by adding the alignment and position errors in quadrature. -- 14 -- Table 4. Fits to the X-ray spectrum of r3-127 Date Model Norm.a Parameterb χ2/dof Qc d LX 23-May-2004 disk blackbody 23-May-2004 04-Oct-2004 04-Oct-2004 disk blackbody power-law power-law 0.010±0.009 (3.2±0.6)×10−5 0.022±0.016 (4.3±0.6)×10−5 0.9±0.2 1.9±0.3 0.8±0.1 2.0±0.2 3.10/4 6.31/4 4.28/5 4.42/5 0.54 0.18 0.51 0.49 10 12 11 16 aNormalization parameter for the model. For the power-law model the units are pho- tons keV−1 cm2 s−1 at 1 keV; for the disk blackbody model the parameter is the quantity ((Rin/km)/(D/10 kpc))2 × cos(θ). Where θ is the inclination angle of the accretion disk to the line of sight, D is the distance to the source, and Rin is the radius of the inner edge of the accretion disk. bThe final parameter in the model fit. For power-law models, this is the photon index of the spectrum. For disk blackbody models, this is the temperature (kT ) in keV of the inner edge of the accretion disk. cProbability, based on χ2 statistics, that the observed spectrum is a sample obtained from a source with an intrinsic spectrum equivalent to the model. cThe absorption-corrected 0.3 -- 7 keV luminosity of the source in units of 1036 erg s−1. -- 15 -- Chandra UT: 20:40, 04−Oct−2004 HST UT: 02:37, 02−Nov−2004 50.0 40.0 41:23:30.0 20.0 33.5 33.0 32.5 41:23:32.0 31.5 ) 0 0 0 2 J ( n o i t a n i l c e D 12.0 11.0 0:43:10.0 09.0 08.0 Right Ascension (J2000) 31.0 10.1 0:43:10.0 09.9 Right Ascension (J2000) Chandra UT: 08:52, 05−Dec−2004 HST UT: 14:48, 01−Jan−2005 50.0 40.0 41:23:30.0 20.0 33.5 33.0 32.5 41:23:32.0 31.5 ) 0 0 0 2 J ( n o i t a n i l c e D ) 0 0 0 2 J ( n o i t a n i l c e D ) 0 0 0 2 J ( n o i t a n i l c e D 12.0 11.0 0:43:10.0 09.0 08.0 Right Ascension (J2000) 31.0 10.1 0:43:10.0 09.9 Right Ascension (J2000) Fig. 1. -- Top left: The Chandra image of brightest X-ray detection of r3-127 is shown with the 1σ position error ellipse overplotted in white. Top right: The HST ACS F435W image of the position of r3-127 taken less than 1 month after the brightest X-ray detection. The black ellipse marks the X-ray position error (semi-major axis = 0.66′′, ∼2.5 pc), and the arrow marks the highest amplitude variable star inside the error ellipse. The black box marks the only source that showed a statistically significant drop in brightness between the two epochs not easily attributable to completeness. The black triangle marks another faint star that varied according to our DAOPHOT analysis. Bottom left: The Chandra image of our final X-ray observation of the position of r3-127 is shown with the 1σ position error ellipse overplotted in white. Although there are 10 counts within a 10′′ aperture around the location of the source, these were not sufficient to provide a 3σ detection. These counts provide a signal-to-noise ratio of 2 for a flux of (6±3)×10−6 ph cm−2 s−1, corresponding to a luminosity of 1.2×1036 erg s−1, showing that the source had faded by at least a factor of eight in the two months between observations. Bottom right: The HST ACS F435W image of the position of r3-127 taken less than 1 month after the X-ray non-detection. The ellipse, arrow, box, and triangle mark the same stars as in the upper-right panel. -- 16 -- HST (02−Nov−2004) HST (01−Jan−2005) (04−Oct−2004) (23−May−2004) (17−Jul−2004) (02−Sep−2004) (05−Dec−2004) −100 0 Days After Peak Detection 100 14 12 10 8 6 4 2 ) 1 − s g r e i 6 3 0 1 ( y t i s o n m u L V e k 7 − 3 . 0 (27−Dec−2003) 0 −300 −200 Fig. 2. -- The X-ray lightcurve of r3-127. The source was active for ∼ 5 months showing two clear peaks in luminosity. The arrow marks the upper-limit measured from the observation closest to the first detection. Each data point is labeled with its observation date. Long, labeled tick marks show the timing of our coordinated HST observations. The final data point is a 2σ peak detection of the source. If the detection is false, the top of the error bar on this point is the 3σ upper-limit to the luminosity of the source during our final observation. -- 17 -- 50 40 30 20 10 0 25.0 25.5 26.0 26.5 27.0 27.5 28.0 B magnitude of Detection i e c w T d e t c e t e D t o N e g a t n e c r e P Fig. 3. -- Solid histogram: The fraction of stars detected by DAOPHOT and ALLSTAR in the first epoch of ACS data, but not detected in the second epoch, as a function of the magnitude of the detection. Dotted histogram: The fraction of stars detected by DAOPHOT and ALLSTAR in the second epoch of ACS data, but not detected in the first epoch, as a function of the magnitude of the detection.
astro-ph/9806304
1
9806
1998-06-23T01:25:37
Starbursts, dark matter, and the evolution of dwarf galaxies
[ "astro-ph" ]
Optical and HI imaging of gas rich dwarfs, both dwarf irregulars (dI) and blue compact dwarfs (BCD), reveals important clues on how dwarf galaxies evolve and their star formation is regulated. Both types usually show evidence for stellar and gaseous disks. However, their total mass is dominated by dark matter. Gas rich dwarfs form with a range of disk structural properties. These have been arbitrarily separated them into two classes on the basis of central surface brightness. Dwarfs with \mu_{0}(B) <~ 22 mag arcsec^{-2} are usually classified as BCDs, while those faintwards of this limit are usually classified as dIs. Both classes experience bursts of star formation, but with an absolute intensity correlated with the disk surface brightness. Even in BCDs the bursts typically represent only a modest <~ 1 mag enhancement to the B luminosity of the disk. While starbursts are observed to power significant galactic winds, the fractional ISM loss remains modest. Dark matter halos play an important role in determining dwarf galaxy morphology by setting the equilibrium surface brightness of the disk.
astro-ph
astro-ph
STARBURSTS, DARK MATTER, AND THE EVOLUTION OF DWARF GALAXIES Gerhardt R. Meurer The Johns Hopkins University Baltimore MD, USA Abstract Optical and H i imaging of gas rich dwarfs, both dwarf irregulars (dI) and blue com- pact dwarfs (BCD), reveals important clues on how dwarf galaxies evolve and their star formation is regulated. Both types usually show evidence for stellar and gaseous disks. However, their total mass is dominated by dark matter. Gas rich dwarfs form with a range of disk structural properties. These have been arbitrarily separated them into two classes on the basis of central surface brightness. Dwarfs with µ0(B) <∼ 22 mag arcsec−2 are usually classified as BCDs, while those faintwards of this limit are usually classified as dIs. Both classes experience bursts of star formation, but with an absolute intensity correlated with the disk surface brightness. Even in BCDs the bursts typically represent only a modest <∼ 1 mag enhancement to the B luminosity of the disk. While starbursts are observed to power significant galactic winds, the fractional ISM loss remains modest. Dark matter halos play an important role in determining dwarf galaxy morphology by setting the equilibrium surface brightness of the disk. 1 Introduction Dwarf galaxies come in a variety of morphologies including: (1) dwarf elliptical (dE) - defined by smooth elliptical isophotes, and which invariably have a red color; (2) dwarf irregular (dI) - characterized by an irregular structure, blue colors and H ii regions scattered haphazardly over the optical face of the galaxy; and (3) blue compact dwarf (BCD) [45] - also an irregular morphology, differing from dI in that the H ii emission is usually highly concentrated into one or two high intensity patches near the center. BCD galaxies frequently are given other classifications such as Amorphous [35, 13] (sometimes prefaced with the adjectives "dwarf" or "blue"), or H ii galaxies [44]. However, the properties of dwarf (MB > −18 mag) galaxies 1 having these different monickers are virtually identical [22, 23], indicating that they represent the same physical phenomenon, hence I will refer to them all as BCDs. Despite their morphological differences dIs, BCDs, and dEs have similar optical structures - their radial profiles are exponential, at least at large radii [3, 4, 34, 43, 22]. Naturally this leads to speculation: are there evolutionary connections between these different morphologies? One commonly accepted evolutionary path was expounded by Davies & Phillips [7]. In it, the initial morphology is that of a dI. If its ISM manages to concentrate at the center of the galaxy a tremendous occurs starburst occurs resulting in a BCD morphology. Such starbursts are known to be capable of powering a powerful galactic wind (e.g. [9, 21, 25]). If the wind is strong enough all of the ISM is expelled resulting in a dE morphology. If some ISM remains, the system fades back into a dI, and undergoes a few more dI ⇔ BCD transitions before eventually expelling all of its ISM to become a dE. Here I will address the validity of this scenario by piecing together some work that I have been involved in [22, 23, 25, 26, 28, 30] (citation implicit throughout this review), as well as that of other researchers, that answers some smaller questions. What kind of dwarf galaxies host starbursts? What is the effect of starbursts on the ISM of dwarf galaxies? In addition I will consider another issue - dark matter (DM). This is the dominant form of mass in dwarf galaxies. Are the differences in dwarf galaxy morphology related to DM content? How does DM effect the evolution of dwarf galaxies? This review is heavily weighted towards gas rich dwarfs and in particular BCDs. In the Davies and Phillips scenario they are the active link between dI and dE galaxies, and hence should provide the best clues to deciphering dwarf galaxy evolution. The paper is laid out as follows: §2 compares the optical structure of dIs and BCDs; §3 details the H i structure and dynamics of two BCDs: NGC 1705 (D = 6.2 Mpc) and NGC 2915 (D = 3.1 Mpc), and compares them to dI galaxies; and §4 synthesizes the optical and radio results to form a new scenario where DM plays a dominant role in determining the morphology of gas rich dwarfs. Throughout this review I adopt distances based on H0 = 75 km s−1 Mpc−1. 2 The optical structure and classification of gas rich dwarfs Figure 1 illustrates typical radial surface brightness (µ) and color profiles of BCDs using NGC 1705 and NGC 2915 as examples. The exponential nature of the outer profiles is clearly demonstrated by the linear fits in the µ versus radius R plane. While in dwarf galaxies it is not clear that the stars responsible for this light are rotationally supported (in the lowest luminosity dEs they are not [24]), I will call this structure the disk for brevity's sake and be- cause both dIs and BCDs contain rotationally supported H i (§3). The central region usually displays relatively blue emission in excess of the extrapolated disk. I will call this structure the starburst because it is responsible for the starburst characteristics of BCDs, as is clear from the following evidence. Firstly, Hα imaging shows that the most intense H ii emission is confined to this region in BCDs. Secondly, HST [6, 27] and ground based [1, 5, 25, 26] imaging reveal numerous young blue star clusters in these structures. Finally, their blue broad band colors indicate they must be due to young stellar populations: they have ages of ∼ 10 Myr if instantaneous burst models are adopted or ∼ 100 Myr if constant star formation rate models are adopted. The strong Hα fluxes from the cores are more consistent with the constant star formation rate models (because ∼ 10 Myr old instantaneous bursts are no longer ionizing). In either case this is much less than the Hubble time, thus confirming their starburst nature. In comparison, the colors of the disk are typically like those of stellar populations forming con- [15, 34]), or a bit redder suggesting an tinuously over a Hubble time (i.e. like dI galaxies, cf. 2 Figure 1: Surface brightness profiles in B and R, with corresponding (B − R) color profiles of NGC 1705 [25] and NGC 2915 [26]. The dashed lines show exponential fits to the outer portions of the µ profiles. Note: the optical profiles of NGC 1705 exclude the light of the bright cluster NGC1705-1. inactive population. Hence, the disk is best attributed to the progenitor, or host, galaxy. Surface brightness profile fitting provides a means to determine both the relative strength of the starburst, and the structural properties of the disk. The method is simple. The outer portions of the profile are fitted with an exponential. This yields the extrapolated central surface brightness µ0 (which is corrected for inclination), and scale length α−1 of the disk. The burst and disk are separated by assuming that the disk remains exponential all the way into the center. Figure 2 shows the ratio of B band fluxes of BCD bursts relative to their host disks. The strongest starbursts are about twice as bright as their hosts. Hence, while starbursts can outshine the host disk they are nevertheless modest <∼ 1 mag enhancements to the total B flux of BCDs (cf. [39]). Typical flux enhancements are only a few tens of percent. This figure does not include upper limits to the burst/disk ratio: about 20% of BCDs have exponential profiles all the way into their cores (see below). The mass contribution of the starbursts is even smaller, typically <∼ 5%. These are not the ∼ 6 mag starburst enhancements proposed to explain the excess of faint blue galaxies at moderate redshifts [2]. Figure 3 compares the disk parameters µ0, α−1 of both BCDs and dI galaxies. I reemphasize that µ0 does not include the contribution of the starburst core. While there is some overlap, we see that BCD distribution is offset from that of dIs in terms of both µ0 and α−1. The offset in α−1 is in the sense that BCDs tend to be smaller than dIs. However this difference may be due to the (loose) selection criterion MB >∼ − 18 mag typically applied to dwarf galaxy samples: large scale length, high surface brightness galaxies are not dwarfs. More striking is the difference in µ0 (cf. [32]). Typically µ0 is 2.5 mag arcsec−2 more intense in BCDs than in dIs. Structurally BCD disks are very different from those of dIs. The absence of BCDs on the left half of Fig. 3 is puzzling. Does this mean that dI galaxies do not experience starbursts? Examination of the µ and color profiles of Patterson & Thuan [34] reveal several dIs (e.g. UGC 5706, UGC 7636) with µ profiles of like those in Fig. 1, that 3 Figure 2: (Left) Ratio of B band fluxes of starbursts to host disks in two samples of BCDs (symbol correspondence in right panel). Figure 3: (Right) Structural parameters of the exponential disks for a variety of samples of BCDs, and one dI sample. is containing a blue central "high" surface brightness excess relative to the exponential disk. This is structural evidence for starbursts in dIs. The episodic star-forming nature of dIs is best demonstrated using color-magnitude diagrams of the nearest ones (e.g. [14]). However the observed central surface brightness of the bursting dIs including the light of this central excess is typically µ(B) >∼ 22 mag arcsec−1, much fainter than the central regions of BCDs. While dIs do experience short duration bursts of star formation, they are pathetic and not usually recognized as starbursts because they are not intense enough. On a similar note, about 20% of BCDs (data from [22, 33, 42]) have µ profiles that are nearly exponential and have fairly flat color profiles (examples include Haro 14 [22] and UM483 [42]) - that is with no structural evidence for a starburst. The colors of these galaxies are consistent with fairly long duration ( >∼ 1 Gyr) star formation. These results highlight the importance of absolute surface brightness in morphological classification. For example, in the Virgo Cluster Atlas [36] galaxies are recognized as dwarfs by their low surface brightness. Similarly BCDs are recognized by their high surface brightness. This is largely an unstated, perhaps even unconscious, selection criteria. There are clearly low surface brightness dwarfs (e.g. GR8) that meet the usual luminosity, emission line, and size criteria for BCD classification [45] but are invariably classified as dIs. From Fig. 3 the BCD/dI dividing line is µ0(B) ≈ 22 mag arcsec−2. Clearly the presence of a starburst makes it more likely that a dwarf will be classified as a BCD. However, such a classification does not guarantee the presence of a strong burst. Nor does the dI classification exclude galaxies experiencing strong bursts relative to the host disk brightness. 3 H i structure and dynamics of BCDs Radio array observations of the H i structure and dynamics of BCDs can tell us much about star formation feedback (e.g. galactic winds) and allow the distribution of mass (including DM) 4 to be determined. Compared to dIs there are not that many H i imaging studies of BCDs; in part because it is harder to find BCDs that are resolvable with radio arrays, due to their small numbers and usual compact angular sizes. My collaborators and I therefore decided to address these issues with Australia Telescope Compact Array observations of NGC 1705 and NGC 2915, two of the nearer BCDs. The resultant composite H i and optical images are shown in the color section of this volume. Although both galaxies have some kinematic irregularities their dominant structures are extended rotating disks which are strongly centrally peaked. These are typical properties of BCDs imaged in HI [40, 41, 47]. The disk of NGC 2915 is so extended that it has the H i appearance of a late type barred spiral. Similar galaxies include IC 10 and NGC 4449 [48]. 3.1 Starburst -- ISM feedback Both NGC 1705 and NGC 2915 show evidence of star formation churning up the neutral ISM. In NGC 2915, kinks and enhancements in the velocity dispersion map correspond well to Hα bubbles and peculiar knots associated with recent star formation. However it does not appear that H i is being ejected from the system. In the center of the galaxy, where star formation is the most vigorous, σHI ≈ 40 km s−1 which is the same as the one dimensional velocity dispersion derived for the DM particles (see Fig. 4 below). Hence, star formation appears to be maintaining the central H i in virial equilibrium with the DM halo. This suggests that DM plays a role in the feedback process: if the starburst energizes H i to have σHI much larger than the halo velocity dispersion, then neutral ISM is thrown into the halo (or beyond) and star formation shuts down. Jutting out obliquely from NGC 1705's edge-on H i disk out to R ≈ 4.5 kpc, is a gaseous spur containing 8% of the H i flux (see color plate). Its position angle is similar to that of the optical outflow, hence this may be the one-sided blow out of the prominent Hα wind. Its outflow is inferred to be predominantly transverse, as is the Hα outflow, hence the expansion velocity Vexp, or similarly the expansion timescale texp, is uncertain. The Hα outflow has texp ≈ 10 Myr, like the age of the most prominent star cluster NGC1705-1. Adopting this texp for the H i outflow M ≈ 1.7 M⊙ yr−1, and total kinetic yields Vexp ≈ 300 km s−1, an average mass loss rate of energy Ek(gas) ≈ 1.5 × 1055 erg. If texp ≈ 100 Myr, closer to the age of the continuosly star M ≈ 0.17 M⊙ yr−1, and forming population in the core of NGC 1705, then Vexp ≈ 30 km s−1, M⋆ ≈ 0.13 M⊙ yr−1. Ek(gas) ≈ 1.5 ×1053 erg. In comparison, the current star formation rate is Mass loss is at least competitive with star formation in regulating the gas content of NGC 1705. Can the energetics of this gas be accounted for by NGC 1705's stellar populations? Using the solar metallicity Salpeter IMF (mass range 1 to 100 M⊙) population models of Leitherer & Heckman [20] I estimate the mechanical energy output (from stellar winds, supernovae) of the young stellar populations in NGC 1705 integrated over texp = 10 (100) Myr to be Ek(⋆) ≈ 2.3 (11.3) × 1055 erg. The uncertainties on both Ek(gas) and Ek(⋆) are probably a factor of few. Hence, within the considerable uncertainties, the energetics of the spur are are consistent with it being a starburst driven wind. The fate of NGC 1705's ejected gas depends critically on Vexp. If it is as high as 300 km s−1, then even the large amounts of DM in the system do not gravitationally bind it. If Vexp ≈ 30 Myr, some gas will be retained. At least 8% of the neutral ISM has been ejected into the halo of NGC 1705 if not out of the system entirely. Hence we are witnessing a significant mass ejection event. Nevertheless, even in NGC 1705, a BCD with one of the most spectacular Hα outflows, the majority of the ISM is retained in a disk. Even this starburst is incapable of totally blowing away the ISM. 5 Figure 4: (Left) Rotation curves of NGC 1705 [30] and NGC 2915 [28] with mass model fits. The thick solid lines shows the full models, the dotted, dashed, and dot-dashed lines show the contributions of the stars, H i disk, and DM halo respectively. The model parameters are listed in each panel. The vertical arrows indicate the optical Holmberg radius of each galaxy. Figure 5: (Right) DM halo central density ρ0 plotted against disk central surface brightness. Open squares are from de Blok & McGaugh [8], while diamonds represent NGC 1705 and NGC 2915. The top panel shows the results for maximimum disk model fits, while the bottom panel shows Bottema disk fits. The circles on the right side of the top panel mark crude estimates of ρ0 in 12 BCDs with published and unpublished RCs. The dotted line, at bottom, is a fit to the Bottema disk results with the relationship log(ρ0) = 0.4 log(µ0) + Constant. 3.2 Mass distributions Figure 4 shows the derived rotation curves (RCs) of NGC 1705 and NGC 2915 with mass model fits to them. These are the first two BCDs that have been mass modelled. The models consist of three components to the mass distribution: (1) the stellar distribution which is given by the projected luminosity profile (Fig 1) scaled by (M/LB)⋆ -- the mass to light ratio of the stars; (2) the neutral ISM distribution which is set by the H i profile scaled by a constant 1.33 to account for the Helium contribution (i.e. no free parameters); and (3) a dark matter halo. This halo is taken to be a pseudo-isothermal sphere with a density distribution given by ρ = ρ0 1 + (R/Rc)2 (1) where the free parameters are the central density ρ0 and the core radius Rc. From these, the rotational velocity at large R, V∞, and halo velocity dispersion σ0 are given by [19]: V 2 ∞ = 4πGρ0R2 c = 4.9σ2 0. (2) The models shown are maximum disk mass models, where (M/LB)⋆ is set by the first points in the RC, and then held fixed. Maximum disk models are by nature minimum halo models, hence ρ0 is a lower limit. We also made minimum disk models ((M/LB)⋆ = 0) and "Bottema Disk" models where (M/LB)⋆ is set by the color of the optical disk [8]. 6 There are a few things to note about Fig. 4. First, the form of the RCs is like that seen in normal disk galaxies, consisting of a rising inner portion and a more or less flat RC thereafter. Second, the optical extent of the galaxy, as marked by the arrows is contained within the rising portion of the RC. In fact the optical extent marks very well the RC knee (or equivalently Rc). The optical portions of dIs are also commonly contained within the rising portion of their RCs [37]. This is where the rotation is almost solid body, and thus shear is minimized, enhancing the ability of clouds to form stars [16]. Thirdly, in both cases DM dominates the mass distribution, even within the optical radius of the galaxy. In comparison, the stellar component has a mass equal to or less than the neutral gas disk. Overall, the global dynamics of BCDs appear to be similar to dIs: they are dominated by rotating disks with normal looking RCs. A distinction between the two types is seen when the DM halo densities ρ0 are compared, as shown in Fig. 5. Central densities found by maximum disk and Bottema disk fits are shown in separate panels. The comparison sample is taken from de Blok & McGaugh [8], and includes only galaxies with MB > −18 mag. It is com- prised mostly of dIs (left side of panel), but also includes three low luminosity spirals having µ0(B) ≈ 22 mag arcsec−2. This comparison shows that NGC 1705 and NGC 2915 have two of the highest ρ0 measurements of any dwarf galaxies. In order to check that these galaxies are typical, I crudely estimated ρ0 from the central velocity gradient for 12 BCDs with published or unpublished RCs, and plotted them as circles at arbitrary µ0 in the top panel of Fig. 5. These estimates are upper limits, since the contribution of the baryonic components to the velocity gradients have not been removed. Nevertheless, the comparison indicates that NGC 1705 and NGC 2915 have normal ρ0 for BCDs. Figure 5 shows a weak but noticeable correlation be- tween log(ρ0) and µ0(B), with higher surface brightness disks corresponding to higher ρ0 halos. This result holds for both maximum disk and Bottema disk solutions, and was first noted (for Bottema disks) by de Blok & McGaugh [8], who show that it extends to high luminosity disks. Here we show that the correlation also includes BCDs. 4 Evolutionary Connections The correlation in Fig. 5 can readily be explained by considering the response of a self gravi- tating disk immersed in a DM halo core of constant density ρ0, i.e. where the rotation curve is linearly rising. Then, the angular frequency Ω, ρ0, and dynamical time tdyn are constant with radius and related by ρ0 = 3Ω2 4πG , tdyn = π 2Ω , (3) (4) and Toomre's [46] disk stability parameter Q is given by Q = 2σgΩ πGΣg , where σg is the gas velocity dispersion and Σg is the disk surface density. Lower values of Q indicate a higher degree of self gravity. In normal disk galaxies Q typically remains radially constant at Q ≈ 2, and star formation correlates with regions of somewhat lower Q [17, 12]. This suggests that star formation regulates disk structure to retain constant Q. Kennicutt [18] finds that for normal disk galaxies the star formation rate per area Σ⋆ is given by Σ⋆ ∝ Σg tdyn . 7 (5) As noted in §2 the optical disks of dwarf galaxies typically have colors indicating a long duration of star formation. Let us assume that these are normal disks and hence form stars with the above star formation law and at a constant and universal Q. Then equations 3, 4 and 5 can be combined to yield Σ⋆ ∝ σgρ0. (6) For a constant star formation rate population and a sufficiently blue passband (e.g. B) the linearly surface brightness is proportional to Σ⋆. Therefore, for DM dominated galaxies we expect a simple correlation between linear surface brightness and ρ0. This is consistent with the observed correlation, as shown by the dotted line in Fig. 5. Meurer et al. [29] find a similar correlation between surface brightness and ρ0 holds in the center of starburst galaxies. However for them it is normal baryonic matter that dominates ρ0 rather than DM. In essence, the central mass density determines the equilibrium star formation rate of the embedded disk. Following the discussion in §2, the disk intensity largely determines whether a dwarf galaxy is classified as a BCD or dI. As noted earlier, the optical size of dwarfs seems to be limited to Rc. Hence, both DM halo parameters are important in governing the morphology of gas rich dwarfs. Bottema disk mass model decompositions [8] indicate that the two DM halo parameters are uncorrelated. This is contrary to cosmological simulations showing DM halos forming a one parameter (mass) sequence [31]. The origins and implications of this contradiction deserve further investigation. Can there be evolution between dI and BCD classes? While some evolution in ρ0 may be allowed, it is unlikely that there can be enough to change a typical dI into a typical BCD. That would require a 2.5 mag arcsec−2 change in µ0 or equivalently a factor of ten change in ρ0. The most obvious way to do that is to expand or contract the halo by a factor of e = 101/3 ≈ 2.2 following a mass loss (wind) or gain (accretion or spin down of an extended disk). The mass loss/gain fraction f required to effect this change is given by e = 1 1 − f (7) for homologous expansion/contraction that is slow relative to tdyn [49]. A factor of 10 change in ρ0 thus requires a 55% mass loss or gain. The problem is that there isn't that much baryonic mass in a dwarf galaxy. To effect this large of a change would require DM loss or gain. This is not feasible if DM is non-dissipative and feels only the force of gravity, as is usually assumed. I conclude that there is probably little dI ⇔ BCD evolution. If the ISM were removed from a dI or BCD, it could still plausibly evolve into a dE (or dwarf spheroidal) galaxy. However, as noted in § 3 even in a dwarf galaxy undergoing a strong starburst with a spectacular galactic wind (NGC 1705), the fractional loss of the ISM is modest. If this is typical, it would take on order of 10 bursts to expel all the ISM from a BCD. The bursts aren't strong enough, and the ISM distributions are too flattened to allow a single burst expulsion of the ISM [11]. Single bursts should have a more profound effect on the chemical evolution of dwarf galaxies, since the hot metal enriched ISM is preferentially lost in a starburst driven wind [10]. This view is consistent with arguments by Skillman & Bender [38] against evolution between gas rich and gas poor morphologies occurring commonly at the present epoch. The demographics of dwarf galaxy morphologies point to an environmental component to their evolution. Gas rich dwarfs are found in low density environments where the frequency of external starburst triggers is low. They survive easily. The clock runs faster (more frequent triggers) in clusters, and in addition ram pressure stripping would accelerate the removal of gas from dwarfs, while tidal truncation of DM halos would assist galactic wind losses. Hence it is not surprising that gas poor dEs are found more often in clusters than the field. 8 5 Conclusions We are now at a position to re-evaluate the Davies and Phillips [7] scenario for dwarf galaxy evolution. The mechanisms they invoke have clearly been verified. Dwarf galaxies do experience starbursts and these can expel some of the ISM. Mass expulsion can rival or surpass lock up into stars in regulating the gas content of dwarfs. However, the results of any single burst are not so severe. Cataclysmic bursts are not common at the present epoch, and the milder bursts that are observed may not be sufficient to change a galaxy's morphological classification. The morphology of a dwarf galaxy is largely set by its enveloping dark halo, and is relatively impervious to starbursts. Acknowledgements. My collaborators on the papers relevant to this review were Sylvie Beaulieu, Carla Cacciari, Claude Carignan, Michael Dopita, Ken Freeman, Tim Heckman, Neil Killeen, Glen Mackie, Amanda Marlowe, and Lister Staveley-Smith. I thank them for all their efforts. I am grateful to Liese van Zee, Eric Wilcots, and Crystal Martin for providing during the conference additional data that went into Fig 5, and to John Salzer for useful discussions. References [1] Arp, H. & Sandage, A. 1985, Astron. J. 90, 1163 [2] Babul, A., & Ferguson, H.C. 1996, Astrophys. J. 458, 100 [3] Bothun, G.D., Mould, J.R., Caldwell, N., & MacGillivray, H.T. 1986, Astron. J. 92, 1007 [4] Caldwell, N., & Bothun, G. 1987, Astron. J. 94, 1126 [5] Caldwell, N., & Phillips, M.M. 1989, Astrophys. J. 338, 789 [6] Calzetti, D., Meurer, G.R., Bohlin, R.C., Garnett, D.R., Kinney, A.L., Leitherer, C., & Storchi-Bergmann, T. 1997, Astron. J. 114, 1834 [7] Davies, J.I., & Phillips, S. 1988, MNRAS 233, 553 [8] de Blok, W.J.G., & McGaugh, S.S. 1997, MNRAS 290, 533 [9] Dekel, A., & Silk, J. 1986, Astrophys. J. 303, 39 [10] De Young, D.S., & Gallagher, III, J.S. 1990, Astrophys. J. 356, L15 [11] De Young, D.S., & Heckman, T.M. 1994, Astrophys. J. 431, 598 [12] Ferguson, A.M.N. 1997, Ph.D. Thesis, The Johns Hopkins University [13] Gallagher, III, J.S. & Hunter, D.A. 1987, Astron. J. 94, 43 [14] Gallagher, III, J.S., Tolstoy, E., Dohm-Palmer, R.C., Skillman, E.D., Cole, A.A., Hoessel, J.G., Saha, A., & Mateo, M. 1998, Astron. J. 115, 1869 [15] Hunter, D.A., & Gallagher, III, J.S. 1985, Astrophys. J. Suppl. Ser. 58, 533 [16] Kenney, J.D.P., Carlstrom, J.E., & Young, J.S. 1993, Astrophys. J. 418, 687 [17] Kennicutt, R.C. 1989, Astrophys. J. 344, 685 [18] Kennicutt, R.C. 1998, Astrophys. J. 498, 541 [19] Lake, G., Schommer, R.A., & van Gorkom, J.H. 1990, Astrophys. J. 320, 493 [20] Leitherer, C., & Heckman, T.M. 1995, Astrophys. J. Suppl. Ser. 96, 9 [21] Marlowe, A.T., Heckman, T.M., Wyse, R.F.G., & Schommer, R. 1995, Astrophys. J. 438, 563 9 [22] Marlowe, A.T., Meurer, G.R., Heckman, T.M., & Schommer, R. 1997, Astrophys. J. Suppl. Ser. 112, 285 [23] Marlowe, A.T., Meurer, G.R., & Heckman, T.M. 1998, Astrophys. J., submitted [24] Mateo, M., Olszewski, E., Welch, D.L., Fischer, P., & Kunkel, W. 1991, Astron. J. 102, 914 [25] Meurer, G.R., Freeman, K.C., Dopita, M.A., & Cacciari, C. 1992, Astron. J. 103, 60 [26] Meurer, G.R., Mackie, G., & Carignan, C. 1994, Astron. J. 107, 2021 [27] Meurer, G.R., Heckman, T.M., Leitherer, C., Kinney, A., Robert, C., & Garnett D.R. 1995, Astron. J. 110, 2665 [28] Meurer, G.R., Carignan, C., Beaulieu, S., & Freeman, K.C. 1996, Astron. J. 111, 1551 [29] Meurer, G.R., Heckman, T.M., Lehnert, M.D., Leitherer, C., & Lowenthal, J. 1997, Astron. J. 114, 54 [30] Meurer, G.R., Staveley-Smith, L., & Killeen, N.E.B. 1998 MNRAS, accepted (astro- ph/9806261) [31] Navarro, J.F., Frenk, C.S., & White, S.D.M. 1990, Astrophys. J. 490, 493 [32] Norton, S., & Salzer, J.J. 1997, Bull. American Astron. Soc. 190, 40.01 [33] Papaderos, P., Loose, H.-H., Thuan, T.X., & Fricke, K.J. 1996, Astr. Astrophys. Suppl. Ser. 120, 207 [34] Patterson, R., & Thuan, T. 1996 Astrophys. J. Suppl. Ser. 107, 103 [35] Sandage, A. & Brucato, R. 1979, Astron. J. 84, 472 [36] Sandage, A., Binggeli, B., and Tammann, G.A. 1985, Astron. J. 90, 1759 [37] Skillman, E.D. 1987, in Star Formation in Galaxies, ed. C.J. Lonsdale Persson, (NASA Conf. Pub. CP-2466), p. 263 [38] Skillman, E.D., & Bender, R. 1995, Rev. Mex. A.A. 3, 25 [39] Sudarsky, D.L, & Salzer, J.J. Bull. American Astron. Soc. 186, 39.04 [40] Simpson, C. 1998, this volume [41] Taylor, C.L., Brinks, E., Pogge, R.W., & Skillman, E.D. 1994, Astron. J. 107, 971 [42] Telles, E., & Terlevich, R. 1997, MNRAS 286, 183 [43] Telles, E., Melnick, J., & Terlevich, R. 1997, MNRAS 288, 78 [44] Terlevich, R., Melnick, J., Masegosa, J., Moles, M., & Copetti, M.V.F. 1991, Astr. Astro- phys. Suppl. Ser. 91, 285 [45] Thuan, T.X. & Martin G.E. 1981, Astrophys. J. 247, 823 [46] Toomre, A. 1964, Astrophys. J. 139, 1217 [47] van Zee, L. 1998, this volume [48] Wilcots, E. 1998, this volume [49] Yoshii, Y., & Arimoto, N. 1987, Astr. Astrophys. 188, 13 10 This figure "colorplate.jpg" is available in "jpg"(cid:10) format from: http://arxiv.org/ps/astro-ph/9806304v1
astro-ph/9611071
1
9611
1996-11-10T18:03:36
Numerical Simulations Of Advective Flows Around Black Holes
[ "astro-ph" ]
We present fully time-dependent solutions of accretion processes on black holes and compare them with analytical solutions. We use Smoothed Particle Hydrodynamics and Total Variation Diminishing mathods as the numerical tenchniques. Apart from steady state solutions, we also obtain time dependent results where the shock oscillates with the characteristic period of quasi-periodic oscillations (QPOs) observed in galactic and extragalactic black hole candidates. This oscillation produces 10-15% variation in the hard X-rays in galactic candidates and in soft X-rays in Extragalactic candidates.
astro-ph
astro-ph
Numerical Simulations Of Advective Flows Around Black Holes Sandip K. Chakrabarti1, D. Ryu2, D. Molteni3, H. Sponholz4, G. Lanzafame5, G. Eggum6 1. Tata Institute Of Fundamental Research, Mumbai, 400005, INDIA 2. Chungnam National University, Daejeon, SOUTH KOREA 3. Istitut di Fisica, Via Archirafi 36, 90123 Palermo, ITALY 4. University of Kentucky, Lexington, USA 5. Osservatorio di Catania, Catania, Sicily, ITALY 6. Los Alamos National Laboratory, Los Alamos, NM, USA Observational results of compact objects are best understood using advec- tive accretion flows (Chakrabarti, 1996, 1997). We present here the results of numerical simulations of all possible types of such flows. Two parameter (specific energy E and specific angular momentum λ) space of solutions of inviscid advective flow is classified into 'SA' (shocks in accretion), 'NSA' (no shock in accretion), 'I' (inner sonic point only), 'O' (outer sonic point only) etc. (Fig. 1 of Chakrabarti, 1997 and references therein). Fig. 1a shows examples of solutions (Molteni, Ryu & Chakrabarti, 1996; Eggum, in prepara- tion) from 'SA', 'I' and 'O' regions where we superpose analytical (solid) and numerical simulations (short dashed curve is with SPH code and medium dashed curve is with TVD code; very long dashed curve is with explicit/implicit code). The agreement is excellent. In presence of cooling effects, shocks from 'SA' os- cillate (Fig. 1b) when the cooling timescale roughly agrees with postshock infall time scale (Molteni, Sponholz & Chakrabarti, 1996). The solid, long dashed and short dashed curves are drawn for T 1/2 (bremsstrahlung), T 0.4 and T 0.75 6 9 9 1 v o N 0 1 1 v 1 7 0 1 1 6 9 / h p - o r t s a : v i X r a Fig. 1: (a) Numerical simulations and theoretical calculations of shock-free (O, I) and shocked (SA) flows. (b) Oscillation of shocks when cooling effects are included. 1 cooling laws respectively. In the absence of steady shock solutions, shocks for parameters from 'NSA' oscillate (Fig. 2) even in the absence of viscosity (Ryu et al. 1997). The oscillation frequency and amplitude roughly agree with those of quasi-periodic oscillation of black hole candidates. When the flow starts from a cool Keplerian disk, it simply becomes sub-Keplerian before it enters through the horizon. Fig. 3a shows this behaviour where the ratio of λ/λKeplerian is plotted. When the flow deviates from a hot Keplerian disk, it may develop a standing shock as well (Fig. 3b) (Molteni et al. 1996). Fig. 2: Oscillation of shock location when the flow has two sonic points but a steady shock condition is not satisfied. Fig. 3: (a) Ratio λdisk/λKep as obtained from a 1D numerical simulation. (b) Another 1D simulation which shows formation of sub-Keplerian flows and shocks from a hot Keplerian disk. References Chakrabarti, S.K. 1996, Phys. Rep., 266, 229 (Chapter V) Chakrabarti, S.K. 1997 (this volume) Molteni, D., Chakrabarti, S.K., Bisikalo, D. & Kuznetzov, O. 1996, MNRAS. Molteni, D., Ryu, D. & Chakrabarti, S.K. 1996, ApJ(Oct. 10th) Molteni, D., Sponholz, H., & Chakrabarti, S.K. 1996, ApJ, 457, 805 Ryu, D., & Chakrabarti, S.K., & Molteni, D. 1997, ApJ(Jan. 1st) 2 To appear in: 'Accretion Phenomena and Related Outflows', proceedings of 163rd IAU Symposium, July-1996, Eds. D. Wickramsinghe, L. Ferrario and G. Bicknell. SKC's address AFTER November 26th, 1996: Prof. S.K. Chakrabarti S.N. Bose National Center for Basic Sciences JD Block, Sector -III, Salt Lake Calcutta 700091, INDIA e-mail: [email protected] OR [email protected] 3 n o i t a c o L k c o h S 45 40 35 30 25 20 15 10 5 0 5000 10000 15000 Time 20000 25000 30000
0811.3046
2
0811
2010-01-27T07:14:55
New SETI Sky Surveys for Radio Pulses
[ "astro-ph" ]
Berkeley conducts 7 SETI programs at IR, visible and radio wavelengths. Here we review two of the newest efforts, Astropulse and Fly's Eye. A variety of possible sources of microsecond to millisecond radio pulses have been suggested in the last several decades, among them such exotic events as evaporating primordial black holes, hyper-flares from neutron stars, emissions from cosmic strings or perhaps extraterrestrial civilizations, but to-date few searches have been conducted capable of detecting them. We are carrying out two searches in hopes of finding and characterizing these mu-s to ms time scale dispersed radio pulses. These two observing programs are orthogonal in search space; the Allen Telescope Array's (ATA) "Fly's Eye" experiment observes a 100 square degree field by pointing each 6m ATA antenna in a different direction; by contrast, the Astropulse sky survey at Arecibo is extremely sensitive but has 1/3,000 of the instantaneous sky coverage. Astropulse's multibeam data is transferred via the internet to the computers of millions of volunteers. These computers perform a coherent de-dispersion analysis faster than the fastest available supercomputers and allow us to resolve pulses as short as 400 ns. Overall, the Astropulse survey will be 30 times more sensitive than the best previous searches. Analysis of results from Astropulse is at a very early stage. The Fly's Eye was successfully installed at the ATA in December of 2007, and to-date approximately 450 hours of observation has been performed. We have detected three pulsars and six giant pulses from the Crab pulsar in our diagnostic pointing data. We have not yet detected any other convincing bursts of astronomical origin in our survey data. (Abridged)
astro-ph
astro-ph
New SETI Sky Surveys for Radio Pulses Andrew Siemiona,d, Joshua Von Korffc, Peter McMahond,e, Eric Korpelab, Dan Werthimerb,d, David Andersonb, Geoff Bowera, Jeff Cobbb, Griffin Fostera, Matt Lebofskyb, Joeri van Leeuwena, Mark Wagnerd aUniversity of California, Berkeley - Department of Astronomy, Berkeley, California, USA bUniversity of California Berkeley - Space Sciences Laboratory, Berkeley, California, USA cUniversity of California, Berkeley - Department of Physics, Berkeley, California, USA dUniversity of California, Berkeley - Berkeley Wireless Research Center, Berkeley, California, USA eStanford University - Department of Computer Science, Stanford, California, USA 0 1 0 2 n a J 7 2 ] h p - o r t s a [ 2 v 6 4 0 3 . 1 1 8 0 : v i X r a Abstract Berkeley conducts 7 SETI programs at IR, visible and radio wavelengths. Here we review two of the newest efforts, Astropulse and Fly's Eye. A variety of possible sources of microsecond to millisecond radio pulses have been suggested in the last sev- eral decades, among them such exotic events as evaporating primordial black holes, hyper-flares from neutron stars, emissions from cosmic strings or perhaps extraterrestrial civilizations, but to-date few searches have been conducted capable of detecting them. The recent announcement by Lorimer et al. of the detection of a powerful (≈ 30 Jy) and highly dispersed (≈ 375 cm−3 pc) radio pulse in Parkes multi-beam survey data has fueled additional interest in such phenomena. We are carrying out two searches in hopes of finding and characterizing these µs to ms time scale dispersed radio pulses. These two observing programs are orthogonal in search space; the Allen Telescope Array's (ATA) "Fly's Eye" experiment observes a 100 square degree field by pointing each 6m ATA antenna in a different direction; by contrast, the Astropulse sky survey at Arecibo is extremely sensitive but has 1/3,000 of the instantaneous sky coverage. Astropulse's multibeam data is transferred via the internet to the computers of millions of volunteers. These computers perform a coherent de-dispersion analysis faster than the fastest available supercomputers and allow us to resolve pulses as short as 400 ns. Overall, the Astropulse survey will be 30 times more sensitive than the best previous searches. Analysis of results from Astropulse is at a very early stage. The Fly's Eye was successfully installed at the ATA in December of 2007, and to-date approximately 450 hours of observation has been performed. We have detected three pulsars (B0329+54, B0355+54, B0950+08) and six giant pulses from the Crab pulsar in our diagnostic pointing data. We have not yet detected any other convincing bursts of astronomical origin in our survey data. Keywords: SETI, radio transients, Allen Telescope Array, ATA, Arecibo 1. Introduction The Berkeley SETI group conducts seven searches at visible, IR and radio wavelengths, covering a wide vari- ety of signal types and spanning a large range of time scales: SETI@home searches for radio signals with time scales ranging from mS to seconds. SEVENDIP searches for nS time scale pulses at visible wavelengths. ASTROPULSE and Fly's Eye search for dispersed uS and mS time scale radio pulses from extraterrestrial civilizations, pulsars, or evaporating primordial black holes. SERENDIP and SPOCK search for continuous narrow band signals in the radio and optical bands re- Preprint submitted to Elsevier spectively. DYSON searches for infrared excess from advanced civilizations that use a lot of energy. SETI@home II, ASTROPULSE and SERENDIP V are radio sky surveys at the 300 meter Arecibo tele- scope. Commensal observations have been conducted almost continuously for the past ten years and are ongo- ing. Most beams on the sky visible to the Arecibo tele- scope have been observed four or more times. We rank SERENDIP and SETI@home candidate signals based on the number of independent observations, the strength of the signals, the closeness of the signals in frequency and sky position, and the proximity to stars, planetary systems, galaxies, and other interesting astronomical October 22, 2018 objects. SETI@home uses the CPU power of volun- teered PCs to analyze data. Five million people in 226 countries have participated. Combined, their PCs form Earth's second most powerful supercomputer, averaging 482 TeraFLOPs and contributing over two million years of CPU time. Here we describe two of the newest SETI searches, Astropulse and Fly's Eye. 2. SETI Pulse Searches One common assumption of SETI is that an alien civilization wishing to make contact with others would broadcast a signal that is easily detected and easily dis- tinguished from natural sources of radio emission. One way of achieving these goals is to send a narrow band signal. By concentrating the signal power in a very nar- row frequency band, the signal can be made to stand out among the natural, broad-band sources of noise. By the same token, signals leaked from a civilization's internal communications may be narrow band, but would be sig- nificantly weaker than a direct attempt at extraterrestrial contact. Because of this, radio SETI efforts have concentrated on detecting narrow band signals. When searching for narrow band signals it is best to use a narrow search window (or channel) around a given frequency. The wider the channel, the more broadband noise is included in addition to any signal. This broadband noise limits the sensitivity of the system. Early systems used analog technology to create narrow bandpass filters that could observe at a single frequency channel. More recent sys- tems use massive banks of dedicated fast Fourier trans- form (FFT) processors to separate incoming signals into up to a billion channels, each of width ∼1 Hz. There are, however, limitations to this technique. One limitation is that extraterrestrial signals are unlikely to be stable in frequency due to accelerations of the trans- mitter and receiver. For example, a receiver listening for signals at 1.4 GHz located on the surface of the earth un- dergoes acceleration of up to 3.4 cm/s2 due to the earth's rotation. This corresponds to a Doppler drift rate of 0.16 Hz/s. If uncorrected, an alien transmission would drift out of a 1 Hz channel in about 6 seconds, effectively lim- iting the maximum integration time to 6 seconds. Be- cause of the inverse relationship between maximum fre- quency resolution and integration time (∆ν = 1 ∆t ) there is an effective limit to the frequency resolution that can be obtained without correcting the received signal for this effect. There are other parameters of the signal that are un- known, for example: At what frequency will it be trans- mitted? What is the bandwidth of the signal? Will the signal be pulsed, if so at what period? Fully investigat- ing a wide range of these parameters requires innova- tive instruments and enormous computing power. An- other possibility that, until now, has not been included in SETI searches is that rather than directing large amounts of power into a narrow frequency band, an extraterres- trial intelligence might direct large amounts of power into a narrow time window by sending a short duration wide-band pulse. A wide band signal has the advan- tage of reducing the importance of the choice of obser- vation frequency, and the unavoidable interstellar dis- persion of a broad-band pulse reduces the confusion be- tween terrestrial and extraterrestrial pulses. In an ion- ized medium, high frequencies propagate slightly faster than low frequencies at radio frequencies. Therefore RF pulses will be dispersed in frequency during their travel through the interstellar medium. The dispersion across a bandwidth ∆ν is given by δt = (8.3 µs ) ∆ν( MHz ) ν3( GHz ) DM (1) (cid:82) L where dispersion measure DM is defined as is usually quoted in units of cm−3 pc. 0 nedl and It is possible that a civilization intentionally creating a beacon for extraterrestrial astronomers would choose to create "pulses" which have a negative DM. Natu- ral dispersion always causes higher frequency compo- nents to arrive first. A signal in which the low frequen- cies arrive first would stand out as obviously artificial. As a check on this, as well as to establish our back- ground noise limit, we examine both positive and nega- tive dispersion cases. Unfortunately, sensitive detection of broad-band pulses at an unknown dispersion measure also requires enormous computing power. We are currently conducting two searches for these dispersed radio pulses. Astropulse at Arecibo Observa- tory and Fly's Eye at the Allen Telescope Array (ATA). The Allen Telescope Array is a joint project of the SETI Institute and the University of California, Berkeley [3]. 2.1. Pulses from Extraterrestrial Intelligence Astropulse and Fly's Eye consider the possibility that extraterrestrials communicate using high intensity and wideband but short timescale pulses. It turns out that both the conventional (narrowband) method of data transfer assumed by SETI@home and the wideband, short timescale method assumed by Astropulse and Fly's Eye require the same amount of energy to send a message. In each case, the energy required to send one bit of information is proportional to the energy (per area) re- quired to send the minimum detectable signal. And this 2 energy is the same for the two methods, so energy con- siderations cannot rule out short timescale pulses as a medium for extraterrestrials' communication. SETI@home's sensitivity in searching for narrow- band signals is given by: of mass M emits radiation like a blackbody with a tem- perature TBH given by: TBH = c3 8πkGgravM = 10−6 K (5) (cid:18) M(cid:12) (cid:19) M (2) This radiation emanates from the black hole event hori- zon and comes completely from the black hole's mass. For a non-accreting black hole, Stefan-Boltzmann yields a lifetime of: (cid:32) M (cid:33)3 τBH = 1010years 1012kg (6) For stellar mass black holes, this theory predicts life- times of order 1034 years, much too long to ever expect to observe. However, some cosmologies predict the cre- ation of numerous small (M ∼ 1012g) primordial black holes in the early universe, which according to theory could be evaporating now [5]. The specific mechanism by which the evaporating black hole produces a strong radio pulse has not been fully elucidated, but in short, it is thought that the pro- cess is similar to the EMP that accompanies super- nova explosions. In such a process, a highly conduc- tive plasma fireball expanding into an ambient magnetic field can exclude the field and create an electromagnetic pulse. For typical values of the interstellar magnetic field, this pulse would be peaked near 1GHz [1]. An observation of these pulses would not only pro- vide a significant confirmation of Hawking radiation, but would also give strong evidence of the existence of primordial black holes. 3. Fly's Eye: Searching for Bright Pulses with the ATA The Allen Telescope Array has several advantages over other telescopes worldwide for performing tran- sient searches, particularly when the search is for bright pulses. The ATA has 42 independently-steerable dishes, each 6m in diameter. The beam size for individual ATA dishes is considerably larger than that for most other telescopes, such as VLA, NRAO Green Bank, Parkes, Arecibo, Westerbork and Effelsberg. This means that the ATA can instantaneously observe a far larger portion of the sky than is possible with other telescopes. Conversely, when using the ATA dishes independently, the sensitivity of the ATA is far lower than that of other telescopes. The Fly's Eye instrument was purpose built to search for bright radio pulses of millisecond duration at the G(cid:112)NpolBtint ασTsys where α = 2 is a loss from 1-bitting the data twice, σ = 24 is the threshold above noise, Tsys = 28 K is the system temperature, G = 10 K Jy−1 is the gain, and Npol = 1 is the number of polarizations. SETI@home searches for narrowband signals with a bandwidth of 0.075 Hz, and its longest integration time is (0.075 Hz)−1 = 13.4 s. To compute the energy (per area) required to send one bit, multiply by Btint = 1 to get 134 · 10−26 J m−2. On the other hand, Astropulse and Fly's Eye are searching for short pulses, so the sensitivity of these ex- periments will be given in Jansky microseconds, by a slightly different formula: √ ασTsys G(cid:112)NpolB tint (3) (4) where the variables have the meanings as before, ex- cept that now tint is the timescale of the pulse and B the entire bandwidth. For Astropulse, tint is 0.4 µs and B = 2.5 MHz. Also, α = 1.4 because Astropulse 1-bits the data only once, Npol = 2, and σ = 21.5. The resulting sensitivity is 24 Jy µs. To compute the energy per area required to send one bit, multiply by B = 2.5·106 Hz to get 60·10−26 J m−2. For both narrow frequency and broadband pulse cases, the expression for the minimum detectable en- ergy has the form ασTsys √ G(cid:112)Npol Btint √ Btint = 1, so similar answers are expected. with Because these energies are comparable, we have a chance to detect ETI communications in this new regime. 2.2. Pulses from Evaporating Black Holes Another intriguing possible source of short duration pulses comes from a suggestion by Martin Rees in 1977 [8] that primordial black holes, evaporating via the Hawking Process, could emit a large electromag- netic signature. According to Hawking [4], a black hole 3 ATA. The instrument consists of 44 independent spec- trometers using 11 CASPER IBOBs. Each spectrom- eter processes a bandwidth of 210MHz, and produces a 128-channel power spectrum at a rate of 1600Hz (i.e. 1600 spectra are outputted by each spectrometer per second). Therefore each spectrum represents time domain data of length 1/1600Hz=0.000625s=0.625ms, and hence pulses as short as 0.625ms can be resolved1. We have to-date performed roughly 400 hours of ob- serving with the Fly's Eye. Figures 1 and 2 show the beam pattern and sky coverage using all 42 antennas. Figure 1: From [2]. The beam pattern of the 42 beams at ATA, with the diameters equal to the half-power width. This hexagonal pack- ing is pointing north. A south pointing results in poor interference properties, due to the highly populated areas south of the ATA. 3.1. System Architecture The overall architecture of the Fly's Eye system is shown in Figure 3. Each IBOB can digitize four ana- logue signals, and 11 IBOBs are provided so that 44 sig- nals can be processed. The ATA has 42 antennas, each with two polarization outputs. A selection2 of 44 of the available 84 signals is made, and these are connected to the 44 iADC inputs. The IBOBs are connected to a control computer and a storage computer via a standard Ethernet switch. 3.2. Fly's Eye Offline Processing The analysis required for the Fly's Eye experiment is, in principle, fairly simple -- we wish to search over a 1Pulses of duration <0.625ms can also be detected provided that they are sufficiently bright, but their length cannot be determined with a precision greater than the single spectrum length. 2The selection is made with consideration for the goal of maximis- ing field-of-view -- in practice we selected at least one polarization signal from every functioning antenna. 4 Figure 2: From [2]. The sky coverage of the ATA for an observing pe- riod of 24 hours. Both the coverages for southern and northern point- ings (corresponding to the respective contiguous regions) are shown. Figure 3: Fly's Eye System Architecture. A selection of 44 analogue signals from 42 dual polarization antennas are connected to 44 inde- pendent spectrometers implemented in 11 IBOBs. wide range of dispersion measures to find large individ- ual pulses. Specifically our processing requires that all the data be dedispersed with dispersion measures rang- ing from 50 cm−3 pc to 2000 cm−3 pc. At each disper- sion measure the data needs to be searched for 'bright' pulses. The processing chain is in practice significantly more complicated than this description suggests. Processing is performed on compute clusters, with input data formatted, divided and assigned to worker nodes for processing. In the worker node flow, the data is equalized, RFI rejection is performed, and finally a pulse search is performed through the range of dispersion measures. The results are written to a database where they can be subsequently queried. The key feature of the results is a table that lists, in order of decreasing significance, the pulses that were found and the dispersion measures they were located at. (cid:80)N−1 i=0 P(cid:48) Average power equalization is performed on the fre- quency spectrum equalized values P(cid:48) i(t). We compute the average power over all frequency channels for a single integration (time sample t). The power average is defined as P(cid:48)(t) = 1 i(t). N is the number N of channels (for Fly's Eye this is always 128). The motivation for why it is possible to normalize the power is that we expect pulses to be dispersed over many time samples, so this procedure should not remove extraterrestrial pulses. Our strategy for mitigating constant narrowband RFI is simply to identify the channels that are affected, and to exclude them from further processing. This channel rejection is typically performed manually by looking at a set of spectra and identifying obviously infected channels, which are then automatically excluded in subsequent processing runs. (cid:18) = 1 T0 (cid:16) T0 Intermittent RFI is often quite difficult to automat- ically distinguish from genuine astronomical pulses, and we followed a conservative approach to try to ensure that we do not accidentally excise dispersed pulses. Our statistic for intermittent RFI is the variance of a single channel over a 10 minute data chunk, P(cid:48)(cid:48) σ2 i (t) i fitting determines a σ2 i outside which it is likely that channel i contains time-varying RFI. Future reprocess- ing will likely use a more robust method, such as that based on a kurtosis estimator [7]. (cid:17)(cid:17)2. Curve- (cid:17)2(cid:19) −(cid:16) 1 (cid:80)T0−1 t=0 (cid:80)T0−1 t=0 P(cid:48)(cid:48) i (t) (cid:16) Our final RFI mitigation technique is manual -- in 5 Figure 4: Fly's Eye Rack at the ATA. Two 6U CompactPCI crates (top and bottom) house the 11 IBOBs. The switch, data recorder computer and storage server are all visible. our results it is easy to see high-σ hits that are a result of RFI: these hits appear as simultaneous detections at many dispersion measures. 3.3. Detection of Giant Pulses from the Crab Nebula A suitable test of transient detection capability is to observe the Crab pulsar and attempt to detect giant pulses from it. Figure 5 shows a diagnostic plot generated from a one-hour Crab observation. The data is an incoherently summed set from the 35 best inputs. The diagnostic plot was generated after the raw data had been dedis- persed using a range of dispersion measures from 5 to 200. The Crab pulsar has dispersion measure ≈ 57 cm−3 pc, so three giant pulses from the Crab pulsar can be easily identified in the lower plot. The giant pulses appear only at the expected dispersion measure, whereas wideband RFI appears across a wide range of dispersion measures. Figure 5: Diagnostics on data taken from the Crab pulsar in a 60- minute observation conducted on 22 December 2007. The data was dedispersed using dispersion measures ranging from 0 to 200 cm−3 pc. Top-left: single-pulse SNR histogram. Top-centre: noise appears at all DMs, but bright pulses (SNR> 7) from the Crab correctly appear at DM ≈ 57 cm−3 pc. Top-right: inset of Figure 6. Bottom: pulse de- tections plotted on the DM versus time plane. Higher SNR detections appear as larger circles. Three giant pulses from the Crab are clearly visible in this plot. A frequency vs. time plot of the raw (summed) data at the time when the brightest giant pulses was detected is shown in Figure 6. The pulse is clearly visible in the data, and a fit to the dispersion measure shows that the pulse is, nearly without doubt, from the Crab (as opposed to RFI). 6 Figure 6: A giant pulse from the pulsar in the Crab nebula. This pulse was detected in a 60-minute dataset taken on 22 December 2007. The dispersion of the pulse, correctly corresponding to DM = 56.78 cm−3 pc, is clearly visible. 4. Astropulse 4.1. Algorithm Astropulse uses a unique approach to the problem of searching for dispersed pulses. This problem is well known to pulsar astronomers, and many solutions have been devised over the years. The traditional method of searching for pulses of unknown dispersion measure (DM) is through incoherent de-dispersion, which is ba- sically a filter bank - for each channel, the power is mea- sured and integrated, then appropriate delays are added to compensate for the dispersion. This can be made very efficient through the use of a binary tree type algorithm. We use a coherent de-dispersion algorithm, which pre- serves the phase of the signal, thus increasing sensitiv- ity. It also allows us to have much better time resolu- tion (down to the band limit at 0.4 µs ) than incoherent methods, whose theoretical maximum time resolution is determined by the individual channel bandwidths. (In practice, the time resolution for pulsar searches is typi- cally 100 µs or greater.) The problem with coherent de- dispersion is that it is very computationally intensive. We are able to afford this by implementing the pulse search in a distributed computing environment. The coherent de-dispersion process is basically a con- volution. The time domain data needs to be convolved with an appropriate chirp function in order to remove the dispersion smearing effect. fch(t) = exp(2πiν(t)t) = exp(2πi t δt ∆νt) Here, δt is the amount of time stretching caused by dispersion over a bandwidth of ∆ν, which can be deter- mined from equation 1. chirp. (b) Inverse FFT to go back to the time domain. (c) Threshold the (de-dispersed) time data, recording strong pulses. (d) Co-add adjacent bins, and threshold to look The most efficient way to do this is through the use of FFT convolution. Here is a brief description of the Astropulse detection algorithm: 1. Take 13.4 s time-domain data, perform FFT on 32k-sample chunks. Overlap chunks by 50% to re- cover pulses which span a chunk boundary. Save results on disk for later use. 2. For each DM, (a) Multiply frequency domain data by correct for broad pulses. (e) Search repeating pulses by using a folding or harmonic search algorithm. 3. Repeat for next DM. 4. When DM range is covered, start at step 1 for next 13.4 s of data. The optimal step size for DM is to step by the amount of dispersion which would cause 1 extra time sample length of stretching. Using the above ex- pression for δt = 0.4 µs , this gives us a DM res- olution of 0.055 pc cm−3 . We want to cover DMs from 55 pc cm−3 to 830 pc cm−3 , which corresponds to 14,000 DM steps at this resolution. (DMs smaller than about 55 are inaccessible to us, or to any search that employs one-bit sampling.) The maximum DM we wish to search determines the length of the FFTs. We want to look at DM up to 830 pc cm−3 , which will cover most of the galaxy. Using the same formula, this is a time stretch of δt = 5600 µs , or about 14,000 samples. Our FFT size should be at least twice this big, so we will use 32k point FFTs. A FFT takes about 5N log N floating point operations (FLOPs) to compute, where N is the number of points in the FFT. We are doing N = 215 point FFTs, and we need to do one inverse transform for each DM, plus one original forward transform. In addition, there are an- other N multiplies per DM for the chirp function, and a factor of 2 for the 50% overlap. This makes our total number of operations for a 32k sample chunk of data: Nops = 2(NDM + 1)N log N + 2NDMN = 1.8 × 1010 To calculate how much computation we would need to analyze the data in real time, divide the above num- ber by 0.013 s (length of 32k samples). We need an- other factor of 14 from multiple beams (7) and multiple 7 polarizations (2). Inefficiencies in memory allocation processes may contribute another factor of 5. In all, 100 TeraFLOP/s or more may be required. For comparison, an average desktop PC can compute at a speed of about 1 GigaFLOP/s. This analysis does not take into account the amount of computation required for folding. Astropulse folds on two different time scales, spending enough time on these to triple the overall computation requirement. This computation would take far too long on one computer, or even on a modest sized Beowulf cluster. Our solution to the problem is to distribute the data publicly to volunteers who download a program which will analyze it on their home computers. This is set up as a screen saver, which will turn on and start com- puting when their computers would otherwise be idle. [6] This approach has been remarkably successful for SETI@home, which has an average computation rate of about 40 TeraFLOP/s. We are able to implement this though the BOINC software platform, also developed by our group at Berkeley. The package, the Berkeley Open Infras- tructure for Network Computing (BOINC), is a gen- eral purpose distributed computing framework. It takes care of all the "management" aspects inherent to a dis- tributed computing project - keeping records of user ac- counts, distribution of data and collection of results over the network, error checking via redundant processing, sending out updates of the science code, etc. Many lessons learned over the course of running the origi- nal SETI@home project went into BOINC, so that new groups wishing to start computing projects don't have to "re-learn" these. An interesting aspect of running a public distributed computing project is that it has education and pub- lic outreach automatically built into it. People invest their computer time into the project, and as a result become interested in learning more about the science. SETI@home has been enormously successful in this re- gard - the website averages 1.25 million hits per day, and the project has spawned many internet discussion groups and independent websites. It is also used as part of the science curriculum by thousands of K-12 teachers nationwide. References [1] Blandford, R. D. 1977, "Spectrum of a radio pulse from an ex- ploding black hole," MNRAS, 181, 489 [2] G. Bower, J. Cordes, G. Foster, P. McMahon, A. Siemion, J. van Leeuwen, M. Wagner and D. Werthimer, "A Fly's Eye Search for Extragalactic Transients," ATA User Proposal (2008). [3] D. Deboer et al. 2004 Experimental Astronomy 17:1934 Springer 2005 [4] Hawking, S. 1974, "Black hole explosions?" Nature, 248, 30. [5] Hawking, S. 1971 "Gravitationally collapsed objects of very low mass," MNRAS, 152, 75. [6] Korpela, E., Werthimer, D., Anderson, D., Cobb, J., & Lebofsky, M. 2001, "SETI@home - Massively Distributed Computing for SETI," Computing in Science and Engineering, 3, 79. [7] G. Nita, D. Gary, Z. Liu, G. Hurford and S. White. Radio Fre- quency Interference Excision Using Spectral-Domain Statistics. Publ. Astro. Soc. Pacific, 119, 857, 805 -- 827 (2007). [8] Rees, M.J. 1977, "A better way of searching for black-hole ex- plosions?" Nature, 266, 333. 8
0802.2659
1
0802
2008-02-19T14:42:43
The diffuse X-ray emission from the Galactic center with Simbol-X
[ "astro-ph" ]
Similarly to the larger Galactic ridge, the Galactic center region presents a hard diffuse emission whose origin has been strongly debated for the past two decades: does this emission result from the contribution of numerous, yet unresolved, discrete point sources ? Or does it originate in a truly diffuse, hot plasma ? The Galactic center region (GC) is however different on many respects from the outer parts of the Galaxy, which makes the diffuse emission issue at the Galactic center unique. Although recent observations seem to favour a point sources origin in the far Galactic ridge, the situation is still unclear at the GC and new observations are required. Here we present results on the modeling of the truly diffuse plasma. Interestingly, such a plasma would strongly affect the dynamics of orbiting molecular clouds and thus the central engine activity. Discriminating between the two hypothesis has thus become a crucial issue in the understanding of this central region that makes the link between the inner small accretion disk and the large scale Galactic dynamics. We investigate the new inputs we can expect from Simbol-X on this matter.
astro-ph
astro-ph
Mem. S.A.It. Vol. , 1 c(cid:13) SAIt 2004 Memorie della The di use X-ray emission from the Galactic center with Simbol-X R. Belmont1 and M. Tagger2 1 Centre d'Etude Spatiale des Rayonnements, 9 rue du Colonel Roche, BP44346, 31028 Toulouse Cedex 4, France, e-mail: [email protected] 2 CEA Service d'Astrophysique, UMR "AstroParticules et Cosmologie", Orme des Merisiers, 91191 Gif-sur-Yvette, France Abstract. Similarly to the larger Galactic ridge, the Galactic center region presents a hard diffuse emission whose origin has been strongly debated for the past two decades: does this emission result from the contribution of numerous, yet unresolved, discrete point sources ? Or does it originate in a truly diffuse, hot plasma ? The Galactic center region (GC) is however different on many respects from the outer parts of the Galaxy, which makes the diffuse emission issue at the Galactic center unique. Although recent observations seem to favour a point sources origin in the far Galactic ridge, the situation is still unclear at the GC and new observations are required. Here we present new results on the modeling of the truly diffuse plasma. Interestingly, such a plasma would strongly affect the dynamics of orbiting molecular clouds and thus the central engine activity. Discriminating between the two hypothesis has thus become a crucial issue in the understanding of this central region that makes the link between the inner small accretion disk and the large scale Galactic dynamics. We investigate the new inputs we can expect from Simbol-X on this matter. Key words. Galaxy: center - X-rays: general - Plasmas - Magnetohydrodynamics (MHD) 1. Introduction For two decades, X-ray observations with Einstein, HEAO, Ginga, ASCA, and now XMM-Newton and Chandra have reported a diffuse emission from the Galactic ridge. This emission extends out to more than 4 kpc in radius (25o with a 8 kpc sun-Galactic center distance) from the central super massive black hole Sgr A*, but in the 1-10 keV band, it is very peaked in the first 2o (250 pc) at the Galactic center. The typical spectrum from this central region results from the contribution from sev- eral phases (1; 2). 1- Many K-α and K-β lines of very ionized metals (Si, S, Ar, and Ca) are observed in the 1-4 keV band. Those lines have been at- tributed to the soft (kBT ∼ 0.8 keV) plasma of young supernova remnants. 2- A fluorescence iron line at 6.4 keV, re- sulting from the interaction of cosmic rays with cold gas (kbT ∼50 K), is also present. 3- The diffuse emission is mainly character- ized by two iron lines at 6.7 and 6.9 keV and by the underlying continuum. These lines correspond to very ionized states of Fe and therefore cannot originate in the soft phase. Two main ideas have been sug- 8 0 0 2 b e F 9 1 ] h p - o r t s a [ 1 v 9 5 6 2 . 2 0 8 0 : v i X r a 2 Belmont: Diffuse X-ray emission from the GC gested to account for these lines. a) They originate in the hot plasma of many unre- solved discrete point sources (mostly CVs). b) They originate in a hot (kBT ∼ 7 keV), truly diffuse plasma that bathes the emit- ting region. 4- Above 8 keV, the spectrum is line free. The continuum cannot originate in a thermal 7 keV plasma and rather corresponds to a non thermal emission. Again, its not clear whether these processes take place in dis- crete points source or in a diffuse medium. Here we address the issue of the diffuse emission at the GC. In the first section, we re- mind briefly the key points that are common to the Galactic Ridge emission (see the pro- ceeding by S Mereghetti for more details). In the second section, we describe more precisely the central region and the truly diffuse plasma issue. In the last section, we detail the inputs Simbol-X might bring on these questions. 2. Origin of the Galactic Ridge and Galactic center diffuse emissions Since the first space-based X-ray observations in the 80s, more and more sources have been resolved by the increasing power of observato- ries. A cosmic contribution from extra-galactic sources was first identified. Galactic sources were also detected and it is now estimated that 85% of the X-ray emission above 20 keV is re- solved in the Galactic ridge (3). The emission at lower energy is however unclear. The typical spectrum is very similar to that of sources such as Cataclysmic Variables, but so far only a frac- tion of the total X-ray emission in the 1-10 keV band has been resolved. To account for un- detected sources, Log(N)-Log(S) diagrams are built and extrapolated to weak sources. This work is subject to possibly large biases and un- certainties. Source confusion can also limit the detection capabilities of the instruments, espe- cially at high energy. Recent deep observations combined with new estimates of the unresolved population seem to show that the fraction from discrete point sources could be large in the Galactic ridge (4). However, observations with Suzaku have shown that the remaining diffuse emission has not the same profile as the point sources in the first fractions of degrees at the Galactic center (2). The point sources may thus con- tribute differently in the Galactic ridge and at the Galactic center. The idea of a diffuse plasma was early sug- gested since the fits with such a thermal plasma match best the spectrum, but it was also very debated since it raises severe paradoxes. The temperature of this plasma is so high that is should not be confined by the gravitational po- tential, so that the power required to heat it be- fore it leaves the Galactic plane exceeds any known energy source. Also, even if the plasma was confined, no heating mechanism had been identified. However, as we show in the next section, recent advances in the modeling of a truly hot plasma have solved its paradoxes at the Galactic center. 3. Possibility of a truly diffuse plasma at the Galactic center As can be observed at all wavelengths, the GC region is a very particular region. 1- The X- ray emission is much stronger than in the rest of the Galactic ridge. 2- Infrared observations show that the GC corresponds to a strong con- centration of cold molecular gas. At a radius of about 150 pc, the mean density jumps by a fac- tor of 20 from outside to inside. All this cold molecular content is condensed in clouds and forms the so-called Central Molecular Zone (5; 6) that well matches the central X-ray emit- ting region. 3- At radio wavelengths, obser- vations also show numerous non thermal fil- aments that are observed nowhere else in the Galaxy (7). These filaments are beautifully aligned with the direction perpendicular to the galactic plane and suggest a strong, vertical magnetic field (5). All these unique features probably result from the same global galactic dynamics and may contribute to make the situ- ation at the GC very different from that farther out in the Galaxy. And the problems of a dif- fuse plasma at the GC, (confinement and heat- ing) can be solved by a consistent interaction of these different phases. Belmont: Diffuse X-ray emission from the GC 3 3.1. Plasma confinement The idea that a diffuse 7 keV plasma should escape from the Galactic plane directly results from the (mono-) fluid assumption: the global sound speed of the plasma is larger than the escape velocity required for bodies to leave the gravitational wheel. represent a large reservoir of kinetic and grav- itational energy that can be dissipated by the viscosity of the diffuse plasma. Because of its high temperature, the dif- fuse plasma is highly viscous (ν ∝ T 5/2, (9)). However, the strong magnetic field inhibits the usual shear viscosity, only leaving the other component: the so called bulk viscosity related to the compression of the fluid. As the clouds motion is rather slow (subsonic), the associated compression is weak, which limits the dissipa- tion efficiency. The overall dissipation depends on the exact wake of the clouds. The wake of conducting bodies in a magne- tized plasma has been studied in space science to investigate the motion of satellites in space magnetospheres. In first approximation, it is dominated by Alfv´en perturbations that prop- agate along the field lines. These Alfv´en waves forms a wing structure and carry a strong en- ergy flux away from the cloud (10; 11). This flux depends on the magnetic field strength and for any value in the expected range (10µG- 1mG), the cumulative flux associated to the motion of all the clouds in the central region is much larger that the X-ray luminosity. Only a fraction of this power is required to heat the diffuse plasma. To first order, Alfv´en wave can- not be dissipated by the bulk viscosity since they are not compressible. However, it has been shown that non linear effects or a signif- icant curvature of the field line (Rc ∼ 100 pc) provide sufficient dissipation to account for the hot plasma temperature (12). 4. Observations with Simbol-X It has long been though that such diffuse plasma could simply not exist in the Galactic plane, so that the only observational chal- lenge was to make deeper observations and improve the knowledge on the weak sources population in order to find enough of them. Observational inputs of Simbol-X for the de- tection of weak point sources are discussed in a different talk (see S. Mereghetti's proceed- ing). We emphasize that the identification ca- pabilities of Simbol-X at high energy (kBT > 10 keV) will be even more crucial at the Galactic center at where the source confusion However, the plasma is composed by many species (electrons, protons, He ions and other heavy ions) and, on the typical escape time (τesc ∼ 5 × 104 yr), the plasma is not colli- sional (τcoll ≈ 105 yr). This basically means that the different species can have different be- haviors and must be studied separately. Such an analysis is very similar to that in plane- tary atmospheres (except that the gas is a ion- ized plasma) and the results are comparable. Namely, by comparing the effective thermal velocity of an ion and its electrons (vth = , where µ is a mean molecular weight) with the escape velocity (vesc ≈1200 km/s), it is found that: -- The thermal velocity of protons (vth ≈1300 km/s) is larger than the escape velocity. At this hot temperature, protons are too light to be confined by gravity. This is basically the same result as the fluid one. (cid:113) kBT µmp -- The thermal velocities of heavier ions (vth < 750 km/s) are smaller than the es- cape velocity. Even with this high temper- ature, any ion other than a proton is heavy enough to be confined by gravity As a consequence, the Galactic center can undergo a selective evaporation that natu- rally leads to the formation of a heavy, He- dominated plasma, is confined by the Galactic potential in this region (8). that 3.2. Plasma heating If there is no other cooling mechanism, such a bound plasma only cools by radiation, which is much more reasonable since the total X- ray luminosity of the central region is only L ≈ 4 × 1037erg/s. Such a power can actually be balanced by viscous heating. The numerous molecular clouds that flow in the central region 4 Belmont: Diffuse X-ray emission from the GC is high. On the other hand, the recent results presented here show that a diffuse plasma at kBT ∼ 7 keV can survive at the Galactic center and that its hot temperature can be explained by an efficient viscous dissipation. Thus, they give new observational diagnostics that must be investigated as well. At 7 keV, hydrogen and helium are fully ionized, so that there is no direct check for the dominant species. Re-interpretation of the X- ray data show that assuming a helium dom- inated plasma leads to densities and abun- dances about three times lower than when as- suming a hydrogen dominated plasma. The most recent observations with Suzaku give iron abundances about 3.5 solar ones assuming a H-dominated plasma, which gives back solar abundances when assuming a He-dominated plasma. Also, if there is no strong mixing mecha- nism, such a gravitationnaly confined plasma should be stratified, the heavier elements con- centrating at low latitude whereas light ele- ments extend higher above the Galactic plane. If observed, such a stratification would be a strong evidence for a diffuse hot plasma. So far, only the iron line has been observed precisely. The comparison of the scale height of its emis- sion with that of the emission in the He con- tinuum could reveal the stratification. However the spectral region where the iron lines and the continuum are observed is very confused. Contributions from the iron lines (i.e. from the hot phase), from the line complex (i.e. from the soft phase), from the fluorescence line (i.e. from the cold phase) and from the hard possi- bly non-thermal tail that extends above 10 keV, as well as their respective continuum all add in the same narrow spectral range. As a result, it is difficult to measure with a good accuracy the relative strength of these different components, in particular the He continuum. Simbol-X will provide the best spectra of the Galactic center region above 10 keV. This is of first interest for three reasons: 1. The spectrum above 8 keV is less confused than at lower energy. The direct estimate of the He continuum will thus be easier. 2. The continuum however remains confused by the flat power law tail extending to higher energy: the thermal continuum of a 7 keV plasma is supposed to drop quickly above 10 keV but observations show that the emission evolves from a thermal to a non-thermal one around 10 keV. Previous high resolution X-ray observatories were not able to get precise spectra above 8- 10 keV from the central region and could not get a reliable estimate of the non ther- mal contribution. Such observations are however critical to measure the cut-off of the thermal emission and it strength in or- der to characterize the stratification. 3. Given its low background level, Simbol-X will probably be able to resolve the He-like nickel line at 7.8 keV, first observed last year by Suzaku (2). This new line corre- sponds to the emission of a third species constituting the hot plasma. Its detection, compared to the He continuum and the iron line emission may provide more constrains on the stratification. Several pointings at different latitudes will thus allow for the first time to investigate the strati- fication of a plasma confined by gravity. References Muno, M. P., et al. 2004, ApJ, 613, 326 Koyama, K., et al. 2007, PASJ, 59, 245 Lebrun, F., et al., 2004, Nature, 428, 293 Revnivtsev, M., A.Vikhlinin, Sazonov, 2006, astro-ph/0611952 Morris, M., & Serabyn, E. 1996, ARA&A, 34, 645 Oka et al. 2001, ApJ, 562, 348 LaRosa, T. N., Lazio, T. J. W., Kassim, N. E., & Lang, C. C. 2000, BAAS, 197, 406 Belmont, R., Tagger, M., Muno, M., Morris, M., & Cowley, S. 2005, ApJ, 631, L53 Spitzer, L. 1962, Physics of Fully Ionized Gases, New York: Interscience, 1962 Drell, S. D., Foley, H. M., & Ruderman, M. A. 1965, JGR, 70, 3131 Neubauer, F. M. 1980, JGR, 85, 1171 Belmont, R., & Tagger, M. 2006, A&A, 452, 15
0706.3667
2
0706
2007-09-05T14:04:14
Affine equation of state from quintessence and k-essence fields
[ "astro-ph", "gr-qc", "hep-th" ]
We explore the possibility that a scalar field with appropriate Lagrangian can mimic a perfect fluid with an affine barotropic equation of state. The latter can be thought of as a generic cosmological dark component evolving as an effective cosmological constant plus a generalized dark matter. As such, it can be used as a simple, phenomenological model for either dark energy or unified dark matter. Furthermore, it can approximate (up to first order in the energy density) any barotropic dark fluid with arbitrary equation of state. We find that two kinds of Lagrangian for the scalar field can reproduce the desired behaviour: a quintessence-like with a hyperbolic potential, or a purely kinetic k-essence one. We discuss the behaviour of these two classes of models from the point of view of the cosmological background, and we give some hints on their possible clustering properties.
astro-ph
astro-ph
Affine equation of state from quintessence and k-essence fields Claudia Quercellini,1 Marco Bruni,1, 2, 3 and Amedeo Balbi1, 2 1Dipartimento di Fisica, Universit`a di Roma "Tor Vergata", via della Ricerca Scientifica 1, 00133 Roma, Italy 2INFN Sezione di Roma "Tor Vergata", via della Ricerca Scientifica 1, 00133 Roma, Italy 3Institute of Cosmology and Gravitation, University of Portsmouth, Mercantile House, Portsmouth PO1 2EG, Britain We explore the possibility that a scalar field with appropriate Lagrangian can mimic a perfect fluid with an affine barotropic equation of state. The latter can be thought of as a generic cosmological dark component evolving as an effective cosmological constant plus a generalized dark matter. As such, it can be used as a simple, phenomenological model for either dark energy or unified dark matter. Furthermore, it can approximate (up to first order in the energy density) any barotropic dark fluid with arbitrary equation of state. We find that two kinds of Lagrangian for the scalar field can reproduce the desired behaviour: a quintessence-like with a hyperbolic potential, or a purely kinetic k-essence one. We discuss the behaviour of these two classes of models from the point of view of the cosmological background, and we give some hints on their possible clustering properties. PACS numbers: 98.80.-k; 98.80.Jk; 95.35.+d; 95.36.+x I. INTRODUCTION Known forms of matter such as baryons and radiation are not sufficient to explain the observed universe, hence we need to assume the existence of an unknown dark component, whose properties are then deduced from in- direct detections. About one third of it, named dark matter, is needed to account for the inhomogeneities that we observe up to very large scales, namely to ex- plain both large scale structure and the cosmic microwave background anisotropy peaks [1, 2]. The remaining two thirds, dubbed dark energy, are needed to explain the observed flatness of the universe and its late time accel- eration [3, 4]. Assuming the existence of heavy particles, non collisional and cold, dark matter is usually modelled as a pressureless perfect fluid. In its simplest form dark energy takes the form of vacuum energy density, i.e. a cosmological constant Λ. More generally, it can be mod- eled as a perfect fluid with an equation of state (EoS from now on) that can violate the strong energy condi- tion (SEC, see e.g. [5]), such that it can dominate at late times and have sufficiently negative pressure to account for the observed accelerated expansion. For both forms of energy several tentatives have been done in different frameworks to make them descend from a scalar field, re- lated to a Lagrangian (see [6] for a recent review and ref- erences therein). Since we do not know much about these dark components, the idea that they may be ascribed to a unique source of energy is appealing, especially if their origin can be easily related to some fundamental the- ory, as it is for a scalar field. A considerable effort at describing the dark side of the universe with a unified phenomenological model [7, 8] has been made in the last few years, also aiming at constraining the parameters of the models (e.g. see [9, 10]). In this context scalar field models have been proposed, both with a canonical ki- netic term in the Lagrangian [11, 12] and in k-essence scenarios [13, 14, 15, 16, 17]. In this paper we develop a different approach: we in- vestigate whether a minimally coupled scalar field with a specific Lagrangian can mimic the dynamics of the dark fluid discussed in [9, 18, 19, 20, 21] and similarly in [22, 23]. More precisely, in [9] the working hypothesis was the existence of a single dark perfect fluid with a sim- ple 2-parameter barotropic equation of state, the affine EoS PX = p0 + αρX , which naturally yields an energy density evolution of an effective cosmological constant plus a generalized dark matter [21]. This EoS can also be seen as a first order Taylor expansion of a wider class of functions of the energy density, where the dark compo- nent more generally can either only provide the late time acceleration or mimic also a matter-like background evo- lution. In addition, this EoS can be derived from the simple assumption that the speed of sound is constant and potentially positive, alleviating the unpleasant con- sequences of a negative value for structure formation [24]. In [9] we tested this model against observables related to the homogeneous and isotropic expansion and compared it to the standard ΛCDM model. Here we will investigate if and to what extent such a EoS, in general represent- ing either a dark energy or a unified dark matter com- ponent, can arise from scalar field dynamics, considering two possibilities: a quintessence model, with a hyperbolic potential, and a purely kinetic k-essence model. II. AFFINE EQUATION OF STATE In an homogeneous isotropic universe, modeled with a Robertson-Walker metric, the energy momentum tensor must take the perfect fluid form, with a total energy den- sity ρT = Pi ρi and pressure PT = Pi Pi. Under stan- dard assumptions, if the various components are non- interacting, each satisfies the usual conservation equa- tions, ∇µT µν (i) = 0, independently of the theory of gravity. To fix ideas, let us assume that X, one of the above com- ponents, is represented by a barotropic fluid with EoS PX = PX (ρX ). Let's assume that this EoS allows for vi- olation of SEC at least below some redshift, so that X can become the dominant component and drive the observed cosmic acceleration. Assuming now and in the following Einstein equations with no cosmological constant Λ, vi- olation of SEC is equivalent to ρX + 3PX ≤ 0, which in a Friedmann-Robertson-Walker universe is sufficient for acceleration. Then, denoting with H the Hubble expan- sion scalar related to the scale factor a by H = a/a, it follows from the energy conservation equation ρX = −3H(ρX + PX ) (1) that, if there exists an energy density value ρX = ρΛ such that PX (ρΛ) = −ρΛ, then ρΛ has the dynamical role of an effective cosmological constant: ρΛ = 0 (see [21] for a more detailed discussion). Another reasonable assumption is that the square c2 of the speed of sound of X is non-negative: X := dPX /dρX ≥ 0. Indeed, this assumption ensures that the adiabatic perturbations of X do not blow up (be- cause it follows from momentum conservation that, for a barotropic fluid and for dP/dρ < 0, pressure gradients do not work anymore as a restoring force against gravity, and instead act as gravity). Actually, the assumption c2 X ≥ 0 is strong enough to imply violation of SEC and a sort of cosmic no-hair theorem, if PX < 0 at some point (in an expanding phase, H > 0, with ρX + PX > 0). That is, under these conditions it follows that the en- ergy density decreases while the pressure becomes nega- tive enough for SEC to be violated, thereby driving an accelerated expansion in a Friedmann-Robertson-Walker universe, till ρ → ρΛ, so that the universe becomes de Sitter at late times, i.e. ρΛ is an attractor for (1). When ρΛ becomes the dominant component, the same holds true in Bianchi models, as proved by Wald [25], as well as in other cases, see e.g. [26]. In [9] we have considered a flat Friedmann-Robertson- Walker universe in general relativity, with radiation, baryons and a single unified dark matter component with energy density ρX represented by a barotropic fluid. Given that the EoS PX = PX (ρX ) is unknown, we as- sumed a constant speed of sound dPX /dρX ≃ α, leading to the 2-parameter affine form [21] PX ≃ p0 + αρX . (2) This allows for violation of SEC even with c2 s = α ≥ 0. Then, using (2) in (1) and asking for ρΛ = 0 leads to the effective cosmological constant ρΛ = −p0/(1+α). Eq. (2) may also be regarded (after regrouping of terms) as the Taylor expansion, up to O(2), of any EoS PX = PX (ρX ) about the present energy density value ρXo [18]. The EoS (2), if taken as an approximation, could be used to parametrize a dark component (either unified dark matter or dark energy) at low and intermediate redshift. Actually, with α → 0 (and ρm > 0, see Eq. (3) below) the EoS above is equivalent to a ΛCDM. This allows for 2 a straightforward comparison of models, as done in [9]. Indeed, in [9] we made a more radical assumption, that is, we extrapolated the validity of Eq. (2) to any time, thereby building a cosmological model based on a unified dark component with EoS (2), which we tested against observations and compared with the standard ΛCDM model. We found that ΩΛ ≃ 0.7 and α ≃ 0.01. The evolution of ρ with the expansion can be found us- ing the EoS (2) in the conservation equation (1), leading to ρX (a) = ρΛ + ρma−3(1+α), (3) where today ρm = ρXo − ρΛ and a = 1. Formally, with the EoS (2) one can then interpret our dark component as made up of the effective cosmological constant ρΛ and an evolving part with present "density" ρm. It is clear that the standard ΛCDM model is recovered for α = 0, if ρm > 0 is identified with the density of pressureless dark matter. More in general, a priori no restriction on the values of α and po is required, but one needs po < 0 and α > −1 in order to satisfy the conditions that ρΛ > 0 to have that ρΛ is an and ρ → ρΛ in the future, i.e. attractor for Eq. (2). In this case, our affine dark matter is phantom if ρm < 0, but without a "big rip", cf. [21]. In summary, the barotropic affine EoS (2) can be used to model either a dark energy component, if standard dark matter is also assumed to be present, or a unified dark matter, as we did in [9]. It is well known that, from a given EoS describing a fluid, a scalar field model can be derived (see e.g. [27]). However, the correspondence is in general not one to one: scalar fields in general have an extra degree of freedom and thus a more complicated dynamics. The question then arises, whether this dynamics allows for a solution which acts somehow as an attractor, so that it can mimic, at least to a certain extent and with little or no fine tun- ing, the fluid evolution. In the following sections we are going to investigate to what extent a scalar field with ap- propriate Lagrangian can mimic the cosmological dynam- ics arising from the affine EoS (2), i.e. the density evo- lution (3). We shall assume a flat Friedmann-Robertson- Walker universe with units c = 1 and 8πG = 1, so that the Hubble parameter H 2 = a/a and the energy density ρ obey the Friedmann equation H 2 = ρ/3. III. SCALAR FIELDS In the most general form, in general relativity, the ac- tion for a minimally coupled scalar field can be written as S =Z d4x√−g(cid:16) R 2 + L(χ, φ)(cid:17), (4) where χ = − 1 energy tensor consequently is 2 gµν∂µφ∂ν φ is the kinetic term. The stress T φ µν = ∂L(φ, χ) ∂χ ∂µφ∂ν φ + L(φ, χ)gµν (5) and, as long as the scalar field 4-gradient is time-like, i.e. gµν∂µφ∂ν φ < 0, it can be cast in the perfect fluid µν = (ρφ + pφ)uµuν + pφgµν, where uµ = ∂µφ form T φ √2χ is the 4-velocity (coinciding with the unit normal to the φ = constant slices) and the Lagrangian plays the role of the pressure of the fluid. For an observer comoving with the fluid, the energy density reads ρφ(φ, χ) = 2χ ∂pφ(φ, χ) ∂χ − pφ(φ, χ), (6) while the effective EoS and the speed of sound [28] are wφ = 2χ ∂pφ(φ,χ) pφ(φ, χ) ∂χ − pφ(φ, χ) ; c2 φ = ∂pφ(φ, χ)/∂χ ∂ρφ(φ, χ)/∂χ . (7) Thus a scalar field admits a perfect fluid description, in general not of barotropic type. Indeed, comparing with a barotropic fluid, a scalar field has an extra de- gree of freedom: therefore, in general each initial condi- tion (equivalently, each trajectory in phase space) for a given scalar field Lagrangian corresponds to a different barotropic EoS. Hence, the question is to what extent a given EoS can be mimicked by a given scalar field model, at least in a asymptotic regime, i.e. to what extent that model admits a solution that acts as attractor for other trajectories in phase space, thereby avoiding a strong fine tuning problem that would spoil the value of the model itself. In the next subsections we will consider separately two scalar field models: we will first analyse a quintessence model and derive a potential that returns the affine EoS as an exact solution in phase space, then we will recon- struct the pressure in a purely k-essence model where the potential is set to be constant. A. Quintessence In the quintessence model the Lagrangian is just the difference between a canonical kinetic term and a poten- tial, L = χ − V (φ) = λ 1 2 φ2 − V (φ) (8) at the zero order, and the energy density and pressure take the form pφ = L ρφ = λ 1 2 φ2 + V (φ), (9) where λ = 1 for the standard quintessence scenario and λ = −1 for phantom cases (not yet excluded by current observations [29]). The Klein Gordon equation for the scalar field is φ+3H φ+V ′(φ) = 0, which will be analysed later in this section. Combining Eq. (9) with Eq. (2) we can express the derivative of the field and the potential energy as functions of the scale factor: λ φ2 = ρm(1 + α)a−3(1+α) V (φ) = ρΛ + ρm(1 − α) 2 a−3(1+α). 3 (10) (11) The dot here refers to the derivative with respect to cosmic time t. Phantom evolution for the energy den- sity, namely solutions for which the energy density is a growing function of the scale factor, can be set by either α < −1 or ρm < 0. We now want to work out the shape of the potential in- tegrating and inverting Eq. (10), eventually substituting it in Eq. (11). Let us assume that the fluid is the only dominating component (i.e. H 2 = ρφ/3) and rewrite Eq. (10) using dφ dt = aH dφ da : dφ da =s 3(1 + α) λa2 1 a3(1+α)) (1 + ρΛ ρm , (12) where we choose the positive sign for the square root of H 2, connected to a universe that is expanding at least at present time, although one can also have a change of sign in this function (see point iii) below). Also, the outcome of our calculations will be independent of the sign of dφ da , thanks to the simmetry of the potential about its minimum (see below). Integrating out Eq. (12) we find r 3(1 + α) λ φ = −2 log(cid:16)a− 3(1+α) 2 +r ρΛ ρm + a−3(1+α)(cid:17), (13) which inverted and substituted in Eq. (11) returns the expression for the potential: (ρme−√3λ(1+α)φ (14) 4 V (φ) = h 3 + α + (cid:16) ρ2 Λ ρΛ + (1 − α) 8 ρm(cid:17)e√3λ(1+α)φ)i. Again here λ accounts for α < −1 values. We can distinguish three different cases: i) ρΛ > 0 and ρm > 0 The minimum of clearly V (φmin) = ρΛ and the field rolls down to it, reaching the final de Sitter attractor (see Fig. 1). the potential here is 1 ∗ = (ρm/ρΛ) ii) ρΛ > 0 and ρm < 0, phantom The universe evolves from a contracting phase, bounce 3(1+α) and then re-expands [21]. The at a potential here exhibits a maximum and the field is then forced to climb up the hill towards the maximum (see Fig. 2) reaching the ensuing De Sitter attractor. The scalar field thus exhibits phantom-like behaviour with the energy density growing with time. iii) ρΛ < 0 and ρm > 0, non phantom The case ρΛ < 0 necessarly gives ρm > 0, if we assume a positive total energy density of the fluid. The universe 3(1+α) and then expands to a maximum amax = (ρm/ρΛ) 1 WL=0.7 5 4 3 2 1 V Φ 2 3 H0 Α= 0.6 Α= 0.3 Α= 0 Α= -0.3 Α= -0.6 -4 -2 0 2 Φ FIG. 1: Scalar field potential with ρΛ, ρm > 0, non phantom case. WL=1 1 0 -1 -2 -3 -4 V Φ 2 3 H0 Α= 0.6 Α= 0.3 Α= 0 Α= -0.3 Α= -0.6 -6 -4 -2 0 Φ FIG. 2: phantom case. Scalar field potential with ρΛ > 0 and ρm < 0, recollapses. At a = amax the field has reached the min- imum of the potential, which in this case is not ρΛ, but is V (φmin) = ρΛ(1 + α)/2 (see Fig. 3 for negative value of ΩΛ). Α=0.1 4 3 2 1 0 V Φ 2 3 H0 WL = 0.8 WL = 0.7 WL = 0.6 WL = -0.2 -4 -2 0 Φ 2 4 6 FIG. 3: Scalar field potential for different values of ρΛ, non phantom case. The minimum of the potential is ρΛ(1 + α)/2 for ρΛ < 0. The potential (14) has been derived imposing an affine equation of state for a barotropic fluid. However, as said above, the scalar field has an extra degree of freedom, 4 hence the universe expansion in general will not be the same that in the barotropic fluid case. In order to study the dynamics of the scalar field we need to look at the Klein Gordon equation and its phase space. In Ref. [21] it has been shown that current background observables exclude phantom behaviour at more than 3σ. Moreover, as we will see, the de Sitter critical point in the phase space is a pure attractor only in non phantom cases. For these reasons till the end of this section we will put λ = 1, α > −1 and ρm > 0. Rescaling the scalar field so as to have the minimum at ϕ = φ − φmin = 0, the potential (14) can be written as V (ϕ) = ρΛh 3 + α cosh (ϕp3(1 + α))i. (15) A related form was derived in [20]. Here the case ρΛ = 0 is not included, since we put e−√3(1+α)φmin = ρΛ/ρm, but it is recovered in the limit φmin → ∞ (correspond- ing to the pure exponential potential, where α takes the standard role of the linear EoS parameter: Px = αρx). Thus, assuming as a starting point the affine EoS we fall into the class of quintessence "exponential potentials", whose properties are well known ([11, 30, 31, 32]) and which we will comment more in details further on. (1 − α) + 4 4 In order to analyse the behaviour of the field from a dynamical system point of view, let us define the new dimensionless variables X := ϕ, Y := dϕ dη , η := √ρmt. (16) Then the Klein-Gordon and Friedmann equations are equivalent to the system X′ = Y, Y ′ = − Y − 1 ρm dV dX , (17) (18) √3(cid:16) Y 2 2 + V ρm(cid:17)1/2 where the prime represents the derivative with respect to η. System (17)-(18) obviously exhibits a fixed point (0, 0), corresponding to the field lying at rest at the min- imum of the potential and driving the expansion with an effective EoS parameter (7) wφ = −1, i.e. an effec- tive cosmological constant. This point therefore repre- sents a de Sitter model. The eigenvalues of the lineariza- tion of system (17)-(18) at this critical point are e1 = −p3ρΛ/(4ρm)(1 − α) and e2 = −p3ρΛ/(4ρm)(1 + α), both always negative for α < 1 and ρΛ, ρm > 0. Within these bounds, which we always assume, this fixed point is therefore an attractor: a stable node in general, and an improper node in the degenerate case e1 = e2 = −p3ρΛ/4, i.e. for α = 0. Therefore, the scalar field forcely approaches asymptotically the de Sitter attractor (0, 0) in the phase space, which corresponds to the total domination of the constant part of the potential and the late time accelerated expansion (see Fig. 4). It is easy to derive an analytic expression in terms of X and Y for the trajectories corresponding to the affine EoS (2): they lie on the curve 5 Yaf f = ±s ρΛ ρm (1 + α) 2 (cosh (Xaf fp3(1 + α)) − 1), (19) which we may call the affine curve in phase space. This is the union of three exact solutions of system (17)-(18): 1) the fixed point (0, 0); 2) a positive branch for X < 0; 3) a negative branch for X > 0. 2.0 1.6 1.2 0.8 0.4 Y 0.0 −0.4 −0.8 −1.2 −1.6 −2.0 −2 −1 0 X 1 2 FIG. 4: Phase space for system (17)-(18) with α = 0.8 and ΩΛ = 0.7. The fixed point at the origin is a stable node for α 6= 0 and represents a de Sitter asymptotic state. The thicker line represents the curve (19), i.e. the solutions corresponding exactly to the affine EoS fluid. It is rather clear that generic trajectories approach the fixed point along the eigenvector corresponding to e2 and not along the thick curve, see text. An analysis of the linearization of system (17)-(18) shows that, the eigenvector E2 corresponding to e2 is tangent at the origin to the affine curve (19). For α > 0 e2 < e1 < 0, thus generic trajectories approach the de Sitter stable node along the other eigenvector E1 (they are tangent to it), as it is rather clear from the represen- tative example in Fig. 4. For the improper node case α = 0 there is a single eigenvector, and all trajecto- ries approach the stable node from that direction. In other words, for α > 0 the scalar field dynamics mimics the affine EoS fluid rather poorly, unless α is very small or zero. In practice, for the potentially interesting case α ≪ 1 [9, 33, 34], the angle between the two eigenvectors becomes smaller and smaller, and the scalar field evolu- tion is better and better represented by that of the affine fluid. The situation changes for for α < 0: in this case e1 < e2 < 0 and generic trajectories approach the de Sitter stable node along the eigenvector E2 (tangent to it), which remains tangent to the affine curve (19). In this case therefore, the affine curve (19) attracts generic 2.0 1.6 1.2 0.8 0.4 Y 0.0 −0.4 −0.8 −1.2 −1.6 −2.0 −2 −1 0 X 1 2 FIG. 5: Phase space for system (17)-(18) with α = 0 and ΩΛ = 0.7. Here generic trajectories approach the fixed point along the single existing eigenvector e1 = e2, see text. 2.0 1.6 1.2 0.8 0.4 Y 0.0 −0.4 −0.8 −1.2 −1.6 −2.0 −2 −1 0 X 1 2 FIG. 6: Phase space for system (17)-(18) with α = −0.8 and ΩΛ = 0.7. For negative values of α the qualitative behavior is rather clear: generic trajectories approach the fixed point along the eigenvector corresponding to e2, driven towards the affine curve. trajectories, better and better for 1 + α → 0+ (see Fig. 5). As analysed in Ref. [11] the scalar field crosses two distinct phases: first of all it acts as in a standard pure exponential quintessence model, in which phase, if coex- isting together with other components (e.g. radiation and CDM), it can track the EoS of the background and, sec- ondly, it slips into the minimum following a parabolic-like potential. However, differently from Ref. [11], here the minimum is non-zero. Thus, when entering the parabolic phase an effective cosmological constant drives the accel- erated expansion. It is often asserted that, since the speed of sound for a scalar field is equal to the speed of light, then the scalar field cannot form sub-horizon structures. One should ex- ercise extreme caution to relate the definition of the speed of sound to the evolution of density perturbations like in the hydrodynamic models. Indeed it is true that in the perturbation equation of a scalar field the term which is proportional to the wave number k2 and define the phase velocity is exactly the unity, and that this term is the well known sound of speed contribution to a standard wave equation, see Eq. (7). On the other hand, the second derivative of the potential with respect to the field plays an important role in the clustering. Let us write down the relativistic equation for the scalar perturbations: δϕk + 3H δϕk +(cid:16) k2 a2 + V ′′(ϕ)(cid:17)δϕk = − ϕ h, 1 2 (20) where the prime here indicates the derivative with respect to the field, h is the trace of the spatial metric tensor perturbation in the synchronous gauge and k = 2π/λ is the wave number. At early time the potential (15) is undistinguishable from a simple exponential potential and the energy den- sity of the scalar field tracks the background radiation evolution [31]. However, soon after the equivalence of matter and radiation, one would like the scalar field to mimic a matter component. If the scalar field then leaves the exponential regime and enters the parabolic phase, its fast oscillations around the minimum could generate an average energy density which acts as a matter-like en- ergy density, see e.g. [35]. The second derivative of the potential today is the square of the mass associated to the fluctuations (m2 ϕ) and scalar perturbations can only grow if the term proportional to k2 in (20) is subdomi- nant with respect to a2V ′′, i.e. for k < amϕ. This mass defines the value of the Compton length: , (21) λc = 2π √V ′′ = 4π p3ρΛ(1 − α2) In the most optimistic view we can set the stage of os- cillations to begin just before the equivalence, i.e. at z ≃ 105. In this case, if one thinks about a scalar field related to ultra-light particles capable to condensate and set an energy scale of the potential V0 ≃ 106M pc−2 [11, 35], a Compton length which is of order of hun- dreds parsec is obtained. Besides, under the assump- tion of a unified dark matter suitable for driving also the late-time acceleration, this description is no longer valid. As already mentioned, after the early exponen- tial phase, the potential enters a parabolic phase, where its shape is well represented by the quadratic expression 6 V (ϕ) ≃ ρΛ/4((3 + α) + (1 − α)(1 + 3(1 + α)ϕ2)). The mass associated to the fluctuations stabilizes when the frequency of the oscillations is bigger than the Hubble expansion, namely when mϕ ≥ H, otherwise it varies un- til it asymptotically mimics the parabolic oscillations set by this potential, where V ′′, calculated in the minimum, is the value entering Eq. (21). Indeed the energy scale related to the potential (15) is V0 = ρΛ ≃ 10−9M pc−2, for which we find a Compton length much larger than the horizon, even for tiny values of α. Unavoidably, scalar perturbations at scales smaller than the horizon must have been erased, with unpleasant consequences for structure formation. Hence one is forced either to trig- ger two different energy scales in the potential, no longer describing an affine EoS, or to use the scalar field only as a dark energy component. Infact, when a scalar field is invoked merely as a dark energy component, a naturally vanishing mass is recovered. Moreover the homogeneity of the dark energy is a desirable property, as we do not observe its clustering at sub-horizon scales. In the next subsection we will present a purely kinetic Lagrangian which can avoid this problem. B. K-essence The possibility of k-essence was first taken into con- sideration to explain the dynamics of inflation [36], but has been subsequently proposed to generate the late time acceleration [37]. As already mentioned in the introduc- tion, k-essence is particularly suitable for describing uni- fied fluids with a matter-like component and a subluminal sound of speed. We now discuss a purely kinetic k-essence model equivalent to an affine EoS. Let us assume that the Lagrangian is a generic function of χ only, i.e. L = P (χ), (22) which hides a simple constant term playing the role of a potential [13]. More general types of Lagrangian in- cluding, as a factor in (22), a function of the field can also be motivated by low energy effective string the- ory (see Ref. [6]). A general expression for the La- grangian of a purely kinetic k-essence model obtained from a barotropic EoS was derived in [38]. In a Friedman-Robertson-Walker metric the equation of motion for the field is (cid:16) dP dχ + 2χ d2P dχ2(cid:17)χ′ + 6χ dP dχ = 0, (23) where ′ = d/d log (a). From this equation a straight in- dipendence of the solution on other possible components is easily noticeable. Moreover the EoS and the speed of sound do not seem to be influenced by the choice of the constant potential, but they actually are, in that the po- tential affects the evolution of χ and, in turn, through the Friedman equation the other components. Assuming a specific function P (χ) one can in princi- ple solve Eq. (23) for χ(a) and then substitute it in the ∗ ∗ expression (6) for ρφ(χ) to figure out the evolution of the energy density with the scale factor. In [17] the sta- ble nodes of Eq. (23) have been analysed, corresponding to solutions for which either ∂P/∂χχ∗ = 0 or χ = 0 (both with wφ = −1). Thus, modelling P by means of a parabolic form, or equivalently expanding the pressure around its minimum in χ (as previously done in[16]), (a/a1)−3, where g0 and g2 they found ρφ = −g0 + 4g2χ2 ∗ are constants. This corresponds to an energy density of a unified dark fluid composed by a cosmological constant and a pure CDM. However this solution happens to be an approximation only valid in the neighborhood of the minimum, and for a ≫ a1, where a1 depends on the con- stants. Moreover the speed of sound is not vanishing at all out of this regime. Looking for an explicit solution with a χ2 Lagrangian leads to undesirable radiation-like evolution in the past. Here our approach moves along an inverse path: we start imposing the evolution for the energy density corresponding to the affine EoS (2) and derive the expression for P (χ) and the speed of sound. Connecting Eq.(2) to Eq. (6), a first order differential equation in P as a function of χ is derived: P − 2αχ dP dχ + ρΛ = 0. (24) This equation can be solved for the pressure and then used in Eq. (6) to derive the following: P = −ρΛ + cχ 1+α 2α ; ρφ = ρΛ + 1+α 2α , χ c α (25) 1+α 0 2α where c = ρmα/χ is the integration constant derived imposing the value of the fluid energy density at present and χ0 is χ at present time. These two function are clearly solutions of the affine EoS (2), and the stable node of Eq. (23) is the one for which χ0 = 0 and wφ = −1. The expressions for the pressure and the energy density (25) represent the sum of a power-law solution, related to an unchanging and non-zero EoS (p = αρ), plus a constant potential term that acts as a cosmological constant [39]. It has been shown in [40] that if α ≥ 0 the constant EoS is an attractor for the k-essence field, holding also for superluminal EoS. From Eq. (7) it follows that the "effective" speed of sound in this model is c2 s = In many k-essence model, whenever ∂2P/∂χ2 6= 0 α. and positive, the speed of sound can be arbitrarily small between last scattering and today. As a purely kinetic k- essence model, one main property is that the scalar field possesses a single degree of freedom: therefore its EoS is naturally barotropic and reads: wφ = −ρΛa3(1+α) + ρmα ρΛa3(1+α) + ρm . (26) Hence the scalar field dynamics derived from a purely kinetic Lagrangian, related to the simplest first order parameterization of a perfect fluid EoS, is completely equivalent to the perfect fluid description: while −1 ≤ wφ ≤ α, the speed of sound is constant and 7 can reproduce a "fuzzy" dark matter [41], i.e. a low sound-speed fluid inhibiting growth of matter inhomo- geneities at very small scales. The density and pressure perturbations of the constant part of the Lagrangian are null independently of the speed of sound, and the pressure perturbations are defined in the rest frame of the matter-like component. Moreover, in this case, the scalar field does not have the problem of an early radiation-like behavior. In addition, evolution (3) is not a transient solution, but is valid at all times, implying no unnatural fine-tuning on the parameters (apart from the energy scale). Besides, the effective speed of sound defines the stabilization scale for a perturbation [28]; in this model perturbations of physical wavelength a/k < csH−1 0 must have been completely erased. Thus, the Compton length related to the growth of inhomo- geneities is λc = αH−1 and, in order to obtain formed structures at least at Kpc scale, we expect α < very strongly constrained, although non-vanishing. This is also the bound obtained in [33] for a generalized dark matter from a direct comparison with the matter power spectrum. A larger value of α would indeed distort also the large scale CMB anisotropy power spectrum via the integrated Sachs Wolfe effect [34]. ∼ 10−6, i.e. 0 IV. CONCLUSIONS We have investigated the prospects of reproducing the behaviour of a dark fluid with affine barotropic EoS within the context of a scalar field model with an appro- priate Lagrangian. The importance of this approach is twofold: first, this kind of EoS can be regarded as a first order approximation to that of any generic barotropic fluid; second, describing this kind of fluid in a scalar field framework gives a more fundamental representation of the dark component. Our main finding is that it is in- deed possible to obtain an equivalent description of the dark fluid with affine EoS in two cases: a quintessence- like Lagrangian, with a hyperbolic potential, and a purely kinetic Lagrangian, or k-essence. Both models are able to reproduce the correct behaviour of the cosmological back- ground from early epochs up to the present, although in the quintessence case this is strictly true only for α ≤ 0. In the case of the quintessence potential we have found (see also [20]), the dynamics of the scalar is driven by an "exponential-like" potential including a constant term. The field then first follow a trajectory where it tracks the dominant cosmological component (radiation or matter), then it reaches an attractor solution where it behaves as a cosmological constant, resulting in an accelerated ex- pansion of the universe. However, when one addressed the problem from the point of view of structure forma- tion, this kind of Lagrangian does not appear to have the necessary characteristics to act as a unified dark compo- nent, although it could be used as a dark energy. In fact, since for quintessence scalar field the speed of sound is such that c2 s = 1 and the energy scale related to the min- imum is set by the cosmological constant, perturbations are erased on scales smaller than a Compton length po- tentially bigger than the horizon. The only way to use the scalar field as a unified dark fluid would be to intro- duce a new energy scale in the potential. In this regard, the dark fluid obtained from a quintessence potential has a worse behaviour than the barotropic fluid that it tries to reproduce, since in the latter case the speed of sound can be made arbitrary small. The k-essence Lagrangian seems more suitable to treat a unified dark fluid, even when considerations based on structure formation are taken into account. We found that the energy density scaling behavior of an affine per- fect fluid is an exact solution of a specific pure kinetic Lagrangian. Furthermore, the absence of an early phase mimicking the radiation component makes the model suitable to reproduce a matter-like behaviour at all times. Moreover, in this case the speed of sound is related to one of the parameters of the affine EoS, c2 s = α. This suggests 8 that the model might have the right clustering properties both at galactic and cosmological scale (although requir- ing a fine-tuning of the α parameter). Based on simple arguments related to the expected Compton length, we conclude that the α parameters should be strongly con- strained to small (∼ 10−6) but not necessarily vanish- ing values. This is consistent with values obtained by a comparison of a similar model (namely, a generalized dark matter plus a cosmological constant) with the mat- ter power spectrum[33]. A full analysis in linear pertur- bation theory and a comparison of the predictions with CMB anisotropy data is currently being undertaken and will be the subject of a forthcoming paper [34]. A. Acknowledgements We would like to thank Luca Amendola and Davide Pietrobon for useful discussions about the topics. [1] D. N. Spergel, R. Bean, O. Dor´e, M. R. Nolta, C. L. Ben- nett, J. Dunkley, G. Hinshaw, N. Jarosik, E. Komatsu, L. Page, et al., ArXiv Astrophysics e-prints (2006), astro- ph/0603449. [17] D. Bertacca, S. Matarrese, and M. Pietroni, ArXiv As- trophysics e-prints (2007), astro-ph/0703259. [18] M. Visser, Classical and Quantum Gravity 21, 2603 (2004), arXiv:gr-qc/0309109. [2] V. Springel, C. S. Frenk, and S. D. M. White, Nature [19] T. Chiba, N. Sugiyama, and T. Nakamura, MNRAS 289, (London) 440, 1137 (2006), arXiv:astro-ph/0604561. L5 (1997), arXiv:astro-ph/9704199. [3] S. Perlmutter, G. Aldering, G. Goldhaber, R. A. Knop, P. Nugent, P. G. Castro, S. Deustua, S. Fabbro, A. Goo- bar, D. E. Groom, et al., Astrophys. J. 517, 565 (1999), arXiv:astro-ph/9812133. [4] A. G. Riess, L.-G. Strolger, S. Casertano, H. C. Fergu- son, B. Mobasher, B. Gold, P. J. Challis, A. V. Filip- penko, S. Jha, W. Li, et al., Astrophys. J. 659, 98 (2007), arXiv:astro-ph/0611572. [5] M. Visser, Science 276, 88 (1997). [6] E. J. Copeland, M. Sami, and S. Tsujikawa, ArXiv High [20] V. Gorini, A. Y. Kamenshchik, U. Moschella, and V. Pasquier, Phys. Rev. D69, 123512 (2004), hep- th/0311111. [21] K. N. Ananda and M. Bruni, Phys. Rev. D 74, 023523 (2006), arXiv:astro-ph/0512224. [22] E. Babichev, V. Dokuchaev, and Y. Eroshenko, Classi- cal and Quantum Gravity 22, 143 (2005), arXiv:astro- ph/0407190. [23] R. Holman and S. Naidu, ArXiv Astrophysics e-prints (2004), astro-ph/0408102. Energy Physics - Theory e-prints (2006). [24] W. Hu, Astrophys. J. 506, 485 (1998), arXiv:astro- [7] A. Kamenshchik, U. Moschella, and V. Pasquier, Physics ph/9801234. Letters B 511, 265 (2001), arXiv:gr-qc/0103004. [8] M. C. Bento, O. Bertolami, and A. A. Sen, Phys. Rev. D 70, 083519 (2004), arXiv:astro-ph/0407239. [9] A. Balbi, M. Bruni, and C. Quercellini, ArXiv Astro- physics e-prints (2007), astro-ph/0702423. [10] L. Amendola, F. Finelli, C. Burigana, and D. Carturan, Journal of Cosmology and Astro-Particle Physics 7, 5 (2003), arXiv:astro-ph/0304325. [25] R. W. Wald, Phys. Rev. D28, 2118 (1983). [26] J. Wainwright and G. F. R. Ellis, Dynamical Sys- tems in Cosmology (Dynamical Systems in Cosmol- ogy, Edited by J. Wainwright and G. F. R. Ellis, pp. 357. ISBN 0521673526. Cambridge, UK: Cambridge University Press, June 2005., 2005). [27] G. F. R. Ellis and M. S. Madsen, Classical and Quantum Gravity 8, 667 (1991). [11] V. Sahni and L. Wang, Phys. Rev. D 62, 103517 (2000), [28] J. Garriga and V. F. Mukhanov, Physics Letters B 458, arXiv:astro-ph/9910097. [12] R. Mainini, L. P. L. Colombo, and S. A. Bonometto, Astrophys. J. 632, 691 (2005). [13] D. Giannakis and W. Hu, Phys. Rev. D 72, 063502 (2005). 219 (1999), arXiv:hep-th/9904176. [29] U. Seljak, A. Slosar, and P. McDonald, Journal of Cosmology and Astro-Particle Physics 10, 14 (2006), arXiv:astro-ph/0604335. [30] E. J. Copeland, A. R. Liddle, and D. Wands, Phys. Rev. [14] L. M. G. Be¸ca and P. P. Avelino, MNRAS 376, 1169 D 57, 4686 (1998), arXiv:gr-qc/9711068. (2007). [31] P. G. Ferreira and M. Joyce, Phys. Rev. D 58, 023503 [15] L. P. Chimento, Phys. Rev. D 69, 123517 (2004), (1998), arXiv:astro-ph/9711102. arXiv:astro-ph/0311613. [32] A. R. Liddle and R. J. Scherrer, Phys. Rev. D 59, 023509 [16] R. J. Scherrer, Physical Review Letters 93, 011301 (1999), arXiv:astro-ph/9809272. (2004), arXiv:astro-ph/0402316. [33] C. M. Muller, Phys. Rev. D71, 047302 (2005), astro- ph/0410621. [34] D.Pietrobon et al., in preparation (2007). [35] T. Matos and L. Arturo Urena-L´opez, Phys. Rev. D 63, nal of Modern Physics D 14, 1561 (2005), arXiv:gr- qc/0501101. [39] A. D´ıez-Tejedor and A. Feinstein, Phys. Rev. D 74, 063506 (2001), arXiv:astro-ph/0006024. 023530 (2006), arXiv:gr-qc/0604031. 9 [36] C. Armend´ariz-Pic´on, T. Damour, and V. Mukhanov, arXiv:hep- (1999), Physics Letters B 458, th/9904075. 209 [40] L. P. Chimento and A. Feinstein, Modern Physics Letters A 19, 761 (2004), arXiv:astro-ph/0305007. [41] W. Hu, R. Barkana, and A. Gruzinov, Physical Review [37] T. Chiba, T. Okabe, and M. Yamaguchi, Phys. Rev. D Letters 85, 1158 (2000), arXiv:astro-ph/0003365. 62, 023511 (2000), arXiv:astro-ph/9912463. [38] A. Diez-Tejedor and A. Feinstein, International Jour-
astro-ph/0410636
2
0410
2005-06-28T17:05:35
A Faraday Rotation Template for the Galactic Sky
[ "astro-ph" ]
Using a set of compilations of measurements for extragalactic radio sources we construct all-sky maps of the Faraday Rotation produced by the Galactic magnetic field. In order to generate the maps we treat the radio source positions as a kind of "mask" and construct combinations of spherical harmonic modes that are orthogonal on the masked sky. As long as relatively small multipoles are used the resulting maps are quite stable to changes in selection criteria for the sources, and show clearly the structure of the local Galactic magnetic field. We also suggest the use of these maps as templates for CMB foreground analysis, illustrating the idea with a cross-correlation analysis between the Wilkinson Microwave Anisotropy Probe (WMAP) data and our maps. We find a significant cross-correlation, indicating the presence of significant residual contamination.
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 000 -- 000 (0000) Printed 2 July 2018 (MN LATEX style file v1.4) A Faraday Rotation Template for the Galactic Sky Patrick Dineen⋆ & Peter Coles School of Physics & Astronomy, University of Nottingham, University Park, Nottingham, NG7 2RD, United Kingdom 2 July 2018 5 0 0 2 n u J 8 2 2 v 6 3 6 0 1 4 0 / h p - o r t s a : v i X r a ABSTRACT Using a set of compilations of measurements for extragalactic radio sources we con- struct all-sky maps of the Faraday Rotation produced by the Galactic magnetic field. In order to generate the maps we treat the radio source positions as a kind of "mask" and construct combinations of spherical harmonic modes that are orthogonal on the masked sky. As long as relatively small multipoles are used the resulting maps are quite stable to changes in selection criteria for the sources, and show clearly the structure of the local Galactic magnetic field. We also suggest the use of these maps as templates for CMB foreground analysis, illustrating the idea with a cross-correlation analysis between the Wilkinson Microwave Anisotropy Probe (WMAP) data and our maps. We find a significant cross-correlation, indicating the presence of significant residual contamination. Key words: magnetic fields -- methods: data analysis -- Galaxy: structure -- cosmic microwave background 1 INTRODUCTION The origin of large-scale magnetic fields observed on galac- tic and cluster scales is unknown. The magnetic fields, with observed strengths of ∼ 10−6G, could be the consequence of an amplification of a tiny seed (<∼ 10−20G) by a dynamo mechanism. Alternatively, the compression of a primordial seed (∼ 10−9G) by protogalactic collapse could lead to the fields we see today. Both scenarios require an initial pri- mordial field. Furthermore, the two mechanisms need to ex- plain the high redshift magnetic fields observed in galaxies (Kronberg, Perry & Zukowski 1992) and damped Lyman-α clouds (Wolfe, Lanzetta & Oren 1992). Magnetic fields play a crucial role in star formation as well as possibly playing an active role in galaxy formation as a whole (Wasserman 1978; Widrow 2002). Thus, one of the most significant tasks in cosmology is unravelling the mystery behind magnetic fields; from the primordial field to our own Galactic field. The cosmic microwave background (CMB) provides us with the most distant and extensive probe of the early uni- verse. A primordial magnetic field will leave an imprint in this radiation. Various methods have been developed that seek these signatures. Barrow, Ferreira and Silk (1997) use the anisotropic expansion caused by the presence of a ho- mogeneous primordial field to place limits on its size from large-angle CMB measurements. Others have sought correla- tion between different scales in the temperature anisotropies (Chen et al. 2004; Naselsky et al. 2004) or computed the ⋆ E-mail: [email protected] c(cid:13) 0000 RAS effects the field has on the polarisation-temperature cross- correlation (Scannapieco & Ferreira 1997; Lewis 2004). The existence of a magnetic field at last-scattering also leads to a possible measurable Faraday rotation of the polarised CMB light (Kosowsky & Loeb 1996). At the other end of the scale, investigation of our own Galaxy's magnetic field has led to the development of a num- ber of different techniques; Han (2004) provides a concise review of the subject. Techniques include observing Zeeman splitting, polarised starlight, synchrotron radiation, Faraday rotated light and polarised dust emission. However, even with all these methods, there are still outstanding problems. This has led to a lack of consensus on key issues: from the number of spiral arms; how the arms are connected; to the direction the field takes along the arms (Vall´ee 1997; Han 2004). To fully understand magnetic fields, we need a coherent picture throughout different epochs. Theories and models can then be tested against this observational picture. A ro- tation measure (RM) map of the full sky has the potential to fulfil such a goal. RM values probe the integral of the magnetic field from the radiation source to the observer. Obviously, the information encoded in such a map depends on the location of the radiation that is rotated. Faraday ro- tated polarised CMB radiation will offer both a picture of the primordial field (at the surface of last scattering) and that of our Galaxy. The recent detection of CMB polarisa- tion by DASI (Kovac et al. 2002) and confirmation of this via the Wilkinson Microwave Anisotropy Probe (WMAP; Kogut et al. 2003), have opened up a new avenue in CMB 2 P. Dineen & P. Coles research. Future results from WMAP and Planck satellites will offer polarised data covering the full-sky over a range of frequencies (seven in the case of Planck). By forming a RM map with the data and looking at differing scales, we should be able to untangle the information in the data; on the largest scales the local magnetic field can be studied, whereas the primordial field can be studied on the smaller scales. On the other hand, one of the main tools for probing local magnetic fields (such as our Galaxy's) involves utilising RM values from extragalactic sources. Catalogues contain- ing RM values of extragalactic sources have been used to map the Galactic field (eg. Frick et al. 2001). Combining this data with rotation measures from pulsars that are located within our Galaxy, it is conceivable that a 3-dimensional image of the Galactic magnetic field can be built. However, current RM catalogues are both sparsely populated and un- evenly sampled. Thus, astute methods are required to pro- duce a RM map with the data at hand. So, what do we intend to do? We attempt to map the RM values as a function of angular position giving R(Ω), where we use R to denote the Faraday rotation measure and Ω for the angular position. Catalogues containing RM values of extragalactic sources are used to construct the function. The observed spatial distribution of the RM values can be expanded over a set of orthogonal basis functions. For anal- ysis of data distributed on the sky, expansion over spherical harmonics seems natural (de Oliveira-Costa et al. 2004), labelled foreground X, that is spatially correlated with 100µm dust emission. Spinning dust grains (Draine & Lazarian 1998) are the most popu- lar candidates for causing this anomalous emission. A RM map can provide insight into the role of these grains as they may align with the local magnetic field. Foreground removal will be particularly challenging in CMB polarisation stud- ies as foregrounds are more dominant than in the temper- ature anisotropies. Also, single frequency polarisation mea- surements will not be able to remove the effects of Faraday rotation through the Galactic magnetic field. Thus, the ex- tent to which the results have been effected by the E-mode signal rotating into the B-mode signal (and vice versa) is unknown. The layout of the paper is as follows. In the next sec- tion we describe the three rotation measure catalogues used in our analysis. In describing the data we clarify the mean- ing of extragalactic rotation measures. In Section 3 we il- lustrate a method to generate orthonormal basis functions for each catalogue. From the coefficients of the new basis, the spherical harmonic coefficients are calculated. In Sec- tion 4 we present the resulting RM maps and discuss the observed features. In Section 5 we give a brief application of the maps. Correlations are sought between the RM maps and cleaned CMB-only maps. The conclusions are presented and discussed in Section 6. al,mYl,m(Ω), (1) 2 ROTATION MEASURE CATALOGUES R(Ω) = ∞ Xl=1 m=+l Xm=−l where the al,m are the spherical harmonic coefficients and the Yl,m are the spherical harmonics. The properties of the spherical harmonics are well understood and the calcula- tion of the al,m will allow us to utilise routines within the HEALPix† package (G´orski, Hivon & Wandelt 1998) for vi- sualisation purposes and further analysis. However, a non- uniform distribution of data points compromises spherical harmonic analysis due to the loss of orthogonality (G´orski 1994). It is more fruitful to analyse a system using orthogo- nal functions: the statistical properties of the coefficients are simplified. If nonorthogonal functions are used, the proper- ties of the system and the basis are confused. Therefore, we would like to construct an orthonormal basis with func- tions closely related to those of the spherical harmonics. The spherical harmonic coefficients can then be obtained from the resultant coefficients of the orthonormal basis. Spherical harmonic analysis of extragalactic sources has been previously performed by Seymour (1966,1984). How- ever, the analysis was carried out using a different form of orthogonalisation and only on a set of 65 sources. The RM map resulting from our method will be a use- ful tool for probing Galactic magnetic structure. The map will also be a valuable point of reference when investigating mechanisms that involve the Galactic magnetic field. For ex- ample, CMB foregrounds (synchrotron, dust, free-free emis- sion) are correlated with rotation measures (Dineen & Coles 2004). There is evidence of another foreground component † http://www.eso.org/science/healpix/ Faraday rotation measures of extragalactic radio sources are direct tracers of the Galactic magnetic field. When plane- polarised radiation propagates through a plasma with a com- ponent of the magnetic field parallel to the direction of prop- agation, the plane of polarisation rotates through an angle φ given by φ = Rλ2, where the Faraday rotation measure is measured in rad m−2 where (2) R = e3 2πm2 ec4 Z neBk ds. (3) Note that Bk is the component of the magnetic field along the line-of-sight direction. The observed RM of extragalac- tic sources is a linear sum of three components: the intrinsic RM of the source (often small); the value due to the in- tergalactic medium (usually negligible); and the RM from the interstellar medium of our Galaxy (Broten et al. 1988). The latter component is usually assumed to form the main contribution to the integral. If this is true, studies of the distribution and strength of RM values can be used to map the Galactic magnetic field (Vall´ee & Kronberg 1975). Even if the intrinsic contribution were not small, it could be ig- nored if the magnetic fields in different radio sources were uncorrelated and therefore simply add noise to any measure of the Galactic field (Frick et al. 2001). In a similar vein, the distributions of RM values have been used to measure local distortions of the magnetic field, such as loops and filaments, and attempts have also been made to determine the strength of intracluster magnetic fields (Kim, Tribble & c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 Kronberg 1991). In what follows we shall use RM values ob- tained from three catalogues in an attempt to map R(Ω) over the whole sky. All three catalogues are sparsely populated and have non-uniform distributions. It is essential to remove the struc- ture due to the spatial distribution of the sources. This structure is unique to each catalogue. Therefore, for each catalogue a new set of orthonormal functions has to be gen- erated. The three catalogues used are those of Simard- Normandin et al. (1981; hereafter S81), Broten et al. (1988- updated in 1991; B88) and Frick et al. (2001; F01). S81 present an all-sky catalogue of rotation measures for 555 extragalactic radio sources (ie. galaxies and quasars). B88 and F01 contain 674 and 800 sources respectively. In F01, the two other catalogues are combined with smaller studies of specific regions in the sky (see paper for details). They also provide slightly reduced versions of the other two cat- alogues. Sources with significantly larger RM values than those in the other studies are removed, leaving catalogues of 551 sources for S81 and 663 for B88. In our analysis we will use these versions of the two catalogues. Finally, we reject sources with R > 300 rad m−2. Such large RM values are unlikely to represent real features of the Galactic magnetic field: probably they arise from incorrect determination of R due to the nπ ambiguity in polarisation angle; magnetic fields within the sources; Equation (2) being incorrect; and so on. This final selection criteria reduces the catalogues to 540 sources for S81, 644 for B88, and 744 for F01. 3 GENERATING A NEW BASIS We can only observe RM values where there happens to be a line of sight. This means we see the RM sky through a pe- culiar "mask". We wish to generate a new orthogonal basis that takes account of the spatial structure of this mask. In particular, we need to find a set of functions that are orthog- onal on the incomplete sphere. Ideally, these new functions should be related to the spherical harmonics (which are or- thogonal functions on a complete sphere). This will enable us to determine the spherical harmonic coefficients from the new functions and their coefficients. G´orski (1994) tackles the problem from the point of view of CMB analysis. The determination of the angular power spectrum is a crucial element of much work in the field. If the temperature anisotropies form a Gaussian ran- dom field then they can be completely characterised by the angular power spectrum. In order to obtain the angular power spectrum, one needs to estimate the spherical har- monic coefficients. At low Galactic latitudes (b < 20o) fore- ground contamination is severe. Therefore, it is preferable to obtain an estimate of the spherical harmonic coefficients out- side this region. G´orski (1994) calculates a new set of func- tions that are orthogonal to this cut sphere. These functions are used to calculate the spherical harmonic coefficients and thus estimate the angular power spectrum from the two-year COBE-DMR data (Bennett et al. 1994). In order to see how the method works it is prudent to look at the definition of orthogonal functions. Let us consider two complex functions A(x) and B(x). If c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 Faraday Rotation Template 3 A∗(x)B(x) dx = 0, (4) Z b a then A(x) and B(x) are orthogonal over on the interval {a , b}. If we incorporate these two functions into a vector v=[A(x),B(x)], the orthogonality of the functions can be expressed through the scalar product hv · vTi{a,b} = I. We shall now look at orthogonal functions in the con- text of a complete sphere. A function can be de- scribed by spherical harmonics Yl,m(Ω) up to an order lmax. We can form an (lmax + 1)2 -- dimensional vector y=[Y0,0(Ω), Y1,−1(Ω), Y1,0(Ω), Y1,1(Ω), . . . , Ylmax,lmax(Ω)]. The scalar product is then defined as hy · yTi{full sky} = I. The sky can therefore be fully described by (5) (6) (lmax+1)2 R(Ω) = aiYi(Ω) ≡ aT · y. (7) Xi=1 (8) However, when the sphere is incomplete due to a Galaxy cut or a more complex mask being applied, we have hy · yTi{cut sky} = W 6= I, where W is the coupling matrix. A new basis, where the equality is true, can be constructed in the following manner. The procedure is a type of Gram-Schmidt orthogonalisation. W can be Choleski-decomposed into a product of a lower triangular matrix L and its transpose W = L · LT . The inverse matrix Γ = L−1 is then computed. The new set of functions on the cut sky is (9) ψ = hΓ · yi{cut sky}. By construction, we have hψ · ψTi{cut sky} = Γ · y · yT·ΓT = Γ · L · LT·ΓT = L−1 · L · LT · (L−1)T = I. (10) (11) This can be a useful cross-check for testing the code. Finally, the new basis functions can be used to describe R (lmax+1)2 R(Ω) = ciΨi(Ω) ≡ cT · ψ. (12) Xi=1 At this point, only the angular positions Ωn of the sources have been required. In order to obtain the coeffi- cients ci of the new basis, the rotation measure Rn them- selves are required. It is worth putting it into the context of the data we have. Let the number of sources in our cat- alogue be N and let (lmax + 1)2 = M . It should be evident that we have a set of simultaneous equations which have to be solved in order to obtain the coefficients of the new basis functions R(Ω1) = c1ψ1(Ω1) + c2ψ2(Ω1) + . . . + cM ψM (Ω1) R(Ω2) = c1ψ1(Ω2) + c2ψ2(Ω2) + . . . + cM ψM (Ω2) 4 P. Dineen & P. Coles ... ... R(ΩN ) = c1ψ1(ΩN ) + c2ψ2(ΩN ) + . . . + cM ψM (ΩN ).(13) So, as long as N > M , these equations should be solvable. Ultimately, we wish to obtain the spherical harmonic coeffi- cients al,m. Using Equation (10), we see R = aT · y = cT · ψ = cT · Γ · y and therefore (14) aT = cT · Γ → a = ΓT · c. So, we have obtained the spherical harmonic coefficients. (15) There are some practical points that have been glossed over in the above description of the method. Firstly, the RM values in the catalogues need to be smoothed. Otherwise, as the series in Equation (12) is finite, we will be attempting to fit large-scale waves to small-scale features. Ideally, the smoothing will take place in the new basis, however, this is impractical. Therefore, we chose to smooth in harmonic space. Around each source, a hoop of 20o is thrown and the average RM value is taken of the sources within the hoop. This could have been done via a more sophisticated method, say a Gaussian-weighted mean of RM values. However, we chose to use the simple approach. The size of the hoop was chosen to match lmax (∼ 16) closely in angular size. Further- more, it coincided with the limiting resolution of the wavelet method used in Frick et al. (2001) on the same catalogues. Secondly, we need to determine to what order we take the series up to, i.e. the value of lmax. We do this through trial and error. As we will expand upon in the next section, RM maps were generated for lmax values of 8, 10, 15, 16, 17 and 18. The power spectrum for each map was studied. At some limiting value of lmax the shape at low l alters as features become unstable. Maxima and minima points explode since we are trying to fit more and more function to the same amount of data. That is to say, N is getting too similar to M . Finally, the convention for spherical harmonics has to be chosen carefully. The new basis functions were calculated using the convention of G´orski (1994) where Figure 1. Power spectra (measured in rad2 m−4). Red lmax=15, green lmax=16, blue lmax=17, dotted lmax=8 and 10, and dash- dot lmax=18. From top to bottom: S81, B88 and F01 catalogues. (cos θ)f (φ)(16) INTERPRETATION 4 ALL-SKY RM MAPS AND THEIR Yl,m(θ, φ) =r(cid:16) 2l + 1 2 (cid:17)s(cid:18) (l − m)! (l + m)!(cid:19)P m l and f (φ) = π−1/2 cos(mφ), (2π)−1/2, or π−1/2 sin(mφ) for m > 0, = 0, or < 0. The spherical harmonic coefficients can then be trivially converted into those that adhere to the HEALPix definition of the harmonics where now f (φ) = (2π)−1/2[cos(mφ) + i sin(mφ)] or (−1)m(2π)−1/2[cos(mφ) + i sin(mφ)] for m ≥ 0 or < 0. The convention of G´orski (1994) was chosen since the information within the coefficients is more highly compressed. Whereas in the HEALPix defini- tion, the coefficient are complex with a symmetry between +m and -m, those following the convention of G´orski (1994) are real and contain no such symmetry. The redundant in- formation in the HEALPix coefficients leads to confusion when solving Equation (13). For all three catalogues, sets of spherical harmonic coeffi- cients were calculated with lmax being set to 8, 10, 15, 16, 17 and 18. From these coefficients, RM maps were produced using the 'synfast' routine in the HEALPix package. To see whether a RM map was displaying real features or whether the series expansion had been extended too far, we looked at the angular power spectrum of each map. The angular power spectrum is the harmonic space equivalent of the au- tocovariance function in real space. It is defined as al,m2. (17) Cl = 1 2l + 1Xm The spectra of the RM maps are shown in Figure 1. For all three catalogues, it is clear that the shape of the spectra is c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 Faraday Rotation Template 5 Figure 3. B88 catalogue with 644 sources. Top: lmax =10. Bot- tom: lmax =16. Figure 4. F01 catalogue with 744 sources. Top: lmax =10. Bot- tom: lmax =16. From Figure 5, it is clear that the maxima at lII∼ 270o and minima at lII∼ 90o are the dominant features in all three plots (here we use lII to denote Galactic longitude to avoid confusion with the angular scale l). This maxima/minima pair corresponds to the large-scale magnetic field in the lo- cal Orion spur (sometimes referred to as an arm). These two spots are displaced from the equator to negative Galactic co- ordinates. This asymmetry between the two hemispheres has been widely reported before (eg. Vall´ee & Kronberg 1975; Figure 2. S81 catalogue with 540 sources. Top: lmax =10. Bot- tom: lmax =16. All maps are shown in Galactic coordinates with the Galactic centre in the middle and longitude increasing from right to left. The temperature-colour scales are measured in rad m−2. consistent up to lmax = 16. Extending the series expansion to higher values of lmax leads to fluctuations in the power on the largest-scales (low l). The method finds it harder to reconcile the data with the increasing number of basis functions. The result is that maxima and minima explode as too much freedom is given. This is clearly visible in the maps for lmax=17 and 18 (not displayed). Although looking at the bottom sets of spectra for F01, we see the spectra for lmax=17 is consistent until the octupole (l=3) where it spikes. This suggest that due to the larger data size of F01 it is more able to cope with the demands of increasing the series expansion. Therefore, at times throughout this sec- tion, we will focus our analysis on the RM map from the F01 catalogue with lmax=16. In Figures 2, 3 and 4, we show the RM maps for the S81, B88 and F01 catalogues, respectively. We display only the lmax=10 and 16 maps in order not to overload the reader with information. The r.m.s. values of R for the S81, B88 and F01 RM maps (lmax=16) are 26.4, 23.5 and 21.5 rad m−2, respectively. Maxima (large positive R) are white, whereas, minima (large negative R) are black. The two lim- iting scales enable us to see the progress of structure as the series expansion is extended to include higher modes. We can observe how feature at small l develop as the se- ries extends. Reassuringly, the main features in the lmax=16 maps are also present in the lmax=10 maps. The positions of the maxima and minima remain roughly unchanged. This is compelling evidence that the observed features are real. However, comparison of the maps from the three catalogues is inhibited by the temperature-colour scale varying from map to map. Therefore, for the lmax=16 maps, we force the maximum and minimum scale to be R = 100 rad m−2 ; the results of which are shown in Figure 5. c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 6 P. Dineen & P. Coles Figure 6. Galactic equator cross-section (lmax = 16).Dashed: S81 catalogue; Dotted: B88 catalogue; and Solid line: F01 catalogue Figure 5. RM maps with identical temperature-colour scaling (lmax = 16). From top to bottom: S81, B88 and F01 catalogues. Figure 7. F01 catalogue with both the dipole and quadrupole removed (lmax = 16). Frick et al. 2001); it is usually attributed to the local radio Loop I (the North Galactic spur). Such local distortions are associated with interstellar magnetised superbubbles with typical diameters of 200 pc (Vall´ee 1997). There is also a prominent maxima/minima pair towards the Galactic centre in the RM map formed from the F01 catalogue. The centres of the maxima and minima are at lII = 1o and lII = 346o, respectively . On closer inspection, this feature is present in the RM maps from the other two catalogues. Finally, in the S81 RM map, there is a strong maxima at lII ∼ 50o in the northern hemisphere. This feature is only suggested in the other two maps. Cross-sections, along the Galactic equator, were taken of the RM maps (lmax=16) in order to further understand the features. These are shown in Figure 6. Ideally, with such a slice, maxima/minima locations should indicate the tan- gential direction to spiral arms and directional field changes should correspond to R = 0. However, local distortions and flaws in the map-making process, make this not entirely true. The orion spur location is clear for all three maps. Further- more, the maxima/minima pair towards the Galactic centre (described in the previous paragraph) is evident in all three cross-sections. However, the picture is hazy from lII=30-50o: there is clear field reversal in S81 map; a hint of a reversal in F01 map; and none in B88 map. The maxima and min- ima in the cross-sections could be attributed to the named inner arms (eg. the minima near the Galactic centre to the Norma arm), however, this seems quite speculative given the variation from catalogue to catalogue. It is clear that the Orion spur is the dominant feature. Since the associated maxima/minima pair is separated by 180o, it will be the main source of the dipole (l=1). More- over, we see from the spectra that the quadrupole (l=2) is also strong. Therefore, we remove both the dipole and quadruple from the RM map (lmax=16) compiled from the F01 data. The results of which are displayed in Figure 7. This enable us to view some of the smaller scale features. These details will be hard to explain solely from Galactic magnetic field models. It will be interesting to see, if these small-scale features persist with larger data sets. A study of the global Galactic magnetic field structure would benefit from the removal of local distortions. Conse- quently, we applied the method with the region containing Loop I removed (b > 0o, 0 < lII < 40o, 270 < lII < 360o) (Ruzmaikin & Sokoloff 1977). However, the removed seg- ment was too large to successfully reconstruct the sky given the remaining sources. The lack of restrictions in the seg- ment meant large maxima/minima formed there. This high- lights one disadvantage of spherical harmonic analysis over c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 wavelet analysis that can be localised in both physical and wavenumber spaces. The removal of these local structures is useful for getting a clear picture of the Galactic magnetic field. In CMB foreground studies, however, these structures are essential components of a template. We now turn to the question of errors in the derived RM maps. In principle there are two distinct types of uncertainty that could arise in the analysis we described above. The first concerns : errors intrinsic to the measurement of R and the second relates to errors resulting from sample selection. We have tackled the first type of error by removing sources with the most extreme values of R (those with R > 300 rad m−2), on the grounds that these are least likely to be galactic. We expect the remaining experimental errors to be stochastic and therefore the process of extracting in- formation on large -- scales (as we do) should be unaffected by any underlying noise. The spectrum of this noise should be flat and only dominate on scales where the "real" power is weakest. Experimental errors were addressed in some detail by Frick et al. (2001) in the construction of their catalogue and their versions of the two other catalogues used in our analysis. We feel it would be inappropriate to repeat such a detailed analysis here. The second type of error corresponds to the sampling uncertainties. This error could be computed reliably if we had a large number of independent samples. This is, of course, impossible but even in cases of non-repeatable ob- servations there are resampling techniques that can be used to make reasonable estimates of these errors. The purpose of resampling the data is to generate further sets with the same population distribution as the original. Small perturbations to the original data set will lead to this. For example, Ling, Frenk & Barrow (1986) apply a 'bootstrap' resampling tech- nique to estimate the sampling errors in the two -- point cor- relation function estimated from galaxy and cluster redshift data. The bootstrap technique involves sampling N points (with replacement) from the original data set of N sources in order to create pseudo data sets. The variation over an ensemble of such resamplings is used to estimate the error in the statistic in question for the original data. We applied this method to the F01 catalogue with the intention of find- ing the sample errors in the RM map with lmax set to 16. In total, we produced 50 bootstrap samples from the origi- nal data set and from each of these constructed a RM map. The resampling technique is only designed to test the in- ternal variance of the true data set. Mean values obtained from the ensemble of pseudo data sets are not expected to be good estimators of the true mean values. Therefore, from the bootstrap samples, we calculate the standard deviation σ at each pixel position p σ(p) =vuut Ri(p) − hR(p)i 49 , 50 Xi=1 (18) where i corresponds to the bootstrap sample. From this, we constructed a signal -- to -- noise map where the signal is taken as R and the noise is σ. This map is displayed in Figure 8. The map saturates at a signal -- to -- noise of unity so the regions with low signal to noise are more obvious. The mean signal -- to -- noise across the whole sky is 117. We found this value to be very stable as the number of bootstrap samples increased and this dictated the number of pseudo data sets c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 Faraday Rotation Template 7 Figure 8. Signal -- to -- noise map constructed from 50 bootstrap samples of the F01 catalogue. Dark regions are where the esti- mated signal -- to -- noise exceeds unity. produced. Clearly, the majority of the sky is unaffected by sample errors and we feel reassured that the features seen are not the result of sample selection. It appears that the boundaries between positive and negative regions of the sky where the signal is correspondingly low are the most suscep- tible to sample selection. set The spherical harmonic coefficients generated for three catalogues with lmax to 16 are avail- all able at http://www.nottingham.ac.uk/~ppxptd/rm_maps. Hopefully, this will enable our method to be compared with other techniques and allow the maps to used in the investiga- tion of observables affected by the Galactic magnetic field. Instructions on the generation of full -- sky maps using the HEALPix package are also given at this address. 5 CORRELATIONS WITH CMB MAPS As mentioned in the introduction, RM maps have a gen- eral importance beyond trying to map the Galactic mag- netic field. In what follows, we hope to display one particular function. In this section, we focus solely on the RM map pro- duced from the F01 catalogue with lmax set to 16. Dineen & Coles (2004) developed a diagnostic of foreground con- tamination in CMB maps. The method measured the cross- correlation between the RM of extragalactic sources and the observed microwave signals at the same angular position. In what follows, we seek correlations between the spheri- cal harmonic modes of the RM map and CMB-only maps. In doing so, we shall look at the phases of the (complex) coefficients of the modes from l=2 to l=16. Phase correla- tions have been used before to hunt for evidence of depar- tures in the CMB temperature field from a Gaussian random field (Coles et al. 2004). In Dineen, Rocha & Coles (2004) a certain form of phase correlation was found to be associ- ated with non-trivial topologies. Phase correlations between CMB and foreground maps have been sought before (Nasel- sky, Doroshkevich & Verkhodanov 2003; Chiang & Nasel- sky 2004), however, here we wish to emphasise the virtue of having independent probes of Galactic foreground contami- nation. In order to seek evidence of phase correlations between the RM map and CMB data, we turned to two WMAP- derived maps. Both were constructed in a manner that min- imises foreground contamination and detector noise, leaving a pure CMB signal. The ultimate goal of these maps is to 8 P. Dineen & P. Coles build an accurate image of the last scattering surface (LSS) that captures the detailed morphology. Following the release of the WMAP 1 yr data, the WMAP team (Bennett et al. 2003), and Tegmark, de Oliveira-Costa & Hamilton (2003; TOH) have released CMB-only sky maps (see papers for de- tails). We use the WMAP team's internal linear combination (ILC) map and the Wiener-filtered map of TOH. The latter was chosen since the map was found to be correlated with RM values in Dineen & Coles (2004). Two measures of phase association were used: the circu- lar cross-correlation coefficient R and Kuiper's statistic V . Both statistics will be evaluated at each scale l from 2 to 16. If we let ΦRM and ΦCMB be the phases of the RM and CMB maps, respectively. Then, following Fisher (1993), R is defined as: l cos(Φm,RM − Φm,CMB). (19) R(l) = l−1 Xm=1 The expectation value of R is 0, and hence highly corre- lated phases are associated with large values of R. Kuiper's statistic is calculated from the available set of phase differ- ences (Φm,RM − Φm,CMB) at a given scale. First, the phase differences are sorted into ascending order, to give the set {Θ1, . . . , Θp}. Each angle Θj is divided by 2π to give a set of variables Xj , where j = 1 . . . p. From the set of Xj we derive two values S+ p and S− p where S+ 2 p − X1, p = max(cid:26) 1 p = max(cid:26)X1, X2 − p − X2, . . . , 1 − Xp(cid:27) p (cid:27) . p − 1 , . . . , Xp − 1 p and S− Kuiper's statistic is then defined as V (l) = (S+ p + S− p ) ·(cid:18)√p + 0.155 + 0.24 √p (cid:19) . (20) (21) (22) The form of V is chosen so that it is approximately inde- pendent of sample size for large p. Anomalously large values of V indicate a distribution that is more clumped than a uniformly random distribution, while low values mean that angles are more regular. To access the significance of the values of R and V ob- tained from the comparison of the RM map with the two CMB maps, we make use of Monte Carlo (MC) skies with uniformly random phases. The statistics were calculated for 10,000 MC skies contrasted with a further 10,000 MC skies. Thus, we are left with 10,000 values of R and V for each scale. The results from both CMB maps suggests that there is strong correlations between the phases at l=11. For the ILC, the values of R(11) and V (11) are greater than 99 percent of the MC values. Whereas, the values of R(11) and V (11) corresponding to the TOH Wiener-filtered map are greater than 97 and 98 percent of the MC skies, respectively. In Figure 9 we plot the ILC map constructed with the al,m for l=11. Overlapping this image with that of the RM in Fig 4, we can see that the central maxima/minima pair in the RM map are similar in location, size and shape to structure in the ILC image (but with colours reversed). This is prob- ably what determines the specific scale l=11. Interestingly, Figure 9. Internal linear combination map constructed with al,m for l=11 only. l=11 corresponds to the scale that Naselsky, Doroshkevich & Verkhodanov (2003) found the greatest level of correla- tion between the ILC phases and those of the foreground maps. 6 CONCLUSION In this paper, we have presented a new method to generate all-sky RM maps from uneven and sparsely populated data samples. The method calculates a set of functions orthonor- mal to the data set. With these basis functions, the spherical harmonic coefficients are calculated and converted into sky maps using the HEALPix package. The method was applied to three catalogues; S81, B88 and F01 catalogues. Maps from each catalogue showed evidence of the magnetic field in our local Orion spur, the North-South asymmetry attributed to radio Loop I and a maxima/minima pair close to the Galac- tic centre that possibly corresponds to the magnetic field of two inner Galactic arms. A RM map constructed from the S81 catalogue also had a prominent maxima at lII ∼ 50o in the northern hemisphere. In Section 5, we showed the benefits a RM map has to CMB foreground analysis. Phase correlations were sought between RM maps and those of CMB-only maps derived from the WMAP data. For both the WMAP team's ILC map and the Wiener-filtered map of TOH, phases corre- sponding to l=11 were found to be highly correlated. Nasel- sky, Doroshkevich & Verkhodanov (2003) found the same scale to display phase correlations when carrying out a sim- ilar analysis on the ILC map and foreground maps. Their detection of correlations at the same scale as our analysis, reaffirms that the RM catalogues are valuable independent tracers of CMB foregrounds (Dineen & Coles 2004). Modelling foregrounds will play a crucial role in CMB polarisation studies. Foreground contamination is expected to be more severe than in the temperature measurements (Kosowsky 1999). Consequently, superior templates for the individual foreground components are required. RM maps will help trace these components. Besides this, extrapolation of low frequency measurement of synchrotron polarisation to CMB frequencies has been shown to be complicated by Faraday rotation (de Oliveira-Costa et al. 2003). Again, this underlines the importance of developing templates of the Faraday rotation of the Galactic sky. Efforts to map the RM sky will be greatly enhanced by increased source catalogues for both extragalactic sources and pulsars within our Galaxy. This may enable the formation of a 3-dimensional image of c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 Faraday Rotation Template 9 Tegmark M., de Oliveira-Costa A. & Hamilton A., 2003, Phys. Rev. D, 68, 123523 Vall´ee J.P., 1997, Fundamentals of Cosmic Physics, 19, 1 Vall´ee J.P. & Kronberg P.P., 1975, A&A, 43, 233 Wasserman I., 1978, ApJ, 224, 337 Widrow L.M., 2002, Rev. Mod. Phys., 74, 775 Wolfe A.M., Lanzetta K.M, & Oren A.L., 1992, ApJ, 388, 17 the Galactic magnetic field. Furthermore, attempts to map the RM sky will be enhanced by upcoming satellite CMB polarisation experiments which present unprecedented sky- coverage and resolution. ACKNOWLEDGEMENTS We thank Anvar Shukurov and Rodion Stepanov for provid- ing us with the rotation measure catalogues and useful com- ments. We gratefully acknowledge the use of the HEALPix package and the Legacy Archive for Microwave Background Data Analysis (LAMBDA). Support for LAMBDA is pro- vided by the NASA Office of Space Science. REFERENCES Barrow J.D., Ferreira P.G. & Silk J., 1997, Phys. Rev. Lett., 78, 3610 Bennett C.L. et al., 1994, ApJ, 436, 423 Bennett C.L. et al., 2003, ApJS, 148, 97 Broten N.W., MacLeod J.M. & Vall´ee J.P., 1988, Ap&SS, 141, 303 Chen G., Mukherjee P., Kahniashvili T., Ratra B. & Wang Y., 2004, ApJ, 611, 655 Chiang L.-Y. & Naselsky P., 2004, astro-ph/0407395 Coles P., Dineen P., Earl J. & Wright D., 2004, MNRAS, 350, 983 de Oliveira-Costa A., Tegmark M., O'Dell C., Keating B., Timbie P., Efstathiou G. & Smoot G., 2003, Phys. Rev. D, 68, 83003 de Oliveira-Costa A., Tegmark M., Davies R.D., Guti´errez C.M., Lasenby A.N., Rebolo R. & Watson R.A., 2004, ApJ, 606, L89 Dineen P. & Coles P., 2004, MNRAS, 347, 52 Dineen P., Rocha G. & Coles P., 2004, submitted to MNRAS, astro-ph/0404356 Draine B.T. & Lazarian A., 1998, ApJ, 494, L19 Fisher N.I., 1993, Statistical Analysis of Circular Data. Cam- bridge University Press, Cambridge. Frick P., Stepanov R., Shukurov A. & Sokoloff D., 2001, MNRAS, 325, 649 G´orski K.M., 1994, ApJ, 430, L85 G´orski K.M., Hivon E. & Wandelt B.D., 1999, in Proceedings of the MPA/ESO Conference Evolution of Large-Scale Struc- ture, eds. A.J. Banday, R.S. Sheth and L. Da Costa, Print- Partners Ipskamp, NL, pp. 37-42 (also astro-ph/9812350) Han J.L., 2004, astro-ph/0402170 Kim K.-T., Tribble P.C., Kronberg P.P. 1991, ApJ, 379, 80 Kogut A. et al., 2003, ApJS, 148, 161 Kosowsky A., 1999, New Astronomy Reviews, 43, 157 Kosowsky A. & Loeb A., 1996, ApJ, 469, 1 Kovac J. M., Leitch E. M., Pryke C., Carlstrom J. E., Halverson N. W. & Holzapfel W. L., 2002, Nature, 420, 772 Kronberg P.P., Perry J.J. & Zukowski E.L., 1992, ApJ, 387, 528 Lewis A., 2004, accepted by Phys. Rev. D, astro-ph/0406096 Ling E.N., Frenk C.S. & Barrow J.D., 1986, MNRAS, 223, 21 Naselsky P., Doroshkevich A. & Verkhodanov O., 2003, ApJ, 599, L53 Naselsky P.D., Chiang L.-Y., Olesen P. & Verkhodanov O.V., 2004, accepted by ApJ, astro-ph/0405181 Ruzmaikin & Sokoloff, 1977, A&A, 58, 247 Scannapieco E.S. & Ferreira P.G., 1997, Phys. Rev. D, 56, R7493 Sc´occola C., Harari D. & Mollerach S., 2004, astro-ph/0405396 Seymour P.A.H., 1966, MNRAS, 134, 389 Seymour P.A.H., 1984, QJRAS, 25, 293 Simard-Normandin M., Kronberg P. & Button S., 1981, ApJS, 45, 97 Sofue Y. & Fujimoto M., 1983, ApJ, 265, 722 c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
astro-ph/0603590
2
0603
2006-05-25T13:56:47
Swift Observations of the highly X-ray variable Narrow Line Seyfert 1 galaxy RX J0148.3-2758
[ "astro-ph" ]
We report on Swift observations of the Narrow-Line Seyfert 1 galaxy (NLS1) RX J0148.3--2758. It was observed for 41.6 ks in 2005 May and for 15.8 ks in 2005 December. On short as well as on long timescales RX J0148.3--2758 is a highly variable source. It doubles its X-ray flux within 18-25 ks. The observation of 2005 December 09, which had a flux 4 times lower than during the 2005 May observations, shows a significant hardening of the X-ray hardness ratio compared with the 2005-May and 2005-December 20/21 observations. A detailed analysis of the X-ray spectra shows that we actually observe two spectral changes in RX J0148.3-2758: First a decrease of the soft X-ray component between 2005 May and December 09, which is most likely due to an increase of the intrinsic absorber column, and second a decrease of the hard X-ray flux in the December 20/21 observations. The soft X-ray spectral slope $\alpha_{\rm X, soft}$=2.58$^{+0.15}_{-0.12}$ during the high state in 2005 May agrees well with that measured by ROSAT (\axs=2.54\plm0.82). In contrast to the strong X-ray variability, the analysis of the Swift UVOT photometry from December 2005 of RX J0148.3--2758 shows no significant variability in any of the 6 UVOT filters. From the simultaneous X-ray and UV observations in 2005 December we measured the X-ray loudness alpha-ox varies between alpha-ox=1.5 and 1.8. Our Swift observations of RX J0148.3-2758 demonstrate the great potential the multi-wavelength observatory Swift has for AGN science. (shortened)
astro-ph
astro-ph
Draft version September 30, 2018 Preprint typeset using LATEX style emulateapj v. 6/22/04 SWIFT OBSERVATIONS OF THE HIGHLY X-RAY VARIABLE NARROW LINE SEYFERT 1 GALAXY RX J0148.3 -- 2758 Dirk Grupe1 , Karen M. Leighly2, Stefanie Komossa3, Patricia Schady1,4, Paul T. O'Brien5, David N. Burrows1, John A. Nousek1 Draft version September 30, 2018 ABSTRACT We report on Swift observations of the Narrow-Line Seyfert 1 galaxy (NLS1) RX J0148.3 -- 2758. It was observed for 41.6 ks in 2005 May and for 15.8 ks in 2005 December. On short as well as on long timescales RX J0148.3 -- 2758 is a highly variable source. It doubles its X-ray flux within 18-25 ks. The observation of 2005 December 09, which had a flux 4 times lower than during the 2005 May observations, shows a significant hardening of the X-ray hardness ratio compared with the 2005-May and 2005-December 20/21 observations. A detailed analysis of the X-ray spectra shows that we actually observe two spectral changes in RX J0148.3 -- 2758: first, a decrease of the soft X- ray component between 2005 May and December 09, which is most likely due to an increase of the intrinsic absorber column, and second, a decrease of the hard X-ray flux in the December 20/21 observations. The soft X-ray spectral slope αX,soft=2.58+0.15 −0.12 during the high state in 2005 May agrees well with that measured by ROSAT (αX,soft=2.54±0.82). This soft X-ray spectrum is superimposed on a hard X-ray component with αX,hard=0.96+0.15 −0.12 which in consistent with the hard X-ray spectral slope αX,hard=1.11+0.16 −0.42 measured during the December 09 observation, agrees well with αX,soft=2.03+0.23 −0.20 measured from the ASCA observation when RX J0148.3 -- 2758 was also in a low state. In contrast to the strong X-ray variability, the analysis of the Swift UVOT photometry from December 2005 of RX J0148.3 -- 2758 shows no significant variability in any of the 6 UVOT filters. From the simultaneous X-ray and UV observations in 2005 December we measured the X-ray loudness αox and found it to vary between αox=1.5 and 1.8. Our Swift observations of RX J0148.3 -- 2758 demonstrate the great potential that the multi-wavelength observatory Swift has for AGN science. Subject headings: galaxies: active, galaxies: individual (RX J0148.3 -- 2758) −0.19 found by ASCA. The soft X-ray slope αX,soft=1.93+0.58 1. INTRODUCTION Most of the power in the spectral energy distribution (SED) of an AGN is contained in the Big Blue Bump (BBB, Shields 1978). As suggested by Walter & Fink (1993), the BBB may stretch from the UV into the soft X-ray regime. The soft X-ray part of the BBB may be UV photons from the accretion disk which are shifted into the soft X-ray band by Comptonization in the ac- cretion disk corona (e.g. Pounds et al. 1995). Based on their sample of soft X-ray selected ROSAT AGN, Grupe et al. (1998a) showed that the BBB extends as far as the optical band and that sources with steeper X- ray spectra tend to have bluer optical spectra, suggesting that Narrow Line Seyfert 1 galaxies are the AGN with the strongest BBB component. However, from a study of the IUE spectra of NLS1s, Rodr´iguez-Pascual et al. (1997) came to the conclusion that NLS1s have weaker UV emission than Broad Line Seyfert 1s. All these stud- ies, however, were hampered by the lack of simultaneous 1 Department of Astronomy and Astrophysics, Pennsylvania State University, 525 Davey Lab, University Park, PA 16802 2 Homer L. Dodge Department of Physics and Astronomy, Uni- versity of Oklahoma, 440 West Brooks Street, Norman, OK 73019; email: [email protected] 3 MPI fur extraterrestrische Physik, Giessenbachstr., D-85748 Garching, Germany; email: [email protected] 4 Mullard Space Science Laboratory, Holmbury St. Mary, Dork- ing, Surrey RH5 6NT, U.K.; email: [email protected] 5 Department of Physics & Astronomy, University of Leicester, Leicester, LE1 7R, UK, email: [email protected] observations in the optical/UV and X-ray bands; the ob- servations available frequently had been performed years apart. This situation has changed now with the availabil- ity of the multi-wavelengths observatories XMM-Newton and Swift. Swift (Gehrels et al. 2004) is a multi-wavelength mis- sion equipped with three telescopes that together cover the electromagnetic spectrum between 6000A to 150 keV: the Burst Alert Telescope (BAT, Barthelmy 2005), the X-Ray Telescope (XRT, Burrows et al. 2005), and the UV-Optical Telescope (UVOT, Roming et al. 2005). Swift was designed to chase Gamma-Ray Bursts (GRBs), and was launched on 20-November-2004. At the low- energy side of Swift's observing window, the UVOT covers the wavelengths range between 1700-6000A. The UVOT is a sister instrument of XMM's Optical Monitor (OM, Mason et al. 2001), equipped with a similar set of filters (Mason et al. 2001; Roming et al. 2005). The XRT covers the 0.3-10.0 keV range, and uses a CCD detector identical to the EPIC MOS on-board XMM (Turner et al. 2001). As described by Hill et al. (2004) the XRT operates in three observing modes: the Photon Counting (PC) which is equivalent to the full-frame mode on XMM, Window Timing (WT), and Low-Rate Photo- diode mode (LrPD). Due to the nature of the Swift mis- sion, the XRT switches automatically between the ob- serving modes according to the brightness of a source. Only for specific purposes, e.g. in case of calibration ob- servations, are the modes set manually in the observing 2 Grupe et al. schedule. The BAT is a coded-mask experiment that operates in the 15-150 keV energy range, at the high- energy part of Swift's observing window. Although the main purpose of the Swift mission is to detect and observe GRBs, fill-in targets are used in the observing schedule to optimize the scientific return of the mission when GRBs are not observable. Swift´s UV and X-ray capabilities, and rapid and flexible scheduling make it an ideal obser- vatory to study AGN. With the launch of 1996; Grupe 1996; Grupe et al. the X-ray satellite ROSAT (Trumper 1982) the X-ray energy range down to 0.1 keV became accessible for the first time. During the half-year ROSAT All-Sky Survey (RASS, (RASS, Voges et al. 1999) a large number of sources with steep X-ray spectra were detected (Thomas et al. 1998; Beuermann et al. 1999; Schwope et al. 2000). About one third to one half of these sources are AGN. Grupe (1996) and Grupe et al. (1998a, 2004a) found that about 50% of bright soft X-ray selected AGN are Narrow-Line Seyfert 1 galaxies (NLS1s, Osterbrock & Pogge 1985; Goodrich 1989). They turned out to be the class of AGN with the steepest X-ray spectra (e.g., Puchnarewicz et al. 1992; Boller et al. 1998a; Grupe et al. 2001a; Vaughan et al. 2001; Grupe et al. 2004a; Williams et al. 2002) and often show very strong X-ray variability (e.g., Boller et al. 1996; Nandra et al. 1997; Leighly 1999a; Turner et al. 1999; Grupe et al. 2001a). NLS1s are AGN with extreme properties which seem to be linked to one another: a steeper X-ray spec- tral index αX correlates with the strength of the op- tical FeII emission and anti-correlates with the widths of the Broad Line Region (BLR) Balmer lines and the strength of the Narrow-Line Region (NLR) forbidden lines (e.g., Grupe 1996; Grupe et al. 1999; Grupe 2004; Laor et al. 1994, 1997; Sulentic et al. 2000). All these relationships are governed by a set of fundamental under- lying parameters, usually called the Boroson & Green (1992) 'Eigenvector-1' relation in AGN. The most ac- cepted explanation for these Eigenvectors is the Edding- ton ratio L/LEdd or the mass of the central black hole MBH (Boroson 2002; Sulentic et al. 2000; Grupe 2004; Yuan & Wills 2003) in which NLS1s are AGN with the highest Eddington ratios and smallest black hole masses for a given luminosity. The Eddington ratio has also been found to be correlated with the X-ray spectral slope αX (Grupe 2004). Alternatively, this can also be inter- preted as the age of an AGN in which NLS1s are AGN in an early stage of their development (Grupe 1996; Grupe 2004; Mathur 2000). RX J0148.3 -- 2758 (α2000=01 48 22.3, δ2000= -- 27 58 26, z=0.121) was discovered during the RASS as a bright and variable X-ray source (Grupe et al. 1998a; Thomas et al. 1998; Schwope et al. 2000). It was iden- tified as a NLS1 by Grupe (1996) and Grupe et al. (1999). Besides a later 6.7 ks ROSAT PSPC ob- servation (Grupe et al. 2001a), RX J0148.3 -- 2758 was also observed for 34 ks by ASCA (Turner et al. 1999; Vaughan et al. 1999). The 2-10 keV ASCA light curve shows that the source is highly variable (Turner et al. 1999). Its 2-10 keV spectral slope α2−10 keV=0.99±0.17 is typical for a Seyfert 1 galaxy (Vaughan et al. 1999). In this paper we present our observations of RX J0148.3 -- 2758 with Swift and we compare those with the data pre- viously taken by ROSAT and ASCA. RX J0148.3 -- 2758 was one of the most X-ray variable AGN in the soft X- ray selected AGN sample of Grupe et al. (2001a). The outline of this paper is as follows: in § 1 we de- scribe the Swift, ROSAT and ASCA observations and the data reduction, in § 3 we present the results of the Swift data analysis, and in § 4 we discuss the results. Through- out the paper, spectral indexes are quoted as energy spec- tral indexes with Fν ∝ ν−α. Luminosities are calculated assuming a ΛCDM cosmology with ΩM=0.27, ΩΛ=0.73 and a Hubble constant of H0=75 km s−1 Mpc−1 using the luminosity distances given by Hogg (1999). All er- rors are 1σ unless stated otherwise. 2. OBSERVATIONS AND DATA REDUCTION ′′ ′′ RX J0148.3 -- 2758 was observed by Swift between 2005- May-05 and 2005-May-13 (segments 002-006) for a to- tal of 41.6 ks and between 2005-December-07 and 2005- December-21 for 15.8 ks (segments 008-011). Table 1 lists the segment numbers of the Swift observations, the start and end times, the total observing times and the 0.2-2.0 keV rest-frame luminosities. All XRT observations were performed in PC mode. The event files were created with the standard Swift XRT analysis task xrtpipeline version 0.9.9. For both spectral and temporal analysis, source counts between 0.3-10 keV were extracted from a circle with a radius of 50 , and background photons were ex- tracted from a 100 circle in a source-free region near the source. We created source and background spectra and event files using XSELECT version 2.3. Background- subtracted light curves were created by using ESO's Mu- nich Image Data Analysis System MIDAS version 04Sep as described in Nousek et al. (2006). The data were binned to have 250 source + background photons per bin except for the 2005 December 07 observation where we used a binning of 150 photons per bin. Note that on 2005 May 27 the Swift XRT detector was hit by a micro-meteorite that caused some damages. In partic- ular the CCD columns DETX=294 and 320 had to be turned off afterwards. While our 2005 May and the 2005 December 20/21 observations are not affected by those dead columns, the 2005 December 07 and 09 observations were in part. For the latter data sets, a correction was applied if the source felt on one of the dead columns to account for the loss of photons this caused. The spectra were rebinned using grppha version 3.0.0 to have at least 20 photons per bin and analyzed using XSPEC 12.2.1 (Arnaud 1996). The Auxiliary Response files were cre- ated using the Swift analysis task xrtmkarf. We used the standard response matrix version 007 with grade selec- tion 0 to 12. Due to the low count rate, the data were not affected by pileup. Swift UVOT data were obtained during 2005 May 11 and 13 (segments 004 and 006), and 2005 December. The UVOT was blocked during 2005 May 05 and 07 (segments 002 and 004) observations. The UV grism was used for the observations of 2005 May 11 and 13, and during 2005 December UVOT photometry was performed. Due to the on-going calibration of the UV grisms and the require- ment for well-calibrated UV grism data in our analysis, we do not present these data at this point, and discuss only the UVOT photometry results of the 2005 December observations. Observations were taken in the three optical and three UV filters available on the UVOT (Roming et al. 2005) Swift observation of RX J0148.3 -- 2758 3 with the exception of 2005 December 07 (segment 008) observation, in which no B band observations were made. This covers the wavelength range from 1700 to 6000 A. There was a bright star (B∼12.0 mag) about 10′′ from RX J0148.3-27758 that made it necessary to carry out the UVOT photometry using a 4.5′′ source extraction re- gion. This is smaller than the 6′′ and 12′′ radii that are used for the optical and UV filters, respectively, for the compatibility with the current effective area calibrations. An aperture correction was therefore applied to account for source photon counts that lay outside of this extrac- tion region, in the wings of the PSF. The background region was taken from an annulus around the source, off- centered by ∼ 7 to avoid excessive contamination from the nearby star. Source photon counts, magnitudes and fluxes were then extracted using the UVOT tool uvot- maghist version 1.0 for every individual exposure taken in each filter, as well as from the co-added exposures within each segment number. All UVOT magnitudes were cor- rected for Galactic reddening with EB−V=0.017. ′′ In order to be able to carry out broadband spectral fit- ting, source and background data files compatible with XSPEC were created from the co-added exposures from the 2005 December 09 (segment 009) observations. This was done using the tool uvot2pha version 1.1. This pro- vided a single spectral file per filter. The same source and background extraction regions were used as before, and the exposure times in the headers were changed to normalize the count rates to the rates with the aperture correction taken into account. The field of RX J0148.3 -- 2758 was also observed by the BAT. However, a preliminary analysis of the BAT pointed and survey data does not show a detection of the source. So far more than 100 AGN have been detected by the BAT, of which about 50 have had the results pub- lished (Markwardt et al. 2006). RX J0148.3 -- 2758 was observed by ROSAT with the Position Sensitive Proportional Counter (PSPC, Pfeffermann et al. 1987) three times during the RASS for a total of 504 s and for 6.7 ks in a pointed obser- vation (Table 1). Source counts were selected in a cir- cular region with R=200 . For the RASS observations background photons were taken from two circular regions with R=400 in the ROSAT scan direction (for details see Belloni et al. 1994). For the pointed observation the background was estimated from a close-by circular region with R=400 . Spectra were rebinned to have at least a S/N=5 in each bin. The light curves were binned in 400s bins. The ROSAT data were processed using the EXSAS version Apr01 (Zimmermann et al. 1998). ′′ ′′ ′′ ASCA observed RX J0148.3-2758 on 1997 November 7 for a total of 33.2 ks with its Solid-state Imaging Spectrometers (SIS) and 36.4 ks with the Gas Image Spectrometers (GIS) on 1997-07-11 (Table 1). A stan- dard configuration was used during the observation. The Gas Imaging Spectrometers (GISs) were operated in PH mode throughout the observation. The Solid-state Imag- ing Spectrometers were operated in 1-CCD Faint mode. The SIS energy gain was reprocessed using the latest cal- ibration file (sisph2pi 290301.fits). We used standard cri- teria for reducing the ASCA data. For the SIS detectors, source photons were extracted from a circular region 3.5′ in radius, and for the GIS detectors, the source extraction region is 5.25′ in radius. In both cases, the background was drawn from source-free regions of the detectors. For spectral fitting, the spectra were grouped so that there at least 20 photons pre bin. It has been demon- strated that the SIS spectra suffered degradation during the mission. The SIS efficiency loss can be parameterized by adding additional absorption to the model, where the amount of additional absorption depends on the time of the observation6. For the time of the RX J0148.3-2758, the appropriate additional column is 4.01 × 1020 cm−2. We fit the SIS0 spectrum between 0.5 and 8.0 keV, the SIS1 spectrum between 1 and 0.8 keV, and the two GIS spectra between 0.8 and 8.0 keV. 3. RESULTS 3.1. X-rays 3.1.1. Light curves The left panel of Figure 1 shows the Swift-XRT light curve of RX J0148.3 -- 2758 during the 2005 May observa- tions (segments 002-006). The middle and right panels show the observations from 2005 December 07-09 (seg- ments 008+009) and 2005 December 20/21 (segments 010+011), respectively. The XRT count rate light curves shown in the upper panels of Figure 1 suggest that RX J0148.3-2758 is a highly variable source. In general the AGN varies between ≈ 0.1 to 0.4 count s−1. The most dramatic variability can be seen in the May 2005 light curve (left panel of Figure 1), where the count rate dou- bled in 25 ks between 90 -- 115 ks, followed by a rapid drop between 120 -- 150 ks by a factor of more than 2. A simi- lar increase in count rate was also observed at the end of the May 2005 observation when RX J0148.3 -- 2758 dou- bled its count rate within 18 ks. The 2005 December observations show that during the December 09 obser- vation (middle panel) RX J0148.3 -- 2758 became signifi- cantly fainter with a count rate of about 0.18 counts s−1. The AGN count rate fell by a further factor of 3 when it was re-observed on 2005-December-20 (segment 010, right panel). Whether RX J0148.3 -- 2758 remained in this low state during the 10 day gap . We do not know what happened during the between 2005 December 09 and 20, whether RX J0148.3 -- 2758 remained in this low state dur- ing the 10 day gap, or whether further variability took place. By the end of segment 011 the count rate went back to its 'normal' level, having increased by a factor of about 4 within 30 ks. The observations of 2005 De- cember 21st were discontinued at the end of segment 011 due to the trigger of GRB 051221A (Parsons et al. 2005; Burrows et al. 2006) which superseded the RX J0148.3 -- 2758 observation. The hardness ratio7 plots suggest the presence of some spectral variability. While the hardness ratios during the December 07 and 20/21 observations are similar to the ones measured during May 2005, the hardness ratio of the December 09 observations are significantly harder, suggesting a change in the X-ray spectrum. The left panel of Figure 2 shows the RASS light curves and the right panel displays the pointed ROSAT PSPC 6 see http://heasarc.gsfc.nasa.gov/docs/asca/calibration/nhparam.html for details. 7 The hardness ratio is defined by (hard-soft)/(hard+soft) with soft are the count in the 0.3-1.0 keV band and hard in the 1.0-10.0 keV band. 4 Grupe et al. light curve. In both light curves RX J0148.3 -- 2758 dis- plays a similar variability, in agreement with the Swift- XRT light curve (Figure 1). RX J0148.3 -- 2758 was one of the most variable of the soft X-ray selected AGN sample of Grupe et al. (2001a). RX J0148.3 -- 2758 was also observed by ASCA for a pe- riod of about 1 day. Light curves were extracted in the 0.5 -- 10 keV band for SIS detectors, and 0.8 -- 10 keV band for the GIS detectors. The average net source count rates were 0.054, 0.043, 0.022, and 0.028 counts s−1 in the SIS0, SIS1, GIS2, and GIS3 detectors, respectively. The back- ground fraction in the source regions are estimated to be 20%, 22%, 36% and 30% in the SIS0, SIS1, GIS2, and GIS3 detectors, respectively. The SIS0+SIS1 net count rate light curve was binned by orbit, and is displayed in Figure 3. The light curve shows that RX J0148.3 -- 2758 is varied by a factor of about 2, consistent with pre- vious findings (e.g. Nandra et al. 1997), which found RX J0148.3 -- 2758 to show the largest excess variance among the AGNs observed by ASCA (Turner et al. 1999). The long term light curve shown in Figure 4 displays the complete set of X-ray observations performed on RX J0148.3 -- 2758. The rest-frame 0.2-2.0 keV luminosi- ties were derived from the unabsorbed fluxes determined from the best-fit spectral models as described in § 3.1.2. All luminosities are listed in Table 1. The light curve shows that during the ASCA observation RX J0148.3 -- 2758 was in a low state - about 13 times fainter than dur- ing the RASS observation in December 1990. During the Swift observation in May 2005 and at the beginning of the 2005 December observations, RX J0148.3 -- 2758 was at a similar brightness as in first RASS and the 1992 pointed ROSAT observations. By the end of the 2005 December Swift observations, RX J0148.3 -- 2758 had become signif- icantly fainter, only a factor of about 2 brighter than during the ASCA low-state. 3.1.2. Spectral Analysis Figure 5 displays the spectra of the 2005-May (left panel) and the 2005-December (right panel) observa- tions. All spectra were initially fitted using a single ab- sorbed power law with the absorption column density at z=0 fixed to the Galactic value (1.50× 1020 cm−2 Dickey & Lockman 1990). A simple absorbed power law did not produce acceptable fits (Table 2). With the ex- ception of the 2005 December 09 observations, all spectra require multi-component spectral models such as black- body plus power law or a broken power law model, with an additional absorption component above the Galactic column density. A broken power law model as well as a blackbody plus power law model yield similar χ2/ν and we cannot distinguish between the models. Using a blackbody model over a hard power law component yields a temperature kT≈100-120 eV which is typical for a NLS1 and agrees with the value kT=120 eV found by ASCA (Table 3). A broken power law model simul- taneously fitted to the 2005 May XRT spectra results in a soft X-ray spectral slope αX,soft=2.58+0.15 −0.12 which is in good agreement with the results found by ROSAT (both the RASS and pointed observations; Table 3). The XRT hard X-ray spectral slope αX,hard=0.96+0.16 −0.14 is also in good agreement with the hard X-ray spectral slope from the ASCA data (Table 3). In addition to the phe- nomenological models we fitted the partial covering ab- sorber model pcfabs, the 'warm', ionized absorber model absori, the reflection model pexrav, and the disk black- body model to the 2005 May spectra. We found the absori and disk blackbody models to show no improve- ment over a simple powerlaw model, and the parameters of the reflection model pexrav could not be constrained. However, the partial covering model pcfabs yields rea- sonable results. We found the 2005 May spectra to be well fit by a partial covering absorber with a column density NH=7.2+1.7 −1.3 × 1022 cm−2, a covering fraction f = 0.80+0.03 −0.04, and a spectral index αX=2.34±0.09 where χ2/ν=428/302, which is significantly better than a single power law fit as listed in Table 2. Although the 2005 December 07 observations provided poorly constrained spectral fits due to the small num- ber of photons (290 in 822 s), some interesting spectral variability is observed in the 2005 December data set. The hardness ratio light curve (lower right panel of Fig- ure 1) suggests that RX J0148.3 -- 2758 had a similar spec- trum as during the 2005 May and 2005 December 20/21 observations. However, a fit to the 6.3 ks observation from 2005 December 09 with a single absorber at z=0 yields an absorption column density consistent with the Galactic value, suggesting there is no intrinsic absorp- tion. Furthermore, an absorbed broken power law model yields a soft X-ray spectral slope αX,soft=1.93+0.58 −0.42 flat- ter than during the 2005-May observations. To exam- ine this spectral change, a Target-of-Opportunity obser- vation was made with Swift on 2005 December 20/21. The spectral analysis of these data show RX J0148.3 -- 2758 to have once again become intrinsically absorbed, with NH,intr = 11.6+7.5 −5.7 × 1020 cm−2, and the soft X-ray spectral slope αX,soft=3.41+0.78 −0.64 to have become signifi- cantly steeper. It is also interesting to note that between the 2005 May and December observations the best-fit spectral break from a broken power law model fit has shifted towards softer energies. During the 2005 May observations the break energy was at Ebreak = 1.68+0.12 −0.14 keV while during the 2005 December observations the break energy shifted to Ebreak ≈ 1.2 keV in agreement with the break energy Ebreak=1.36+0.16 −0.19 found by ASCA when RX J0148.3 -- 2758 was also in a low state. As a result of the low number of photons in the December 09 observation the parameters of a partial covering ab- sorber were poorly constrained if all parameters were left to vary. However, by fixing the spectral index to the value during the 2005 May observations, αX=2.33, the fit resulted in a partial covering absorber with essen- tially the same covering fraction f =0.75+0.06 −0.09 as during the 2005 May observation, but with a significantly lower NH=1.9+1.7 −0.9 × 1022 cm−2 with χ2/ν=35/35. A partial covering absorber model fit to the December 20/21 spec- trum find an increase of the absorption column of the partial covering absorber to NH=3.6+3.2 −1.4 × 1022 cm−2, a covering fraction f = 0.87+0.06 −0.44, with χ2/ν=30/27. −0.12 and αX=3.98+0.48 The left panel of Figure 6 displays the spectra from the merged data sets of the 2005 May observations, the data set from 2005 December 09, and the merged data set from December 20/21. The right panel shows the corresponding contour plots between the intrinsic col- umn density and the photon index Γ = αX + 1. The Swift observation of RX J0148.3 -- 2758 5 lack of overlaps between the 2005-May, 2005 December 09, and 2005 December 20/21 observations suggest the presence of significant spectral variability in RX J0148.3 -- 2758. The spectra in the left panel of Figure 6 show how the spectra change: compared with the 2005 May obser- vations, the spectrum from December 09 has a similar hard X-ray flux, but a significantly lower flux in the soft X-ray component. Then in the December 20/21 obser- vations the soft X-ray component remained at a similar level as the December 09 observation. The hard X-ray flux, however, decreased by a factor of 4. By the end of the December 21st observation RX J0148.3 -- 2758 in- creased its 0.3-10.0 observed flux by a factor of 3 (Figure 1). The short-term variability we observed in RX J0148.3 -- 2758 during the Swift observations seems to reflect the previous measurements by ROSAT and ASCA. During the high state observations during the RASS and ROSAT pointed observations the soft X-ray spectral slope was steep with αX,soft=2.62 and 2.25, respectively. For the spectral analysis of the ASCA data, we first constrain the power law index by fitting the region between 2 and 5 keV with a power law model. We obtain a good fit (χ2 = 158 for 173 degrees of freedom) and measure the energy index to be 1.16+0.27 −0.26. Next, we include the pho- tons between 5 and 8 keV. The residuals show a slight excess that may indicate the presence of a reprocessing component. Indeed, when we plot the spectrum in this bandpass, we find the photon index flattens to 2.01, al- though the difference is not significant. We add a narrow iron line at 6.4 keV, but find no significant decrease in χ2 (∆χ2 = 1.78). Allowing the line energy to vary yields a better fit with 6.70+0.18 −0.24 keV, and a greater reduction in χ2. However, the line equivalent width is very large (550 eV), and an F-test indicates the change in χ2 of 5.9 compared with the no-line model to not be a significant improvement. Similarly, a broad line does not improve the fit significantly, and produces a line with unphysi- cally large equivalent width. We conclude that evidence for a line in these data is weak, most likely because of the low signal-to-noise ratio at high energies in the spectrum. Since there is some flattening that distorts the powerlaw, we ignore the spectra above 5 keV henceforth. Note that due to the lower effective area in the Swift XRT at 6 keV we were not able to identify the line with the XRT. Next, we examine the spectrum at low energies. Ex- trapolating down to the lower limits described above, we find that the continuum subtly steepens toward low ener- gies. Indeed, fitting between 1 and 5 keV gives an energy index of 1.25 ± 0.10, while fitting down to the lowest lim- its on the spectrum yields 1.48+0.09 −0.08. We conclude that there is a weak soft excess present. We can model the soft excess with either a blackbody or broken power law. The fit parameters between 0.5 and 5 keV are given in Table 3. The soft X-ray spectral slope αX,soft=2.03+0.23 −0.20 is in good agreement αX,soft=1.93+0.58 −0.42 found during the 2005 December 09 observation by Swift. 3.2. UVOT Photometry Table 4 summarizes the results of the analysis of the photometry of the co-added UVOT images. During segment 008, no observations of RX J0148.3 -- 2758 were made in the B filter. Figure 7 displays the UVOT light curves of all 6 filters plus the XRT light curve from seg- ments 008 to 011. The figure might suggest that there is some variability in the UV. However a comparison with 4 field stars, as listed in Table 5, shows that the variation seen in the UVOT light curves are still within the er- ror margins. Figure 8 displays the UVOT measurements of these comparison stars. This figure shows that the trends seen in the RX J0148.3 -- 2758 UVOT light curves are also present in the light curves of the comparison stars. Therefore we consider RX J0148.3 -- 2758 not to be variable in the UV/optical band during 2005 December observations. Figure 9 displays the UVOT V image of the field around RX J0148.3 -- 2758 with the 4 compari- son stars marked. In part, the variations seen in the UV light curves of RX J0148.3 -- 2758 are due to the relatively small extraction radius of 4.5 and the variable PSF of the UVOT. However, the lack of photometric data during the 2005 May observations provide us with no knowledge on the UV flux/magnitudes during a high-state. ′′ 3.3. Spectral Energy Distribution Figure 10 displays the Spectral Energy Distribution (SED) of RX J0148.3 -- 2758. The Swift XRT data taken in 2005 May are shown as triangles and the 2005 De- cember 20/21 observations are represented by diagonal crosses. For the UVOT data, only the 2005 December data are shown. This AGN was not detected in the NVSS or the FIRST radio catalogues. The Far-Infrared IRAS and NIR 2MASS luminosities were derived with the GATOR catalogue search engine at NASA/IPAC (irsa.ipac.caltech.edu/applications/Gator/). The IRAS luminosities deviate slightly from those given in Grupe et al. (1998a) due to the improved extraction software at IPAC. We measured the optical-to-X-ray spectral slopes αox of the 2005 December 09 and 20/21 observations from the SED plot Figure 10. During the December 09 obser- vation we found a rest-frame αox=1.53. At a luminos- ity density log lo=22.78 [W Hz−1] and redshift z=0.121 this value is in good agreement with the mean of radio- quiet AGN ROSAT sample of Yuan et al. (1998a,b) and Strateva et al. (2005) for the same redshift and lumi- nosity intervals. Following the relation given in equation (4) in Strateva et al. (2005), we would expect αox=1.42. However, during the December 20/21 observation the source became more X-ray quiet with an αox=1.81. We are yet to analyze the grism data taken during May 2005, during which the source was in a high state. We therefore have no current measure of αox during this time period. Although the 2005 December UVOT observations show no significant variability, the value of αox during the ob- servations of 2005 May cannot be determined because we do not know what the flux in the UV filters was during this time period. 8 This NLS1 has been observed once before in the UV, in 1992 by IUE (SWP 45107). The spectrum is dis- played in the left panel of Figure 11. The right panel of Figure 11 shows the optical spectrum of RX J0148.3 -- 2758 taken in September 1995 at the ESO 1.52m tele- scope in La Silla for a total of 4 hours. Details of this observing run are given in Grupe et al. (2004a). 8 The X-ray loudness is defined by Tananbaum et al. (1979) as αox= -- 0.384 log(f2keV /f 2500A ). 6 Grupe et al. With a FWHM(Hβ)=1030±100 km s−1 we derived a central black hole mass of 1.3×107M⊙using equation (5) in Vestergaard & Peterson (2006). From the IUE spec- trum shown in Figure 11 we derived a FWHM(CIV) = 2300 km s−1. By using the relation given in equation (7) in Vestergaard & Peterson (2006) we estimated the black hole mass MBH = 3.4 × 107M⊙. Both black hole masses estimates agree with each other within their un- certainties. Estimated from these black hole masses, the Ed- dington luminosity is 1.6-4.3×1038 W. As described in Grupe et al. (2004a) we modeled the BBB by a power- law with exponential cutoff plus an absorbed power law. This model is displayed in Figure 10. From the 2005 De- cember 20/21 data we measured a bolometric luminosity Lbol = 5×1038 W which is similar to the value given by Grupe et al. (2004a) based on the optical spectrum and the ROSAT RASS data. This results in an Eddington ration L/LEdd = 1-3. The [OIII] lines can be separated into a narrow and a blueshifted broad component. The broad [OIII] lines are blueshifted by 600±200 km s−1 with respect to the narrow [OIII] and Hβ lines. Similar results on NLS1s have been previously reported by e.g. Grupe & Leighly (2002); Zamanov et al. (2005) and Bian et al. (2005). (2002); Aoki et al. 4. DISCUSSION In this paper we have presented the Swift observations In addi- of the high variable NLS1 RX J0148.3 -- 2758. tion to the strong X-ray flux variability, our main results are composed of the spectral changes. We observed a hardening followed by a softening of the spectrum of RX J0148.3 -- 2758 over a time during which the X-ray flux was on a continual decrease. Both types of spectral changes have been observed in AGN, although the hard- ening of the X-ray spectrum with decreasing flux is more common (e.g. Gallo et al. 2004a; Dewangan et al. 2002; Lee et al. 2001; Chiang et al. 2000). However, softening of the X-ray spectrum with decreasing flux has been reported on NLS1s in e.g., RX J2217.9 -- 5941 (Grupe et al. 2004b), RX J0134.2 -- 4258 (Grupe et al. 2000; Komossa & Meerschweinchen 2000), PKS 0558 -- 504 (Gliozzi et al. 2001), and 1H 0707 -- 495 (Gallo et al. 2004b; Fabian et al. 2004). A simple way to explain a hardening in the spectrum with decreasing observed X-ray flux is with a cold ab- sorber cloud in the line of sight. Variable absorbers in AGN are often observed in Seyfert galaxies, e.g. the Seyfert 2 sample of Risaliti et al. (2002), NGC 1365 (Risaliti et al. 2004), NGC 4388 (Elvis et al. 2004), the Seyfert 1.8 galaxy NGC 3786 (Komossa & Fink 1997b) the Seyfert 1.5 galaxies NGC 4151 (Puccetti et al. 2004) and NGC 3227 (Komossa & Fink 1997a), or 1H0419 -- 577 (Pounds et al. 2004). A variable cold absorber could also, in part, provide a plausible explanation for the spectral variability between the 2005 May and 2005 December 09 observation. As listed in Table 2, we fit- ted an absorbed broken power law to the 2005 December 09 data, where all parameters were fixed to those deter- mined from the May 2005 observations, except for the absorption column density and the normalization, which were allowed to vary. This provided a best-fit column density NH=8.1±1.4×1020 cm−2, although there were strong residuals below 0.5 keV. Although NLS1s often resemble AGN with only minor intrinsic absorption, this is not a true picture in general (e.g. Grupe et al. 1998b, 2004c). A different result is found if all parameters are left to vary. As listed in Table 2 and shown in Figure 6, NH actually becomes consistent with the Galactic value and the soft X-ray spectral index αX,soft flattens out. However, in a fitting routine like XSPEC, NH and the spectral index are not independent parameters. A larger value of the absorption column density NH will result in a steeper spectral index, and vice-versa. Given the column density observed in December 20/21, which was again in the order of 1021 cm−2, we can conclude that the spectral change seen between the 2005 May and 2005 De- cember 09 observations is most likely due to an increase in the absorber column density. A soft X-ray spectrum fitted by a spectral model can mimic a low column den- sity as shown by e.g. Puchnarewicz et al. (1995) and Grupe et al. (1998a), even though the real column den- sity is much larger. Grupe et al. A softening with decreasing X-ray flux can be cause by several processes such as a change in the accretion disk corona (e.g. 2000) or the presence of a variable ionized absorber (e.g. Komossa & Meerschweinchen 2000). Another possibil- ity is the presence of a partial covering absorber as dis- cussed for e.g. RX J2217.9 -- 5941 (Grupe et al. 2004b), 1H 0707 -- 495 (Gallo et al. 2004b; Tanaka et al. 2004), and Mkn 1239 (Grupe et al. 2004c). As an alternative, Fabian et al. (2004) discussed the variability observed in 1H 0707 -- 495 in the context of X-ray reflection on an ionized disk. Of all these models only the partial cov- ering absorber model yields reasonable results. Interest- ingly, the coverage fraction observed in 2005 May, 2005 December 09, and December 20/21 observation remains the same, at around f =0.8. The column density of the partial covering absorber follows the same trend as the cold absorber column density, suggesting that it is at a low value during the December 09 observation. The soft X-ray slope αX,soft=2.58+0.15 −0.12 is rather steep even for a NLS1. The mean soft X-ray slope for the sam- ple of 51 NLS1s from Grupe et al. (2004a) is αX=1.96 with a standard deviation σ=0.41, and αX=2.1 for the sample taken from Boller et al. (1996). However, the hard X-ray spectral slope of αX=0.96+0.16 −0.12 is slightly flatter than that found in the sample of NLS1s from Leighly (1999b), who found a mean hard X-ray slope of αX=1.19±0.10 and the sample from Brandt et al. (1997) with αX=1.15. This is in better agreement with the values found for BLS1, for which Leighly (1999b) found αX=0.78±0.11 and Brandt et al. (1997) found αX=0.87. A possible explanation for this 'discrepancy' is that whereas the soft X-ray spectral slope is driven by the Eddington ratio L/LEdd, the hard X-ray spectral slope is more dependent on the black hole mass. The Eddington ratio L/LEdd is one of the highest in the sam- ple from Grupe et al. (2004a), with L/LEdd = 4. As shown by Grupe & Mathur (2004); Mathur & Grupe (2005a,b), NLS1s with a high Eddington ratio L/LEdd deviate significantly from the MBH - stellar velocity dispersion σ relation (e.g. 2000a; Ferrarese & Merritt 2000; Tremaine et al. 2003). With a FWHM([OIII])=700±500 and a black hole mass in the Gebhardt et al. Swift observation of RX J0148.3 -- 2758 7 order of a few 107M⊙ RX J0148.3 -- 2758 shows one of the most extreme deviations from the Tremaine et al. (2003) MBH − σ relation. RX J0148.3 -- 2758 also shows a variation in its optical- to-X-ray spectral slope αox. While during the 2005 De- cember 09 observations αox=1.5, during the December 20/21 observations RX J0148.3 -- 2758 became X-ray weak with αox=1.81. Similar changes in αox have also been recently reported by Gallo (2006). Changes like these can explain in part the large scatter seen in the αox dia- grams of e.g. Yuan et al. (1998a,b) and Strateva et al. (2005). The Swift observations of RX J0148.3 -- 2758 have shown the great potential of Swift for AGN science. The X-ray light curves are highly variable, in particular those of NLS1s, and require long-term coverage in a range of wavelengths. Due to its low-earth orbit, Swift is very similar to ROSAT and ASCA, but has also the added advantage of being a multi-wavelength observatory. Our study has shown the importance of simultaneous UV and X-ray observations over a time-span of days, and Swift is the only observatory that can obtain such observa- tions. Our observations of RX J0148.3 -- 2758 utilize the multi-wavelength capabilities of Swift as well as its flexi- ble observing scheduling. The simultaneous observations in the UVOT and XRT allow us to measure the X-ray loudness αox directly without assuming any optical/UV spectral slopes. We are also able to measure the total power in the Big Blue Bump and therefore the bolomet- ric luminosity directly. The strong change in its spectrum between the 2005 May and 2005 December 09 observa- tions prompted us to execute further observations, which took place a few days later, on December 20/21. These additional observations allowed us to observe a hardening and a softening in the same source. Based on this inter- esting spectral behavior, we plan to continue observing RX J0148.3 -- 2758 with Swift. We would like to thank the whole Swift-team for mak- ing this observation possible, especially the Swift sci- ence planners Jamie Kennea, Sally Hunsberger, Clau- dio Pagani, Judy Racusin and Antonino Cucchiara for scheduling RX J0148.3 -- 2758 for such a long observing time, and Neil Gehrels for approving the ToO observa- tions of 2005-December 20/21. We would also like to thank Marco Ajello and Jochen Greiner (MPE) and Jack Tueller (GSFC) for checking the BAT pointed and sur- vey data for any detection of RX J0148.3 -- 2758. We also thank the anonymous referee for valuable comments and suggestions to improve this paper. This research has made use of the NASA/IPAC Extra-galactic Database (NED) which is operated by the Jet Propulsion Labo- ratory, Caltech, under contract with the National Aero- nautics and Space Administration. This research was supported by NASA contract NAS5-00136 (D.G., D.B., & J.N.). REFERENCES Aoki, K., Kawaguchi, T., & Ohta, K., 2005, ApJ, 618, 601 Arnaud, K. A., 1996, ASP Conf. Ser. 101: Astronomical Data Grupe, D., Wills, B.J., Wills, D., Beuermann, K., 1998b, A&A, 333, 827 Analysis Software and Systems V, 101, 17 Grupe, D., Beuermann, K., Mannheim, K., & Thomas, H.-C., 1999, Barthelmy, S.D., 2005, Space Science Reviews, 120, 143 Belloni, T., Hasinger, G., & Izzo, C., 1994, A&A, 283, 1037 Beuermann, K., Thomas, H.-C., Reinsch, K., et al., 1999, A&A, 347, 47 Bian, W., Yuan, Q., & Zhao, Y., 2005, MNRAS, 364, 187 Boller, T., Brandt, W.N., & Fink, H.H., 1996, A&A, 305, 53 Boroson, T.A., & Green, R.F., 1992, ApJS, 80, 109 Boroson, T.A., 2002, ApJ, 565, 78 Brandt, W.N., Mathur, S., & Elvis, M., 1997, MNRAS, 285, L25 Brandt, W.N., Laor, A., & Wills, B.J., 2000, ApJ, 528, 637 Burrows, D.N., et al., 2005, Space Science Reviews, 120, 165 Burrows, D.N., et al., 2006, ApJ, submitted, astro-ph/0604320 Chiang, J., Reynolds, C.S., Blaes, O.M., Nowak, M.A., Murray, N., Madejeski, G., Marshall, H.L., & Magdziarz, P., 2000, ApJ, 292 Dewangan, G.C., Boller, Th., Singh, K.P., & Leighly, K.M., 2002, A&A, 390, 65 Dickey, J.M., & Lockman, F.J., 1990, ARA&A, 28, 215 Elvis, M., Risaliti, G., Nicastro, F., Miller, J., Fiore, F., & Puccetti, S., 2004, ApJ, 615, L25 Fabian, A.C., Miniutti, G., Gallo, L.C., Boller, Th, Tanaka, Y., Vaughan, S., & Ross, R., 2004, MNRAS, 353, 1071 Ferrarese, L., & Merritt, D., 2000, ApJ, 539, L9 Gallo, L.C., Boller, Th., Brandt, W.N., Fabian, A.C., & Grupe, D., A&A, 350, 805 Grupe, D., Leighly, K.M., Thomas, H.-C., & Laurent-Muehleisen, S.A., 2000, A&A, 356, 11 Grupe, D., Thomas, H.-C., & Beuermann, K., 2001a, A&A, 367, 470 Grupe, D., & Leighly, K.M., 2002, MPE report 279, p287 Grupe, D., Wills, B.J., Leighly, K.M., & Meusinger, H., 2004a, AJ, 127, 156 Grupe, D., Leighly, K.M., Burwitz, V., Predehl, P., & Mathur, S., 2004b, AJ, 128, 1524 Grupe, D., Mathur, S., & Komossa, S., 2004c, AJ, 127, 3161 Grupe, D., & Mathur, S., 2004, ApJ, 606, L41 Hill, J.E., et al., 2004, SPIE, 5165, 217 Hogg, D., 1999, astro-ph/9905116 Komossa, S., & Fink, H.H., 1997a, A&A, 327, 483 Komossa, S., & Fink, H.H., 1997b, A&A, 327, 555 Komossa S., & Fink H., 1998, in: Highlights in X-ray astronomy, B. Aschenbach & M.J. Freyberg (eds.), MPE Report 272, 147 Komossa, S., & Meerschweinchen, J., 2000, A&A, 354, 411 Laor, A., Fiore, F., Elvis, M., Wilkes, B.J., & McDowell, J.C., 1994, ApJ, 435, 611 Laor, A., Fiore, F., Elvis, M., Wilkes, B.J., & McDowell, J.C., 1997, ApJ, 477, 93 2004a, MNRAS, 352, 744 Lee, J.C., Fabian, A.C., Reynolds, C.S., Brandt, W.N., & Iwasawa, Gallo, L.C., Tanaka, Y., Boller, Th., Fabian, A.C., Vaughan, S., & K., 2001, MNRAS, 318, 857 Brandt, W.N., 2004b, MNRAS, 353, 1064 Gallo, L.C., 2006, MNRAS accepted, astro-ph/0602145 Gebhardt, K., Bender, R., Bower, G., et al., 2000, ApJ, 539, L13 Gehrels, N., et al., 2004, ApJ, 611, 1005 Gliozzi, M., Brinkmann, W., O'Brien, P.T., Reeves, J.N., Pounds, K.A., Trioglio, M., & Gianotti, F., 2001, A&A, 365, L128 Goodrich, R.W., 1989, ApJ, 342, 224 Grupe, D., 1996, PhD Thesis, Universitat Gottingen Grupe, D., 2004, AJ, 127, 1799 Grupe, D., Beuermann, K., Thomas, H.-C., Mannheim, K., & Fink, H.H., 1998a, A&A 330, 25 Leighly, K.M., 1999a, ApJS, 125, 297 Leighly, K.M., 1999b, ApJS, 125, 317 Markwardt, C.B., Tueller, J., Skinner, G.K., Gehrels, N., Barthelmy, S.D., & Mushotzky, R.F., 2006, ApJ, submitted, astro-ph/0509860 Mason, K.O., et al., 2001, A&A, 365, L36 Mathur, S., 2000, MNRAS, 314, L17 Mathur, S., & Grupe, D., 2005a, A&A, 432, 463 Mathur, S., & Grupe, D., 2005b, ApJ, 633, 688 Nandra, K., George, I.M., Mushotzky, R.F., Turner, T.J., & Yaqoob, T., 1997, ApJ, 476, 70 Nousek, J.A., et al., 2006, ApJ, 642, 389 8 Grupe et al. Osterbrock, D.E., & Pogge, R.W., 1985, ApJ, 297, 166 Parsons, A., et al. 2005, GCN 4363 Pfeffermann, E., Briel, U.G., Hippmann, H., et al., 1987, SPIE, Tremaine, S., Gebhardt, K., Bender, R., et al., 2003, ApJ, 574, 740 Trumper, J., 1982, Adv. Space Res., 4, 241 Turner, M.J.L., Abbey, A., Arnaud, M., et al., 2001, A&A, 365, 733, 519 L27 Pounds, K.A., Done, C., & Osborne, J.P., 1995, MNRAS, 277, L5 Pounds, K.A., Reeves, J.N., Page, K.L., & O'Brien, P.T., 2004, Turner, T.J., George, I.M., Nandra, K., & Turcan, D., 1999, ApJ, 524, 667 ApJ, 616, 696 Vaughan, S., Reeves, J., Warwick, R., & Edelson, R., 1999, Puccetti, S., Risaliti, G., Fiore, F., Elvis, M., Nicastro, F., Perola, MNRAS, 309, 113 G.C., & Capalbi, M., 2004, Nucl. Phys. B Suppl., 132, 225 Vaughan, S., Edelson, R., Warwick, R.S., Malkan, M,A., & Goad, Puchnarewicz, E.M., et al., 1992, MNRAS, 256, 589 Puchnarewicz, E.M., Mason, K.O., Siemiginowska, A., & Pounds, K.A., 1995, MNRAS, 276, 20 Risaliti, G., Elvis, M., & Nicastro, F., 2002, ApJ, 571, 234 Risaliti, G., Elvis, M., Fabbiano, G., Baldi, A., & Zezas, A., 2004, ApJ, 623, L93 Rodr´iguez-Pascual, P.M., Mas-Hesse, J.M., & Santos-Lle´o, M., 1997, A&A, 327 Roming, P.W.A., et al., 2005, Space Science Reviews, 120, 95 Schwope, A.D., Hasinger, G., Lehmann, I., et al., 2000, AN, 321, 1 Shields, G.A., 1978, Nature, 272, 706 Strateva, I.V., Brandt, W.N., Schneider, D.P., Vanden Berk, D.G., & Vignali, C., 2005, ApJ, 130. 387 Sulentic, J.W., Zwitter, T., Marziani, P., & Dultzin-Hacyan, D., 2000, ApJ, 536, L5 Tanaka, Y., Boller, Th., Gallo, L.C., Keil, R., & Ueda, Y., 2004, PASJ, 56, L9 Tananbaum, H., et al., 1979, ApJ, 234, L9 Thomas, H.-C., Beuermann, K., Reinsch, K., et al., 1998, A&A, 335, 467 M.R., 2001, MNRAS, 327, 673 Vestergaard, M., & Peterson, B.M., 2006, ApJ, 641, 689 Voges, W., Aschenbach, B., Boller, T., et al., 1999, A&A, 349, 389 Walter, R., & Fink, H.H., 1993, A&A, 274, 105 Williams, R.J., Pogge, R.W., & Mathur, S., 2002, AJ, 124, 3042 Williams, R.J., Pogge, R.W., & Mathur, S., 2004, ApJ, 610, 737 Yuan, W., Brinkmann, W., Siebert, J., Voges, W., 1998a, A&A, 330, 108 Yuan, W., Siebert, J., & Brinkmann, W., 1998b, A&A, 334, 498 Yuan, M.J., & Wills, B.J., 2003, ApJ, 593, L11 Zamanov, R., Marziani, P., Sulentic, J.W., Galvani, M., Dultzin- Hacyan, D., & Bachev, R., 2002, ApJ, 576, L9 Zimmermann, 1998, U., 'EXSAS al., (http://wave.xray.mpe.mpg.de/exsas/users-guide) Guide', MPE User's Boese, G., Becker, W., et report Swift observation of RX J0148.3 -- 2758 9 Fig. 1. -- Swift XRT 0.3-10.0 keV light curves The left panel shows the 2005 May observation; the middle panel shows the December 07 and 09 observations, and the right one shows the December 20 and 21 observations. Start times are 2005-May-06 00:05 UT, 2005- December-07 00:34 UT, and 2005-December-20 13:40 UT, for the left, middle and right panels, respectively. The start and end times of each segment are given in Table 1. The upper panel displays the count rate light curve and the lower panel the light curve of the hardness ratio = (H-S)/(H+S) with S and H are the number of photons in in the 0.3-1.0 and 1.0-10.0 keV band, respectively. Fig. 2. -- ROSAT All-Sky Survey and pointed observation light curves. The start times are 1990-07-15 15:26 UT, 1990-12-28 01:01 UT, and 1991-01-15 09:24 UT for the RASS observations (left panels), and 1992-07-09 09:54 for the pointed ROSAT PSPC observation (right panel). 10 Grupe et al. Fig. 3. -- ASCA SIS0 + SIS1 0.5 -- 10 keV light curve. The start time was 1997-July-1 21:05 UT. Fig. 4. -- Long-term light curve of RX J0148.3 -- 2758. The luminosities are rest-frame 0.2-2.0 keV and are determined from unabsorbed fluxes based on the best-fit models as given in Tables 2 and 3. Swift observation of RX J0148.3 -- 2758 11 0.1 0.01 10−3 2.5 2 1.5 1 0.5 1 1 − V e k 1 − s s t n u o C d e z i l a m r o N o i t a R 0.1 0.01 10−3 8 6 4 2 1 1 − V e k 1 − s s t n u o C d e z i l a m r o N o i t a R 0.5 1 2 5 0.5 Energy [keV] 1 Energy [keV] 2 5 Fig. 5. -- Swift XRT spectra of RX J0148.3 -- 2758. The 2005 May spectra (left panel) were fitted by a single power low with the absorption column fixed to the Galactic value. For the spectra of the 2005 December observations (right panel), the X-ray spectral slope was also fixed to αX=2.4 (see text). The colors represent the spectra of different segments: in the left panel: segment 002 = black, segment 003 = red, segment 004 = green, and segment 006 = blue; right panel: segment 008 = black, segment 009 = red, segment 010 = green, and segment 011 = blue. 1 − V e k 1 − s s t n u o C d e z i l a m r o N 1 . 0 1 0 . 0 a m m a G x e d n I t n o o h P 5 5 5 4 4 4 3 3 3 + 0.5 1 2 5 0 0 0 0.05 0.05 0.05 Energy [keV] 0.1 0.1 0.1 0.15 0.15 0.15 NH [1022 cm−2] 0.2 0.2 0.2 0.25 0.25 0.25 Fig. 6. -- Comparison between the spectra of the 2005 May, 2005 December 09, and December 20/21 observations. The left panel displays the spectra with the average of the 2005 May spectra (segment 002-006) = black, December 09 (segment 009) = red, and December 20/21 (segment 010/011) = green. The right panel shows the contour plot between the intrinsic column density NH,intr and the soft X-ray photon index Γ = αX,soft+1 of the XRT spectra. The colors of the contours refer to the same segments as in the left panel. 12 Grupe et al. Fig. 7. -- XRT and UVOT light curves. Swift observation of RX J0148.3 -- 2758 13 Fig. 8. -- UVOT light curves of the comparison stars S1-S4 as given in Table 5 and displayed in Figure 9. 14 Grupe et al. N E S2 S1 RX J0148-27 S4 S3 Fig. 9. -- UVOT V-image of the field around RX J0148.3 -- 2758. The 4 comparison stars as listed in Table 5 are marked as S1 - S4 in the figure. Fig. 10. -- Spectral Energy Distribution of RX J0148.3 -- 2758. The luminosities are the observed luminosities as given in Table 6. The crosses of the Swift UVOT and XRT data represent the 2005 December 20/21 observations and the triangles the XRT observations from 2005 May. The dotted line displays the power law plus exponential cutoff and absorbed power law model to describe the BBB. Swift observation of RX J0148.3 -- 2758 15 Fig. 11. -- IUE and optical spectra of RX J0148.3 -- 2758 16 Grupe et al. Observation log of RX J0148.3 -- 2758 TABLE 1 Observatory observation T-start1 T-stop1 Texp 2 log LX 3 Swift ASCA ROSAT ROSAT segment 0024 2005-05-06 00:05 segment 0034 2005-05-07 00:05 segment 0044 2005-05-11 00:27 segment 0064 2005-05-13 00:45 segment 0084 2005-12-07 00:34 segment 0094 2005-12-09 12:17 segment 0104 2005-12-20 13:40 segment 0114 2005-12-21 01:01 1997-07-11 21:11 SIS 0, 1 GIS 2, 3 1997-07-11 21:11 pointed PSPC 1992-07-09 09:54 1990-07-15 15:26 RASS 1990-12-28 01:11 1991-01-15 09:24 2005-05-06 24:00 2005-05-09 22:58 2005-05-11 23:11 2005-05-13 10:26 2005-12-07 10:35 2005-12-09 23:54 2005-12-20 23:32 2005-12-21 14:00 1997-07-13 03:07 1997-07-13 03:07 1992-07-10 01:16 1990-07-16 07:28 1990-12-29 07:37 1991-01-17 07:49 7637 21966 8840 3132 822 6346 5110 3506 33243 36381 6652 89 140 274 37.855 37.83 37.27 37.26 37.21 36.92 37.79 37.87 38.05 38.00 1Start and End times are given in UT 2Observing time given in s 3Rest-frame 0.2-2.0 keV luminosity given in units of W 4The term 'segment' is from the way Swift is scheduled. Swift is scheduled on a day by day basis. A segment contains the observations of a source of a single day (except during the weekends when Swift is scheduled for 3 days) 5Average luminosity of segments 002-006 Swift observation of RX J0148.3 -- 2758 17 Spectral parameters of the fits to the Swift-XRT spectra of RX J0148.3 -- 2758 TABLE 2 Obs. Date Model1 NH,Gal 1020 cm−2 NH,intr 1020 cm−2 May 06 May 07 May 11 May 13 May 06-133 December 07 December 09 December 07+093 December 20 December 21 December 20+213 1 1 1 2 3 4 1 1 1 2 3 4 1 1 2 3 4 1 1 2 3 3 4 1 1 1 2 3 4 1 2 3 1 2 3 3 4 1 2 3 4 1 4 1 1 2 4 1 1 2 3 4 1.50 (fix) 3.33+0.14 −0.12 1.50 (fix) 1.50 (fix) 4.80±1.16 1.50 (fix) 1.50 (fix) 2.77±0.05 1.50 (fix0) 1.50 (fix) 4.85+0.93 −0.67 1.50 (fix) 1.50 (fix) 3.88±0.09 1.50 (fix) 6.18+1.81 −1.56 1.50 (fix) 1.50 (fix) 1.60+0.17 −0.15 1.50 (fix) 1.50 (fix) 3.01+1.80 −1.60 1.50 (fix) 1.50 (fix) 2.92+0.55 −0.53 1.50 (fix) 1.50 (fix) 4.96+0.68 −0.46 1.50 (fix) 1.50 (fix) 1.50 (fix) 1.50 (fix) 1.50 (fix) 1.50 (fix) 1.50 (fix) 8.14+1.49 −1.36 1.50 (fix) 1.50 (fix) 1.50 (fix) 2.27+3.85 −1.80 1.50 (fix) 1.50 (fix) 1.50 (fix) 1.50 (fix) 4.44+3.25 −2.67 1.50 (fix) 1.50 (fix) 1.50 (fix) 3.30+2.42 −2.10 1.50 (fix) 7.80+4.31 −3.60 1.50 (fix) -- -- 3.01+0.22 −0.19 -- -- 7.95+0.25 −0.34 -- -- 2.34+1.17 −1.09 -- -- 7.61+0.20 −0.17 -- -- -- -- 7.88+0.36 −0.41 -- -- -- -- -- 3.42+0.33 −0.11 -- -- 2.42+0.86 −0.82 -- -- 7.15+1.48 −1.21 -- -- -- -- -- -- -- 2.92+6.07 −2.92 -- -- -- 6.93+7.91 −5.96 -- 2.86+6.80 −2.80 -- -- -- 12.42+9.91 −6.72 -- -- -- -- 11.61+7.49 −5.73 αX,soft 2 1.84±0.07 2.03+0.16 −0.15 2.08+0.19 −0.17 -- 2.28±0.14 2.59+0.21 −0.29 1.82±0.03 1.96±0.05 2.02+0.11 −0.10 -- 2.30+0.12 −0.11 2.61+0.19 −0.16 1.93±0.05 2.13±0.10 kT eV -- -- -- 126+8 −11 -- -- -- -- -- 118+5 −7 -- -- -- -- -- 120±7 2.53+0.25 −0.20 2.75+0.32 −0.41 1.83±0.10 1.84x+0.20 −0.18 -- 1.93±0.08 2.10+0.21 −0.19 2.24±0.16 1.84±0.03 2.00±0.07 2.05±0.08 -- 2.33±0.08 2.58+0.15 −0.12 2.17+0.27 −0.25 -- 2.26±0.32 1.48+0.10 −0.09 -- 1.60+0.38 −0.15 2.33 (fix) 1.93+0.58 −0.42 1.57±0.05 -- 1.73+0.64 −0.23 2.45+0.79 −0.73 2.07+0.21 −0.20 2.47+0.62 −0.42 1.96+0.16 −0.15 2.33+0.43 −0.37 -- 3.38+1.00 −0.73 2.00+0.13 −0.12 2.22+0.31 −0.21 -- 3.05+0.70 −0.55 3.41+0.78 −0.64 -- -- -- -- 108+108 −102 -- -- -- -- -- -- 119±4 -- -- -- 104±16 -- -- 100±20 -- -- -- -- 103+23 −19 -- -- -- -- -- -- 110±14 -- -- -- 101±13 -- -- Ebreak KeV -- -- -- -- 1.87±0.25 1.80+0.16 −0.19 -- -- -- -- 1.72+0.15 −0.16 1.61±0.14 -- -- -- 1.75+0.25 −0.30 1.70+0.33 −0.18 -- -- -- 1.70 (fix) 1.42±0.50 1.40+1.02 −3.24 -- -- -- -- 1.76±0.11 1.68+0.12 −0.14 -- -- 1.07±1.39 -- -- 1.28+4.30 −0.85 1.76 (fix) 1.16+0.30 −0.19 -- -- 2.01+1.88 −0.65 1.08+1.10 −0.15 -- 1.68 (fix) -- -- -- 1.30+0.43 −0.26 -- -- -- 1.20+0.49 −0.20 1.17+0.34 −0.17 αX,hard χ2/ν -- -- -- 1.27+0.30 −0.21 0.95±0.24 0.85+0.35 −0.23 -- -- -- 1.23+0.14 −0.13 1.01+0.18 −0.17 1.00+0.19 −0.13 -- -- 0.87+0.14 −0.23 0.69+0.42 −0.36 0.68+0.33 −0.32 -- -- 1.64+0.25 −0.26 1.37±0.23 1.47±0.30 1.47+1.90 −3.11 -- -- -- 1.23+0.04 −0.06 0.96±0.13 0.96+0.16 −0.14 -- 1.36+1.21 −1.36 1.86±0.80 -- 1.29+0.22 −0.11 1.26+0.29 −5.26 0.96 (fix) 1.27+0.27 −0.15 -- 1.23+0.23 −0.12 1.29+0.41 −0.71 1.34+0.23 −0.61 -- 0.96 (fix) -- -- 0.92+0.37 −0.56 0.98+0.73 −0.95 -- -- 1.13+0.37 −0.39 1.28+0.44 −0.73 1.32+0.45 −0.52 113/68 107/67 106/67 83/65 83/65 71/64 268/130 258/129 254/129 179/127 174/126 146/126 155/67 145/66 84/65 98/63 92/63 27/36 27/35 25/34 23/34 24/33 23/32 567/304 546/303 541/303 386/299 386/300 347/301 12/7 10/6 11/6 30/37 27/35 29/34 48/37 25/34 60/47 38/43 54/43 41/42 22/12 18/11 29/17 26/16 16/15 12/14 52/30 57/29 33/27 32/27 30/27 1Model fit to the data: 1) Single power law with Galactic absorption; 2) Blackbody plus power law with Galactic absorption; 3) Broken power law with Galactic absorption; 4) Broken power law with Galactic absorption and intrinsic absorption at z=0.121 2This spectral slope also refers to the 0.3-10.0 keV slope in case only a single power law has been used. 3Simultaneous fits in XSPEC 18 Grupe et al. TABLE 3 Spectral parameters of the fits to the ROSAT and ASCA spectra of RX J0148.3 -- 2758 Mission observation Model1 NH,Gal 2 NH,intr 2 αX,soft 3 ASCA ROSAT RASS ROSAT po 1 2 3 1 1 1 1 1.50 (fix) 1.50 (fix) 1.50 (fix) 1.50 (fix) 2.54±0.82 1.50 (fix) 2.35±0.22 -- -- -- -- -- -- -- 1.48+0.09 −0.08 -- 2.03+0.23 −0.20 2.12±0.11 2.62±0.30 1.88±0.03 2.25±0.08 kT4 -- 127±16 -- -- -- -- -- Ebreak 5 αX,hard χ2/ν -- -- 1.36+0.16 −0.19 -- -- -- -- -- 1.10+0.13 −0.14 1.11+0.16 −0.19 -- -- -- -- 254/237 208/235 210/235 30/23 21/22 178/110 111/110 1Model fit to the data: 1) Single power law with Galactic absorption; 2) Blackbody plus power law with Galactic absorption; 3) Broken power law with Galactic absorption; 4) Broken power law with Galactic absorption and intrinsic absorption at z=0.121 2Column density NH given in units of 1020 cm−2 3This spectral slope also refers to the 0.3-10.0 keV slope in case only a single power law has been used. 4kT in units of eV 5Broken Power law break energy Ebreak in units of keV UVOT photometry from the co-added images of RX J0148.3 -- 2758 TABLE 4 Filter V B U UVW1 UVM2 UVW2 Segment 008 Mag Flux1 Segment 009 Mag Flux1 Segment 010 Mag Flux1 Segment 011 Mag Flux1 15.358±0.062 26.90±1.18 -- 14.625±0.042 14.531±0.031 14.701±0.031 14.598±0.018 -- 47.19±1.62 64.62±1.52 74.77±1.71 112.15±1.46 15.350±0.018 15.645±0.015 14.482±0.011 14.485±0.009 14.602±0.010 14.537±0.007 27.10±0.35 33.45±0.41 53.80±0.49 67.39±0.49 81.85±0.57 118.60±0.59 15.321±0.020 15.647±0.017 14.480±0.013 14.471±0.011 14.603±0.011 14.530±0.008 27.84±0.39 33.37±0.39 53.91±0.55 68.29±0.56 81.78±0.64 119.39±0.66 15.299±0.017 15.652±0.088 14.523±0.030 14.425±0.024 14.647±0.024 14.533±0.006 28.41±0.34 33.24±2.40 51.81±1.27 71.25±1.32 78.57±1.37 119.11±0.56 1The fluxes are given in units of 10−19 W m−2 A−1. UVOT photometry from the co-added images of the four comparison stars The light curves of these stars is shown in Figure 8. TABLE 5 Star segment V 1 2 3 4 008 009 010 011 008 009 010 011 008 009 010 011 008 009 010 011 13.102±0.036 13.072±0.011 13.074±0.012 13.074±0.010 13.79±0.037 13.81±0.011 14.02±0.013 14.00±0.011 15.09±0.057 15.09±0.016 15.05±0.018 15.08±0.016 16.77±0.162 16.70±0.038 16.69±0.043 16.68±0.036 B -- 13.638±0.018 13.634±0.020 13.676±0.010 -- 14.38±0.013 14.62±0.014 -- -- 15.66±0.015 15.63±0.017 15.74±0.092 -- 17.02±0.028 17.02±0.031 17.16±0.174 U UVW1 UVM2 UVW2 13.695±0.038 13.600±0.010 13.599±0.012 13.600±0.028 14.29±0.040 14.20±0.011 14.42±0.012 -- 15.57±0.061 15.50±0.015 15.55±0.017 15.54±0.042 16.89±0.126 16.84±0.023 16.85±0.033 16.74±0.073 15.063±0.041 15.058±0.012 15.046±0.013 15.026±0.031 15.56±0.056 15.60±0.016 15.78±0.019 -- 16.70±0.114 16.87±0.032 16.84±0.035 16.80±0.080 17.45±0.196 17.52±0.047 17.53±0.053 17.45±0.116 16.842±0.087 16.667±0.024 16.670±0.027 16.672±0.061 16.96±0.100 16.96±0.029 17.17±0.036 -- 18.90±0.382 18.14±0.060 18.20±0.069 18.27±0.160 18.34±0.249 18.09±0.056 18.12±0.062 18.02±0.131 16.804±0.050 16.729±0.017 16.728±0.020 16.707±0.016 17.20±0.063 17.20±0.023 17.40±0.028 17.36±0.024 18.41±0.157 18.50±0.053 18.50±0.061 18.41±0.048 18.13±0.120 18.38±0.048 18.40±0.053 18.33±0.044 Swift observation of RX J0148.3 -- 2758 19 Measurements of the spectral energy distribution shown in Figure 10. TABLE 6 Observatory Filter 1 λc log ν [Hz] Magnitude2 NVSS IRAS 2MASS UVOT -- 1.40 GHz 100µm 100µm 60µm 60µm 25µm 25µm 12µm 12µm Ks 2.159µm H 1.662µm J 1.235µm 5460A V 4340A B 3450A U 2600A UVW1 2200A UVM2 1930A UVW2 9.146 12.477 12.699 13.079 13.398 14.143 14.302 14.385 14.740 14.840 14.939 15.062 15.135 15.191 <1mJy 746±190 mJy 237±50 mJy 107±25 mJy 113±30 mJy 12.250±0.026 13.399±0.032 14.214±0.025 15.37±0.02 15.66±0.02 14.49±0.01 14.49±0.01 14.60±0.01 14.54±0.01 νLν 3 4 7.41±1.89 3.98±0.83 4.27±1.00 9.55±2.54 3.85±0.05 2.99±0.10 2.67±0.07 5.08±0.25 4.84±0.24 6.22±0.31 5.88±0.30 6.01±0.30 7.70±0.38 Comments Segment 010 Segment 010 Segment 010 Segment 010 Segment 010 Segment 010 1Central wavelength of the filter 2For the NVSS and the IRAS data we give the flux densities in units of mJy. All others are given in units of mag. 3Observed Luminosities are given in units of 1037 W. 4The upper limit of the NVSS observation is νL1.4GHz < 4.7 × 1031 W.
astro-ph/0608153
1
0608
2006-08-07T19:04:03
Gravitational Lensing of Supernovae Type Ia by Pseudo Elliptic NFW Haloes
[ "astro-ph" ]
We present the effects of ellipticity of matter distribution in massive halos on the observation of supernovae. A pseudo elliptical Navarro-Frenk-White (NFW) mass model is used to calculate the introduced gain factors and observation rates of type Ia supernovae due to the strong lensing. We investigate how and to what extent the ellipticity in mass distribution of the deflecting halos can affect surveys looking for cosmologically distant supernovae. We use halo masses of $1.0 \times 10^{12} h^{-1} M_{\odot}$ and $1.0 \times 10^{14} h^{-1} M_{\odot}$ at redshifts $z_{d}=0.2$, $z_{d}=0.5$, and $z_{d}=1.0$, with ellipticities of up to $\epsilon=0.2$.
astro-ph
astro-ph
Gravitational Lensing of Supernovae Type Ia by Pseudo Elliptic NFW Haloes Hamed Bagherpour 1 Homer L. Dodge Department of Physics and Astronomy, University of Oklahoma, Norman, OK 73019, USA ABSTRACT We present the effects of ellipticity of matter distribution in massive halos on the observation of supernovae. A pseudo elliptical Navarro-Frenk-White (NFW) mass model is used to calculate the introduced gain factors and observation rates of type Ia supernovae due to the strong lensing. We investigate how and to what extent the ellipticity in mass distribution of the deflecting halos can affect surveys looking for cosmologically distant supernovae. We use halo masses of 1.0 × 1012h−1M⊙ and 1.0 × 1014h−1M⊙ at redshifts zd = 0.2, zd = 0.5, and zd = 1.0, with ellipticities of up to ǫ = 0.2. Subject headings: gravitational lensing -- supernovae: general 1. Introduction Supernovae have emerged as the most promising standard candles. Due to their signifi- cant intrinsic brightness and relative ubiquity they can be observed in the local and distant universe. Observational efforts to detect high-redshift supernovae have proved their value as cosmological probes. The systematic study and observation of these faint supernovae (mainly type Ia) has been utilized to constrain the cosmic expansion history (Goobar & Perlmutter 1995; Perlmutter et al. 1999; Schmidt et al. 1998). Light emitted from any ce- lestial object is subject to lensing by intervening objects while traversing the large distances involved (Kantowski, Vaughan, & Branch 1995) and the farther the light source, the higher its chance of being significantly lensed. Apart from the fact that gravitational lensing can limit the accuracy of luminosity distance measurements (Perlmutter & Schmidt 2003), it can change the observed rate of supernovae as well. [email protected] -- 2 -- Studying supernovae and their rates at high redshifts provide us with much needed information for constraining the measurements of the ellusive dark energy, as well as under- standing the cosmic star formation rate and metal enrichment at high redshifts. In order to observe and, hence, study the faint high-redshift supernovae, one can raise the chance of observation by looking through clusters of galaxies or even massive galaxies (see Smail, et al. (2002) and the references therein). These 'gravitational telescopes' amplify the high-redshift supernovae and thereby increase the chance of their detection. However, this boost in ob- servation is offset by the competing effect of depletion (Fig. 1), due to the field being spread by the deflector (amplification bias). For an assumed lens model and a given field of view it is not obvious which effect dominates the observation of supernovae through the halo. The net result depends on the deflector and source parametrs as well as the observational setup (Gunnarsson & Goobar 2003). Some research has been conducted on the feasibility of observing supernovae through cluster of galaxies (see, for instance, Saini, Raychaudhary, & Shchekinov 2002; Gal-Yam, Maoz, & Sharon 2002; Gunnarsson & Goobar 2003). These studies have not taken into ac- count how the morphology (mainly the ellipticity) of these clusters as gravitational telescopes could change the expected supernova rate. In this paper, we investigate whether introduc- ing ellipticity into the mass distribution of the deflecting halos can affect the observation of supernovae. For this purpose, we use a pseudo elliptical Navarro-Frenk-White (NFW) halo model with different values of ellipticity. Throughout the paper we assume the so-called concordance cosmology where Ωm = 0.3, ΩΛ = 0.7, and h100 = 0.67, with h100 = H0/100 km s−1Mpc−1. In § 2 we briefly go over the NFW model and show how an analytical formalism for a pseudo elliptical NFW mass profile can be introduced. Strong lensing by thin deflectors as well as the way ellipticity can afffect the amplification is explained in § 3. We present and discuss the results of our calculations in § 4. 2. The NFW Halo Model Profile 2.1. NFW Haloes High resolution N-body numerical simulations (Navarro, Frenk, & White 1995, 1996, 1997) have indicated the existence of a universal density profile for dark matter halos resulting from the generic dissipationless collapse of density fluctuations. This density profile does not (strongly) depend on the mass of halo, on the power spectrum of initial fluctuations, or on the cosmological parameters. These halo models which are formed through hierarchical clustering diverge with ρ ∝ r−1 near the halo center and behave as ρ ∝ r−3 in its outer regions. Inside the virial radius, this so-called NFW halo profile appears to be a very good -- 3 -- description of the mass distribution of objects spanning 9 orders of magnitude in mass: ranging from globular clusters to massive galaxy clusters (see Wright & Brainerd (2000) and references therein). The NFW halo model is similar to Hernquist profile (Hernquist 1990) that gives a good description of elliptical galaxy photometry. However, the two models differ significantly at large radii, possibly due to the fact that elliptical galaxies, countrary to the dark halos, are relatively isolated systems. The spherically symmetric NFW density profile takes the form of ρ(r) = r rs δcρc (1 + r rs )2 (1) where ρc = [3H 2(z)]/(8πG) is the critical density for closure of the universe at the redshift z of the halo, H(z) is the Hubble parameter at the same redshift, and G is the universal gravity constant. The scale radius rs ≡ r200/c is the charactristic radius of the halo where c is a dimensionless number refered to as the concentration parameter, and δc = 200 3 c3 ln(1 + c) − c 1 + c (2) is a charactristic overdensity for the halo. The virial radius r200 is defined as the radius inside which the mass density of the halo is equal to 200ρc. It is then easy to see that M(r200) ≡ M200 = 800 3 ρcr3 200 . (3) Therefore, NFW halos are defined by two parameters; c, and either r200 or M200. For any spherical NFW profile with a given mass, the concentration parameter c can be calculated using the Fortran 77 code charden.f publicly available on the webpage of Julio Navarro1. NFW halos can be shown to always produce odd number of images, as opposed to the commonly-used singular isothermal sphere (SIS) model which produces either one or two images (Schneider, Ehlers, & Falco 1992). Although baryons are expected to isothermalize the matter distribution for halos of galaxy mass and below (Kochanek & White 2001), taking all of the matter in the universe in isothermal spheres is a great oversimplification (Holz 2001). It is, hence, reasonable to model halos (at least massive halos) with NFW mass profile instead of SIS model. 1http : //pinot.phys.uvic.ca/∼jfn/mywebpage/jfn I.html -- 4 -- 2.2. Elliptical Potential Model Here we present the introduced ellipticity ǫ in the circular lensing potential ϕ(θ), as- suming that angular position θ can be scaled by some scale radius/angle θs. The reader is encouraged to see Golse & Kneib (2002) and Meneghetti, Bartelmann, & Moscardini (2003) for illuminating discussions. We first introduce the dimensionless radial coordinates x = (x1, x2) = R/rs = θ/θs where R is the radial coordinate in the deflector plane, and θs = rs/Dd. Then, one can introduce the ellipticity in the expression of the lens potential by substituting xǫ for x, using the following elliptical coordinate system: x1ǫ = √a1ǫ x1 x2ǫ = √a2ǫ x2 xǫ = px2 φǫ = arctan (x2ǫ/x1ǫ) 1ǫ + x2 2ǫ = pa1ǫx2 1 + a2ǫx2 2 (4)   where a1ǫ and a2ǫ are the two parameters used to define the ellipticity, as explained below. From the elliptical lens potential ϕǫ(x) ≡ ϕ(xǫ), we can calculatete the elliptical deflec- tion angle (see § 3.2): αααǫ(x) =  ∂ϕǫ ∂x1 ∂ϕǫ ∂x2 = α(xǫ)√a1ǫ cos φǫ = α(xǫ)√a2ǫ sin φǫ   (5) Notice that the expressions above hold for any definition of a1ǫ and a2ǫ. Here, we follow Golse & Kneib (2002) who, in order to be able to analytically derive the convergence and shear, chose the following elliptical parameters: a1ǫ = 1 − ǫ a2ǫ = 1 + ǫ (6) (7) which for small values of ellipticity ǫ results in the same ellipticity along the x1 as the standard elliptical model of a1ǫ = 1 − ǫ a2ǫ = 1/(1 − ǫ) (8) (9) with ǫ = 1 − b/a, where a and b are the semi-major and semi-minor axis of the projected elliptic potential, respectively. -- 5 -- 3. Gravitational Lensing: a Reminder 3.1. General Formalism In the thin-lens approximation, we define z as the optical axis and Φ(R, z) as the 3- 2. The so-called reduced 2-dimensional dimensional Newtonian potential, with r = √R2 + z potential which is defined in the deflector plane is given by +∞ ϕ(θ) = 2 2 c Dds DdDs Z−∞ Φ(Dd θ, z) dz (10) (Schneider, Ehlers, & Falco 1992) where c is the speed of light, and θ = (θ1, θ2) is the angular position in the image plane. Dd, Ds, and Dds are angular distances of observer-deflector, observer-source, and deflector-source, respectively. The deflection angle α, convergence κ and the shear γ are given by the following set of equations: 1 ααα(θ) = ∇θϕ(θ) 2(cid:18) ∂2ϕ κ(θ) = ∂θ2 1 γ 2(θ) = kγ(θ)k2 = + ∂2ϕ ∂θ2 1 2(cid:19) 4(cid:18)∂2ϕ 1 − ∂θ2 ∂2ϕ ∂θ2 2(cid:19)2 +(cid:18) ∂2ϕ ∂θ1∂θ2(cid:19)2 .   The lensing equation then reads: β = θ − α = θ − ∇θϕ(θ) (11) (12) (13) where β = (β1, β2) is the angular location of the source. The amplification amp of a point image formed at θ is: amp(θ) = 1 (1 − κ)2 − γ 2 To calculate the angular distances in our work, we use the solution to the Lam´e equation for the distance-redshift equation in a partially filled beam Friedmann-Lemaitre-Robertson- Walker (FLRW) cosmology. For a filled-beam flat FLRW cosmology, the angular distance D as a function of redshift z is D(z) = where 2cz (1 + z)H0 (g(z))1/2 2F1 1 6 , 1 2 ; 7 6 ,−(cid:20) (Ω2 mΩΛ)1/3z2 g(z) See Kantowski (2003) for more detail. g(z) ≡ 2p1 + Ωmz(3 + 3z + z2) + 2 + Ωmz(3 + z) . 3! (cid:21) (14) (15) -- 6 -- 3.2. Lensing Parameters of Spherically symmetric NFW Model Several authors have developed the lensing equations for the ordinary, spherical NFW halos (e.g. Bartelmann 1996; Wright & Brainerd 2000; Golse & Kneib 2002). Following § 2.2 we can introduce a dimensionless radial coordinate in the lens plane x = (x1, x2) = R/rs = θ/θs where θs = rs/Dd. The surface mass density then becomes +∞ Σ(x) = Z−∞ ρ(rs x, z)dz = 2δcρcrsF (x) (16) with F (x) = 1 1 3 x2 − 1(cid:18)1 − x2 − 1(cid:18)1 − 1   1 √1 − x2 arcch 1 √x2 − 1 arccos 1 (x < 1) x(cid:19) x(cid:19) (x > 1) (x = 1) 1 and the mean surface density inside the radius x can be written as Σ(x) = 1 πx2 x Z0 2πxΣ(x)dx = 4δcρcrs g(x) x2 with g(x) = (see Golse & Kneib (2002)). ln x 2 + 1 + ln ln x 2 +   1 2 1 1 √1 − x2 √x2 − 1 arcch 1 x arccos 1 x (x < 1) (x = 1) (x > 1) The deflection angle α, convergence κ and shear γ turn out as α(x) = κ(x) = γ(x) = θ Σ(x) Σcrit Σ(x) Σcrit = 4κs θ x2 g(x)ex = 2κs F (x) Σ(x) − Σ(x) Σcrit = 2κs(cid:18)2g(x) x2 − F (x)(cid:19)   where κs = δcρcrsΣ−1 crit, with Σcrit ≡ c 2Ds/(4πGDdDds). (17) (18) (19) (20) -- 7 -- By integrating the deflection angle, the potential ϕ(x) can be found: with ϕ(x) = 2κsθ2 s h(x) ln2 x ln2 x 2 2 − arcch2 1 x + arccos2 1 x (x < 1) (x ≥ 1) h(x) =  (21) (22) 3.3. Lensing Parameters of Pseudo Elliptical NFW Model For the particular choice of ǫ in § 2.2, the corresponding convergence and shear can be calculated: κǫ(x) = + 1 2θ2 ∂2ϕǫ ∂x2 s (cid:18)∂2ϕǫ 2 (cid:19) s (cid:18)∂2ϕ(xǫ) 2ǫ − = κ(xǫ) + ǫ cos 2φǫ γ(xǫ). ∂x2 1 ǫ 2θ2 = κ(xǫ) + ∂x2 ∂2ϕ(xǫ) 1ǫ (cid:19) ∂x2 and γ 2 ǫ (x) = 1 4θ4 s ((cid:18)∂2ϕǫ 1 − ∂x2 ∂2ϕǫ ∂x2 2 (cid:19)2 +(cid:18)2 ∂2ϕǫ ∂x1∂x2(cid:19)2) = γ 2(xǫ) + 2ǫ cos 2φǫγ(~xǫ)κ(xǫ) + ǫ2(κ2(xǫ) − cos2 2φǫγ 2(xǫ)). Also, the elliptic projected mass density reads: Σǫ(x) = Σ(xǫ) + ǫ cos 2φǫ(Σ(xǫ) − Σ(xǫ)). The lensing equation now becomes (see the appendix): β1 = θsx1(cid:18)1 − 4ksǫ1 β2 = θsx2(cid:18)1 − 4ksǫ2 g(xǫ) g(xǫ) ǫ (cid:19) ǫ (cid:19) x2 x2   and as one expects, the amplification amp reads: amp(x) = 1 (1 − κǫ(x))2 − γ 2 ǫ (x) (23) (24) (25) (26) (27) -- 8 -- It can be shown that ellipticities beyond ǫ = 0.2 result in unrealistic 'peanut' shaped projected densities, hence in this work we focus on lower values of ǫ. Figure 2 shows the multiple images produced by a 1.0 × 1014h−1M⊙ halo with ellipticity ǫ = 0.1 (courtesy of Golse & Kneib). Dashed lines are the contours with constant surface density Σǫ and the solid lines are the critical and caustic lines. Redshifts of source and deflector are 0.2 and 1.0, respectively. 4. The Method The main reason for studying supernovae magnified by gravitational lensing is to inves- tigate the chance of observing supernovae too faint to be observed in the absence of lensing, which is usually the case for cosmologically distant supernovae, specifically type Ia's. To calculate the observed rate of type Ia supernovae we use the result of predicted rates by Dahl´en & Fransson (1999) for a hierarchical star formation rate model with a charactristic time of τ = 1 Gyr (Fig. 3), which limits our calculation to the redshift depth of zM ax = 5. In order for a supernova to be detected, its apparent magnitude m should not exceed the limiting magnitude of the survey mlimit. Using the definitions of the apparent magnitude and amplification, we get: mamp = mo + 2.5 log((cid:12)(cid:12)(1 − κ)2 − γ 2(cid:12)(cid:12)) in which, mamp is the observed magnitude, and mo is the apparent magnitude of the su- pernova in the absence of the lensing. We can further write mo in terms of the absolute magnitude Mabs of the supernova and rewrite the detection criterion as (28) (cid:0)(1 − κ)2 − γ 2(cid:1) D2 L(zs) 6 10(cid:16) m abs+5 (cid:17) (29) limit −M 2.5 where DL(zs) is the luminosity distance of the supernova at redshift zs. The absolute mag- nitude of type Ia SNe has a very narrow Gaussian distribution around Mabs = −19.16 at a confidence level of 89% (Richardson et al. 2002) . Here, we assume that the supernova is detected as soon as its absolute magnitude becomes brighter than Mabs = −18. We take the deflecting halo to be at redshifts zs = 0.2, 0.5, and 1.0, and with virial masses of md1 = 1.0 × 1012h−1M⊙ and md2 = 1.0 × 1014M⊙h−1. Concentration parameter c, overdensity δc, and virial radius r200 (in units of Kpch−1) for each case are given in Table 1. The field of view is taken to be the spatial angle subtending the virial area of the halo. By breaking the projected halo into pixels with the angular size of δx1 and δx2 (which are taken to be smaller than the angular resolution of the observation, Figure 4), we calculate -- 9 -- the amplification across the halo and hence, find the number of observable supernovae in redshift shells with the width of δz = 0.05. We find the corresponding (spatial angular) element δβ1 × δβ2 in the area behind the halo (in redshift space) where the supernovae are bright enough to be detected. Assuming we can arbitrarily minimize δx1 and δx2 , we have δβ1 × δβ2 =(cid:12)(cid:12)(cid:12)(cid:12) ∂βββ ∂xxx(cid:12)(cid:12)(cid:12)(cid:12) is the determinant of Jaccobian matrix. δx1 × δx2 (30) where (cid:12)(cid:12)(cid:12) ∂βββ ∂xxx(cid:12)(cid:12)(cid:12) The reader can refer to the appendix for the derivation of the Jaccobian. The gain factor, defined as the ratio of the number of observable lensed supernovae over the number of observable supernovae in the absence of lensing (Nlensed/NN oLensing) can be calculated by integrating over the predicted rates of both cases across the whole observable area (Fig. 1) for any given lensing configuration, considering the ellipticity ǫ. 5. Results and Discussion First, we consider the effect of ellipticity in the number rate of SN Ia in every redshift bin δz = 0.05. Upper panels of Figures 5 (md1) and 6 (md2) show the number of expected supernovae per year occuring in the redshift bins. We present the results for ǫ = 0.0 and ǫ = 0.2 with the deflecting halo at redshifts zd = 0.2, 0.5, and 1.0. The survey magnitude is assumed to be mlim = 27. The number rate peaks at around z = 1.3 as expected (see Fig. 3) and dies off rapidly beyond that. It can be seen that the farther the deflector, the slightly higher the slope of the curves up to z = 1.3 as a result of higher number of supernovae observed in front of the deflector. Middle and lower panels in Figures 5 and 6 show the cumulative number rates and the gains, respectively. The dominance of amplification bias as a result of the narrowing of the md zd 0.2 0.5 1.0 δcδcδc 1.0 × 1012h−1M⊙ r200 152.61 136.24 111.79 38468.6 33096.0 25118.2 c 9.40 8.83 7.87 δcδcδc 1.0 × 1014h−1M⊙ r200 708.36 632.38 518.88 15741.7 14426.7 12086.5 c 6.46 6.22 5.77 Table 1: NFW halo parameters for the two halo masses md at the given redshifts zd used in the paper. -- 10 -- field in a region immediately behind the deflectors at the assumed redshifts is clear, as the gains fall below 1. Beyond that region amplification takes over and more (lensed) supernovae are observed. In the absence of an intercepting halo, the number rate of the survey drops to zero at the redshift limit of the survey. With the deflecting halo present, the observed rate goes to zero at a higher redshift. This can be seen in figures 7 (md1) and 8 (md2) where the deflector is at redshift zd = 0.5 and the survey magnitude limit is mlim = 27. The three upper panels depict the expected rates for lensing and no-lensing sccenarios for ellipticities ǫ = 0.0, 0.1, and 0.2. The number rates per redshift bin (left) and the cumulative rate (right) are given. The reader can readily notice the effect of bias behind the halo. With the galactic size halo md1, the survey can detect supernovae up to redshift z ∼ 3 (Fig. 7). This limit increases to z ∼ 5 (Fig. 8) for the cluster-size halo md2. The lowest panel in these two figures show the relative difference of the cases with ǫ = 0.1 and ǫ = 0.2 with respect to ǫ = 0.0. The 2 curves do not show significant difference for the redshift bins in front of the halo. In the regime behind the halo, the difference becomes remarkable: it increases up to redshift z = 1.4 for md1 and z = 1.7 for md2. The difference doesn't vary remarkably beyond the maximum point. To further see how ellipticity changes the expected rate of observed supernovae we put the result of our calculations for different ellipticities for a given range of magnitude limits on the same plot. Figure 9 shows the number rate of observed type Ia supernovae (upper panel) for ellipticities ǫ = 0.1 and ǫ = 0.2 together with their relative difference with respect to the case with no ellipticity (lower panel). Both halo masses, md1 and md2 are at redshift zd = 0.2. Figures 10 and 11 show the results of the same calculations with halos at redshifts z = 0.5 and z = 1.0, respectively. The number rates in each figure increase smoothly up to the magnitude limit at which the survey is deep enough to detect the supernovae as far as the halo itself, e.g, mlim = 22.4 for a concordance cosmology of (Ωm, ωΛ, h100) = (0.3, 0.7, 0.67). From that point on the rates increase very rapidly as the magnitude limit goes up. That is caused by the halo lensing and hence amplifying the supernovae which would otherwise be too dim to be observed. The relative differences depicted in these figures show that even at a magnitude limit of 25, effect of ellipticity cannot be ignored as it significantly changes the number/percentage of the observed supernovae; for instance, the relative difference for ǫ = 0.2 with deflecting halo md2 at redshift z = 1.0 (Fig. 11) exceeds 9% for the magnitude limit of mlim = 27. -- 11 -- 6. Conclusion Aiming behind massive halos seem to be a good way to enhance the high-redshift su- pernovae surveys. The cumulative gains of such surveys seem insignificant at low redshifts (zs < 0.2) but the results are remarkable at higher redshifts. For deep observations where mlim > 25, the geometry of the intervening halo cannot be ignored. We have shown that introducing ellipticity in the (gravitational potential of) the mass distribution of a deflecting halo (here, for a galactic halo of mass 1.0 × 1012M⊙h−1 as well as a middle-size cluster of galaxies with a mass of 1.0 × 1014M⊙h−1) can affect the rate of observed supernovae by a few percent. It was shown that the farther the supernova survey probes, the more significant the effects of introduced ellipticity become. It should be noted that this work does not involve a broad range of mass profiles for the halos (although we specify that the survey is limited to the virial area of the halo), nor does it address the much needed k-correction. Our calculations are actually an oversimplification due to the fact that a large, massive halo like a galaxy cluster has substructure which consists of the member galaxies, as well as large clouds of gas. A more sophisticated lens model with ellipticity should be employed to calculate the number rate of observed supernovae. The author wishes to thank D. Branch for enlightening discussions and suggestions. The author would also like to thank C. Gunnarsson and A. Goobar for generously offer- ing him their data set on the rate of SNe Ia, and J. Navarro for allowing him to use the NFW code. This work was in part supported by NSF grant AST0204771 and NASA grant NNG04GD36G. A. Appendix Here we derive the lensing equation for an elliptical NFW halo with ellipticity of ǫ ∂xxx needed to introduced in its 2-dimensional potential, and proceed to calculate Jaccobian ∂βββ get the spatial angular element δβ1 × δβ2 in the source frame. Lensing Equation Introducing the dimensionless coordinate system x = (x1, x2) = R/rs = θ/θs, the lensing equation becomes (cid:26) β1 = θsx1 − α1 [x1, x2] β2 = θsx2 − α2 [x1, x2] (A1) Given the elliptical deflection angle of αǫ(x) =  and the deflection angle of α as -- 12 -- ∂ϕǫ ∂x1 ∂ϕǫ ∂x2 = α(xǫ)√a1ǫ cos φǫ = α(xǫ)√a2ǫ sin φǫ   the lensing equation now reads ααα(x) = θ Σ(x) Σcrit = 4κs θ x2 g(x)ex β1 = θsx1(cid:18)1 − 4ksǫ1 β2 = θsx2(cid:18)1 − 4ksǫ2   Jaccobian g(xǫ) ǫ (cid:19) ǫ (cid:19) x2 x2 g(xǫ) To calculate spatial angular element δβ1 × δβ2 we use the Jaccobian equation . ∂β2 δx1 × δx2 Given Eq. A4, we get: δβ1 × δβ2 = (cid:12)(cid:12)(cid:12)(cid:12)   ∂β1 ∂x1 ∂β2 ∂x1 ∂β1 ∂x2 ∂β2 ∂x2 ∂βββ ∂β1 ∂x2 . ∂β1 ∂x1 ∂β2 ∂x2 − δx1 × δx2 = (cid:12)(cid:12)(cid:12)(cid:12) ∂x1(cid:12)(cid:12)(cid:12)(cid:12) ∂xxx(cid:12)(cid:12)(cid:12)(cid:12) = θs (1 − 4ksa1ǫG(xǫ)) θsx1(cid:18)1 − 4ksa1ǫ ∂x1 (cid:19) = θsx2(cid:18)1 − 4ksa2ǫ = θsx1(cid:18)1 − 4ksa1ǫ = θs (1 − 4ksa2ǫG(xǫ)) θsx2(cid:18)1 − 4ksa2ǫ ∂x1 (cid:19) ∂x2 (cid:19) ∂x2 (cid:19) ∂G(xǫ) ∂G(xǫ) ∂G(xǫ) ∂G(xǫ) G(xǫ) ≡ g(xǫ) x2 ǫ where function G is defined as and g(xǫ) x2 ǫ = 3 2 xǫ (1 − x2 ǫ ) 1 − 6 (1 + x2 ǫ ) 2xǫ (1 − x2   arcch ǫ ) − 1 xǫ − (1 + x2 ǫ ) 2xǫ (1 − x2 ǫ ) xǫ (1 − x2 ǫ ) 3 2 arccos 1 xǫ (xǫ < 1) (xǫ = 1) (xǫ > 1). (A2) (A3) (A4) (A5) (A6) (A7) (A8) -- 13 -- REFERENCES Bartelmann, M. 1996, A&A, 313, 697 Dahl´en, T., & Fransson, C. 1999, A&A, 350, 349 Gal-Yam, A., Maoz, D., & Sharon, K. 2002, MNRAS, 332, 37 Golse, G., & Kneib, J. P. 2002, A&A, 390, 821 Goobar, A., & Perlmutter. S. 1995, ApJ, 450, 14 Gunnarsson, C., & Goobar, A. 2003, A&A, 405, 859 Hernquist, L. 1990, ApJ, 356, 359 Holz, D. 2001, ApJ, L71, 2001 Kantowski, R., Vaughan, T., & Branch, D. 1995, ApJ, 447, 35 Kantowski, R. 2003, Phys. Rev. D, 68, 123516 Kochanek, C. S., & White, M. 2001, ApJ, 559, 531 Meneghetti, M., Bartelmann, M.,& Moscardini L. 2003, MNRAS, 340, 105 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 Navarro, J. F., Frenk, C.S., & White, S. D. M. 1997, ApJ, 490, 493 Perlmutter, S., et al. 1999, ApJ, 517, 565 Perlmutter, S., & Schmidt B. P. 2003, Lecture Notes in Physics, ed. K. Weiler (Berlin: Springer-Verlag) Richardson, D., et al. 2002, AJ, 123, 745 Schmidt, B. P., et al. 1998, ApJ, 507, 46 Schneider, P., Ehlers, J., & Falco, E. E. 1992, Gravitational Lenses (Berlin: Springer-Verlag) Saini, T. D., Raychaudhary, S., & Shchekinov, Y. A. 2000, A&A, 363, 349 Smail, I., et al. 2002, MNRAS, 331, 495 -- 14 -- Sullivan, M., et al. 2000, MNRAS, 319, 549 Wright, C. O., & Brainerd, T. G. 2000, ApJ, 534, 34 This preprint was prepared with the AAS LATEX macros v5.2. -- 15 -- Fig. 1. -- Schematic picture of the lensing configuration by a deflecting halo. zhalo is the redshift of the halo and zlimit is the redshift corresponding to the limiting magnitude mlimit. The shaded area shows the volume where SNe are bright enough to be observed. Fig. 2. -- Multiple images produced by a 1.0 × 1014h−1M⊙ halo with ellipticity ǫ = 0.1. Dashed lines are the contours with constant surface density and the solid lines are the critical and caustic lines. Redshifts of source and deflector are 0.2 and 1.0, respectively (courtesy of Golse & Kneib). Fig. 3. -- Rates of type Ia supernovae in intervals of δz = 0.05. Dilution factor of 1 + z is taken into account (courtesy of Gunnarsson & Goobar). Fig. 4. -- This figure shows how the projected deflector is 'pixellated' in order to calculate the observable area behind the halo. Each pixel has dimensions of δω× δω with δω being smaller than the angular resolution of the observation. Fig. 5. -- Rates of observed supernovae Ia per redshift bin δz = 0.05 (upper panel), cumulative rate (middle panel), and the lensing gain (lower panel) for a deflecting halo of mass md = 1.0 × 1012h−1M⊙ at redshifts of zd = 0.2, 0.5, and 1.0 with ellipticities ǫ = 0.1 and 0.2. Fig. 6. -- Same as Figure 5, with md = 1.0 × 1014M⊙h−1. Fig. 7. -- In this picture the 3 upper panels show observed rates of lensed (solid line) and unlensed (dash line) for three different ellipticies ǫ = 0.0, 0.1, and 0.2. The deflecting halo has a mass of md = 1.0 × 1012h−1M⊙ and is located at redshift zs = 0.5. The lowermost panel depicts the relative difference of ǫ = 0.1 (solid line) and ǫ = 0.2 (dash line) with respect to ǫ = 0.0. Fig. 8. -- Same as Figure 7, with md = 1.0 × 1014M⊙h−1. -- 16 -- Fig. 9. -- Rates of observed supernovae Ia as a function of survey magnitude limit m (upper panel). Result sare shown for halo masses md = 1.0 × 1012h−1M⊙ and md = 1.0 × 1014h−1M⊙ with ellipticities ǫ = 0.1 and 0.2. The halo is at redshift zd = 0.2. The relative difference of cases with ǫ = 0.1 and 0.2 with respect to ǫ = 0.0 is given in the lower panel. Fig. 10. -- Same as Fig. 9 with zd = 0.5. Fig. 11. -- Same as Fig. 9 with zd = 1.0. -- 17 --                                   haloZ limitZ Figure 1 d n o c e s c r a o r c i m -- 18 -- micro arcsecond Figure 2 z δ / r y / N 80 60 40 20 0 0 -- 19 -- δz = 0.05 1 2 z Figure 3 3 4 5 δ 1X δ 2X -- 20 -- b X 2 X 1 a Figure 4 -- 21 -- ε = 0.0 ε = 0.2 zd = 0.2 zd = 0.5 zd = 1.0 zd = 0.2 zd = 0.5 zd = 1.0 80 60 40 20 0 1200 900 600 300 0 1.4 1.2 1 80 60 40 20 z δ / r y / N r y / N 0 1200 900 600 300 0 1.4 1.2 n i a G 1 0 1 2 z 3 4 5 1 2 3 4 5 z Figure 5 80 60 40 20 z δ / r y / N 0 2500 2000 1500 1000 500 r y / N 0 3 2.5 2 1.5 n i a G 1 0 -- 22 -- ε = 0.0 ε = 0.2 zd = 0.2 zd = 0.5 zd = 1.0 zd = 0.2 zd = 0.5 zd = 1.0 1 2 z 3 4 0 5 1 2 3 4 z Figure 6 80 60 40 20 0 2500 2000 1500 1000 500 0 3 2.5 2 1.5 1 5 80 60 40 20 0 80 60 40 20 0 80 60 40 20 z δ / r y / N z δ / r y / N z δ / r y / N ) % ( e c n e r e f f i D e v i t a l e R 0 -20 -15 -10 -5 0 5 0 ε = 0.0 -- 23 -- zd = 0.5 ε = 0.0 Lensing No Lensing ε = 0.1 ε = 0.1 Lensing No Lensing ε = 0.2 ε = 0.2 Lensing No Lensing ε = 0.1 ε = 0.2 ε = 0.1 ε = 0.2 1 z 2 3 1 z 2 3 Figure 7 1200 900 600 300 0 1200 900 600 300 0 1200 900 600 300 r y / N r y / N r y / N ) % ( e c n e r e f f i 0 -6 -5 -4 -3 -2 -1 0 R D e v i t a l e 80 60 40 20 0 80 60 40 20 0 80 60 40 20 z δ / r y / N z δ / r y / N z δ / r y / N 0 -20 -15 -10 -5 0 5 ) % ( e c n e r e f f i D e v i t a l e R ε = 0.0 -- 24 -- zd = 0.5 ε = 0.0 Lensing No Lensing ε = 0.1 ε = 0.1 Lensing No Lensing ε = 0.2 ε = 0.2 Lensing No Lensing ε = 0.1 ε = 0.2 ε = 0.1 ε = 0.2 0 1 2 z 3 4 5 1 2 3 4 5 z Figure 8 2000 1500 1000 r y / N 500 0 2000 1500 1000 r y / N 500 0 2000 1500 1000 r y / N 500 ) % ( e c n e r e f f i 0 -7 -6 -5 -4 -3 -2 -1 0 R D e v i t a l e -- 25 -- (zd = 0.2) md = 1014h-1 Msun ε = 0.1 ε = 0.2 md = 1012h-1 Msun ε = 0.1 ε = 0.2 2000 1500 1000 500 0 15 10 5 0 -5 2000 1500 r y / N 1000 500 0 0 -1 -2 -3 -4 -5 -6 ) % ( e c n e r e f f i D e v i t a l e R 22 23 24 25 mlim 26 27 28 23 24 26 27 28 25 mlim Figure 9 -- 26 -- (zd = 0.5) md = 1014h-1 Msun ε = 0.1 ε = 0.2 md = 1012h-1 Msun ε = 0.1 ε = 0.2 r y / N 2500 2000 1500 1000 500 0 0 -1 -2 -3 -4 -5 ) % ( e c n e r e f f i D e v i t a l e R 2500 2000 1500 1000 500 0 2 0 -2 -4 -6 22 23 24 25 mlim 26 27 28 23 24 26 27 28 25 mlim Figure 10 -- 27 -- (zd = 1.0) md = 1014h-1 Msun ε = 0.1 ε = 0.2 md = 1012h-1 Msun ε = 0.1 ε = 0.2 r y / N 3000 2500 2000 1500 1000 500 ) % ( e c n e r e f f i D e v i t a l e R 0 -1 -2 -3 -4 3000 2500 2000 1500 1000 500 0 -2 -4 -6 -8 -10 22 23 24 25 mlim 26 27 28 23 24 26 27 28 25 mlim Figure 11
0711.2305
1
0711
2007-11-14T21:11:57
IMAGES I. Strong evolution of galaxy kinematics since z=1
[ "astro-ph" ]
(abbreviated) We present the first results of the ESO large program, ``IMAGES'' which aims at obtaining robust measurements of the kinematics of distant galaxies using the multi-IFU mode of GIRAFFE on the VLT. 3D spectroscopy is essential to robustly measure the often distorted kinematics of distant galaxies (e.g., Flores et al. 2006). We derive the velocity fields and $\sigma$-maps of 36 galaxies at 0.4<z<0.75 from the kinematics of the [OII] emission line doublet, and generate a robust technique to identify the nature of the velocity fields based on the pixels of the highest signal-to-noise ratios (S/N). We have gathered a unique sample of 63 velocity fields of emission line galaxies (W0([OII]) > 15 A) at z=0.4-0.75, which are a representative subsample of the population of M_stellar>1.5x10^{10} M_sun emission line galaxies in this redshift range, and are largely unaffected by cosmic variance. Taking into account all galaxies -with or without emission lines- in that redshift range, we find that at least 41+/-7% of them have anomalous kinematics, i.e., they are not dynamically relaxed. This includes 26+/-7% of distant galaxies with complex kinematics, i.e., they are not simply pressure or rotationally supported. Our result implies that galaxy kinematics are among the most rapidly evolving properties, because locally, only a few percent of the galaxies in this mass range have complex kinematics.
astro-ph
astro-ph
Astronomy & Astrophysics manuscript no. YangetalAccepted April 22, 2018 c(cid:13) ESO 2018 IMAGES(cid:63) I. Strong evolution of galaxy kinematics since z =1 Y. Yang1, H. Flores1, F. Hammer1, B. Neichel1, M. Puech2, N. Nesvadba1, A. Rawat1,3, C. Cesarsky2, M. Lehnert1, L. Pozzetti4, I. Fuentes-Carrera1, P. Amram5, C. Balkowski1, H. Dannerbauer6, S. di Serego Alighieri7, B. Guiderdoni8, A. Kembhavi3, Y. C. Liang9, G. Ostlin10, C. D. Ravikumar11, D. Vergani12, J. Vernet2, and H. Wozniak8 7 0 0 2 v o N 4 1 ] h p - o r t s a [ 1 v 5 0 3 2 . 1 1 7 0 : v i X r a 1 GEPI, Observatoire de Paris, CNRS, University Paris Diderot; 5 Place Jules Janssen, Meudon, France 2 ESO, Karl-Schwarzschild-Strasse 2, D-85748 Garching bei Munchen, Germany 3 Inter-University Centre for Astronomy and Astrophysics, Post Bag 4, Ganeshkhind, Pune 411007, India 4 INAF - Osservatorio Astronomico di Bologna, via Ranzani 1, 40127 Bologna, Italy 5 Laboratoire d'Astrophysique de Marseille, Observatoire Astronomique de Marseille-Provence, 2 Place Le Verrier, 13248 Marseille, 6 MPIA, Konigstuhl 17, D-69117 Heidelberg, Germany 7 INAF, Osservatorio Astrofisico di Arcetri, Largo Enrico Fermi 5, I-50125, Florence, Italy 8 Centre de Recherche Astronomique de Lyon, 9 Avenue Charles Andr, 69561 Saint-Genis-Laval Cedex, France 9 National Astronomical Observatories, Chinese Academy of Sciences, 20A Datun Road, Chaoyang District, Beijing 100012, PR France China 10 Stockholm Observatory, AlbaNova University Center, Stockholms Center for Physics, Astronomy and Biotechnology, Roslagstullsbacken 21, 10691 Stockholm, Sweden 11 Department of Physics, University of Calicut, Kerala 673635, India 12 IASF-INAF - via Bassini 15, I-20133, Milano, Italy Received ...... ; accepted ...... ABSTRACT Nearly half the stellar mass of present-day spiral galaxies has formed since z = 1, and galaxy kinematics is an ideal tool to identify the underlying mechanisms responsible for the galaxy mass assembly since that epoch. Here, we present the first results of the ESO large program, "IMAGES", which aims at obtaining robust measurements of the kinematics of distant galaxies using the multi-IFU mode of GIRAFFE on the VLT. 3D spectroscopy is essential to robustly measure the often distorted kinematics of distant galaxies (e.g., Flores et al. 2006). We derive the velocity fields and σ-maps of 36 galaxies at 0.4 < z < 0.75 from the kinematics of the [Oii] emission line doublet, and generate a robust technique to identify the nature of the velocity fields based on the pixels of the highest signal-to-noise ratios (S/N). Combining these observations with those of Flores et al., we have gathered a unique sample of 63 velocity fields of emission line galaxies (W0([Oii]) ≥ 15 Å) at z =0.4 – 0.75, which are a representative subsample of the population of Mstellar≥1.5×1010M(cid:12) emission line galaxies in this redshift range, and are largely unaffected by cosmic variance. Taking into account all galaxies -with or without emission lines- in that redshift range, we find that at least 41± 7% of them have anomalous kinematics, i.e., they are not dynamically relaxed. This includes 26± 7% of distant galaxies with complex kinematics, i.e., they are not simply pressure or rotationally supported. Our result implies that galaxy kinematics are among the most rapidly evolving properties, because locally, only a few percent of the galaxies in this mass range have complex kinematics. It is well-established that galaxies undergoing a merger have complex large-scale motions and thus are likely responsible for the strong evolution of the galaxy kinematics that we observe. Key words. Galaxies: formation - Galaxies: evolution - Galaxies: kinematics and dynamics 1. Introduction The resolved 3D kinematics of distant galaxies are a power- ful tracer of the major processes governing star-formation and galaxy evolution in the early universe such as merging, accre- tion, and hydrodynamic feedback related to star-formation and active galactic nucleus (e.g., Barnes & Hernquist 1996; Barton et al. 2000; Dressler 2004). Thus, robustly measuring the inter- nal kinematics of galaxies in the distant universe plays a crucial role for our growing understanding of how galaxies formed and evolved. Over the last decade, great efforts have been made to study the properties of galaxies in the distant universe (at z∼1), reveal- Send offprint requests to: [email protected] (cid:63) Intermediate MAss Galaxy Evolution Sequence, ESO programs 174.B-0328(A), 174.B-0328(A), 174.B-0328(E) ing a strong evolution with cosmic time. For instance, the cosmic star formation rate (SFR) has declined by a factor ∼10 from z∼1 to the present (Lilly et al. 1996; Madau et al. 1996; Hammer et al. 1997; Cowie et al. 1999; Flores et al. 1999). Such a strong evolution of cosmic SFR is consolidated by subsequent works, e.g., Haarsma et al. (2000), Wilson et al. (2002). Although the conclusions are made from different data, they are consistent within a factor of 3 (Hopkins 2004). Heavens et al. (2004) sug- gest that the cosmic SFR may have reached its peak as late as z ∼ 0.6. Overall, about half of the stellar mass in intermediate- mass galaxies was formed since z = 1, mostly in luminous in- frared galaxies (Hammer et al. 2005). Galaxy interactions and merging may be mechanisms that played a significantly larger role for star-formation in the distant universe than today. Le F`evre et al. (2000) found that the merger rate in the distant universe was about a factor of 10 times higher 2 Y. Yang et al.: IMAGES I. Observations than at low redshift (see also Conselice et al. 2003; Bell et al. 2006; Lotz et al. 2006). The high merger rate detected in the dis- tant universe raises a challenge to the standard scenario of disk formation (Hammer et al. 2007). If major mergers generate the ellipticals inevitably, we would find a large fraction of elliptical galaxies rather than about ∼70% of spiral galaxies among the intermediate-mass galaxies in the local universe. Similarly, the fraction of luminous compact blue galaxies (LCBGs) increases with redshift by about an order of magnitude out to z∼ 1 (Werk et al. 2004; Rawat et al. 2007). LCBGs may be the progenitors of local spheroidal or irregular galaxies at low redshift (e.g., Koo et al. 1995; Guzman 1999), or of the bulges of massive spirals (Hammer et al. 2001; Noeske et al. 2006). 3D spectroscopy of the internal kinematics of LCBGs suggests that they are likely merger remnants ( Ostlin et al. 2001; Puech et al. 2006a). Strong evolution as a function of cosmic time has also been claimed for the Tully-Fisher relationship (TFR, Tully & Fisher 1977; Giovanelli et al. 1997), which relates the luminosity and the rotation velocity of disk galaxies. Out to z ∼ 1, the B-band TFR has been found to have evolved by ∼ 0.2 – 2 mag (e.g., Portinari & Sommer-Larsen 2007 and references therein). This brightening of the B-band TFR can be explained by the enhanced star-formation rates at higher redshifts (Ferreras & Silk 2001; Ferreras et al. 2004), but is still a matter of debate. Conselice et al. (2005) do not find significant evolution in either the stellar mass or K-band TFR's slope or zero point. However, the most striking evolution of the TFR is provided by its large scatter at high redshifts (Conselice et al. 2005), which may be related to the disturbed kinematics of distant galaxies (e.g., Kannappan & Barton 2004). The rapid time decrease of cosmic SFR, the role of merg- ing in the early evolution of galaxies, and the possible evolu- tion of the TFR are only examples of why measuring the kine- matics of distant galaxies precisely and robustly is a sine qua non for studying galaxy evolution. However, this is often be- yond what can be achieved with classical long-slit spectroscopy. The morphologies and kinematics of distant galaxies are often complex, and their small sizes make it very difficult to precisely position and align the slit. Both limitations can be overcome with integral-field spectroscopy, although the method is rela- tively complex and time-consuming. Flores et al. (2006) presented the first study of a statistically meaningful sample of 35 intermediate-mass galaxies at z=0.4- 0.7, using the integral-field multi-object spectrograph GIRAFFE on the ESO-VLT. They defined a classification scheme to distin- guish between rotation and kinematic perturbances, which may stem from interactions and mergers, from the 3D kinematics and high-resolution HST imaging. Intriguingly, they find that the large scatter of distant TFR shown in previous studies is due to non-relaxed systems while the pure rotational disks exhibit a TFR that is similarly tight as that of local spirals. Here, we present another sample of 36 galaxies with very similar selec- tion criteria, to enlarge the total sample size and put the con- clusions on statistically more robust grounds. This is the first of a series of publications related to the ESO-VLT large program IMAGES, which aims at studying the evolutionary sequence of galaxies over the last 8 Gyrs (see Ravikumar et al. 2007 for more details). The paper is organized as follows. In Sect. 2 we describe the observations and the sample selection. The methodology to describe and classify the distant galaxy kinematics is shown in Sect. 3, as well as a detailed description of the 36 observed ve- locity fields (VFs). Sections 4 and 5 include the discussion and the conclusion. In this paper, we adopt the Concordance cos- Table 1. Journal of observations. Run ID 174.B-0328(A) 073.A-0209(A) 174.B-0328(A) 174.B-0328(E) Setup L04 L05 L05 L05 Exposure (hr) 10 4.5 10.6 10.4 mological parameters of H0 = 70 km s−1 Mpc−1, ΩM = 0.3 and ΩΛ =0.7. 2. Data 2.1. Observations We used the FLAMES-GIRAFFE multi-object integral-field spectrograph on the ESO-VLT in the multi-IFU mode to mea- sure the velocity and dispersion fields of a statistically mean- ingful sample of galaxies at redshifts z = 0.4 – 0.75 from their [Oii]λ3726,3729 emission. Each integral-field unit (IFU) of GIRAFFE consists in 20 micro-lenses with 0.52(cid:48)(cid:48) spatial sam- pling, resulting in a 2(cid:48)(cid:48)×3(cid:48)(cid:48) field of view per IFU. We used the LR04 and LR05 set-ups, which correspond to spectral resolu- tions of 0.55 Å (30 km s−1) and 0.45 Å (22 km s−1), respectively. Observations were carried out as part of the IMAGES large program, complemented by guaranteed time observations (pro- grams 174.B-0328(A), 073.A-0209(A), 174.B-0328(A), 174.B- 0328(E), see also Table 1). The total observing time was 5 nights, with integration times ranging from 4.5 to 15 hrs for indi- vidual targets. The seeing ranged from 0.4(cid:48)(cid:48)to 1(cid:48)(cid:48), with a median value of 0.8(cid:48)(cid:48). Data reduction and the construction of the final data cubes are described in detail in Flores et al. (2006). 2.2. Sample selection Our targets are a subset of the Chandra Deep Field South, with redshifts z∼ 0.4– 0.75, IAB ≤ 23.5 and detected [Oii]λ3726,3729 emission lines (W0([Oii]) ≥ 15 Å, Ravikumar et al. 2007). Our goal is to investigate a sample of intermediate mass galaxies (see Hammer et al. 2005), therefore we required J-band absolute magnitudes brighter than MJ(AB) = −20.3. Such a limit corre- sponds approximately to a stellar mass of Mstellar ≥ 1.5×1010M(cid:12) when converting the J-band luminosity using the prescription discussed in Bell et al. (2003; see also Hammer et al. 2005). Ravikumar et al. (2007) has convincingly shown that within the redshift range of 0.4– 0.75, IAB ≤ 23.5 galaxies include almost all intermediate mass galaxies (e.g., at least 95% of MJ(AB) ≤ -20.3 galaxies, see their Sect. 3.4). Given all the above, our sam- ple comprises a total of 46 targets. The relatively small number of suitable galaxies and small bandpass of the GIRAFFE set-ups make it difficult to fill all 15 IFUs with galaxies with spectro- scopic redshifts. We therefore used empty "bonus" IFUs to ob- serve galaxies for which only photometric redshifts were known, but could not detect any because of the large uncertainties of photometric redshifts. Moreover, we rejected 2 galaxies with spurious features that were identified as [Oii] emission lines in their spectra (J033221.42-274231.2, J033241.08-274853.0), 4 targets due to faint line emission (i.e., W0([Oii]) < 15 Å: J033211.41- 274650.0, J033232.13-275105.5, J033254.50-274703.6), and one due to the CCD defects (J033213.85-274248.9). Another 2 galaxies (J033212.36- 274835.6, J033236.72-274406.4) were rejected by our mini- J033226.00-274150.6, Y. Yang et al.: IMAGES I. Observations 3 3. Kinematics of distant galaxies 3.1. Measuring galaxy kinematics using the [Oii] doublet Our method to extract kinematic fields from 3D spectroscopy of the [Oii]λ3737,3729 emission line doublet has been described in detail in Flores et al. (2006). Here, we only give a brief overview, and highlight recent improvements. The 20 individual spectra of each object were inspected visually to detect possible artefacts or contamination with night sky lines. We then constructed the 3D data cube around the expected observed [Oii] wavelength with and without sky subtraction, and fitted the [Oii] doublet with two Gaussian, keeping the wavelength difference between the two lines at rest-frame λ2 − λ1 = 2.783 Å fixed, and requiring that both lines have the same dispersion, σ1 = σ2. The line ratio is a free parameter except when the fit failed, in which case we impose a ratio of R(3729/3727) =1.4, which corresponds to the low density limit and is appropriate for most galaxies (see also Puech et al. 2006b; Weiner et al. 2006). We estimate the systemic velocity of each galaxy from the σ-clipped mean of the spatially-resolved velocities, and mea- sure the width of night sky lines to correct the dispersion maps (σ-maps) for instrumental resolution. We also derive S/N-maps to quantify the uncertainty of the kinematics following the def- inition of Flores et al. (2006); in particular, we use only those spectra where the [Oii] line emission is detected with a S/N ≥ 3, and apply a simple 5×5 linear interpolation to the VF and σ- map. We show the VF, σ-maps and S/N-maps in Fig. 2 with the high resolution (0.03(cid:48)(cid:48)/pixel) ACS F775W image for each ob- ject. The analysis of the full sample was done independently by several of us (HF, BN, MP and YY), before comparing and fi- nalizing the results. Since Flores et al. (2006), we have improved our analysis software to better account for the contamination of the emis- sion line spectrum with night sky lines, fitting the [Oii] emission lines and superimposed night sky lines simultaneously. This is particularly relevant with the L05 set-up (5741 – 6524 Å), where the risk of overlaps is important due to a relatively large num- ber of night sky lines. At the relatively high effective resolving power of GIRAFFE of R≥ 10 000, the [Oii] emission lines are significantly broader than the night sky lines, which is essen- tial to successfully isolate the signal (see Fig. 3 for an exam- ple). By fitting the sky and object simultaneously, we have been able to recover the kinematics of 3 galaxies that were particularly strongly blended with night sky lines, and would have been oth- erwise rejected (J033217.62-274257.4, J033220.48-275143.9, J033244.20-274733.5). 3.2. Classification of the kinematics of distant galaxies Flores et al. (2006) developed a simple kinematic classification scheme for distant galaxies based on their 3D kinematics and their morphology in the ACS F775W images. It relies on the fact that at low spatial resolution, a rotating disk should show a well defined peak in the center of the σ-map, which corresponds to the convolution of the large scale motion (i.e., the rotation) with the (relatively small) dispersion of the gas in the disk or in the bulge. Indeed, the central parts of the galaxy, where the rotation curve rises most quickly, are not spatially resolved with ground- based optical spectroscopy and our classification fully accounts for its convolution with the actual PSF (point spread function). To summarize, we distinguish between the following classes : 1. Rotating disks (RD): the VF shows an ordered gradient, and the dynamical major axis is aligned with the morphological Fig. 1. Number counts (in logarithmic scale) of selected galax- ies versus AB absolute magnitude in J-band. The GTO sample refers to Flores et al. (2006); the IMAGES sample refers to this paper. The vertical dotted line indicates the limit of the IMAGES program. Two luminosity functions derived from Pozzetti et al. (2003) are shown (full line: z = 0.5; dashed line: z = 1). The galaxies of our sample have redshifts ranging from z = 0.4 to z = 0.75. This implies that the combined sample of 63 galaxies with MJ(AB) ≤ −20.3 is representative of galaxies with stellar masses larger than 1.5 × 1010M(cid:12) at z ∼ 0.6. mum quality criterion: at least 4 GIRAFFE spatial pixels with spectral S/N > 4. The galaxy J033229.71-274507.2 was also rejected because emission was detected in only 4 GIRAFFE pix- els, which is not sufficient to classify its kinematics. Finally, we obtained a sample of 36 well resolved galaxies of intermediate stellar mass with good S/N values. 2.3. Representativeness/completeness of the sample Figure 2.3 shows the distribution of the J-band absolute magni- tudes (also listed in Table 2) for the sample studied in this paper, combined with the Flores et al. (2006) sample of 35 galaxies. Both samples can be merged because the selection of Flores et al. is very similar to that of this paper and because both stud- ies used essentially the same instrumental set-ups. Applying our criteria of MJ(AB) ≤ −20.3 and W0([Oii]) ≥ 15 Å, we are left with 63 galaxies with data of appropriate quality to carry out our analysis. We compared the luminosity distribution of the sample to the luminosity function at redshift of 0.5 and 1 (Fig. 2.3). Kolmogorov-Smirnov tests support that our sample follows the luminosity function in the z = 0.4 – 0.75 range at > 99.9% confidence level. Furthermore, the combined two sam- ples include galaxies from 4 different fields, namely the CDFS, HDFS, CFRS03h and CFRS22h. It is then unlikely that our con- clusions are strongly affected by statistical effects, possibly re- lated to large scale structures (see Sect. 4 for more analysis, also Ravikumar et al. 2007). Our sample is a completely representa- tive sub-sample of MJ(AB)≤−20.3 emission line selected galax- ies at z =0.4 – 0.75. To our knowledge, it is the only such existing sample of distant galaxies with measured 3D kinematics. 4 Y. Yang et al.: IMAGES I. Observations Fig. 2. Kinematics of the individual galaxies. Each row corresponds to one galaxy. From left to right, we show the HST/ACS F775W images, the observed VFs, σ-maps and S/N-maps, the model VFs and σ-maps. A grid of GIRAFFE IFU superposed on the HST image indicates the position and the scale of IFU with respect to the galaxy. We have applied a 5×5 linear interpolation to the VFs and σ-maps for visualization. -147.3 63.0km/s 31.5 62.3km/s 3.3 4.7-131.6 74.7km/s 55.3 90.7km/s-152.3 169.0km/s 31.0 98.3km/s 3.3 21.7-171.7 156.6km/s 45.0 116.7km/s -69.5 50.3km/s 42.4 77.3km/s 3.0 5.3 -11.0 130.6km/s 55.9 87.2km/s -89.3 73.5km/s 32.4 69.6km/s 3.0 12.4 -99.3 88.9km/s 34.5 71.4km/s-103.6 105.0km/s 28.2 61.9km/s 3.1 20.9 -99.6 108.2km/s 34.4 74.7km/s-127.0 85.0km/s 23.0 77.3km/s 4.9 17.1-110.1 113.5km/s 25.5 81.9km/s Y. Yang et al.: IMAGES I. Observations 5 Fig. 2. continued. -74.4 162.1km/s 16.7 57.3km/s 3.1 7.8-141.5 92.4km/s 36.5 80.7km/s -91.8 67.2km/s 27.4 48.8km/s 3.0 20.0 -74.1 83.3km/s 32.7 52.6km/s -89.4 99.5km/s 56.8 107.8km/s 3.1 10.4-119.2 73.8km/s 63.9 110.7km/s -69.3 149.7km/s 30.4 85.7km/s 3.0 10.5-116.9 112.6km/s 40.0 61.6km/s -43.9 79.7km/s 5.6 62.4km/s 4.1 10.1 -60.0 61.7km/s 9.8 40.6km/s -43.4 119.6km/s 24.2 73.6km/s 3.4 37.3 -89.4 80.6km/s 34.4 86.2km/s 6 Y. Yang et al.: IMAGES I. Observations Fig. 2. continued. -155.0 144.1km/s 13.6 76.8km/s 3.1 6.5-138.4 169.6km/s 31.1 93.1km/s -27.4 35.3km/s 16.7 53.0km/s 3.1 17.6 -32.8 30.4km/s 19.7 29.7km/s -54.2 37.2km/s 25.3 73.0km/s 3.6 30.1 -52.3 57.2km/s 26.6 46.9km/s -53.8 62.9km/s 22.5 66.2km/s 3.6 32.0 -57.4 63.3km/s 26.2 56.6km/s -36.7 39.2km/s 3.7 62.5km/s 3.3 10.2 -32.2 39.8km/s 5.6 28.7km/s -85.2 66.8km/s 27.7 63.9km/s 4.2 13.9 -87.5 66.7km/s 30.1 54.7km/s Y. Yang et al.: IMAGES I. Observations 7 Fig. 2. continued. -65.8 81.8km/s 25.8 90.0km/s 3.7 16.6 -66.4 79.4km/s 31.7 65.7km/s -60.5 81.4km/s 36.5 50.6km/s 5.1 10.7 -69.4 73.8km/s 45.4 62.0km/s-195.4 176.1km/s 54.5 99.2km/s 3.5 7.8-180.5 174.6km/s 74.6 140.3km/s -81.1 99.3km/s 13.0 80.0km/s 3.0 13.5 -89.7 85.7km/s 30.4 63.7km/s-181.4 81.4km/s 13.8 125.1km/s 3.2 12.2-131.5 140.2km/s 23.2 116.1km/s-135.5 96.4km/s 34.4 90.6km/s 3.1 8.7 -24.5 183.8km/s 62.0 162.6km/s 8 Y. Yang et al.: IMAGES I. Observations Fig. 2. continued. -49.9 27.4km/s 25.9 88.3km/s 3.1 16.7 -41.5 37.2km/s 27.2 38.9km/s -62.2 64.0km/s 31.1 54.2km/s 3.0 18.2 -62.6 63.7km/s 36.6 52.9km/s -57.1 37.1km/s 30.8 77.6km/s 3.0 22.8 -43.1 49.0km/s 33.5 48.7km/s -28.9 113.7km/s 19.9 47.9km/s 3.7 18.8 -72.4 69.5km/s 41.6 58.6km/s -40.6 61.0km/s 24.5 55.8km/s 3.2 6.2 -61.0 36.6km/s 33.9 47.0km/s-201.2 237.0km/s 12.4 92.7km/s 3.5 14.4-307.3 103.0km/s 23.1 146.0km/s Y. Yang et al.: IMAGES I. Observations 9 Fig. 2. continued. -32.7 48.3km/s 41.5 78.5km/s 3.6 15.6 -43.4 40.1km/s 42.1 52.8km/s -25.8 23.2km/s 13.8 34.3km/s 4.2 24.3 -38.4 8.4km/s 26.6 38.7km/s -15.6 16.7km/s 16.6 54.7km/s 3.3 55.7 -16.8 17.1km/s 16.9 20.0km/s -81.3 133.7km/s 10.2 111.4km/s 3.0 7.1-124.8 98.2km/s 19.6 118.0km/s -55.0 36.8km/s 47.4 105.3km/s 4.3 49.4 -45.0 48.3km/s 52.8 64.1km/s -57.0 107.7km/s 31.6 88.1km/s 3.1 11.2 -88.2 76.7km/s 41.9 76.8km/s 10 Y. Yang et al.: IMAGES I. Observations Table 2. Properties of the 39 galaxies of the IMAGES sample. GOODS ID J033212.39-274353.6 J033219.68-275023.6 J033230.78-275455.0 J033231.58-274121.6 J033234.04-275009.7 J033237.54-274838.9 J033238.60-274631.4 J033241.88-274853.9 J033245.11-274724.0 J033210.25-274819.5 J033214.97-275005.5 J033219.61-274831.0 J033226.23-274222.8 J033232.96-274106.8 J033233.90-274237.9 J033239.04-274132.4 J033243.62-275232.6 J033248.28-275028.9 J033249.53-274630.0 J033250.53-274800.7 J033210.76-274234.6 J033213.06-274204.8 J033217.62-274257.4 J033219.32-274514.0 J033220.48-275143.9 J033224.60-274428.1 J033225.26-274524.0 J033227.07-274404.7 J033228.48-274826.6 J033230.43-275304.0 J033230.57-274518.2 J033234.12-273953.5 J033239.72-275154.7 J033240.04-274418.6 J033244.20-274733.5 J033250.24-274538.9 J033212.36-274835.6 J033229.71-274507.2 J033236.72-274406.4 RA & DEC (J2000.0) 03:32:12.387 −27:43:53.59 03:32:19.678 −27:50:23.57 03:32:30.780 −27:54:54.99 03:32:31.575 −27:41:21.63 03:32:34.037 −27:50:09.69 03:32:37.538 −27:48:38.94 03:32:38.595 −27:46:31.37 03:32:41.883 −27:48:53.86 03:32:45.108 −27:47:24.00 03:32:10.250 −27:48:19.49 03:32:14.971 −27:50:05.45 03:32:19.606 −27:48:30.97 03:32:26.229 −27:42:22.81 03:32:32.959 −27:41:06.78 03:32:33.897 −27:42:37.93 03:32:39.044 −27:41:32.43 03:32:43.623 −27:52:32.63 03:32:48.281 −27:50:28.88 03:32:49.525 −27:46:29.98 03:32:50.534 −27:48:00.67 03:32:10.761 −27:42:34.58 03:32:13.061 −27:42:04.81 03:32:17.620 −27:42:57.44 03:32:19.317 −27:45:14.04 03:32:20.484 −27:51:43.93 03:32:24.601 −27:44:28.12 03:32:25.260 −27:45:23.97 03:32:27.074 −27:44:04.66 03:32:28.477 −27:48:26.55 03:32:30.429 −27:53:04.02 03:32:30.569 −27:45:18.24 03:32:34.120 −27:39:53.53 03:32:39.719 −27:51:54.68 03:32:40.040 −27:44:18.63 03:32:44.199 −27:47:33.48 03:32:50.239 −27:45:38.92 03:32:12.360 −27:48:35.64 03:32:29.707 −27:45:07.20 03:32:36.715 −27:44:6.435 za 0.42130 0.55955 0.68573 0.70411 0.70162 0.66377 0.62066 0.66702 0.43462 0.60874 0.66652 0.66992 0.66713 0.46811 0.61801 0.73186 0.67823 0.44464 0.52212 0.73604 0.41686 0.42150 0.64565 0.72411 0.67780 0.53680 0.66479 0.73814 0.66857 0.64533 0.67988 0.62734 0.41510 0.52201 0.73605 0.73099 0.56210 0.73170 0.66500 b MB −19.55 −20.88 −20.51 −20.16 −19.88 −21.29 −20.03 −20.32 −20.13 −19.76 −21.50 −20.36 −20.71 −19.50 −20.99 −20.46 −19.27 −19.35 −20.05 −19.99 −21.78 −19.53 −19.78 −20.31 −19.93 −19.58 −20.90 −20.34 −20.10 −19.96 −21.93 −23.08 −20.10 −20.55 −21.08 −19.79 −20.13 −20.05 −20.23 b MJ −21.58 −22.37 −21.90 −20.69 −20.61 −22.07 −21.54 −21.00 −22.06 −20.93 −22.53 −20.99 −22.01 −20.45 −21.91 −20.75 −20.03 −20.47 −21.09 −20.50 −23.70 −20.67 −21.23 −21.24 −20.72 −20.44 −21.63 −21.04 −21.74 −21.71 −22.95 99.99 −21.04 −22.04 −21.86 −20.70 −21.16 −20.88 −22.01 ic,e Cd ∆re 90.0(5.0) RD 0.08(0.32) RD 0.11(0.46) 58.3(6.7) RD 0.11(0.31) 66.1(4.3) 42.2(9.6) RD 0.11(0.57) 59.3(3.4) RD 0.09(0.21) 30.7(12.) RD 0.11(0.35) RD 0.16(0.34) 60.2(4.3) RD 0.03(0.50) 66.5(6.2) RD 0.14(0.18) 43.3(6.2) 68.5(4.5) PR 0.83(0.19) 1.32(0.27) PR 21.9(8.2) 0.40(0.35) PR 49.1(7.0) 0.72(0.42) PR 76.4(3.1) 0.64(0.37) PR 16.0(3.8) 17.2(10.) PR 0.07(0.26) 0.44(0.25) PR 43.2(10.) 0.15(0.63) PR 70.8(1.6) 0.34(0.82) PR 80.8(3.3) PR 45.6(2.2) 1.10(0.58) 62.3(3.9) PR 0.22(0.62) 26.0(7.4) CK 0.74(0.54) CK 0.68(0.30) 78.7(2.8) CK 1.07(0.21) 46.3(8.4) CK 0.15(0.17) 72.1(4.5) 63.4(3.6) CK 1.05(0.30) CK 0.99(0.48) 64.8(1.2) CK 0.87(0.19) 60.5(6.0) CK 0.69(0.14) 84.2(1.7) CK 0.67(0.24) 22.4(5.0) 70.3(2.4) CK 1.02(0.24) 34.5(9.4) CK 1.04(0.52) 32.1(4.5) CK 0.12(0.18) 35.4(1.3) CK 1.55(0.45) 15.6(11.) CK 0.37(0.22) 39.0(5.8) CK 0.64(0.15) 31.3(8.8) CK 1.37(0.38) 65.8(1.4) UC 42.3(6.2) UC 58.7(1.8) UC – – – e 0.13(0.03) 0.06(0.01) 0.12(0.02) 0.01(<.01) 0.17(0.02) 0.04(0.01) 0.08(0.02) 0.08(0.01) 0.09(0.02) 0.40(0.07) 0.65(0.12) 0.18(0.01) 0.40(0.09) 1.15(0.14) 1.35(0.17) 0.24(0.01) 1.54(0.25) 0.40(0.04) 0.38(0.03) 0.30(0.04) 0.47(0.09) 0.02(<.01) 0.04(0.01) 0.66(0.12) 2.94(0.42) 0.42(0.04) 1.00(0.09) 0.91(0.13) 1.35(0.26) 0.44(0.09) 0.16(0.02) 0.53(0.08) 5.15(0.26) 0.36(0.08) 0.62(0.03) 0.59(0.08) – – – a Redshift measured by [Oii] emission. b Absolute magnitudes in B- and J- band. c Inclination of galaxies and its error (in unit of degree). d Kinematical classification (see Sect. 3.2 for details): RD-rotating disks; PR-perturbed rotations; CK-complex kinematics; UC-unclassified. e The corresponding error of each quantity is given into brackets. major axis. The σ-map indicates a single peak close to the dynamical center; 2. Perturbed rotations (PR): the kinematics shows all the fea- tures of a rotating disk (see above), but the peak in the σ-map is either absent or clearly shifted away from the dynamical center; 3. Complex kinematics (CK): neither the VF nor the σ-map are compatible with regular disk rotation, including VFs that are misaligned with the morphological major axis. According to these definitions of 3 kinematical classes, we have classified the 36 galaxies of our sample. To do so, each galaxy has been examined by 5 of us (HF, FH, BN, MP and YY), before comparing our classification. The results are listed in Table 2, which also includes absolute magnitudes and incli- nations. Absolute magnitudes are derived using the procedure described by Hammer et al. (2005), based on photometry at near IR and optical wavelengths (see also Ravikumar et al. 2007). Inclinations were measured using ellipsoidal isophotes near the optical radius from HST/ACS F775W and F814W imaging. Comparison between estimates from different team members and with values derived from Sextractor (Bertin et al. 1996) sug- gests that typical uncertainties are about 5◦. In some cases, classification is not an easy task. There are two galaxies (J033234.04-275009.7, J033245.11-274724.0) that possess double σ-peaks in GIRAFFE IFU view. They are classi- fied to be RD because the center of the galaxy is just located in between the two adjacent GIRAFFE IFU pixels. These phenom- ena have been reproduced in our model (see Fig. 2). However, another galaxy J033219.61-274831.0, which is classified as PR, has its two σ-peaks located at one side of the galaxy center, which is not expected from a rotating disk. In the next paragraphs, we will describe quantitatively the differences between kinematical classes, a test of the robustness Y. Yang et al.: IMAGES I. Observations 11 during the process to define the geometrical extent of the VF in the plane of the sky, assuming a thin disk. Further, we com- pute the corresponding IFU σ-map by considering the effects of seeing. By comparing the observed and model σ-maps, we can estimate whether the observed kinematics are consistent or not with a rotating disk. Two parameters are then computed to characterize the dif- ferences between the two σ-maps, taking the model σ-map as a reference. The first parameter is the spatial separation (∆r, in GIRAFFE pixels) between the peaks in the two σ-maps. This indicates how the center of rotation is recovered by our observa- tion. For each σ-map the pixel including the σ peak is identified, and then the peak location is calculated as the barycenter of the surrounding pixels. We verify that weighting the barycenter by S/N does not change the result, so we choose a uniform weight for each pixel. The second parameter is the relative difference () between the amplitudes of the modeled and observed σ peaks. We define  as: σobs − σmod (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)location of the peak of model σ-map  = σmod . (1) This definition significantly improves the robustness of the test compared with what was done in Flores et al. (2006), because the test is now applicable to the pixels of highest S/N in the center of the σ-map . One may wonder whether the test is re- liable, given the possible uncertainties due to our observational set-up with low spatial resolution and limited S/N, especially in the outskirts of the galaxies. In fact, some parts of the VF of distant galaxies might have been missed, for example, when the galaxy may extend further than the IFU. Alternatively, low S/N near our cut-off limit may generate an absence of detection in some extended parts of the galaxy. An illustration of this is given by J033230.78-275455.0 (see Fig. 2), for which the S/N ratio is below the cut-off for almost half the disk . This means that for some objects we may have missed part of the rotation curve, or we may have introduced an artificial asymmetry. The immediate consequence of this would be to generate a σ peak that is smaller in amplitude than the real one and/or that is slightly offset from the rotational center. However, the impact of the above has to be quite marginal and has no effect on our classification after a care- ful examination of individual VFs. Moreover, with our method to build model rotational VFs, the maximal model rotation is produced by the observed large scale motions, and we do use the same hot pixels for both model and observed maps. Figure 4 shows the diagnostic diagram of ∆r versus . We have recalculated  for the 32 classified galaxies from Flores et al. 2006, and plotted them in the same figure. While modify- ing the definition clearly changes the location of the points, we confirm all the results of Flores et al., i.e., that rotating disks have locations near ∆r ∼  ∼ 0, while galaxies with anoma- lous VFs are clearly offset. We have performed simulations of two local interacting pairs: ARP 271 and KPG 468 (Fuentes-Carrera et al. 2004; Hern´andez-Toledo et al. 2003), which are supposed to be CK systems. We assume that they are located at z=0.6 and observed by GIRAFFE IFU mode in the simulations. Following our method, we find these two simulated CKs (indicated by black crosses) are located far from the rotating disks, supporting our di- agnostic method. 3.3.2. Using χ2-test to recognize disk kinematics Given the spatial resolution of IFU and the seeing during ob- servations, we only have a few degrees of freedom to confine Fig. 3. Example of a simultaneous fit to the emission line of a galaxy with superimposed night sky lines. The top panel shows the observed [Oii] emission, which has three night sky lines su- perimposed. The mid panel shows the isolated components of the night sky lines after the fit, and the bottom panel shows the reconstructed [Oii] emission line doublet. of our classification scheme and a description of each individual target. 3.3. A powerful diagnostic of the classification 3.3.1. Measuring the discrepancy from a pure rotational disk Flores et al. (2006) developed a diagnostic method to test the validity of the classification, and to quantify the deviation of a given VF from that of a pure rotational disk. First, we identify the VF, based on the largest scale motion in each galaxy. We then assume that this VF is the result of pure rotational motion, whatever the true nature of the dynamics. We then model the VF of a rotating disk that matches the observed velocity gradient es- timated from the measured minimal and maximal velocities to obtain the expected σ-map corresponding to the observed VF (a "model VF"). In particular, we take into account that most of the velocity gradient in the central region (from 45% to 70%) falls within one spatial pixel of GIRAFFE, as we observe in well- identified rotating disks. To generate the model VF and σ-map shown in Fig. 2, we then use a single rotation curve that con- centrates an equivalent fraction of the velocity gradient in one GIRAFFE pixel. While this should not affect the location of the σ peak, we are aware that this simplistic assumption may affect the σ peak intensity for some galaxies with flatter rotation curves or with a prominent bulge. Finally, for model σ-map, we assume that the intrinsic dispersion (due to the random gas motion in the disk) is the smallest dispersion observed in the data. The model VF for each galaxy has then been generated by assuming a rotation curve that is been used to generate a model data cube. Note that, whenever possible, we tried to align the model rotational axis with the major axis of the galaxy, in agree- ment with the rotating disk hypothesis. Unlike what was done in Flores et al. (2006), we do not try to correct for inclination, as the observed and the model VFs are affected in a similar way by inclination effects. However, note that the inclination is used 12 Y. Yang et al.: IMAGES I. Observations Fig. 4. Diagnostic diagram of the galaxy kinematics. ∆r is the spatial separation between the peaks of the modeled and the ob- served σ-maps.  gives the relative difference between the veloc- ity dispersions of the model and the observation at the reference location of the peak of the model σ-map. The blue dots, green squares and red triangles represent the RD, PR and CK galaxies, respectively. Typical uncertainties are 0.3 spatial pixels in ∆r, and 15% in . (see Sect. 3.3 for more details.) Fig. 5. The χ2-diagram for the best-fitting rotation model of galaxy kinematics. The reduced χ2 for VF-map and σ-map are set to be abscissa and ordinates, respectively. The blue dots, green squares and red triangles represent the RD, PR and CK galaxies, respectively. The two black crosses indicate the two simulated galaxy pairs: ARP271 and KPG468 (see Sect. 3.3.1 for more details). Although the CKs tend to have higher χ2 than the PRs and the RDs, it is still less efficient to identify different kinematics with this diagram. (see Sect. 3.3.2 for more discus- sion.) the model. Hence, χ2-test may not be an ideal tool for compar- ing the model and the observations. A series of tests have been performed in order to explore the validity of the χ2 estimator of goodness of the fits. In Fig. 5 we present a best-fitting χ2- diagram for the CDFS sample. First, we fit the observational VF-map with a rotation model, then we calculate the reduced χ2 which is set as abscissa. The corresponding σ-map is gener- ated following the same method described in the previous sec- tion, then the reduced χ2 for the σ-map is computed and pre- sented as ordinates. The two simulated galaxy pairs, ARP271 and KPG468 (see the caption of Fig. 4 for more details), are also plotted. Although, the CK systems tend to have larger χ2 than the RD and PR, it is difficult to distinguish different kinematical classes with such diagram. Taking into account the spatial resolution of the GIRAFFE IFU, the most important features of a rotating disk are (a) a well ordered VF with dynamical axis following the optical major- axis of the galaxy, (b) a clear σ-peak located at the position of the galaxy center, (c) and the σ-peak having a reasonable value, which can be generated by rotation. These important features are somehow diluted by the χ2-test as it is presented in Fig. 5. For example, J033230.78-275455.0, a RD galaxy, has a reduced χ2 close to 1 in both VF- and σ- maps because of its relatively low averaged S/N (= 4); other galaxies (which have typically aver- aged S/N ∼ 8) have χ2 of ∼ 10. The χ2-test appears to be too sensitive to the averaged S/N. Furthermore, using this type of test we cannot distinguish whether one of the simulated galaxy pairs (KPG468) is a RD or a PR, since the χ2 value of this pair is similar to that of both RDs and PRs. Since the χ2 values are to some extent scaled by the errors that are related to the S/N, we also tried to apply an unweighted χ2-test. In this case, we still have the same disordered distribution over the χ2 diagram. Consequently, than our method (see Sect. 3.3.1 and Fig. 4) in recognizing disk kine- the χ2-test seems less efficient matics from perturbed and complex kinematics. Therefore, we choose to compare the observations with the rotating disk model using the methodology developed in Sect. 3.3.1. 3.3.3. Error estimates We used Monte-Carlo simulations to evaluate the uncertainty of our velocity dispersion measurements. First, we generate ∼50 000 artificial [Oii] emission line doublets with S/N rang- ing from 3 to 40. The flux ratio within the doublets is fixed at 1.4. Simulated dispersions range from 10 to 100 km s−1, cor- responding to the minimum and maximum of our observations, respectively. In order to remove possible artefacts in the data, we analyzed the spectrum of each spatial pixel manually in each data cube be- fore thoroughly analyzing the spectral fit of each individual line. To ensure that the Monte-Carlo simulations are a fair represen- tation of this process, we did not only consider the results from automatic fitting routines, but use the same procedure that was used for the observed data-cubes. This is particularly important for spectra with low S/N (i.e., 3 – 5) spectra, and it somewhat reduces the uncertainties of our measurement. We then investigated the uncertainties in the velocity disper- sion as a function of the S/N. Figure 6 shows the relative uncer- tainty in the velocity dispersion (∆σ/σ) as a function of the S/N. We find a slight trend that we possibly underestimate velocity dis- persion when we have spectra with S/N lower than ∼ 5, because the noise affects the line wings in particular. However, this systematic uncertainty is negligible compared to the statistical error. Based on this result, we estimate the uncertainty of the observed velocity dispersion at a given S/N. Standard error propagation then yields the uncertainty on  (Eq. 1). Typically, we find that uncertainties Y. Yang et al.: IMAGES I. Observations 13 are ∼ 15%, ranging from 5% to 25% (see Table 2 for the uncer- tainties of individual objects). Fig. 6. The red solid line indicates the median differences be- tween the measured and the simulated σ values varying with the S/N. The two dashed lines give the 1σ errors for each bin in S/N: typically, these amount to ∼20% at 3 <S/N <4, 15% at S/N∼10 and about 5% at S/N >15, respectively. Uncertainties in ∆r are dominated by the error on the loca- tion of the peak of the velocity dispersions. This uncertainty is dominated by the profile of the σ-map and its S/N. Again, we use Monte-Carlo simulations to quantify these uncertainties. We calculate the uncertainty of each value in our σ-maps using the S/N-map and Fig. 6. As a next step we generate a set of sim- ulated σ-maps from the observed maps by randomly shuffling the σ values within the error distribution, and derive the uncer- tainty in the peak location from the statistics of the Monte-Carlo simulations. We thus find that the largest uncertainties are related to the spatial sampling of GIRAFFE, i.e., the sub-pixel position adopted for Vmax and Vmin. In the models this yields a geometri- cal error of about 0.14 pixels in the location of the peak. For some very extended galaxies (e.g., J033230.78-275455.0 and J033226.23-274222.8), parts of the galaxy fall outside of the IFUs; in these cases, we extrapolated the position of Vmax and/or Vmin according to the partial VFs and the optical images. Here, we obviously underestimate the corresponding errors. However, most galaxies in the sample are smaller than the field of view of an individual GIRAFFE IFU and are fully sampled; since we are mainly concerned about the ensemble properties of our sam- ple, we did not include this into the total error budget. In total, the uncertainties in the observed σ-map and those found from the models together correspond to an error of 0.14 to 0.8 spatial pixels on ∆r, with a median of ∼ 0.3 pixels (see Table 2). The clear distinction between rotating disks and more com- plex kinematics may serve as additional evidence for the validity of our classification scheme. Galaxies classified as rotating disks are well concentrated near the ∆r ∼  ∼ 0 region, which implies that they are well modeled by a simple rotating disk. Galaxies classified as PR or CK fall outside of this region, and with large scatter. Note that there are five objects with ∆r ∼ 0 and  > 1. Their VFs resemble that of a rotation disk, but the amplitude of their σ-peaks show a significant deficiency. Three of them, (J033243.62-275232.6, J033219.32-274514.0 and J033234.12- 273953.5) have an obviously disturbed σ-map and VF, which is why we classify them as having perturbed or complex kine- matics. The two remaining galaxies (J033233.90-274237.9 and CFRS031032) are dominated by bulges that are much bluer than those of present day galaxies and presumably are experiencing star formation (see Neichel et al. 2007). The presence of a star forming bulge may affect the central velocity dispersion in such a way that a pure rotating disk model underestimates the true σ-peak amplitude. It is then possible that these two galaxies are indeed supported by the combination of a rotation and disper- sion, as expected for S0 galaxies. During the final phase of the classification process (see Table 2), we have indeed used Fig. 4 to verify our results. It had led us to change one galaxy from PR to RD class (J033230.78- 275455.0, see discussion above) and one galaxy from RD to PR class (J033248.28-275028.9), thus evidencing that classification errors are marginal. Furthermore, one PR galaxy (J033219.61- 274831.0) falls close to the region of rotating disks and has rel- atively large error bars in ∆r (see Fig. 4 and Table 2). Its σ-peak is located near the edge of the σ-map, which causes a relatively large uncertainty on the peak location. However, given the S/N of the σ-peak and the presence of a secondary peak near the galaxy center, there is no doubt about its classification as PR. 3.4. Comments on individual targets J033212.39-274353.6: Kinematically classified as a RD. The dynamical major axis is aligned with the morphological major axis. The σ-map shows a clear peak near the center. J033219.68-275023.6: This is an excellent example of a RD. The velocity gradient and the sigma peak are well consistent with a RD, while the morphology clearly shows spiral arms. J033230.78-275455.0: It has the morphology of a spiral galaxy. Its [Oii] emission is relatively faint, and only the lower half of the galaxy was detected. Nevertheless, the peak of the σ-map is well centered on the nucleus of the galaxy. The σ-map is in agreement with our simple rotational model, where the center and PA have been chosen according to the geometry of the optical image. Thus, we classify its VF to be a RD. J033231.58-274121.6: The kinematics indicates that this is a RD, in spite of its irregular morphology (see Neichel et al. 2007). J033234.04-275009.7: We find a well ordered velocity gradient consistent with disk rotation, and dispersion peaks near the morphological and kinematic center. Notice that the σ peak falls just between two GIRAFFE pixels. This galaxy is classified as a RD, in spite of its irregular morphology (Neichel et al. 2007). J033237.54-274838.9: Kinematically, this galaxy appears to be a RD. Morphologically, we find a nearly face-on disk galaxy with an asymmetric outer region and prominent arms. Both the optical images and kinematic maps show the typical features of a rotating system. J033238.60-274631.4: The VF shows the characteristics of a RD. Morphologically it has been classified as an S0 galaxy. Thus, we classify this galaxy as a RD. J033241.88-274853.9: A RD with a dynamical axis centered on the luminous peak of the red HST/ACS image. The morphology appears however asymmetric, and this galaxy has been classified as peculiar/tadpole by Neichel et al. (2007). Indeed, the ionized gas detected by the IFU has no stellar counterpart on one side of the galaxy. J033245.11-274724.0: We classify this galaxy as a RD, in spite of a secondary peak, which is offset from the dispersion map. However, this is a compact object with symmetric spiral arms. The double peak in the dispersion map can be reproduced by our simulation and may be an artefact caused by the relatively low spatial resolution of the data, due to the compactness of the source. J033210.25-274819.5: The VF is not well aligned with the major axis of the galaxy because of one high S/N pixel with the highest velocity. The σ peak is found to be close to this pixel, 14 Y. Yang et al.: IMAGES I. Observations and cannot be reproduced by rotation along the major axis, implying that the rotation is perturbed. J033214.97-275005.5: The peak of the σ-map significantly deviates from the center of this face-on galaxy, we therefore classify it as a PR. J033219.61-274831.0: Another PR for which the σ-map shows a peak that is well offset from the dynamical center. Morphologically, this appears to be a peculiar, perhaps merging galaxy (Neichel et al. 2007). In the spectrum with the largest velocity dispersion, we identify a narrow component superim- posed onto a broad component. This may be related to merging. J033226.23-274222.8: The VF of this galaxy is classified as PR. The VF shows the dynamical axis parallel to the morpho- logical major axis. However, the peak in the σ-map is not well aligned with this axis. This perturbed component in the σ-map corresponds to a blue clump in the HST data (Neichel et al. 2007), which might be a gas rich satellite galaxy. J033232.96-274106.8: The peak in the σ-map is offset from the dynamical axis. Thus, this is clearly a PR. An unresolved component near the bottom left corner of the kinematic maps may correspond to a small companion of this galaxy, which perhaps is the cause of the perturbation. J033233.90-274237.9: The σ-map shows a peak near the dynamical center. Our rotating disk model can reproduce the σ-peak in position but not its amplitude. The morphological analysis by Neichel et al. (2007) suggests that this is a S0 galaxy with a significant bulge (B/T=0.8). Our purely rotating disk model does not take into account the bulge: if we add a bulge with a typical velocity dispersion of ∼ 50 km s−1, then we are able to reproduce the data. Conservatively, we classify this galaxy as a PR, although the perturbation may simply be caused by the prominent bulge. J033239.04-274132.4: The dynamical and morphological major axes are aligned, but the peak of the σ-map is strongly offset from the this axis. Its VF is obviously a PR. The center of this galaxy shows very luminous [Oii] line emission. J033243.62-275232.6: Classified as a PR because the σ-map does not show a single peak but a very extended large σ region, which cannot be reproduced with our rotating disk model. J033248.28-275028.9: The VF is classified as PR. The σ-map shows an elongated peak across two spatial pixels, and has a secondary peak. The extended peak of the σ-map cannot be reproduced by our model of a rotating disk. The overall blue colour (Neichel et al. 2007) of the galaxy may be related to enhanced star-formation as a result of the perturbed gravitational potential. J033249.53-274630.0: The VF of this galaxy is classified as PR because of the irregularity of its σ-map, which cannot be reproduced assuming pure rotation. The morphological analysis by Neichel et al. (2007) suggests that this may be a merger. J033250.53-274800.7: We detect the dynamical axis follows the major axis of the galaxy, but the peak of the σ-map is clearly offset from the center. J033210.76-274234.6: It is classified to be CK without evidence for a dynamical axis. The peak of the σ-map is not located at the center of the galaxy. J033213.06-274204.8: It has possibly a spiral morphology (maybe with a bar), but the kinematics are complex. We find the dynamical axis is clearly misaligned with the optical major axis. The σ-map is irregular without any peak at the center, and with an overall relatively high velocity dispersion (∼ 50 km s−1) over the whole field. J033217.62-274257.4: The dynamical axis is misaligned with the major axis, implying complex kinematics. We also detected that that a very broad and high dispersion σ-peak covers the majority of the galaxy. J033219.32-274514.0: Its kinematics is complex. Both VF and σ-map are perturbed. No dynamical axis can be determined from the VF. J033220.48-275143.9: It clearly shows a CK, from both VF and σ-map patterns. J033224.60-274428.1: A complex VF, which shows a velocity gradient along one component major axis while the σ-map shows a very large region of high dispersion region that deviates from the center of galaxy. Its morphology resembles that of an on-going merger. J033225.26-274524.0: The VF shows a well ordered velocity gradient but skewed. Its σ-map does not show any peak near the center. Thus, it was classified to be a CK. J033227.07-274404.7: We detect a very small velocity gradient over most of the galaxy, with an amplitude of less than 20 km/s. The top end clump in the optical image is responsible for the highest velocity detected. The peak of the σ-map is offset from the center of the galaxy and corresponds to the maximum gradient in the VF. Its VF and coma-like morphology make us suspect that this is a merger between two or three galaxies. J033228.48-274826.6: Both VF and σ-map show the signs of perturbation. No clear dynamical axis can be found. Its morphology is classified as irregular. J033230.43-275304.0: Its VF is classified CK, because the dynamical axis is not oriented along the major axis and the VF does not show a clear velocity gradient consistent with rotation. The peak of the σ-map is strongly offset from the center of the galaxy. J033230.57-274518.2: It is classified as possessing complex kinematics. Its VF shows a very asymmetric gradient while the σ-map shows a clear difference from a single peak pattern. J033234.12-273953.5: This galaxy has CK. We have detected a narrow VF and a low velocity dispersion for this object. The peak of the σ-map is not corresponding to the center of the galaxy. It has an irregular morphology and a nearby companion, which we do not detect in the [Oii] line emission. This may be an on-going merger or a simple projection of two objects at different redshift. We also notice that VVDS misidentified the [Oii] emission of this galaxy. J033239.72-275154.7: This is a galaxy with CK showing no evidence for rotation. Its dynamical axis seems to be parallel to a possible bar-like structure but it is clearly offset from this structure. Because of its low inclination, it is not expected to have a large amplitude of the velocity field. The σ-map shows a peak clearly deviating from the galaxy center. J033240.04-274418.6: The kinematics are classified as com- plex. Its VF is perturbed and its σ-map is offset from the center. J033244.20-274733.5: We find no evidence for rotation in the VF. The σ-map shows a single peak close to the center of galaxy, but the central velocity dispersion is much higher than expected for a rotating system. J033250.24-274538.9: This is a low surface brightness galaxy with complex kinematics showing no evidence for rotation. Its large large amplitude of the VF is unexpected for such a low inclination system. Both VF and σ-map show deviations from a normal RD. 4. Discussion Table 3 shows how the galaxies in our sample fall into the differ- ent kinematical classes. We find 32% RDs, 25% PRs and 43% Table 3. Statistics of each kinematical class in different fields for the complete sample (63 galaxies with MJ(AB)<−20.3). Y. Yang et al.: IMAGES I. Observations 15 HDFS CFRS22h CFRS03h CDFS Total ( fraction) 20 (32%±12%) 16 (25%±12%) 27 (43%±12%) 6 (9%) 9 10 16 3 RD PR CK UC 4 2 3 2 2 2 3 in total 5 2 6 Note: the first column gives the kinematical classifications (see Sect. 3.2 for details): RD-rotating disks; PR-perturbed rotations; CK-complex kinematics; UC-unclassified. Five galaxies (CFRS220619, CFRS220919, HDFS4070, HDFS4090, J033243.62-275232.6) are not included in this statistic because they are fainter than the limiting magnitude of MJ(AB) =−20.3. CKs which are limited by an error of 12%. This confirms the preliminary result of Flores et al. (2006) that at z = 0.4 – 0.75, few massive emission line galaxies are kinematically relaxed. Furthermore, the combined two samples include galaxies from 4 different fields, namely the CDFS, HDFS, CFRS03h and CFRS22h. We refer to the later three fields as the GTO sample (see Flores et al. 2006). It is based on 3 different lines of sight, therefore its result is unlikely to be affected by the cosmic vari- ance effect. The GTO sample gives RD:PR:CK = 11:6:11 (i.e., 39%:22%:39%), which is consistent with the results from CDFS (25%:28%:47%) within the statistical error. Notice however that in the field of CDFS we discovered a relatively large fraction of CKs, which is possibly related to the presence of large-scale structure (Ravikumar et al. 2007). In total our sample includes 6 galaxies that are unclassified (UC), because the data are not well resolved spatially. The relatively small fraction of RD galaxies is intriguing, in particular if compared with the significantly larger fraction of disk galaxies found at low redshift (see below). We emphasize the robustness of this result, in particular because: – the sample is representative of galaxies with Mstellar ≥ 1.5 × 1010M(cid:12) (see Sect. 2.3 and Fig. 2.3); – it is unaffected by cosmic variance since galaxies are selected from four independent fields, and we find that the fraction of galaxies with a particular classification does not vary signif- icantly from field to field (see Table 3); – it is based on a representative sample of 63 galaxies, and the uncertainties to the above fractions are smaller than 12%. Let us now consider the general population of galaxies at z =0.4 – 0.75, with Mstellar≥1.5×1010M(cid:12), including emission line galaxies (with W0([Oii])≥ 15 Å) like those studied in this paper, and also more quiescent galaxies with very faint or without emis- sion lines (e.g., quiescent late type galaxies such as E/S0 and some early type spirals). Hammer et al. (1997, see also Hammer et al. 2005) have found that at z =0.65 (the average redshift of our sample), 60% of intermediate mass galaxies have emission lines, and this result has been confirmed by all galaxy surveys. We fur- ther assume that all quiescent galaxies have relaxed kinematics, e.g., pressure (or dispersion) supported for spheroids and rota- tionally supported for disks. This implies that at z =0.4 – 0.75, at least 41±7% of galaxies are not dynamically relaxed, including 26± 7% of galaxies with CKs. This result is in sharp contrast with the kinematics of present-day galaxies, which are almost all relaxed, and indicates a strong evolution over the last 5 Gyr. Indeed, at z = 0, we find that 70% of intermediate mass galax- ies are spirals, while irregulars, compact galaxies and mergers contribute to less than 1% (e.g., Hammer et al. 2005). This result is unlikely to be affected by artefacts of our methodology. To illustrate this, we will now critically evaluate the possible sources of error leading to misclassifications, and quantify their impact on the ensemble properties of our sample. As discussed earlier (Sect. 3.4), we suspect that two galaxies classified as a perturbed rotation may in fact have an enhanced central velocity dispersion due to the effect of a prominent star- forming, and possibly gas-rich, bulge. This does not change our result significantly, as it would only reduce the fraction of kine- matically perturbed galaxies from 41% to 39%). Moreover, for objects with small spatial coverage, i.e., galaxies extending over less than 6 spatial pixels, our classification may be less robust than for more extended galaxies. This is the case for a few com- pact galaxies, in particular those with half light radii less than one GIRAFFE pixel (0.52(cid:48)(cid:48)), and for a few more extended galax- ies that have relatively weak emission (i.e., a mean S/N < 4). However, these galaxies represent less than 10% of the whole sample. For more than 90% of the sample, the kinematics are well sampled, with a median spatial coverage of 9 pixels at S/N > 4), allowing us to robustly and uniquely classify the kinemat- ics. We note explicitly that we verified that the classification does not depend on the mean S/N of the galaxies. Moreover, Flores et al. (2006) and Puech et al. (2007) have convincingly shown that galaxies with non-relaxed kinematics are responsible for the large dispersions of both the Tully-Fisher and the jdisk–Vmax relationships. So it is beyond doubt that a significant fraction of z = 0.4 – 0.75 galaxies have kinematics that deviate signifi- cantly from those of their local descendants, i.e., the present-day intermediate-mass galaxies, which include 70% of spirals. Which physical process could explain such a dramatic evo- lution in the kinematics of galaxies within a relatively modest amount of time (4 – 7 Gyr)? The morphology of several galax- ies in the sample strongly advocates that merging is one such process. For example, a minor merger may cause perturbed ro- tation: the in-fall of a gas-rich satellite would unavoidably lead to a local increase of the dispersion shifting the peak of the σ- map. A major merger will significantly affect a rotational VF by destroying the pre-existing disk, and lead to a signature resem- bling a complex VF (see Puech et al. 2006a and 2007). It is then probable that merging may explain most of the discrepancies in the observed VFs at z =0.4 – 0.75. However, this fact alone does not necessarily imply that merging is the only physical process explaining the strong evolution of galaxy kinematics. If major mergers are responsible for complex VFs, then 26± 7% of the galaxies within z = 0.4 – 0.75 will either be on- going mergers or merger remnants. This has to be compared with only 5± 1% of on-going mergers, as revealed by pair counts, two-point correlation or morphological classifiers (see a sum- mary in Hammer et al. 2007 and references therein). Indeed, these morphological analyzes essentially account for the ap- proaching phase of a merger, while kinematics are affected by large scale peculiar motions induced before and after the merger. This leads Hammer et al. (2007) to postulate that the merger remnant phase may be 4 to 5 times longer than the approaching phase. Assuming 0.35 – 0.4 Gyr for the duration of the latter, this 16 Y. Yang et al.: IMAGES I. Observations results in a merger remnant phase with a duration of 1.5 – 2 Gyrs. Indeed, simulations by Robertson et al. (2006) and Governato et al. (2007) predict such a duration for the rebuilding of a disk after a major merger of gas rich galaxies. Furthermore, the re- quirement for gas rich interacting galaxies in Robertson et al. (2006), finds some support from the evolution of the gas content in galaxies as a function of cosmic time, although this evolution is derived indirectly from the gas phase metal abundance in dis- tant galaxies. Liang et al. (2006) estimated that the gas content in intermediate-mass galaxies at z ∼ 0.6 was two times larger than in galaxies at the current epoch. However, the predominance of mergers, and especially ma- jor mergers, leads to the requirement that many present-day galactic disks have in fact been rebuilt at a relatively recent epoch. Taking into account the number fraction of both on-going mergers and galaxies with complex VFs, Hammer et al. (2005, 2007) conclude that between 50% and 70% of galaxies may have experienced a major merger and subsequent disk rebuilding since z =1. Although this alternative may explain many observa- tions in the z = 0.4−1 redshift range (e.g., Hammer et al. 2005), could a less dramatic mechanism be at the origin of the peculiar kinematics z =0.4 – 0.75? In fact our observations can only account for the large scale motions traced by the ionized gas and not those of the stellar component. Observations of the latter are certainly crucial, al- though for most distant galaxies this is beyond the reach of 8 to 10 meter telescopes. One may then postulate that rapid gas mo- tions may be superimposed on a normal rotational stellar com- ponent. There are two difficulties with such an assumption. First, at z = 0.4 – 0.75 the gaseous fraction is much higher than today, and represents a significant fraction of the baryonic mass. To illustrate this, recall that since z = 1, about half of the present- day stellar mass has been formed from gas in intermediate mass galaxies (Hammer et al. 2005; Bell et al. 2005). One may then wonder how under such conditions an ordered rotational VF of about half of the baryons (stars) may survive against highly per- turbed VFs for the other half of the baryons (gas). The origin of a large-scale motion gas component is certainly a second diffi- culty, especially if no merger is advocated. Another possibility may be to invoke internal processes, such as bars, that may per- turb the gaseous VFs. One may wonder whether the presence of bars in the central region could create additional dispersion leading to apparently unrelaxed VFs according to our VF diag- nostic (see Fig. 4). However, our spatial resolution may be too poor to kinematically characterize most of the bars, except possi- bly the most giant ones. Could some vigorously enhanced inter- nal processes contribute to the complex VFs? Indeed, this is not expected from the apparent non-evolution of the frequency of barred galaxies (e.g., Zheng et al. 2005 and references therein), which does not support that bars are the process to explain the large evolution of galaxy kinematics. The observed motions at large scales rather suggest another mechanism, probably related to asymmetric gas accretions (such as provided by a merger) or perhaps to gas outflows. 5. Conclusion We have been able to measure the VFs of 36 galaxies at z = 0.4 – 0.75 using deep exposures of the multiplex integral-field spectrograph GIRAFFE at the VLT in the multi-IFU mode, mea- suring the kinematics of the spatially and spectrally well re- solved [Oii] emission line doublet. In combination with a sim- ilar study by Flores et al. (2006), we have a relatively large and representative sample of 63 galaxies with Mstellar ≥ 1.5 × 1010M(cid:12). Thus, our results are representative for the population of intermediate-mass galaxies in this redshift range, and it cannot be affected by cosmic variance. To date, this is the only existing representative sample of distant galaxies with measured VFs. We confirm and consolidate the results of Flores et al. (2006), that a significant fraction of intermediate mass galaxies had perturbed or complex kinematics 5 Gyrs ago. Our method to classify the kinematics of the galaxies is particularly robust. It attributes a large weight to the velocity dispersion in the central region of the galaxy, where the S/N are the highest. Even if we assume that all quiescent galaxies at z = 0.4 – 0.75 had well or- dered VFs, we find that 41± 7% of the galaxies are not kinemat- ically relaxed, including 26± 7% of galaxies that show complex kinematics. Undoubtedly, galaxy kinematics are evolving very rapidly, since most present-day galaxies in the same mass range are likely to have ordered VFs. This result may be combined with the fact that anomalous VFs are responsible for most of the large observed dispersion of both the Tully-Fisher and the jdisk–Vmax relationships (see Flores et al. 2006 and Puech et al. 2007). It suggests a random walk evo- lution of galaxies related to a high fraction of merging events, in- cluding major mergers (Puech et al. 2007). Mergers may indeed reproduce all the peculiar kinematics at z = 0.4 – 0.75, as well as being responsible for the dispersion of fundamental relations. Other mechanisms, such as in-fall of high velocity gas clouds, gas outflows or bars may also contribute to the observed evolu- tion in the kinematics. To understand their influences, and more- over the underlying mechanisms that activate them, requires de- tail analyzes of individual objects and as a prerequisite, a full model of the significance of the GIRAFFE measurements. If major merging is the main mechanism responsible for the large fraction of complex VFs, then this implies that, since z =1, from 50% to 70% of intermediate mass galaxies have experienced a major merger. This is quantitatively in good agreement with the spiral rebuilding scenario as proposed by Hammer et al. (2005). Acknowledgements. We would like to thank the referee for helpful com- ments and discussions on the paper. We thank all the GIRAFFE team at the Observatories in Paris and Geneva, and ESO for this unique instrument. We are grateful to Albrecht Rudiger for helping us in the writing of the paper. We thank Francoise Combes and Denis Burgarella for their useful comments and for their kind support of this program since the beginning, as well as Andrea Cimatti, Emanuele Daddi, David Elbaz, Olivia Garrido, Dominique Proust, Xianzhong Zheng. References Barnes, J. E., & Hernquist, L. 1996, ApJ, 471, 115 Barton, E. J., Geller, M. J., & Kenyon, S. J. 2000, ApJ, 530, 660 Bell, E. F., McIntosh, D. H., Katz, N., & Weinberg, M. D. 2003, ApJS, 149, 289 Bell, E. F., et al. 2005, ApJ, 625, 23 Bell, E. F., Phleps, S., Somerville, R. S., Wolf, C., Borch, A., & Meisenheimer, K. 2006, ApJ, 652, 270 Bertin, E., Arnouts, S. 1996, A&AS, 117, 393 Conselice, C. J., Gallagher, J. S., III, & Wyse, R. F. G. 2003, AJ, 125, 66 Conselice, C. J., Bundy, K., Ellis, R. S., Brichmann, J., Vogt, N. P., & Phillips, A. C. 2005, ApJ, 628, 160 Cowie, L. L., Songaila, A., & Barger, A. J. 1999, AJ, 118, 603 Dressler, A. 2004, Clusters of Galaxies: Probes of Cosmological Structure and Galaxy Evolution, 206 Ferreras, I., & Silk, J. 2001, ApJ, 557, 165 Ferreras, I., Silk, J., Bohm, A., & Ziegler, B. 2004, MNRAS, 355, 64 Flores, H., et al. 1999, ApJ, 517, 148 Flores, H., Hammer, F., Puech, M., Amram, P., & Balkowski, C. 2006, A&A, 455, 107 Fuentes-Carrera, I., et al. 2004, A&A, 415, 451 Giovanelli, R., Haynes, M. P., da Costa, L. N., Freudling, W., Salzer, J. J., & Wegner, G. 1997, ApJ, 477, L1 Governato, F., Willman, B., Mayer, L., Brooks, A., Stinson, G., Valenzuela, O., Wadsley, J., & Quinn, T. 2007, MNRAS, 374, 1479 Y. Yang et al.: IMAGES I. Observations 17 Guzman, R. 1999, in Proc. 19th Moriond Conf., Building Galaxies: From the Primordial Universe to the Present, ed. F. Hammer, T. X. Thuan, V. Cayatte, B. Guiderdoni, & J. T. T. Van (Paris: Editions Frontires), 269 Haarsma, D. B., Partridge, R. B., Windhorst, R. A., & Richards, E. A. 2000, ApJ, Hammer, F., et al. 1997, ApJ, 481, 49 Hammer, F., Gruel, N., Thuan, T. X., Flores, H., & Infante, L. 2001, ApJ, 550, 544, 641 570 Hammer, F., Flores, H., Elbaz, D., Zheng, X. Z., Liang, Y. C., & Cesarsky, C. 2005, A&A, 430, 115 Hammer, F., Puech, M., Chemin, L., Flores, H., & Lehnert, M. 2007, ArXiv Astrophysics e-prints, arXiv:astro-ph/0702585 Heavens, A., Panter, B., Jimenez, R., & Dunlop, J. 2004, Nature, 428, 625 Hern´andez-Toledo, H. M., Fuentes-Carrera, I., Rosado, M., Cruz-Gonz´alez, I., Franco-Balderas, A., & Dultzin-Hacyan, D. 2003, A&A, 412, 669 Hopkins, A. M. 2004, ApJ, 615, 209 Kannappan, S. J., & Barton, E. J. 2004, AJ, 127, 2694 Koo, D. C., Guzman, R., Faber, S. M., Illingworth, G. D., Bershady, M. A., Kron, R. G., & Takamiya, M. 1995, ApJ, 440, L49 Le F`evre, O., et al. 2000, MNRAS, 311, 565 Liang, Y. C., Hammer, F., & Flores, H. 2006, A&A, 447, 113 Lilly, S. J., Le Fevre, O., Hammer, F., & Crampton, D. 1996, ApJ, 460, L1 Lotz, J. M., Madau, P., Giavalisco, M., Primack, J., & Ferguson, H. C. 2006, ApJ, 636, 592 Madau, P., Ferguson, H. C., Dickinson, M. E., Giavalisco, M., Steidel, C. C., & Fruchter, A. 1996, MNRAS, 283, 1388 Neichel et al. 2007, in preparation Noeske, K. G., Koo, D. C., Phillips, A. C., Willmer, C. N. A., Melbourne, J., Gil de Paz, A., & Papaderos, P. 2006, ApJ, 640, L143 Ostlin, G., Amram, P., Bergvall, N., Masegosa, J., Boulesteix, J., & M´arquez, I. 2001, A&A, 374, 800 Portinari, L., & Sommer-Larsen, J. 2007, MNRAS, 375, 913 Pozzetti, L., et al. 2003, A&A, 402, 837 Puech, M., Flores, H., Hammer, F., & Lehnert, M. D. 2006a, A&A, 455, 131 Puech, M., Hammer, F., Flores, H., Ostlin, G., & Marquart, T. 2006b, A&A, 455, Puech, M., Hammer, F., Lehnert, M. D., & Flores, H. 2007, A&A, 466, 83 Ravikumar, C. D., et al. 2007, A&A, 465, 1099 Rawat, A., Kembhavi, A. K., Hammer, F., Flores, H., & Barway, S. 2007, ArXiv e-prints, 704, arXiv:0704.2177 Robertson, B., Bullock, J. S., Cox, T. J., Di Matteo, T., Hernquist, L., Springel, V., & Yoshida, N. 2006, ApJ, 645, 986 Tully, R. B., & Fisher, J. R. 1977, A&A, 54, 661 Weiner, B. J., et al. 2006, ApJ, 653, 1027 Werk, J. K., Jangren, A., & Salzer, J. J. 2004, ApJ, 617, 1004 Wilson, G., Cowie, L. L., Barger, A. J., & Burke, D. J. 2002, AJ, 124, 1258 Zheng, X. Z., Hammer, F., Flores, H., Ass´emat, F., & Rawat, A. 2005, A&A, 435, 507 119
0707.1930
1
0707
2007-07-13T06:59:51
Improved Simulation of the Mass Charging for ASTROD I
[ "astro-ph" ]
The electrostatic charging of the test mass in ASTROD I (Astrodynamical Space Test of Relativity using Optical Devices I) mission can affect the quality of the science data as a result of spurious Coulomb and Lorentz forces. To estimate the size of the resultant disturbances, credible predictions of charging rates and the charging noise are required. Using the GEANT4 software toolkit, we present a detailed Monte Carlo simulation of the ASTROD I test mass charging due to exposure of the spacecraft to galactic cosmic-ray (GCR) protons and alpha particles (3He, 4He) in the space environment. A positive charging rate of 33.3 e+/s at solar minimum is obtained. This figure reduces by 50% at solar maximum. Based on this charging rate and factoring in the contribution of minor cosmic-ray components, we calculate the acceleration noise and stiffness associated with charging. We conclude that the acceleration noise arising from Coulomb and Lorentz effects are well below the ASTROD I acceleration noise limit at 0.1 mHz both at solar minimum and maximum. The coherent Fourier components due to charging are investigated, it needs to be studied carefully in order to ensure that these do not compromise the quality of science data in the ASTROD I mission.
astro-ph
astro-ph
October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd International Journal of Modern Physics D c(cid:13) World Scientific Publishing Company 7 0 0 2 l u J 3 1 ] h p - o r t s a [ 1 v 0 3 9 1 . 7 0 7 0 : v i X r a Improved Simulation of the Mass Charging for ASTROD I GANG BAO1,2,3, WEI-TOU NI1,3, D. N. A. SHAUL4, H. M. ARAUJO4, LEI LIU1,2, T. J. SUMNER4 1. Center for Gravitation and Cosmology, Purple Mountain Observatory, Chinese Academy of Sciences, Beijing West Road No.2, Nanjing, 210008, China 2. Graduate University of Chinese Academy of Sciences, Beijing, 100049, China [email protected] 3. National Astronomical Observatories, Chinese Academy of Sciences, Beijing, 100049, China 4. Department of Physics, Imperial College London, London, SW7 2BZ, UK Received Day Month Year Revised Day Month Year Communicated by The electrostatic charging of the test mass in ASTROD I (Astrodynamical Space Test of Relativity using Optical Devices I) mission can affect the quality of the science data as a result of spurious Coulomb and Lorentz forces. To estimate the size of the resultant disturbances, credible predictions of charging rates and the charging noise are required. Using the GEANT4 software toolkit, we present a detailed Monte Carlo simulation of the ASTROD I test mass charging due to exposure of the spacecraft to galactic cosmic- ray (GCR) protons and alpha particles (3He, 4He) in the space environment. A positive charging rate of 33.3 e+/s at solar minimum is obtained. This figure reduces by 50% at solar maximum. Based on this charging rate and factoring in the contribution of minor cosmic-ray components, we calculate the acceleration noise and stiffness associated with charging. We conclude that the acceleration noise arising from Coulomb and Lorentz effects are well below the ASTROD I acceleration noise limit at 0.1 mHz both at solar minimum and maximum. The coherent Fourier components due to charging are investi- gated, it needs to be studied carefully in order to ensure that these do not compromise the quality of science data in the ASTROD I mission. Keywords: ASTROD I; charging; GEANT4; disturbances. 1. Introduction The ASTROD I (Astrodynamical Space Test of Relativity using Optical Devices I) mission concept is a down-scaled version of ASTROD. The main objectives of ASTROD I are: to improve the precision of measurement of solar-system dynam- ics, solar-system constants and ephemeris; to measure the relativistic gravitational effects; to test the fundamental laws of space-time more precisely and to improve the measurement of the rate of change of the gravitational constant with time.1 The basic scheme of the ASTROD I space mission is to use two-way laser interfer- ometric ranging and laser pulse ranging between a drag-free ASTROD I spacecraft in a solar orbit and deep space laser stations on Earth. The ASTROD I spacecraft 1 October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd 2 G. BAO et al. is 3-axis stabilized with a total mass of 300-350 kg. The mass of the payload is 100-120 kg. The science data rate is 500 bps. The spacecraft is cylindrical with OD (Outer Diameter) 2.5 m and height 2 m and has its side surface covered with solar panels. In orbit, the cylindrical axis will be perpendicular to the orbit plane with the telescope pointing toward the ground laser station. The effective area to receive sunlight is about 5 m2 and can generate over 500 W of power.1 -- 2 The spacecraft will be launched into the solar orbit from a low earth orbit. The injection correction will be made using a medium-sized ion thruster. A launch on August 4, 2010 would provide a suitable orbit. The orbit in the X-Y plane of the heliocentric ecliptic coordinate system is shown in Figure 1.1,3 This solar orbit will initially have a period of 290 days. After two gravity-assist encounters with Venus, the period will be shortened to about 165 days. After about 370 days from launch, the spacecraft will arrive at the other side of the Sun. The spacecraft will have the first closest approach to Venus 107.8 days after launch with a distance of 31606.0 km to the centre of Venus. The spacecraft crosses the Venus trajectory in front of Venus and gets a swing toward the Sun to achieve the Venus orbital period. After the first encounter, the spacecraft has the same orbit's period as Venus and encounters Venus again after about 1 period (224.7 days) with a closest approach distance of 16151.7 km from the centre of Venus.1 -- 2 0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8 -1.0 ) U A ( s i x A Y Mercury Venus spacecraft -0.8 -0.6 -0.4 -0.2 Sun 0.4 0.6 0.8 1.0 0.0 0.2 X axis (AU) Fig. 1. The ASTROD I orbit in the X-Y plane of the heliocentric ecliptic coordinate system. To achieve its goal, the ASTROD I residual acceleration noise target is: S1/2 △a (f ) = 3 × 10−14[ 0.3 mHz f + 30( f 3 mHz )2] ms−2Hz−1/2, (1) over the frequency range of 0.1 mHz < f < 100 mHz.2 Here S1/2 △a (f ) is the residual acceleration noise spectral density. It is compared to the LISA Pathfinder LISA October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd Improved Simulation of the Mass Charging for ASTROD I 3 Technology Package,4 LISA5 and ASTROD6 noise target curves in Figure 2. 1E-9 1E-10 1E-11 1E-12 1E-13 1E-14 1E-15 y t i s n e d l a r t c e p s e s i o n n o i t a r e l e c c A ) 2 / 1 - z H 2 - s m ( ASTROD I LISA LTP 1E-16 1E-4 ASTROD 0.01 0.1 1E-3 Frequency(Hz) Fig. 2. A comparison of the target acceleration noise curves of ASTROD I, the LTP, LISA and ASTROD. The test mass is key to guarantee the drag-free condition for ASTROD I. It is a 1.75 kg, rectangular parallelepiped, made of an extremely low magnetic suscepti- bility (< 5 × 10−5) Au-Pt alloy, to minimize magnetic disturbances. The test mass is placed in the centre of spacecraft and surrounded by electrodes on all six sides. Any relative displacement of test mass to the surrounding electrodes will be capaci- tively detected and the spacecraft will follow it using FEEP (Field Emission Electric Propulsion) to ensure drag-free flight. Cosmic rays and solar energetic particles will easily penetrate the light shielding of the spacecraft to transfer heat, momentum and electrical charge to the test mass. Electrical charging is the most significant of these disturbances. It will result in forces on the test mass, due to Coulomb and Lorentz interactions, which will disturb the geodesic motion. The characteristics of the test mass charging process depend on the incident flux, spacecraft geometry and the physical processes that occur. The three main disturbances associated with this charge are an increase in the test mass acceleration noise, coupling between the test mass and the spacecraft and the appearance of coherent Fourier components in the measurement bandwidth.7 To limit the acceleration noise associated with Coulomb and Lorentz forces to meet the ASTROD I noise requirement, the test mass must be discharged in orbit. Our previous work predicted the charging rates for ASTROD I test mass from galactic cosmic rays at solar minimum using a simplified geometry, and using these predictions, estimated the magnitude of disturbances associated with charging.8 In this paper, we present the detailed calculation of the ASTROD I net test mass charging rate and shot noise, due to cosmic rays at solar minimum and solar maximum, with a more realistic geometry model. Based on these results, we estimate the magnitude of acceleration noise, stiffness and the coherent signals associated with charging. October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd 4 G. BAO et al. 2. Modelling the Charging Process 2.1. Radiation environment model We have simulated the fluxes of the 3 most abundant primaries, proton, 3He and 4He primary particles which represent approximately 98% of the total cosmic ray flux. Near-Earth cosmic ray spectra were adopted, as used in similar LISA simulations. These are shown in Figure 3.9 The primary particles are generated from points ) n / V e G ( / r S / 2 m / s / s e l c i t r a p , x u l f 1000 100 10 1 0.1 0.01 1E-3 1E-4 1E-5 0.1 Protons_solmin Protons_solmax 4He_solmin 4He_solmax 3He_solmin 3He_solmax solmin: solar minimum solmax: solar maximum 1 10 GCR Energy (GeV) 100 1000 Fig. 3. Differential energy spectra for cosmic ray protons and He nuclei at near earth orbit. For each species, the upper curve indicates the solar minimum spectrum, the lower curve indicates the solar maximum spectrum. sampled uniformly from a spherical surface of 3000 mm diameter, encompassing the whole ASTROD I geometry model. The total time T for bombardment by N0 primaries is given by T = N0/F · πR2, where F is the integral flux (unit: particles/cm2/s) for each species at solar minimum or solar maximum, and R = 1500 mm. The N th particle event occurs at time t = N · T /N0.10 The effects on charging of other particle species (C, N, O, e−) are determined separately, based on a LISA study.11 Work is under way to study the impact on the charging disturbances from variations in the incident flux due to the ASTROD I heliocentric position changes. 2.2. Physics model Test mass charging depends heavily on the physics processes that occur during the passage of the particles through matter and the geometry model used in simula- tion. The GEANT4 toolkit employs Monte Carlo particle ray-tracing techniques to follow all primary and secondary particles. Due to their high energy and hadronic nature, cosmic rays can produce the complex nuclear reactions which have large October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd Improved Simulation of the Mass Charging for ASTROD I 5 final-state multiplicities, producing many secondary particles. A low energy thresh- old of 250 eV was imposed for secondary particle production in our simulation. The physics processes simulated include electromagnetic, hadronic and photonuclear in- teractions. Fluorescence and non-radiative (Auger) atomic deexcitation have been implemented. The hadronic physics is mainly implemented by elastic and inelastic scattering processes. The inelastic reactions were based on the LEP (Low Energy Particles) and HEP (High Energy Particles) parameterized models. The inelastic re- actions also use evaporation models to treat the deexcitation of nuclei with A > 16, comprising gamma emission, fragment evaporation (p, n, α, 2H and 3H) and fission of heavier residual nuclei. A variety of decay, capture and annihilation processes has also been included in our physics processes list based on LISA GEANT4 model of Araujo et al.12 The charging potential of several additional physics processes, such as the kinetic emission of very low energy electrons which has not been modeled in the present simulation, has been assessed based on LISA studies.10 2.3. Geometry model The basic payload configuration of ASTROD I is sketched in Ref. 1. The geometry model built for ASTROD I using GEANT4 in present work is sketched in Figure 4 and Figure 5. Table 1 lists the dimensions and weight of main constituent parts of the ASTROD I geometry model. Table 2 lists the composition and density of materials used in the model. The cylindrical structure of the spacecraft consists of a layer with diameter 2.5 m, height 2 m and thickness 10 mm made of CFRP (Carbon Fibre Reinforced Plastic) honeycomb. The top and bottom of the spacecraft are covered by the upper deck and lower deck. To prevent sunlight from striking the inside directly and to reduce the temperature perturbation inside, all surfaces of spacecraft are covered by a thermal shield consistent of five layers of materials (face sheet, honey comb core, face sheet, foam, face sheet). The edges of upper deck and thermal shield are shown as large ellipses in Figure 4. The inner lower deck is shown as the grey part in Figure 4 and Figure 5. The payload structure is used for shielding the optical bench, inertial sensor and primary telescope etc. The primary telescope, which collects the incoming light is a 500 mm diameter f/1 Cassegrain telescope.13 Some 30 boxes represent the components above ∼ 0.1 kg based on the payload configuration.1 The mass of the spacecraft and payload are estimated as 341 kg and 109 kg in this study. October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd 6 G. BAO et al. Fig. 4. The schematic diagram for the geometry model with a simulated GEANT4 cosmic-ray event. Fig. 5. The overhead view of the GEANT4 geometry model for ASTROD I. The six black boxes are laser heads and the other white boxes represent the components above ∼ 0.1 kg. October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd Improved Simulation of the Mass Charging for ASTROD I 7 Table 1. Dimension and weight of the main constituent parts of the ASTROD I geometry model Constituent part Material Weight Dimension (mm) Comment Solar Panels Scell (kg) ∼61.5 Spacecraft CFRP ∼7.8 Tube: radius 1250; height 2000; thickness 0.5 Tube: radius 1250; height 2000; thickness 10 ∼100.3 Tube: radius 1240; covers the side surface of the spacecraft contains the payloads covers the spacecraft top of the spacecraft above the MLI-blanket bottom of the spacecraft Shields the optical bench, inertial sensor contains the Ti-house located in the optical bench contains the test mass collects the incoming light Shields the secondary mirror Power Condition- ing and Distribution Unit Centralised Processor System Radio Frequency Distribution Unit Thermal Shield Upper Deck Lower Deck MLI-blanket CFRP Al-Honeycomb foam CFRP Al-Honeycomb CFRP Al-Honeycomb MLImat ∼15 ∼15 ∼6.8 Payload Shield CFRP ∼15.6 Optical Bench ULEglass Ti House Ti Alloy ∼5 ∼1.9 Mo House Molybdenum ∼2.5 Test Mass AuPt Alloy ∼1.75 SiC Primary Telescope Shield/Mounting CFRP Plate Telescope Shield CFRP PCDU Al6061 Transponder Al6061 CPS Interferometer Electronic Boxes Gyroscope RFDU Al6061 Al6061 Al6061 Al6061 ∼6.4 ∼1.6 ∼12.6 ∼15.9 ∼3.5 ∼15.9 ∼3.5 ∼1 ∼1 height 2000; thickness 41.8 Cylinder: radius 1240; thickness 30 Cylinder: radius 1240; thickness 30 Cylinder: radius 1240; thickness 1 Tube: radius 300; length 1000; thickness 5 Rectangular: length 350; width 200; height 40 Tube: radius 62.5; height 224; thickness 5 Cube: length 75 Rectangular: length 50; width 50; height 35 Dish: radius 250; thickness 10.5 Cylinder: radius 175; thickness 10 Tube: radius 285; length 860; thickness 5 Rectangular: length 350; width 200; height 300 Rectangular: length 220; width 184; height 178 Rectangular: length 240; width 356; height 140 Rectangular: length 200; width 200; height 150 Cylinder: radius 42.5; height 89 Rectangular: length 160; width 60; height 80 October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd 8 G. BAO et al. Table 2. The composition and density of materials used in the ASTROD I geometry model (CFRP:Carbon Fibre Reinforced Plastic) Material Vacuum Al6061 Al Honeycomb MLImat Ti Alloy AuPt Alloy ULE Glass foam SiGlass TiGlass SHAPAL SiC Scell Molybdenum Gold CFRP carbon CFRP Honeycomb Composition (by weight) gas Density (g/cm3) 1.0×10−25 Al(98%), Mg(1%), Si(0.6%), Fe(0.4%) Al(98%), Mg(1%), Si(0.6%), Fe(0.4%) H(4.1958%), C(62.5017%), O(33.3025%) Ti(90%), Al(6%), V(4%) Au(70%), Pt(30%) SiGlass(92.5%), TiGlass(7.5%) C(90%), H(10%) O: 2, Si: 1 O: 2, Ti: 1 Al: 2, N: 1 Si: 1, C: 1 Si Mo Au C C C 2.70 0.05 1.40 4.43 19.92 2.21 0.05 2.20 4.25 2.90 3.10 7.82 10.22 19.32 1.66 2.10 0.05 A 50 × 50 × 35 mm3 test mass is at the centre of the spacecraft. The test mass is housed inside capacitance sensors located in optical bench mounted behind the telescope. The test mass is surrounded by sensing and actuation electrodes lodged in a molybdenum housing. A 0.3 µm gold layer is plated on the entire inner surface of the sensor housing. The assembly is accommodated in a titanium vacuum (< 10 µPa) enclosure. The gap between test mass and electrodes along X axis or Y axis is 4 mm; that along Z axis is 2 mm.13 The GEANT4 model for inertial sensor is shown in Figure 6. Fig. 6. ASTROD I inertial sensor model implemented in Geant4. The test mass, located at the centre of the figure, is surrounded by sensing electrodes (white) and injection electrodes (grey). October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd Improved Simulation of the Mass Charging for ASTROD I 9 3. Charging Results We have run 6 independent GEANT4 simulations to determine the charging of the ASTROD I test mass by cosmic ray protons, 3He and 4He, at solar minimum and maximum. In total, about 8,500,000 events were simulated. The details of each event that resulted in test mass charging were recorded, including the event time, net charge deposited on the test mass and the energy of the primary. 3.1. Charging simulation results The variation of the net test mass charge with time, due to GCR proton, 3He and 4He fluxes are shown respectively in Figure 7, Figure 8 and Figure 9 at solar minimum and maximum. The straight lines in these figures correspond to least squares fits of the simulated data, giving mean net charging rates attributable to the proton, 3He and 4He fluxes. The proton flux is responsible for positive charging rates of 26.5 ± 0.5 e+/s at solar minimum and 9.0 ± 0.5 e+/s at solar maximum; The 3He flux is responsible for positive charging rates of 0.8 ± 0.05 e+/s at solar minimum and 0.3 ± 0.05 e+/s at solar maximum; The 4He flux is responsible for positive charging rates of 6.0 ± 0.20 e+/s at solar minimum and 2.4 ± 0.20 e+/s at solar maximum. The uncertainties quoted are only associated with the Monte Carlo fluctuations. Our simulation indicates that ∼ 97% of the charge accumulated comes from primary cosmic ray protons and 4He at both solar minimum and maximum and all three fluxes lead to positive charging of the test mass. The proton flux dominates these rates. However, 4He, which constitutes only 8% of the total cosmic rays flux, is responsible for ∼ 18% of the test mass charging at solar minimum and ∼ 20% at solar maximum. October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd 10 G. BAO et al. 300 250 200 150 100 50 ) + e ( e g r a h c t e n Proton solar minimum solar maximum 0 0 5 10 15 Time(s) 20 25 30 Fig. 7. The charging timeline for protons at solar minimum and maximum. The straight line is a least squares fit. ) + e ( e g r a h c t e n 140 120 100 80 60 40 20 0 0 3He solar minimum solar maximum 50 100 150 200 250 300 350 400 Time(s) Fig. 8. The charging timeline for 3He at solar minimum and maximum. The straight line is a least squares fit. October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd Improved Simulation of the Mass Charging for ASTROD I 11 4He solar minimum solar maximum 180 160 140 120 100 80 60 40 20 0 ) + e ( e g r a h c t e n 0 5 10 15 20 25 30 Time(s) 35 40 45 50 55 Fig. 9. The charging timeline for 4He at solar minimum and maximum. The straight line is a least squares fit. Two histograms of the net charge deposited in an event are given in Figure 10 and Figure 11, for the proton data set, showing that most events result in the transfer of one unit of charge. The effects of the positive and negative chargings cancel to some extent. An imbalance in these currents gives rise to the net positive charging rate. 200 150 100 50 s t n e v E f o r e b m u N 0 -20 -15 -10 -5 0 5 10 15 20 Net Charge Deposited in Envent Fig. 10. Histogram of the net charge deposited in an event, for incident protons at solar minimum. The total number of proton events simulated was 2,290,000. October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd 12 G. BAO et al. s t n e v E f o r e b m u N 350 300 250 200 150 100 50 0 -20 -15 -10 -5 0 5 10 15 20 Net Charge Deposited in Event Fig. 11. Histogram of the net charge deposited in an event, for incident protons at solar maximum. The total number of proton events simulated was 4,000,000. The charging rate is plotted as a function of primary energy in Figure 12. The low energy cut-off is due to the shielding provided by the spacecraft, which prevents incident protons with energies below ∼ 100 MeV from charging the test mass. At solar minimum, the most significant charging mechanism is primary cosmic ray particles stopping in the test mass. This occurs mainly for protons of energy between ∼ 100 - 720 MeV. Protons with energies in excess of ∼ 720 MeV have sufficient energy to traverse the distance through the spacecraft to the test mass and the longest path through the test mass, without being stopped. This explains the peak observed in this energy interval in Figure 12 at solar minimum. The scenario at solar maximum is distinct: a peak is visible at higher energies because the primary proton flux peak shifts towards higher energy. 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 -0.5 100 -1.0 n i b / s / + e , e t a r g n i g r a h C Solar Minimum Solar Maximum 101 102 103 104 105 106 107 GCR Energy, MeV Fig. 12. Charging rate of test mass as a function of primary proton energy at solar minimum and at solar maximum. October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd Improved Simulation of the Mass Charging for ASTROD I 13 In addition to the Monte Carlo uncertainty, we added an error of ±30% in the net charging rates to account for uncertainties in the GCR spectra, physics models and geometry implementation. Further, based on LISA studies10,11, a potential contribution to the charging rate of 28.4 e+/s at solar minimum and of 17.0 e+/s at solar maximum from kinetic low energy secondary electron emission should be considered; the effect of cosmic ray fluxes of particle species not included in this simulation is expected to increase the net charging rate by ∼ 4.2% at solar minimum and ∼ 7.3% at solar maximum13. 3.2. Charging noise Following to Ref. 12, the charging flux is considered to be made up of independent currents, Iq, each composed solely of charges qe (q=+1 for protons), with shot noise of single-sided spectral density Sq = p2qeIq, where e is the magnitude of electron charge. The total noise, SR, is then given by the quadrature sum of Sq, over all values of q. Considering the Monte Carlo currents alone, Sq = 17.6 es−1Hz−1/2 at solar minimum and Sq = 9.6 es−1Hz−1/2 at solar maximum. Based on the LISA study10, low energy secondary electron emission is estimated to contribute an extra 10.3 es−1Hz−1/2 at solar minimum and 8.0 es−1Hz−1/2 at solar maximum; the charging noises from other species (C, N, O, e−) are estimated to be 8.9 es−1Hz−1/2 at solar minimum and 3.7 es−1Hz−1/2 at solar maximum. Integrating in the time domain gives the charging fluctuations at frequency f , SQ(f ) = SR/2πf. (2) The charging rate and noise contributions from the different sources mentioned above are summarized in Table 3. Given the results in Table 3, by adding the con- tributions of all the independent sources, we estimate the total worst case charging rate to be 73.8 e+/s at solar minimum and 33.8 e+/s at solar maximum. The total worst case noise is 22.3 es−1Hz−1/2 at solar minimum and 13.0 es−1Hz−1/2 at solar maximum. Table 3. The charging rate and noise contributions from different sources. Source p 3He 4He Secondary Electron Other Species(C, N, O, e−) Uncertainty charging rate(e+/s) min max 26.5 0.8 6.0 28.4 1.4 10.0 9.0 0.3 2.4 17.0 0.9 3.5 charging noise(e/s/Hz1/2) min 15.9 2.5 7.2 10.3 8.9 - max 8.6 1.6 3.9 8.0 3.7 - October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd 14 G. BAO et al. 4. Acceleration Disturbances 4.1. Coulomb noise and stiffness The charge-dependent Coulomb acceleration aQK in direction k is given by: aQK = Q2 2mC2 T ∂CT ∂k + QVT mCT ∂CT ∂k − Q mCT N −1 X i=1 Vi ∂Ci,N ∂k . (3) The first two terms in equation (3) are dependent on the overall sensor geometric symmetry, through ∂CT /∂k, and the third term is dependent on the symmetry of the sensor voltage distribution.The corresponding acceleration noise, δaQk, due to random fluctuations of the test mass position relative to the spacecraft, δk, of the potentials of the conductors that surround the test mass, δVi, and of the test mass free charge δQ, is given by: δa2 QK = ( ∂aQK ∂k )2δk2 + N −1 ( X i=1 ∂aQK ∂Vi )2δVi 2 + ( ∂aQK ∂Q )2δQ2 (4) where k is a displacement in direction k; m is the mass of the test mass; Q is the free charge accumulated on the test mass; Ci,j is the capacitance between conductors i and j which surround the test mass; Vi is the potential to which conductor i is raised; CT ≡ PN −1 i=1 Ci,N is the coefficient of capacitance of the test mass, which is defined as the N th conductor, with potential VN = Q/CT + VT , and VT ≡ i=1 Ci,N · Vi.7 The estimates for acceleration noise have assumed typical CT PN −1 1 parameter values for the ASTROD I mission: Q was taken as the amount of charge accumulated in 1 day, assuming a net test mass charging rate of 73.8 e+/s at solar minimum and 33.8 e+/s at solar maximum, which corresponds to the Monte Carlo rate, with error margins, estimated contributions from particle species not included in the Monte Carlo model and the potential contribution from kinetic low energy secondary electron emission, that is likely to almost cancel in the actual sensor,10 added; m =1.75 kg; mean voltages on opposing conductors Vi = 0.5 V; the potential difference between conductors on opposing faces of the sensor compensated to 10 mV; the asymmetry in gap across opposite sides of test mass is 10 µm; capacitances and capacitance gradients were calculated using parallel plate approximations: CT = 53 pF; VT = 0.5 V; position noise δk = 1×10−7 mHz−1/2; voltage noise δVi = 1×10−4 VHz−1/2 and charge noise δQ = 4.6×10−15 CHz−1/2, which includes, as for the charging rate, the unmodelled contributions. These Coulomb acceleration noise due to the test mass charging are listed in Table 4. The total noise estimated here is a factor of ∼ 30 less than the ASTROD I acceleration noise target. The stiffness associated with test mass charging, SQk, is given by SQk = −m · ∂aQk/∂k. These acceleration noise figures are lower than the results liberally estimated by Shiomi and Ni14 because of the smaller charging rate and charging noise we obtained here. The requirements on Coulomb noise, stiffness and other associated noises for ASTROD I in Ref. 14 are satisfied. October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd Improved Simulation of the Mass Charging for ASTROD I 15 Table 4. The magnitude of charging disturbances for ASTROD I at solar minimum and maximum, at frequency = 0.1 mHz. Solar activity Coulomb noise Lorentz noise Stiffness (×10−15ms−2Hz−0.5) (×10−15ms−2Hz−0.5) (×10−8s−2) δk δVi δQ total minimum 1.06 2.18 1.32 2.80 maximum 0.49 1.08 0.77 1.40 2.80 1.30 -1.52 -0.69 δk: displacement noise; δVi: voltage noise; δQ: charge noise. 4.2. Lorentz noise Lorentz effects arise from the motion of the test mass through the interplanetary magnetic field, ~BI and its residual motion through the field generated within the spacecraft, ~BS . The test mass will be housed in a conducting enclosure, which will reduce the effect of the interplanetary field, via the Hall effect, with efficiency η. Hence, to first order, the Lorentz acceleration noise, aL , is given by: m2(aL )2 = (ηQVI δBI )2 + (ηQδVI BI )2 + (QδVS BS )2 + (ηδQVI BI )2. (5) where VI is the speed of the test mass through the interplanetary field; δVI and δVS are the magnitudes of random fluctuations in the test mass velocity through the interplanetary field and relative to the spacecraft, respectively and δBI gives the magnitude of fluctuations in the interplanetary field.7 aL also increases with decreasing frequency, and is estimated to be ∼ 2.8×10−15 ms−2Hz−1/2 (0.1 mHz) at solar minimum and ∼ 1.3×−15 ms−2Hz−1/2 (0.1 mHz) at solar maximum, which is a factor of ∼ 30 below the ASTROD I acceleration noise target. We have assumed that Q = 73.8 e+/s at solar minimum and 33.8 e+/s at solar maximum, as in section 4.1; η = 0.1; ~VI = 4×104 m/s; δVI = 4.78×10−12 ms−1Hz−1/2; δVS = 6.28×10−11 ms−1Hz−1/2; ~BS = 9.6×10−6 T; δBS = 1×10−7 THz−1/2; ~BI = 1.2×10−7 T (this is a conservative estimate of the field at 0.5 AU, used to give the worst-case noise, for the ASTROD I orbit) and δBI = 1.2×10−6 THz−1/2. The estimates of acceleration noises and stiffness due to the Coulomb and Lorentz effects are listed in Table 4. 4.3. Coherent Fourier components The charging of the test mass may also result in the appearance of unwanted, coherent Fourier components in the ASTROD I measurement bandwidth through Coulomb and Lorentz interactions, due to the time dependence of the amount of charge accumulated on the test mass. It is shown that the signals associated with Lorentz interactions are expected to fall below the ASTROD I test mass residual acceleration noise, but the signals from Coulomb interactions may exceed both the instrumental noise over a fraction of the bandwidth, and may not be eliminated by October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd 16 G. BAO et al. daily discharging of the test mass. The free charge on the test mass at time t can be expressed as follows: Q(t) ≈ Qt. (6) where t is the time for which the test mass has been allowed to charge and Q is the Q is assumed to be constant. Substituting Q(t) ≈ Qt into mean charging rate. Here, the expressions for the Coulomb and the Lorentz accelerations gives the terms15: fk(t) ≡ Ξkt2 ≡ 2 Q 2mC2 T ∂CT ∂k t2, ek(t) ≡ Θkt ≡ − ∂VT ∂k Q m t, lk(t) ≡ Φkt ≡ η Qt m ( ~VI × ~BI ) · k, (7) (8) (9) where fk(t) is dependent on the overall sensor geometric symmetry; ek(t) is depen- dent on the symmetry of the sensor voltage distribution; lk(t) is caused by the inter- planetary magnetic field. The Fourier transforms of these signals with fixed-interval discharge will be described by a series of sinc functions (sinc(x) = sinπx/πx). The equivalent one-sided power spectral density for these coherent signals is given by F T = 2F T 2/τ , where F T = Fourier transform of the signal and τ is the length of P 2 the data sample time. Implementing the parameter values given in section 4.1 and 4.2, taking the mean charging rate as constant and assuming that the test mass is discharged once every 24 hours (as described in Ref. 15), the spectral densities of fk(t), ek(t) and lk(t) are estimated. The curves in figure 13 and figure 14 trace PFk(f ), PEk(f ) and PLk(f ), at the primary peaks of the sinc functions, where Fk(f ), Ek(f ) and Lk(f ) are the Fourier transform of fk(t), ek(t) and lk(t). fk(t) and ek(t) exceed the ASTROD I acceleration noise limit of 3 × 10−14[0.3 mHz/f + 30(f /3 mHz)2] ms−2Hz−1/2 in a fraction of frequency bins in the frequency range of 0.1 mHz < f < 4 mHz at solar minimum (see Figure 13); only ek(t) exceeds the limit in a fraction of frequency bins in the frequency range of 0.1 mHz < f < 2 mHz at solar maximum (see Figure 14). These effects are more severe as frequency decreases. Several schemes could be used to minimize a potential loss of the ASTROD I science data, including continuously discharging the test mass, minimizing sensor voltage and geometrical offsets and through spectral analysis.15 Variations in, for example, the mean charging rate, could result in these signals exceeding the noise target in a larger fraction of the bandwidth. Hence, variations in these signals need to be studied carefully as they will influence the accuracy with which the solar-system and relativistic parameters can be determined. A charge management system has been developed by Imperial College London, for LISA Pathfinder. This system has been extensively tested, both via simulations and laboratory tests.16 -- 18 A similar system could easily be used for ASTROD I. October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd Improved Simulation of the Mass Charging for ASTROD I 17 From ek(t) ASTROD I acceleration target ek(t) fk(t) lk(t) ASTROD I From fk(t) From lk(t) ) 5 . 0 - z H 2 - s m ( y t i s n e d l a r t c e p s n o i t a r e l e c c A 1E-9 1E-10 1E-11 1E-12 1E-13 1E-14 1E-15 1E-16 1E-17 1E-18 1E-19 1E-4 1E-3 0.01 0.1 Frequency (Hz) Fig. 13. The curves trace the spectral densities at the primary peaks of the sinc functions at solar minimum, of the coherent Fourier components, for τ =1 year: ek(t) is given by the dotted line, fk(t) is given by the dashed line and lk(t) is given by the dashed-dotted line. The bold full line gives the ASTROD I acceleration noise limit. ) 5 . 0 - z H 2 - s m ( y t i s n e d l a r t c e p s n o i t a r e l e c c A 1E-9 1E-10 1E-11 1E-12 1E-13 1E-14 1E-15 1E-16 1E-17 1E-18 1E-19 from ek(t) ASTROD I acceleration target ek(t) fk(t) lk(t) ASTROD I from fk(t) from lk(t) 1E-4 1E-3 0.01 0.1 Frequency (Hz) Fig. 14. The curves trace the spectral densities at the primary peaks of the sinc functions at solar maximum, of the coherent Fourier components, for τ =1 year: ek(t) is given by the dotted line, fk(t) is given by the dashed line and lk(t) is given by the dashed-dotted line. The bold full line gives the ASTROD I acceleration noise limit. 5. Conclusion The charging of the ASTROD I test mass by cosmic ray protons and alpha particles (3He and 4He) has been simulated using the GEANT4 toolkit at solar minimum October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd 18 G. BAO et al. and maximum. The Monte Carlo model predicted a net charging rate of ∼ 11.7 e+/s at solar maximum, rising to 33.3 e+/s at solar minimum. Although the proton flux is the dominant charging flux, 4He, which constitutes only 8% of the total cosmic ray flux, is responsible for ∼ 18% and ∼ 20% of this rate at solar minimum and maximum, respectively. We have also included an additional net charging rate contribution due to particle species that were not included in the Monte Carlo model, and a potential charging mechanism, due to kinetic low energy secondary electron emission based on LISA studies.10,11 There is an additional uncertainty of ± 30% in the net charging rate, due to uncertainties in the cosmic ray spectra, physics models and geometry implementation. A recent, preliminary, comparison of a simplified GEANT4 simulation and actual GP-B charging rates, indicate 45% agreement (GP-B simulation is larger than experimental value by 45%), reinforcing confidence in these predictions.18,19 The ASTROD I acceleration noise limit is 10−13 ms−2Hz−1/2 at 0.1 mHz, which is less stringent than the LISA requirement. The magnitudes of the Coulomb and Lorentz acceleration noise associated with test mass charging increase with decreas- ing frequency. At the lowest frequency in the ASTROD I bandwidth, 0.1 mHz, the estimates of the Coulomb and Lorentz acceleration noise are both well below the acceleration noise target. These results agree, to within 30%, with those from our earlier study8, which was based on a simple geometry model. The variations in the test mass charging rate will alter the spectral description of the coherent Fourier components. Hence, further work is needed to ensure that these do not compromise the quality of the science data of the ASTROD I mission. The charging process of the ASTROD I test mass by SEPs (Solar Energetic Particles) has also been simulated using the GEANT4 toolkit, and the charging rate is much larger than the values due to GCR (Glactic Cosmic Ray) proton flux at solar maximum and solar minimum.20 -- 22 However, the charge management hardware described in reference 18 could be used to discharge a test mass even during solar events, provided safe operation could be ensured. The effect of cosmic ray fluxes of particle species not included in the Monte Carlo simulation needs to be verified for the ASTROD I geometry. According to the recent work by Grimani et al.,21 the charging rate in each LISA test mass induced by primary and interplanetary electrons is comparable (absolute value) to that released by the nuclei of the C, N, O group at solar minimum. The acceleration disturbances due to charging of the ASTROD I test mass by electrons are also under study. Further, we will evaluate the variation in the ASTROD I test mass charging rate over the orbit over the solar cycle, including a detailed study of SEP events, and its variation due to modulation of cosmic ray flux over the ASTROD I orbit. For this, a SCoRE (Solar And Cosmic Ray Physics And The Space Environment: Studies For And With LISA) study along the line of Shaul et al.22 would be useful. October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd Improved Simulation of the Mass Charging for ASTROD I 19 Acknowledgments This work is funded by the National Natural Science Foundation (Grant Nos 10475114 and 10573037) and the Foundation of Minor Planets of Purple Moun- tain Observatory. References 1. W.-T. Ni, Y. Bao, H. Dittus, T. Huang, C. Lammerzahl, G. Li, J. Luo, Z.-G. Ma, J.F. Mangin, Y.-X. Nie, A. Peters, A. Rudiger, E. Samain, S. Shiomi, T. Sumner, C.-J. Tang, J. Tao, P. Touboul, H. Wang, A. Wicht, X.-J. Wu, Y. Xiong, C. Xu, J. Yan, D.-Z. Yao, H.-C. Yeh, S.-L. Zhang, Y.-Z. Zhang, Z.-B. Zhou.: ASTROD I: Mission Concept and Venus Flybys, Proc. 5th IAA Intnl Conf. On Low-Cost Planetary Missions, ESTEC, Noordwijk, The Netherlands, 24-26 September 2003, ESA SP-542, November 2003, pp.79-86; ibid., Acta Astronautica 59, 598 (2006) 2. W.-T. Ni, H. Araujo, G. Bao, H. Dittus, T.-Y. Huang, S. Klioner, S. Kopeikin, G. Krasinsky, C. Lammerzahl, G.-Y. Li, H.-Y. Li, L. Liu, Y.-X. Nie, A. Paton, A. Peters, E. Pitjeva, A. Rudiger, E. Samain, D. Shaul, S. Schiller, J.-C. Shi, S. Shiomi, M.H. Soffel, T. Sumner, S. Theil, P. Touboul, P. Vrancken, F. Wang, H.-T Wang, Z.-Y. Wei, A. Wicht, X.-J. Wu, Y. Xia, Y.-H. Xiong, C.-M. Xu, J. Yan, H.-C. Yeh, Y.-Z. Zhang, C. Zhao, and Z.-B. Zhou, Journal of Physics: Conference Series, 32, 154 (2006). 3. C.-J. Tang, C.-H. Chang, W.-T. Ni, A.-M. Wu, Orbit Design for ASTROD I, 2003 report. 4. S. Vitale, P. Bender, A. Brillet, S. Buchman, A. Cavalleri, M. Cerdonio, M. Cruise, C. Cutler, K. Danzmann, R. Dolesi, W. Folkner, A. Gianolio, Y. Jafry, Gunther. Hasinger, G. Heinzel, C. Hogan, M. Hueller, J. Hough, S. Phinney, T. Prince, D. Richstone, D. Robertson, M. Rodrigues, A. Rudiger, M. Sandford, R. Schilling, D. Shoemaker, B. Schutz, R. Stebbins, C. Stubbs, T. Sumner, K. Thorne, M. Tinto, P. Touboul, H. Ward, W. Weber, W.Winkler, Nucl. Phys. B (Proc. Suppl), 110, 209 (2002). 5. P. L. Bender, et al. LISA Pre-Phase A Report 2nd edn MPQ233 (1998). 6. W.-T. Ni, ASTROD (Astrodynamical Space Test of Relativity using Optical Devices) and ASTROD I, Nucl. Phys. B (Proc. Suppl), 166, 153 (2007). 7. D. Shaul, H. Araujo, G. Rochester, T. Sumner, P. Wass, Class. Quantum Grav., 22, 297 (2005). 8. G. Bao, D. Shaul, H. Araujo, W.-T. Ni, T. Sumner, L. Liu, General Relativity and Gravitation, 39, No. 00, 000 (2007); arXiv: 0704.3303v1. 9. C. Grimani, H. Vocca, M. Barone, R. Stanga, F. Vetrano, A. Vicere, P. Amico, L. Bosi, F. Marchesoni, M. Punturo, and F. Travasso, Class. Quantum Grav., 21, S629 (2004). 10. H. Araujo, P. Wass, D. Shaul, G. Rochester, T. Sumner, Astroparticle Physics, 22, 451 (2005). 11. C. Grimani, H. Vocca, G. Bagni, L. Marconi, R. Stanga, F. Vetrano, A. Vicere, P. Amico, L. Gammaitoni, and F. Marchesoni, Class. Quantum Grav., 22, S327 (2005). 12. H. Araujo, A. Howard, D. Shaul, and T. Sumner, Class. Quantum Grav., 20, S311 (2003). 13. G. Bao, L. Liu, D. Shaul, H. Araujo, W.-T. Ni, and T. Sumner, Nucl. Phys. B (Proc. Suppl), 166, 246 (2007). 14. S. Shiomi and W.-T. Ni, Class.Quantum Grav., 23, 1 (2006). 15. D. Shaul, T. Sumner, G. Rochester, International Journal of Modern Physics D, 14, 51 (2005). 16. M. Schulte, G. Rochester, D. Shaul, T. Sumner, C. Trenkel and P. Wass, AIP Confer- October 29, 2018 12:29 WSPC/INSTRUCTION FILE ws-ijmpd 20 G. BAO et al. ence Proceeding -- November 29, 2006 -- Volume 873, Laser Interferometer Space Antenna 6th International LISA Symposium, edited by S. M. Merkowitz and J. C. Livas, 165-171 (2006). 17. P. Wass, L. Carbone, A. Cavalleri, G. Ciani, R. Dolesi, M. Hueller, G. Rochester, M. Schulte, T. Sumner, D. Tombolato, C. Trenkel, S. Vitale, and W. Weber, AIP Conference Proceeding -- November 29, 2006 -- Volume 873, Laser Interferometer Space Antenna 6th International LISA Symposium, edited by S. M. Merkowitz and J. C. Livas, 220-224 (2006). 18. D. Shaul, Charge Management for LISA and LISA Pathfinder, IJMPD, this issue. 19. J.P. Turneaure et al., Advances in Space Research, 32, 1387 (2003). 20. L. Liu, G. Bao, W.-T. Ni and D. Shaul, Simulation of ASTROD I test mass charg- ing due to solar energetic particles, submitted to Adv. Space Res., November, 2006; arXiv:0704.3493v1. 21. C. Grimani, M. Fabi, R. Stanga, and L. Marconi, AIP Conference Proceeding -- November 29, 2006 -- Volume 873, Laser Interferometer Space Antenna 6th International LISA Symposium, edited by S. M. Merkowitz and J. C. Livas, 184-188 (2006). 22. D. Shaul, K. Aplin, H. Araujo, R. Bingham, J. Blake, G. Branduardi-Raymont, S. Buchman, A. Fazakerley, L. Finn, L. Fletcher, A. Glover, C. Grimani, M. Hapgood, B. Kellet, S. Matthews, T. Mulligan, W.-T. Ni, P. Nieminen, A. Posner, J. Quenby, P. Roming, H. Spence, T. Sumner, H. Vocca, P. Wass and P. Young, AIP Conference Proceeding -- November 29, 2006 -- Volume 873, Laser Interferometer Space Antenna 6th International LISA Symposium, edited by S. M. Merkowitz and J. C. Livas, 172-178 (2006).
astro-ph/9806185
3
9806
1999-10-27T20:52:16
Supernova pencil beam survey
[ "astro-ph" ]
Type Ia supernovae (SNe Ia) can be calibrated to be good standard candles at cosmological distances. We propose a supernova pencil beam survey that could yield between dozens to hundreds of SNe Ia in redshift bins of 0.1 up to $z=1.5$, which would compliment space based SN searches, and enable the proper consideration of the systematic uncertainties of SNe Ia as standard candles, in particular, luminosity evolution and gravitational lensing. We simulate SNe Ia luminosities by adding weak lensing noise (using empirical fitting formulae) and scatter in SN Ia absolute magnitudes to standard candles placed at random redshifts. We show that flux-averaging is powerful in reducing the combined noise due to gravitational lensing and scatter in SN Ia absolute magnitudes. The SN number count is not sensitive to matter distribution in the universe; it can be used to test models of cosmology or to measure the SN rate. The SN pencil beam survey can yield a wealth of data which should enable accurate determination of the cosmological parameters and the SN rate, and provide valuable information on the formation and evolution of galaxies. The SN pencil beam survey can be accomplished on a dedicated 4 meter telescope with a square degree field of view. This telescope can be used to conduct other important observational projects compatible with the SN pencil beam survey, such as QSOs, Kuiper belt objects, and in particular, weak lensing measurements of field galaxies, and the search for gamma-ray burst afterglows.
astro-ph
astro-ph
astro-ph/9806185 September 10, 1999 to appear in ApJ, 531, #2 (March 10, 2000) Supernova pencil beam survey Yun Wang1 Princeton University Observatory Peyton Hall, Princeton, NJ 08544 email: [email protected] ABSTRACT Type Ia supernovae (SNe Ia) can be calibrated to be good standard candles at cosmological distances. We propose a supernova pencil beam survey that could yield between dozens to hundreds of SNe Ia in redshift bins of 0.1 up to z = 1.5, which would compliment space based SN searches, and enable the proper consideration of the systematic uncertainties of SNe Ia as standard candles, in particular, luminosity evolution and gravitational lensing. We simulate SNe Ia luminosities by adding weak lensing noise (using empirical fitting formulae) and scatter in SN Ia absolute magnitudes to standard candles placed at random redshifts. We show that flux-averaging is powerful in reducing the combined noise due to gravitational lensing and scatter in SN Ia absolute magnitudes. The SN number count is not sensitive to matter distribution in the universe; it can be used to test models of cosmology or to measure the SN rate. The SN pencil beam survey can yield a wealth of data which should enable accurate determination of the cosmological parameters and the SN rate, and provide valuable information on the formation and evolution of galaxies. The SN pencil beam survey can be accomplished on a dedicated 4 meter telescope with a square degree field of view. This telescope can be used to conduct other important observational projects compatible with the SN pencil beam survey, such as QSOs, Kuiper belt objects, and in particular, weak lensing measurements of field galaxies, and the search for gamma-ray burst afterglows. 1. Introduction Toward the end of the millennium, cosmology has matured into a phenomenological science. Observational data now dominates aesthetics in the evaluation of cosmological 1Present address: Dept. of Physics, 225 Nieuwland Science Hall, University of Notre Dame, Notre Dame, IN 46556-5670. email: [email protected] -- 2 -- models. Of fundamental importance is the determination of cosmological parameters, in particular, the Hubble constant H0, the matter density faction Ωm, and the density fraction contributed by the cosmological constant ΩΛ. Observation of distant Type Ia supernovae (SNe Ia) has become an increasingly powerful means of measuring cosmological parameters (Perlmutter et al. 1997, 1998; Riess et al. 1998; Schmidt et al. 1998), because SNe Ia can be calibrated to be good standard candles at cosmological distances. (Riess, Press, & Kirshner 1995) A type Ia SN is the thermonuclear explosion of a carbon-oxygen white dwarf in a binary when the rate of the mass transfer from the companion star is high. The SN explosion blows the white dwarf completely apart. The radioactive decay of the isotopes 56Ni and 56Co is responsible for much of the light emitted. The SN lightcurve reaches a maximum about 15 days after the explosion and then declines slowly over years. A SN can outshine the galaxy in which it lies. Two independent groups (Riess et al. 1998, Perlmutter et al. 1999) have made systematic searches for SNe Ia for the purpose of measuring cosmological parameters. Their preliminary results seem to indicate a low matter density universe, possibly with a sizable cosmological constant. Even though the uncertainty in these results is large, they clearly demonstrate that the observation of SNe Ia can potentially become a reliable probe of cosmology. However, there are important systematic uncertainties of SNe Ia as standard candles, in particular, luminosity evolution and gravitational lensing. To constrain the evolution of SN Ia peak absolute luminosities, we need a large number of SNe Ia at significantly different redshifts (low z and z > 1), which is not available at present. Both groups have assumed a smooth universe in their data analysis, although they include lensing in their error budgets. Since we live in a clumpy universe, the effect of gravitational lensing must be taken into account adequately for the proper interpretation of SN data. At present, the small number of observed high z SNe Ia prevents adequate modeling of gravitational lensing effects. In this paper, we propose a pencil beam survey of SNe Ia that could yield between dozens to hundreds of SNe Ia in redshift bins of 0.1 up to z = 1.5 (see §3) which would allow the proper modeling of gravitational lensing, as well as a quantitative understanding of luminosity evolution. Such a survey would yield a wealth of data which can be used to make accurate measurement of cosmological parameters and the SN rate, and provide powerful constraints on various aspects of the cosmological model. In §2, we consider the weak lensing of SNe Ia. We simulate SN Ia luminosities by adding weak lensing noise and scatter in SN Ia absolute magnitudes to standard candles placed at random redshifts. We show how flux-averaging reduces the combined noise of gravitational lensing and scatter in SN Ia absolute magnitudes. In §3, we show that the number count of SNe from a pencil beam survey is not sensitive to matter distribution in the universe; it can be used as a test of models of cosmology and SN progenitors, or to measure the SN rate accurately. In §4, we discuss the observational feasibility of the SN pencil beam survey. §5 contains conclusions. -- 3 -- 2. Weak lensing of supernovae In a SN Ia Hubble diagram, one must use distance-redshift relations to make theoretical predictions. Unlike angular separations and flux densities, distances are not directly measurable, but they are indispensable theoretical intermediaries. The distance-redshift relations depend on the distribution of matter in the universe. In a smooth Friedmann-Robertson-Walker (FRW) universe, the metric is given by ds2 = dt2 − a2(t)[dr2/(1 − kr2) + r2(dθ2 + sin2 θ dφ2)], where a(t) is the cosmic scale factor, and k is the global curvature parameter (Ωk = 1 − Ωm − ΩΛ = −k/H 2 distance r is given by 0 ). The comoving r(z) = 0 cH −1 Ωk1/2 sinn(cid:26)Ωk1/2Z z dz ′ hΩm(1 + z ′)3 + ΩΛ + Ωk(1 + z ′)2i−1/2(cid:27) , (1) 0 where "sinn" is defined as sinh if Ωk > 0, and sin if Ωk < 0. If Ωk = 0, the sinn and Ωk's disappear from Eq.(1), leaving only the integral. The angular diameter distance is given by dA(z) = r(z)/(1 + z), and the luminosity distance is given by dL(z) = (1 + z)2dA(z). However, our universe is clumpy rather than smooth. According to the focusing theorem in gravitational lens theory, if there is any shear or matter along a beam connecting a source to an observer, the angular diameter distance of the source from the observer is smaller than that which would occur if the source were seen through an empty, shear-free cone, provided the affine parameter distance (defined such that its element equals the proper distance element at the observer) is the same and the beam has not gone through a caustic. An increase of shear or matter density along the beam decreases the angular diameter distance and consequently increases the observable flux for given z. (Schneider, Ehlers, & Falco 1992) The observation of SNe Ia at z > 1 is important for the determination of cosmological parameters, but the dispersion in SN Ia luminosities due to gravitational lensing can become comparable to the intrinsic dispersion of SNe Ia absolute magnitudes because the optical depth for gravitational lensing increases with redshift. 2.1. Direction dependent smoothness parameter If only a fraction α (known as the smoothness parameter) of the matter density is smoothly distributed, the largest possible distance (for given redshift) for light bundles which have not passed through a caustic is given by the appropriate solution to the following equation: g(z) d dz "g(z) dDA dz # + 3 2 α Ωm(1 + z)5DA = 0, (2) -- 4 -- where g(z) ≡ (1 + z)3q1 + Ωmz + ΩΛ[(1 + z)−2 − 1]. (Kantowski 1998) The ΩΛ = 0 form of Eq.(2) has been known as the Dyer-Roeder equation. (Dyer & Roeder 1973, Schneider et al. 1992) Fig.1(a) shows magnitude versus redshift for the three cosmological models considered by Riess et al. (1998), SCDM (Ωm = 1, ΩΛ = 0), OCDM (Ωm = 0.2, ΩΛ = 0), and ΛCDM (Ωm = 0.2, ΩΛ = 0.8). For each cosmological model, the upper curve represents the completely clumpy universe (empty beam, α = 0), while the lower curve represents the completely smooth universe (filled beam, α = 1). Fig.1(b) shows the same models relative to smooth OCDM ( filled beam, α = 1), the middle curve for each model now represents a universe with half of the matter smoothly distributed (half-filled beam, α = 0.5). Clearly, at z > 1, there is degeneracy of distances in a flat clumpy universe and an open smooth universe, and also in an open clumpy universe and a flat smooth universe with a sizable cosmological constant, as has been noted by a number of previous authors. (Kantowski 1998, Linder 1998, Holz 1998) We can generalize the angular diameter distance DA(z) by allowing the smoothness parameter α to be direction dependent, i.e., a property of the beam connecting the observer and the standard candle. The smoothness parameter α essentially represents the amount of matter that causes weak lensing of a given source. Since matter distribution in our universe is inhomogeneous, we can think of our universe as a mosaic of cones centered on the observer, each with a different value of α. This reinterpretation of α implies that we have α > 1 in regions of the universe in which there are above average amounts of matter which can cause magnification of a source. (Wang 1999a) In order to derive a unique mapping between the distribution in distances and the distribution in the direction dependent smoothness parameter for given redshift z, we define the direction dependent smoothness parameter α to be the solution of Eq.(2) for given distance DA(z). At given redshift z, the magnification of a source can be expressed in terms of the apparent brightness of the source f ( αz), or in terms of the angular diameter distance to the source DA( αz): µ = f ( αz) f ( α = 1z) = "DA( α = 1z) DA( αz) #2 , (3) where f ( α = 1z) and DA( α = 1z) are the flux of the source and angular diameter distance to the source in a completely smooth universe (filled beam), and α is the direction dependent smoothness parameter. Since distances are not directly measurable, we should interpret Eq.(3) as defining a unique mapping between the magnification of a standard candle at redshift z and the direction dependent smoothness parameter α at z; α parametrizes the direction dependent matter distribution in a well-defined manner. From the magnification distributions of standard candles at various redshifts, p(µz), -- 5 -- with z =0.5, 1, 1.5, 2, 2.5, 3, 5, found numerically by Wambsganss et al. for Ωm = 0.4, ΩΛ = 0.6 (Wambsganss et al. 1997, Wambsganss 1999), Wang (1999a) has obtained simple empirical fitting formulae for the distribution of α: p( αz) = Cnorm exp"−(cid:18) α − αpeak w αq (cid:19)2# , where Cnorm, αpeak, w, and q depend on z and are independent of α. They are given by Cnorm(z) = 10−2"0.53239 + 2.79165 (cid:18)z αpeak(z) = 1.01350 − 1.07857 (cid:18) 1 w(z) = 0.06375 + 1.75355 (cid:18) 1 q(z) = 0.75045 + 1.85924 (cid:18)z 5(cid:19)2 5(cid:19) − 2.42315 (cid:18)z + 1.13844 (cid:18)z 5z(cid:19)2 5z(cid:19)3 5z(cid:19) + 2.05019 (cid:18) 1 − 2.14520 (cid:18) 1 5z(cid:19)3 5z(cid:19)2 5z(cid:19) − 4.99383 (cid:18) 1 + 5.95852 (cid:18) 1 5(cid:19)2 5(cid:19)3 5(cid:19) − 2.91830 (cid:18)z + 1.59266 (cid:18)z . 5(cid:19)3# , , , (4) (5) Cnorm(z) is the normalization constant for given z. The parameter αpeak(z) indicates the average smoothness of the universe at redshift z, it increases with z and approaches αpeak(z) = 1 (filled beam) at z = 5; the parameter w(z) indicates the width of the distribution in the direction dependent smoothness parameter α, it decreases with z. The z dependences of αpeak(z) and w(z) are as expected because as we look back to earlier times, lines of sight become more filled in with matter, and the universe becomes smoother on the average. The parameter q(z) indicates the deviation of p( αz) from Gaussianity (which corresponds to q = 0). Models with different cosmological parameters should lead to somewhat different matter distributions p( αz). In the context of weak lensing of standard candles, we expect the cosmological parameter dependence to enter primarily through the magnification µ to direction dependent smoothness parameter α mapping at given z (the same α corresponds to very different µ in different cosmologies). 2.2. Flux averaging of SN luminosities Gravitational lensing noise in the Hubble diagram can be reduced by appropriate flux averaging of SNe Ia in each redshift bin. Because of flux conservation, the average flux of a sufficient number of SNe Ia at the same z from the same field should be the same as the true flux of the SNe Ia without gravitational lensing if the sample is complete. It is convenient to compare the distance modulus of SNe Ia, µ0, with the theoretical prediction µp 0 = 5 log dL Mpc! + 25, (6) -- 6 -- where dL(z) is the luminosity distance. Before flux-averaging, we convert the distance modulus µ0(zi) of SNe Ia into "fluxes", f (zi) = 10−µ0(zi)/2.5. We then obtain "absolute luminosities", {L(zi)}, by removing the redshift dependence of the "fluxes", i.e., L(zi) = 4π d2 L(ziH0, Ωm, ΩΛ) f (zi), (7) where (H0, Ωm, ΩΛ) are the best-fit cosmological parameters derived from the unbinned data set {f (zi)}. We then flux-average over the "absolute luminosities" {Li} in each redshift bin. The set of best-fit cosmological parameters derived from the binned data is applied to the unbinned data {f (zi)} to obtain a new set of "absolute luminosities" {Li}, which is then flux-averaged in each redshift bin, and the new binned data is used to derive a new set of best-fit cosmological parameters. This procedure is repeated until convergence is achieved. This iteration should lead to the optimal removal of gravitational lensing noise and the accurate determinations of the cosmological parameters. Wang (1999b) has applied this method to analyze the combined data from the two groups (Riess et al. 1998, Perlmutter et al. 1999). To illustrate how flux-averaging can reduce the dispersion in SN Ia luminosities caused by weak lensing, let us simulate the data by drawing NSN random points from the redshift interval [z0 giving each standard candle a direction dependent smoothness parameter α (corresponding to µ = [DA( α = 1)/DA( α)]2) drawn at random from the distribution p( αz) given in §2.1. The scatter in SN Ia absolute magnitudes, ∆mabm, can be written as 2], each point represents a standard candle. We add weak lensing noise by 1, z0 ∆mabm = ∆mint + ∆mobs, (8) where ∆mint is the intrinsic scatter and ∆mobs is the observational noise. We assume that both intrinsic scatter and observational noise are Gaussian distributed in magnitude, with dispersions σint and σobs respectively. Then the total scatter in SN Ia absolute magnitudes is also Gaussian distributed, i.e., p(∆mabm) = 1 √2π σabm exp"− (∆mabm)2 abm # , 2σ2 (9) with σabm = qσ2 extracted from the data is int + σ2 obs. We take σabm = 0.20. The absolute luminosity of each SN Ia L(z) = 10−∆mobs/2.5 µ( αz) Lint(z), = 10−∆mobs/2.5 µ( αz)n10−∆mint/2.5 L( α = 1z)o = L( α = 1z) µ( αz) 10−∆mabm/2.5. (10) We have used Lint(z) = 10−∆mint/2.5 L( α = 1z). -- 7 -- For each SN Ia, the total noise is ∆m = −2.5 log L( α = 1z)! . L(z) Let us average the fluxes of all SNe Ia in the redshift bin [z0 1, z0 2]: The flux averaged noise is where L = 1 NSN NSN Xi=1 L(zi). (∆m)avg = −2.5 log L L( α = 1z)! , z = PNSN i=1 zi NSN . (11) (12) (13) (14) Table 1 lists the means and dispersions (in the form of mean±dispersion) of ∆m (which are h∆mi and σ = qh[∆m − h∆mi]2i ), and (∆m)avg (which are h(∆m)avgi and σavg = qh[(∆m)avg − h(∆m)avgi]2i ) for various redshift bins with NSN =2, 4, and 9 SNe Ia in each bin, for 104 random samples. We have taken Ωm = 0.4, ΩΛ = 0.6. Table 1: Means and dispersions of ∆m and (∆m)avg for a ΛCDM model. z [0.5, 0.6] [1, 1.1] [1.5, 1.6] [1.5, 2] [2, 2.5] [2.5, 3] [3, 3.5] [3.5, 4] [4, 4.5] [4.5, 5] h∆mi ± σ 0.001 ±0.200 0.002 ± 0.204 0.003±0.210 0.003±0.213 0.005 ±0.218 0.006 ±0.223 0.007 ±0.228 0.007 ±0.233 0.008 ±0.238 0.008±0.243 NSN =2 NSN =4 NSN =9 -0.007±0.141 -0.007 ± 0.145 -0.006± 0.149 -0.006± 0.151 -0.005 ±0.155 -0.005±0.159 -0.004± 0.162 -0.004±0.166 -0.004±0.170 -0.004± 0.174 -0.012 ±0.101 -0.013 ± 0.103 -0.012 ± 0.107 -0.012 ± 0.108 -0.012 ± 0.111 -0.012±0.114 -0.012±0.117 -0.012±0.119 -0.012±0.122 -0.012± 0.125 -0.015±0.067 -0.015 ± 0.069 -0.015 ± 0.071 -0.015 ± 0.072 -0.015 ± 0.074 -0.015±0.076 -0.015± 0.078 -0.015 ± 0.080 -0.015± 0.082 -0.016± 0.084 The dispersion decreases roughly as 1/√NSN . The dispersion would reduce by 30% if the sample contains 2 SNe Ia; 50% if the sample contains 4 SNe Ia. Even though gravitational lensing noise increases with redshift, the combined gravitational lensing and -- 8 -- SN Ia absolute magnitude scatter noise in the redshift interval z = [4.5, 5] can be reduced to the same level as in the absence of lensing by flux averaging over two SNe Ia. Note that the flux averaged luminosities are biased towards slightly higher luminosities. This is as expected, because we have assumed that the intrinsic scatter and observational noise in the SN Ia absolute magnitude are Gaussian in magnitude. It is straightforward to show that the mean of 10−∆mabm/2.5 is exp [(ln 10/2.5)2 σ2 abm/2], which corresponds to a bias of −σ2 abm ln 10/5 ≃ −0.018 for σabm = 0.2. If we assume that the intrinsic scatter and observational noise are Gaussian in luminosity, the flux averaged luminosities become unbiased. 3. Supernova number count The SN number count is most sensitive to the SN rate as a function of z, which depends on the specific rate of SNe, as well as the number density of galaxies and their luminosity distribution. The frequency of SNe is a key parameter for describing the formation and the evolution of galaxies; the winds driven by SNe tune the energetics, and their production of metals determines the chemical evolution of galaxies and of clusters of galaxies (Ferrini & Poggianti 1993, Renzini et al. 1993). The SN II rate is related (for a given initial mass function) to the instantaneous stellar birthrate of massive stars because SNe II have short-lived progenitors; the SN Ia rate follow a slower evolutionary track, and can be used to probe the past history of star formation in galaxies. Accurate measurements of the SN rates at intermediate redshifts are important for understanding galaxy evolution, cosmic star formation rate and the nature of SN Ia progenitors. (Madau 1998, Ruiz-Lapuente & Canal 1998, Sadat et al. 1998, Yungelson & Livio 1998, Madau et al. 1998) The SN rates are very uncertain at present due to the small number of SNe discovered in systematic searches. (van den Bergh & McClure 1994, Pain et al. 1996). Kolatt & Bartelmann (1997) have estimated the SN Ia average rate per proper time unit per comoving volume to be nSN (z) = [A + Bz] (100 yr)−1(h−1Mpc)−3, A = 0.0136, B = 0.067, (15) for q0 = 0.5. Changing q0 to lower values leads to lower comoving densities but higher luminosities of galaxies at high z (Lilly et al. 1995). Eq.(15) has been derived assuming that all SNe reside in galaxies, neglecting the redshift dependence of the specific SN rate (number per unit luminosity per unit time), and using the number density of galaxies (as function of redshift) and the Schechter function parameters derived from the Canada France redshift survey (Lilly et al. 1995), the APM survey (Loveday et al. 1992), and the AUTOFIB survey (Ellis et al. 1996). We use Eq.(15) for all cosmological models considered in this paper, because it is a very crude and conservative estimate, and we use it for the purpose of illustration only. -- 9 -- The expected number of SNe Ia in a field of angular area θ2 for an effective observation duration of ∆t up to redshift z is N(z) = θ2Z z = 22.4 θ 0 drp(z ′) a(t′) r2(z ′) nSN (z ′) (1 + z ′) , (16) 0.0136 + 0.067z ′ , 1′!2 ∆t 1 yr!Z z 0 0 #2 dz ′ " r(z ′) cH −1 (1 + z ′)qΩm(1 + z ′)3 + ΩΛ + Ωk(1 + z ′)2 where drp = −cdt is the proper distance interval, r is comoving distance, and the factor (1 + z)−1 accounts for the cosmological time dilation. Note that the number counts fundamentally probe a different aspect of the global geometry of the universe than do the distance measures (Carroll, Press, & Turner 1992). Fig.2 shows the number of SNe Ia expected per 0.1 redshift interval as function of redshift for the same three cosmological models as in Fig.1, SCDM (Ωm = 1, ΩΛ = 0), OCDM (Ωm = 0.2, ΩΛ = 0), and ΛCDM (Ωm = 0.2, ΩΛ = 0.8). For a one square degree field, and an effective observation duration of one year, the total numbers of expected SNe Ia are 464, 899, and 1705 for SCDM, OCDM, and ΛCDM respectively for z up to 1.5. The number of SNe which are strongly lensed by galaxies is given by Nlensed(z) = Z z 0 dz ′ τ (z ′) dN dz ′ , 1′!2 = 3.81 × 10−2 θ 0.05(cid:19) ∆t 1 yr! (cid:18) F 0 #5 dz ′ " r(z ′) cH −1 ·Z z 0 0.0136 + 0.067z ′ . (17) (1 + z ′)qΩm(1 + z ′)3 + ΩΛ + Ωk(1 + z ′)2 We have used the optical depth for gravitational lensing by galaxies (Turner 1990, Fukugita & Turner 1991) τ (z) ≃ F 30 " (1 + z)dA(z) cH −1 0 #3 , (18) where F parametrizes the gravitational lensing effectiveness of galaxies (as singular isothermal spheres). Fig.3 shows the number of strongly lensed SNe as function of survey depth z, for the same cosmological models as in Fig.1 and Fig.2. For a one square degree field, and an effective observation duration of one year, the total numbers of strongly lensed SNe Ia are 0.2, 0.6, and 1.8 for SCDM, OCDM, and ΛCDM respectively for z up to 1.5. Fig.2 and Fig.3 show that the SN number counts from a pencil beam survey can be used to measure the SN rate at high redshifts and perhaps to probe cosmology. Because of the large number of SNe in each redshift bin and the smallness of gravitational lensing optical depth, the SN number count should be insensitive to matter distribution in the -- 10 -- universe, and should therefore provide a robust probe of the cosmological model. Note that the strongly lensed SNe can be easily removed from the survey sample, because they appear as unusually bright SNe. The number of strongly lensed SNe is very sensitive to the cosmological model and might be used to further constrain the cosmological model. But given the small number of strongly lensed SNe Ia expected in any realistic observational program, their usefulness may be limited. The SN number counts provide a combined measure of the cosmological parameters and the SN rate. Fig. 4 shows the parameter dependence of the SN Ia number count per 0.1 redshift interval (z − 0.05, z + 0.05) as function of redshift z. Note that the dependences of the SN number count on the SN rate (parametrized by A and B) and the cosmological parameters are not degenerate and can be differentiated in principle. In practice, we are ignorant of the functional form of the SN rate as a function of z. Hence, we may apply the measurements of the cosmological parameters as priors to the SN number counts to obtain a direct measure of the SN rate in the universe for z up to 1.5, which can be used as a powerful constraint on the cosmological model. For a one square degree field, and an effective observation duration of one year, the SN rate per 0.1 redshift interval can be determined to 14-16%, 9-11%, 7-8% at 1 < z < 1.5 for SCDM, OCDM, and ΛCDM respectively. 4. Observational feasibility A pencil beam survey would be efficient in the discovery of SNe at z >∼ 1 through the combination of data from successive nights and the comparison of the latest frame of images with all previous frames. SNe Ia are quite faint at z ∼ 1.5, with AB magnitude in the I band of IAB ∼ 26. Because of the UV suppression due to line blanketing and the apparent IR suppression in the rest frame SN Ia spectrum, one should use a passband that corresponds to the wavelength range of 3000A − 10000A in the SN rest frame; this means using the I, J, H, or K passbands to observe the z > 1 SNe. SN searches from space (HST and NGST) are limited by small fields of view, a large scale SN search is only possible from the ground, where one is limited to the I band by the atmosphere. Here we discuss the observational feasibility of a pencil beam survey of SNe Ia up to z ∼ 1.5 in terms of exposure times in the I band; although multiple band photometry should be obtained to constrain extinction and evolution of the SNe. Using values for the photometric parameters from the Sloan Digital Sky Survey (1 arcsec seeing, effective sky brightness of 20.3 mag/arcsec2 in the I band, seeing-dominated PSF, etc), we find that the exposure time for a point source with AB magnitude of IAB can -- 11 -- be written as 10 !2 t = 13.94 hours S/N D (cid:19)2 (cid:18)4 m 100.8(IAB −26) (19) where S/N is the signal-to-noise ratio, D is the aperture of the telescope. The above equation shows that the supernova pencil beam survey can be accomplished from a modest dedicated 4 meter telescope, which can image the same fields (two pencil beams should be observed to keep the fields close to zenith) every night, which can lead to the discovery of SNe Ia up to z = 1.5 via appropriate combination of data from successive nights, and the light curves of SNe Ia at z < 1.5. It should be feasible to monitor the faintest SNe Ia from the Keck 10 meter telescope or the HST. Spectra of SNe are required to determine whether they are type Ia. SNe Ia have a Si II absorption line at ∼ 4000 A that may be used to identify the 1 < z <∼ 1.5 SNe Ia in ground-based observations. The follow-up spectroscopy can be attempted on the Keck Low-Resolution Imaging Spectrometer (LRIS). Assuming that we use the 300 grooves/mm grating which provide dispersions of 4.99 A per 48 µm (2 pixels on the CCD), we find the exposure time for 0.5 arcsec seeing to be 3 !2 t = 28 hours S/N 3 !2 = 4.44 hours S/N 5 !2 = 12.33 hours S/N (cid:18) W 1 arcsec(cid:19) 100.4[2(IAB −26)−(I sky (cid:18) W 1 arcsec(cid:19) 100.4[2(IAB −25)−(I sky (cid:18) W 1 arcsec(cid:19) 100.4[2(IAB −25)−(I sky AB AB −21)], −21)], AB −21)], (20) where W is the width of the slit, and I sky AB is the sky brightness in the I band in units of mag/arcsec2. Note that the Si II absorption feature at ∼ 4000 A is not deep enough to be useful in a very noisy (S/N = 3) spectrum. Clearly, spectroscopic follow-up of the z ∼ 1.5 (IAB ∼ 26) SNe Ia will require substantial observational resources; NICMOS or NGST will be more suitable for the spectroscopy of the SNe Ia at the highest redshifts discovered from the ground. The Keck LRIS can be used to obtain the spectra of the SNe Ia at more modest redshifts (z ∼ 1.3, IAB ∼ 25). A dedicated 4 meter telescope with a square degree field of view can be used to conduct other important scientific projects compatible with the SN pencil beam survey, such as QSOs, Kuiper belt objects, and in particular, weak lensing and the search of gamma-ray burst (GRB) afterglows. Weak lensing is a powerful tool in mapping the mass distribution in the universe. The large field of view and the depth of a pencil beam survey would be ideal for weak lensing measurements of field galaxies, which can be used to constrain the large scale structure in the universe. -- 12 -- GRBs are perhaps the most energetic astrophysical events in the universe. Currently, there are many competing theories to explain GRBs. GRB afterglows contain valuable information, a statistically significant sample of GRB afterglows can provide strong constraints on the GRB theory. If beaming is involved in GRBs, we expect most of the GRB afterglows not to be associated with observable bursts. Schmidt et al. (1998) have observed a few optical transients which were too short in duration to be SNe; these could be GRB afterglows. Since present observational data seem to indicate that GRBs have associated host galaxies, the detection of host galaxies associated with short optical transients would support the interpretation of the latter as GRB afterglows. Since the GRB host galaxies are typically fainter than R = 25, a pencil beam survey would be ideal in detecting these host galaxies of candidate GRB afterglows. 5. Summary and discussion We have proposed a pencil beam survey of SNe Ia which can yield from tens to hundreds of SNe Ia per 0.1 redshift interval for z up to 1.5, which would enable the quantitative consideration of the systematic uncertainties of SNe Ia as standard candles, in particular, luminosity evolution and gravitational lensing. Using the Perlmutter et al. "batch" search technique repetitively over the same field in the sky, the pencil beam survey would be efficient in the discovery of SNe at z >∼ 1 by allowing the comparison of the latest frame of images (which may consist of combined data from successive nights) with all previous frames. The direct product of such a survey is the SN number count as a function of redshift (see §3), which is a combined measure of the cosmological parameters and the SN rate. When the measurements of the cosmological parameters are applied as priors to the number count, we obtain a direct measure of the SN rate, which is a key parameter in the formation and evolution of galaxies. The non-type-Ia SNe discovered by the pencil beam survey may be comparable to the type Ia SNe in number. The most important and straightforward application of the data from the SN pencil beam survey is to reduce the gravitational lensing noise in a SN Ia Hubble diagram via flux averaging. We have simulated SN Ia luminosities by adding weak lensing noise (using empirical fitting formulae given by Wang 1999a) and scatter in SN Ia absolute magnitudes to standard candles placed at random redshifts. We have shown that flux-averaging is powerful in reducing the combined noise of gravitational lensing and SN Ia absolute magnitude scatter (see §2). Because of the non-Gaussian nature of the luminosity distribution of SNe Ia at given z due to weak lensing, the large number of SNe Ia in a given redshift interval at high z is essential for the proper modeling and removal of the gravitational lensing effect. The SN Ia luminosity distribution in each redshift interval can be used to constrain the cosmological model (in particular, the fraction of matter in compact objects) by comparison -- 13 -- with predictions of numerical ray-shooting. (Holz & Wald 1998) The completeness of the SN Ia sample determines the effectiveness of the removal of gravitational lensing noise from the SN Ia Hubble diagram and the amount of information contained in the SN Ia luminosity distribution in each redshift interval. Note that the magnitude limit of the survey can lead to observational bias against the most distant demagnified SNe, therefore, the SNe which are close to the magnitude limit of the survey should not be used to probe cosmology in the manner described in this paper. To ensure the maximum usefulness of the data, scrupulous attention will have to be paid to photometric calibration, uniform treatment of nearby and distant samples, and an effective way to deal with reddening. (Riess et al. 1998) Note that although we can remove or reduce the effect of gravitational lensing on the SN Ia Hubble diagram, other systematics can affect the observed luminosity of SNe Ia. For example, grey dust, an evolution of the reddening law, or evolution in the peak absolute luminosity of SNe Ia. While it is possible to constrain grey dust and determine dust evolution through multi-band photometry at significantly different redshifts, the dimming with z in peak absolute luminosity of SNe Ia is degenerate with the effect of low matter density. (Aguirre 1999, Riess et al. 1999, Drell, Loredo, & Wasserman 1999, Wang 1999b) Evolution will remain a caveat in the usage of SNe Ia as cosmological standard candles, unless one can somehow correct for the effect of evolution. It is critical to obtain up to hundreds of SNe Ia at z > 1 to constrain luminosity evolution, because we expect the luminosity evolution and low matter density to affect the distance modulus of SNe Ia through different functionals of z, which should become distinguishable at z > 1. We have proposed a survey cutoff of z = 1.5 mainly for two reasons. First, going to higher redshift makes obtaining the spectra of the SNe (which are needed to distinguish different types of SNe) practically impossible from the ground; even obtaining the spectra of z ∼ 1.5 SNe Ia may prove impractical from the ground (as it would require several clear nights per spectrum on the Keck, see §4). Observers have already demonstrated that SNe Ia up to z = 1 can be found in the ongoing searches. (Goobar & Perlmutter 1995, Garnavich et al. 1998, Perlmutter et al. 1999) At z ∼ 1.5, the SNe Ia should be ∼ 2.5 magnitudes fainter than at z ∼ 1, it should be possible to discover them on the ground through deep imaging, and the follow-up spectroscopy can be done on NICMOS or NGST (see §4). Second, predicted rest-frame SN Ia rate per comoving volume as function of redshift seems to peak at z ∼ 1. (Yungelson & Livio 1998, Sadat et al. 1998) A pencil beam survey of SNe up to z = 1.5 will enable the accurate determination of the SN rate as function of redshift in the redshift region important for studying cosmic star formation rate and the SN Ia progenitor models. Although the Next Generation Space Telescope can detect SNe at as high redshifts as they exist (Stockman et al. 1998), the estimated rate of detection is of order 20 SNe II per 4×4 arcmin2 field per year in the interval 1 < z < 4 (Madau et al. 1998), and the detection rate of SNe Ia is likely smaller. Thus a ground-based pencil beam survey of SNe is essential to complement the space-based SN searches. -- 14 -- The most challenging aspect of the SN pencil beam survey is obtaining spectra for the SNe Ia at redshifts close to 1.5. Instead of waiting for future space equipments, we may find innovative ways of obtaining spectra from the ground. The referee has pointed out that since we can find multiple SNe at the same time on the same one square degree field (the number of SNe depends on cosmology), it may be possible to get multiple spectra at once with fibers. The nominal numbers we have used for the proposed SN pencil beam survey, a one square degree field, a depth of z = 1.5, and an effective observation duration of one year (which is equivalent to several years of actual observation), are optimistic but not implausible. The large sky coverage and the long effective observation duration will probably require a large consortium of existing and new SN search teams through, for example, a dedicated 4 meter telescope which can be used at the same time for other important observational projects compatible with the SN pencil beam survey (see §4). The goal of going up to z = 1.5 in spectroscopy will require support from the Keck LRIS and NICMOS/NGST. We conclude by noting that a SN pencil beam survey can yield enormous scientific return. The observational efforts directed towards a SN pencil beam survey should be very rewarding. Acknowledgements It is a pleasure for me to thank Joachim Wambsganss for generously providing unpublished magnification distributions; Zeljko Ivezic and Todd Tripp for explaining technical details concerning photometry and spectroscopy; Saul Perlmutter for communicating details of the current supernova search by the Supernova Cosmology Project; Ed Turner for a careful reading of a draft of the manuscript and for helpful comments, the referee for encouraging and useful comments; Christophe Alard, Jim Gunn, David Hogg, Rocky Kolb, Robert Lupton, Michael Strauss, and Tony Tyson for helpful discussions. -- 15 -- REFERENCES Aguirre, A.N. 1999, astro-ph/9904319 Carroll, S.M., Press, W.H., & Turner, E.L. 1992, ARA&A, 30, 499 Drell, P.S.; Loredo, T.J.; & Wasserman, I. 1999, astro-ph/9905027 Dyer, C.C., & Roeder, R.C. 1973, ApJ, 180, L31. Ellis, R.S., Colless, M., Broadhurst, T., Heyl, J., & Glazebrook, K. 1996, MNRAS, 280, 235 Ferrini, F., & Poggianti 1993, ApJ, 410, 44 Frieman, J.A. 1997, Comments Astrophys., 18, 323, astro-ph/9608068. Fukugita, M., & Turner, E.L. 1991, MNRAS, 253, 99 Garnavich, P.M. et al. 1998, ApJ, 493, L53 Goobar, A. & Perlmutter, S. 1995, ApJ, 450, 14 Holz, D.E., & Wald, R.M. 1998, Phys. Rev. D, 58, 063501. Holz, D.E. 1998, ApJ, 506, L1. Lilly, S.J., Tresse, L., Hammer, F., Crampton, D., & Le F`evre, O. 1995, ApJ, 455, 108. Linder, E.V. 1998, ApJ, 497, 28 Loveday, J., Peterson, B.A., Efstathiou, G., Maddox, S.J. 1992, ApJ, 390, 338 Kantowski, R., Vaughan, T., & Branch, D. 1995, ApJ, 447, 35 Kantowski, R. 1998, ApJ, 507, 483 Kolatt, T.S., & Bartelmann, M. 1998, MNRAS, 296, 763 Madau, P. 1998, in D'Odorico, S., Fontana, A. & Giallongo, E. eds, PASP, The Young Universe: Galaxy Formation and Evolution at Intermediate and High Redshift Madau, P. 1998, Della Valle, M., & Panagia, N. 1998, MNRAS 297, L17 Metcalf, R.B. 1999, MNRAS, 305, 746 Pain, R. et al. 1996, ApJ, 473, 356 Perlmutter, S., et al. 1997, ApJ, 483, 565 -- 16 -- Perlmutter, S.; Aldering, G.; Goldhaber, G.; Knop, R.A.; Nugent, P.; Castro, P.G.; Deustua, S; Fabbro, S; Goobar, A; Groom, D.E.; Hook, I.M.; Kim, A.G.; Kim, M.Y.; Lee, J.C.; Nunes, N.J.; Pain, R.; Pennypacker, C.R.; Quimby, R.; Lidman, C.; Ellis, R.S.; Irwin, M.; McMahon, R.G.; Ruiz-Lapuente, P.; Walton, N.; Schaefer, B.; Boyle, B.J.; Filippenko, A.V.; Matheson, T.; Fruchter, A.S.; Panagia, N.; Newberg, H.J.M.; & Couch, W.J. 1999, ApJ, 517, 565 Renzini, A., Ciotti, L., D'Ercole, A., & Pellegrini, S. 1993, ApJ, 419, 52 Riess, A.G., Press, W.H., & Kirshner, R.P. 1995, ApJ, 438, L17 Riess, A.G.; Filippenko, A.V.; Challis, P.; Clocchiatti, A.; Diercks, A.; Garnavich, P.M.; Gilliland, R.L.; Hogan, C.J.; Jha, S.; Kirshner, R.P.; Leibundgut, B.; Phillips, M.M.; Reiss, D.; Schmidt, B.P.; Schommer, R.A.; Smith, R.C.; Spyromilio, J.; Stubbs, C.; Suntzeff, N.B.; & Tonry, J 1998, AJ, 116, 1009 Riess, A.G., et al. 1999, astro-ph/9907037 Ruiz-Lapuente, P. & Canal, R. 1998, ApJ, 497, L57 Sadat, R.; Blanchard, A.; Guiderdoni, B.; & Silk, J. 1998, Astron. Astrophys., 331, L69 Schneider, P., Ehlers, J., & Falco, E.E. 1992, Gravitational Lenses, Springer-Verlag, Berlin Schmidt, B.P., et al. 1998, ApJ, 507, 46 Stockman, H.S.; Stiavelli, M.; Im, M.; & Mather, J. 1998, in Smith, E.P.,& Koratkar, A. eds, ASP Conf. Ser. Vol. 133, Science with the Next Generation Space Telescope, Astron. Soc. Pac., San Francisco Turner, E.L. 1990, ApJ, 365, L43 van den Bergh, S., & McClure, R.D. 1994, ApJ, 425, 205 Wambsganss, J., Cen, R., Xu, G., & Ostriker, J.P. 1997, ApJ, 475, L81 Wambsganss, J. 1999, private communication. Wang, Y. 1999a, astro-ph/9901212, ApJ, in press Wang, Y. 1999b, astro-ph/9907405, ApJ, accepted Yungelson, L. & Livio, M. 1998, ApJ, 497, 168 This preprint was prepared with the AAS LATEX macros v4.0. -- 17 -- Fig. 1. -- (a) Magnitude versus redshift for three cosmological models, SCDM (Ωm = 1, ΩΛ = 0), OCDM (Ωm = 0.2, ΩΛ = 0), and ΛCDM (Ωm = 0.2, ΩΛ = 0.8). For each cosmological model, the upper curve represents the completely clumpy universe (empty beam, α = 0), while the lower curve represents the completely smooth universe (filled beam, α = 1). (b) The same models relative to smooth OCDM (filled beam, α = 1), the middle curve for each model represents a universe with half of the matter smoothly distributed (half-filled beam, α = 0.5). Fig. 2. -- The number of SNe Ia expected per 0.1 redshift interval as function of redshift for three cosmological models, SCDM (Ωm = 1, ΩΛ = 0), OCDM (Ωm = 0.2, ΩΛ = 0), and ΛCDM (Ωm = 0.2, ΩΛ = 0.8), for a one square degree field, and an effective observation duration of one year. Fig. 3. -- The number of strongly lensed SNe as function of survey depth z, for the same cosmological models as in Fig.1 and Fig.2, for a one square degree field, and an effective observation duration of one year. Fig. 4. -- The parameter dependence of the SN Ia number count per 0.1 redshift interval as function of redshift. Note that the dependence of the SN number count on the SN rate is parametrized by A and B. -- 18 -- Fig. 1. -- (a) Magnitude versus redshift for three cosmological models, SCDM (Ωm = 1, ΩΛ = 0), OCDM (Ωm = 0.2, ΩΛ = 0), and ΛCDM (Ωm = 0.2, ΩΛ = 0.8). For each cosmological model, the upper curve represents the completely clumpy universe (empty beam, α = 0), while the lower curve represents the completely smooth universe (filled beam, α = 1). -- 19 -- Fig. 1. -- (b) The same models relative to smooth OCDM (filled beam, α = 1), the middle curve for each model represents a universe with half of the matter smoothly distributed (half-filled beam, α = 0.5). -- 20 -- Fig. 2. -- The number of SNe Ia expected per 0.1 redshift interval as function of redshift for three cosmological models, SCDM (Ωm = 1, ΩΛ = 0), OCDM (Ωm = 0.2, ΩΛ = 0), and ΛCDM (Ωm = 0.2, ΩΛ = 0.8), for a one square degree field, and an effective observation duration of one year. -- 21 -- Fig. 3. -- The number of strongly lensed SNe as function of survey depth z, for the same cosmological models as in Fig.1 and Fig.2, for a one square degree field, and an effective observation duration of one year. -- 22 -- Fig. 4. -- The parameter dependence of the SN Ia number count per 0.1 redshift interval as function of redshift. Note that the dependence of the SN number count on the SN rate is parametrized by A and B.
0811.0673
1
0811
2008-11-05T12:31:32
Planet formation in the habitable zone of alpha Centauri B
[ "astro-ph" ]
Recent studies have shown that alpha Centauri B might be, from an observational point of view, an ideal candidate for the detection of an Earth-like planet in or near its habitable zone (0.5-0.9AU). We study here if such habitable planets can form, by numerically investigating the planet-formation stage which is probably the most sensitive to binarity effects: the mutual accretion of km-sized planetesimals. Using a state-of-the-art algorithm for computing the impact velocities within a test planetesimal population, we find that planetesimal growth is only possible, although marginally, in the innermost part of the HZ around 0.5AU. Beyond this point, the combination of secular perturbations by the binary companion and gas drag drive the mutual velocities beyond the erosion limit. Impact velocities might later decrease during the gas removal phase, but this probably happens too late for preventing most km-sized objects to be removed by inward drift, thus preventing accretion from starting anew. A more promising hypothesis is that the binary formed in a crowded cluster, where it might have been wider in its initial stages, when planetary formation was ongoing. We explore this scenario and find that a starting separation roughly 15 AU wider, or an eccentricity 2.5 times lower than the present ones are required to have an accretion-friendly environment in the whole HZ.
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 000 -- 000 (0000) Printed 27 October 2018 (MN LATEX style file v2.2) Planet formation in the habitable zone of alpha Centauri B P. Th´ebault1⋆, F. Marzari2, H.Scholl3 1Stockholm Observatory, Albanova Universitetcentrum, SE-10691 Stockholm, Sweden, and Observatoire de Paris, Section de Meudon, F-92195 Meudon Principal Cedex, France 2Department of Physics, University of Padova, Via Marzolo 8, 35131 Padova, Italy 3Laboratoire Cassiop´ee, Universit´e de Nice Sophia Antipolis, CNRS, Observatoire de la Cote d'Azur, B.P. 4229, F-06304 Nice, France Draft version 27 October 2018 ABSTRACT Recent studies have shown that alpha Centauri B might be, from an observational point of view, an ideal candidate for the detection of an Earth-like planet in or near its habitable zone (0.5-0.9AU). We study here if such habitable planets can form, by numerically investigating the planet-formation stage which is probably the most sensitive to binarity effects: the mutual accretion of km-sized planetesimals. Using a state-of-the-art algorithm for computing the im- pact velocities within a test planetesimal population, we find that planetesimal growth is only possible, although marginally, in the innermost part of the HZ around 0.5AU. Beyond this point, the combination of secular perturbations by the binary companion and gas drag drive the mutual velocities beyond the erosion limit. Impact velocities might later decrease during the gas removal phase, but this probably happens too late for preventing most km-sized ob- jects to be removed by inward drift, thus preventing accretion from starting anew. A more promising hypothesis is that the binary formed in a crowded cluster, where it might have been wider in its initial stages, when planetary formation was ongoing. We explore this scenario and find that a starting separation roughly 15 AU wider, or an eccentricity 2.5 times lower than the present ones are required to have an accretion-friendly environment in the whole HZ. Key words: planetary systems: formation -- stars: individual: α Centauri -- planets and satel- lites: formation. 1 INTRODUCTION The study of planet formation in binary stars is an important issue, as approximately ∼ 20% of all detected exoplanets inhabit such systems (Desidera & Barbieri 2007). Even if most of these binaries are very wide, at least 3 planets have been found in binaries of sepa- ration ∼20 AU, the most interesting of them being the M > 1.6MJup planet at 2.1AU from the primary of the γ Cephei system, which has been the subject of several planet-formation studies (Th´ebault et al. 2004; Kley & Nelson 2008; Jang-Condell et al. 2008). However, the system which has been most thoroughly investi- gated is one where no planet has been detected so far: alpha Cen- tauri 1. Holman & Wiegert (1997) have shown that, for the copla- nar case, the < 3 AU region around α Cen A can harbour planets on stable orbits, while Barbieri et al. (2002) and Quintana et al. (2002, 2007) have found that the final stages of planetary formation, the ones leading from large planetary embryos to the final planets, are possible in the inner < 2.5 AU region. However, Th´ebault et al. (2008,hereafter TMS08) showed that the earlier stage leading, through runaway growth in a dynamically quiet environment (e.g. ⋆ E-mail:[email protected] 1 Endl et al. (2001) rule out the presence of a M > 2.5 MJupiter planet at any radial distance Lissauer 1993), from kilometre-sized planetesimals to the embryos is much more affected by the companion star's perturbations. In- deed, the coupled effect of these secular perturbations and drag due to the gas in the nebulae leads to a strong orbital phasing according to planetesimal sizes. This differential phasing induces high colli- sion velocities for any impacts between objects of sizes s1 , s2, which can strongly slow down or even halt the accretion process. As a result, the region beyond 0.5 AU (0.75 AU for the most ex- treme cases explored) from the primary is hostile to mutual accre- tion of planetesimals, and thus to the in situ formation of planets. The presence of planets in these regions, however, cannot be ruled out, in the light of possible additional mechanisms not taken into account in the simulations, such as a change of the dynamical en- vironment once gas has been dispersed, or outward migration of planets formed closer to the star, or a change in the binary's sepa- ration during it's early history. We reinvestigate here these issues for the case of the secondary star α Centauri B. Indeed, a recent study by Guedes et al. (2008) has shown that, in addition to the final embryos-to-planets stage be- ing possible within ∼ 2.5 AU from it, this star would be an almost perfect candidate for the detection of a potential terrestrial planet by the radial velocity method. This is mainly because of its great proximity, but also because it is exceptionally quiet. Moreover, be- cause of its lower luminosity, its habitable zone (HZ) lies much 2 P. Th´ebault, F. Marzari, H. Scholl closer, 0.5 to 0.9 AU (Guedes et al. 2008), than that of α Cen A (∼1 to 1.3 AU, e.g. Barbieri et al. 2002). In this letter, we first numerically study the dynamics of the planetesimal accretion phase in a similar way as for α Cen A, by es- imating the extent of the accretion-friendly region around the cen- tral star (section II). We then push our studies a step further than in TMS08, by quantitatively exploring two mechanisms which could potentially increase the odds for planet formation in the HZ (Sec- tion III). Conclusions and perspectives are given in the last Section. 2 SIMULATIONS 2.1 model and setup We consider a disc of 104 test planetesimals extending from 0.3 to 1.6 AU from the central star, starting on quasi-circular orbits such as initial encounter velocities h∆vi < 1 m.s−1, and with a physical radius s in the 1km< s <10km range. The binary's parameters are taken from Pourbais et al. (2002): semi-major axis ab = 23.4 AU, eccentricity eb = 0.52, MA = 1.1M⊙ and MB = 0.93M⊙. We fol- low the dynamical evolution of the planetesimal population for t = 104years, the typical timescale for runaway growth, under the coupled influence of the companion star's (here α Cen A) per- turbations and gas drag, using the deterministic code developed by Th´ebault & Brahic (1998). All test planetesimal collisions are tracked with the code's build-in close-encounter search routine, looking at each time step for all intersecting particle trajectories with the help of the classical "inflated radius" procedure (see sec- tion 2.1 of TMS08 for more details). The gas disc is assumed ax- isymmetric with a volumic density profile following ρg = ρg0 rq (AU), where we consider as a nominal case the Minimum Mass Solar Nebula (MMSN), with q = −2.75 and ρg0 = 1.4 × 10−9g.cm−3 (Hayashi 1981), exploring ρg0 and q as free parameters in separate runs. All mutual encounters and corresponding encounter velocities are tracked and recorded in order to derive h∆vis1 ,s2 matrices for all possible impactor pairs of sizes s1 and s2. These velocity values are then interpreted in terms of accreting or eroding encounters by comparing them, for each impacting pair, to 2 threshold velocities (for more details, see Th´ebault et al. 2006, 2008): • vesc(s1 ,s2), the escape velocity of the impacting pair. This ve- locity gives approximately the limit below which single star-like runaway accretion is possible (when accretion rates are enhanced by the gravitational focusing factor). • v∗(s1 ,s2), the velocity above which impacts preferentially lead to mass loss rather than mass accretion. This value is obtained from estimates of the critical threshold energy for catastrophic fragmen- tation Q∗(s1,s2). Given the large uncertainties regarding the possible values for Q∗(s1 ,s2), we consider 2 limiting values, v∗(s1 ,s2)low and v∗(s1 ,s2)high, corresponding to a weak and hard material assumption respectively. Finally, at any given location in the system, the global balance be- tween accreting and eroding impacts is estimated assuming a size distribution for the planetesimal population, weighting the contri- bution of each (s1, s2) impact by its normalized probability accord- ing to the assumed size distribution. As a nominal case, we assume a Maxwellian peaking at the median value smed = 5 km, but other distributions (Gaussians, power laws) are also explored. Figure 1. eccentricity vs. semi-major axis graph at the end (104years) of the nominal case run. The colour scale shows the different particle sizes. The two vertical lines indicate the estimated location of the habitable zone. Figure 2. Relative importance of different types of collision outcomes, as a function of distance to the primary star (α Cen B), at t=104years for our nominal case. Each ∆v(s1,s2) for an impact between bodies of sizes s1 and s2 is interpreted in terms of eroding impacts (if ∆v > v∗(s1 ,s2)high), accreting impacts despite of increased velocities ("perturbed accretion" for vesc(s1 ,s2) < ∆v < v∗(s1 ,s2)low), or unperturbed "normal" accretion (∆v < vesc(s1 ,s2)). For v∗(s1 ,s2)low < ∆v < v∗(s1 ,s2)high, we do not draw any conclusions. The contribution of each impact is then weighted assuming that the size distribution for the planetesimals follows a Maxwellian cen- tered on 5km (see text for details). 2.2 results: nominal case The (e,a) graph displayed in Fig.1 clearly illustrates the effect of differential orbital phasing as a function of sizes. Smaller planetes- imals align to orbits with lower eccentricities than that of the bigger objects. 2. Note that in the outer regions (beyond ∼ 1 AU), residual eccentricity oscillations are observed for the larger bodies. This is because these objects, which are less affected by gaseous friction, 2 This differential alignment in e is strengthened by the associated differ- ential alignment of the pericenter angle (e.g., Th´ebault et al. 2006) Planet formation in the habitable zone of alpha Centauri B 3 Table 1. Outer limit racc(out) of the accretion-friendly inner zone, as defined by the region where the "green" and "blue" areas make up more than 50% of the collision outcomes (see Fig.2), for test runs exploring different free parameters: gas density ρg0 at 1AU, slope q of the ρg ∝ rq gas density profile, and assumed size distribution for the planetesimal population discussion in Th´ebault et al. 2006). For any "reasonable" distri- butions, we do not observe a significant displacement of the ac- cretion/erosion radial limit. Only a very peaked, almost Dirac-like Gaussian of variance σ2 = (0.5 km)2 leads to mostly accreting im- pacts in the whole HZ (Tab.1). Set-Up Nominal case ρg0 = 1.4 × 10−10g.cm−3 (0.1xMMSN) ρg0 = 1.4 × 10−8g.cm−3 (10xMMSN) ρg ∝ r−2.25 (α viscous disc) ρg ∝ r−1.75 Size dist.: Gaussian, σ2 = (2 km)2 Size dist.: Gaussian, σ2 = (1 km)2 Size dist.: Gaussian, σ2 = (0.5 km)2 Size dist.: power law dN ∝ s−3.5ds racc(out) 0.5 AU 0.3 AU 0.8 AU 0.45 AU 0.4 AU 0.55 AU 0.6 AU 1.1 AU 0.5 AU have not had the time yet to fully reach the equilibrium orbit forced by gas drag. Fig.2 shows how h∆vis1 ,s2 resulting from this differential phas- ing can be interpreted in terms of accreting versus eroding impacts. This graph displays the relative weights of the possible types of col- lision outcomes as a function of distance to α Cen B, for our nomi- nal case (MMSN gas disc), assuming that the planetesimal size dis- tribution follows a Maxwellian centered on 5km. By comparing this graph to the equivalent graph for α Cen A (Fig.3 of TMS08), we see that results are roughly comparable for both stars. Schematically, the region beyond ∼0.5 AU from the primary is hostile to planetary accretion. This similarity of the results is mostly due to the fact that the mass ratio between the 2 stars is close to 1 (∼ 0.85), and that the strongest perturbations of α Cen A on α Cen B are partly com- pensated by the lower Keplerian velocities around that smaller star. One important difference, however, is that the HZ around α Cen B lies closer to the star: 0.5 to 0.9 AU (Guedes et al. 2008). This means that its innermost part lies at the outer-edge of the accretion- friendly region, thus opening a possibility, although marginal, for planet formation there. 2.3 parameter exploration To check the robustness of these results with respect to our choice of parameters for the nominal case, we firstly explore other possible gas disc profiles, by varying both the slope q and the density ρg0 at 1AU. We find that the only case which leads to a wider accretion- friendly zone is that of a dense 10xMMSN gas disc, for which this region extends up to ∼ 0.8 AU (Tab.1). All other gas-disc cases lead to equally narrow or even narrower accretion-friendly regions than for our nominal case, except for a purely academic gas-free case. Of course, our axisymmetric gas disc assumption is probably a crude simplification of the real behaviour of a gas disc perturbed by a companion star. However, preliminary Hydro+N-body runs performed by Paardekooper et al. (2008) show that the build up of an eccentricity for the gas-disc, altough this mechanism is yet not fully understood, always lead to higher perturbations and impact velocities in the planetesimal population. As a consequence, our axisymmetric disc case might be regarded as a best-case scenario and our results to be on the conservative side (for a more detailed discussion and justification for our choice of an axisymmetric disc, see TMS08). Another crucial parameter which has been explored is the as- sumed size distribution for the planetesimal population, as it is a parameter poorly constrained from planet-formation models (see 3 DISCUSSION The results of the previous section show that, as was the case for α Cen A, the region allowing km-sized planetesimal accre- tion is much more limited than the one allowing the final stages of planet formation (starting from large embryos): the accretion- friendly zone only extends up to ∼ 0.5 AU for our nominal case, as compared to the estimated 2.5 AU for the embryos-to-planets phase (Guedes et al. 2008, and reference therein). The only cases leading to an accretion-friendly environment in a wider area than for the nominal run are: a) the gas free case, b) the very-peaked Gaussian size distribution, and c) the 10xMMSN run. Cases a) and b) are obviously very unrealistic test runs, but even case c) is probably here of limited physical significance. In- deed, assuming such a high-mass disc around a 0.9M⊙ star appears as a rather extreme hypothesis, which is not backed by theoretical estimates or observational data (e.g. Andrews & Williams 2007). For a binary system, this issue gets even more critical, as the ex- pected outer truncation of the circumprimary gas disc probably depletes it from a substancial amount of it mass reservoir (e.g., Jang-Condell et al. 2008), making the 10xMMSN hypothesis even more questionable, although it cannot be completely ruled out. Apart from this massive gas disc case, the racc(out) ∼ 0.5 AU limit allows in principle for planetesimal accretion to occur in the innermost part of the habitable zone, which is estimated to extend from 0.5 to 0.9AU (as compared to ∼ 1 to 1.3 AU for α Cen A). However, even if some planetesimal accretion is possible, it could probably not be what it is in the standard planet formation sce- nario. Indeed, even if the ∼ 0.5 AU region is accretion-friendly, relative velocities there have values higher than what they should be in an unperturbed system (that is, approximately the escape ve- locities of km-sized bodies, e.g Lissauer 1993). This means that the gravitational focusing factor, which is the cause of the fast runaway growth mode, is significantly reduced or cancelled. Growth would thus have to be much slower, maybe orderly or of the "type II" run- away identified by Kortenkamp et al. (2001). In short, planetesimal accretion is not possible in the HZ, except in the innermost region around 0.5 AU where it should be significantly different from what it is around a single star. However, no conclusions regarding the presence of planets in the HZ can be directly drawn from these results, because of the possible effect of additional processes not taken into account in our simulations. In TMS08, we reviewed some of these possible mech- anisms, and concluded that, while some of them might not signifi- cantly affect our conclusions (like bigger "initial" planetesimals or re-accretion in the subsequent gas-free disc), others might indeed open for a possible presence of a planet in the terrestrial region. We reinvestigate here these mechanisms in more details, focusing specifically on the two most promising scenarios, that of orbital re-phasing during an extended gas dispersion phase and that of a greater separation for the early binary. 3.1 re-phasing during gas dispersion Xie & Zhou (2008) have investigated, for the specific case of the γ 4 P. Th´ebault, F. Marzari, H. Scholl Figure 3. (e,a) graph, after t = 2×105years, for the run with gas dissipation. Gas dissipation starts at t=104years and has a timescale τdiss = 105years (see text for details) Figure 4. Dominant collision outcome (erosion, perturbed accretion, un- perturbed accretion, see Fig.2) in the habitable zone (0.5-0.9 AU) around α Centauri B, for different configurations of the binary's orbit. Cephei binary, what happens to a planetesimal swarm during the gas dispersal period. They have identified an interesting mecha- nism: if the gaseous component is slowly removed from the disc, all planetesimal orbits are progressively re-phased towards the same orbits regardless of their sizes. This leads to lower, possi- bly accretion-friendly impact velocities. They concluded that this might allow, under certain circumstances, for planetesimal accre- tion to start anew once this realignment is achieved. We explore this possibility for the present α Cen B case, assuming the same param- eters for gas dispersion as Xie & Zhou (2008): ρg ∝ (t/τdiss)−1.5, with τdiss = 105years. The dissipation is started at 104years, the end of our nominal run, when all orbits have reached their size- dependent alignment in the HZ. Fig.3 shows the dynamical state of the system after 2τdiss = 2x105years and clearly illustrates the efficient re-phasing of all or- bits, which is almost complete in the > 0.8 AU region. In this outer region, h∆vi are low enough to allow accretion of all remaining ob- jects. However, this encouraging result is undermined by several problems. The first one is that, in this accretion-friendly outer re- gion, no object smaller than ∼ 4 km is left after 2τdiss, because the time it takes for most of the gas to be dispersed is long enough (a few 105years) to have a significant inward drift of all smaller bodies 3. True, in most of the habitable zone small planetesimals are still present, but here full orbital re-phasing is not achieved yet, and the dynamical environment, even if the situation is slighly im- proved with respect to the nominal run, is still globally hostile to accretion. Only after ∼ 5τdiss = 5 × 105years do h∆vi become low enough to allow accretion in the whole HZ, by which time all small objects have here also been removed. The second problem is that, even for the larger > 4 km planetesimals, the favourable conditions shown in Fig.3 are only reached after an accretion-hostile transition period of a few τdiss. The question is then how these objects might survive this long pe- riod during which most impacts will erode them. These erosion processes cannot be followed with the N-body models used here, 3 This issue had already been identified by Xie & Zhou (2008) but didn't show up in their simulations because of the numerically imposed re- injection at the outer boundary of all objects lost at the inner one but it is likely that most large planetesimals will be fragmented into smaller debris which will be quickly removed from the system by fast inward drift. For all these reasons, we do not believe that gas dispersal opens the possiblity for planetesimal accretion to start anew after a few τdiss. However, the re-phasing mechanism during gas disper- sal identified by Xie & Zhou (2008) is definitely worth studying in more details. One crucial, and yet almost unexplored issue is for in- stance the relative timing between the planetesimal accretion phase (in particular its start) and the parallel viscous evolution of the gas disc. Indeed, as pointed out by Xie & Zhou (2008), the steady evo- lution of the gas disc profile, due to viscous angular momentum redistribution and accretion, can lead to a gas density decrease of several orders of magnitudes long before its "definitive" and abrupt dispersion, at 5 to 10Myrs, resulting from photoevaporation- induced disc truncature (see e.g., Fig.1 of Alexander et al. 2006). The question is where (or when) to place the planetesimal accre- tion phase in this picture. 3.2 Wider initial binary separation Most stars are born in clusters. As such they spend their early life in an environment where the risk for close stellar encounters is rel- atively high. For a binary system, such encounters could have dra- matic effects, breaking it up or significantly modifying its initial orbit. Conversely, a present day binary could be what is left of an initially triple (or more) system (Marzari & Barbieri 2007a,b). As a consequence, there might be an important difference between the present day orbit of a binary system and the one it had when plan- etary formation processes were at play. In order to have an idea of the statistical chances for an α Centauri-like binary to have an initially different orbit, we take as a reference the work of Malmberg et al. (2007), who studied the dynamical evolution of a typical open cluster. They found that most binaries of initial separation 6 200 AU are not broken up during the cluster's lifetime. However, a fraction of these "hard" binaries suffer stellar encounters which alter their orbits. The net result of these encounters is to shrink the binaries' orbits, so that separations of young binaries tend to be larger, with an environment which is thus more favourable to planet formation. Fig.4 of Malmberg et al. Planet formation in the habitable zone of alpha Centauri B 5 (2007) indicates that ∼ 50% of binaries with separation ∼ 20AU have suffered an orbit-modifying encounter. There is thus roughly 1 chance in 2 that α Centauri had a wider separation during the early stages of planet formation if it formed in a dense cluster. The crucial question is here which initial orbital configura- tions could have allowed planetesimal accretion to occur in the hab- itable zone of α Cen B. We investigate this issue by running a series of 31 test runs for 20 < ab < 50 AU and 0 < eb < 0.8 (Fig.4). Not surprisingly, we see that present day α Cen B lies almost at the limit where accretion, although highly perturbed, becomes possible in the innermost part of the HZ. For the same eb = 0.52 as today, par- tial perturbed accretion in the HZ is possible for 23 6 ab 6 37 AU. In order to have the HZ become fully accretion-friendly, one needs ab > 37 AU, while unperturbed single star-like accretion requires ab > 52 AU, more than twice the present day separation. The sit- uation gets of course more favourable if one assumes a lower ini- tial eb for the binary. As an example, with eb = 0.26 (half of the present day value), then the whole HZ becomes accretion friendly for ab > 26 AU and allows unperturbed accretion for ab > 40 AU. These results are of course only a first step towards under- standing this complex issue. A crucial point is to quantitatively es- timate what the probability for a given ∆ab and ∆eb "jump" are. As an example, how likely is it for a present day α Centauri-like system to have originated from the accretion-friendly regions (3) or (4) in Fig.4? The answer to these questions requires very detailed and en- compassing numerical explorations which have, to our knowledge, not been performed yet. Marzari & Barbieri (2007a,b) have studied in great detail the dynamical outcomes of individual encounters, but decoupled from the general cluster context, while the cluster evolution study of Malmberg et al. (2007) does not give statistical information about the amplitudes of binary orbital changes. 4 CONCLUSIONS AND PERSPECTIVES We have numerically investigated planetesimal accretion around α Centauri B. Our main conclusions can be summarized as follows: • Planetesimal accretion is marginally possible in the innermost parts, ∼0.5 AU, of the estimated habitable zone. Beyond this point, high collision velocities, induced by the coupling between gas fric- tion and secular perturbations, lead to destructive impacts. More- over, even in the ∼ 0.5 AU region, h∆vi are increased compared to an unperturbed case. Thus, "classical", single-star like runaway accretion seems to be ruled out. • These results are relatively robust with respect to the planetes- imal size distribution or gas disc profile, except for a very massive, and probably unrealistic 10xMMSN gas disc. • We confirm the conclusions of several previous studies that the planetesimal-to-embryo stage is much more affected by binarity effects than the subsequent embryos-to-final-planets stage. • As in Xie & Zhou (2008), we find that later progressive gas dispersal reduces all h∆vi to values that might allow accreting im- pacts. However, we find that the system has first to undergo a long accretion-hostile transition period during which most of the smaller planetesimals are removed by inward drift and most bigger objects are probably fragmented into small debris. Thus, the positive effect on planetesimal accretion is probably limited. • We quantitatively investigate to what extent a wider initial bi- nary separation, later reduced by stellar encounters in an early clus- ter environment, could have favoured planetesimal accretion. We find that, for a constant eb, an entirely accretion-friendly HZ re- quires an initial ab > 37 AU, while normal unperturbed accretion is only possible for ab > 52 AU. The statistical likehood of such orbital changes in early open clusters remains to be quantitatively estimated. We conclude that, although the presence of planets further out cannot be fully excluded, the most likely place to look for an habitable terrestrial planet would be around 0.5 AU. According to Guedes et al. (2008), the detection of such a planet could be pos- sible, if its mass is > 1.8M⊕, after 3 years of high cadence obser- vations, provided that the noise spectrum is white. Let us however point out again that the ∼ 0.5 AU region around α Cen B is dy- namically perturbed by the binary, so that the formation of a planet there cannot have followed the standard single-star scenario (unless the initial binary separation was much larger, see Section 3.2). The planet formation process in such a perturbed environment, which could concern many binary systems of intermediate (∼ 20 AU) sep- arations, is an important issue which remains to be investigated. Another crucial issue which remains to be investigated is what hap- pens before the phase studied here. Indeed, our study implicitly as- sumes that km-sized planetesimals could form from smaller grains and pebbles, but the validity of this assumption has to be critically examined. REFERENCES Alexander, R. D., Clarke, C. J., Pringle, J. E., 2006, MNRAS, 369, 229 Andrews, S. M.; Williams, J. P., 2007, ApJ, 659, 705 Barbieri, M.; Marzari, F.; Scholl, H., 2002, A&A, 396, 219 Desidera, S., Barbieri, M., 2007, A&A 462, 345-353 Endl, M.; Kurster, M.; Els, S.; Hatzes, A. P.; Cochran, W. D., 2001, A&A, 374, 675 Guedes, J. M.; Rivera, E. J.; Davis, E.; Laughlin, G.; Quintana, E.V.; Fischer, D.A., 2008, ApJ, 679, 1582 Hayashi, C., 1981, PthPS 70, 35 Holman, M.J., Wiegert, P. A. 1997, AJ, 113, 1445 Jang-Condell, H.; Mugrauer, M.; Schmidt, T., 2008, ApJL, 683, 191 Kley, W., Nelson, R., 2008, A&A, 486, 617 Kortenkamp, S., Wetherill, G., Inaba, S., 2001, Science, 293, 1127 Lissauer, J.J., 1993, ARA&A 31, 129 Malmberg, D.; Davies, M. B.; Chambers, J. E., 2007, MNRAS, 377, L1 Marzari, F. and Barbieri, M. 2007, A&A, 467, 347-351 Marzari, F. and Barbieri, M. 2007, A&A, 472, 643-647 Paardekooper, S.-J., Th´ebault, P., Mellema, G. 2008, MNRAS, 386, 973 Pourbaix, D.; et al., 2002, A&A, 386, 280 Quintana, E. V., Lissauer, Jack J.; Chambers, John E.; Duncan, Martin J., 2002, ApJ, 576, 982 Quintana, E. V., Adams, F. C., Lissauer, J. J., & Chambers, J. E. 2007, ApJ, 660, 807 Th´ebault, P., Brahic, A. 1998, P&SS, 47, 233 Th´ebault, P., Marzari, F., Scholl, H., 2006, Icarus, 183, 193 Th´ebault, P., Marzari, F., Scholl, H., 2008, MNRAS, 388, 1528 Th´ebault, P., Marzari, F., Scholl, H., Turrini, D., Barbieri, M., 2004, A&A, 427, 1097 Xie, Ji-Wei; Zhou, Ji-Lin, 2008, ApJ, 686, 570
astro-ph/0005535
1
0005
2000-05-26T07:05:00
On the Formation of Massive Stars by Accretion
[ "astro-ph" ]
(Abriged) At present, there are two scenarios for the formation of massive stars: 1) The accretion scenario and 2) The coalescence scenario, which implies the merging of intermediate mass stars. We examine here some properties of the first one. We calculate three different sets of birthlines, i.e. tracks followed by a continuously accreting star. First, three models with a constant accretion rate ($\dot{M}_{\rm{accr}}$ = $10^{-6}$, $10^{-5}$, $10^{-4}$ M$_{\odot}$ yr$^{-1}$). Then several birthlines following the accretion models of Bernasconi and Maeder (\cite{BM96}), which have $\dot{M}_{\rm{accr}}$ increasing only slightly with mass. Finally we calculate several birthlines for which $\dot{M}_{accr} = \dot{M}_{\mathrm{ref}} ({\frac{M}{M_{\odot}}}) ^{\phi}$, with values of $\phi$ equal to 0.5, 1.0 and 1.5 and also for different values of $\dot{M}_{\mathrm{ref}}$. The best fit to the observations of PMS stars in the HR diagram is achieved for $\phi$ between 1.0 or 1.5 and for $\dot{M}_{\mathrm{ref}} \simeq 10^{-5}$ M$_{\odot}$ yr$^{-1}$. Considerations on the lifetimes favour values of $\phi$ equal to 1.5. These accretion rates do well correspond to those derived from radio and IR observations of mass outflows. We emphasize the importance of the accretion scenario for shaping the IMF, and in particular for determining the upper mass limit of stars. In the accretion scenario, this upper mass limit will be given by the mass for which the accretion rate is such that the accretion induced shock luminosity is of the order of the Eddington luminosity.
astro-ph
astro-ph
A&A manuscript no. (will be inserted by hand later) Your thesaurus codes are: 03(08.06.2, 08.16.5, 08.05.10 ASTRONOMY AND ASTROPHYSICS On the Formation of Massive Stars by Accretion Peder Norberg1 and Andr´e Maeder2 1 Department of Physics, University of Durham, Durham DH1 3LE, UK 2 Observatoire de Gen`eve, CH-1290 Sauverny, Switzerland the date of receipt and acceptance should be inserted later Abstract. At present, there are two scenarios for the for- mation of massive stars: 1) The accretion scenario and 2) The coalescence scenario, which implies the merging of in- termediate mass stars. We examine here some properties of the first one. Radio and IR observations by Churchwell (1999) and Henning et al. (2000) of mass outflows around massive Pre-Main Sequence (PMS) stars show an increase by several orders of magnitudes of the outflow rates with stellar luminosities, and thus with stellar masses. As typ- ically, a fraction of 1 6 of the infalling material is esti- mated to be accreted, this suggests that the accretion rate is also quickly increasing with the stellar mass. 3 to 1 Maccr = Mref (cid:16) M M⊙(cid:17)ϕ We calculate three different sets of birthlines, i.e. tracks followed by a continuously accreting star. First, Maccr = three models with a constant accretion rate ( 10−6, 10−5, 10−4 M⊙ yr−1). Then several birthlines fol- lowing the accretion models of Bernasconi and Maeder Maccr increasing only slightly with (1996), which have mass. Finally we calculate several birthlines for which , with values of ϕ equal to 0.5, 1.0 Mref . The best fit and 1.5 and also for different values of to the observations of PMS stars in the HR diagram is Mref ≃ 10−5 achieved for ϕ between 1.0 or 1.5 and for M⊙ yr−1. Considerations on the lifetimes favour values of ϕ equal to 1.5. These accretion rates do well correspond to those derived from radio and IR observations of mass outflows. Moreover they also lie in the "permitted region" of the dynamical models given by Wolfire and Cassinelli (1987). We emphasize the importance of the accretion scenario for shaping the IMF, and in particular for determining the upper mass limit of stars. In the accretion scenario, this upper mass limit will be given by the mass for which the accretion rate is such that the accretion induced shock luminosity is of the order of the Eddington luminosity. Key words: stars: evolution - stars: pre-main sequence - Hertzsprung-Russell (HR) diagram - accretion Send offprint requests to: A. Maeder 1. Introduction The formation of massive stars is still a very uncertain domain of stellar astrophysics. Schematically, there are at present two very different scenarios. 1) The coalescence scenario proposed by Bonnell et al. (1998) and Stahler et al. (2000). The formation of massive stars (M ≥ 10M⊙) is assumed to occur by coalescence of stars of interme- diate masses, which form through accretion onto initially lower mass protostars. The basic reason for the develop- ment of this formation scenario was the difficulty of ac- creting mass onto very luminous stars. 2) The accretion scenario was initially proposed by Stahler et al. (1980a, 1980b, 1981) and was further developed for low and in- termediate mass stars (see for example Palla and Stahler, 1993). It was then investigated as a formation mechanism for massive stars by Beech and Mitalas (1994), Bernasconi and Maeder (1996). In this scenario, the massive stars no longer cross horizontally the HR diagram, coming from the red to the blue, on the Kelvin-Helmholtz timescale, but rise upwards in the HR diagram along the so-called birthline. The birthline is defined as the path in the HR diagram followed by a continuously accreting star. For low and intermediate mass stars, the birthline forms an upper envelope of individual evolutionary tracks in the HR dia- gram. The location of the birthline and the timescales on it Maccr (Bernasconi strongly depend on the accretion rates and Maeder, 1996; Tout et al. 1999). Thus, in this scenario it is very important to know how the accretion rate varies with the mass already accreted onto the star. Both scenarios have their own advantages and difficul- ties. They both influence the upper limit of stellar masses, but the physical mechanism determining this limit is of course different for each of the two above scenarios. In this paper we shall examine some properties of the accre- tion scenario, in order to provide further arguments in the debate. By Maccr, it is usually meant the accretion rate onto the central body, and this is our adopted viewpoint throughout this whole paper. In further more complete models, we will distinguish between the accretion from the molecular cloud to the disk and the accretion from the disk to the central protostar. 2 P. Norberg & A. Maeder: Massive Star Formation In Section 2, we examine some recent results on mass accretion and outflows in ultra-compact HII (UC HII) re- gions. In Section 3, we compare to the observations of pre-main sequence (PMS) stars some standard birthlines, i.e. with constant or slowly variable Maccr. In Section 4, we calculate new birthlines with quickly increasing accretion rates. In Section 5, we briefly give our conclusions, and in particular we discuss the issue of the maximum stellar mass in the accretion paradigm. 2. The accretion scenario for massive stars The accretion scenario, despite its successes (Palla and Stahler, 1993), has been considered to be impossible for massive stars, due to their high luminosities, which are able to reverse the collapse (Bonnell et al. 1998; Stahler et al. 2000). In this case, the radiation pressure on the dust is high enough, so that its momentum can be transferred to the gas (Wolfire and Cassinelli, 1987). In the accretion scenario, the accretion rate Maccr is an essential param- eter, since it determines the momentum of the infalling material. In their discussion on the study of the difficulties to form massive stars with the accretion scenario, Stahler et al. (2000) assume that the accretion rate behaves like Maccr ≃ 3 cs G (1) where cs is the sound speed in the molecular cloud. They stress that a remarkable property of this rate is that it is independent of the density of the parent cloud. The value of cs, of course, depends very much on the temperature of the cloud. Stahler et al. (2000) assume that the pre- collapse temperature is independent of the core density and mass, and also suggest that this does not change too much, if the infalling material goes via a disk instead of landing directly on the stellar surface. With the assump- tion that clouds where massive stars form have the same typical temperature T = 10-20 K as those where low mass stars form, Stahler et al. (2000) get accretion rates of 10−5 to 10−6 M⊙ yr−1. It is with this kind of assumptions that the constant accretion rates models by Beech and Mitalas (1994), and those with slowly varying accretion rates by Bernasconi and Maeder(1996) have been constructed. The birthline of these models joins the zero-age sequence when the heat released by the nuclear reactions stops the stellar contraction. This occurs typically around 8 to 10 M⊙. For massive stars with accretion rates of the order of 10−5 M⊙ yr−1, it is true that the momentum of the in- falling material is much smaller than the outwards radi- ation momentum of the star, which is thus able to re- verse the accretion process. Therefore such low mass ac- cretion rates are impossible for massive stars. This is by the way also confirmed by the existence, around these mas- sive stars, of stellar winds of similar magnitudes but blow- ing in the opposite direction. Moreover, we note that for 6 4 2 4.5 4 Fig. 1. Comparison between birthlines calculated with constant accretion rates and those calculated with expres- sion 3. applied to the models by Bernasconi and Maeder (1996). The models are given for F = 0.1, 1.0 and 10.0 Mcst = 10−6, 10−5 and and they correspond to about 10−4 M⊙ yr−1. The dot-broken lines are the PMS tracks for constant mass with the indicated value (Bernasconi and Maeder, 1996). The tracks with broken lines in the upper part are post-Main Sequence (post-MS) tracks for massive stars of 15 to 85 M⊙ by Schaller et al. (1992). Maccr the formation lifetimes would be longer this kind of than the main sequence lifetime! The dynamics of the in- falling material on protostars has been studied in detail by Wolfire and Cassinelli (1987). They found that the abun- dance of dust as well as the size of the grains should be reduced to allow infall. They examined the permitted re- gions in a plane log Maccr vs. log M , where M is the mass of the newly formed star. E.g. for a star of 40 M⊙, the allowed region lies between Maccr ≃ 10−4 and 10−2 M⊙ yr−1. For lower Maccr, the accretion is halted by the radia- Maccr the total luminosity (the tion field, while for higher protostellar luminosity and the accretion induced shock luminosity) exceeds the Eddington luminosity. There should be some relation between the accre- tion rates from collapsing clouds and their temperature T (Wolfire and Cassinelli (1987)). However, this is true only if the thermal support is the only source of support in the clouds. In this case, which is likely not realistic, the temperature of the collapsing clouds leading to massive stars should be as high as T ≥ 200 K and maybe up to 103 K. This results from the expressions of the Jeans mass and the free fall timescale, which imply that the average P. Norberg & A. Maeder: Massive Star Formation 3 6 4 2 4.5 4 Fig. 2. This grid of birthlines (continuous lines) is made out of twelve tracks with the following values of F = 0.1, 0.15, 0.2, 0.3, 0.5, 0.75, 1.0, 1.25, 1.75, 2.5, 3.5, 5.0. In general, the luminosity increases with F for a fixed value of the effective temperature. Pre-main sequence evolution tracks with constant mass are in dot-broken lines (Bernasconi and Maeder, 1996) and the post-main sequence tracks from Schaller et al. (1992) are in short dashed lines. The labels correspond to the mass along each track. Observations are issued from: 1. compilation done by Bernasconi and Maeder (1996); 2. Hillenbrand et al. (1992); 3. Damiani et al. (1994); 4. Cohen and Kuhi (1979); 5. van den Ancker et al. (1997a); 6. Berrilli et al. (1992); 7. de Winter et al. (1997); 8. Th´e et al. (1990); 9. van den Ancker et al. (1998); 10. van den Ancker et al. (1997b). For those stars with an error estimate on the luminosity and/or the effective temperature, the error bars are plotted. Moreover if several measures exist for a same star, we give the average value and indicate Maccr is given by the size of the symbol, and an open symbol signifies that the existing dispersion on it. The value of several groups have measured L and Teff for this star. inflow rate and the temperature of the collapsing cloud are increasing simultaneously. There is a variety of results on the temperature of UC HII regions (Churchwell 1999 and Henning et al. 2000). From their infrared emission, Churchwell finds, around them, a sizeable (∼ 1016 cm) dust evacuated cavity and he estimates that the temperature at the inner face of the dust shell is typically 300 K. In the extreme case of W3, a massive star forming region, there is even a hard X-ray emission, implying T up to 7 107 K in the wind-shocked cavity surrounding its central UC HII region (Hofner and Churchwell, 1997). The above results are likely to concern environments that have been altered by the presence of ex- isting massive stars. Indeed, for the first stars to be formed one need to consider the temperature of the cloud core just before star formation begins. In regions like Orion, the so-called massive dense cores, which presumably are the birthsites of massive stars, are not as hot as the UC 4 P. Norberg & A. Maeder: Massive Star Formation 4 3 2 1 4.6 4.4 4.2 4 3.8 3.6 Fig. 3. The continuous lines show the tracks calculated Mref = 10−5 and for ϕ = 1.5, 1.0 and 0.5 with expr. 4 for from top to bottom. The broken lines are the birthlines calculated with expression 3 and F= 0.5, 1.0 and 1.5. The dot-broken lines are the PMS tracks for constant mass with the indicated value (Bernasconi and Maeder 1996). HII regions studied by Churchwell (1999) and Henning et al. (2000) indicate them to be. The temperature is more likely to be well below 100K (Caselli & Myers 1995). Thus, the temperature of the molecular cloud is not necessarily the only parameter responsible for the enhanced accretion rates necessary to form massive stars. In massive star forming regions, like Orion, the velocity width of an observed line is dominated by non-thermal, su- personic motions (e.g. Caselli & Myers 1995). Thus, there is a significant contribution of turbulent motions to the support of the clouds. In equation (1) for the accretion rate, the thermal sound speed should be replaced by the sum of a thermal and non–thermal contribution. Turbu- lence takes a long time to be dissipated and persists dur- ing a significant part of the formation process. Caselli & Myers (1995) find a higher density and pressure in mas- sive cores leading to a fast mass infall. Thus, it may be that the turbulence and the higher density of the ambient gas favour higher accretion rates in massive star forming molecular clouds (we are indebted to Prof. F. Palla for this very important remark). There are remarkable results concerning the presence of huge, likely bipolar, molecular outflows coming out of the regions of massive star formation. The luminosities of these regions are estimated from the radio free-free fluxes and/or from the integrated IR fluxes. The outflow rates come from the expansion velocities of the CO(1-0) line Mout behave con- (Churchwell, 1999). The outflow rates tinuously like L0.7 bol over 6 decades of luminosity (Shepherd and Churchwell, 1996). Around the solar luminosity, the Mout are about 10−5M⊙ yr−1, i.e. of the same or- values of der as the currently estimated accretion rates. However for massive stars with L from 104 to 106L⊙, the outflow rates Mout are in the range of 10−3 to 10−2M⊙ yr−1. There are several possible origins for these large outflows, however the review of the arguments by Churchwell (1999) favours the possibility that the massive outflows are driven by accretion, although it is not yet proved. From the large masses present in the outflows and the luminosity of the central object, he estimates that the fraction f of the infalling material incorporated into the star is about 15 %, while 85% are deflected in the outflows. Adopting a mass-luminosity relation of the form L ∼ M 3, Church- well (1999) suggests that the outflow rates behave like Mout ∼ M 2.1. If we specifically consider, for example, the mass-luminosity relation from the models by Schaller et al. (1992) in the broad mass interval of 2 to 85 M⊙, we would get Maccr = 1.5 10−5f (cid:18) M M⊙(cid:19)1.54 M⊙yr−1 , (2) where f is the accreted fraction of the infalling material. Such results suggests that constant accretion rates of the order of 10−5M⊙ yr−1 do not necessarily apply for all ranges of stellar masses and that much larger values of Maccr may have to be considered for larger masses. If this is true, several of the arguments against the accretion sce- nario for massive stars may not apply. We also note that there is a class of theoretical mod- els of cloud equilibria which do in fact predict a strong dependence of the mass accretion rate on protostellar mass. These are the so-called logatropic spheres studied by McLaughlin & Pudritz (1996,1997) and Galli et al. (1999), where internal pressure varies like the logarithm of the density. The equation of state departs from isother- mal, and the sound speed increases with decreasing den- sity. These models are useful to account for the line width density relation observed for molecular clouds. The accre- tion rate onto a protostar is not constant in logatropic Maccr varies models, but grows like t3, which implies that 4 (McLaughlin & Pudritz 1997). These models have like M also been applied to study the collapse of hot molecular cores leading to the formation of massive stars (Osorio et al. 1999). 3 In view of all the above arguments in favour of possible large accretion rates, we do think it is worth to further ex- amine the accretion scenario for the formation of massive stars. P. Norberg & A. Maeder: Massive Star Formation 5 (3) 3. Simple birthlines with constant or slowly varying accretion In this Section, we briefly show for the purpose of compari- son some new sets of birthlines obtained with constant and slowly varying Maccr. The initial quasi-static model with a mass Mini = 0.7 M⊙ are fully convective (see Stahler et al. 1988; Bernasconi and Maeder, 1996) and are started before the deuterium-burning sequence, which is treated with a time dependent convection scheme. It is well known that the initial D-abundance and that of the accreting matter influence the PMS evolution; here the initial and cosmic D abundance is taken 5 · 10−5 in mass fraction con- sistently with the results by Geiss (1993). From many test models by Bernasconi and Maeder (1996), it appears that the exact value chosen for the initial model has no signifi- cant influence for the PMS structure and on the evolution of intermediate and massive stars. Only the age is possi- bly influenced by this choice (see comments made about Tables 1, 2 and 3 below). 3.1. Constant values of Maccr Our simplest accretion model considers a constant mass accretion rate. In a recent study of the Orion Nebula, Palla Maccr equal and Stahler (1999) have used constant values to 10−5M⊙ yr−1. In Fig.1, the birthlines corresponding to Maccr to 10−6, 10−5 and 10−4M⊙ yr−1 are shown. Higher accretion rates lead to birthlines with higher luminosities. For high accretion rates, a star gains more mass and thus luminosity during its PMS contraction and reaches the ZAMS at a higher luminosity. Deuterium is also contin- uously brought to the star and contributes to the stellar luminosity (cf. Bernasconi and Maeder, 1996). Tout et al. (1999) have recently shown that the PMS tracks and age estimates of PMS stars are very much influenced by the accretion rates and this is indeed in agreement with the present results. 3.2. Slowly varying Maccr The models by Bernasconi and Maeder (1996) did not Maccr were use constant accretion rates, but the values of slightly increasing with the already accreted mass, accord- ing to the prescriptions of a simple model of collapsing clouds. This model needs several input parameters such as the temperature T of the cloud (currently 30 K), the mean molecular weight µ (currently 2.4). The effect of tur- bulent pressure were accounted for according to the veloc- ity dispersion vs. size of the clouds given by Larson (1981). In Fig. 1, a model with these prescriptions is also shown (model with F=1.0). We also show two other models with rates 10 times smaller and 10 times larger than the rates MBM by Bernasconi and Maeder (1996), according to the expression Maccr = F MBM Fig. 1 indicates clearly the differences between the two sets of models. These differences are small, especially for the low accretion rates. The model with F=0.1 reaches the ZAMS at about 4.4 M⊙, the one with F=1.0 at 9.5 M⊙, and at 27.5 M⊙ for F = 10.0. After having reached the ZAMS, the continuously accreting protostar carries on its evolution along the ZAMS. On the upper main sequence, the tracks are also in- fluenced by the accretion rates. For low Maccr, the time necessary to build up a massive star is so long that the star begins to burn a large fraction of its central hydrogen. Thus it has already moved away from the ZAMS, when it becomes visible at the end of its very long accretion phase. For high accretion rates, the time necessary to form a mas- sive star is so short that a small amount of hydrogen is burnt and therefore the star becomes visible close to the ZAMS. Theses features are well shown in Fig. 1. In Fig. 2, we compare 12 birthlines made with F val- ues between 0.1 and 5.0 with recent observations of PMS stars in various clusters (the references are given in the figure caption). Indications are given for stars which have observed values of the accretion rates. Let us note with caution that there are considerable uncertainties in the derivations of luminosities and Teff , as shown by the error bars. The birthlines have to be seen as upper envelopes, since a fraction of the stars may already have ended their accretion phase and are moving towards the ZAMS along the canonical PMS tracks. We do not know whether one single birthline should apply to various star groups; how- ever let us for simplicity adopt such a view. From Fig.2, we can make the following remarks. –1. A birthline with F between 1 and 5 fits best as an upper envelope, this corresponding to Maccr in the range of 10−5 and 10−4M⊙ yr−1. Thus, it is very likely that accretion rates higher than the currently used values of 10−5M⊙ yr−1 are necessary, at least for the most luminous PMS stars. –2. The upper part of the tracks at the time the star becomes visible also depends on the previous accre- tion rates (Figs. 1 and 3). There is some support towards high Maccr from the observations by Hanson et al. (1997). They find some very massive PMS objects (M ≥ 60 M⊙) close to the ZAMS and as seen above this also constrains the accretion rates. Thus these stars have to be formed in a time of the order of 106 yr at most. This implies an average accretion rate of about 10−4 M⊙ yr−1. Thus, a most critical difficulty for all these birthlines with slowly increasing Maccr is their too long lifetimes (cf. also Hanson et al. 1997). –3. The observations suggest that the birth- line should join the ZAMS between about 9 and 15 M⊙. –4. The distribution of stars in the HR diagram do not tightly constrain the slope of the birthline. However, we note that the observed distribution of stars in Fig. 2 seems 6 P. Norberg & A. Maeder: Massive Star Formation Table 1. Birthline properties for accretion rate given by expr.4 with ϕ = 0.5 and Mref = 10−5 M⊙ yr−1. The columns are in general self explicit; however, Mcore gives the convective core extension (in mass fraction), Tb is the temperature at the base of the convective envelope, when present; the surface content 2Hsurf in deuterium and the central hydrogen content 1Hcent are both given in mass fraction. NB Age yr Mass M⊙ Log L Log Te L⊙ K dM/dt M⊙ yr−1 Lgrav Mcore L⊙ M/M⊙ Log Tb K 2Hsurf 1Hcent 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0.000E+00 3.271E+05 1.156E+06 1.791E+06 2.327E+06 2.799E+06 3.230E+06 3.628E+06 3.993E+06 4.341E+06 4.656E+06 5.301E+06 5.828E+06 6.355E+06 6.834E+06 7.313E+06 8.354E+06 9.305E+06 1.018E+07 1.099E+07 1.175E+07 1.248E+07 1.317E+07 0.700 1.000 2.001 3.000 4.000 5.000 6.011 7.024 8.022 9.037 10.008 12.147 14.047 16.085 18.058 20.146 25.080 30.062 35.073 40.028 45.018 50.009 55.012 1.241 0.838 0.648 1.767 2.456 2.738 3.077 3.336 3.489 3.646 3.790 4.055 4.247 4.425 4.572 4.708 4.974 5.189 5.375 5.530 5.670 5.798 5.923 3.609 3.649 3.705 3.857 4.176 4.241 4.294 4.334 4.362 4.388 4.411 4.450 4.478 4.502 4.522 4.539 4.568 4.585 4.595 4.590 4.569 4.512 4.395 0.836E-06 0.999E-06 0.141E-05 0.173E-05 0.200E-05 0.223E-05 0.244E-05 0.264E-05 0.282E-05 0.299E-05 0.315E-05 0.346E-05 0.372E-05 0.398E-05 0.422E-05 0.446E-05 0.499E-05 0.547E-05 0.591E-05 0.632E-05 0.670E-05 0.706E-05 0.741E-05 17.36 4.70 1.22 55.02 133.86 -116.69 -145.66 -155.52 -171.74 -199.69 -219.64 -263.28 -301.07 -366.45 -426.49 -486.96 -646.14 -915.03 -1234.56 -1702.94 -2319.24 -3345.30 -2531.52 1.000 1.000 0.000 0.000 0.132 0.267 0.249 0.256 0.291 0.319 0.335 0.331 0.356 0.385 0.411 0.470 0.465 0.486 0.513 0.530 0.503 0.467 0.417 5.832 6.250 6.474 5.00E-05 5.68E-08 4.87E-10 4.20E-05 1.10E-05 1.14E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 6.800E-01 6.800E-01 6.800E-01 6.800E-01 6.793E-01 6.780E-01 6.755E-01 6.730E-01 6.706E-01 6.677E-01 6.640E-01 6.538E-01 6.426E-01 6.315E-01 6.189E-01 6.024E-01 5.620E-01 5.052E-01 4.564E-01 3.818E-01 2.939E-01 1.699E-01 2.512E-02 to suggest a slope, which may be steeper than the one pre- dicted by Bernasconi and Maeder (1996). Especially the birthline could be lower at lower luminosities. The various difficulties met with these accretion mod- els as a formation mechanism of massive stars lead us to examine in the next Section models where the accre- tion rate Maccr increases more rapidly with mass. Indeed this will give rise to birthlines with steeper slopes in the HR diagram and at the same time shorten the formation timescales. 4. Models with strongly increasing accretion rates The results of the two previous Sections suggest that the accretion rates may increase relatively quickly with the stellar mass. To further test this hypothesis, we examine the consequences of a power law with exponent ϕ Maccr(M ) = Mref (cid:18) M M⊙(cid:19)ϕ (4) We consider that the accretion rate is increasing with an exponent ϕ of the mass. Models with ϕ = 0.5, 1.0 and 1.5 are calculated. We provide in the Tables 1, 2 and 3 some important data for these models, assuming a ref- Mref of 10−5 M⊙ yr−1. We shall erence mass loss rate also test this value below. The models are started with an initial mass Mini = 0.7 M⊙ and the zero of the age scales are placed at this time. Another possible choice (Bernasconi, 1996) would be to take as initial age 0.7M⊙ , < Maccr > where < Maccr > represents the average accretion rate since the start of the formation process. As a matter of fact, we do not strictly know the zero point in the age scale, and therefore any convention is slightly arbitrary. Fig. 3 shows 3 birthlines obtained with expression 4. We notice that the tracks by Bernasconi and Maeder (1996; corresponding to F = 1.0) are rather close to those with ϕ = 0.5. For higher values of ϕ, we obtain, as ex- pected, a much steeper slope of the birthlines. As shown in the Tables 1, 2 and 3, the models with a higher ϕ have much shorter formation times. In models of high ϕ the cen- tral H-content is still much higher, and the size of their convective cores are larger. Both points are consistent with the fact that the ages are shorter and the evolution much less advanced. For ϕ = 0.5, and even more for the con- Maccr, the PMS lifetimes tPMS are longer than the stant Main Sequence (MS) lifetimes tMS for massive stars above about 20 M⊙. In this case, tMS = 8.1 106 yr, and tPMS = 7.3 106yr. This is a very severe problem for slowly in- creasing Maccr and a very important argument in favour of models with an accretion rate quickly increasing with stellar mass. For ϕ = 1.0, the equality of the two lifetimes tPMS and tMS is realized around a mass of about 45 M⊙; however a large fraction of tPMS is still spent in the low mass regime. For ϕ = 1.5, the situation with respect to the lifetimes is much more satisfactory, as tPMS is shorter than tMS up to at least 120 M⊙. Moreover, half of tPMS is spent be- low a mass of 2 M⊙, and the time for the star to evolve from 2 to 120 M⊙ is about 106 yr, which is an accept- P. Norberg & A. Maeder: Massive Star Formation 7 Table 2. Same as in Table 1, but for ϕ = 1.0 . NB Age Mass Log L Log Te dM/dt Lgrav Mcore Log Tb 2Hsurf 1Hcent yr M⊙ L⊙ K M⊙ yr−1 L⊙ M/M⊙ K 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 0.000E+00 3.575E+05 1.050E+06 1.456E+06 1.744E+06 1.967E+06 2.149E+06 2.308E+06 2.445E+06 2.561E+06 2.667E+06 2.851E+06 3.005E+06 3.141E+06 3.257E+06 3.363E+06 3.586E+06 3.769E+06 3.924E+06 4.059E+06 4.175E+06 4.278E+06 4.376E+06 4.466E+06 4.622E+06 4.753E+06 4.977E+06 0.700 1.000 2.001 3.000 4.002 5.000 6.000 7.031 8.059 9.046 10.056 12.073 14.082 16.113 18.085 20.104 25.084 30.116 35.128 40.193 45.113 49.946 55.062 60.217 70.340 80.182 100.23 1.241 0.793 0.666 1.558 2.776 2.661 3.086 3.280 3.515 3.680 3.800 4.034 4.236 4.405 4.546 4.672 4.921 5.116 5.272 5.401 5.509 5.600 5.682 5.758 5.887 5.991 6.160 3.609 3.649 3.704 3.802 4.197 4.213 4.305 4.331 4.366 4.393 4.415 4.452 4.483 4.508 4.528 4.546 4.579 4.603 4.621 4.635 4.645 4.654 4.660 4.665 4.671 4.683 4.689 0.700E-06 0.999E-06 0.199E-05 0.299E-05 0.399E-05 0.499E-05 0.599E-05 0.699E-05 0.798E-05 0.896E-05 0.996E-05 0.119E-04 0.139E-04 0.159E-04 0.179E-04 0.199E-04 0.248E-04 0.298E-04 0.347E-04 0.398E-04 0.446E-04 0.491E-04 0.546E-04 0.597E-04 0.698E-04 0.796E-04 0.996E-04 17.36 4.03 0.05 29.24 -33.63 89.33 -374.60 -425.53 -516.27 -564.90 -667.72 -880.94 -1084.87 -1159.10 -1623.84 -1725.76 -2599.08 -3324.35 -4241.38 -5319.48 -6289.63 -6833.45 -8946.76 -10319.23 -6300.715 -6455.416 -23297.530 1.000 1.000 0.000 0.000 0.197 0.211 0.293 0.276 0.266 0.278 0.313 0.359 0.365 0.412 0.439 0.466 0.503 0.566 0.606 0.640 0.669 0.656 0.715 0.686 0.718 0.739 0.768 5.832 6.274 6.427 5.00E-05 2.80E-08 1.24E-09 4.79E-05 1.04E-05 1.05E-05 1.08E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 6.800E-01 6.800E-01 6.800E-01 6.800E-01 6.799E-01 6.793E-01 6.783E-01 6.770E-01 6.756E-01 6.744E-01 6.735E-01 6.712E-01 6.682E-01 6.660E-01 6.635E-01 6.606E-01 6.545E-01 6.469E-01 6.408E-01 6.351E-01 6.293E-01 6.262E-01 6.194E-01 6.144E-01 6.097E-01 6.012E-01 5.852E-01 able value. Due to this we must stress that accretion rates strongly increasing with mass become a necessity in order to form stars on a timescale significantly smaller than the MS lifetime, so that they have not evolved too far from the ZAMS when they become visible. Also, a too long lifetime will allow a massive star to ionize a too large surrounding region, preventing any further accretion. In this respect, we emphasize that the models with ϕ ≃ 1.5 are much more favourable than the other ones studied here. These calculations raise the question whether, at a given value of the stellar mass, the initial formation of a massive star occurs with the same accretion rate as for a star with a low final mass. This is what is assumed here. This looks reasonable, since according to Newton's Theo- rem the gravitational potential in a spherical configuration is determined by the matter within the considered radius. However, further studies may show the importance of en- vironmental effects, such as the local density and temper- ature in a cluster, which are not taken into account in the present study. In order to further examine the accretion rates and their dependence on ϕ and Mref in expr. 4, some other models have been calculated. Fig. 4 shows models for ϕ = 1.0 and Mref = 10−6, 2 · 10−6, 5 · 10−6 and 10−5 M⊙ yr−1 respectively. These birthlines are compared to the observations already shown in Fig. 2. Clearly the highest value for Mref gives the best fit as an upper envelope. This confirms that in the range of intermediate mass stars the accretion rates are not as low as 10−5 M⊙ yr−1. As already noticed for Fig. 2, it is difficult to obtain constraints on the slope ϕ from the point distributions in the HR diagram. However, the agreement for the highest curve in Fig. 4 is rather better than for Fig.2. Also, it is possible that the theoretical slope is not steep enough in view of the observations. From the envelopes in the HR diagram, the Mref are much better determined than the value values of of ϕ. Fig. 5 shows the same kind of results, but for an ex- ponent ϕ = 1.5 and Mref = 10−6, 5 · 10−6 and 10−5 M⊙ yr−1 respectively. Clearly the two upper curves give the best fit, and maybe the highest one is the best. From the envelope fits in the HR diagram, it is hard to say whether a value ϕ = 1.0 or 1.5 is best, however from the consid- erations on the lifetimes, the case with ϕ = 1.5 is clearly favoured. Therefore, our preferred choice of parameters for the model given by expr. 4 is an exponent ϕ ≃ 1.5 and a multiplying factor Mref ≃ 10−5 M⊙ yr−1. We notice that it is amazing how this slope and the multiplying factor are close to the results obtained by Churchwell (1999) and Henning et al. (2000) and in par- ticular to the values given in expr. 2. Without saying that the physical behaviour of the accretion rates is determined by a law which is exactly of the form given by expr. 4, it is interesting that these two very different approaches give a rather similar dependence with respect to the stellar mass. Further theoretical and observational analyses are needed to give more insight into the dependence of the accretion rate on various possible parameters. 8 P. Norberg & A. Maeder: Massive Star Formation Table 3. Same as in Table 1, but for ϕ = 1.5 . NB Age yr Mass M⊙ Log L Log Te L⊙ K dM/dt M⊙ yr−1 Lgrav Mcore L⊙ M/M⊙ Log Tb K 2Hsurf 1Hcent 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 0.000E+00 3.907E+05 9.767E+05 1.236E+06 1.391E+06 1.496E+06 1.574E+06 1.635E+06 1.684E+06 1.725E+06 1.760E+06 1.818E+06 1.857E+06 1.896E+06 1.925E+06 1.949E+06 1.997E+06 2.031E+06 2.060E+06 2.079E+06 2.099E+06 2.113E+06 2.128E+06 2.142E+06 2.161E+06 2.176E+06 2.200E+06 0.700 1.000 2.000 3.000 4.002 5.000 6.000 7.001 8.004 9.038 10.053 12.164 13.961 16.185 18.220 20.222 25.350 30.187 35.516 39.887 45.110 49.724 55.077 61.335 71.464 80.783 100.86 1.241 0.745 0.721 1.120 2.548 3.018 2.944 3.297 3.560 3.774 3.829 4.065 4.220 4.395 4.538 4.662 4.915 5.099 5.261 5.372 5.485 5.571 5.658 5.746 5.869 5.964 6.128 3.609 3.649 3.708 3.766 4.067 4.268 4.266 4.342 4.369 4.402 4.419 4.457 4.484 4.511 4.532 4.550 4.585 4.609 4.629 4.642 4.655 4.664 4.673 4.681 4.691 4.699 4.710 17.36 0.585E-06 3.42 0.999E-06 -1.22 0.282E-05 7.92 0.519E-05 340.19 0.799E-05 -476.96 0.111E-04 -1851.30 0.146E-04 -1275.61 0.185E-04 -1535.61 0.226E-04 -1468.14 0.269E-04 -2105.96 0.311E-04 -3042.56 0.413E-04 -3639.83 0.508E-04 -5075.49 0.632E-04 -6693.96 0.754E-04 -8078.64 0.880E-04 -12403.20 0.123E-03 -17836.61 0.159E-03 -24182.66 0.203E-03 -30703.26 0.241E-03 -39344.84 0.289E-03 -47801.27 0.333E-03 -58971.57 0.388E-03 -72931.53 0.454E-03 0.569E-03 -98576.05 0.682E-03 < -1.0E+05 0.945E-03 < -1.0E+05 1.000 1.000 0.000 0.000 0.000 0.350 0.454 0.340 0.263 0.255 0.288 0.335 0.359 0.423 0.460 0.484 0.535 0.577 0.616 0.644 0.672 0.697 0.718 0.737 0.767 0.796 0.833 5.832 6.297 6.129 6.065 5.00E-05 1.46E-08 5.89E-06 1.68E-05 2.11E-05 1.13E-05 1.12E-05 1.17E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 5.00E-05 6.800E-01 6.800E-01 6.800E-01 6.800E-01 6.800E-01 6.797E-01 6.794E-01 6.790E-01 6.784E-01 6.778E-01 6.773E-01 6.766E-01 6.760E-01 6.757E-01 6.751E-01 6.746E-01 6.732E-01 6.721E-01 6.711E-01 6.704E-01 6.697E-01 6.692E-01 6.685E-01 6.679E-01 6.670E-01 6.664E-01 6.653E-01 5. Conclusions and remarks on the maximum stellar mass The main result of this work is that the accretion rate of a forming protostar strongly depends on the protostar's mass. A birthline described by a power law (5) M⊙(cid:19)ϕ Maccr = Mref (cid:18) M with Mref ≃ 10−5 and ϕ = 1.5 gives the best upper en- velope for PMS stars in the HR diagram and, more im- portantly, it satisfies also the constraints coming from the formation lifetimes. The above law is also quite consistent with radio and IR studies of protostellar outflows (Church- well, 1999; Henning et al., 2000), which show Maccr quickly increasing with the luminosity of the UC HII region. It is interesting that the high values of Maccr suggested here well correspond to the permitted domain of accretion rates found by Wolfire and Cassinelli (1987). The limits of this permitted domain are defined, on the low side of Maccr, by the condition that the momen- the values of tum in the accretion flow is larger than the outwards ra- diation momentum. On the high side, it is fixed by the condition that the shock luminosity due to the accretion process is smaller than the Eddington luminosity. In ad- dition, Wolfire and Cassinelli (1987) also found, for the occurrence of inflows onto massive stars, that the dust abundance has to be reduced by at least a factor of 4 and that the larger graphite grains are absent from the dust distribution function. In the context of the star models presented here, which do not follow the properties of the surrounding interstellar matter, we do not know whether this additional condition is met. If this accretion scenario for forming massive stars proves to be the correct one, it has several further impli- cations: on the luminosities and Teff of the progenitors of massive stars; on the lifetimes of PMS evolution; on the initial stellar structure on the ZAMS; on surface abun- dances of light elements; etc... It has also an impact on the slope of the initial mass function (IMF), since the final mass spectrum is not only shaped by the size of the col- lapsing fragments, which determine the reservoir of mat- ter potentially available, but also by the accretion process which leads the star to reach its final mass. Moreover the maximum stellar mass is determined by the physics intervening in the accretion process. In par- ticular, the maximum stellar mass is the mass for which the accretion rate Maccr is such that the shock luminosity Lshock due to the accretion plus the protostellar luminosity L∗ is equal to the Eddington luminosity LEdd, i.e. Lshock + L∗ = LEdd. If Rshock is the radius where the shock occurs, one has: G MaccrM Rshock + L∗ = 4πcGM κdust . (6) The relevant opacity to be considered here is the dust opacity κdust near the inner face of the dust cavity, since the grain opacity is the largest opacity source which may prevent further material accretion. As shown by Pollack P. Norberg & A. Maeder: Massive Star Formation 9 6 4 2 4.5 4 Fig. 4. The continuous lines represent birthlines obtained with ϕ = 1.0 and Mref being equal to 10−6, 2 10−6, 5 · Mref is the higher the birthline is in the HR 10−6 and 10−5 M⊙ yr−1 from bottom to top respectively. The higher diagram. The other curves have the same meaning as in Fig. 2 and the observations are the same. et al. (1994), the main opacity source are organics below the vaporization temperature (T ≃ 500 K) and silicates and metallic iron at higher temperatures. The opacity to be considered is likely the Planck opacity, i.e. the flux weighted opacity appropriate for small optical depths. The typical values of the Planck opacity κdust range between 2 and 8 cm2/g. Now, for the high accretion rates considered in the upper mass range, the shock luminosity dominates over the stellar luminosity by about 3 to 4 orders of magni- tudes so that we may ignore L∗. Thus, we obtain a simple expression for the limiting accretion rate Maccr = 4πc Rshock κ . (7) If we consider for simplicity that the shock radius is some multiple α of the stellar radius, we can apply the mass- radius relation obtained from the models of Schaller et al. (1992). The relation, valid between 40 and 120 M⊙, is thus: Rshock R⊙ = α (cid:18) M M⊙(cid:19)0.557 (8) Now, we search for the intersection of the expression (5) Maccr(M ) with relation (7) also accounting for (8). for This gives the maximum possible stellar mass Mshock M⊙ = (cid:18) 41.6 α κdust (cid:19)1.06 (9) with κdust expressed in cm2/g. The main interest is to emphasize that in the present context the maximum mass Maccr(M ) with the the is defined by the intersection of accretion rate (7) giving a shock luminosity equal to the 10 P. Norberg & A. Maeder: Massive Star Formation 6 4 2 4.5 4 Fig. 5. The continuous lines represent birthlines obtained with ϕ = 1.5 and Mref being equal to 10−6, 5 · 10−6 and 10−5 M⊙ yr−1 respectively. Same remarks as for Fig.4 appropriate Eddington luminosity. More detailed models of the stellar surroundings are of course necessary. The above simple derivation may at most give an order of mag- nitude, if we know both the location of Rshock given by α and κdust. While the approximate range of values of κdust is known as seen above, the value of α is uncertain, since it depends on the adopted structure and turbulence of the clouds (cf. Pollack et al., 1994). As an example with α ≃ 10, we would have a maximum mass between 70 and 300 M⊙. The accretion scenario thus leads to a new simple con- cept for the maximum stellar mass. A change of metallic- ity will influence both the opacity and the location of the shock radius, thus the resulting effect of metallicity cannot be estimated without detailed models of collapsing clouds. manuscript. They also express their thanks to Dr. Georges Meynet for his advice and helpful discussions. References van den Ancker, M.E., de Winter, D., Tjin A Djie, H.R.E., 1998, A&A 330, 145 van den Ancker, M.E., Th´e, P.S., Tjin A Djie, H.R.E., et al., 1997a, A&A 324, L33 van den Ancker, M.E., Th´e, P.S., Feinstein, A., et al., 1997b, A&AS 123, 63 Beech, M, Mitalas, R., 1994, ApJS 95, 517 Bernasconi, P.A., 1996, A&AS 120, 57 Bernasconi, P.A., Maeder, A., 1996, A&A 307, 829 Berrilli, F., Corcuilo, G., Ingrosso, G., et al., 1992, ApJ 398, 254 Bonnell, I.A., Bate, M.R., Zinnecker, H., 1998, MNRAS 298, Acknowledgements. The authors want to express their grat- itude to Prof. F. Palla for his most useful remarks on the 93 Caselli, P., Myers, P.C., 1995, ApJ 446, 665 P. Norberg & A. Maeder: Massive Star Formation 11 Churchwell, E., 1999, "Massive Star Formation: The Role of Bipolar Outflows", in UN solved Problems in Stellar Evo- lution, Proc. STScI Meeting, Ed. M. Livio, in press Cohen, M., Kuhi, L.V., 1979, ApJS 41, 743 Damiani, F., Micela, G., Sciortino, S., et al., 1994, ApJ 436, 807 Galli, D., Lizano, S., Li, Z.Y., et al., 1999, ApJ 521, 630 Geiss, J., 1993, In Origin and Evolution of the Elements, Eds. Prantzos, N., Vangioni-Flam, E., Cass, M., Cambridge Univ. Press, p. 89 Hanson, M.M., Howarth, I.D., Conti, P.S., 1997, ApJ 489, 698 Henning, Th., Schreyer, K., Launhardt, R., et al., 2000, A&A 353, 211 Hillenbrand, L.A., 1992, ApJ 397, 613 Hofner, P., Churchwell, E. 1997, ApJL 486, L39 Larson, R.B., 1981, MNRAS 145, 271 McLaughlin, D.E., Pudritz, R.E., 1996, ApJ 469, 194 McLaughlin, D.E., Pudritz, R.E., 1997, ApJ 476, 750 Osorio, M., Lizano, S., D'Alessio, P., 1999, ApJ 525, 808 Palla, F., Stahler, S.W., 1993, ApJ 418, 414 Palla, F., Stahler, S.W., 1999, ApJ 525, 772 Pollack J.B., Hollenbach D., Beckwith S., et al., 1994, ApJ 421, 615 Schaller, G., Schaerer, D., Meynet, G., et al., 1992, A&AS 96, 269 Shepherd, D.S., Churchwell, E., 1996, ApJ 472, 225 Stahler, S.W. 1988, ApJ 332, 804 Stahler, S.W., Palla, F., Ho, P.T., 2000, in Protostars and Plan- ets IV, Eds. V. Mannings et al., Univ. of Arizona Press, Tucson. Stahler, S.W., Shu, F.H., Taam, R.E., 1981, ApJ 248, 727 Stahler, S.W., Shu, F.H., Taam, R.E., 1980a, ApJ 241, 637 Stahler, S.W., Shu, F.H., Taam, R.E., 1980b, ApJ 242, 226 Th´e, P.S., de Winter, D., Feinstein, A., et al. 1990, A&AS 82, 319 Tout, C.A., Livio M., Bonnell, I.A. 1999, MNRAS 310, 360 de Winter, D., Koulis, C., Th´e, P.S., et al., 1997, A&AS 121, 223 Wolfire, M.G., Cassinelli, J.P., 1987, ApJ 319, 850
0708.2505
1
0708
2007-08-18T20:05:36
Radio Emission Signatures in the Crab Pulsar
[ "astro-ph" ]
Our high time resolution observations of individual pulses from the Crab pulsar show that both the time and frequency signatures of the interpulse are distinctly different from those of the main pulse. Main pulses can occasionally be resolved into short-lived, relatively narrow-band nanoshots. We believe these nanoshots are produced by soliton collapse in strong plasma turbulence. Interpulses at centimeter wavelengths are very different. Their dynamic spectrum contains regular, microsecond-long emission bands. We have detected these bands, proportionately spaced in frequency, from 4.5 to 10.5 GHz. The bands cannot easily be explained by any current theory of pulsar radio emission; we speculate on possible new models.
astro-ph
astro-ph
RADIO EMISSION SIGNATURES IN THE CRAB PULSAR T. H. Hankins1 J. A. Eilek1 [email protected] [email protected] ABSTRACT Our high time resolution observations of individual pulses from the Crab pulsar show that both the time and frequency signatures of the interpulse are distinctly different from those of the main pulse. Main pulses can occasionally be resolved into short-lived, relatively narrow-band nanoshots. We believe these nanoshots are produced by soliton collapse in strong plasma turbulence. Inter- pulses at centimeter wavelengths are very different. Their dynamic spectrum con- tains regular, microsecond-long emission bands. We have detected these bands, proportionately spaced in frequency, from 4.5 to 10.5 GHz. The bands cannot easily be explained by any current theory of pulsar radio emission; we speculate on possible new models. Subject headings: pulsars: individual (Crab Nebula pulsar) - radiation mecha- nisms: non-thermal 1. INTRODUCTION What is the pulsar radio emission mechanism? Does the same mechanism always op- erate, in all stars or throughout the magnetosphere of one star? What are the physical conditions in the magnetosphere that allow the emission to happen? Despite forty years of effort, these questions still have not been answered conclusively. Most models of pulsar radio emission fall into three groups. These are (1) antenna-type emission from coherent charge bunches; (2) strong plasma turbulence (SPT), in which non- linear effects convert plasma waves to electromagnetic waves which can escape the plasma; 1Physics Department, New Mexico Tech, Socorro, NM 87801 – 2 – and (3) several variants of maser emission. In Hankins et al. (2003; "HKWE") we sug- gested these different emission mechanisms can be differentiated by their time signatures, because the characteristic variability timescales of each model differ. To test this idea, we designed and developed data acquisition systems to probe the radio emission signatures at the highest possible time resolutions. At low radio frequencies scattering by electron den- sity inhomogeneities in the Crab Nebula and the interstellar medium mask the highest time and frequency resolution structure of the pulsar emission. The observations we describe here were made at high enough frequencies to avoid the pulse broadening, due to multipath propagation through the interstellar medium, which occurs at lower frequencies. 1.1. The Crab Pulsar We have focused on the Crab pulsar because its occasional very strong "giant" pulses are ideally suited to our data acquisition systems. The mean profile of this star is dominated by a Main Pulse (MP) and an Interpulse (IP), as shown in Figure 1. Although the rela- tive amplitudes and detailed profiles of these features change with frequency, they can be identified from low radio frequencies ( < ∼ 300 MHz) up to the optical and hard X-ray bands. The similarity of the mean profile across this broad frequency range suggests that the radio emission and high-energy emission arise from the same regions of the magnetosphere in this star. Several geometrical models have been proposed for the origin of the MP and IP emission in pulsars. Traditional radio-pulsar models ascribe MP/IP pairs to low-altitude emission (a few to a few tens of stellar radii) from highly relativistic outflows above the star's two magnetic poles. Some models of high-energy pulsed emission also locate the emission regions at low altitudes (e.g., Daugherty & Harding 1996). If this is the case, the magnetic axis of the Crab pulsar must be nearly orthogonal to its rotation axis in order to see the highly beamed emission from both poles. Alternatively, some authors have suggested that the magnetic and rotation axes are nearly aligned, and the MP and IP emission comes from a wide emission cone (e.g., Manchester and Lyne 1977). Still other models relax constraints on the angle between the rotation and magnetic axes, and locate both radio and high-energy emission sites in the outer magnetosphere, possibly at the outer gap described by Cheng and Ruderman (1977). Yet another variant is the caustic model of Dyks et al. (2004), in which emission extends over a wide range of altitudes, from the star's surface nearly to the light cylinder. (The Dyks et al. model suggests IP emission comes from higher altitudes than MP emission, as does one of the models discussed by Hankins and Cordes 1981). We do not know which, if any, of these models are correct, but most of them suggest – 3 – physical conditions in the two emission regions should be similar. One would expect the same radio emission mechanism to be active in the IP and the MP. We were quite surprised, therefore, to find that the IP and MP have very different properties at high radio frequencies (5-10 GHz), as we report in this paper. 1.2. Observations and Post-processing In 2002 we captured strong, individual Crab pulses at the Arecibo Observatory1 at 2-ns time resolution, from 1.4− 5 GHz, as reported in HKWE. For our new observations, reported here, we went to higher frequencies, 6− 8.5 GHz and 8− 10.5 GHz, in order to obtain 2.5-GHz bandwidth and consequent 0.4-ns time resolution. For pulses that exceeded a preset threshold the received voltages from both polarizations were digitized with 8-bit resolution and stored for off-line coherent dedispersion. This allowed us to reach intrinsic time resolutions down to the limit imposed by the inverse of the receiver bandwidth, 0.4 ns. In §2 and §3 we show dedispersed individual pulses and their dynamic spectra2 recorded at 8-10.5 GHz for all of our figures; the results at 6-8.5 GHz are similar in all characteristics. After capturing a pulse our data acquisition system requires more than a pulse period to store the data and be reset to capture another pulse. Therefore we record only pulses which exceed a preset threshold, which we set high enough to trigger the data acquisition system only for the brightest individual pulses. The trigger detector bandwidth was typically 0.5 GHz, centered on the 2.5-GHz sampled bandwidth. The pulses we record coincide with the high-flux power law tail of the number-vs.-flux histogram for single pulses, as seen by Argyle and Gower (1972) and Lundgren et al. (1995) at lower frequencies; thus they might be loosely called "giant" pulses. However, it is not yet clear whether these high-flux pulses are physically similar to, or different from, the more common "weak" pulses; we are not aware of any compelling evidence for either case. In what follows we do not attempt to distinguish between the two, but just discuss MPs or IPs. At the high time resolution we achieve, details of the pulses are sensitive to the ex- act value of the dispersion measure (DM) used for the coherent dedispersion operation. We generally started with the DM value given by the Jodrell Bank Crab Pulsar Monthly 1The 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 Foundation. 2The dynamic spectrum is computed from the dedispersed voltage time series. It shows how the received pulse intensity is distributed in time and radio frequency. – 4 – Ephemeris3 for our observing epochs. However, we found evidence in the dynamic spectra that individual pulses could be more or less dispersed than the tabulated value, and that IPs are systematically more dispersed than MPs, as discussed in §3. We then attempted to find the "optimum" DM for most of the pulses we show in this paper. For a narrow pulse, such as most components of a MP, we used the DM value which maximized the peak intensity and the intensity variance, minimized the width, and aligned the arrival times of emission throughout our bandwidth. As shown in §3 and Figures 6, 7 and 8 the typical IPs are broader than the MPs; alignment of the dynamic spectra provided more reliable DM estimates for the IPs. We found it necessary to refine our optimum DM to a resolution of 10−5 pc-cm−3. In the remainder of this paper we present our observations of the main pulse in §2, then the interpulse and its narrow emission bands in §3. In §4 we discuss some possible causes of the interpulse emission bands and model limitations, and summarize our results in §5. 2. THE MAIN PULSE In our first high time resolution observations of the Crab pulsar (reported in HKWE), we concentrated on the MP, because it is brighter at low frequencies, and strong pulses are more common at the phase of the MP (Cordes et al. 2004). Our new observations at higher frequencies, 6-8.5 GHz and 8-10.5 GHz at Arecibo, confirm and extend our original results on the MP. 2.1. Microbursts and Nanoshots Most MPs consist of one to several "microbursts"; the brightest microburst in an MP can occur anywhere within the pulse average envelope. The microbursts can often be resolved into overlapping, short-lived "nanoshots". Figures 2 and 3 shows typical examples of MPs; other examples are shown in Sallmen et al. (1999) and Kern (2004). The dynamic spectrum of the microbursts is broadband, filling our entire observing bandwidth. The emission is sometimes, but not always, slightly weaker toward the high frequency edge of the receiver band (as illustrated in Figures 2 and 3). This is unlikely to be due to instrumental or interstellar effects. We normalized our system gain, as a function of frequency, by the Crab Nebula background which dominates the off-pulse system 3www.jb.man.ac.uk/∼pulsar/crab.html – 5 – temperature. The nebular spectrum is quite flat (proportional to ν−0.26 at these frequencies; Baars & Hartsuijker 1972). The correlation bandwidth due to interstellar scintillation (ISS) is predicted to be ∼ 1 − 2 GHz at 8-10 GHz, and scales approximately as ν4 (Cordes et al. 2004, Kern 2004). We might expect interstellar effects occasionally to be seen in the dynamic spectrum, but such effects should be equally likely to be seen at the lower edge of the passband, as at the high edge. We therefore conclude that the high-frequency fading sometimes seen in the dynamic spectrum of the MP is intrinsic to the MP emission mechanism. This is consistent with the known steep radio spectrum of the Crab pulsar (ν−3.1; Manchester, et al. 2005). While most MPs resemble the examples in Figures 2 and 3, occasionally the nanoshots are sufficiently sparse to be seen individually. Figure 4 shows one example. When this is the case, some of the nanoshots turn out to be relatively narrow-band. We believe these new data support our argument, in HKWE, that each MP microburst is a collection of short-lived nanoshots. When the time resolution is high enough, and the nanoshots are well separated in time, the individual shots can be resolved. We note that this picture is consistent with previous modeling of pulsar emission as amplitude-modulated noise, produced by the ensemble of a large number of randomly occurring nanoshot pulses, modulated by a more slowly varying amplitude function (Rickett 1975). 2.2. Our Interpretation: Strong Plasma Turbulence We argued in HKWE that the nanoshots represent the fundamental emission mechanism in MPs. In that paper we compared the nanoshots to predictions of the three competing theoretical models of the radio emission mechanism. We found that the short durations and narrow bandwidths of the nanoshots are consistent only with predictions of the SPT model. They are not consistent with predictions of scaling arguments describing emission from masers or from coherent charge bunches (both of which predict longer characteristic times). In particular, Weatherall (1998) modeled SPT and predicted narrow-band radiation, δν/ν ∼ 0.1−0.2, centered on the co-moving plasma frequency. He also predicted a distinctive time signature, arising from the coupling of the electromagnetic modes to the turbulence: νδt ∼ O(10). The relatively narrow-band spectra of the nanoshots revealed by our new observations match Weatherall's models well. We thus confirm our suggestion, in HKWE, that coherent radio emission in MPs is plasma emission produced by collapsing solitons in strong plasma turbulence. We note, however, that the SPT model makes no predictions on the spectrum of a collection of nanoshots. The spectral steepening we sometimes see in the MP dynamic – 6 – spectrum could be due either to fewer nanoshots within a high-frequency microburst, or to less energy released in a single high-frequency nanoshot, or both. This model does make one important prediction: plasma flow in the radio emission region is likely to be unsteady. The plasma flow will be smooth only if the local charge density is exactly the Goldreich-Julian (1969; "GJ") value, ρGJ ≃ ΩB/2πc (for a rotation rate Ω), so that the rotation-induced electric field is fully shielded. If the charge density differs from ρGJ, the emitting plasma feels an unshielded electric field, and feeds back on that field as its charge density fluctuates, leading to unsteady plasma flow (and consequently unsteady radio emission). For the Crab pulsar, making the usual assumption that its spindown is due to magnetic dipole radiation, we estimate a field B(r) ≃ 4×1012 G, and a GJ density nGJ(r) = ρGJ(r)/e ≃ 8 × 1012 cm−3, close to the star's surface. At the light cylinder, ∼ 160r∗, the field drops to ∼ 1 × 106 G and the GJ density to ∼ 2 × 106 cm−3. Current pair cascade models predict that the (neutral) pair density exceeds the primary beam (GJ) density by a factor λ ∼ 102 − 103 (Arendt & Eilek 2002). Now, if SPT is the emission mechanism, we can determine the plasma density directly, because SPT emission is centered on the comoving plasma frequency (νp ∝ √γbn, for bulk Lorentz factor γb). Thus, emission at frequency ν GHz cm−3. Noting that current models predict comes from plasma density nγb ≃ 1.2 × 1010ν2 γb ∼ 102 − 103 (e.g., Arendt & Eilek), and using λ to convert from plasma density to charge density, our SPT argument predicts that 1-GHz emission comes from a region with number density of excess charge ∼ 104 − 106 cm−3, and 9.5-GHz emission comes from ∼ 106 − 108 cm−3. The higher density values may be consistent with GJ conditions at moderate to high altitudes; the lower values are very unlikely to satisfy GJ conditions anywhere in the magnetosphere. We therefore suggest the highly unsteady, bursty MP emission we see from 1 to 10 GHz reflects unsteady plasma flow due to sub-GJ charge densities in the emission region (cf. also Kunzl et al. 1998, who drew the same conclusion from a somewhat different argument). 2.3. Extreme Nanoshots The nanoshots can occasionally be extremely intense. In Figure 5 we show a single MP which exceeds 2 MJy, and has an unresolved duration of less than 0.4 ns. If we ignore relativistic effects (following, e.g., Cordes et al. 2004), we estimate a light-travel size cδt ≃ 12 cm. From this we find an implied brightness temperature 2 × 1041K, which we believe is the highest ever reported for pulsar emission. Alternatively, we might assume the emitting structure is moving outwards with Lorentz factor γb ∼ 102 − 103. If this is the case, our size – 7 – estimate increases to 103−105 cm, and the brightness temperature decreases to 1035−1037K. The extremity of this pulse can also be demonstrated in terms of local quantities. If the pulse is emitted from a structure moving at γb, its energy density is urad ∼ 4 × 1023/γ4 b erg cm−3. This high energy density can be compared to the plasma energy density, upl = γbnmc2, which we can estimate either from our assumption of SPT emission (which gives a lower density), or by assuming that GJ conditions hold (which gives a higher density). In either case, we find urad ≫ upl, unless γb is extremely large. As noted in HKWE, this emphasizes the need for a collective emission process. For another comparison, we can convert the energy density in the radiation pulse to an equivalent electric field, E ∼ 3.2 × 1012/γ2 b G, giving a wave-strength parameter eE/2πmecν ≫ 1. It follows that magnetospheric propagation of such strong nanoshots will be complex and nonlinear (e.g., Chian & Kennel 1983). 3. THE INTERPULSE In order to test our hypothesis that strong plasma turbulence governs the emission physics in the Crab pulsar, we went to higher frequencies to get a larger bandwidth and shorter time resolution. In addition to the MP, we observed single pulses from the IP, because at high frequencies strong pulses are far more common at the rotation phase of the IP. When we observed IPs and MPs with a broad bandwidth, from 6-8 or 8-10.5 GHz, we were astonished to find that IPs have very different properties from MPs. 3.1. Characteristics We recorded about 220 individual MPs, and about 150 individual IPs, between 4.5 and 10.5 GHz, during 20 observing days from 2004 to 2006. Together with our earlier results (Moffett & Hankins 1999), these data reveal that the high-frequency IP differs from the MP in intensity, time signature, polarization, dispersion and spectrum. In this subsection we discuss the first four properties; we defer discussion of the spectrum to the next subsection. We illustrate our discussion with Figures 6, 7 and 8 which show three typical IPs. 3.1.1. Intensity Strong IPs are at least an order of magnitude more frequent than strong MPs at 9 GHz, but the MPs can be considerably stronger than the IPs when they occur. This can be seen in the examples shown in this paper, as well as in the the signal-to-noise-ratio histograms of – 8 – Figure 3 of Cordes et al. (2004) and the scatter plots of their Figure 5 (where the signal-to- noise-ratio of MPs and IPs are shown as a function of pulse phase). 3.1.2. Polarization High-frequency IPs are more strongly polarized than MPs. Moffett & Hankins (1999) showed that the IP is strongly linearly polarized, 50-100% at 4.9 GHz, while the MP is only weakly polarized. We showed in HKWE that individual nanoshots in the MP can be strongly polarized, but the polarization changes dramatically from one nanoshot to the next. This leads to to weak MP polarization when the nanoshot density is high or the pulse is smoothed to ∼ 1µs. 3.1.3. Time signature IP emission is not broken up into the short-lived microbursts that characterize the MP. Instead, it is more continuous in time, spread out over a few microseconds. When optimally dedispersed, IPs usually have a very rapid onset, followed by a slower decay and often similar secondary bursts. To quantify the time duration of the IPs, we used the equivalent width of the intensity autocorrelation function to estimate the IP duration. In Figure 9 we show the distribution of equivalent widths for all of the IPs we recorded at and above 6 GHz. This figure shows that IPs typically last several microseconds at 6-8 GHz, and become shorter at higher frequencies. Because the temporal behavior of MPs is much more complex (as we discussed in §2; cf. Figures 2 and 3), the question "what is the characteristic time signature of an individual MP" is difficult to answer. We discuss MP time scales in a forthcoming paper. 3.1.4. Dispersion IPs are more dispersed than MPs measured at the same time. As an example, the IP in Figure 6 was observed 12 minutes after the MP shown in Figure 2, and processed identically. The dynamic spectrum of the IP shows that lower frequencies arrive later than high frequen- cies; we take this as evidence for extra dispersion in the IP. Because we consistently found IPs more dispersed than MPs observed on the same day, we conclude this extra dispersion must occur in the pulsar's magnetosphere. – 9 – It is hard to compare the data to predictions of magnetospheric dispersion, because we do not know the correct dispersion relation for the magnetospheric plasma. As a simple example, consider dispersion from a cold, unmagnetized plasma. The 0.65-µs delay between 8.4 and 10.4 GHz, for the pulse shown in Figure 6, would correspond to an excess dispersion measure, ∼ 0.032 pc-cm−3 (∼ 10−3 of the total DM measured by Jodrell Bank). If the magnetosphere were filled with cold plasma at the GJ density, it would have a column density ∼ 1.3 pc-cm−3, far more than enough to account for the excess DM of the IP. But this estimate is naive. Most of the magnetosphere is strongly magnetized, and at low altitudes charges are constrained to move only along field lines. A more realistic dispersion law is needed, but without knowing conditions through which the pulse propagates it is not clear which law to choose. We discuss these complex issues more fully in a separate paper (Crossley et al. 2007). 3.2. Emission Bands The most striking difference between the interpulse and the main pulse is found in the dynamic spectrum. An IP contains microsecond-long trains of emission bands, as illustrated in Figures 6, 7 and 8. Every IP we have recorded displays these emission bands. However, MPs recorded during the same observing sessions and processed identically do not show the bands. The bands are, therefore, not due to instrumental or interstellar effects, but are intrinsic to the star. 3.2.1. Properties of the Emission Bands The emission bands are grouped into regular "sets"; 2 or 3 band sets, regularly spaced, can usually be identified in a given IP. Individual band sets last no more than a few mi- croseconds. All bands in a particular set appear almost simultaneously, certainly to within ∆tstart < 100 ns for optimally dedispersed pulses. This requires that all of the bands must originate from a region no more than d = c∆tstart < 30 m in size. The IPs show a sharp onset, which is often associated with a very short-lived (≤ 100 ns) band set. Additional band sets often turn on part way through the pulse, producing secondary bursts of total intensity which also show the characteristic fast-rise-slow-decay time signature. Band sets that begin later tend to last longer, up to the few-microsecond duration of the total pulse. At first glance the bands appear to be uniformly spaced. However, closer inspection of – 10 – our data shows that the bands are proportionally spaced. Figure 10 shows that the spacing between two adjacent bands depends on the mean frequency, as ∆ν/ν ≃ 0.06. Thus, two bands near 6 GHz are spaced by ∼ 360 MHz; two bands near 10 GHz are spaced by ∼ 600 MHz. This proportional spacing is robust; a set of emission bands can drift in frequency (usually upwards, as in Figures 6, 7 and 8), but their frequency spacing stays constant. We note that the least-squares fitted line in Figure 10 is consistent with zero spacing at zero frequency; however we do not think that the bands continue to very low frequencies (as we discuss in the next section). The frequency profile of a given band tends to be peaked about a central frequency, i.e. closer to Gaussian than to rectangular or impulsive. The frequency width of an emission band, estimated by eye as a half power width, is typically 10-20% of the spacing between bands. Within a single IP the center frequency of an emission band often remains steady until the band disappears. In some instances, however, the center frequency drifts upwards during the band duration, by no more than 20-30% of the band spacing. We have occasionally seen bands that appear to drift slightly downward in frequency, but this is rare. Bands sets that begin later in the pulse tend to start at frequencies slightly higher than the early bands, but againw the frequency shift is less than the spacing between bands. These features are all illustrated in Figures 6, 7 and 8. We note that quantitative analysis of the structure of a single band is limited by the frequency resolution we can achieve for short-lived bands, by signal-to-noise limits and the overlap of separate band sets. 3.2.2. Frequency Extent of the Bands We suspect the bands extend over at least a 5-6 GHz bandwidth in a single IP, but do not occur below ∼ 4 GHz. While we have not been able to observe more than 2.2 GHz simultaneously, we have seen no evidence that a given band set cuts off within our observable bandwidth. The characteristics of the bands (proportional spacing, duration, onset relative to total intensity microbursts) are unchanged from 5 to 10 GHz. We have captured a few IPs between 4 and 5 GHz, but the bands are unclear in all of them. We suspect the dynamic spectrum in this frequency range has been corrupted by ISS, for which the correlation bandwidth is predicted to be ∼ 100 MHz at these frequen- cies, somewhat less than the band spacing projected from Figure 10. Technical limitations, involving terrestrial radio interference and the high-speed memory capacity of our data ac- quisition system, kept us from observing with enough bandwidth to investigate the existence of bands below ∼ 4 GHz. We note, however, that mean profiles suggest the nature of the IP changes between 1 and 4 GHz. From Figure 1 (also Cordes et al. 2004) we see that the – 11 – IP between 4 and 8 GHz appears at an earlier phase than at ≤ 1.4 GHz, and there is no IP at all around 3 GHz. This leads us to believe that the low and high radio frequency IPs are unrelated; we therefore expect no IP band emission below 4 GHz. Unlike the MPs, there is no indication in our data that the band intensity is weaker at high frequencies. From this we infer that the IP spectrum is flatter than the MP spec- trum. The IP spectrum may, in fact, be atypical of radio emission from the general pulsar population, which is known to be steep spectrum. We reiterate that the emission bands in the IP cannot be due to ISS. Their clear reg- ularity, the fact that they exist only in the IP and not in MPs observed at the same time, and their ∆ν ∝ ν spacing, all disagree with known properties of ISS, and point to an origin intrinsic to the IP. 4. POSSIBLE CAUSES OF THE EMISSION BANDS The dynamic spectrum of the interpulses does not match any of the three types of emission models described in §1. Because each of the models predicts narrow-band emission at the plasma frequency, none of them can explain the dynamic spectrum of the IP. A new approach is required here, which may "push the envelope" of pulsar radio emission models. While we remain perplexed by the dramatic dynamic spectrum of the interpulse, we have explored possible models. This exercise is made particularly difficult by the fact that the emission bands are not regularly spaced. Because of this, models that initially seemed attractive must be rejected. As an example, if the emission bands were uniformly spaced they could be the spectral representation of a regular emission pulse train. Many authors have invoked regularly spaced plasma structures (sparks or filaments), whose passage across the line of sight could create such a pulse train. Alternatively, strong plasma waves with a characteristic frequency will also create a regular emission pulse train. The dynamic spectrum of either of these models would contain emission bands at constant spacing; the proportional spacing we observe disproves both of these hypotheses. We looked to solar physics for insight. We initially considered split bands in the dynamic spectra of Type II solar flares, which are thought to be plasma emission from low and high density regions associated with a shock propagating through the solar corona. This does not seem to be helpful for the Crab pulsar emission bands, because the radio-loud plasma would have to contain 10 or 15 different density stratifications, which seems unlikely. However, "zebra bands" seen in Type IV solar flares may be germane. These are parallel, drifting, narrow emission bands seen in the dynamic spectra of Type IV flares. Band sets – 12 – containing from a few up to ∼ 30 bands have been reported, with fractional spacing ∆ν/ν ∼ .01 − .03 (e.g., Chernov et al. 2005; Sawant et al. 2005). While zebra bands have not yet been satisfactorily explained, two classes of models have been proposed, namely resonant plasma emission and geometrical effects. It may be that these models will provide clues to understanding the emission bands in the Crab pulsar. 4.1. Resonant cyclotron emission One possibility is plasma emission at the cyclotron resonance, ω − kkvk − sΩ0/γ = 0 (where γ is the particle Lorentz factor, Ω0 = eB/mc, and s is the integer harmonic number). Kazbegi et al. (1991) proposed that this resonance operates at high altitudes in the pulsar magnetosphere, possibly generating both X-mode waves (which can escape the plasma directly) and O-mode waves; a related model is being applied to the Crab emission bands by Lyutikov (2007, private communication). Alternatively, double resonant cyclotron emission at the plasma resonant frequency has been proposed for zebra bands in solar flares (e.g., Winglee & Dulk 1986). In solar conditions, this resonance generates O-mode waves, which must mode convert in order to escape the plasma. The emission frequency in these models is determined by local conditions where the resonance is satisfied; the band separation is ∆ν ≃ Ω0/2πγ. It is not clear how the specific, proportional band spacing can be explained in such models; perhaps a local gradient in the magnetic field can be invoked. Such models face several additional challenges before they can be considered success- ful. The emission must occur at high altitudes, in order to bring the resonant (cyclotron) frequency down to the radio band. Close to the light cylinder, where B ∼ 106 G, particle energies γ ∼ 500 − 1000 could radiate at 5 - 10 GHz. In addition, such models must be developed by means of specific calculations which address the fundamental plasma modes and their stability, under conditions likely to exist at high altitudes in the pulsar's magne- tosphere. 4.2. Geometrical models Alternatively, the striking regularity of the bands calls to mind a special geometry. If some mechanism splits the emission beam coherently, so that it interferes with itself, the bands could be interference fringes. For instance, a downwards beam which reflects off a high density region could return and interfere with its upwards counterpart on the way back up (e.g., Ledenev et al. 2001 for solar zebra bands). Simple geometry suggests that fringes – 13 – occur if the two paths differ in length by only c/∆ν < ∼ 1 m. Another geometrical possibility is that cavities form in the plasma and trap some of the emitted radiation, imposing a discrete frequency structure on the escaping radiation (e.g., LaBelle et al. 2003). The scales required here are also small; the cavity scale must be some multiple of the wavelength, i.e., several centimeters. Geometrical models need an underlying broad-band radiation source, with at least 5- GHz bandwidth, in order to produce the emission bands we observe. Because standard pulsar radio emission mechanisms lead to relatively narrow-band radiation, at the local plasma frequency, they seem unlikely to work here. A double layer might be the radiation source; charges accelerated within the layer should radiate broadband, up to ν ∼ L/2πc, if L is the thickness of the double layer (e.g., Kuijpers 1990; Volwerk 1993). This is also a small-scale effect; emission at 10 GHz requires L ∼ 1 cm. This may be consistent with the thickness of a relativistic, lepton double layer, which we estimate as several times c/2πνp (following, e.g., Carlqvist 1982). Geometrical models also face several challenges before they can be considered success- ful. The basic geometry is a challenge: what long-lived plasma structures can lead to the necessary interference or wave trapping? Once again, it is not clear how the proportional band spacing can be explained; perhaps a variable index of refraction in the interference or trapping region can be invoked. 5. Summary and conclusions We have observed individual pulses from the Crab pulsar with 2.2-GHz bandwidth and 0.4-ns time resolution. We observed both the main pulse (MP) and the interpulse (IP), at high radio frequencies (5−10 GHz). We were very surprised when our observations revealed that the MP and the IP are strikingly different. At these frequencies, MPs consist of many short-lived, relatively narrow-band, nanoshots. Both the time and frequency signatures of the nanoshots are consistent with predictions of one current model of pulsar radio emission, namely, coherent emission from strong plasma turbulence. IPs, however, differ from MPs in their time, polarization, dispersion and spectral sig- natures. It seems that the MP and the IP differ in their emission mechanisms and their propagation within the magnetosphere. The dynamic spectrum of the IP contains regularly spaced emission bands, which do not match the predictions of any current model. Our result is especially surprising because magnetospheric models generally ascribe the MP and the IP to physically similar regions, which simply happen to be on opposite sides of the star. One – 14 – would therefore expect the MP and the IP to have similar characteristics, which is exactly not what we find. The work of Moffett & Hankins (1996) may provide an important clue. They discovered new components in the mean profile at 5 and 8 GHz, which appear at offset rotation phases, and only at high radio frequencies. They also found that the IP at high radio frequencies appears at an earlier rotation phase than its counterpart at both lower (radio below 2 GHz) and higher (optical to X-ray) frequencies. It is just this high-radio-frequency IP which we discuss in this paper. We speculate that this IP, and the other high-radio-frequency compo- nents Moffett & Hankins found, originate in an unexpected part of the star's magnetosphere, where different physical conditions produce quite different radiation signatures. We appreciate helpful conversations with Joe Borovsky, Alice Harding, Axel Jessner, Jan Kuijpers, Maxim Lyutikov, and the members of the Socorro pulsar group. We thank our referee for constructive comments which improved the paper. We particularly thank our students, Jared Crossley, Eric Plum and James Sheckard for their help with observations. This work was partially supported by the National Science Foundation, through grants AST0139641 and AST0607492. REFERENCES Arendt, P.N. Jr. & Eilek, J.A., 2002, ApJ, 581, 451 Argyle, E., & Gower, J.F.R. 1972, ApJ, 175, L89 Baars, J.W.M. & Hartsuijker, A.P., 1972, A&A, 17, 172 Carlqvist, P., 1982, ApSS, 87, 21 Cheng, A.F. & Ruderman, M.A., 1977, ApJ, 216, 865 Chian, A.C.-L.& Kennel, C.F., 1982, Ap&SS, 97, 9 Chernov, G.P., Yan, Y.H., Fu, Q.J. & Tah, Ch.M., 2005, A&A, 437, 1047 Cordes, J.M., Bhat, N.D.R., Hankins, T.H., McLaughlin, M.A. & Kern, J., 2004, ApJ, 612, 375 Crossley, J.M., Hankins, T.H. & Eilek, J.A., 2007, in preparation Daugherty, J.K. & Harding, A.K., 1996, ApJ, 300, 500 – 15 – Dyks, J., Harding, A.K. & Rudak, B., 2004, ApJ, 606, 1125 Goldreich, P. & Julian, W.H., 1969, ApJ, 157, 869 Goodman, J., & Narayan, R. 1985, MNRAS, 214, 519 Hankins, T.H., & Cordes, J.M., 1981, ApJ, 249, 241 Hankins, T.H., Kern, J.S., Weatherall, J.C. & Eilek, J.A., 2003, Nature, 422, 141 Isaacman, R., & Rankin, J. M. 1977, ApJ, 214, 214 Kazbegi, A.Z., Machabeli, G.Z. & Melikidze, G.I., 1991, MNRAS, 253, 377 Kern, J., 2004, Ph.D. thesis, New Mexico Tech Kuijpers, J., 1990, in Plasma Phenomena in the Solar Atmosphere, eds. M. Dubois, F. Bely- Dubau & D. Gresillon (Editions de Physique, France), p. 17 Kunzl, T., Lesch, H., Jessner, A. & von Hoensbroech, A., 1998, ApJ, 505, L139 LaBelle, J., Treumann, R.A., Yoon, P.H. & Karlicky, M., 2003, ApJ, 593, 1195 Ledenev, V.G., Karlick´y, M., Yan, & Fu, Q., 2001, Sol Phys., 202, 72 Lundgren, S.C., Cordes, J.M., Ulmer, M., Matz, S.M., Lomatch, S., Foster, R.S., & Hankins, T. 1995, ApJ, 453, 433 Manchester, R.N., Hobbs, G.B., Teoh, A., & Hobbs, M. 2005, AJ, 129, 1993 Moffett, D.A. & Hankins, T.H., 1996, ApJ, 468, 779 Manchester, R.N. & Lyne, A.G., 1977, MNRAS, 181, 761 Moffett, D.A. & Hankins, T.H., 1999, ApJ, 522, 1046 Rickett, B.J., 1975, ApJ, 197, 185 Sallmen, S., Backer, D.C., Hankins, T.H., Moffett, D., & Lundgren, S. 1999, ApJ, 517, 460 Sawant, H.S., Karlick´y, M., Fernandes, F.C.R. & Cecatto, J.R., 2002, AA, 396, 1015 Volwerk, 1993, J Phys D, 26, 1192 Weatherall, J.C., 1998, ApJ, 506, 341 Winglee, R.M. & Dulk, G.A., 1986, ApJ, 307, 808 – 16 – This preprint was prepared with the AAS LATEX macros v5.2. – 17 – Fig. 1.- The mean profile of the Crab pulsar, over a wide range of frequencies (from Moffett & Hankins 1996). The main pulse and interpulse, shown by dashed lines at pulse phases 70◦ and 215◦, persist from radio to hard X-ray bands. However, between 4.7 and 8.4 GHz the interpulse is offset from the interpulse at lower and higher frequencies, and new components appear (labeled HFC1 and HFC2). These intermediate-frequency components may have a different origin from the lower and higher frequency interpulse and main pulse. – 18 – Fig. 2.- An example of a "normal" main pulse, processed with optimal dispersion measure, plotted with total intensity time resolution 6.4 ns, and dynamic spectrum resolution 51.2 ns, 19.5 MHz. The pulse seen in total intensity (upper panel) consists of several short-lived microbursts, which themselves contain shorter-duration nanoshots. The dynamic spectrum (lower panel) reveals that the microbursts are broad-band, spanning the 2.2-GHz receiver bandwidth. The lack of emission at the lower band edge is because the 2.5-GHz sampled bandwidth is slightly larger than the receiver passband; the high-ν fading is intrinsic to the star. The spectrum contour levels are 50, 100, 200, 500, and 1000 Jy. – 19 – Fig. 3.- Another example of a "normal" main pulse, also processed with optimal disper- sion measure. Similarly to the example in Figure 2, this pulse contains several short-lived microbursts, each of which contains shorter-duration nanoshots. This pulse differs from that in Figure 2 in that its dynamic spectrum does not fade to high frequencies. Total intensity time resolution 6.4 ns; dynamic spectrum resolution 51.2 ns, 19.5 MHz. Spectrum contour levels are 20, 50, 100, and 200 Jy. – 20 – Fig. 4.- An example of a sparse main pulse at 9.25 GHz. The pulse is plotted with the same total intensity and dynamic spectrum resolution as the pulse in Figures 2 and 3. Occasionally nanobursts are sufficiently sparse that individual bursts with relatively narrow bandwidths can be identified, as seen here. The spectrum contour levels are 20, 50, 100, and 200 Jy. – 21 – Fig. 5.- A single main pulse recorded at 9.25-GHz center frequency over a 2.2-GHz band- width and optimally dedispersed. The nanopulse shown is unresolved with the 0.4-ns time resolution afforded by our system. Despite the high peak intensity of this pulse, it is unlikely that it saturated the data acquisition system. The dispersion sweep time across the band- width is about 1.5 ms, so as sampled by our data acquisition system, the dispersed pulse energy is spread over ≈ 7.5 × 106 samples. – 22 – Fig. 6.- A typical interpulse, observed 12 minutes after the main pulse shown in Figure 2, and processed identically. The later arrival of bands at lower frequency implies that this pulse is more dispersed than the main pulse in Figure 2. The spectrum contour levels are 20, 50, 100, 200, 500 Jy. Total intensity time resolution is 51.2 ns; dynamic spectrum resolution is 51.2 ns, 19.5 MHz. – 23 – Fig. 7.- Another interpulse, processed with the "optimum" dispersion measure. All three of the pulses shown in Figures 6, 7 and 8 show the emission bands we discovered in the dynamic spectrum of the interpulse. As these examples show, several band sets can be identified within an interpulse. The interpulse usually starts with a short-lived band set, and continues with longer-lived sets, which either start at slightly higher frequencies or drift upwards in frequency. The onset of later band sets often coincides with a second burst in total intensity. As in Figures 2 and 3, the apparent lack of low-ν mission is because the sampled bandwidth is slightly larger than the receiver bandwidth. Unlike the main pulse in Figure 2, however, the band intensity does not fade toward high frequencies. Here and in Figure 8, the spectrum contour levels are 10, 20, 50, 100, 200 Jy. Total intensity time resolution is 51.2 ns; dynamic spectrum resolution is 51.2 ns, 19.5 MHz. – 24 – Fig. 8.- Another example of an interpulse, also processed with "optimum" DM and the same time and spectral resolution as in Figures 6 and 7. – 25 – Fig. 9.- The distribution of time durations of our interpulses, as measured by the equivalent width of the autocorrelation function. Nearly all of the interpulses captured at frequencies between 6 and 8 GHz were recorded with a 1-GHz bandwidth; for these pulses we used the full bandwidth in computing the autocorrelation width. These pulses are labeled as 7 GHz in the figure. All of the interpulses between 8 and 10.5 GHz were recorded with a 2.5-GHz bandwidth. For these we divided the full band into high and low half-bands, each 1.25-GHz wide, centered at 8.625 and 9.875 GHz, and computed the autocorrelation width separately for each half-band. From the distribution of time durations it is clearly seen that the interpulse autocorrelation equivalent width is frequency dependent. The mean widths are 4.2, 3.0, and 2.7 µs for 7, 8.6 and 9.9 GHz, respectively. – 26 – Fig. 10.- The emission band spacing, measured for 460 band sets in 105 pulses recorded on 20 observing days, is shown as a function of center frequency. The line is fitted to all of the points, and has the form ∆ν = 0.058(±0.001)ν − 0.007(±0.011).
0806.0719
1
0806
2008-06-04T09:16:59
Misaligned spin-orbit in the XO-3 planetary system?
[ "astro-ph" ]
The transiting extrasolar planet XO-3b is remarkable, with a high mass and eccentric orbit. The unusual characteristics make it interesting to test whether its orbital plane is parallel to the equator of its host star, as it is observed for other transiting planets. We performed radial velocity measurements of XO-3 with the SOPHIE spectrograph at the 1.93-m telescope of Haute-Provence Observatory during a planetary transit, and at other orbital phases. This allowed us to observe the Rossiter-McLaughlin effect and, together with a new analysis of the transit light curve, to refine the parameters of the planet. The unusual shape of the radial velocity anomaly during the transit provides a hint for a nearly transverse Rossiter-McLaughlin effect. The sky-projected angle between the planetary orbital axis and the stellar rotation axis should be lambda = 70 +/- 15 degrees to be compatible with our observations. This suggests that some close-in planets might result from gravitational interaction between planets and/or stars rather than migration due to interaction with the accretion disk. This surprising result requires confirmation by additional observations, especially at lower airmass, to fully exclude the possibility that the signal is due to systematic effects.
astro-ph
astro-ph
Astronomy & Astrophysics manuscript no. article February 14, 2018 c(cid:13) ESO 2018 Misaligned spin-orbit in the XO-3 planetary system?(cid:63) G. H´ebrard1, F. Bouchy1, F. Pont2, B. Loeillet1,3, M. Rabus4, X. Bonfils5,6, C. Moutou3, I. Boisse1, X. Delfosse6, M. Desort6, A. Eggenberger6, D. Ehrenreich6, T. Forveille6, A.-M. Lagrange6, C. Lovis7, M. Mayor7, F. Pepe7, C. Perrier6, D. Queloz7, N. C. Santos5,7, D. S´egransan7, S. Udry7, A. Vidal-Madjar1 1 Institut d'Astrophysique de Paris, UMR7095 CNRS, Universit´e Pierre & Marie Curie, 98bis boulevard Arago, 75014 Paris, France 2 Physikalisches Institut, University of Bern, Sidlerstrasse 5, 3012 Bern, Switzerland 3 Laboratoire d'Astrophysique de Marseille, Universit´e de Provence, CNRS (UMR 6110), BP 8, 13376 Marseille Cedex 12, France 4 Instituto de Astrof´ısica de Canarias, La Laguna, Tenerife, Spain 5 Centro de Astrof´ısica, Universidade do Porto, Rua das Estrelas, 4150-762 Porto, Portugal 6 Laboratoire d'Astrophysique de Grenoble, CNRS (UMR 5571), Universit´e J. Fourier, BP53, 38041 Grenoble, France 7 Observatoire de Gen`eve, Universit´e de Gen`eve, 51 Chemin des Maillettes, 1290 Sauverny, Switzerland Received TBC; accepted TBC ABSTRACT The transiting extrasolar planet XO-3b is remarkable, with a high mass and eccentric orbit. These unusual characteristics make it interesting to test whether its orbital plane is parallel to the equator of its host star, as it is observed for other transiting planets. We performed radial velocity measurements of XO-3 with the SOPHIE spectrograph at the 1.93-m telescope of Haute-Provence Observatory during a planetary transit, and at other orbital phases. This allowed us to observe the Rossiter-McLaughlin effect and, together with a new analysis of the transit light curve, to refine the parameters of the planet. The unusual shape of the radial velocity anomaly during the transit provides a hint for a nearly transverse Rossiter-McLaughlin effect. The sky-projected angle between the planetary orbital axis and the stellar rotation axis should be λ = 70◦ ± 15◦ to be compatible with our observations. This suggests that some close-in planets might result from gravitational interaction between planets and/or stars rather than migration due to interaction with the accretion disk. This surprising result requires confirmation by additional observations, especially at lower airmass, to fully exclude the possibility that the signal is due to systematic effects. Key words. Techniques: radial velocities - Stars: individual: GSC03727-01064 - Stars: planetary systems: individual: XO-3b 1. Introduction Johns-Krull et al. (2008) announced the detection of XO-3b, an extra-solar planet transiting its F5V parent star with a 3.2-day or- bital period. Transiting planets are of particular interest as they allow measurements of parameters including orbital inclination and planet radius, mass and density. Moreover, follow-up ob- servations can also be performed during transits or anti-transits, yielding physical constraints on planetary atmospheres. Among the forty transiting extra-solar planets known to date, XO-3b is particular as it is among the few on an eccentric or- bit, together with HD 147506b (Bakos et al. 2007), HD 17156b (Fischer et al. 2007; Barbieri et al. 2007), and GJ 436b (Butler et al. 2004; Gillon et al. 2007). XO-3b is also the most mas- sive transiting planet known to date. Most of the sixty known extrasolar planets, with and without transits, with orbital peri- ods shorter than five days have masses below 2 MJup; XO-3b is actually one of the rare massive close-in planets. It is just at the limit between low-mass brown dwarfs and massive plan- ets, 13 MJup, which is defined by the deuterium burning limit. There was a quite large uncertainty on the planetary parame- ters of XO-3b and its host star. Indeed, Johns-Krull et al. (2008) presented a spectroscopic analysis favoring large masses and radii (Mp (cid:39) 13.25 MJup, Rp (cid:39) 1.95 RJup, M(cid:63) (cid:39) 1.41 M(cid:12), and R(cid:63) (cid:39) 2.13 R(cid:12)), whereas their light curve analysis suggests lower (cid:63) Based on observations collected with the SOPHIE spectrograph on the 1.93-m telescope at Observatoire de Haute-Provence (CNRS), France, by the SOPHIE Consortium (program 07A.PNP.CONS). values (Mp (cid:39) 12.03 MJup, Rp (cid:39) 1.25 RJup, M(cid:63) (cid:39) 1.24 M(cid:12), and R(cid:63) (cid:39) 1.48 R(cid:12)) [see however Sect. 5 and Winn et al. (2008a)]. The fast rotating star XO-3 (V sin I = 18.5 km s−1; Johns-Krull et al. 2008) is a favorable object for Rossiter- McLaughlin effect observations. This effect (Rossiter 1924; McLaughlin 1924) occurs when an object transits in front of a rotating star, causing a distortion of the stellar lines profile, and thus an apparent anomaly in the measured radial velocity of the star. The shape of the disturbed radial velocity curve allows one to determine whether the planet is orbiting in the same direction as its host star is rotating, and more generally to measure the sky-projected angle between the planetary orbital axis and the stellar rotation axis, usually noted λ (see, e.g., Ohta et al. 2005; Gim´enez 2006a; Gaudi & Winn 2007). A stellar spin axis not aligned with the orbital angular momentum of a planet (λ (cid:44) 0◦) could reflect processes in the planet formation and migration, or interactions with perturbing bodies (see, e.g., Malmberg et al. 2007, Chatterjee et al. 2007, Nagasawa et al. 2008). Solar System asteroids are examples of objects whose orbital axes can be misaligned from the Sun spin axis by over 30◦. Up to now, spectroscopic transits have been detected for eight exoplanets: HD 209458b (Queloz et al. 2000), HD 189733b (Winn et al. 2006), HD 149026b (Wolf et al. 2007), TrES-1 (Narita et al. 2007), HD 147506b (Winn et al. 2007; Loeillet et al. 2008), HD 17156b (Narita et al. 2008), CoRoT- Exo-2b (Bouchy et al. 2008) and TrES-2 (Winn et al. 2008b). For all of these targets the stellar rotation is prograde relative to the planet orbit, and the sky-projected λ angle is close to 8 0 0 2 n u J 4 ] h p - o r t s a [ 1 v 9 1 7 0 . 6 0 8 0 : v i X r a 2 H´ebrard et al.: Misaligned spin-orbit in the XO-3 planetary system? Table 1. Radial velocities of XO-3 measured with SOPHIE. BJD -2 400 000 RV (km s−1) Planetary transit: zero for most of them. So the axes of the stellar spins are prob- ably parallel to the orbital axes, as expected for planets that formed in a protoplanetary disc far from the star and that later migrated closer-in. Three systems have error bars on the λ angle that do not include 0◦: TrES-1 (λ = 30◦ ± 21◦), CoRoT-Exo- 2b (λ = 7.2 ± 4.5◦) and HD 17156b (λ = 62◦ ± 25◦). However, those cases have the largest error bars on λ, and no firm detection of misalignment has yet been claimed. Barbieri et al. (2008) re- cently presented new radial velocity measurements of HD 17156 secured during a transit, which agree with a spin-orbit alignment. Approximate spin-orbit alignment therefore seems typical for exoplanets, as it is for planets in the Solar System. The un- usual parameters of XO-3b make a test of whether it agrees with this apparent behavior interesting. We present here new mea- surements of XO-3 radial velocity performed during a transit and at other orbital phases. These data refine the orbital param- eters and provide a hint of detection for a transverse Rossiter- McLaughlin effect, i.e. a λ angle possibly near 90◦. We also present a revised analysis of the transit light curve. 2. Observations We observed the host star XO-3 (GSC 03727-01064, mV = 9.91) with the SOPHIE instrument at the 1.93-m telescope of Haute- Provence Observatory, France. SOPHIE is a cross-dispersed, en- vironmentally stabilized echelle spectrograph dedicated to high- precision radial velocity measurements (Bouchy et al. 2006). We used the high-resolution mode (resolution power R = 75, 000) of the spectrograph, and the fast-read-out-time mode of the 4096 × 2048 15-µm-pixel CCD detector. The two optical-fiber circular apertures were used; the first one was centered on the target, and the second one was on the sky to simultaneously mea- sure its background. This second aperture, 2' away from the first one, was used to estimate the spectral pollution due to the moon- light, which can be quite significant in these 3"-wide apertures (see Sect. 3). We acquired 36 spectra of XO-3 during the night of January 28th, 2008 (barycentric Julian date BJD = 2 454 494.5), where a full coverage of the planetary transit was observed. Another 19 spectra were acquired at other orbital phases dur- ing the following two months. Table 1 summaries the 55 spectra finally acquired. The exposure times range from 6 to 30 minutes in order to reach as constant the signal-to-noise ratio as possible. Indeed, SOPHIE radial velocity measurements are currently affected by a systematic effect at low signal-to-noise ratio, which is proba- bly due to CCD charge transfer inefficiency that increases at low flux level. A constant signal-to-noise ratio through a sequence of observations reduces this uncertainty. The different exposure times needed to reach similar signal-to-noise ratios reflect the variable throughputs obtained, due to various atmospheric con- ditions (seeing, thin clouds, atmospheric dispersion). The sky was clear on the night whom the XO-3b transit was observed, but the airmass ranged from 1.2 to 3.1 during this ∼6-hour ob- servation sequence; the exposure times therefore increased dur- ing the transit observation. They remain short enough to provide a good time sampling (20 measurements during the ∼3 hours of the transit). The 19 measurements outside the transit night were performed at airmasses better than 1.4 but with conditions vary- ing from photometric to cloudy. Exposures of a thorium-argon lamp were performed every 2-3 hours during each observing night. Over 2-3 hours, the ob- served drifts were typically ∼ 3 m s−1, which is thus the accuracy of the wavelength calibration of our XO-3 SOPHIE spectra; this ±1 σ (km s−1) exp. time (sec) S/N p. pix. (at 550 nm) 0.020 0.026 0.027 0.027 0.028 0.028 0.029 0.028 0.027 0.027 0.028 0.028 0.028 0.028 0.028 0.028 0.027 0.027 0.036 0.035 0.034 0.034 0.035 0.034 0.036 0.033 0.033 0.033 0.035 0.034 0.035 0.033 0.034 0.036 0.037 0.049 0.050 0.029 0.030 0.031 0.033 0.024 0.030 0.040 0.028 0.023 0.025 0.038 0.036 0.052 0.046 0.071 0.032 0.033 0.019 600 403 373 370 370 370 370 381 380 380 380 380 380 380 380 434 403 392 395 509 500 500 554 531 530 602 622 653 650 682 775 801 906 964 1321 1300 1202 655 1003 775 775 614 907 1806 645 635 755 600 600 600 600 802 1274 999 999 54 42 40 40 39 39 38 39 40 40 39 39 39 40 39 39 40 40 40 43 44 42 44 44 44 45 44 45 46 45 45 45 44 42 43 38 22 37 36 34 34 44 35 29 39 46 43 48 41 48 39 32 37 36 57 54494.4461 54494.4526 54494.4578 54494.4625 54494.4675 54494.4721 54494.4767 54494.4813 54494.4861 54494.4913 54494.4960 54494.5007 54494.5054 54494.5101 54494.5152 54494.5202 54494.5254 54494.5303 54494.5352 54494.5413 54494.5474 54494.5535 54494.5599 54494.5665 54494.5738 54494.5806 54494.5880 54494.5957 54494.6036 54494.6123 54494.6210 54494.6305 54494.6407 54494.6532 54494.6668 54494.6822 54496.2649 54497.2609 54499.2765 54501.2926 54501.4628 54502.2730 54503.2614 54503.4700 54504.4321 54505.2889 54506.2904 54511.4534 54512.4618 54513.3091 54516.3517 54516.4540 54551.3044 54553.3002 54554.3114 -11.298 -11.300 -11.345 -11.379 -11.390 -11.354 -11.460 -11.448 -11.419 -11.411 -11.457 -11.463 -11.549 -11.482 -11.589 -11.652 -11.611 -11.679 -11.691 -11.667 -11.782 -11.700 -11.649 -11.785 -11.811 -11.783 -11.867 -11.767 -11.785 -11.740 -11.783 -11.880 -11.843 -11.920 -11.964 -11.913 -12.723 -10.156 -13.006 -12.433 -12.756 -13.068 -10.936 -10.182 -12.398 -13.132 -11.593 -13.041 -12.246 -10.360 -10.176 -10.267 -10.316 -13.135 -11.004 Other orbital phases: is good enough for the expected signal. We did not use simulta- neous calibration to keep the second aperture available for sky background estimation. During the night of January 28th, 2008, we performed a thorium-argon exposure before the transit and another one after the sequence, about six hours later. The mea- sured drift was particularly low this night, 1 m s−1 in six hours, H´ebrard et al.: Misaligned spin-orbit in the XO-3 planetary system? 3 which makes us confident the wavelength calibration did not un- expectedly drift during the observation of the XO-3b transit. 3. Data reduction We extracted the spectra from the detector images and mea- sured the radial velocities using the SOPHIE pipeline. Following the techniques described by Baranne et al. (1996) and Pepe et al. (2002), the radial velocities were obtained from a weighted cross-correlation of the spectra with a numerical mask. We used a standard G2 mask constructed from the Sun spectrum atlas including more than 3500 lines, which is well adapted to the F5V star XO-3. We eliminated the first eight spectral orders of the 39 available ones from the cross-correlation; these blue or- ders are particularly noisy, especially for the spectra obtained at the end of the transit, when the airmass was high. The resulting cross-correlation functions (CCFs) were fitted by Gaussians to get the radial velocities, as well as the width of the CCFs and their contrast with respect to the continuum. The uncertainty on the radial velocity was computed from the width and contrast of the CCF and the signal-to-noise ratio, using the empirical re- lation detailed by Bouchy et al. (2005) and Collier Cameron et al. (2007). It was typically around 25 m s−1 during the night of the transit, and between 20 and 45 m s−1 the remaining nights. The large V sin I of this rotating stars makes the uncertainty slightly larger than what is usually obtained for such signal-to- noise ratios with SOPHIE. Some measurements were contaminated by the sky back- ground, including mainly the moonlight. As the G2 mask matches the XO-3 spectrum as well as the Sun spectrum re- flected by the Moon and the Earth atmosphere, the moonlight contamination can distort the shape of the CCF and thus shift the measured radial velocity. During our observations, the 29-km s−1 wide (FWHM) CCF of XO-3 is at radial velocities between −13 and −10 km s−1, whereas the moonlight was centered near the barycentric Earth radial velocity, between −23 and −20 km s−1. Thus moonlight contamination tends to blueshift the measured radial velocities. Following the method described in Pollacco et al. (2008) and Barge et al. (2008), we estimated the Moon con- tamination thanks to the second aperture, targeted on the sky, and then subtracted the sky CCF from the star CCF (after scaling by the throughput of the two fibers). Five exposures with too strong contamination were not used. We estimated the accuracy of this correction on one hand by correcting uncontaminated spectra, and on the other hand by correcting uncontaminated spectra on which we have added moonlight contaminations. Comparisons of the corrected velocities to the uncontaminated ones show that the method works well up to ∼ 500 m s−1 shifts, with an uncer- tainty of 1/9 of the correction to which a minimum uncertainty of 25 m s−1 is quadratically added. The second half of the transit night measurements was con- taminated by moonlight, with sky CCFs contrasted between 2 and 5 % of the continuum, whereas the XO-3 CCF has a contrast of 8 %. This implied sky corrections < 150 m s−1 (except for the very last exposure where it was ∼ 300 m s−1), with uncertainties in the range 25-30 m s−1 (40 m s−1 for the last exposure). Five ex- posures obtained later at different phases were contaminated by the moonlight; corrections of 100 to 500 m s−1 were computed, with uncertainties in the range 30-60 m s−1. The final radial velocities are given in Table 1 and displayed in Figs. 1 and 2. The error bars are the quadratic sums of the different error sources (photon noise, wavelength calibration and drift, moonlight correction). Fig. 1. Top: Radial velocity measurements of XO-3 as a func- tion of time, and Keplerian fit to the data (without transit). Only the 23 measurements used for the fit are displayed. The or- bital parameters corresponding to this fit are reported in Table 2. Bottom: Residuals of the fit with 1-σ error bars. 4. Determination of the planetary system parameters 4.1. Refined orbit The radial velocities measurements presented by Johns-Krull et al. (2008) have a typical accuracy of ∼ 160 m s−1. Those secured with SOPHIE are about five times more accurate, so they allow for a refinement of the original parameters of the system. We made a Keplerian fit of the first four SOPHIE measurements per- formed during the transit night and those performed afterwards, at other phases, first using the orbital period from Johns-Krull et al. (2008). For the refinement of the orbit we did not use most of the data secured during the night of January 28th, 2008, in order to remain free from alteration due to transit anomalies and possible systematic effects due to large-airmass observations. The standard deviation of the residuals to the fit is σ(O−C) = 29 m s−1, implying a χ2 of 15.3, which is acceptable according the low degrees of freedom, ν = 18. The 29 m s−1 dispersion of the measurements around the fit is similar to the errors on the in- dividual radial velocity measurements; these estimated error bars thus are approximatively correct. The fits are plotted in Figs. 1 and 2; the derived orbital parameters are reported in Table 2, to- gether with error bars, which were computed from χ2 variations and Monte Carlo experiments. They agree with the Johns-Krull et al. (2008) parameters but the error bars are reduced by fac- tors of three to six. The largest difference is on the eccentricity, which we found 1.6 σ larger than Johns-Krull et al. (2008). The residuals, plotted as a function of time in the bottom panel of Fig. 1, do not show any trend that might suggest the presence of another companion in the system over two months. 4 H´ebrard et al.: Misaligned spin-orbit in the XO-3 planetary system? and is used for the fits plotted in Figs. 1 and 2. Adding HJS and HET data does not significantly change the other orbital parame- ters nor their uncertainties. For the global fit using the radial ve- locities from the three instruments, we did not use the last HET measurement, performed during a transit (see Sect. 6). The Keplerian fit of the new SOPHIE radial velocity mea- surements also improves the transit ephemeris, as the photomet- ric transits reported by Johns-Krull et al. (2008) were secured between December 2003 and March 2007, one hundred or more XO-3b revolutions before the January 28th, 2008 transit. The mid-point of this transit predicted from the Keplerian fit of the SOPHIE radial velocity measurements is tc = 2 454 494.549 ± 0.014 (BJD), i.e. just a few minutes earlier than the prediction from Johns-Krull et al. (2008). The uncertainty on this transit mid-point is ±20 minutes (or ±0.004 in orbital phase). In order to reduce this uncertainty, we observed a recent photometric transit of XO-3b with a 30-cm telescope at the Teide Observatory, Tenerife, Spain, on February 29th, 2008 (Fig. 3). Weather conditions were poor and we therefore ana- lyzed the transit with a fixed model based on the algorithm of Gim´enez (2006b). The fixed parameters were the ratio between the radii of the star and of the planet k = 0.0928, the sum of the projected radii rr = 0.2275, the inclination i = 79.3◦, and the ec- centricity e = 0.26. We then scanned different mid-transit times and found tc = 2 454 526.4668 ± 0.0026 (BJD) from χ2 varia- tions. This reflects photon noise only; fluctuations due to poor weather may introduce additional uncertainties. By taking into account for the uncertainty on the orbital period, this translates into tc = 2 454 494.5507 ± 0.0030 (BJD) for the spectroscopic transit that we observed with SOPHIE on January 28th, 2008, i.e. ten revolutions earlier. That is just two minutes after the above prediction from SOPHIE ephemeris, and the uncertainty on this transit mid-point is ±4.3 minutes (or ±0.0009 in orbital phase). Fig. 2. Phase-folded radial velocity measurements of XO-3 (cor- rected from the velocity Vr = −12.045 km s−1) as a function of the orbital phase, and Keplerian fit to the data. Orbital parame- ters corresponding to this fit are reported in Table 2. For display purpose, all the measurements performed during the transit night are plotted here. However, only the first four measurements of the transit night are used for the orbit fit, together with 19 mea- surements secured at other orbital phases (see § 4.1). Figs. 5 and 7 display a magnification on the transit night measurements. Table 2. Fitted orbit and planetary parameters for XO-3b. Parameters Vr P e ω K T0 (periastron) σ(O − C) reduced χ2 N tc (transit) M(cid:63) R(cid:63) Mp sin i i Mp Rp λ †: using M(cid:63) = 1.3 ± 0.2 M(cid:12) Values and 1-σ error bars −12.045 ± 0.006 3.19161 ± 0.00014 0.287 ± 0.005 −11.3 ± 1.5 1.503 ± 0.010 2 454 493.944 ± 0.009 2 454 494.549 ± 0.014 29 0.85 23 1.3 ± 0.2 1.6 ± 0.2 12.4 ± 1.9† 82.5 ± 1.5 12.5 ± 1.9† 1.5 ± 0.2 70 ± 15 Unit km s−1 days ◦ km s−1 BJD m s−1 BJD M(cid:12) R(cid:12) MJup ◦ MJup RJup ◦ Fitted alone, the 23 SOPHIE measurements have too short time span (60 days) to measure the period more accurately than Johns-Krull et al. (2008) from photometric observations of twenty transits. A 1.5-year time span is obtained when the SOPHIE measurements are fitted together with the radial ve- locities measured by Johns-Krull et al. (2008) using the tele- scopes Harlan J. Smith (HJS) and Hobby-Eberly (HET). This longer time span allows a more accurate period measurement. We obtained P = 3.19168 ± 0.00015 days from the fit using the three datasets, in agreement with the photometric one, and with a similar uncertainty. The final period reported in Table 2 (P = 3.19161 ± 0.00014 days) reflect these two measurements Fig. 3. Light curve of XO-3 observed at the Teide Observatory, Tenerife, during the transit of February 29th, 2008. The tran- sit fit (solid line) provides tc = 2 454 526.4668 ± 0.0026 ≡ 2 454 494.5507 ± 0.0030 (BJD). −0.10.00.10.20.30.40.50.60.70.80.91.01.1−1000−500 0 500 1000 1500 2000XO−3SOPHIEfRV [m/s] . . . . H´ebrard et al.: Misaligned spin-orbit in the XO-3 planetary system? 5 4.2. Transit light curve fit revisited Johns-Krull et al. (2008) point out that the host star radius ob- tained from the spectroscopic parameters (temperature, grav- ity, metallicity) combined with stellar evolution models, R(cid:63) (cid:39) 2.13 R(cid:12), is incompatible with the value obtained from the shape of the transit light curve, namely R(cid:63) (cid:39) 1.48 R(cid:12). Indeed, a large stellar radius implies a large planetary radius (to account for the depth of the transit) and a large inclination angle (to account for the duration of the transit), but the time from the first to the second contacts (ingress) and third to fourth contacts (egress) predicted for such an inclination are too long when compared to the observed transit light curve (see the upper panel of Fig. 9 in Johns-Krull et al. 2008). Formal uncertainties on the stellar spectroscopic parameters and the photometric measurements are insufficient to account for the mismatch. Since there can be only one value of the real stellar radius, this must be due to systematic uncertainties on the spectroscopic parameters, or the parameter derivation from the photometric data, or both. We revisit these analyses below, using the photometric data from Johns-Krull et al. (2008) and the parameters of the Keplerian orbit obtained in § 4.1 from the SOPHIE radial velocity measurements. Regarding the spectroscopic parameters, the formal uncer- tainties stated by Johns-Krull et al. (2008), e.g. 0.06 dex for the gravity log g or 0.03 dex for the metallicity Z, are particularly small. Since these are used in combination with stellar evolu- tion models, even if the actual uncertainties on the observations are small, systematic uncertainties are known to be present in the models themselves. Also, precise gravity measurements are difficult to obtain from stellar spectra. We therefore set a floor level of effective uncertainties in the confrontation with stellar evolution models of 100 K in temperature, of 0.1 dex in log g, or 0.1 dex in Z (see e.g. discussion in Santos et al. 2004 and Pont & Eyer 2004). Regarding the photometric data, we estimated the uncertain- ties including systematics effects with "segmented bootstrap" analysis (Jenkins et al. 2002; Moutou et al. 2004). According to Pont et al. (2006), correlated noise usually dominates the to- tal parameter error budget for ground-based transit light curves. The segmented bootstrap consist of repeating the fit on realiza- tions of the data with individual nights selected at random. The photometric follow-up for XO-3 by Johns-Krull et al. (2008) consists of ten individual nights. Since the sequencing of the data within each night is preserved, this method provide error estimates that takes into account the actual correlated noise in the data. We find much larger uncertainties on the impact pa- rameter than the photon-noise uncertainties. This is corrobo- rated by the discussion in Bakos et al. (2006) of the case of HD 189733. With a much deeper transits and a similar number of high-precision photometry transits covered from several obser- vatories, they found that the determination of the stellar radius from the photometric data produced an error of ∼ 15 %, consis- tent with the discussion in Pont et al. (2006). To estimate the probability distribution of the radius of XO-3 given the available photometric and spectroscopic ob- servations, and the a priori assumption that the star is located near theoretical stellar evolution tracks, we use a Bayesian ap- proach. As discussed in Pont & Eyer (2004), such approach is needed for realistic parameter estimates when the uncertainties are not small compared to the total parameter space and the re- lation between parameters and observable quantities are highly non-linear. We thus calculate the posterior probability distribu- tion of the stellar radius R(cid:63), using Bayes' theorem with stel- lar evolution models and prior probability distributions suitable Fig. 4. Posterior probability distribution function for the stellar radius of XO-3 obtained from Bayesian approach. Dashed line: Using only the constraints from the light curves of Johns-Krull et al. (2008) and the parameters of the Keplerian orbit (§ 4.1). Dotted line: Using only the constraints from the spectroscopic parameters. Solid line: Using all these constraints together. for a Solar-Neighbourhood magnitude-limited sample, as dis- cussed in Pont & Eyer (2004) in the context of the Geneva- Copenhaguen survey. The posterior probability distribution for the stellar radius is (cid:90) calculated according to Bayes' theorem: P(PR(cid:63))P(SR(cid:63))P(R(cid:63)) P(R(cid:63)PS ) = where R(cid:63) is the stellar radius, P the photometric observations and S the spectroscopic observations. The first two terms on the right are the likelihood of the photometric and spectroscopic observations, exp(−1/2χ2), the last term is the a priori distri- bution of R(cid:63). The integral covers the mass, age and metallic- ity parameters. The stellar evolution models provide the func- tion R(cid:63) = R(cid:63)(M(cid:63), Z, age). For more detailed explanations of the method see Pont & Eyer (2004). Fig. 4 displays the posterior probability distribution function for the stellar radius obtained from this Bayesian approach given the spectroscopic and photometric data from Johns-Krull et al. (2008), the stellar evolution models from Girardi et al. (2002), and the orbit parameters determined from the Keplerian fit of the SOPHIE radial velocity measurements (§ 4.1). The probabil- ity distribution function for the radius of XO-3 is centered near R(cid:63) (cid:39) 1.5 R(cid:12), but extends with non-negligible density from 1.3 to 2.0 R(cid:12). It is well described by R(cid:63) = 1.6 ± 0.2 R(cid:12). The corre- sponding masses are M(cid:63) = 1.3 ± 0.2 M(cid:12). This is a quantifica- tion of our "best guess" from the present observational data and prior knowledge about field stars. These parameters are reported in Table 2. 6 H´ebrard et al.: Misaligned spin-orbit in the XO-3 planetary system? 4.3. Transverse Rossiter-McLaughlin effect? The radial velocities of XO-3 measured with SOPHIE during the transit of January 28th, 2008 are plotted in Fig. 5. Surprisingly, they do not show the ordinary anomaly seen in case of prograde transit, i.e. a red-shifted radial velocity in the first half of the transit, then blue-shifted in its second half. During the full transit of XO-3b, the radial velocity is blue-shifted from the Keplerian curve, by about 100 m s−1. Such shape is expected for a trans- verse Rossiter-McLaughlin effect, i.e. when the λ misalignment angle is near 90◦ so the the planet crosses the stellar disk nearly perpendicularly to the equator of the star. This is apparently the case for XO-3b, whose transit seems to only hide some red- shifted velocity components, i.e. a part of the star rotating away from the observer. A schematic view of the XO-3 system with a transverse transit is shown in Fig. 6. We overplot in Fig. 5 models of Rossiter-McLaughlin ef- fects for XO-3b, for λ = 0◦ (upper panel) and λ = 90◦ (lower panel). Following Loeillet et al. (2008) and Bouchy et al. (2008), we used the analytical Ohta et al. (2005) description of the Rossiter-McLaughlin anomaly. We adopted the orbital param- eters of Table 2, a projected stellar rotation velocity V sin I of 18.5 km s−1, and a linear limb-darkening coefficient  = 0.69 from Claret (2004), for Teff = 6250 K and log g = 4.0 dex. The transit was centered on the tc time determined above from the February 29th, 2008 photometric transit. To take into ac- count for the large uncertainty in the masses and radii of the star and its planet derived from spectroscopic and light curve analyses (§ 1 and § 4.2), we plot the models using two extreme sets of parameters over the SOPHIE radial velocities in each panel of Fig. 5: the solid line is the Rossiter-McLaughlin model with large masses and radii as favored from spectroscopic analy- ses, and the dashed line is the Rossiter-McLaughlin model with smaller masses and radii as favored by the light curve analysis. The dotted line is, for comparison, the Keplerian curve without Rossiter-McLaughlin effect. Table 3 summaries the parameters used for the different mod- els and the quantitative estimations of the quality of the fits. Note that the inclination used by Johns-Krull et al. (2008) for large masses and radii, namely i = 79.32◦, produces slightly too long transits duration when used together with our refined orbit. We used i = 78.6◦ in that case, which remains within the ±1.36◦ er- ror bar obtained on i by Johns-Krull et al. (2008). Models with λ = 0◦, or without Rossiter-McLaughlin effect detection, pro- duce poor fits, with high χ2 values and radial velocity dispersions of 60 to 75 m s−1. This is significantly higher than the expected uncertainties on radial velocity measurements, around 33 m s−1 (see Table 1) and the residuals of the Keplerian fit presented in § 4.1, σ(O − C) = 29 m s−1. Thus, our SOPHIE data seem to exclude such ordinary solutions. The models with λ = 90◦ produces lower χ2, with velocity dispersions of 42 or 44 m s−1. The lower panel of Fig. 5 shows that transverse transits produce better fits of the data, centered on the expected mid-transit and with the adequate duration and depth. One should also note that the SOPHIE measurements per- formed just after the transit (orbital phases from 0.21 to 0.23) are well described by the Keplerian orbit model (see Figs. 2 and 5). We recall that these points were not used to determine the Keplerian orbit (§ 4.1); the orbital parameters were determined using only the first four measurements of the January 28th, 2008 night (filled squares in Fig. 5), together with the measurements secured on other nights. The good match of the data with the λ = 90◦ models argues for a transverse transit. This possible de- tection is independent of the set of stellar parameters adopted in Table 3; both produce similar fits with λ = 90◦. The χ2 is slightly better in the case of large masses and radii but this does not seem to be significant according the noise level. Fig. 5. Rossiter-McLaughlin effect models. Top: λ = 0◦ (spin- orbit alignment). Bottom: λ = 90◦ (transverse transit). On both panels, the squares (open and filled) are the SOPHIE radial- velocity measurements of XO-3 with 1-σ error bars as a func- tion of the orbital phase. Only the first four measurements (filled squares) are used for the Keplerian fit (together with 19 mea- surements at other orbital phases; see § 4.1). The dotted line is the Keplerian fit without Rossiter-McLaughlin effect. The two other lines show Rossiter-McLaughlin models with i = 78.6◦ and a/R(cid:63) = 4.8 (solid line) and i = 84.9◦ and a/R(cid:63) = 7.2 (dashed line). The summary of these parameters is in Table 3. Table 3. Chosen parameters for the Rossiter-McLaughlin effect models plotted in Fig. 5 (see text). a/R(cid:63) M(cid:63) a UA M(cid:12) 1.4 0.048 4.8 1.4 0.048 4.8 0.045 1.2 7.2 7.2 1.2 0.045 without transit: R(cid:63) Mp R(cid:12) MJup 13.7 2.1 13.7 2.1 1.3 12.3 12.3 1.3 Rp RJup 2.0 2.0 1.2 1.2 i ◦ 78.6 78.6 84.9 84.9 λ ◦ 0 90 0 90 σ m s−1 61 42 74 44 59 χ2 196 63 291 79 169 The ∼ 40 m s−1 dispersion of the data from these transverse models remains slightly above the computed uncertainties on ra- dial velocity measurements. This suggests that some extra uncer- tainties might be present and not taken into account in the error budget. This make us considering this observation as a hint of detection for a spin-orbit misalignment. An explanation for this too large dispersion could be the high atmospheric refraction. Indeed, as seen in Sect. 2, the end of the transit was observed at large airmass. This could intro- duce biases in the radial velocity measurements that are difficult to quantify. This agrees with the increasing dispersion of the data from these transverse models, which is in the range 30−35 m s−1 H´ebrard et al.: Misaligned spin-orbit in the XO-3 planetary system? 7 plying a 42 m s−1 velocity dispersions around the model. The best fit with these parameters is plotted in Fig. 7. The residuals are plotted in Fig. 8 in three cases: without transits, with spin- orbit alignment, and with λ = 70◦. Among them, the last case is clearly favored by our data when the parameters from Winn et al. (2008a) are adopted. Fig. 7. Rossiter-McLaughlin effect models with λ = 70◦ and the small Rp value reported by Winn et al. (2008a). The squares (open and filled) are the SOPHIE radial-velocity measurements of XO-3 with 1-σ error bars as a function of the orbital phase. Only the first four measurements (filled squares) are used for the Keplerian fit (together with 19 measurements at other orbital phases; see § 4.1). The dotted line is the Keplerian fit without Rossiter-McLaughlin effcet. The solid and dotted lines are the models with and without Rossiter-McLaughlin effect. Fig. 6. Schematic view of the XO-3 system with transverse tran- sit, as seen from the Earth. The stellar spin axis is shown, as well as the planet orbit and the λ misalignment angle. The scale is in stellar radii. The limit between soft and strong grey on the λ-scale represents the favored value from our observations (λ = 70◦, see Sect. 5). in the first half of the transit, then in the range 40 − 45 m s−1 in the second half. The larger dispersion might also be partly explained as the expected errors are increasing in the second part of the transit because of moonlight correction (see § 3). In addition, it is pos- sible that the planet was crossing the stellar disk above a spot; this could cause extra radial velocity variations (jitter), as the anomaly that is visible near the phase 0.19. Stellar H and K Ca II lines do not show core emissions, but they are less deep = −4.6 ± 0.2, and we than other F5 stars. This implies log R(cid:48) can not exclude XO-3 presents stellar activity, including spots. HK 5. A small radius for XO-3b Shortly after the submission of this paper, photometry of 13 tran- sits of XO-3b were released by Winn et al. (2008a). These new observations strongly favors the smaller values for XO-3 and XO-3b radii and masses. The parameters reported by Winn et al. (2008a) agree with the ones presented here (Table 2); this strengthen the Bayesian approach we used in §4.2. Timing pa- rameters (as P or T0) from Winn et al. (2008a) are more ac- curate thanks to their high-quality transit photometry, whereas orbital parameters (as e, ω or K) are more accurate in the present study due to the high-quality radial velocity measure- ments with SOPHIE. The small Rp value excludes a grazing transit for XO- 3b, and the corresponding models plotted in solid lines in Fig. 5; the Rossiter-McLaughlin anomaly should thus be large and detectable, with an amplitude near the order of magnitude (V sin I)(Rp/R(cid:63))2 (cid:39) 150 m s−1 (Winn et al. 2008a). We fitted the SOPHIE data using the updated parameters from Winn et al. (2008a), in particular Rp = 1.217 RJup, a/R(cid:63) = 7.07 and i = 84.20◦. According χ2 variations, the range of λ compatible with our observations is 70◦ ± 15◦. The lowest χ2 is ∼ 64, im- Fig. 8. Residuals of the Rossiter-McLaughlin effect fits. Top: Without transit. Middle: λ = 0◦ (spin-orbit alignment). Bottom: λ = 70◦. The squares (open and filled) are the SOPHIE radial- velocity measurements of XO-3 with 1-σ error bars as a func- tion of the orbital phase. Only the first four measurements (filled squares) are used for the Keplerian fit (together with 19 measure- ments at other orbital phases; see § 4.1). The vertical, dashed line shows the center of the transit. 8 H´ebrard et al.: Misaligned spin-orbit in the XO-3 planetary system? 6. Conclusion and discussion Table 2 summarizes the star, planet and orbit parameters of the XO-3 system that we obtained from our analyses. The radial ve- locity measurements that we performed with SOPHIE during a planetary transit suggest that the spin axis of the star XO-3 could be nearly perpendicular to the orbital angular momentum of its planet XO-3b (λ = 70◦ ± 15◦). We note that one Johns-Krull et al. (2008) HET measurement was obtained near a mid-transit of XO-3b. This radial velocity is blue-shifted by (260 ± 194) m s−1 from the Keplerian curve, in agreement with the possible trans- verse Rossiter-McLaughlin effect we report here, though with a modest significance. The SOPHIE observation remains noisy, showing more dis- persion around the fit during the transit than at other phases. We consider this result as a tentative detection of transverse transit rather than a firm detection. Indeed, the end of the transit was ob- served at large airmasses, which could possibly biases the radial velocity measurements. Our fits favor a transverse transit, but one can not totally exclude a systematic error that would mimic by chance the shape of a transverse transit. This would imply that the radial velocities measured during the end of the transit night, at large airmasses, would be off by about 100 m s−1, i.e. three to four times the expected errors. Other spectroscopic transits of XO-3b should thus be observed. They will allow the transverse Rossiter-McLaughlin effect to be confirmed or not, and to better quantify its parameters, such as the value of the misalignment angle λ. Narita et al. (2008) estimate that the timescale for spin-orbit alignment through tidal dissipation is longer than thousand Gyrs. This timescale is uncertain, but much longer than the timescale for orbit circularization, which itself is longer than the age of XO-3, estimated in the range 2.4 − 3.1 Gyrs (Johns-Krull et al. 2008); there are thus no obvious reasons to exclude an eccen- tric, transverse system. A strong spin-orbit misalignment would favor of formation scenarii that invokes planet-planet scattering (Ford & Rasio 2006) or planet-star interaction in a binary system (Takeda et al. 2008) rather than inward migration due to inter- action with the accretion disk. This suggests in turn that some close-in planets might result from gravitational interaction be- tween planets and/or stars. Chatterjee et al. (2007) and Nagasawa et al. (2008) have recently shown that scattering with at least three large planets can account for hot Jupiters and predicts high spin-orbit inclinations (see also Malmberg et al. 2007). On an- other hand, XO-3b is an object close to the higher end of plan- etary masses. As discussed for instance by Ribas & Miralda- Escud´e (2007), there are some indications that these objects be low-mass brown dwarfs, formed by gas cloud fragmentation rather than core accretion; so that XO-3b may not necessarily constrain planet formation scenario. Finally, pseudo-synchronization might be questioned in the case of the massive XO-3b (MXO−3 (cid:39) 100 × MXO−3b), which moves on an eccentric orbit with a periastron particularly near its host star. Tidal frictions might be high enough to tune the stellar rotation velocity close to the velocity of its companion on its orbit at the periastron (Zahn 1977). The expected pseudo- synchronized stellar rotation is given by Vrot = Vp× R(cid:63) a(1−e), where Vp = 2π a 1+e 1−e is the planet velocity at the periastron. For the P XO-3 system, this translates into Vrot (cid:39) 30 R(cid:63) R(cid:12) km s−1 accord- ing to the values in Table 2. As the XO-3 radius is larger than 1.1 R(cid:12), its rotation velocity V sin I = 18.5 km s−1 is clearly smaller than the pseudo-synchronized velocity. However, we note that a spin-orbit misalignment would tend to reduce the (cid:113) the thank technical pseudo-synchronized rotation velocity of the star. In that case, the planet approaches at the nearest of its star at low stellar lat- itude, and not above the stellar equator. Pseudo-synchronization might thus be possible if actually there is a significant spin-orbit misalignment in the XO-3 system. Acknowledgements. We team at Haute-Provence Observatory for their support with the SOPHIE instrument and the 1.93-m to the SOPHIE Consortium from the OHP telescope. Financial support "Programme national de plan´etologie" (PNP) of CNRS/INSU, France, and from the Swiss National Science Foundation (FNSRS) are gratefully acknowledged. NCS would like to thank the support from Fundac¸ao para a Ciencia e a Tecnologia, Portugal, in the form of a grant (references POCI/CTE- AST/56453/2004 and PPCDT/CTE-AST/56453/2004), and through program Ciencia 2007 (C2007-CAUP-FCT/136/2006). XB acknowledges support from the Fundac¸ao para a Ciencia e a Tecnologia (Portugal) in the form of a fellowship (reference SFRH/BPD/21710/2005) and a program (reference PTDC/CTE-AST/72685/2006), as well as the Gulbenkian Foundation for funding through the "Programa de Est´ımulo Investigac¸ao". AML, AE, and DE acknowledge support from the French National Research Agency through project grant ANR-NT-05-4 44463. MR is funded by the EARA - Marie Curie Early Stage Training fellowship. References Bakos, G. ´A., Knutson, H., Pont, F., et al. 2006, ApJ, 650, 1160 Bakos, G. ´A., Kov´acs, G., Torres, G., et al. 2007, ApJ, 670, 826 Baranne, A., Queloz, D., Mayor, M., et al. 1994, A&AS, 119, 373 Barbieri, M., Alonso, R., Laughlin, G., et al. 2007, A&A, 476, L13 Barbieri, M., et al. 2008, IAU Symposium No. 253 - "Transiting Planets", May 19-23, 2008, Harvard Barge, P., Baglin, A., Auvergne, M., et al. 2008, A&A, 482, L17 Bouchy, F., Pont, F., Melo, C., et al. 2005, A&A, 431, 1105 Bouchy, F., and the Sophie team, 2006, in Tenth Anniversary of 51 Peg-b, eds. L. Arnold, F. Bouchy & C. Moutou, 319 Bouchy, F., Queloz, D., Deleuil, M., et al. 2008, A&A, 482, L25 Butler, R. P., Vogt, S. S., Marcy, Ge. W., et al. 2004, ApJ, 617, 580 Chatterjee, S., Ford, E. B., Rasio, F. A. 2007, ApJ, submitted [arXiv:0703166] Claret, A., 2004, A&A, 428, 1001 Collier Cameron, A., Bouchy, F., H´ebrard, G., et al. 2007, MNRAS, 375, 951 Fischer, D. A., Vogt, S. S., Marcy, G. W., et al. 2007, ApJ, 669, 1336 Ford, E. B., & Rasio, F. A. 2006, ApJ, 638, L45 Gaudi, B. S., & Winn, J. N. 2007, ApJ, 655, 550 Gillon, M., Pont, F., Demory, B.-O., Mallmann, F., Mayor, M., Mazeh, T., Queloz, D., Shporer, A., Udry, S., Vuissoz, C. 2007, A&A, 472, L13 Gim´enez, A. 2006a, ApJ, 650, 408 Gim´enez, A. 2006b, A&A, 450, 1231 Girardi, M., Manzato, P., Mezzetti, M., Giuricin, G., Limboz, F. 2002, ApJ, Wolf, A. S., Laughlin, G., Henry, G. W., et al. 2007, ApJ, 667, 549 Zahn, J.-P. 1977, A&A, 57, 383 569, 451 Jenkins, J. M., Caldwell, D. A., Borucki, W. J. 2002, ApJ, 564, 495 Johns-Krull, C. M., McCullough, P. R., Burke, C. J., et al. 2008, ApJ, 677, 657 Loeillet, B., Shporer, A., Bouchy, F., et al. 2008, A&A, 481, 529 McLaughlin, D.B., 1924, ApJ, 60, 22 Malmberg, D., Davies, M. B., Chambers, J. E. 2007, MNRAS, 377, L1 Moutou, C., Pont, F., Bouchy, F., Mayor, M. 2004, A&A, 424, L31 Nagasawa, M., Ida, S., Bessho, T. 2008, ApJ, 678, 498 Narita, N., Enya, K., Sato, B., et al. 2007, PASJ, 59, 763 Narita, N., Sato, B., Ohshima, O., Winn, J. N. 2008, PASP, 60, L1 Ohta, Y., Taruya, A., Suto, Y. 2005, ApJ, 622, 1118 Pepe, F., Mayor, M., Galland, F., et al. 2002, A&A, 388, 632 Pollacco, D., Skillen, I., Collier Cameron, A., et al. 2007, MNRAS, 385, 1576 Pont, F., & Eyer, L. 2004, MNRAS, 351, 487 Pont, F., Zucker, S., Queloz, D. 2006, MNRAS, 373, 231 Queloz, D., Eggenberger, A., Mayor, M., et al. 2000, A&A, 359, L13 Ribas, I., & Miralda-Escud´e, J. 2007, A&A, 464, 779 Rossiter, R. A., 1924, ApJ, 60, 15 Santos, N. C., Israelia, G., Mayor, M. 2004, A&A, 415, 1153f Takeda, G., Kita, R., Rasio, F. A. 2008, ApJ in press [arXiv:0802.4088] Winn, J. N., Johnson, J. A., Marcy, G. W., et al. 2006, ApJ, 653, L69 Winn, J. N., Holman, M. J., Bakos, G. A., et al. 2007, ApJ, 665, L167 Winn, J. N., Holman, M. J., Torres, G., et al. 2008a, ApJ, in press Winn, J. N., Asher Johnson, J., Narita, N., et al. 2008b, ApJ, in press [arXiv:0804.4475] [arXiv:0804.2259]
astro-ph/0612298
1
0612
2006-12-12T00:46:15
A Photometric and Spectroscopic Study of the Cataclysmic Variable ST LMi during 2005-2006
[ "astro-ph" ]
We present orbit-resolved spectroscopic and photometric observations of the polar ST LMi during its recent low and high states. In the low state spectra, we report the presence of blue and red satellites to the H-alpha emission line; the velocities and visibility of the satellites vary with phase. This behavior is similar to emission line profile variations recently reported in the low state of AM Her, which were interpreted as being due to magnetically-confined gas motions in large loops near the secondary. Our low-state spectroscopy of ST LMi is discussed in terms of extreme chromospheric activity on the secondary star. Concurrent photometry indicates that occasional low-level accretion may be present, as well as cool regions on the secondary near L1. Furthermore, we report a new ``extreme low-state'' of the system at V~18.5mag. Our orbital high-state spectroscopy reveals changes in the emission line profiles with orbital phases that are similar to those reported by earlier high-state studies. The complicated emission line profiles generally consist of two main components. The first has radial velocity variations identical to that of the major emission H-alpha component seen in the low state. The second is an additional red-shifted component appearing at the phases of maximum visibility of the accreting column of the white dwarf; it is interpreted as being due to infall velocities on the accreting magnetic pole of the white dwarf. At the opposite phases, an extended blue emission wing appears on the emission line profiles. We confirm the presence of a broad absorption feature near 6275Ang which has been previously identified as Zeeman sigma(-) absorption component to H-alpha. This feature appears at just those phases when the accretion pole region is mostly directly visible and most nearly face-on to the observer.
astro-ph
astro-ph
A Photometric and Spectroscopic Study of the Cataclysmic Variable ST LMi during 2005-20061, 2 S. Kafka1,2,3 S.B. Howell3,4,5 R.K. Honeycutt3,6,7 and J.W. Robertson3,8,9 ABSTRACT We present orbit-resolved spectroscopic and photometric observations of the polar ST LMi during its recent low and high states. In the low state spectra, we report the presence of blue and red satellites to the Hα emission line; the velocities and visibility of the satellites vary with phase. This behavior is similar to emission line profile variations recently reported in the low state of AM Her, which were interpreted as being due to magnetically-confined gas motions in large loops near the secondary. Our low-state spectroscopy of ST LMi is discussed in terms of extreme chromospheric activity on the secondary star. Concurrent photometry indicates that occasional low-level accretion may be present, as well as cool regions on the secondary near L1 Furthermore, we report a new "extreme low-state" of the system at V∼18.5 mag. Our orbital high-state spectroscopy reveals changes in the emission line profiles with orbital phases that are similar to those reported by earlier high-state studies. The complicated emission line profiles generally consist of two main components. The first 1CTIO/NOAO, Casilla 603, La Serena, Chile. 2E-mail: [email protected] 3Visiting Astronomer, Kitt Peak National Observatory, National Optical Astronomy Observatory, which is oper- ated by the Association of Universities for Research in Astronomy, Inc. (AURA) under cooperative agreement with the National Science Foundation. 4WIYN Observatory & NOAO, P.O. Box 26732, 950 N. Cherry Ave., Tucson, AZ 85719 5E-mail: [email protected] 6Indiana University, Astronomy Department, 319 Swain Hall West, Bloomington, IN 47405 7E-mail: [email protected] 8Arkansas Tech Univ., Dept. of Physical Sciences, 1701 N. Boulder , Russellville, AR 72801-2222 9E-mail: [email protected] -- 2 -- has radial velocity variations identical to that of the major emission Hα component seen in the low state. The second is an additional red-shifted component appearing at the phases of maximum visibility of the accreting column of the white dwarf; it is interpreted as being due to infall velocities on the accreting magnetic pole of the white dwarf. At the opposite phases, an extended blue emission wing appears on the emission line profiles. We confirm the presence of a broad absorption feature near 6275A which has been previously identified as Zeeman σ− absorption component to Hα. This feature appears at just those phases when the accretion pole region is mostly directly visible and most nearly face-on to the observer. Subject headings: binaries: close -- stars: LMi),stars: magnetic fields, stars: activity stars: general -- stars: individual (ST 1. Introduction Cataclysmic Variables (CVs) are short orbital period (hours to days) interacting binaries con- taining a white dwarf (WD) primary and a K-M (or later) dwarf mass losing secondary star. It is generally accepted that mass transfer occurs as the secondary overflows its Roche lobe and matter falls toward the primary star through the inner Lagrangian point (L1). The future of the accretion stream is then determined by the magnetic field of the WD: in magnetic systems the stream is channeled toward the magnetic poles of the primary along its magnetic field lines; in non-magnetic systems a hot accretion disk is formed around the primary, replenished by continuous mass trans- fer. The abundance of CV properties is reviewed in Warner (1995) and will not be repeated here. Among the various subcategories, there is one which includes both disk and magnetic CVs with large (up to 5 mag) apparently random drops in the optical brightness; these are known as VY Scl stars. Although there is no well-accepted explanation for the cause of these low states, it is clear that accretion from the secondary star is significantly diminished, often exposing the two stellar components and providing a unique opportunity to study their photospheres. Sadly, low states in CVs are erratic in occurrence and duration, making observational study a challenge. Therefore, we generally rely on photometric monitoring efforts by professional or amateur astronomers in order to ascertain when a given system is in a low state. Recently, low state observations of the magnetic CV prototype, AM Her, revealed short outburst-like events in the optical light curves of the system which were attributed to flares (ac- 1Based on observations from the WIYN Observatory, which is a joint facility of the University of Wisconsin- Madison, Indiana University, Yale University, and the National Optical Astronomy Observatories, USA. 2Based on observations obtained with the Mayall 4m telescope at Kitt Peak National Observatory, a division of the National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy Inc., under cooperative agreement with the National Science Foundation -- 3 -- tivity) on the secondary star (Kafka et al. 2005a). Later work (Kafka et al. 2005b) revealed components of the Hα emission line arising from the vicinity of the secondary star. These features were interpreted as being due to stellar activity on the secondary and their study provides a new method to explore the extent and variability of magnetic structures on the low-mass donor star. Stellar activity on the secondary in CVs has been invoked to explain a plethora of phenomena, such as long-term light curve variations (Bianchini 1988) and high/low states (Livio and Pringle 1994). However, correlation of such phenomena with stellar activity has yet to be fully explored. In AM Her, prominence-like loops with gas velocities of ∼300 km/sec persisted for more than three years (Kafka, Honeycutt & Howell 2006), suggesting that they might be present even when the system is in a high state, albeit masked by accretion flux (Kafka et al. 2005b). Recently the polar VV Pup exhibited similar structures in the Hα emission line during its low state (Mason et al. 2007). In VV Pup the Hα emission structures were highly variable on timescales of several weeks. ST LMi (CW 1103+254) was introduced as a magnetic CV in the work of Stockman et al. (1983), in which linear and circular polarization varied with the 114 min orbital period of the system. Several high-state emission line components (Stockman et al. 1983; Bailey et al. 1985) have been attributed to infall velocities and to features arising near the secondary star. Most studies (Stockman et al. 1983; Schmidt, Stockman & Grandi 1983; Bailey et al. 1985; Cropper 1986; Cropper & Horne 1994, Stockman & Schmidt 1996) have concluded that the system is normally a one-pole accretor, in which the active pole is self-eclipsed by the limb of the white dwarf for a major part of the orbit. Peacock et al. (1992) reported polarimetric evidence for accretion onto a second pole at times. Schmidt, Stockman and Grandi (1983) provided a geometrical model of the system, according to which the secondary star slightly lags the single accreting magnetic pole. Using near-IR data, Ferrario, Bailey & Wickramasinghe (1993) modeled the high state cyclotron emission from the active pole and derived a magnetic field strength of 11.5±0.5MG. Since 1990, ST LMi has been a part of the long-term monitoring program of a 0.41-m automated telescope located in central Indiana (RoboScope3). The 1990-2004 light curve of the system, which appeared in Kafka & Honeycutt 2005 (KH05), revealed a 5-year long low state (1992-1997), followed by ∼6 years of high state (Fig 1). During the high state, the light curve of ST LMi shows a ∼1.5 mag orbital modulation caused by the self-eclipse of the single accreting pole by the white dwarf (Stockman & Schmidt 1996). During the low states, the RoboScope light curve showed that ST LMi usually reached a quiescent magnitude of 17.5 with an approximate sinusoidal modulation of ∼0.2 mag (KH05). In this paper, we present a photometric and spectroscopic study of the system during 2005 and 2006. In Section 2 we describe our data and data reductions, in Section 3 we discuss the results, while in Section 4 we compare these results to the behavior of other polars in the low state. Our conclusions are summarized in Section 5. The results of this study show that the secondary star in ST LMi is a highly active star and behaves in a similar manner as those observed in a small but 3Honeycutt et al. 1994 and refs therein growing sample of polar mass donors observed during low states. -- 4 -- 2. Observations and Reductions 2.1. Photometry In the top panel of figure 1 we have combined our V-band photometry of ST LMi from various sources. The data prior to 2004-Nov are RoboScope data from KH05. RoboScope continued to collect photometry of ST LMi until 2005-Mar, which is also included in figure 1; after 2005-Mar RoboScope ceased operation (temporarily we hope). The RoboScope data were processed/reduced through a pipeline that performs aperture photometry and extracts light curves of all the stars in the field based on the method of incomplete ensemble photometry (Honeycutt 1992). The photometry later than 2005-Mar in figure 1 is mostly from the WIYN 0.91-m telescope at Kitt Peak. Processing of the non-RoboScope images followed standard IRAF4 procedures, using aperture photometry for the extraction of instrumental magnitude. We used stars from Henden & Honeycutt (1995) as secondary standards. Much (but not all) of this 0.91-m photometry consists of sequences of exposures covering ∼one orbital period; these sequences are grouped into several observing runs, designated A-D (figure 1 and Table 1). Figures 2-4 show our phased photometric light curves of ST LMi, using only the data from figure 1 that is continuous for ∼1 orbit. In this paper we have adopted an orbital ephemeris which uses the period from Cropper (1986) and a spectroscopic zeropoint (or epoch) from Howell et al. (2000). The spectroscopic epoch is based on IR observation of photospheric absorption lines from the secondary star and phase zero marks the time of inferior conjunction of the secondary. For reference, phase zero for our ephemeris corresponds to phase 0.17 on the polarimetric ephemeris of Cropper (1986), thereby placing the expected location of the observer's most direct view of the active accretion pole region near our phase 0.83. The A-D designations shown in fig. 1, and listed in Table 1, are used as appropriate in figures 2-3. We show phased RoboScope light curves in fig. 4. 2.2. Spectroscopy Our spectroscopic data were obtained in four different observing runs during the interval 2005- Jun to 2006-May (indicated in figure 1, bottom). During runs 2005-Jun (one night) and 2006-Feb (two nights) the HYDRA multi-object fiber spectrograph (MOS) on the WIYN5 3.5-m telescope 4IRAF 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. 5The WIYN Observatory is a joint facility of the University of Wisconsin-Madison, Indiana University, Yale University, and the National Optical Astronomy Observatories, USA. -- 5 -- was used. At WIYN we employed the 600 line mm−1 grating in first order, blazed at 7500A; the spectral coverage was ∼5500-8000A, with a resolution of ∼3A. A number of non-assigned fibers were placed at random locations to serve for sky subtraction, and a CuAr lamp was used for wavelength calibration. The R-C Spectrograph on the 4-m KPNO telescope was used to acquire two spectra of ST LMi in 2005-Dec, and to obtain a sequence of spectra covering a little more than one orbit in 2006-May. For the 2005-Dec spectra we used the KPC-007 grating and the GG-385 order separation filter. The spectral coverage was ∼5500-7500A with a resolution of ∼3A. For the 2006-May data we used grating KPC-24 in second order with a 1" slit to match the seeing. This combination provided ∼1000A coverage at 1.3A resolution. The 2006-May night was clear with photometric conditions and stable seeing, allowing extraction of good quality relative photometric information from these spectra. We produced a synthetic light curve using the fluxed spectra from this night of high state spectroscopy (figure 3 bottom right, see §4.2). An instrumental magnitude was extracted from the reduced 4-m spectra using imstat to sum the counts in rectangular windows (including both lines and continuum) for star and sky. The length of each window was ∼80% of the length of the spectrum, yielding an effective wavelength approximately midway between V and R. Exposure times for all our spectra ranged between 300 and 1800 sec, depending on the weather conditions (Table 1). Data processing and reduction employed standard IRAF procedures such as twodspec with wavelength calibration accomplished via the use of arc lamp exposures obtained near in time and position to our stellar exposures. For the R-C spectrograph data, we extracted the 2-d images, reduced them in the usual fashion, flux calibrated them using the spectrophotometric standard star GD140, and produced final 1-d spectra. No corrections were made for atmospheric extinction as the slit was kept at the parallactic angle and the total range of airmass was small and never larger than 1.6. The usual telluric absorption features remain in all the final spectra. For Hydra, we performed the additional steps to allow correction for scattered light and the more complex sky subtraction necessary for a fiber spectrograph. Partly cloudy weather conditions prevailed during all our runs with the exception of the 2006-May observations. 3. Photometric Results The photometric coverage in figure 1 is not continuous during 2005-2006 (due to the untimely failure of RoboScope) but the available photometry does show that ST LMi was in a nearly- continuous low state during most of our photometric and spectroscopic observations. In 2005-2006, the most common basal magnitude in the low state appears to be ∼17.5 (bottom panel of figure 1); consistent with the brightness during the 1992-1996 low state recorded by RoboScope (top panel of figure 1). However, the photometric sequence of 2006-Feb-12 (UT) (sequence C1 in Table 1) shows that the system reached V∼18.0-18.4, significantly fainter than other data points in figure 1. The 0.91-m exposures showing this unusually faint state were acquired using the same equipment, and were reduced using the same procedures as our other photometric observations near this time (see -- 6 -- table 1) which show the system at V∼17.5. We therefore conclude that this unusually low state is real. Although it appears inconsistent with the rest of the RoboScope data (which never showed such a state over 13 years), the RoboScope magnitude limit is about 17.8; therefore fainter events would escape detection. Although unusual, the presence of an "extreme-low state" is not unique to ST LMi; VV Pup, for example, was also observed to have two different low state levels (Mason et al. 2007). This extreme low state, allows for an interesting experiment for ST LMi: assuming that the extreme low state (mean V=18.2mag) represents a time where activity and accretion temporarily stop then the observed light in V should be dominated by the white dwarf alone. For a DA WD, MV =12.5. This gives a distance for ST LMi of 138pc. This is very similar to the distance of 136 pc derived by Bailey et al. (1985) using the K surface brightness for an M5 main sequence star and a secondary mass of 0.18 Msun. The low state orbital light curves from our photometric monitoring are shown in figures 2 and 3. Light curves A1-A5 show a 0.3-0.8 mag "bump" near phase 0.8, the phase at which we most directly view the main accretion region on the magnetic white dwarf. Although the mean magnitude of these sequences was about at the "normal" low state brightness, the observed bump in the light curves suggest that some low-level variable accretion (likely wind accretion) was taking place in 2005-late Feb/early March. In contrast, the low state orbital light curves for 2005-Jun and 2006-Feb (B1-B4, D1) appear consistent with the 1992-1996 RoboScope low-state orbital light curve shown in fig. 4, bottom panel). These low state light curves are essentially featureless with (at best) only a ∼0.1 mag sinusoidal variation superposed on ∼0.2 mag random variations. Note that even the random variation is largely missing for the low-state orbital light curve for UT 2006- Feb-21 (fig. 3, D1). In 2005-March-1 (UT) we monitored ST LMi (A2) using an R filter. Figure 3 (right, top) shows our phased differential R band light curve with an arbitrary zero point. Again we see an orbital bump during this low state observation but with a larger amplitude than observed in V at nearly the same time (fig. 2). Even with a slightly variable amplitude from night to night, the larger amplitude bump observed in R compared with V during the low state agrees with previous high state observations of ST LMi showing that it becomes redder near phase 0.8. We will see below that for ST LMi, the red color at this phase is due to the contribution of the cyclotron continuum to the overall flux, a contribution that outshines the other components during this phase interval. It is of particular importance to ascertain the photometric state at the time of the spectroscopic observations, using both the brightness of the system and the orbital light curve. For the 2005- Jun-17 WIYN spectra, simultaneous photometry (figure 2, sequence B2) shows V∼17.5 magnitude, with little or no orbital modulation. Therefore the system was in the normal low state, with no evidence of accretion. For the 2005-Dec-31 4-m spectra we have V-band measures on two nights, 7 and 8 nights preceding the spectroscopy, from Tenagra Observatory6, showing ST LMi at V∼17.3, again indicating a normal low state. During the 2006-Feb-21/22 (UT) WIYN spectroscopy, we have simultaneous WIYN 0.91-m photometry showing the system at V∼17.5. The orbital light curve 6http://www.tenagraobservatories.com -- 7 -- for 2006-Feb-22 (UT) (sequence D1, figures 1 and 4) is very "quiet", with no significant orbital variations. Therefore the system showed no evidence of accretion for 2006-Feb-21 (UT) (based on its brightness) nor on 2006-Feb-22 (UT) (based on both the brightness and on the flat orbital light curve). The 2006-Feb-21/22 data is the nearest photometric measurement to our 2006-May-19 4-m spectroscopy available to us. In order to better determine the photometric state for the 2006-May spectroscopy we produced a relative light curve using the summed flux in the 4-m spectra (see section 2.2 for details). The transparency and seeing on this night were very good and stable, resulting in a quantitatively good light curve (figure 3, bottom right) at an effective wavelength which is approximately mid-way between V and R, and has an arbitrary zeropoint. This light curve has a conspicuous 1.7 mag "bump" near phase 0.8, similar to that seen in high-state photometry such as the RoboScope high-state orbital light curve (top panel, figure 47). The bump at phase 0.8 is attributed to accretion onto the white dwarf pole as seen most directly by the observer. While some low state photometry shows residual accretion, likely due to wind accretion (e.g., sequences A1-A5, fig. 2) the amplitude and consistency of the bump seen in the 2006-May light curve (fig. 3) is similar to the RoboScope photometry high state light curve (fig. 4, top). Thus, it seems that during our 2006-May spectroscopy ST LMi was probably no longer in the low state; this will be discussed in Sec. 4.2. In summary, we conclude that the orbit-resolved WIYN spectroscopy of 2005-Jun and 2006- Feb were in the low state, whereas the orbit-resolved 4-m spectra of 2006-May were likely in the high state. 4. Spectroscopic Results 4.1. Low State Spectra Figure 5 shows representative spectra from the three low-state spectroscopic runs in 2005-Jun, 2005-Dec, and 2006-Feb. All spectra were taken at a similar spectral resolution of ∼3A and clearly show the TiO bandheads near 7050A from the secondary star, confirming our earlier conclusion that any accretion luminosity was very low at these epochs. The Hα emission line has considerable structure, as shown in the enlarged inset plots, similar to the profiles of the Hα line of AM Her (Kafka et al. 2005b; 2006) and VV Pup (Mason et al. 2007) during their recent low states. Figures 6 and 7 show the Hα emission line profiles as a function of phase for two adjacent nights of WIYN spectroscopy 2006-Feb-21/22 UT. In both nights, the emission line occasionally develops a red satellite near phases 0.75-1.00 and a blue component to the Hα emission line for phases near 0.4 and 0.5. The complex emission line profiles have been fitted with Gaussian (in a manner identical to that described in Kafka et al. 2006 for AM Her) to determine their radial velocities (RVs). We 7the phasing of the folded light curves of ST LMi in KH05 (figure 5 in that paper) is in error because a wrong epoch for the ephemeris was mistakenly employed. -- 8 -- have been conservative in choosing between noise in the profile and a real emission line component; we used satellites that exceeded the noise level by 2.5σ. The RV curves for the line components we judged to be reliable, are shown in figure 8. Figure 8 also includes the RV of the central peak of the Hα line for the 2005-June (squares) and the 2005-December (triangles) data. The RVs of the satellites are plotted with open symbols; different colors correspond to different cycles. The RV phasing of the central emission line component of Hα indicates that it arises from the secondary star's side of the center of mass. The blue/red crossing is 0.05 phase units away from phase zero (which marks inferior conjunction of the secondary), but is consistent with the uncertainty of ∼0.1 phase units quoted by Howell et al. (2000) for the time of conjunction in the ephemeris. The velocities and phase visibility of the occasional blue and red satellites in figure 8 are quite similar to that those reported for AM Her (Kafka et al. 2005, 2006) and VV Pup (Mason et al. 2007); they have been interpreted as arising from gas motions along two large magnetically-confined loops, perhaps similar to slingshot prominences. Figure 9 (top) shows the low-state equivalent width of the full Hα emission line for the WIYN spectroscopy of 2006-02-21/22 vs orbital phase. For comparison, we reproduce the phased light curve (D1) of figure 3 on a expanded magnitude scale so as to see the structure of the lightcurve. The Hα emission line appears to be stronger between phases 0.1-0.5 (albeit with variable strength); this is the trailing side of the secondary star. Near the same phases the light curve appears to have a V∼0.05 mag dip, which is repeated in two consecutive orbits. Since accretion was absent at this epoch, the light curve dip could be due to the presence of a dark starspot region on the secondary star, with the Hα brightening corresponding to associated chromospheric activity. This behavior suggests that magnetic activity (spots) near the L1 point may indeed be connected to the occurrence of the low states, in agreement with the Livio & Pringle (1994) work. 4.2. High State Spectra In spite of the lack of concurrent photometry we have concluded that ST LMi was likely accreting in the high state during our 2006-May 4-m spectra. ST LMi is often considered a "simple" one-pole accretor, rather similar to VV Pup. There is general agreement in the literature regarding the geometry of the system: the orbital inclination is near 55◦ and the main accreting pole has an inclination of ∼150◦ with respect to the white dwarf spin axis (e.g. Schmidt, Stockman & Grandi 1983; Cropper 1986; Peacock et al. 1992; Ferrario et al. 1993). This geometry leads to a self-eclipse of the main accretion region by the white dwarf, providing good leverage for extracting geometrical properties of this pole. The main accreting pole is the one nearest to the secondary star, having a magnetic field strength of ∼13MG). The derived accretion footprint leads (in the direction of orbital motion) a line joining the stellar centers by ∼0.05-0.15 phase units (Schmidt et al. 1983; Cropper 1988; Howell et al. 2000) being most directly viewed at our phase 0.8. The main accretion region has been shown to have a complex extended pattern on the white dwarf (e.g. Cropper & Horne 1994; Stockman & Schmidt 1996). The second pole has a strength of ∼30 MG and has been -- 9 -- reported to also accrete on occasion (Peacock et al. 1992). Figure 10 presents a representative spectrum from the 2006-May-19 4-m run. Hα is strongly in emission with a complex and variable profile. Furthermore the HeI emission lines at 5876A and 6678A are quite strong, consistent with previous existing high state spectroscopy of the system, and inconsistent with the low state WIYN spectra in which the HeI emission lines are completely missing. The phased high-state optical light curve of ST LMi (Stockman et al. 1983; Cropper 1986; Peacock et al. 1992; also see figure 4) has a bright phase lasting about 0.25 phase units centered near orbital phase 0.8, in which the system becomes ∼1.2 mag brighter in the V-band. The amplitude of this feature becomes larger as one moves from B to V to R, consistent with our measurements. The bright phase corresponds to the time when the primary accreting pole is most directly viewed from the earth. The spectrum of ST LMi is significantly different between the bright orbital phases and the faint orbital phases of the high-state (see below). The high-state optical emission lines are quite complex and have varying interpretations involving infall velocities from an accretion funnel plus other components thought to be produced near the secondary star. (Schmidt, Stockman & Grandi 1983; Bailey et al. 1985). This picture of the origin of the H and He emission lines during the high state bright phase is similar to that developed for other single pole accretors, such as VV Pup. In ST LMi, though, Schmidt et al. (1983) and Bailey et al. (1985) agree, as we discuss below, that some of the emission line components are from the secondary star and/or its vicinity. Stockman et al. (1983) identified two major components in their ST LMi high-state optical emission lines (Hδ, Hγ, HeI4471, HeII 4686 and Hβ): a broad emission component (BEC) which has a sinusoidal velocity curve with K∼500 km s−1 plus a non-sinusoidal narrow emission component (NEC) which is always redshifted with respect to the BEC. They conclude that the BEC emission arises near the primary star, in a cone-like structure having an infall velocity of ∼1000 km/sec. The high velocities imply that the emitting region should be close to the white dwarf, in a curved, organized path that leads to the observed line broadening. Using the same data, Schmidt et al. (1983) constructed a physical model for the accretion stream, and identified the BEC with a "constricted region" approximately mid-way between the two stellar components, where the accretion stream encounters strong magnetic pressure as well as becoming confined by the magnetic field of the white dwarf. The bright phase (0.5-0.95) NEC is attributed to a highly collimated infall of high-density gas, similar to the situation observed in EF Eri in its high state (Schneider & Young 1980; Young et al. 1982). During the faint optical phase (0.0-0.5), the NEC (which dominates the peak of the emission) is thought to arise from the secondary star itself - possibly irradiation on the heated inner hemisphere. However, upon calculating the area that gives rise to this emission, these authors find that it is one order of magnitude less than the area of the irradiated inner hemisphere of the secondary star. This discrepancy is interpreted as being due to limb-darkening effects which were not taken into account in the calculation, and to a degree of self-eclipsing by the accretion funnel. -- 10 -- The high state data of Bailey et al. (1985) found structure in the HeII 4686A and Hβ emission lines similar to that reported in the Stockman et al. (1983) study. However, these authors argue that the profiles are made up of two sinusoids, one of which has variable width with phase. Component A appears narrow and reshifted for the bright part of the orbit, reaching velocities of 600 km/sec, whereas it is faint and broad at other phases. As in the work of Stockman et al. (1983) and Schmidt et al. (1983) this component is attributed to a highly curved inflow region onto the white dwarf. The second component (component B) has a sinusoidal velocity curve with K∼300 km s−1 and was attributed to the secondary star - either its irradiated inner hemisphere or material in its vicinity. Our phase resolved profiles of the Hα, HeI 5876A and HeI 6678A emission lines (figure 11) are complex, having many of the same features as shown in the earlier high-state studies of ST LMi. In particular, we see double-peaked emission lines between phases 0.6 and 0.9, plus an extended blue wing for phases 0.1-0.4. The profiles are difficult to fit with multiple Gaussians because of their asymmetric nature plus the uncertainty regarding the number of components. For those reasons we chose to characterize the radial velocities using two simple methods: 1) the centroid of the full emission line, regardless of the structure, and 2) the velocities at the emission line peaks of one, or at most two, well-defined components. This is similar to the approach of Bailey et al. (1983), whereas Stockman et al. (1983) measured velocities of emission line components using bisectors and centroids. We find no significant differences in the shapes or velocities of the He and Hα emission lines. In figure 12 (top) we show the measured radial velocities of the centroid of the full Hα emission line (open triangles) along with the velocities of the individual peaks. For comparison, we reproduce the velocity curve of the central Hα component from the low-state WIYN data (crosses). The other two panels include the measured RV of the centroid of the full HeI 5876A (middle) and HeI 6678A (bottom). In the top panel of fig. 12 we see that, for about half the orbit (phases 0.0-0.6), the centroid of the full Hα emission line (open triangles) follows the curve established by the 2006-Feb- 21/22 low state data (crosses), as does the radial velocity of the single peak of the line (solid points) over phases 0.0-0.6. However, over the phase interval 0.6-1.0 the centroid deviates sharply redward as do the two distinct emission line components (solid points). The redder component eventually deviates from the low-state curve by nearly 800 km s−1. The radial velocity of the peak of the blue component also deviates somewhat redward from the low-state curve. However, this is likely due to contribution of underlying emission from the redder component. We therefore will proceed under the assumption that the major change that occurs over phases 0.6-1.0 is the appearance of a new emission line component that is redshifted 700-800 km s−1 with respect to the persistant component. Ferrario & Wehrse (1999) have computed model line profiles for polars, finding a narrow, low- velocity component from the magnetically heated coupling region, while a second component arises from inflow near the white dwarf. We think that the line component we observe over phases 0.6-1.0 is indeed due to inflow near the white dwarf, but identifying the "other" line component in our spectra as being from the coupling region does not seem to be possible. The magnetic field in ST -- 11 -- LMi is similar to that in EF Eri, for which the distance of the coupling region above the white dwarf is estimated to be 5RW D (Meggitt & Wickramasinghe 1989). If the coupling region in ST LMi is also ∼5RW D, we find this emission to arise ∼0.13a above the white dwarf, while we find (using the system parameters from Stockman & Schmnidt (1996) that the center of mass is ∼0.29a above the white dwarf. This is well on the white dwarf's side of the center of mass whereas the phasing of the "other" component in our spectra places the origin well on the secondary's side of the center of mass. In figures 13 through 16 we present our high-state spectroscopy of ST LMi from 2006 May. The spectra are shown with assigned phases and each panel has the same identical flux scale. The spectra at phases 0.917, 0.764, and 0.865 correspond to the time of the orbital photometric bump, that is, the time when the accretion pole region is most nearly face-on to the observer (see figure 4 top, for comparison). These spectra have two features not seen at other orbital phases: 1) they have an absorption feature near 6275A and 2) they have a broad continuum hump from ∼6400A redward. The broad absorption feature near 6275A was also reported by Schmidt et al. (1983). Those authors considered several possibilities for the identification: λ6280 TiO from the secondary star, the Hα σ− Zeeman absorption component from the white dwarf photosphere, and the telluric a- band (due to O2). Their preferred identification was the Hα σ− line component from the white dwarf photosphere. Our improved spectral resolution allowed us to make a more careful comparison with the a-band. We used the very high resolution telluric spectrum above Kitt Peak from Hinkle, Wallace & Livingston (2003), convolved with a Gaussian to match our spectral resolution. The match with the a-band is surprisingly good in wavelength, in approximate shape (degraded to the red in the average spectrum of phases 0.917, 0.764, and 0.865), and in approximate strength. Unlike water vapor however, oxygen is uniformly distributed in the earth's atmosphere and could not be modulated with phase to appear only over this small orbital phase range, even by coincidence. Like Schmidt et al. (1983), we too can rule out TiO for a similar reason: the 6275A feature appears only during a small range in orbital phase. Both of these features occur only during the time when the observer has the most direct view of the white dwarf magnetic pole. Given the above, we also conclude that the 6275A absorption feature is due to the Hα Zeeman σ− line. Examination of averaged WIYN low state spectra (figure 17) reveals the same Zeeman absorp- tion component from Hα as we discussed above. The wavelength of the σ− component can be used to calculate the surface magnetic field strength of the WD. Note, this field strength will be different from that estimated from the cyclotron humps as it comes from the surface while the cyclotron emission comes from somewhere in the column above the magnetic pole, probably, in fact a small range of locations. The separation of the Zeeman σ− absorption component from the central π component of Hα is 340A. Eq. 10 in Wickramasinghe & Ferrario (2000) yields a B of 16.8 MG for this Zeeman separation in Hα. Looking at fig. 15 we see a redward hump in the continuum for those data obtained during the -- 12 -- phase interval 0.7-0.9. The location of the hump corresponds to the n=14 harmonic for a B=12 MG field. This harmonic would be weak, if present at all. Cyclotron humps are far more prominent during times of very low mass accretion (e.g., Schmidt et al. 2005). We therefore argue that this redward rise in the continuum slope is not a cyclotron hump but the general cyclotron continuum (the Wien slope) observed during this time of high state mass accretion and seen during the interval of near face-on viewing of the accretion region. The geometry of ST LMi's orbit as well as main accreting pole location allows a unique view of a magnetic CV. The accretion pole is not directly visible for long (or at all) and the binary inclination provides both a near-direct view as well as places the observer in a near perpendicular plane to a normal to the field lines at the pole. Thus, we can observe Zeeman absorption from the near pole white dwarf surface as well as the beamed cyclotron emission from the accretion column simultaneously. The fact that we do not see a direct pole or only for a small time period (phases 0.75-0.85), allows us to study the component stars during both high and low state with little contamination form the accretion flux. 4.2.1. Hyperactivity on CV secondary stars Magnetic activity on the mass-losing secondary star of CVs has been previously inferred from a variety of types of excess emission. Low state x-rays have been attributed to coronal emission from the secondary star in AM Her (de Martino et al. 1998), and ST LMi is one of a few systems for which chromospheric activity has been deduced from structures in the Hα emission line of low- state optical spectra. There are currently only a handful of CVs whose low-state Hα profiles suggest chromospheric activity (hyperactivity). Apart from AM Her (Kafka et al. 2005b, 2006), VV Pup (Mason et al. 2007) and ST LMi, AR UMa (Schmidt et al. 1999) and TT Ari (Shafter et al. 1985) have been reported to show extra components in their low-state Hα emission line. Although these extra components in AR UMa and TT Ari have not been attributed to prominences on the secondary star, the emission line profiles and RV behavior are similar to the ones reported for AM Her and ST LMi. In all cases, the blue/red satellites are present for a portion of the orbit and, with the exception of ST LMi, the red satellite is present during the first part of the orbital cycle and the blue during the second part. In some instances both satellites are present at the same phases (triple-peaked emission lines), and in all cases there is a rapid switchover from red to blue satellite near spectroscopic conjunction. The persistence/duration of the satellites varies from system to system: in AM Her the satellites appear to have remained relatively unchanged for 3 years, the satellites in VV Pup appeared only during a second epoch of spectroscopic observations of the system and in ST LMi the visibility of the blue/red Hα satellites changes from cycle to cycle. Differences in the character and duration of activity is expected for different spectral types of secondaries and different rotation rates (i.e. system orbital periods). It is also worth noting that the systems with reported hyperactivity are polars (diskless MCVs), with the exception of TT Ari. Although TT Ari is classified as a nova-like CV, Jameson et al. (1982) -- 13 -- proposed that the system likely has a magnetic white dwarf therefore it belongs in the DQ Her category. An upper limit of 4MG (Shafter et al. 1985) for the magnetic field of the primary is still within the range of the magnetic field strength of the WDs in DQ Her (intermediate polar) systems. The suspected magnetic nature of TT Ari, along with the known magnetic nature of the other systems, suggests that a combination of fast rotation (all systems have orbital periods less than 4.0 hours) and magnetic field interaction between the two stellar components may be responsible for the presence of the satellites. Activity-induced line emission appears to be concentrated around the inner Lagrangian point, although the observed velocities suggest that the relevant structures are not confined to the orbital plane of the system. Again, an exception to this rule is presented by TT Ari, in which the Hα emission line appears to be stronger at inferior conjunction of the secondary star. Triple-peaked H-alpha emission lines are not the only characteristic of stellar activity in CV secondaries: EF Eri, after spending seven years in a low state, and without changing its optical brightness, began showing weak and narrow Balmer lines and CaII H & K emission, originating on or near the secondary star (Howell et al. 2006). Although there are indications that the EF Eri Hα emission line shows additional components, the S/N of the data was not adequate to reach secure conclusions about the structure of the emission line. On the other hand low states in disk systems are more rare than low states in MCVs. Chromospheric activity and activity cycles have been considered responsible for quasi-periodic modulations in the long-term optical light curves of dwarf nova in quiescence. Furthermore, variations (positive and negative) in the timings of eclipse minima in disk systems have been explained in terms of magnetic activity on the secondary (Baptista et al. 2003). Perhaps the most "direct" observation of stellar activity in a disk CV is provided by Steeghs et al. (1996) who described a peculiar stationary low velocity component in several emission lines of the quiescence spectra of the dwarf novae IP Peg and SS Cyg. The authors attributed this component to slingshot prominences (Collier Cameron & Robinson 1989), co-rotating with the secondary star and kept in the vicinity of the L1 point by the binary potential. Taken at face value, this implies that the secondary stars of all CVs should be chromospherically active. Since the work of Parker (1961), it is known that a tachocline (convective zone - radiative core boundary) is necessary for the formation of a large scale, αΩ (dipole) field, giving rise to all our familiar (from the sun) activity phenomena. Rotation can amplify the field giving rise to large-scale magnetic structures. Furthermore, magnetic field interactions may increase the lifetime of such structures. So far, the systems that show hyperactivity have secondary stars of spectral type M3 or later. Near that spectral type, main sequence stars become fully convective likely leading to a change in the nature of their magnetic field structure, because the formation of an αΩ-induced magnetic field is no longer possible. For fully convective stars, chromospheric emission may be generated by a small scale, local turbulent magnetic field (Durney et al. 1993). Although this might be an alternative to the αΩ dynamo mechanism, it is not expected to produce large-scale features such as large starspots, or a dominant dipole field for magnetic braking. Other mechanisms might allow a fully convective star to produce a large scale field, such as an α2 dynamo (Chabrier and Kuker 2006). Fast rotation can amplify this field, which can easily reach the order of several kG, similar -- 14 -- to the one that is observed in single active late M dwarfs, and similar to the one necessary for the observed activity on CV secondaries. The He I emission line ratio is often used as an indicator of the physical conditions in the chromospheres of solar-type stars. It can be shown (Howell et al. 2006) that the ratio, EW[HeI(5876A)]/EW[HeI(6678A)] can be as high as 45 in the quiet sun but is near 3 in active prominences. Using our 4-m high-state ST LMi spectra, we find that the ratio of EW[HeI(5876A)]/EW[HeI(6678A)] emission lines is ∼2.5-3.0, with the exception of phases 0.7-0.9 when the accretion pole likely con- tributes to the line emission. This line ratio value is that expected for He I emission caused by collisional excitation in an active chromosphere hotter than 8000K. 5. Final Remarks We have presented a spectroscopic and photometric study of ST LMi during its 2005-2006 high and low states. Our main results can be summarized as follows: • In the low state, ST LMi remains at V∼17.5 mag, with an occasional "hump" in the optical light curve, between phases 0.5 and 0.95. This is the same phase interval during which the "bright" phase in the high state occurs due to the visibility of the accretion pole. Therefore episodic low-level accretion apparently takes place in ST LMi during the low state. • The low-state spectra of the system are dominated by strong TiO bands from the secondary star. The only emission line is Hα and it appears to have structure similar that of AM Her (Kafka et al. 2005b, 2006) and VV Pup (Mason et al. 2007) in their low states. At times when accretion seems to have ceased completely (according to the optical light curves), the RVs of all components of Hα originate from the secondary's side of the center of mass. Concurrent photometry supports the lack of accretion and the presence of a large cool region (spot) near the L1 point (between phases 0.3-0.6). • A photometric state ∼ 0.7 mag fainter than the usual low state magnitude was observed. This is the second CV which displays such a state in its optical light curve. It is not clear whether this was a more complete cessation of mass transfer, or some other change in the system. It will be important to study such events if we to understand the properties of a CV in which accretion has truly stopped. • In the high state, the Hα line presents a complicated multi-component profile. For orbital phases between 0.0 and 0.5, the RV of the Hα emission line nearly coincides with the RVs of the low state Hα emission. This line component, which is phased with the motion of the secondary star, cannot be associated with accretion because it is present in both the high and low state. It is apparently not due to irradiation of the inner hemisphere of the secondary star because it is not brightest when the inner hemisphere most nearly faces the observer. It -- 15 -- may be due to activity on the secondary star, although the geometry of the emission region is far from clear. • In the high state a second emission line component of Hα appears near phase 0.8 which is highly red-shifted, likely due to infall onto the white dwarf. We confirm the results of Bailey et al. (1983) that a broad blue wing appears on the Hα profile at phases opposite the appearance of the highly red-shifted component. Although CV secondaries are the fastest rotating lower main sequence stars known, the large- scale activity signatures we have deduced are not expected for a fully convective star. It will be necessary to obtain suitable spectroscopy of a non-magnetic (disk) CV in the low state to test if the presence of Hα satellites are tied to magnetic field interactions of the two stars, or if ultra-fast rotation along suffice for their presence. We would like to thank our anonymous referee for the careful review of the manuscript. Bailey, J., et al. 1985, MNRAS, 215, 179 REFERENCES Baptista, R., Borges, B. W., Bond, H. E., Jablonski, F., Steiner, J. E., & Grauer, A. D. 2003, MNRAS, 345, 889 Bianchini, A. 1988, Informational Bulletin on Variable Stars, 3136, 1 Chabrier, G., Kumlker, M. 2006, A&A, 446, 1027 Collier Cameron, A., & Robinson, R. D. 1989, MNRAS, 238, 657 Cropper, M. 1986, MNRAS, 222, 853 Cropper, M., & Horne, K. 1994, MNRAS, 267, 481 de Martino, D., et al. 1998, A&A, 333, L31 Durney, B. R., De Young, D. S., & Roxburgh, I. W. 1993, Sol. Phys., 145, 207 Ferrario, L., Bailey, J., & Wickramasinghe, D. T. 1993, MNRAS, 262, 285 Ferrario, L., & Wehrse, R. 1999, MNRAS, 310, 189 Hinkle, K. H., Wallace, L., & Livingston, W. 2003, Bulletin of the American Astronomical Society, 35, 1260 Henden, A. A., & Honeycutt, R. K. 1995, PASP, 107, 324 Howell, et al., 2006, ApJ, 652, 709. Howell, S. B., Ciardi, D. R., Dhillon, V. S., & Skidmore, W. 2000, ApJ, 530, 904 -- 16 -- Jameson, R. F., Sherrington, M. R., King, A. R., & Frank, J. 1982, Nature, 300, 152 Kafka, S., Robertson, J., Honeycutt, R. K., & Howell, S. B. 2005, AJ, 129, 2411 (2005a) Kafka, S., Honeycutt, R. K., Howell, S. B., & Harrison, T. E. 2005, AJ, 130, 2852 (2005b) Kafka, S., Honeycutt, R. K., & Howell, S. B. 2006, AJ, 131, 2673 Kafka, S., & Honeycutt, R. K. 2005, AJ, 130, 742 [KH05] King, A. R., & Cannizzo, J. K. 1998, ApJ, 499, 348 Livio, M., & Pringle, J. E. 1994, ApJ, 427, 956 Linnell, A.P., Szkody, P., Gansicke, B., Long, K. S., Sion, E. M., Hoard, D. W., & Hubeny,I. 2005, ApJ, 624, 923 Mason et al. 2007, A&A in press Meggitt, S. M. A., & Wickramasinghe, D. T. 1989, MNRAS, 236, 31 Peacock, T., Cropper, M., Bailey, J., Hough, J. H., & Wickramasinghe, D. T. 1992, MNRAS, 259, 583 Shafter, A. W., Szkody, P., Liebert, J., Penning, W. R., Bond, H. E., & Grauer, A. D. 1985, ApJ, 290, 707 Schmidt, G. D., Stockman, H. S., & Grandi, S. A. 1983, ApJ, 271, 735 Schmidt, G. D., Hoard, D. W., Szkody, P., Melia, F., Honeycutt, R. K., & Wagner, R. M. 1999, ApJ, 525, 407 Schmidt, G. D., et al. 2005, ApJ, 630, 1037 Schneider, D. P., & Young, P. 1980, ApJ, 238, 946 Steeghs, D., Horne, K., Marsh, T. R., & Donati, J. F. 1996, MNRAS, 281, 626 Stockman, H. S., Foltz, C. B., Schmidt, G. D., & Tapia, S. 1983, ApJ, 271, 725 Stockman, H. S., & Schmidt, G. D. 1996, ApJ, 468, 883 Warner, B. 1995, Cambridge Astrophysics Series, Cambridge, New York: Cambridge University Press, -- c1995, Wickramasinghe, D. T., & Ferrario, L. 2000, PASP, 112, 873 Young, P., Schneider, D. P., Sargent, W. L. W., & Boksenberg, A. 1982, ApJ, 252, 269 This preprint was prepared with the AAS LATEX macros v5.2. -- 17 -- Table 1. Log of Observations of ST LMi UT Date Seq. No. Telescope/Instrument No. of Exposures/Filter Exp time(s) A1 A2 A3 A4 A5 B1 B2 B3 C1 D1 PHOTOMETRY: 1991-01 to 2005-03 2005-02-28 2005-03-01 2005-03-02 2005-03-03 2005-03-04 2005-06-13 2005-06-17 2005-06-20 2005-12-23 2005-12-24 2006-02-12 2006-02-21 2006-02-22 SPECTROSCOPY: 2005-06-17 2005-12-31 2006-02-21 2006-02-22 2006-05-19 RoboScope 0.4-m WIYN 0.9-m/S2KB WIYN 0.9-m/S2KB WIYN 0.9-m/S2KB WIYN 0.9-m/S2KB WIYN 0.9-m/S2KB WIYN 0.9-m/S2KB WIYN 0.9-m/S2KB WIYN 0.9-m/S2KB Tenagra 0.8-m Tenagra 0.8-m WIYN 0.9-m/S2KB WIYN 0.9-m/S2KB WIYN 0.9-m/S2KB WIYN 3.5-m/Hydra KPNO 4-m/RC Spec WIYN 3.5-m/Hydra WIYN 3.5-m/Hydra KPNO 4-m/RC Spec 551/V 44/V 35/R 42/V 50/V 33/V 11/V 12/V 13/V 1/V 1/V 23/V 6/V 68/V 6 2 12 31 16 240 120 90 90 90 90 180 120 120 200 200 120 180 180 600 - 1800 1200 600 600 300-600 -- 18 -- 14 15 16 17 18 14 15 16 17 18 1000 2000 3000 JD - 2448600 4000 5000 WIYN 3.5-m WIYN 3.5-m 4-m 4-m A1-5 B1-3 4800 5000 JD - 2448600 D1 C1 5200 Fig. 1. -- Top: V-band light curve of ST LMi 1990-2005 from various sources, mostly RoboScope. The dates of our spectroscopy are noted as vertical dashed lines. Bottom: Expanded view of the latter portion of the top panel, with the spectroscopic observing runs noted. The labeled photometric sequences correspond to the designations in figure 2 and Table 1. -- 19 -- 16 16.5 17 17.5 18 18.5 16 16.5 17 17.5 18 18.5 13 June 2005 (UT) B1 17 June 2005 (UT) B2 20 June 2005 (UT) B3 12 Feb 2006 (UT) C1 16 16.5 17 17.5 18 18.5 16 16.5 17 17.5 18 18.5 16 16.5 17 17.5 18 18.5 16 16.5 17 17.5 18 18.5 16 16.5 17 17.5 18 18.5 16 16.5 17 17.5 18 18.5 28 Feb 2005 (UT) A1 2 March 2005 (UT) A3 3 March 2005 (UT) A4 4 March 2005 (UT) A5 0 0.5 1 1.5 0 0.5 1 1.5 Phase Phase Fig. 2. -- Left: Phased V-band light curves of ST LMi from 4 nights of observing run A, as marked in figure 1 (bottom). Note the maxima near phase 0.8. Filled symbols are data from the second of two adjacent orbital cycles. Right: Phased V-band light curves of ST LMi from 3 nights of observing run B and 1 night of observing run C, as marked in figure 1 (bottom). Note the absence of a maxima near phase 0.8 -- 20 -- 22 Feb 2006 (UT) D1 16 16.5 17 17.5 18 18.5 0 1 March 2005 (UT) A2 0 0.5 0.5 1 1.5 1 Phase 10 May 2006 (UT) 1.5 6 7 8 9 0 0.5 1 1.5 Phase Fig. 3. -- Left: Phased light curve of ST LMi from observing run D, as marked in figure 1 (bottom). Note the weakness of the maximum near phase 0.8 compared to figure 2 (left), and the near absence of scatter in the light curve compared to figure 2 and figure 4 (bottom). Filled symbols are data from the second of two adjacent orbital cycles. Right: Phased R-band light curves of ST LMi. Top is a differential R-band light curve from run A2, as marked in figure 1 (bottom). Bottom is a synthetic light curve extracted from the 4-m spectra of 2006-May-19, with an effective wavelength approximately midway between V and R. Note the large amplitude of the maximum near phase 0.8 in both plots. 14 15 16 17 18 -- 21 -- Phase Fig. 4. -- Phased V-band light curves of ST LMi from RoboScope data 1990-2005. The data consist of single point each clear night, at a random orbital phase. Top is the high state light curve showing a maximum near phase 0.8. Bottom is the low-state light curve, perhaps showing a weak maximum near phase 0.8. The occasional brightward excursions in the bottom panel are likely artifacts due to cosmic rays. -- 22 -- June 2005 Phase: 0.727 Dec 2005 Phase: 0.867 Feb 2006 Phase: 0.519 500 400 300 200 100 0 800 600 400 200 300 200 100 0 6000 6500 7000 7500 8000 Fig. 5. -- Representative low-state spectra of ST LMi from 2005-Jun (top), 2005-Dec (middle) and 2006-Feb (bottom). No corrections for atmospheric absorption have been made. The middle panel shows sharp features due to incomplete cancellation of night sky emission lines. The inset plots show the Hα emission line on an expanded scale. -- 23 -- 0.468 0.581 0.682 0.796 2006-Feb-22 0.293 0.407 0.508 0.622 2006-Feb-21 0.519 0.633 0.772 0.886 0.000 0.113 0.227 0.3411 0.736 0.837 0.950 0.052 0.165 0.279 0.380 0.494 6500 6600 6500 6600 6500 6600 Fig. 6. -- Low-state Hα profiles of ST LMi for orbital phases 0.5192 to 0.7963 on 2006-Feb-21 UT (first night) and for phases 0.2930 to 0.4942 on 2006-Feb-22 UT (second night)). Successive phases run down the page, starting in the upper left corner. -- 24 -- 0.608 0.709 0.823 0.924 0.165 0.265 0.379 0.480 0.594 0.708 0.809 0.555 0.656 0.770 0.923 6500 6600 0.125 0.226 0.340 0.441 6500 6600 6500 6600 Fig. 7. -- Continuation of previous figure (for night 2). -- 25 -- 400 200 0 -200 -400 0 0.5 1 1.5 Phase Fig. 8. -- The RVs of the central Hα emission line component in ST LMi for the low-state WIYN spectra of 2005-June-17 (squares), the 4-m spectra of 2005-Dec-31 (triangles), and the low-state WIYN spectra of 2006-Feb-21/22 (solid points). Open circles represent the RVs of the occasional satellite components in the WIYN 2006-Feb spectra. Colors represent points of the same orbit. -- 26 -- 21 & 22 Feb 2006 (UT) 22 Feb 2006 (UT) D1 0.5 1 1.5 Phase -80 -60 -40 -20 0 17.4 17.45 17.5 17.55 17.6 17.65 0 Fig. 9. -- Top: Total EW of the low-state WIYN Hα emission line vs orbital phase; Bottom: The concurrent phase light curve. -- 27 -- HeI NaD HeI Fig. 10. -- A representative high-state spectrum of ST LMi from the 2006-May-19 (UT) 4-m run. The inset plot is an expanded view of the Hα profile. -- 28 -- Phase 0.917 0.018 0.107 0.208 0.297 0.385 0.474 0.587 5 8 6 4 8 6 4 6 4 2 6 4 2 10 5 10 5 10 5 4 2 6 4 2 6 4 2 10 5 10 5 10 5 10 5 10 5 Phase 0.676 0.764 0.865 0.954 0.042 0.143 0.232 0.320 5800 5900 6500 6600 6650 6700 6750 5800 5900 6500 6600 6650 6700 6750 Fig. 11. -- Line profiles of HeI 5876A (left column), Hα (center column) and HeI 6678A (right column) as a function of orbital phase, for the high-state spectra of 2006-May-19 2006 UT. The NaD emission lines (from Tucson airglow) are often visible near and sometimes within the red wing of HeI 5876A. -- 29 -- 500 0 -500 500 0 -500 500 0 -500 HeI 5876 HeI 6678 0 0.5 1 1.5 Phase Fig. 12. -- High-state (2006-May-19 UT) radial velocity curves of ST LMi for Hα (top), and two HeI emission lines (middle and bottom). The open triangles are velocities of the centroid of Hα. The filled circles are velocities at the peaks of the most distinct one or two emission line components, as seen in the line profiles in figure 10. For comparison, the crosses in the top panel are the ST LMi Hα velocities for the low-state WIYN data of 2006-Feb-21/22 UT. -- 30 -- Phase 0.917 0.018 0.107 0.208 6 4 2 0 6 4 2 0 6 4 2 0 6 4 2 0 5800 6000 6200 6400 6600 6800 Fig. 13. -- The complete spectral region of ST LMi as a function of orbital phase for the high-state 2006-May-19 (UT) data. We are using the same scale for all spectra, to emphasize the continuum and the weaker spectral features. (See text for discussion.) -- 31 -- Phase 0.297 0.385 0.474 0.587 6 4 2 0 6 4 2 0 6 4 2 0 6 4 2 0 5800 6000 6200 6400 6600 6800 Fig. 14. -- The complete spectral region of ST LMi as a function of orbital phase for the high-state 2006-May-19 (UT) data (cont). -- 32 -- Phase 0.676 0.764 0.865 0.954 6 4 2 0 6 4 2 0 6 4 2 0 6 4 2 0 5800 6000 6200 6400 6600 6800 Fig. 15. -- The complete spectral region of ST LMi as a function of orbital phase for the high-state 2006-May-19 (UT) data (cont) -- 33 -- Phase 0.042 0.143 0.232 0.320 6 4 2 0 6 4 2 0 6 4 2 0 6 4 2 0 5800 6000 6200 6400 6600 6800 Fig. 16. -- The complete spectral region of ST LMi as a function of orbital phase for the high-state 2006-May-19 (UT) data (cont) -- 34 -- Phase bin 0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 400 300 200 100 400 300 200 100 400 300 200 100 400 300 200 100 400 300 200 100 400 300 200 100 400 300 200 100 400 300 200 100 400 300 200 100 400 300 200 100 6000 6500 7000 7500 8000 6000 6500 7000 7500 Fig. 17. -- WIYN low-state spectra from 2006-Feb-21/22 UT, averaged in phase bins of 0.1. The narrow weak emission lines, mostly redward of 7200A, are artifacts due to incomplete cancellation of OH features in the night sky from the fiber spectra. (Also see text for comments.)
astro-ph/0703586
4
0703
2010-02-17T14:47:56
Light Propagation and Large-Scale Inhomogeneities
[ "astro-ph", "gr-qc", "hep-ph", "hep-th" ]
We consider the effect on the propagation of light of inhomogeneities with sizes of order 10 Mpc or larger. The Universe is approximated through a variation of the Swiss-cheese model. The spherical inhomogeneities are void-like, with central underdensities surrounded by compensating overdense shells. We study the propagation of light in this background, assuming that the source and the observer occupy random positions, so that each beam travels through several inhomogeneities at random angles. The distribution of luminosity distances for sources with the same redshift is asymmetric, with a peak at a value larger than the average one. The width of the distribution and the location of the maximum increase with increasing redshift and length scale of the inhomogeneities. We compute the induced dispersion and bias on cosmological parameters derived from the supernova data. They are too small to explain the perceived acceleration without dark energy, even when the length scale of the inhomogeneities is comparable to the horizon distance. Moreover, the dispersion and bias induced by gravitational lensing at the scales of galaxies or clusters of galaxies are larger by at least an order of magnitude.
astro-ph
astro-ph
Light Propagation and Large-Scale Inhomogeneities Nikolaos Brouzakis, Nikolaos Tetradis and Eleftheria Tzavara University of Athens, Department of Physics, University Campus, Zographou 157 84, Athens, Greece Abstract. We consider the effect on the propagation of light of inhomogeneities with sizes of order 10 Mpc or larger. The Universe is approximated through a variation of the Swiss-cheese model. The spherical inhomogeneities are void-like, with central underdensities surrounded by compensating overdense shells. We study the propagation of light in this background, assuming that the source and the observer occupy random positions, so that each beam travels through several inhomogeneities at random angles. The distribution of luminosity distances for sources with the same redshift is asymmetric, with a peak at a value larger than the average one. The width of the distribution and the location of the maximum increase with increasing redshift and length scale of the inhomogeneities. We compute the induced dispersion and bias on cosmological parameters derived from the supernova data. They are too small to explain the perceived acceleration without dark energy, even when the length scale of the inhomogeneities is comparable to the horizon distance. Moreover, the dispersion and bias induced by gravitational lensing at the scales of galaxies or clusters of galaxies are larger by at least an order of magnitude. 0 1 0 2 b e F 7 1 4 v 6 8 5 3 0 7 0 / h p - o r t s a : v i X r a Light Propagation and Large-Scale Inhomogeneities 2 1. Introduction The observed deviation from homogeneity of the structure of the Universe at small length scales poses the question of whether the use of the Friedmann-Robertson-Walker (FRW) metric is adequate for the discussion of the cosmological expansion. The argument for the applicability of a homogeneous solution is based on the observation that the matter distribution is homogeneous when averaged over length scales of O(100) h−1 Mpc. On the other hand, the conclusion drawn from observations of distant supernova [1, 2], or the perturbations of the cosmic microwave background (CMB) [3], that the recent cosmological expansion is accelerating has placed the effect of structure formation on the overall expansion under scrutiny. We are interested in the influence of inhomogeneities with sub-horizon characteristic scales on the perceived expansion [4, 5, 6]. The observed inhomogeneities in the matter distribution with length scales of O(10) h−1 Mpc or smaller are large, so that the Universe cannot be approximated as homogeneous at these scales. It is important to have a quantitative estimate of the influence of these inhomogeneities on the data used for the determination of the expansion rate. As all the observations involve the detection of light signals, it is crucial to understand the effect of the inhomogeneous background on light propagation. The transmission of a light beam in a general gravitational background can be studied through the Sachs optical equations [7]. These describe the expansion and shear of the beam along its null trajectory. Apart from the case of a FRW background, the optical equations have been derived by Kantowski for a Scharzschild background [8]. He used them for the study of light transmission within the Swiss-cheese model of the Universe [9]. In this model the light propagates essentially in empty space with a beam expansion larger than the average. In rare instances it passes near a very dense clump of matter, which produces significant shear and focusing of the beam [10]. Because of the randomness of such events, sources with the same intrinsic luminosity and redshift may have different luminosity distances. The distribution is peaked at a value of the luminosity distance larger than the one in a homogeneous background [11, 12], even though the mean value remains unaffected. If the sample of observed sources is small, it unlikely that a clump of matter will be encountered during the beam propagation. The beams essentially propagate in empty space, while the expansion is induced by the average energy density. A phenomenological equation has been proposed by Dyer and Roeder in order to describe the case in which a fraction of the energy density is homogeneously distributed while the remaining is in clumps that do not affect the light propagation [13]. The picture of the Universe we described above is applicable to length scales of O(1) h−1 Mpc or smaller, for which the dominant structures are galaxies or clusters of galaxies. At larger distances the averaged matter distribution has a smaller density Light Propagation and Large-Scale Inhomogeneities 3 contrast. The use of the Schwarzschild geometry for the description of the inhomogeneities is not appropriate. The Lemaitre-Tolman-Bondi (LTB) metric [14] has been employed often for the modelling of the Universe at scales of O(10) h−1 Mpc or larger [15] -- [24]. Its use demonstrates that inhomogeneities can induce deviations of the luminosity distance from its value in a homogeneous background. For example, it has been observed that any form of the luminosity distance as a function of redshift can be reproduced with the LTB metric [15]. This means that the supernova data can be explained through an inhomogeneous matter distribution in the context of this metric. However, reproducing the data requires a variation of the density or the expansion rate over distances of O(100) h−1 Mpc or even larger [20] -- [24]. In order to avoid a conflict with the isotropy of the CMB the location of the observer must be within a distance of O(10) h−1 Mpc from the center of the spherical configuration described by the LTB metric. In this sense, the explanation of the perceived acceleration relies on the position of the observer. On the other hand, there are indications for the presence of a very large void in our vicinity [25, 26]. Our work is based on the fundamental assumption that we do not occupy a special position in the Universe. We are interested in determining the maximum effect that large-scale inhomogeneities can have on the luminosity distance. The density contrast of inhomogeneities with characteristics lengths of O(10) h−1 Mpc or larger is at most of O(1). For this reason it seems unlikely that they can affect the propagation of light more strongly than the inhomogeneities with lengths of O(1) h−1 Mpc or smaller, whose density contrast can be larger than 1 by several orders of magnitude. On the other hand, the validity of the FRW metric is questionable if inhomogeneities with lengths comparable to the horizon distance ∼ 3 × 103 h−1 Mpc develop a density contrast of O(1). We shall allow for such a possibility in order to obtain a quantitative estimate of the effect on the luminosity distance. In a previous publication [27] we derived the optical equations for a general LTB background [14]. We used them in order to study light propagation in a variant of the Swiss-cheese model. The inhomogeneities are modelled as spherical regions within which the geometry is described by the LTB metric. At the boundary of these regions the LTB metric is matched with the FRW metric that describes the evolution in the region between the inhomogeneities. The Universe consists of collapsing or expanding inhomogeneous regions, while a common scale factor exists that describes the expansion of the homogeneous intermediate regions. This model is similar to the standard Swiss- cheese model [9], with the replacement of the Schwarzschild metric with the LTB one. For this reason we refer to it as the LTB Swiss-cheese model. We focus on the effect of inhomogeneities on the luminosity distance if the source and the observer do not occupy special positions in the Universe. This is achieved by placing both the source and the observer within the homogeneous region of the LTB Swiss-cheese Light Propagation and Large-Scale Inhomogeneities 4 model. During its path the light signal crosses several inhomogeneous regions before reaching the observer. As we have mentioned, similar studies [11, 12] have discussed the influence of structures such as galaxies or clusters of galaxies on light propagation. The beam shear plays an important role in this effect, characterized as gravitational lensing [28]. We focus on inhomogeneities of much larger length scale, of O(10) h−1 Mpc or larger. The matter distribution, even though inhomogeneous, is more evenly distributed than in the previous case. In particular, we assume that each spherical region has a central underdensity surrounded by an overdense shell. The densities in the two regions are comparable at early times and differ by a factor O(1) during the later stages of the evolution. The LTB Swiss-cheese model we construct in this way describes a Universe dominated by spherical voids with the compensating matter concentrated in shells surrounding them. The beam shear is negligible in our calculations, and the main effect arises from the variations of the beam expansion because of the inhomogeneities. In ref. [27] we estimated the deviations of the luminosity distance from its value in a homogeneous Universe by considering the extreme case in which the light passes through the centers of all the inhomogeneities it encounters. In this work we perform a more detailed statistical analysis, by considering light beams with random impact parameters relative to the centers. We check that the resulting luminosity distance as a function of the impact parameter is consistent with flux conservation. For a given redshift we estimate the width of the distribution of luminosity distances. We also determine the deviation of the maximum of the distribution from the value in a homogeneous background. For a central underdensity this value is positive and sets the scale for the perceived increase in the expansion rate relative to a homogeneous Universe. An analytical study with similarities with our work is described in ref. [29]. It focuses, however, on the study of the redshift, which determines only partially the luminosity distance. No statistical analysis, which is the central point of our work, is performed either. In the following section we summarize the geodesic and optical equations in a LTB background. In section 3 we describe the cosmological evolution of the inhomogeneous regions. In section 4 we study, both numerically and analytically, the effect of the inhomogeneity on the properties (redshift,beam area) of a light beam that travels through it. In section 5 we study the luminosity distance as a function of the angle at which the beam crosses the inhomogeneity. We verify that the results are consistent with flux conservation. In section 6 we consider multiple crossings of inhomogeneities by the light beam and the effect on the luminosity distance. We calculate the distribution of the deviations of the luminosity distance from its value in a homogeneous background. In section 7 we estimate the effect on the determination of cosmological parameters from supernova data. Light Propagation and Large-Scale Inhomogeneities 5 2. Luminosity distance in a Lemaitre-Tolman-Bondi (LTB) background Under the assumption of spherical symmetry, the most general metric for a pressureless, inhomogeneous fluid is the LTB metric [14]. It can be written in the form ds2 = −dt2 + b2(t, r)dr2 + R2(t, r)dΩ2, where dΩ2 is the metric on a two-sphere. The function b(t, r) is given by b2(t, r) = R′2(t, r) 1 + f (r) , (2.1) (2.2) where the prime denotes differentiation with respect to r, and f (r) is an arbitrary function. The bulk energy momentum tensor has the form T A B = diag (−ρ(t, r), 0, 0, 0) . (2.3) The fluid consists of successive shells marked by r, whose local density ρ is time-dependent. The function R(t, r) describes the location of the shell marked by r at the time t. Through an appropriate rescaling it can be chosen to satisfy R(0, r) = r. The Einstein equations reduce to 1 8πM 2 M(r) R2(t, r) = R M′(r) = 4πR2ρ R′, + f (r) (2.4) (2.5) where the dot denotes differentiation with respect to t, and G = (16πM 2)−1. The generalized mass function M(r) of the fluid can be chosen arbitrarily. Because of energy conservation M(r) is independent of t, while ρ and R depend on both t and r. Without loss of generality we consider geodesic null curves on the plane with θ = π/2. The first geodesic equation is dk0 dλ + R′R′ 1 + f (cid:16)k1(cid:17)2 + RR(cid:16)k3(cid:17)2 with ki = dxi/dλ and λ an affine parameter along the null beam trajectory. The second geodesic equation can be replaced by the null condition = 0, (2.6) −(cid:16)k0(cid:17)2 + R′2 1 + f (cid:16)k1(cid:17)2 + R2(cid:16)k3(cid:17)2 while the third one can be integrated to obtain = 0, k3 = cφ R2 . The equation for the beam area can be written as [27] d2√A dλ2 = − 1 √A 1 4M 2 ρ(cid:16)k0(cid:17)2 − σ2. (2.7) (2.8) (2.9) The shear σ is generated by inhomogeneities, for which the local energy density is different from the average one [27]. It describes the deformations of the cross-section of the beam, Light Propagation and Large-Scale Inhomogeneities 6 induced by the propagation within an inhomogeneous medium. The shear is important when the beam passes near regions in which the density exceeds the average one by several orders of magnitude. Within our modelling of large-scale structure, applicable for scales above O(10) h−1 Mpc, the average density contrast is not sufficiently large for the shear to become important. (We have verified this conclusion numerically.) For this reason we neglect it in our study. The Friedmann-Robertson-Walker (FRW) metric is a special case of the LTB metric with R(t, r) = a(t)r ρ = cρ a3(t) f (r) = cr2, M(r) = 4π 3 c = 0,±1 cρr3. The geodesic equation (2.6) has the solution k0 = ct a(t) . The solution of eq. (2.9) for an outgoing beam is A(λ) = r2(λ) a2(t(λ)) Ωs. (2.10) (2.11) (2.12) (2.13) If we normalize the scale factor so that a(ts) = 1 at the time of the beam emission, we recover the standard expression A = r2Ωs in flat space-time. The constant Ωs can be identified with the solid angle spanned by a certain beam when the light is emitted by a point-like isotropic source. We are interested in light propagation in more general backgrounds. We assume that the light emission near the source is not affected by the large-scale geometry. By choosing an affine parameter that is locally λ = t in the vicinity of the source, we can set This expression, along with = qΩs. d√A dλ (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)λ=0 √A(cid:12)(cid:12)(cid:12)λ=0 = 0, (2.14) (2.15) provide the initial conditions for the solution of eq. (2.9). In order to define the luminosity distance, we consider photons emitted within a solid angle Ωs by an isotropic source with luminosity L. These photons are detected by an observer for whom the light beam has a cross-section Ao. The redshift factor is 1 + z = ωs ωo = k0 s k0 o , (2.16) because the frequencies measured at the source and at the observation point are proportional to the values of k0 at these points. The energy flux fo measured by the observer is fo = L 4πD2 L = L 4π Ωs (1 + z)2Ao . (2.17) Light Propagation and Large-Scale Inhomogeneities 7 The above expression allows the determination of the luminosity distance DL as a function (2.9), with initial of the redshift z. The beam area can be calculated by solving eq. conditions given by eqs. (2.14), (2.15), while the redshift is given by eq. (2.16). 3. Modelling the inhomogeneities We study the effect of inhomogeneities on the luminosity distance without assuming a preferred location of the observer. We model the inhomogeneities as spherical regions within which the geometry is described by the LTB metric. At the boundary of these regions, the LTB metric is matched with the FRW metric that describes the expansion of the homogeneous intermediate regions. Our model is similar to the standard Swiss-cheese model [9], with the replacement of the Schwarzschild metric with the LTB one. For this reason we refer to it as the LTB Swiss-cheese model. The choice of the two arbitrary functions M(r) and f (r) in eq. (2.1) can lead to different physical situations. The mass function M(r) is related to the initial matter (2.4). We distribution. The function f (r) defines an effective curvature term in eq. work in a gauge in which R(0, r) = r. We parametrize the initial energy density as ρi(r) = (1 + ǫ(r)) ρ0,i, with ρi(r) = ρ(0, r) and ǫ(r) < 1. The initial energy density of the homogeneous background is ρ0,i. If the size of the inhomogeneity is r0, a consistent solution requires 4πR r0 (3.1) 0 r2ǫ(r)dr = 0, so that 4π 3 r2ρ(r) dr = r3 0ρ0,i. M(r0) = 4πZ r0 0 This is obvious if we apply eqs. (2.4), (2.5) to the homogeneous part at r > r0. The FRW metric is a special case of the LTB metric, described by eqs. (2.10), (2.11). Consistency of the equations as the shell with r = r0 is approached from both sides imposes the condition (3.1). Moreover, the absence of singularities requires the continuity of f (r) and f ′(r). As we assume that the homogeneous part is flat, this means that f (r0) = f ′(r0) = 0. Alternatively, one may consider the matching of the two metrics at the surface with r = r0 employing junction conditions [30]. If this surface does not contain a singular energy density, the above constraints must be imposed [29, 31]. We assume that at the initial time ti = 0 the expansion rate Hi = R/R = R′/R′ is i = ρ0,i/(6M 2). given for all r by the standard expression in homogeneous cosmology: H 2 Then, eq. (2.4) with R(0, r) = r implies that f (r) = ρ0,i 6M 2 r2 1 − 4πr3ρ0,i! . 3M(r) The spatial curvature of the LTB geometry is (3)R(r, t) = −2 (f R)′ R2R′ . (3.2) (3.3) Light Propagation and Large-Scale Inhomogeneities For our choice of f (r) we find that at the initial time (3)R(r, 0) = −6H 2 i 1 − M′ 4πr2ρ0,i! = −6H 2 i 1 − 8 (3.4) ρi(r) ρ0,i ! . Overdense regions have positive spatial curvature, while underdense ones negative curvature. This is very similar to the initial condition considered in the model of spherical collapse [32]. When the inhomogeneity is denser near the center, we have f (r) < 0 for r < r0 and f (r) = 0 for r ≥ r0. It is then clear from eq. (2.4) that, in an expanding Universe, the central region will have R = 0 at some point in its evolution and will stop expanding. Subsequently, it will reverse its motion and start collapsing. The opposite happens if the inhomogeneity has a central underdensity. In this case, the central region expands faster than the surrounding denser spherical shell. The width of the shell decreases, while its density increases. It is the latter configuration that is relevant if we want to model the Universe as being composed mainly of voids separated by thin dense regions. Our expressions simplify if we switch to dimensionless variables. We define ¯t = tHi, ¯r = r/r0, ¯R = R/r0, where H 2 i = ρ0,i/(6M 2) is the initial homogeneous expansion rate and r0 gives the size of the inhomogeneity in comoving coordinates. The evolution equation becomes 2 ¯R ¯R2 = with ¯M = M/(ρ0,ir3 a derivative with respect to ¯t. 3 ¯M(¯r) 4π ¯R3 + 0) and ¯f = 6M 2f /(ρ0,ir2 ¯f (¯r) ¯R2 , 0) = f / ¯H 2 i , ¯Hi = Hir0. The dot now denotes (3.5) We take the affine parameter λ to have the dimension of time and we define the dimensionless variables ¯λ = Hiλ, ¯k0 = k0, ¯k1 = k1/ ¯Hi, ¯k3 = r0k3. The geodesic equations (2.6) -- (2.8) maintain their form, with the various quantities replaced by barred ones, and i + ¯f . For geodesics going through subhorizon the combination 1 + f replaced by ¯H −2 perturbations with ¯Hi ≪ 1 the effective curvature term ¯f plays a minor role. However, this term is always important for the evolution of the perturbations, as can be seen from eq. (3.5). The optical equation takes the form , (3.6) with ¯ρ = ρ/ρ0,i. We omitted the shear, as it gives a negligible contribution to our results. The initial conditions (2.14), (2.15) become d2√ ¯A d¯λ2 = − 3 2 1 √ ¯A ¯ρ(cid:16)¯k0(cid:17)2 = 1 ¯Hiq ¯Ωs = qΩs. d√ ¯A d¯λ (cid:12)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)¯λ=0 √ ¯A(cid:12)(cid:12)(cid:12)λ=0 = 0, i A and ¯Ω = ¯H 2 i Ω. with ¯A = H 2 (3.7) (3.8) Light Propagation and Large-Scale Inhomogeneities 9 ) t ( W R F / ) r , t ( 3.0 2.5 2.0 1.5 1.0 0.5 0.7 0.8 0.9 1.0 1.1 R(t,r)/R(t,1) Figure 1. The evolution of the density profile for a central underdensity surrounded by an overdensity. 4. Single crossing We use ǫ1 = −0.01. The value of ǫ2 is fixed by the requirement that R 1 The typical cosmological evolution is displayed in fig. 1 for a central underdensity that is surrounded by an overdense region. The initial density ¯ρi(r) = ρi(r)/ρ0,i = 1 + ǫ(r) is constant ¯ρi = 1+ǫ1 in the region ¯r ≤ 0.8, constant ¯ρi = 1+ǫ2 in the region 0.9 ≤ ¯r ≤ 0.95, and ¯ρi = 1 for ¯r ≥ 1. In the intervals 0.8 ≤ ¯r ≤ 0.9 and 0.95 ≤ ¯r ≤ 1 it interpolates linearly between the values at the boundaries. This guarantees that the conditions ¯f (1) = ¯f ′(1) = 0, imposed by the matching of the FRW and LTB metrics, are satisfied. 0 ǫ(¯r)¯r2d¯r = 0. In a previous publication [27], we studied in detail the evolution of such inhomogeneities. We showed analytically that their growth is consistent with the standard theory of structure formation and the spherical collapse model [32]. Inhomogeneities that are initially of horizon size ( ¯Hi ∼ 1) with ǫ1 ∼ 10−5 and have a characteristic scale O(10) h−1 Mpc today also have a density contrast of O(1). In our numerical solutions we do not follow the very early evolution of the inhomogeneities. The perturbations we consider are already of subhorizon size and have a density contrast of O(10−2). They evolve to form present- day structures with size O(10) h−1 Mpc or larger and density contrast of O(1). The profile of the perturbations that we assume in this work differs slightly from the one in ref. [27] with respect to the width of the shell. In ref. [27] the widths of the underdense Light Propagation and Large-Scale Inhomogeneities 10 0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20 -0.25 z/zFRW t/tFRW A1/2/A1/2 FRW 0.0 0.2 0.4 0.8 1.0 0.6 r/r0 Figure 2. The relative difference in redshift, coordinate time, and beam area, between the propagation in the background of fig. 1 and in a homogeneous background, as a function of the radial coordinate. and overdense regions were taken equal. In this work we assume that the initial shells are narrower and denser than in ref. [27]. Clearly, our modelling of the Universe cannot reproduce all the details of the statistical nature of the inhomogeneities, especially during their evolution in the non-linear regime. Our choice results in a model of the Universe dominated by voids, in rough agreement with observations. It is possible, however, that the effect of underdensities is overestimated, as they occupy a slightly larger fraction of the total volume than the overdense regions already in the initial small perturbations. In fig. 1 we display the density profile at times ¯t=10, 100, 200, 250. We follow the evolution at later times as well, even though we do not depict it in fig. 1. We normalize the energy density to that of a homogenous FRW background (given by ¯ρF RW (¯t) = ¯ρ(¯t, 1)). The earliest time corresponds to the curve with the smallest deviation from 1, while the latest to the curve with the largest deviation. We observe that the density contrast grows and eventually becomes of O(1). The central energy density drops relative to the homogeneous background, while the surrounding region becomes denser. The radius of the central underdensity grows relative to the total size of the inhomogeneity, as this region expands faster than the average. The surrounding shell becomes thinner and denser. Light Propagation and Large-Scale Inhomogeneities 11 The effect of the spherical inhomogeneity on the characteristics of a light beam is depicted in figs. 2, 3. (Fig. 3 is a magnification of fig. 2 around ¯r = 0.) We assume a perturbation with ¯Hi = 1 at the initial time ¯ti = 0, so that the effects on the light beam are clearly visible. We use this perturbation for the numerical analysis both in this section and in section 5. Such a perturbation leads to an inhomogeneity at the present time with size that would be in conflict with observations (approximately 350 h−1 Mpc). We consider realistic perturbations in section 6. We consider a beam with cφ = 0 that is emitted from a point with ¯rs = 1.5 and passes through the center of symmetry. The emission time is ¯ts = 250 and the signal reaches ¯r = 1.5 again at a time ¯t ≃ 441. The redshift for the exiting beam at ¯r = 1.5 is z ≃ 0.46. We plot the relative difference in redshift (z − zF RW )/zF RW , coordinate time (¯t−¯tF RW )/¯tF RW , and beam area(cid:18)√ ¯A −q ¯AF RW(cid:19) /q ¯AF RW , between the propagation in the background of fig. 1 and in a homogeneous background, as a function of the radial coordinate ¯r. The arrows indicate the evolution of various quantities as the beam enters the inhomogeneity from one side and exits from the other. In the region 0.1 <∼ ¯r <∼ 0.9 we observe a significant deviation of all quantities from their values in a homogeneous background. However, the deviation becomes small in the region near the origin. When the beam moves out of the inhomogeneous region only the beam area deviates from the value in a homogeneous background. The coordinate time and the redshift are not affected significantly. We can obtain an understanding of the evolution depicted in figs. 2, 3 through an analytical treatment. The effect of the inhomogeneity on the characteristics of the light beam can be estimated analytically for perturbations with size much smaller than the distance to the horizon. These have ¯Hi ≪ 1. Let us consider a beam with cφ = 0 that passes through the center of the spherical inhomogeneity. The null condition (2.7) can be written as d¯t d¯r = ∓ ¯Hi ¯R′ q1 + ¯H 2 i ¯f ≃ ∓ ¯Hi ¯R′ ± ¯R′ ¯f , ¯H 3 i 2 If we keep terms up to O( ¯H 2 (4.1) for incoming and outgoing beams respectively. i ) we can neglect the second term in the above expression. This indicates that the spatial curvature does not play a role if the inhomogeneities are much smaller than the horizon. We can also employ the approximation ¯R′(¯t, ¯r) ≃ ¯R′(¯ts, ¯r) + ¯R (¯ts, ¯r)(¯t − ¯ts), as the time it takes for the light to cross the inhomogeneity is much shorter than the Hubble time. In fact, ¯t − ¯ts = O( ¯Hi) (see eqs. (4.2), (4.3) below). We denote by ¯rs the location of the source and by ¯ts the emission time of the beam. The solution of eq. (4.1) is ′ ¯t − ¯ts = ¯Hi(cid:16) ¯R(¯ts, ¯rs) − ¯R(¯ts, ¯r)(cid:17) + ¯H 2 i Z ¯rs ¯r − ¯H 2 i (cid:16) ¯R(¯ts, ¯rs) − ¯R(¯ts, ¯r)(cid:17) ¯R(¯ts, ¯r) + O( ¯H 3 i ) ¯R′(¯ts, ¯r) ¯R(¯ts, ¯r)d¯r (4.2) Light Propagation and Large-Scale Inhomogeneities 12 0.01 0.00 -0.01 0.00 z/zFRW t/tFRW A1/2/A1/2 FRW 0.05 r/r0 0.10 Figure 3. The relative difference in redshift, coordinate time, and beam area, between the propagation in the background of fig. 1 and in a homogeneous background, as a function of the radial coordinate. for an incoming beam, and ¯t − ¯ts = ¯Hi(cid:16) ¯R(¯ts, ¯r) − ¯R(¯ts, ¯rs)(cid:17) − ¯H 2 i Z ¯r ¯rs ¯R′(¯ts, ¯r) ¯R(¯ts, ¯r)d¯r + ¯H 2 i (cid:16) ¯R(¯ts, ¯r) − ¯R(¯ts, ¯rs)(cid:17) ¯R(¯ts, ¯r) + O( ¯H 3 i ) (4.3) for an outgoing one. These expressions are confirmed by numerical solutions. We can also make a comparison with the propagation of light in a FRW background. In this case we have ¯R(¯t, ¯r) = a(¯t)¯r = ¯R(¯t, 1)¯r. We have expressed the scale factor in terms of the value of the function ¯R(¯t, ¯r) at the boundary of the inhomogeneous region ¯r = 1. Let us consider light signals emitted at ¯rs = 1 and observed at the center (¯ro = 0) of the inhomogeneity. The difference in propagation time within the LTB and FRW backgrounds is ¯to − (¯to)F RW = ¯H 2 i Z 1 0 ¯R′(¯ts, ¯r) ¯R(¯ts, ¯r)d¯r− ¯H 2 i 2 ¯R(¯ts, 1) ¯R(¯ts, 1) +O( ¯H 3 i ).(4.4) For signals originating at ¯rs = 0 and detected at ¯ro = 1 the time difference has the opposite sign. As a result, the time difference for signals that cross the inhomogeneity is of O( ¯H 3 i ). This is in agreement with figs. 2, 3. Light Propagation and Large-Scale Inhomogeneities 13 We can derive similar expressions for the redshift of a light beam that passes through the center of the inhomogeneity. The geodesic equation (2.6) can be written as 1 k0 dk0 d¯r = − d ln(1 + z) d¯r == ± ¯Hi ′ ¯R 1 + ¯H 2 i ¯f ≃ ± ¯Hi ′ ¯R , for incoming and outgoing beams respectively. In this way we find ln(1 + z) = ¯Hi(cid:16) ¯R(¯ts, ¯rs) − ¯R(¯ts, ¯r)(cid:17) ′ − ¯H 2 i Z ¯r ¯rs ¯R (¯ts, ¯r)(cid:16) ¯R(¯ts, ¯rs) − ¯R(¯ts, ¯r)(cid:17) d¯r + O( ¯H 3 i ) for an incoming beam, and ln(1 + z) = ¯Hi(cid:16) ¯R(¯ts, ¯r) − ¯R(¯ts, ¯rs)(cid:17) ′ + ¯H 2 ¯R i Z ¯r ¯rs (¯ts, ¯r)(cid:16) ¯R(¯ts, ¯r) − ¯R(¯ts, ¯rs)(cid:17) d¯r + O( ¯H 3 i ) for an outgoing one. These expressions are confirmed by numerical solutions. For signals originating at ¯rs = 1 and detected at ¯ro = 0 the redshifts obey ln(cid:18) 1 + z i Z 1 1 + zF RW(cid:19) = ¯H 2 − 0 ¯H 2 i 2 ′ ¯R ′ ¯R (¯ts, ¯r)(cid:16) ¯R(¯ts, 1) − ¯R(¯ts, ¯r)(cid:17) d¯r (¯ts, 1) ¯R(¯ts, 1) + O( ¯H 3 i ). (4.5) (4.6) (4.7) (4.8) For signals originating at ¯rs = 0 and detected at ¯ro = 1 the r.h.s. of the above equation has the opposite sign. As a result, the redshift difference for signals that cross the inhomogeneity is of O( ¯H 3 i ). Again, this is in agreement with figs. 2, 3. 5. Luminosity distance and flux conservation The beam area obeys the second-order differential equation (3.6), whose solution depends crucially on the initial conditions. For initial conditions given by eqs. (3.7), (3.8) and symmetric situations, it is possible to determine the solution analytically. For signals emitted from some point ¯rs at a time ¯t = ¯ts and observed at ¯ro = 0 we have [33, 16] √ ¯A = (1 + z) ¯R(¯ts, ¯rs)√ ¯Ω. (5.1) This is in agreement with figs. 2, 3. For signals emitted from the center ¯rs = 0 and observed at ¯ro at a time ¯to we have √ ¯A = ¯R(¯to, ¯ro)√ ¯Ω. (5.2) However, for a signal that crosses the inhomogeneity we need to integrate eq. (3.6) from ¯r = 0 to ¯ro with initial conditions determined by the propagation from ¯rs to ¯r = 0. These are different from (3.7), (3.8), so that an analytical solution is not easy. Using perturbation theory, it is possible to obtain an analytical estimate of the deviation of the luminosity distance from its value in homogeneous cosmology. We Light Propagation and Large-Scale Inhomogeneities 14 0.005 0.000 -0.005 W R F , L D / L D -0.010 -0.015 -0.020 0.0 0.2 0.4 0.6 c /c , max 0.8 1.0 Figure 4. The deviation of the luminosity distance from its value in a homogeneous background, as a function of the impact parameter. consider beam trajectories that start at the boundary of the inhomogeneity, pass through its center and exit from the other side. The optical equation (3.6) can be written in the form We express d¯k1/d¯λ in the above equation using the geodesic equation and obtain d¯r2 + (cid:16)¯k1(cid:17)2 d2√ ¯A d2√ ¯A d¯r2 + ± d√ ¯A d¯r d¯k1 d¯λ 3 2 = − ′ ¯R 2 ¯Hi q1 + ¯H 2 i ¯R′′ ¯R′ + ¯f − ¯ρ(cid:16)¯k0(cid:17)2 √ ¯A. ¯f )  ¯f ′ 2(1 + ¯H 2 i ¯H 2 i (5.3) d√ ¯A d¯r 3 2 ¯ρ R′2 1 + ¯H 2 i √ ¯A(5.4) ¯f = − where the positive sign in the second term corresponds to ingoing and the negative sign to outgoing geodesics. We use a simplified initial configuration for this estimate. We take ¯ρ(0, ¯r) = 0 for 1) for ¯r > ¯r1. The initial conditions for an ingoing beam can ¯r < ¯r1 and ¯ρ(0, ¯r) = 1/(1 − ¯r3 be taken √ ¯A(¯r = 1) = 0, d√ ¯A(¯r = 1)/d¯r = −1, without loss of generality. We use the expansion √ ¯A = √ ¯A(0) + ¯Hi√ ¯A(1) + ¯H 2 √ ¯A(2) + O( ¯H 3 (5.5) and calculate √ ¯A(i) in each order of perturbation theory. The calculation in described in the appendix. We point out that our choice of initial configuration, that involves a discontinuous energy density, results in the appearance of δ-function singularities in the i ), i Light Propagation and Large-Scale Inhomogeneities 15 second derivatives of ¯R with respect to ¯r. These must be taken into account in a consistent calculation, as described in the appendix. We have checked that the expressions (5.1) and (5.2) are reproduced correctly by our results, up to second order in ¯Hi. When the photon exits the inhomogeneity at ¯r = 1 we find √ ¯A(¯r = 1) = 2 + 4 ¯Hi + 5 − d√ ¯A (¯r = 1) = 1 + 4 ¯Hi + 6 − dr For a homogeneous universe we have √ ¯A(¯r = 1) = 2 + 4 ¯Hi + 2 ¯H 2 d√ ¯A (¯r = 1) = 1 + 4 ¯Hi + 3 ¯H 2 i . dr i 3 1 + ¯r1 + 1! ¯H 2 1 + ¯r1 + 1! ¯H 2 3 ¯r2 ¯r2 i + O( ¯H 3 i ) i + O( ¯H 3 i ). (5.6) (5.7) (5.8) (5.9) In fig. It is clear that the effect of the inhomogeneity is of O( ¯H 2 i ). This conclusion has been confirmed by numerical solutions of the exact optical equations, without any approximations. Several beam crossings were studied for a multitude of values of ¯Hi. The coordinate time ¯to needed for the crossing, the redshift z and the beam area ¯A were plotted in terms of ¯Hi. The polynomial fits to these plots verify with good precision the conclusions of the present and previous sections: The deviations of ¯to and z from i ), while the deviations of ¯A are of their values in a homogeneous background are of O( ¯H 3 O( ¯H 2 i ). From the above we can conclude that the main effect of an inhomogeneity is to modify the luminosity distance of a light source behind it, while leaving the redshift 4 we depict the modification of the luminosity distance largely unaffected. inhomogeneity at (DL − (DL)F RW ) / (DL)F RW for light beams that cross a spherical various angles. The inhomogeneity is the one employed in the previous section. Its size is approximately 350 h−1 Mpc, much larger that what is deduced from observations. We consider realistic inhomogeneities in the following section. The crossing at a varying angle is achieved by choosing different values for the constant cφ in eq. (2.8). The light source is always located at the same point with ¯rs = 1.5. The beam is allowed to propagate until the redshift reaches a certain value zo ≃ 0.46. The beam with cφ = 0 that goes through the center reaches ro ≃ 1.5 at this time. The difference of the resulting luminosity distance from the one in a FRW background is depicted in fig. 4 as a function of cφ = cφ/cφ,max. The largest value cφ,max corresponds to beams that are emitted tangentially with respect to the center of symmetry. The angle θ between the initial direction of the beam and the radial direction is determined through the relation sin θ = cφ. The variable cφ plays the role of an impact parameter, normalized to 1 for light beams emitted tangentially. In fig. 4 we observe that, if cφ is sufficiently small for the light to travel through the central underdense region, the luminosity distance is enhanced relative to the homogeneous Light Propagation and Large-Scale Inhomogeneities 16 case. For larger values of cφ the light travels mainly through the overdense shell and the luminosity distance in reduced. If the redshift is not affected significantly by the propagation in the inhomogeneous background, the conservation of the total flux implies that the average luminosity distance must be the same as in the homogeneous case [34, 35]. As we have seen, this is not the case for an observer located at the center of an underdensity. The resulting increase in the luminosity distance, arising mainly from the increase in the redshift, can by employed for the explanation of the supernova data, even though the size of the required inhomogeneity is probably in conflict with the observed large-scale structure [15] -- [24]. On the other hand, if the source and the observers are located outside the inhomogeneity, as for fig. 4, so that the redshift is essentially unaffected, we expect that the energy flux may be redistributed in various directions but the total flux will be the same as in the homogeneous case [34, 35]. This implies that the integral of (DL − (DL)F RW ) / (DL)F RW over all angles must vanish. For an isotropic source the integration over the solid angle is 2π sin θ dθ. This is equivalent to the integration over the impact factor cφ, with a weight cφ/q1 − c2 φ. The integral of the function depicted in fig. 4 is indeed approximately zero. An analytical proof is very difficult, but a numerical analysis shows that a cancellation at the 90% level takes place between the positive and negative contributions. The remaining deviation from zero is caused by numerical errors and the small (of O( ¯H 3 i )), but non-vanishing, difference in redshifts in inhomogeneous and homogeneous backgrounds. 6. Multiple crossings In this section we consider light beams that pass through several of the inhomogeneities described in the previous sections. The light is emitted at some time ¯ts from a point with ¯r = 1 at the edge of the inhomogeneous region. Its initial direction is assumed to be random. The initial conditions for the beam area are given by eqs. (2.14), (2.15). The light moves through the inhomogeneity and exits from a point with ¯r = 1. Subsequently, the beam crosses the following inhomogeneity in a similar fashion. The angle of entry into the new inhomogeneity is assumed to be random again. The initial conditions are set by the values of √ ¯A and d√ ¯A/d¯λ at the end of the first crossing. This process is repeated until the light arrives at the observer. Of course, as time passes the profile of the inhomogeneities changes, as depicted in fig. 1. The assumption of entry into a spherical inhomogeneity at a random azimuthal angle is realized by selecting the impact factor cφ with a probability ∼ cφdcφ. The absence of the denominator q1 − c2 φ that appears in the weight employed in the previous section is justified by the fact that the source is located far from the center of the inhomogeneity, so that cφ ≪ 1. This is not strictly true for the first 1 or 2 crossings, but the induced error is small. In order to place the observer outside the inhomogeneities we always replace the Light Propagation and Large-Scale Inhomogeneities 17 y t i s n e D y t i l i b a b o r P y t i s n e D y t i l i b a b o r P 240 200 160 120 80 40 0 240 200 160 120 80 40 0 z=0.5 -0.01 0.00 DL/DL, FRW 0.01 z=1.5 -0.01 0.01 0.00 DL/DL, FRW y t i s n e D y t i l i b a b o r P y t i s n e D y t i l i b a b o r P 240 200 160 120 80 40 0 240 200 160 120 80 40 0 z=1 -0.01 0.00 DL/DL, FRW 0.01 z=2 -0.01 0.01 0.00 DL/DL, FRW Figure 5. The distribution of luminosity distances for various redshifts in the LTB Swiss-cheese model if the inhomogeneities have a characteristic scale of 40 h−1 Mpc. final spherical region crossed by the beam with a homogeneous configuration. that The essence of our procedure is the light beam encounters various inhomogeneities at various angles along its path. In this section we consider the propagation of the beam only in the intervals 0 ≤ r ≤ 1 of coordinate systems with origins at the center of the various inhomogeneities. This means that we neglect the homogeneous region between the inhomogeneities that we assumed in the previous region. (Essentially we assume that its width is negligible.) The reason for this omission is that the presence of the homogeneous region generates a bias towards negligible deviations of the luminosity distance from its value in a homogeneous cosmology. For example, if we consider light propagation in the intervals 0 ≤ r ≤ 1.5 as in the previous section, a large number of beam trajectories propagate only within the homogeneous regions with 1 ≤ r ≤ 1.5. These give a luminosity distance for the source equal to that in the FRW cosmology. On the other hand, there are trajectories that propagate through the inhomogeneities, for which the cancellation between positive and negative contributions results in a negligible total deviation of the luminosity distance from the value in a homogeneous background. It is the latter events that we are interested in, while the former are rather unphysical. Light Propagation and Large-Scale Inhomogeneities 18 y t i s n e D y t i l i b a b o r P y t i s n e D y t i l i b a b o r P 140 120 100 80 60 40 20 0 140 120 100 80 60 40 20 0 z=0.5 -0.03 -0.02 -0.01 0.00 0.01 0.02 DL/DL, FRW z=1.5 -0.03 -0.02 -0.01 0.00 0.01 0.02 DL/DL, FRW y t i s n e D y t i l i b a b o r P 140 120 100 80 60 40 20 0 y t i s n e D y t i l i b a b o r P 140 120 100 80 60 40 20 0 z=1 -0.03 -0.02 -0.01 0.00 0.01 0.02 DL/DL, FRW z=2 -0.03 -0.02 -0.01 0.00 0.01 0.02 DL/DL, FRW Figure 6. Same as in fig. 5 for a characteristic scale of 133 h−1 Mpc. The procedure we outlined above has an unsatisfactory element. The spherical inhomogeneities are assumed to follow each other continuously along the beam trajectory. If the angles of entry of the beam are non-zero, the resulting geometry implies that there is an overlap of the inhomogeneities outside the beam trajectory. This problem cannot be corrected as long as the assumption of spherical symmetry of the inhomogeneities is maintained. A choice must be made between an artificial bias in the luminosity distance if intermediate homogeneous regions are introduced, or the overlap of the inhomogeneities outside the beam trajectory. We have chosen the second option, as we believe that it provides a more reliable estimate of the distribution of luminosity distances. The total number of crossings determines the redshift and the final beam area, related to the luminosity distance. We determine the arrival time at the observer by requiring that the redshift reach a specific value. We repeat the calculation many times (at least 1000) for each value of the redshift and plot the resulting distribution of luminosity distances. The emission time is such that the arrival time is ¯to ≃ 284 for all the redshifts that we consider. At this time the profile of the inhomogeneity is very similar to the curve in fig. 1 with the largest deviation from 1. The deviations of the exact arrival time ¯to from the time in a homogeneous background is one order of magnitude smaller than the respective deviation for the luminosity distance. This is in agreement with the discussion in the Light Propagation and Large-Scale Inhomogeneities 19 y t i s n e D y t i l i b a b o r P y t i s n e D y t i l i b a b o r P 100 80 60 40 20 0 100 80 60 40 20 0 z=0.5 -0.04 z=1.5 -0.02 0.00 DL/DL, FRW 0.02 -0.04 -0.02 0.00 0.02 DL/DL, FRW y t i s n e D y t i l i b a b o r P 100 80 60 40 20 0 y t i s n e D y t i l i b a b o r P 100 80 60 40 20 0 z=1 -0.04 -0.02 0.00 DL/DL, FRW 0.02 z=2 -0.04 -0.02 0.00 0.02 DL/DL, FRW Figure 7. Same as in fig. 5 for a characteristic scale of 400 h−1 Mpc. previous section. In figs. 5, 6, 7 we depict the distributions of the deviations of the luminosity distances from the value in a homogeneous background for various redshifts. The three figures correspond to inhomogeneities with different characteristic length scales at the time of the emission of light. The background through which the light propagates is constructed as described in the previous sections. At the initial time ¯ti = 0 the inhomogeneities have the profile descibed in the previous section with ǫ1 = −0.01. The subsequent evolution is depicted in fig. 1. The characteristic length scale of the inhomogeneities relative to ) = ¯Hi. This quantity does the distance to the horizon at the time ¯ti = 0 is r0/(H −1 not appear in the rescaled evolution equations for the background. It appears only in the rescaled geodesic equations. For this reason we can use the same background, with an evolution depicted in fig. 1, in order to discuss inhomogeneities of various length scales. The important phenomenological quantity is the scale of the inhomogeneities f ) = R(tf , r0) = ¯R(¯tf , 1) ¯Hi. The rescaled present today. This is given by R(tf , r0)/(H −1 time ¯tf is equal to the time of arrival of light signals to the observer ¯to for all the cases we consider. As we mentioned already, ¯tf = ¯to ≃ 284. Our solution has ¯R(¯to) = 0.133. Using H −1 f = 3 × 103 h−1 Mpc we have R(tf , r0) = 400 h−1 ¯Hi Mpc. i Light Propagation and Large-Scale Inhomogeneities 20 The three figures 5, 6, 7 correspond to ¯Hi = 1/10, 1/3, 1, respectively. They describe the effect on the propagation of light of inhomogeneities with sizes 40, 133, 400 h−1 Mpc today. The present profile of the inhomogeneities has a density contrast O(1). Their evolution, as modelled by the LTB metric, is roughly consistent with the standard theory of structure growth. The choices ¯Hi = 1/3 and 1 lead to present-day inhomogeneities with length scales larger than those in typical observations. The size of the same perturbations at horizon crossing is larger than the value ∼ 10−5 implied by the CMB. However, we have included them for two reasons. Firstly, because there are indications that the presence of such large structures may be supported by observations [25, 26]. Secondly, because we would like to understand if inhomogeneities with sizes comparable to the horizon distance can have a significant effect on the luminosity distance for a random location of the observer. The total integral of the distributions has been normalized to 1 in all cases, so that they are in fact probability densities. They have similar profiles that are asymmetric around zero. Each distribution has a maximum at a value larger than zero and a long tail towards negative values. The average deviation is zero to a good approximation in all cases. This is expected according to our discussion of flux conservation in the previous section. As long as the light propagation in an inhomogeneous background does not modify significantly the redshift, the energy may be redistributed in various directions, but the total flux is conserved and remains the same as in a FRW background. The longer tail of the distribution towards luminosity distances smaller than the one in a homogeneous background is a consequence of the presence of a thin and dense spherical shell around each central underdensity. The number of beam trajectories that propagate through several shells is small. However, the focusing of the beam is substantial for such beams and the resulting luminosity distance much shorter than the average. The effect of the long tail is compensated by the shift of the maximum of the distribution towards positive values. The form of the distribution is very similar to that derived in studies modelling the inhomogeneities through the standard Swiss-cheese model [11]. In that case the strong focusing is generated by the very dense concentration of matter at the center of each spherical inhomogeneity. We emphasize, however, that the two models have a different region of applicability. The standard Swiss-cheese model is appropriate for length scales of O(1) h−1 Mpc or smaller, while the LTB Swiss-cheese model for scales of O(10) h−1 Mpc or larger. 7. Determination of cosmological parameters The form of the distributions can be quantified in terms of two parameters: The width of the distribution δd and the location of its maximum δm > 0. The first one characterizes the error induced to cosmological parameters derived through the curve of the luminosity Light Propagation and Large-Scale Inhomogeneities 21 s 0.6 0.4 0.2 0.0 -1.0 z=2 z=1.5 z=1 z=0.5 -0.8 -0.6 -0.4 -0.2 0.0 w Figure 8. The relative deviation of the luminosity distance from its value for w = 0 (non-relativistic matter) in homogeneous cosmology. distance as a function of redshift, while the second one the bias in such determinations. As discussed extensively in ref. [11], a small sample of data is expected to favour values of the luminosity distance near the maximum of the distribution, and thus generate a bias. In figs. 5, 6, 7 we observe that both δd and δm grow with inreasing redshift z and scale ¯Hi. For ¯Hi = 1/10 the average width δd increases from approximately 0.005 to 0.01 as z increases from 0.5 to 2. The maximum is δm <∼ 0.002 for all z. The asymmetry of the distribution is very small. For ¯Hi = 1/3 the average width increases from approximately 0.005 to 0.02 as z increases from 0.5 to 2. The maximum is δm <∼ 0.002 for all z. The asymmetry of the distribution and the longer tail towards negative values are clearly visible in fig. 6. For ¯Hi = 1 the average width increases from approximately 0.01 to 0.04 as z increases from 0.5 to 2. The maximum increases from 0.005 to 0.01. The asymmetry of the distribution is very distinctive in fig. 7. The values of δd and δm are very small for all the values of ¯Hi that we considered. This implies that we do not expect a significant effect on the cosmological parameters. As an interesting example we consider the parameter w that appears in the equation of state of the cosmological fluid. In homogeneous cosmology the luminosity distance is a function of w and the redshift z. The relative deviation of the luminosity distance from Light Propagation and Large-Scale Inhomogeneities 22 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 ) z ( 0.0 0.0 0.5 1.5 2.0 1.0 z Figure 9. The function α(z) = − ∂s(w, z)/∂ww=0. its value for w = 0 (non-relativistic matter) is s(w, z) = DL(w, z) DL(0, z) − 1 = 1 1 + 3w 1 − (z + 1)−(1+3w)/2 1 − (z + 1)−1/2 − 1. The deviations of δd(z) and δm(z) We depict this function in fig. 8. For w = −1 the luminosity distance is larger by roughly 35% relative to w = 0 for z = 0.5. For z = 1 the relative increase is approximately 70%. the presence of inhomogeneities can be attributed to deviations of w from zero if the cosmology is assumed to be homogeneous. This can be achieved by identifying δd(z) or δm(z) with s(w, z). For small w we have from zero because of (7.1) (7.2) s(w, z) ≃ −3 1 − 1 2 ln(z + 1) (z + 1)1/2 − 1! w = −α(z)w. We depict the function α(z) in fig. 9. For a given value of z we can derive effective values of w from the relations wef f = −δd,m(z)/α(z). It is clear from the values we quoted above for δd, δm and fig. 9 that wef f ≪ 1. As a result, for a random location of the observer the perceived acceleration of the Universe cannot be attributed to the modification of the luminosity distance by large-scale inhomogeneities, even when their characteristic scale is smaller than the horizon distance by less than a factor of 10. On the other hand, the presence of inhomogeneities induces a statistical error in the value of w deduced from astrophysical data, as well as a shift of its average value Light Propagation and Large-Scale Inhomogeneities 23 if the sample is small. According to our results, for ¯Hi = 1/10 (and present size of the inhomogeneities 40 h−1 Mpc) the error is δw ≃ 0.015 for all z between 0.5 and 2, while the average value ¯w for a small sample is negative and of O(10−3). For ¯Hi = 1/3 (and present size of the inhomogeneities 133 h−1 Mpc) the error increases from 0.015 to 0.025 as z increases from 0.5 to 2, while the average value ¯w is again negative and of O(10−3). For ¯Hi = 1 (and present size of the inhomogeneities 400 h−1 Mpc) the error increases from 0.03 to 0.05, while the average is ¯w ≃ −0.015. The shift in the average value is always smaller than a standard deviation. The values of δd and δm can be compared to those generated by the effects of gravitational lensing at scales typical of galaxies or clusters of galaxies. At such scales the Universe is modelled through the standard Swiss-cheese model, with the mass of each inhomogeneity concentrated in a very dense object at its center [11]. The typical values of δd and δm are larger by at least an order of magnitude than the ones we obtained. The reason is that in our model the density contrast is always of O(1). Our study indicates that the effect on the luminosity distance grows when the length scale of the inhomogeneities increases and becomes comparable to the horizon distance. However, the strong lensing effect of a high concentration of mass, such as a galaxy or cluster of galaxies, gives a much larger effect. We conclude that within our model the presence of inhomogeneities with large length scales, even comparable to the horizon distance, and density contrast of O(1) does not influence significantly the propagation of light if the source and the observer have random locations. It is possible that our modelling of the Universe is lacking some essential feature that could generate a significant effect on the luminosity distance. For example, it has been suggested that large fluctuations in the spatial curvature may result in a strong backreaction on the overall expansion [37]. Unfortunately, there are no known exact models that realize such a scenario. In their absence the effect on the luminosity distance cannot be computed. In our model, there are variations of the local curvature which result in the collapse of the overdense regions (either central overdensities, or shells surrounding underdensities). The evolution is in approximate agreement with the standard theory of structure formation. It seems difficult to reconcile a much larger local curvature with a large-scale structure consistent with observations. Another possibility is that the assumed spherical symmetry of the inhomogeneities in our modelling of the Universe is too constraining. It is possible that photon propagation through inhomogeneities without such a symmetry results in a stronger modification of the luminosity distance. This point will be the subject of a future investigation. Acknowledgments This work was supported by the research program "Pythagoras II" (grant 70-03-7992) of the Greek Ministry of National Education, partially funded by the European Union. Light Propagation and Large-Scale Inhomogeneities 24 8. Appendix In this appendix we describe the solution of eq. (5.4) through the use of perturbation theory in ¯Hi. In this way we demonstrate that a spherical inhomogeneity induces a deviation of the luminosity distance from its value in a homogeneous background which is of O( ¯H 2 i ). The null constraint (2.7), keeping terms up to O( ¯Hi), is d¯t/d¯r = ± ¯Hi, with the negative sign corresponding to ingoing and the positive to outgoing geodesics. We can set ts = 0 so the geodesic inside the inhomogeneity is ¯t = − ¯Hi(¯r − ¯r0) + O( ¯H 2 i ) for ingoing, and ¯t = ¯Hi(¯r + ¯r0) + O( ¯H 2 i ) (8.3) (8.4) for outgoing geodesics. We can treat ¯t as an O( ¯Hi) quantity. ¯ρ(0, ¯r) = 1/(1 − ¯r3 ¯t = 0: The initial configuration we use for this estimate has ¯ρ(0, ¯r) = 0 for ¯r < ¯r1 and 1) for ¯r > ¯r1. From (3.5) we can calculate various derivatives of ¯R at ′ (¯t, ¯r) = ¯R ¯R ¯R′′ ¯R′ (¯t, ¯r) = ′ (0, ¯r) + ¯t ¯R ′ (0, ¯r) + O( ¯H 2 i ), ¯t2 2 ′′ ¯R (0, ¯r) + O( ¯H 3 i ). (8.5) (8.6) For ¯r > ¯r1 we have ′ ¯R (0, ¯r) = 2r3 (¯r3 ′′ ¯R r4 (¯r3 (0, ¯r) and ¯R ′ (0, ¯r) = − r3 + 2¯r3 1 1 − 1) 3¯r3 1 1 − 1) (0, ¯r) are zero. For the initial configuration that we is discontinuous at ¯r = ¯r1 and (8.8) (8.7) . ′′ ′ ′′ has δ-function singularities at the same points. The initial conditions for the solution of eq. (5.4) for an ingoing beam can be taken For ¯r < ¯r1 both ¯R assume, ¯R is a continuous function of ¯r. However, ¯R ¯r = 1, while ¯R √ ¯A(1) = 0, d√ ¯A(1)/d¯r = −1, without loss of generality. We use the expansion (8.9) To zeroth order in ¯Hi, eq. (5.4) becomes d2√ ¯A(0)/d¯r2 = 0, with solution √ ¯A(0)(¯r) = −(r − 1) for ingoing and √ ¯A(0)(¯r) = r + 1 for outgoing beams. To first order in ¯Hi, eq. (5.4) gives d√ ¯A(1)/d¯r = −2, with solution √ ¯A(1)(¯r) = r2 − 2r + 1 for ingoing and √ ¯A(1)(¯r) = r2 + 2r + 1 for outgoing beams. √ ¯A = √ ¯A(0) + ¯Hi√ ¯A(1) + ¯H 2 √ ¯A(2) + O( ¯H 3 i ). i Light Propagation and Large-Scale Inhomogeneities 25 To second order in ¯Hi, and for ¯r > ¯r1 and ingoing geodesics, we obtain d2√ ¯A(2) d¯r2 + 2¯t(¯r) ¯R ′ (0, ¯r) − ¯t2 2 ′′ ¯R (0, ¯r) + ¯f ′(¯r) 2 !d√ ¯A(0) + 2 ¯ρ(0, ¯r)√ ¯A(0). d¯r = − 3 2 d√ ¯A(1) d¯r For ¯r < ¯r1 and ingoing geodesics, we have d2√ ¯A(2) d¯r2 + ¯f ′(¯r) 2 d√ ¯A(0) d¯r d√ ¯A(1) d¯r = 0. + 2 For ¯r < ¯r1 and outgoing geodesics, we have d2√ ¯A(2) d¯r2 + ¯f ′(¯r) 2 d√ ¯A(0) d¯r − 2 d√ ¯A(1) d¯r = 0. Finally, for ¯r > ¯r1 and outgoing geodesics, we obtain d2√ ¯A(2) d¯r2 + −2¯t(¯r) ¯R ′ (0, ¯r) − ¯t2 2 ′′ ¯R (0, ¯r) + ¯f ′(¯r) 2 ! d√ ¯A(0) d¯r − 2 ¯ρ(0, ¯r)√ ¯A(0). = − 3 2 (8.10) (8.11) (8.12) (8.13) d√ ¯A(1) d¯r The above equations can be solved analytically through simple integration, with the values at the end of each interval determining the initial conditions for the next one. The only non-trivial point is that the δ-function singularities of ¯R at ¯r = ¯r1 and ¯r = 1 induce discontinuities in the values of d√ ¯A(2)/d¯r at these points. These must be taken into account in a consistent calculation. The discontinuities can be easily determined through the integration of eqs. (8.10) -- (8.13) in an infinitesimal interval around each of these points. The remaining calculation is straightforward. When the photon exits the inhomogeneity at ¯r = 1 we find that √ ¯A is given by eq. (5.6). ′′ Light Propagation and Large-Scale Inhomogeneities 26 References [1] A. G. Riess et al. [Supernova Search Team Collaboration], Astron. J. 116 (1998) 1009 [arXiv:astro- ph/9805201]; Astrophys. J. 607 (2004) 665 [arXiv:astro-ph/0402512]; S. Perlmutter et al. [Supernova Cosmology Project Collaboration], Astrophys. J. 517 (1999) 565 [arXiv:astro-ph/9812133]. [2] W. J. Percival et al. [The 2dFGRS Collaboration], Mon. Not. Roy. Astron. Soc. 327 (2001) 1297 [arXiv:astro-ph/0105252]; J. L. Sievers et al., Astrophys. J. 591 (2003) 599 [arXiv:astro-ph/0205387]. [3] D. N. Spergel et al. [WMAP Collaboration], Astrophys. J. Suppl. 148 (2003) 175 [arXiv:astro- ph/0302209]. [4] S. Rasanen, JCAP 0402 (2004) 003 [arXiv:astro-ph/0311257]; E. W. Kolb, S. Matarrese, A. Notari and A. Riotto, Phys. Rev. D 71 (2005) 023524 [arXiv:hep- ph/0409038]. [5] T. Futamase and M. Sasaki, Phys. Rev. D 40 (1989) 2502. M. Kasai, T. Futamase and F. Takahara, Phys. Lett. A 147 (1990) 97. F. Hadrovic and J. Binney, arXiv:astro-ph/9708110. T. Pyne and M. Birkinshaw, Mon. Not. Roy. Astron. Soc. 348 (2004) 581 [arXiv:astro- ph/0310841]. [6] E. W. Kolb, S. Matarrese, A. Notari and A. Riotto, Phys. Rev. D 71 (2005) 023524 [arXiv:hep- ph/0409038]. E. Barausse, S. Matarrese and A. Riotto, Phys. Rev. D 71 (2005) 063537 [arXiv:astro- ph/0501152]. [7] R. K. Sachs, Proc. Roy. Soc. London A 264 (1961) 309. [8] R. Kantowski, Astrophys. J. 155 (1969) 89. [9] A. Einstein and E. G. Straus, Rev. Mod. Phys. 17 (1945) 120; [10] R. Kantowski, Astrophys. J. 507 (1998) 483 [arXiv:astro-ph/9802208]; Phys. Rev. D 68 (2003) ibid. 18 (1946) 148. 123516 [arXiv:astro-ph/0308419]; R. Kantowski and R. C. Thomas, Astrophys. J. 561 (2001) 491 [arXiv:astro-ph/0011176]. [11] D. E. Holz and R. M. Wald, Phys. Rev. D 58 (1998) 063501 [arXiv:astro-ph/9708036]; D. E. Holz, Astrophys. J. 506, L1 (1998) [arXiv:astro-ph/9806124]; D. E. Holz and E. V. Linder, Astrophys. J. 631, 678 (2005) [arXiv:astro-ph/0412173]. [12] M. Sereno, G. Covone, E. Piedipalumbo and R. de Ritis, Mon. Not. Roy. Astron. Soc. 327 (2001) 517 [arXiv:astro-ph/0102486]; M. Sereno, E. Piedipalumbo and M. V. Sazhin, Mon. Not. Roy. Astron. Soc. 335 (2002) 1061 [arXiv:astro-ph/0209181]. [13] C. C. Dyer and R. C. Roeder, Astrophys. J. 180 (1973) L31; ibid. 189 (1974) 167 [14] G. Lemaitre, Gen. Rel. Grav. 29 (1997) 641; R. C. Tolman, Proc. Nat. Acad. Sci. 20 (1934) 169; H. Bondi, Mon. Not. Roy. Astron. Soc. 107 (1947) 410. [15] N. Mustapha, C. Hellaby and G. F. R. Ellis, Mon. Not. Roy. Astron. Soc. 292 (1997) 817 [arXiv:gr- qc/9808079]. [16] M. N. Celerier, Astron. Astrophys. 353 (2000) 63 [arXiv:astro-ph/9907206]. [17] H. Iguchi, T. Nakamura and K. i. Nakao, Prog. Theor. Phys. 108 (2002) 809 [arXiv:astro- ph/0112419]; K. Bolejko, arXiv:astro-ph/0512103; R. Mansouri, arXiv:astro-ph/0512605; R. A. Vanderveld, E. E. Flanagan and I. Wasserman, Phys. Rev. D 74 (2006) 023506 [arXiv:astro- Light Propagation and Large-Scale Inhomogeneities 27 ph/0602476]; D. Garfinkle, Class. Quant. Grav. 23 (2006) 4811 [arXiv:gr-qc/0605088]; D. J. H. Chung and A. E. Romano, Phys. Rev. D 74 (2006) 103507 [arXiv:astro-ph/0608403]. [18] S. Rasanen, JCAP 0411 (2004) 010 [arXiv:gr-qc/0408097]; JCAP 0611 (2006) 003 [arXiv:astro- ph/0607626]. [19] J. W. Moffat, JCAP 0510 (2005) 012 [arXiv:astro-ph/0502110]; arXiv:astro-ph/0505326. [20] H. Alnes, M. Amarzguioui and O. Gron, JCAP 0701 (2007) 007 [arXiv:astro-ph/0506449]; Phys. Rev. D 73 (2006) 083519 [arXiv:astro-ph/0512006]; H. Alnes and M. Amarzguioui, Phys. Rev. D 75 (2007) 023506 [arXiv:astro-ph/0610331]. [21] K. Enqvist and T. Mattsson, JCAP 0702 (2007) 019 [arXiv:astro-ph/0609120]. [22] C. H. Chuang, J. A. Gu and W. Y. Hwang, arXiv:astro-ph/0512651. [23] P. S. Apostolopoulos, N. Brouzakis, N. Tetradis and E. Tzavara, JCAP 0606 (2006) 009 [arXiv:astro-ph/0603234]. [24] T. Biswas, R. Mansouri and A. Notari, arXiv:astro-ph/0606703. [25] K. Tomita, Astrophys. J. 529, 26 (2000); Astrophys. J. 529, 38 (2000); Mon. Not. Roy. Astron. Soc. 326 (2001) 287 [arXiv:astro-ph/0011484]. [26] W. J. Frith, G. S. Busswell, R. Fong, N. Metcalfe and T. Shanks, Mon. Not. Roy. Astron. Soc. 345 (2003) 1049 [arXiv:astro-ph/0302331]. [27] N. Brouzakis, N. Tetradis and E. Tzavara, JCAP 0702 (2007) 013 [arXiv:astro-ph/0612179]. [28] P. Schneider, J. Ehlers and E. E. Falco, Gravitational Lenses, Springer-Verlag, Berlin. [29] T. Biswas and A. Notari, arXiv:astro-ph/0702555. [30] W. Israel, Nuovo Cim. B 44 (1966) 1. [31] H. Sato, Prog. Theor. Phys. 76 (1986) 1250; V. A. Berezin, V. A. Kuzmin and I. I. Tkachev, Phys. Rev. D 36 (1987) 2919; S. Khakshournia and R. Mansouri, Phys. Rev. D 65 (2002) 027302 [arXiv:gr-qc/0307023]. [32] J. E. Gunn and J. R. I. Gott, Astrophys. J. 176, 1 (1972); A. Cooray and R. Sheth, Phys. Rept. 372 (2002) 1 [arXiv:astro-ph/0206508]. [33] M. H. Partovi and B. Mashhoon, Astrophys. J. 276 (1984) 4. N. P. Humphreys, R. Maartens and D. R. Matravers, Astrophys. J. 477 (1997) 47 [arXiv:astro- ph/9602033]. [34] S. Weinberg, Astrophys. J. 208 (1976) L1 [35] H. G. Rose, Astrophys. J. 560 (2001) L15 [arXiv:astro-ph/0106489]. [36] G. Aldering, A. G. Kim, M. Kowalski, E. V. Linder and S. Perlmutter, Astropart. Phys. 27 (2007) 213 [arXiv:astro-ph/0607030]. [37] T. Buchert, Class. Quant. Grav. 23 (2006) 817 [arXiv:gr-qc/0509124]; T. Buchert, J. Larena and J. M. Alimi, Class. Quant. Grav. 23 (2006) 6379 [arXiv:gr-qc/0606020].
astro-ph/9609137
2
9609
1996-09-20T05:06:05
Formation of the Galaxy
[ "astro-ph" ]
Current ideas on the formation of the Galaxy are reviewed. Many of the observed characteristics of our Milky Way System are consistent with a scenario in which the Galaxy formed inside out, with the inner part of it evolving by rapid collapse of a single protogalaxy, while the outer halo was accreted over an extended period. A number of possible problems with this "Standard Model" are discussed and summarized in section 5.
astro-ph
astro-ph
FORMATION OF THE GALAXY Sidney van den Bergh Dominion Astrophysical Observatory, National Research Council of Canada, 5071 West Saanich Road, Victoria, British Columbia, V8X 4M6, Canada Electronic mail: [email protected] Received: 1996 March 11 ; Accepted: - 2 - ABSTRACT Current ideas on the formation of the Galaxy are reviewed. Many of the observed characteristics of our Milky Way System are consistent with a scenario in which the Galaxy formed inside out, with the inner part of it evolving by rapid collapse of a single protogalaxy, while the outer halo was accreted over an extended period. A number of possible problems with this "Standard Model" are discussed and summarized in § 5. 1. INTRODUCTION - 3 - Observational evidence on the formation and early evolutionary history of galaxies may be derived from images and spectra of very distant galaxies (e.g. Cowie, Hu & Songaila 1995, Abraham et al. 1996, van den Bergh et al. 1996) that are viewed at large look-back times. Alternatively, one can study the fossil evidence for Galactic evolution that is provided by the chemical abundances and kinematics of stars belonging to the oldest population components of the Milky Way System. More than three decades ago, Eggen, Lynden-Bell & Sandage (1962, ELS) used such observations of the metallicities and orbit shapes of high- velocity stars to conclude that the Milky Way System formed by rapid collapse of a single massive protogalaxy. More recently, Searle (1977) showed that, contrary to the predictions of the ELS model, globular clusters in the outer halo of the Galaxy did not exhibit a radial abundance gradient. This conclusion is strengthened by observations of RR Lyrae field stars (Suntzeff, Kinman & Kraft 1991) which show no radial abundance gradient for R > 10 kpc. Searle's GC conclusion, and the remark by Toomre (1977) that "it seems almost inconceivable that there wasn't a great deal of merging of sizable bits and pieces (including quite a few lesser galaxies) early in the career of every major galaxy", led Searle & Zinn (1978, SZ) to propose a scenario for the formation of the Milky Way System in which transient protogalactic fragments "continued to fall into - 4 - dynamical equilibrium with the Galaxy for some time after the collapse had been completed". Since this model was proposed in the late 1970's, various pieces of evidence have emerged which strengthen the SZ scenario, while others appear difficult to reconcile with it. Some of these problems will be discussed in more detail in this review. 2. FORMATION OF THE GALACTIC HALO 2.1 Capture and infall In situ measurements of the radial velocities of halo stars appear to indicate (Majewski, Hawley & Munn 1996, Majewski, Munn & Hawley 1996) that the Galactic halo may not presently be well-mixed dynamically. This suggests that we are now observing the signature of past infall of the "sizable bits and pieces" envisioned by Toomre (1977). Prima facie evidence for such infall is provided by the recently discovered Sagittarius dwarf galaxy (Ibata, Gilmore & Irwin 1994), which appears to be merging with the Milky Way System. Debris from tidal stripping of the Sagittarius dwarf (and of other dwarf companions of the Galaxy that may once have existed) will result in the formation of moving groups in the halo (Johnston, Spergel & Hernquist 1995). The present absence of any other dwarf companions to the Galaxy with R < 50 kpc might be due to GC their removal by capture via dynamical friction (van den Bergh 1994a). - 5 - Bellazzinni, Fusi Pecci & Ferraro (1996) have recently noticed a curious relationship between the surface brightnesses and Galactocentric distances of the dwarf spheroidal companions to the Galaxy. They find that the lowest surface brightness (and, hence, the least stable) dSph galaxies are located closest to the Galactic center. Possibly, this result indicates that many low surface brightness dSph galactics remain to be discovered at large distances from the Galaxy. [The faint UMi system would, for example, have been difficult to discover if its red giants had been fainter than the R _20 plate limit of the Palomar Sky Survey]. The absence of any dSph galaxies with high surface brightnesses at R < 100 kpc, which is noted by Bellazzini et al., is not unexpected because only three such objects are known in the much larger volume with 100 < R (kpc) < 300. If the GC luminosity function of the Local Group does indeed contain more faint objects than presently believed, then the apparent discrepancy between the shallow slope of the Local Group luminosity function, and the steep slope of the luminosity functions of rich clusters (Driver et al. 1994, Bernstein et al. 1995), might be removed. Van den Bergh (1993) found that 11 Galactic globulars on retrograde orbits have < [Fe/H] > = -1.59 ± 0.07, which is significantly higher than the value < [Fe/H] > = -1.86 ± 0.08 that Suntzeff (1992) derived for 13 true globular - 6 - clusters in the Large Magellanic Cloud. Since mean cluster metallicity increases with parent galaxy metallicity (van den Bergh 1975), this might be taken to suggest that Galactic globular clusters on retrograde orbits originated in an ancestral galaxy which was more massive than the LMC. Alternatively (and perhaps more plausibly) the fact that the clusters on retrograde orbits have a mean metallicity < [Fe/H] > = -1.59 ± 0.07, which is similar to those of halo clusters on direct orbits (for which < [Fe/H] > = -1.65 ± 0.11) might be understood by assuming that most halo clusters formed in a single highly turbulent protogalaxy. A Kolmogorov-Smirnov test shows no statistically significant difference in the distribution of [Fe/H] values for 12 halo clusters in prograde orbits and for 11 halo clusters that are in retrograde orbits (van den Bergh 1993). Nevertheless, it is of interest to note that six out of 11 halo clusters in retrograde orbits have metallicities in the narrow range -1.59  [Fe/H]  -1.51. In fact, all three of the clusters with the most extreme retrograde motions [i.e. those designated R! by van den Bergh (1993)] have metallicities that fall in this narrow range. This is a point that had previously also been noted by Rodgers & Paltoglou (1984). However, the fact that the globular clusters in the LMC have a large range in [Fe/H] values (Suntzeff 1992) shows that the small [Fe/H] range of most Galactic globular clusters in retrograde orbits does not constitute evidence for origin in a - 7 - single captured ancestral object. The reason for the small metallicity range of the majority of the Galactic globular clusters on retrograde orbits remains a mystery. An argument against the hypothesis that the Galaxy has experienced early mergers with massive ancestral objects similar to the Large Magellanic Cloud is provided by the observation that the globular clusters in the LMC do not exhibit the Oosterhoff (1939) dichotomy of the mean periods of RR Lyrae variables. A merger with an object similar to the Large Cloud would be expected to have left clusters with intermediate periods, such as that observed in the LMC clusters NGC 1466 ( < P > = 0.59 days) and NGC 1841 ab ( < P > = 0.59 days) in the Galactic halo. Furthermore, a relatively recent ab merger with an object resembling the Large Magellanic Cloud or Small Magellanic Cloud would have left many massive intermediate-age clusters in the halo of the Galaxy. In fact, only half a dozen such objects are known; two of these (Arp 2 and Terzan 7) (Buonanno et al. 1994) appear to be associated with the Sagittarius dwarf. This suggests that mergers with rather massive dwarfs resembling the Sagittarius dSph galaxy were probably not very frequent during most of the history of the Galaxy. This makes it unlikely that such mergers provided a dominant contribution to the population of the Galactic halo. - 8 - Minniti, Meylan & Kissler-Pastig (1996) have argued that the cluster Terzan 7, for which Da Costa & Armandroff (1995) find that [Fe/H] = -0.36, is too metal-rich to be physically associated with the Sagittarius dwarf galaxy. However, the same evidence could also be invoked to argue that Ter 7 cannot be a member of the outer halo of the Galaxy! The youngest known cluster in the Galactic halo is Ruprecht 106 (Kaluzny, Krzeminski & Mazur 1995) which appears to have an age of only 9.3 Gyr (Richer et al 1996). It has been suggested by Lin & Richer (1992) that such relatively young clusters in the Galactic halo might have been tidally captured from the Magellanic Clouds. A possible problem with this hypothesis is that Rup 106 has a half-light radius r = 1.10: pc (Harris 1996a), which is less than half h that of any other Magellanic Cloud globular. Furthermore, the age of Rup 106 falls near the center of the quiescent period, from 12 Gyr to 6 Gyr ago, during which the Large Cloud does not seem to have produced any cluster. The second youngest known halo cluster is Arp 2, to which Richer et al. assign an age of 10.3 Gyr. This object is probably associated with the Sagittarius dwarf, which precludes it having been detached from either of the Magellanic Clouds. Finally, the third youngest known halo clusters is Palomar 12, to which Richer et al. (1996) assign an age of 10.5 Gyr. Van den Bergh (1994b) gave a number of - 9 - reasons why this object is unlikely to have been torn from the Magellanic Clouds. In particular, he noted that, at [Fe/H] = -1.14, Pal 12 falls well outside the metallicity range covered by other LMC globular clusters (Suntzeff 1992). 2.2 Destruction of globular clusters Table 1 shows a comparison between the luminosity of all halo stars and the total luminosity of all true globular clusters in the Galaxy, the Large Magellanic Cloud, the Fornax dwarf spheroidal, and the Virgo giant elliptical M49. Perhaps surprisingly, the data in this table show that globular clusters account for rather similar fractions of the total luminosity of the spheroidal populations of these galaxies. Cluster destruction by disk and bulge shocking will be of only negligible importance in the Fornax dwarf galaxy. The close coincidence between the fraction of all halo Population II light that is in the form of globular clusters in Fornax and in the Milky Way System therefor provides weak evidence against the hypothesis (Fall & Rees 1977) that the present Galactic globular cluster system represents only a faint shadow of its former self. This is consistent with calculations by Hut & Djorgovski (1992) which appear to indicate that only 3.6 ± 2.2% of the total Galactic globular cluster population is presently being destroyed per gigayear. - 10 - 2.3 Evolution of the Galactic halo Hartwick (1978) has argued that the Galactic halo consists of a flattened inner component with c/a ~ 0.6 and a vertical scale-height of 1.6 kpc (which is dominant near the Sun's position), and a more nearly spherical outer component. A somewhat more complex model for the halo is advocated by Norris (1994). According to Norris the Galactic halo consists of (1) a dynamically hot, non- rotating spherical, metal-poor component, within which (2) a somewhat metal- richer ( [Fe/H] > -1.5 ) rotating thick disk is embedded. Superimposed on this structure is (3) a constituent which was accreted à la Searle & Zinn (1978). For a more detailed discussion of possible sub-structures in the Galactic halo the reader is referred to the review by Majewski (1993). Perhaps the strongest evidence against the hypothesis that the outer halo of the Galaxy was mainly formed by capture of "transient protogalactic fragments" (Searle & Zinn 1978) is provided by the observation that the half-light radii r of halo globular clusters grow with increasing Galactocentric distance h R . For clusters with R > 20 kpc, van den Bergh (1995) finds a rank GC GC correlation coefficient , (r , R ) = +0.61 ± 0.18. One of the reasons for the existence of this correlation is that the Galactic halo contains few compact GC h globular clusters; even though such objects would have survived destruction - 11 - much more easily than the distended globulars that actually populate the outer halo of the Galaxy. A Kolmogorov-Smirnov test shows no significant difference between the frequency distribution of half-light radii r of globular clusters in the LMC (van h den Bergh 1994b) and in the outer ( R > 10 kpc) halo of the Galaxy GC (Djorgovski 1993). However, a comparison between the metallicity distributions of LMC clusters (Suntzeff 1992) and of that for Galactic globulars with R > 10 GC kpc shows that the LMC clusters are, on average, more metal deficient than those in the outer halo. A K-S test shows only a 7% probability that the LMC and Galactic halo clusters could have been drawn from the same parent population of [Fe/H] values. It is tentatively concluded that capture of a few LMC-like objects is unlikely to have produced a globular cluster population resembling that presently observed in the outer Galactic halo. In Fig. 1, the half-light radii of the globular clusters in the Fornax dSph galaxy (van den Bergh 1994b) are plotted as arrows. Note that the radii of these clusters are, on average, smaller than those of globular clusters in the outer halo of the Galaxy. The paucity of compact clusters, like those in Fornax, in the outer halo of our Milky Way System militates strongly against the hypothesis that the - 12 - outer halo of the Galaxy was mainly assembled from the debris of Fornax-like dwarf galaxies. Note in particular that all five Galactic globular clusters with RGC > 40 kpc have half-light radii r greater than that of the largest Fornax globular. h In Fig. 2, the half-light radii r of Galactic globular clusters are plotted h versus their perigalactic distances P (van den Bergh 1995). Also shown are the clusters in Fornax plotted at the Fornax perigalactic distance of 100 ± 40 kpc (Hodge & Michie 1970). The figure shows that the Fornax clusters are smaller than Galactic globular clusters at similar perigalactic distances. The Sagittarius dSph galaxy, which appears to contain four globular clusters, does not throw much light on the origin of the halo. The cluster NGC 6715 (M54) has r = 2.8 h pc, which is typical of globular clusters in the main body of the Galaxy. On the other hand the cluster Arp 2 has r = 13.2 pc, which would place it firmly among h the outer Galactic outer halo. Finally, Terzan 7 and Terzan 8 have radii of intermediate size. The half-light radii of globular clusters in the Galactic halo increase with Galactocentric distance R and with perigalactic distance P (see Figs. 1 and 2), GC whereas metallicity [Fe/H] does not (Searle, 1977, van den Bergh 1995). This suggests that the sizes of globular clusters were set by global parameters, whereas - 13 - the metallicities of individual clusters were determined by local enrichment events. For very metal-poor halo stars, it is now possible (Sneden et al. 1996) to see the signatures of individual supernova enrichment events. Lee (1993) and van den Bergh (1993) have independently pointed out that Galactic halo clusters can be divided into two populations on the basis of their metallicities and horizontal branch gradients C (B - R) / (B + V + R). Fig. 3 shows that all clusters with R < 8.5 kpc appear to lie on (or close to) a single C GC versus [Fe/H] relation. Van den Bergh (1993) has assigned all globular clusters of this type to his  Population. On the other hand, most halo clusters with R >GC 8.5 kpc are seen to fall below (or to the left) of this fiducial line. Van den Bergh (1993) has assigned such clusters to his  Population. Six out of nine clusters (67%) of the  Population, for which orbital data are available, appear to be in retrograde orbits, compared to 2 out of 11 (18%) for clusters of the  Population. This suggests that the  Population may, at least in part, have formed in infalling fragments that were subsequently captured by the Galaxy. Van den Bergh (1993), Lee, Demarque & Zinn (1994) and Da Costa (1994) have argued that the young () Population of globulars represents objects that were associated with infalling "bits and pieces", whereas the old () Population of clusters - 14 - belongs to an older protogalactic structure that collapsed à la Eggen, Lynden-Bell & Sandage (1962). A scenario, such as that outlined above, would be consistent with the observation (Da Costa 1994) that clusters of the old halo appear, in the mean, to have direct motion (V = +58 ± 24 km s ), whereas rotation of the younger halo -1 rot (V = -45 ± 81 km s ) is marginally retrograde. -1 rot Figure 4 shows a plot of cluster half-light radius r versus perigalactic h distance P for members of the  Population. Such a correlation would not be expected if clusters of the  Population had been captured at random by the protogalaxy. A recent compilation of the most accurate ages of individual globular clusters by Richer et al. (1996) shows that eight clusters of the  Population have a mean age < T > = 13.8 ± 0.5 Gyr, while 11 clusters of the  Population are found to have < T > = 15.0 ± 0.4, i.e. the  clusters are, on average, younger than those belonging to the  Population. However, there appears to be a large intrinsic dispersion in the relation between globular cluster age and its distance from the fiducial relation shown in Fig. 3. This suggests (Fusi Pecci et al. 1996) that "second parameter" effects may be a function of both age and of some other, as yet unidentified, factor. - 15 - Figure 5 shows plots of M versus R for globular clusters with red [C  (B-R)/(B + V + R) < -0.80], and with blue and intermediate-color [-0.80  C  +1.00] horizontal branches, respectively. Intercomparison of the two panels of this GC V figure shows a significant difference between these two types of clusters in the outer halo (R > 10 kpc) of the Galaxy; but no obvious difference in the inner GC GC halo (R  10 kpc). For clusters with red horizontal branches that are located in the outer halo < M > = -4.82, which is almost ten times less luminous than the V value < M > = -7.30, that is found for outer halo clusters with blue and V intermediate-color horizontal branches. A Kolmogorov-Smirnov test shows that there is only a 0.1% chance that the red and blue HB clusters were drawn from the same parent population. At a given metallicity level globular clusters with red horizontal branches are believed to be younger than ones that have blue horizontal branches (Rood & Iben 1968, Rood 1973, but see Richer et al. (1996)). So the observed effect might be due to a decrease in the luminosity with which clusters are formed over time in the outer halo. Alternatively, this luminosity difference could be related to the fact that red horizontal branch clusters in the outer halo are, in the mean, more metal rich (< [Fe/H] > = -1.32) than are clusters with bluer horizontal branches (< [Fe/H] > = -1.70). Perhaps second generation globular clusters that formed in - 16 - the outer Galactic halo were, on average, both less luminous and slightly more metal-rich than those formed earlier. However, the low metallicity ([Fe/H] = - 1.69) of Ruprecht 106, which is the youngest known halo globular cluster (Kaluzny, Krzeminski & Mazur 1995), would appear to militate against such a simple scenario. Most dwarf spheroidal galaxies formed stars for a very extended period of time. Such dSph galaxies are observed to contain large numbers of carbon stars. Capture of these dSph galaxies would, therefore, be expected to enrich the Galactic halo in carbon stars. From the (uncertain!) estimates of the number of C stars in the halo, van den Bergh (1994b) estimated that no more than ~40% of the Galactic halo could have been produced by disintegration of Fornax-like dwarf spheroidal galaxies. It should, of course, be emphasized that capture and destruction of dwarf spheroidals that took 10 Gyr ago would not have contributed to the carbon star population of the Galactic halo. A much stronger constraint on such captures is set by observations of young blue stars in the Galactic halo (Unavane, Wyse & Gilmore 1996). Such blue objects with B - V  0.4 are younger than the vast majority of halo main sequence stars. Unavane et al. conclude that only ~1% of the halo could have been formed by accretion of dSph galaxies, like the Carina system, that contain a significant intermediate-age - 17 - population. Preston, Beers & Shectman (1994) use their observations of metal- poor blue horizontal branch stars to conclude that the accreted population is comparable to the total stellar content of all known dwarf spheroidal satellites of the Galaxy. Some blue stars found in the Galactic halo by Preston, Beers & Shectman (1994) might be the bluest members of a metal-poor intermediate-age population accreted from dwarf spheroidal satellites. Alternatively, a few moderately metal- poor halo A stars (Rodgers, Harding & Sadler 1981, Lance 1988) could perhaps have formed during collisions between metal-poor intergalactic clouds and the metal-rich gas in the disk of the Galaxy. Such metal-poor gas might also have been swept out of one or more dwarf irregular galaxies that collided with gas in the Galactic disk (Freeman 1996). If the Galactic globular cluster system did, as proposed by Lee (1993), van den Bergh (1993) and Zinn (1993), grow inside out, then the spheroidal cluster component of the Galaxy is presently larger than it was in the past. This contrasts with the situation for the system of open clusters which appears to have been more extended in the past than it is at the present epoch (Friel 1995, Hufnagel 1995). The region with R > 12 kpc is found to have produced numerous GC - 18 - clusters with ages in the range of 2 Gyr - 8 Gyr. However, no clusters with R >GC 12 kpc seem to have formed during the last 2 Gyr. Recent HST observations by Richer et al. (1996) appear to show that the outer halo clusters NGC 2419 and Pal. 3, situated at R ~100 kpc, have ages GC quite similar to those of globulars with R < 10 kpc. This has led Harris (1996) GC to conclude that a "coherent event" must have occurred in the Galactic halo ~15 Gyr ago. In this connection, it is of interest to note that the oldest globular clusters in the LMC appear to have ages that are similar to those of the oldest metal-poor globular clusters in the Galactic halo (Brocato et al. 1996). Taken at face value, these results appear to suggest that the first burst of cluster formation took place almost simultaneously throughout the Galactic halo and in the satellites of the Galaxy. 3. FORMATION OF THE GALACTIC BULGE The dense absorbing clouds in the direction of the Galactic center render the nuclear bulge of the Milky Way System almost invisible at visual wavelengths. However, it shows up prominently as a centrally peaked 20° x 15° (2.8 x 2.1 kpc) concentration of IRAS sources at a wavelength of 12 µm (Habing et al. 1985). Most of these infrared sources are possibly dust-embedded late M - 19 - giant stars. High metallicity increases the fraction of giants that become very cool M stars. The existence of a strong radial metallicity gradient in the inner Galaxy (Terndrup 1988, Frogel et al. 1990, Minniti et al. 1995) will, therefore, enhance the frequency of IR sources close to the Galactic nucleus. The distribution of very late M giants, therefore, presents a somewhat biased picture (King 1993) of the distribution of stars in the nuclear bulge of the Galaxy. An initially steep metallicity gradient in the nuclear bulge will be flattened (Friedli, Benz & Kennicutt 1994) by a central bar (Blitz & Spergel 1991). However, this trend might be partly compensated for by metals produced during ongoing starbursts. Such star formation presently takes place in the thin, fast rotating, nuclear disk (Dejonghe 1993) which has R < 150 pc. GC Baade's (1951) discovery of large numbers of RR Lyrae stars in the bulge of the Galaxy at first appeared to confirm the hypothesis (Baade 1944) that the nuclear bulge consisted of metal-poor stars of Population II. However, Morgan (1959) subsequently demonstrated that the dominant population of the Galactic nuclear bulge consists of strong-lined metal-rich stars. Radial velocity observations of RR Lyrae stars in Baade's Window (Gratton 1987) showed that these objects exhibit a large velocity dispersion (.  130 km s ), which clearly -1 - 20 - marks them as members of the halo population that are just passing through the central region of the Galaxy. A recent study by Minniti (1996) of a bulge field at 4 = 8°, b = +7° shows that metal-rich K giants participate in the Galactic rotation ( < V > = +66 ± 5 km s , < . > = 72 ± 4 km s ), whereas metal-poor [Fe/H] < - 1.5) halo giants ( < V > = -6 ± 20 km s , < . > = 114 ± 14 km s ) have a large velocity dispersion and do not participate in the rotation of the Bulge. The Bulge -1 -1 -1 -1 has a half-light radius r ~200 pc (Frogel et al. 1990). This is an order of h magnitude smaller than that of the halo. From an analysis of the K giants in Baade's window (4 = 3 *.9 Sadler, Rich & Terndrup (1996) find < [Fe/H] > = -0.11 ± 0.04, with more than , b = 1 *.0 ) half of the sample lying in the range -0.4 < [Fe/H] < +0.3. These values probably underestimate the true mean metallicity of bulge stars because (1) the line of sight towards Baade's window intersects the bulge at Z  -0.5 kpc and (b) the sample excludes M giants which will, on average, be more metal-rich than K giants. From studies of the integrated light of stars in Baade's window, Idiart, de Frietas Pacheco & Costa (1996) find [Mg/Fe] = +0.45 in the Galactic nuclear bulge. The very high metallicities of some bulge stars have recently been confirmed with Keck echelle spectra obtained by Castro et al. (1996). These - 21 - authors find [Fe/H] = +0.47 ± 0.17 for the star BW IV -167. This is similar to the value [Fe/H] = +0.46 ± 0.14 obtained for the nearby super metal-rich star µ Leonis. Minniti (1995) concludes that metal-rich globular clusters with R > 3 kpc belong to the Thick Disk, but that those having R < 3 kpc are GC GC kinematically associated with the bulge. The color-magnitude diagrams of these clusters also appear consistent with their assignment to the bulge population. From color-magnitude diagrams obtained with the Hubble Space Telescope (HST), Ortolani et al. (1996) conclude that the ages of the bulge clusters NGC 6528 and NGC 6553 do not differ by more than a few Gyr from that of the Thick Disk cluster 47 Tucanae. It has, however, been emphasized by Catalan & de Frietas Pacheco (1996) that such small age differences are rendered uncertain by possible differences in helium abundance and in the ratios of elements produced by SNe Ia and SNe II. From rather noisy color-magnitude diagrams that extend down to the main sequence turnoff in Baade's Window, Terndrup (1988) concluded that the bulk of the stars in the bulge have ages in the range of 11 - 14 Gyr. Furthermore, he found that the number of objects with ages < 5 Gyr is negligible. - 22 - The conclusion that disks and bulges of galaxies belong to distinct population components is supported by observations of M33. This galaxy has a well-developed halo containing globular clusters (Schommer et al. 1991) and RR Lyrae stars (Pritchet 1988), but appears to have little or no old bulge (Bothun 1991). In other words, a galaxy can have a halo but no bulge. However, the central region of M33 does contain evolved stars that are brighter than those in the Milky Way bulge (Minniti, Olszewski & Rieke 1993). Mighell & Rich (1995) suggest that such stars are associated with a relatively recent burst of star formation that took place well after the oldest stars were formed. Hartwick (1976) proposed that the Galactic disk was formed by gas ejected from halo stars. However, this suggestion appears difficult to reconcile with the high specific angular momentum of disk stars. More recently, Carney, Latham & Laird (1990) and Wyse & Gilmore (1992) have suggested that the Galactic bulge was formed from low angular momentum gas that was left over after most star formations had ended in the halo. The idea that halo stars were formed from leftover halo gas, which was enriched by SNe II on a short time- scale, appears to be supported by McWilliam & Rich (1994) who find that, compared to disk stars near the sun, [Mg/Fe] and [Ti/Fe] are (as is the case in halo stars) elevated by  0.3 dex. However, it is not clear why McWilliam & - 23 - Rich find [Ca/Fe] and [Si/Fe] in bulge stars to closely follow normal trends for Galactic disk giants. Sadler, Rich & Terndrup (1996) find < [CN/Fe] > to be close to solar in metal-poor stars in the bulge, whereas < [CN/Fe] > = -0.47 ± 0.04 in metal-rich stars. The weak CN lines in bulge stars indicate that these objects differ in some respects from stars in elliptical galaxies. An additional complication (Ratag et al. 1992) is that He and N in bulge planetary nebulae appear to be higher than they are in the Galactic disk. All of these results suggest that the evolutionary history of the Galactic bulge was probably complex. This conclusion is supported by the observation that the central region of the Galaxy presently contains much less gas than would have been ejected by the stars in the Galactic bulge during a Hubble time (van den Bergh 1957). Most of the gas lost by first-generation bulge stars was probably used up to form second-generation stars. Such second-generation stars could have incorporated elements produced on a relatively long time-scale by SNe Ia. In elliptical galaxies, and in the large bulges of Sa and Sb galaxies that collapsed rapidly, the absence of gas can probably be accounted for (Mathews & Baker 1971) by invoking galactic winds generated via gas heating caused by supernova blast waves. Sofu & Habe (1992) postulate that such bulges are themselves formed by star bursts in gas clouds ejected from the central regions of - 24 - galaxies. On the other hand, Sellwood (1993) has suggested that nuclear bulges might have formed as the result of bar-like instabilities in disks. However, Minniti (1995) questions whether the steep abundance gradients observed in the nuclear bulges of galaxies could have been formed (or maintained) if disks had first been stirred up by bars. It is still too early (Renzini 1993) to decide if any (or all) of the processes discussed above contributed to the formation of the Galactic bulge. Finally, it is noted that Lee (1992) believes the bulge of the Galaxy to be older than the halo. Renzini & Greggio (1990) have also argued that the bulge formed before the halo because the collapse time scale  = (G,) -½ is much longer for the halo than it is for the bulge. So the bulge may be older than the halo, even though it is metal-richer than the halo. A possible example of an infalling object is the globular cluster NGC 6287 (Stetson & West 1994). This cluster is very metal-poor ( [Fe/H] = -2.05) and is located at only 1.9 kpc (Djorgovski 1993) from the Galactic nucleus. Its small half-light radius of r = 1.3 pc suggests that it might be physically h associated with the central region of the Galaxy. However, Stetson & West point out that the relatively high radial velocity (V = -208 km s ) of NGC 6287 gives -1 it sufficient kinetic energy to travel out (or fall in from) as far as the Solar circle. - 25 - Some insight into the formation of spiral galaxies is provided by the morphology of individual spirals in the Hubble Deep Field (e.g. van den Bergh et al.1996). A good example is the spiral HDF 2-86. This object has a nucleus which is slightly orange in color, indicating the presence of some evolved stars. This nucleus is embedded in a disk (or flattened clustering) of presumably younger blue knots. This indicates that the nuclear bulge may already be present in a proto-spiral before assemblage of the disk is complete. 4. FORMATION OF THE GALACTIC DISK 4.1 Evolution of the Galactic disk The evolutionary relationships between the Galactic halo, the Thick Disk and the Thin Disk remain a subject of lively controversy. In particular, it is not yet clear whether the transition from the halo phase of Galactic evolution to the disk phase was continuous, or if there was an extended hiatus (Berman & Suchkov 1991) between the halo and disk stages of evolution. If there were such a hiatus, then it would no longer be possible to regard the Thick Disk as a structure that formed when pressure support started to build up at the beginning of the dissipational phase of Galactic evolution. It is not yet clear [see Majewski (1993) for an excellent review] whether the Thick Disk and the Thin Disk represent distinct evolutionary phases, or if there was a gradual transition between - 26 - the Thick Disk and Thin Disk eras of Galactic evolution. Observations by Oswalds & Risley (1961) show that short-period Mira variables exhibit halo kinematics, whereas longer period Miras seem to belong to a Thick Disk population. More detailed studies of Mira variables might, therefore, provide some insight into the nature of the Thick Disk/Thin Disk transition. Realistic numerical simulations, which take into account energy and angular momentum transfer to the halo (Barnes 1996), seem to show that the Thick Disk probably does not represent Thin Disk material that was heated by capture of an infalling satellite. [If this view is correct, then the existence of thin disks in many spirals no longer places strong constraints (Tóth & Ostriker 1992) on the rate at which disk galaxies capture companions]. The observation by Gratton et al. (1996) that [Mg/Fe]  +0.4 in the Thick Disk, but that it decreases by 0.2 dex at the Thick Disk/Thin Disk transition, also militates against the suggestion that the Thick Disk consists of dynamically heated Thin Disk stars. However, this conclusion depends critically on how the Thick Disk to Thin Disk transition is defined. Observations of a few nearby edge-on spirals (Morrison 1996) may indicate that only galaxies with central bulges exhibit Thick Disks. The reason for the possible existence of such a relationship between bulges and thick disks is not immediately obvious. The reality of a metal-poor (and hence very old) population component in the Galactic disk is presently in doubt (Twarog & Anthony-Twarog 1996). - 27 - Realistic merger calculations (Barnes 1996) suggest that it may not be possible to account for very metal-poor disk stars by invoking capture (Quinn & Goodman) of, and mergers with, dwarf galaxies. 4.2 The open cluster system Recently, Friel (1995) has studied 74 open clusters with ages larger or equal to that of the Hyades. For these disk clusters, she finds that: (1) clusters with R  7 kpc are absent; presumably because they have been destroyed by GC interactions with giant molecular clouds (van den Bergh & McClure 1980). (2) Open clusters (and, hence, the thin disk) exhibit a steep metallicity gradient with < [Fe/H] > _ 0.0 at R = 7 kpc and < [Fe/H] > _ -0.5 at R = 12.5 kpc. (3) At GC GC any value of R , open clusters show a range in metallicity of about 0.5 dex. (4) GC Over the range 7  R (kpc)  13 open clusters exhibit no evidence for a dependence of metallicity on age. Edvardsson et al. (1993) have recently obtained a similar result for element abundances (excluding Ba) in field stars with ages < 10 Gyr. The most straightforward interpretation of this result is that the heavy elements produced by supernovae are diluted by infall of more-or-less pristine gas into the disk. Presumably, such infalling gas will, on average, have zero angular momentum. Such infall will reduce the angular momentum of the Galactic disk and result in its radial contraction. Possible evidence for such infall - 28 - is also provided by the observation (Edvardsson et al. 1993) that the scatter in [Si/Fe] is about four times smaller than that in [Fe/H]. This is exactly what would be expected if disk stars form from gas that has not been well-mixed after infall of clouds with more-or-less pristine composition. Within the Galactic disk, Edvardsson et al. find that [ / Fe] decreases with increasing R . This suggests that the rate of star formation declined faster in the inner disk than it did at larger GC radii. For a detailed discussion of the chemical and stellar evolution of the Galactic disk the reader is referred to Prantzos & Aubert (1995). However, a problem is that constraints imposed by the disk oxygen abundance are uncertain. This is so because H II regions show a steep [O/H] gradient in the outer disk, whereas observations of B stars appear to exhibit no such gradient. 4.3 Age of the Galactic disk The Thick Disk of the Galaxy is observed to contain many RR Lyrae variables. Since RR Lyrae stars occur in the "young" globular cluster Ruprecht 106 (Kaluzny, Krzeminski & Mazur 1995), to which Richer et al. (1996) assign an age of 9.3 Gyr, it follows that the Thick Disk must have an age of at least 9 Gyr. This conclusion is marginally consistent with the fact that few carbon stars - 29 - (most of which are thought to have ages  10 Gyr) appear to have Thick Disk kinematics. An even greater age of 12 ± 2 Gyr has recently been obtained by Phelps et al. (1996) for the probably open cluster Berkeley 17. If one assumes that the disk formed after the end of the halo phase of Galactic evolution, then an upper limit on the age of the Galactic disk is set by the halo cluster M92, for which Bolte & Hogan (1995) derive an age of 15.8 ± 2.1 Gyr. Other normal halo globular clusters may be younger than M92, but have less well-determined ages. [Young halo clusters such as Rup 106 and Ter 7 probably had unusual evolutionary histories (van den Bergh 1996) and may have formed after the thick disk was assembled]. A weaker, but entirely independent, upper limit to the age of the Galactic halo is set by the thorium abundance in the ultra metal-poor star CS 22893-052, which yields an age of 16 ± 6 Gyr (Sneden et al. 1996). However, the work by Phelps, Janes & Montgomery (1994), and of Kaluzny, Krzeminski & Mazur (1995) shows that there may be some overlap between the ages of the oldest open clusters and those of the youngest globulars. In other words, the disk may have started to form before formation of the halo was completed. If all open clusters are members of the Thin Disk then a lower limit to the age of the Thin Disk is provided by NGC 6791, which is the oldest open cluster - 30 - with a well-determined age. Garnavich et al. (1993), Kaluzny & Rucinski (1995) and Tripicco et al. (1995) find ages of 7 - 10 Gyr for this object. The motion of this cluster, which lags circular motion by more than 60 km s (Scott, Friel & -1 Janes 1995) is, however, somewhat peculiar. An age similar to that of NGC 6791 is obtained from the calculated production ratios of the actinoid pairs U/ U 235 238 232 238 and Th/ U and their presently observed abundance ratios. From these values Chamcham & Hendry (1996) find that star formation in the solar neighborhood began at least 9 Gyr ago. An entirely independent lower limit to the age of the Galactic disk is provided by the colors and trigonometric parallax (Ruiz et al. 1995) of the white dwarf ESO 439-26, which is found to have M = +17.6 ± 0.1. This low V luminosity yields a cooling age of 6 - 7 Gyr. In summary, it appears that all presently available data appear consistent with ages of between 12 Gyr and 15 Gyr for the Thick Disk of the Galaxy. Berkeley 17, the oldest known "open" cluster may, on the basis of its kinematics (Scott, Friel & Janes 1995), be a member of the Thick Disk. The age of this cluster is estimated to be 12 ± 2 Gyr, which is consistent with the Thick Disk age limits given above. 5. SUMMARY AND CONCLUSIONS - 31 - It is now widely believed that the central region of the Milky Way System collapsed from a single protogalaxy à la ELS, while the outer part of the Galaxy is thought to have been assembled by infall (and subsequent capture) of protogalactic fragments, as envisioned by SZ. The observation that 6 out of 9 young globular clusters of -type have retrograde orbits, whereas only 2 out of 11 older -type clusters have retrograde orbits, appears consistent with this scenario. In the "Standard Model", the Galaxy formed inside out, with the bulge being old and the outer halo relatively young. However, the following problems are noted with this Standard Model: (a) The Standard Model does not account for the observation (see Fig. 1) that the half-light radii r of globular clusters grow with increasing h Galactocentric distance R . An even closer (and as yet unexplained!) GC relation (Fig. 2) exists between r and perigalactic distance P. It is h particularly puzzling (see Fig. 4) that young -type globular clusters in the halo, which should mainly be captured objects, appear to exhibit a close correlation between r and P. h (b) The globular clusters associated with the Fornax dwarf spheroidal galaxy are much smaller (see Fig. 2) than the globular clusters in the Galactic - 32 - halo. It follows that the halo cannot have been entirely assembled by capture of Fornax-like dSph systems. The frequency of blue intermediate- age stars in the halo also limits recent accretion of Carina-like dSph systems to ~1% of the total stellar population of the Galactic halo. (c) The periods of RR Lyrae stars in Galactic halo globular clusters exhibits a marked Oosterhoff dichotomy, with mean cluster periods < P > _ 0.55 days and < P > _ 0.65 days. No such dichotomy is ab ab observed among the RR Lyrae variables in globular clusters associated with the LMC. This suggests that the Galaxy did not merge with one or more galaxies resembling the Large Cloud during the course of its evolutionary history. The fact that only about half a dozen Galactic globular clusters are known to have ages as low as ~10 Gyr suggests that the number of mergers with Sagittarius-like dSph galaxies has probably been small. Weak limits on the number of recent mergers with dSph galaxies can also be set from the frequency of C stars in the Galactic halo. (d) Contrary to expectations from the Standard Model, recent HST observations of the clusters NGC 2419 and Pal. 3, at R ~100 kpc, yield ages that are similar to those of typical globular clusters at R < 10 kpc. - 33 - This suggests the possibility that a "coherent event" might have produced a burst of cluster formation in the Galaxy ~15 Gyr ago. (e) It is not clear (see Fig. 5) why globular clusters in the outer halo, that have red horizontal branches, are an order of magnitude less luminous than those that have bluer horizontal branches. Taken at face value, some of these results appear to weakly favor a scenario (Sandage 1989) in which mergers represent "noise" that is superposed on an ELS-like collapse model for the Galactic halo. It should, however, be emphasized that infall might have played a more important role in the evolutionary history of other giant galaxies. The observation that globular cluster radii r correlate with R , but that [Fe/H] does not, appears to favor a scenario in GC h which cluster sizes are set by global parameters, whereas their metallicities are determined by local factors. I thank Drs. Michael Bellazzini, Mike Bolte, Raffaele Gratton, Ron Marzke, Flavio Fusi Pecci, María Teresa Ruiz, Nick Suntzeff and Matt Wood for exchanges of views and useful information. I also wish to thank a particularly helpful referee. - 34 - TABLE 1 Comparison between luminosities of halo Population II and the integrated luminosity of globular clusters Galaxy globulars M (halo) V M (globulars) V Milky Way a -18.4 LMC a -15.1 a -13.0 a -10.9 "M V -5.4 -4.2 L (globulars) 0.7% 2.1% Fornax b -13.7: -8.8 c -4.9 1.1% M49 d -22.9 -17.7 d -5.2 0.8% a b c d Suntzeff (1992) van den Bergh (1995) Total luminosity Harris (1991) - 36 - REFERENCES Abraham, R.G., Tanvir, N.R., Santiago, B.X., Ellis, R.S., Glazebrook, K. & van den Bergh, S. 1996, MNRAS, 279, L47 Baade, W. 1994, ApJ, 100, 137 Baade, W. 1951, Pub. Obs. U. Michigan, 10, 7 Barnes, J.E. 1996 in Formation of the Galactic Halo, eds. H. Morrison and A. Sarajedini (San Francisco: ASP), 415 Bellazzini, M., Fusi Pecci, F. & Ferraro, F.R. 1996, MNRAS, 278, 952 Berman, B.G. & Suchkov, A.A. 1991, Ap. Space. Sci. 184, 169 Bernstein, G.M., Nichol, R.C., Tyson, J.A., Ulmer, M.P. & Wittman, D. 1995, AJ, 110, 1507 Blitz, L. & Spergel, D.N. 1991, ApJ, 379, 631 Bolte, M. & Hogan, C.J. 1995, Nature, 376, 399 Bothun, G.D. 1992, AJ, 103, 104 Brocato, E., Castellani, V., Ferraro, F.R., Piersimoni, A.M. & Testa, V. 1996, MNRAS, in press Buananno, R., Corsi, C.E., Fusi Pecci, F., Fahlman, G.G. & Richer, H.B. 1994, ApJ, 430, L121 Carney, B.W., Latham, D.W. & Laird, J.B. 1990, AJ, 99, 572 Castro, S., Rich, S.M., McWilliam, A., Ho, L.C., Spinrad, H., Filippenko, - 37 - A.V. & Bell, R.A. 1996, AJ, 111, 2439 Catalan, M. & de Freitas Pacheco, J.A. 1996, PASP (in press) Chamcham, K. & Hendry, M.A. 1996, MNRAS, 279, 1083 Cowie, L.L., Hu, E.M. & Songaila, A. 1995, Nature, 377, 603 Da Costa, G.G. 1994, in the Local Group = ESO Workshop Proceedings No. 51, eds. A. Layden, R.C. Smith and J. Strom (Garching: ESO), p. 101 Da Costa, G.G. & Armandroff, T.E. 1995, AJ, 109, 2533 Dejonghe, H. 1993, in Galactic Bulges = IAU Symposium No. 153, eds. H. Dejonghe and H.J. Habing (Dordrecht: Kluwer), p. 73 Djorgovski, S. 1993 in Structure and Dynamics of Globular Clusters = ASP Conf. Series No. 50, eds. S.G. Djorgovski and G. Meylan (San Francisco: ASP), p. 373 Driver, S.P., Phillips, S., Davies, J.I., Morgan, I. & Disney, M.J. 1994, MNRAS, 268, 393 Edvardsson, B., Andersen, J., Gustafsson, B., Lambert, D.L., Nissen, P.E. & Tomkin, J. 1993, A&A, 275, 101 Eggen, O.J., Lynden-Bell, D. & Sandage, A.R. 1962, ApJ, 136, 748 (= ELS) Fall, M.S. & Rees, M.J. 1977, MNRAS, 181, 37p Freeman, K.C. 1996, in Formation of the Galactic Halo, eds. H. Morrison and - 38 - A. Sarajedini (San Francisco: ASP), 3 Friedli, D., Benz, W. & Kennicutt, R. 1994, ApJ, 430, L105 Friel, E.D. 1995, ARA&A, 33, 381 Frogel, J.A., Terndrup, D.M., Blenco, V.M. & Whitford, A.E. 1990, ApJ, 353, 494 Fusi Pecci, F., Bellazzini, M., Ferraro, F.R., Buonanno, R. & Corsi, C.E. preprint, Astro-ph/9606108 Garnavich, P.M., VandenBerg, D.A., Zurek, D.R. & Hesser, J.E. 1993, AJ, 107, 1097 Gratton, R.G. 1987, MNRAS, 224, 175 Gratton, R.G. & Carretta, E., Matteucci, F. & Sneden, C. 1996, in Formation of the Galactic Halo, eds. H. Morrison and A. Sarajedini (San Francisco: ASP), 307 Habing, H.J., Olnon, F.M., Chester, T., Gillett, F., Rowan-Robinson, M. & Neugebauer, G. 1985, A&A, 152, L1 Harris, W.E. 1991, ARA&A, 29, 543 Harris, W.E. 1996, in Formation of the Galactic Halo, eds. H. Morrison and A Sarajedini (San Francisco: ASP), 231 Hartwick, F.D.A. 1976, ApJ, 209, 418 Hartwick, F.D.A. 1987, in The Galaxy, ed. G. Gilmore & B. Carswell - 39 - (Dordrecht: Reidel), p. 281 Hodge, P.W. & Michie, R.W. 1969, AJ, 74, 587 Hufnagel, B. 1995, PASP, 107, 1016 Hut, P. & Djorgovski, S. 1992, Nature, 359, 806 Ibata, R.A., Gilmore, G. & Irwin, M.J. 1994, Nature, 370, 194 Idiart, T.P., de Freitas Pacheco, J.A. & Costa, R.D.D. 1996, AJ, 113, 1169 Johnston, K.V., Spergel, D.N. & Hernquist, L. 1995, ApJ, 451, 598 Kaluzny, J., Krzeminski, W. & Mazur, B. 1995, AJ, 110, 2206 Kaluzny, J. & Rucinski, S.M. 1995, (preprint) King, I.R. 1993, in Galactic Bulges = IAU Symposium No. 153, eds. H. Dejonghe and H.J. Habing (Dordrecht: Kluwer), p. 3 Lance, C.M. 1988, ApJ, 334, 927 Lee, Y.-W. 1992, PASP, 104, 798 Lee, Y.-W. 1993, in The Globular Cluster - Galaxy Connection = ASP Conference Series Vol. 48, eds. G.H. Smith and J.P. Brodie (San Francisco: ASP), p. 142 Lee, Y.-W., Demarque, P. & Zinn, R. 1994, ApJ, 423, 248 Lin, D.C.N. & Richer, H.B. 1992, ApJ, 388, L57 Majewski, S.R. 1993, ARA&A, 31, 575 - 40 - Majewski, S.R., Munn, J.A. & Hawley, S.L. 1996, ApJ, 459, L73 Majewski, S.R., Hawley, S.L. & Munn, J.A. 1996, in Formation of the Galactic Halo, eds. H. Morrison and A. Sarajedini (San Francisco: ASP), 119 Mathews, W.G. & Baker, J.C. 1971, ApJ, 170, 241 McWilliam, A. & Rich, R.M. 1994, ApJS, 91, 749 Mighell, K.J. & Rich, R.M. 1995, AJ, 110, 1649 Minniti, D., Olszewski, E.W. & Rieke, M. 1993, ApJ, 410, L79 Minniti, D. 1995, AJ, 109, 1663 Minniti, D. 1996, ApJ, 459, 599 Minniti, D., Meylan, G. & Kissler-Patig, M. 1996, A&A, in press Morgan, W.W. 1959, AJ, 64, 432 Morrison, H. 1996, in Formation of the Galactic Halo, eds. H. Morrison and A. Sarajedini (San Francisco: ASP), 453 Norris, J.E. 1994, ApJ, 431, 645 Oosterhoff, P.T. 1939, Observatory, 62, 104 Ortolani, S., Renzini, A., Gilmozzi, R., Marconi, G., Barbuy, B., Bica, E. & Rich, R.M. 1995, Nature, 377, 701 Osvalds, V. & Risley, A.M. 1961, Publ. Leander McCormick Observatory, 11, 147 - 41 - Phelps, R.L., Janes, K.A., Friel, E.D. & Montgomery, K.A. 1996, in the Formation of the Milky Way, ed E.A. Alfaro (Cambridge: Cambridge University Press), in press Phelps, R.L., Janes, K.A. & Montgomery, K.A. 1994, AJ, 107, 1079 Prantzos, N. & Aubert, O. 1995, A&A, 302, 69 Preston, G.W., Beers, T.C. & Shectman, S.A. 1994, AJ, 108, 538 Pritchet, C.J. 1988, in the Extragalactic Distance Scale, ASP Conf. Series Vol. 4, eds. S. van den Bergh and C.J. Pritchet (Provo: ASP), p. 59 Quinn, P.J. & Goodman, J. 1986, ApJ, 309, 472 Ratag, M.A., Pottasch, S.R., Dennefeld, M. & Menzies, J.W. 1992, A&A, 255, 255 Renzini, A. & Greggio, L. 1990, in Bulges of Galaxies = ESO Workshop No. 35, eds. B. Jarvis & D. Terndrup (Garching: ESO), p. 47 Renzini, A. 1993, in Galactic Bulges = IAU Symposium No. 153, eds. H. Dejonghe & H.J. Habing (Dordrecht: Kluwer), p. 151 Richer, H.B. et al. 1996 ApJ, 463, 602 Rodgers, A.W., Harding, P. & Sadler, E. 1981, ApJ, 244, 912 Rodgers, A.W. & Paltoglou, G. 1984, ApJ, 283, L5 Rood, R.T. 1978, ApJ, 184, 815 Rood, R.T. & Iben, I. 1968, ApJ, 154, 215 - 42 - Ruiz, M.T., Bergeron, P., Leggett, S.K. & Anguita, C. 1995, ApJ, 455, L159 Sadler, E.M., Rich, R.M. & Terndrup, D.M. 1996, AJ, 112, 171 Sandage, A. 1990, J RASC, 84, 70 Schommer, R.A., Christian, C.A., Caldwell, N., Bothun, G.D. & Huchra, J. 1991, AJ, 101, 873 Scott, J.E., Friel, E.D. & Janes, K.A. 1995, AJ, 109, 1706 Searle, L. 1977, in The Evolution of Galaxies and Stellar Populations, eds. B.M. Tinsley and R.B. Larson (New Haven: Yale Observatory), p. 219 Searle, L. & Zinn, R. 1978, ApJ, 225, 357 (= SZ) Sellwood, J.A. 1993, in IAU Symposium No. 153 = Galactic Bulges, eds. H. Dejonghe and H.J. Habing (Dordrecht: Kluwer), p. 391 Sneden, C., McWilliam, A., Preston, G.W. & Cowan, J.J. 1996, in Formation of the Galactic Halo, eds. H. Morrison and A. Sarajedini (San Francisco: ASP), 387 Sofu, Y. & Hube, A. 1992, PASJ, 44, 325 Stetson, P.B. & West, M.J. 1994, PASP, 106, 726 Suntzeff, N.B. 1992, in the Stellar Populations of Galaxies = IAU Symposium No. 149, eds. B. Barbuy and A. Renzini (Dordrecht: Kluwer), p. 23 Suntzeff, N.B., Kinman, T.D. & Kraft, R.P. 1991, ApJ, 367, 528 Terndrup, D.M. 1988, AJ, 96, 884 - 43 - Toomre, A. 1977, in The Evolution of Galaxies and Stellar Populations, eds. B.M. Tinsley and R.B. Larson (New Haven: Yale Observatory), p. 401 Tóth, G. & Ostriker, J.P. 1992, ApJ, 389, 5 Tripicco, M.J., Bell, R.A., Dorman, B. & Hufnagel, B. 1995, AJ, 109, 1697 Twarog, B.A. & Anthony-Twarog, B.J. 1996, AJ, 111, 220 Unavane, M., Wyse, F.G. & Gilmore, G. 1996, MNRAS, 278, 727 van den Bergh, S. 1957, Zs. f. Astrophys., 43, 236 van den Bergh, S. 1975, ARAA, 13, 217 van den Bergh, S. & McClure, R.D. 1980, A&A, 80, 360 van den Bergh, S. 1993, ApJ, 411, 178 van den Bergh, S. 1994a, in the Local Group = ESO Workshop Proceedings No. 51, eds. A. Layden, R.C. Smith and J. Storm (Garching: ESO), p. 3 van den Bergh, S. 1994b, AJ, 108, 2145 van den Bergh, S. 1995, AJ, 110, 1171 van den Bergh, S. 1996, in preparation van den Bergh, S., Abraham, R.G., Ellis, R.S., Tanvir, N.R., Santiago, B.X. & Glazebrook, K. 1996, AJ, 112, xxx Wyse, R.F.G. & Gilmore, G. 1992, AJ, 104, 144 - 44 - Zinn, R. 1993, in the Globular Cluster - Galaxy Connection = ASP Conference Series No. 48, eds. G.H. Smith and J.P. Brodie (San Francisco: ASP), p. 38 - 45 - FIGURE CAPTIONS Fig. 1 Half-light radii of Galactic globular clusters versus Galactocentric distance. The radii of globulars are seen to increase with distance from the Galactic center. Globular clusters in the Fornax dwarf are shown as horizontal arrows. Note that the Fornax clusters are more compact than most globular clusters in the outer halo of the Galaxy. Fig. 2 Half-light radii of Galactic globular clusters versus cluster perigalactic distances. The Fornax clusters (open circles) are plotted at the 100 ± 40 kpc perigalactic distance of this dwarf spheroidal galaxy. The Fornax clusters are seen to be smaller than Galactic globulars at such large perigalactic distances. Fig. 3 Metallicity versus horizontal branch population gradient in the inner (top) and outer (bottom) regions of the Galactic halo. Clusters that fall close to the fiducial line are assigned to the  Population. Clusters below (or to the left of) this line belong to the (possibly younger)  Population. Fig. 4 Relation between the perigalactic distance P (van den Bergh 1995) and - 46 - half-light radius r (Djorgovski 1993) for young metal-poor halo globular h clusters belonging to the  Population (van den Bergh 1993). After excluding the only collapsed-core cluster in the sample (shown as a cross), it is found that the correlation coefficient between r and P is r = 0.95 ± h 0.03: . Such a strong correlation between cluster radius and perigalactic distance is not expected for a scenario in which such clusters were captured, more or less at random, from the neighborhood of the Galaxy. Fig. 5 Integrated magnitudes versus Galactocentric distances for clusters with red horizontal branches having C < -0.80 (bottom) and for clusters with C  -0.80 (top). For R > 10 kpc clusters with red horizontal branches are seen to be almost ten times fainter than those with bluer horizontal branches.
astro-ph/0301343
1
0301
2003-01-17T15:41:15
Clues on the Evolution of Cluster Galaxies From The Analysis of Their Orbital Anisotropies
[ "astro-ph" ]
We study the evolution of galaxies in clusters by the analysis of a sample of about 3000 galaxies, members of 59 clusters from the ESO Nearby Abell Cluster Survey (ENACS). We distinguish four cluster galaxy populations, based on their radial and velocity distributions within the clusters. Using the class of ellipticals and S0's (excluding the very bright ellipticals), we determine the average cluster mass profile, that we compare with mass models available from numerical simulations. We then use this cluster mass profile to solve for the anisotropy profiles of the three other cluster galaxy populations, viz. the very bright ellipticals, the early spirals, and the late spirals with the emission-line galaxies. We discuss the implications of our findings for the evolution of cluster galaxies.
astro-ph
astro-ph
To appear in "Galaxy Evolution: Theory and Observations (2002)" RevMexAA(SC) CLUES ON THE EVOLUTION OF CLUSTER GALAXIES FROM THE ANALYSIS OF THEIR ORBITAL ANISOTROPIES A. Biviano1 P. Katgert2 T. Thomas2 A. Mazure3 We study the evolution of galaxies in clusters by the analysis of a sample of ∼ 3000 galax- ies, members of 59 clusters from the ESO Nearby Abell Cluster Survey (ENACS, Kat- gert et al. 1998), for which redshifts, R-band magnitudes, as well as morphologies are avail- able (Thomas 2003; Biviano et al. 2002, B02 hereafter, and references therein). In order to make the most efficient use of our data, we combine the 59 clusters into a single ensem- ble cluster as described in B02. In the ensemble clus- ter, after excluding galaxies in substructures, we find that there are 4 cluster galaxy populations that must be distinguished because they have different phase- space distributions: (i) the brightest ellipticals, with MR ≤ −22 (using H0 = 100 km sec−1Mpc−1), (ii) the other ellipticals together with the S0 galaxies (we refer to this class as E + S0 hereafter), (iii) the early spirals (Sa -- Sb), and (iv) the late spirals and irregu- lars (Sc -- Ir) together with the emission-line galaxies (ELG's). About 2/3 of all cluster galaxies (outside sub- structures) belong to the E + S0 class. The shape of the E + S0 velocity distribution indicates that these galaxies move on nearly isotropic orbits, β ≈ 0 (see also van der Marel et al. 2000). We can therefore use E + S0 as isotropic tracers of the cluster gravita- tional potential. We solve the Abel and Jeans equa- tions (see, e.g., Binney & Tremaine 1987) using both a direct non-parametric approach, and the inverse method described by van der Marel (1994). We find that a NFW (Navarro, Frenk, & White 1997) mass profile with rs/r200 = 0.25+0.15 −0.10 (68% confidence lim- its) provides a very good fit to our data. We use this mass profile to estimate the anisotropy profiles for the other three cluster galaxy populations, using the method of Solanes & Salvador-Sol´e (1990). We do not find any acceptable solution for the brightest ellipticals, most likely because these galax- ies do not fulfil the conditions for the applicability of 1INAF -- Osservatorio Astronomico di Trieste, via G.B. Tiepolo, 11 -- 34131 Trieste, Italy. 2Leiden Sterrewacht, P.O. Box 9513 NL-2300 RA The Netherlands. 3OAMP, LAM, Traverse du Siphon-Les trois Lucs 13012 Marseille, France. the collisionless Jeans equations. As a matter of fact, the brightest ellipticals mostly sit at the bottom of the cluster potential well, and move very slowly, if at all (B02). They have probably been slowed down by dynamical friction, and could have grown by mergers of other massive galaxies (Brough et al. 2002). We do find acceptable solutions for the early spi- rals. We cannot exclude fully isotropic orbits for these galaxies, but the data taken at face value in- dicate that in the inner cluster region they move on radially-anisotropic orbits (β ≈ 0.6). Since there is evidence that these galaxies evolve into S0's (Thomas & Katgert 2003), it is possible that the early spirals that are still visible near the cluster center are those that have managed to avoid trans- formation, by the amplitude and direction of their velocities. We also find acceptable solutions for the class of late spirals + ELG's. The anisotropy is close to zero in the center, but there are not many galaxies (if any) of this class there. For radii ≥ 0.5 r200 the anisotropy grows almost linearly, reaching β ≈ 0.6 at a radius ∼ 1.5 r200, which is the limit of our observa- tional data. Such an anisotropy profile suggests that the late spirals + ELG's are field galaxies infalling into the cluster. The lack of these galaxies in the central cluster region suggests that they get trans- formed (into dwarf galaxies) or destroyed, once they reach the high density central regions of the clusters. REFERENCES Binney J., Tremaine S., 1987, Galactic Dynamics, Prince- ton University Press, Princeton (New Jersey) Biviano, A., Katgert, P., Thomas, T., & Adami, C. 2002, A&A, 387, 8 (B02) Brough, S. et al. 2002, MNRAS, 329, L53 Katgert, P., Mazure, A., den Hartog, R., Adami, C., Bi- viano, A., & Perea, J. 1998, A&AS, 129, 399 Navarro, J.F., Frenk, C.S., & White, S.D.M., 1997, ApJ 490, 493 Solanes, J.M., & Salvador-Sol´e E., 1990, A&A, 234, 93 Thomas, T. 2003, A&A, submitted Thomas, T. & Katgert, P. 2003, A&A, submitted van der Marel, R. 1994, MNRAS, 270, 271 van der Marel, R., et al. 2000, AJ, 119, 2038 1
astro-ph/9410091
1
9410
1994-10-28T19:11:44
Chaos, Regularity, and Noise in Self-Gravitating Systems
[ "astro-ph", "gr-qc", "nlin.CD" ]
This paper summarises a number of new, potentially significant, results, obtained recently by the author and his collaborators, which impact on various issues related to the gravitational N-body problem, both Newtonianly and in the context of general relativity. Topics addressed include: (1) direct N-body simulations and their interpretation, with reference to the observed exponential instability towards small changes in initial conditions and the phenomenon of ``nonviolent relaxation;'' (2) the Hamiltonian structure of the collisionless Boltzmann equation of general relativity, i.e., the Vlasov-Einstein system; (3) ``transient ensemble dynamics,'' i.e.,,the short time statistical characterisation of collections of orbits in nonintegrable mean field potentials; and (4) the structural stability of the smooth potential approximation typically used in galactic dynamics.
astro-ph
astro-ph
an invited plenary talk at: The Seventh Marcel Grossmann Meeting July 1994 CHAOS, REGULARITY, AND NOISE IN SELF-GRAVITATING SYSTEMS HENRY E. KANDRUP Department of Astronomy and Department of Physics and Institute for Fundamental Theory, University of Florida, Gainesville, FL 32611, USA Abstract This paper summarises a number of new, potentially significant, results, obtained recently by the author and his collaborators, which impact on various issues related to the gravitational N-body problem, both Newtonianly and in the context of general relativity. 4 9 9 1 t c O 8 2 1 v 1 9 0 0 1 4 9 / h p - o r t s a : v i X r a 1 Introduction and motivation The overall objective of the research reported herein is the application of ideas and techniques from modern nonlinear dynamics and nonequilibrium statistical mechanics to self-gravitating systems, both Newtonianly and in the context of general relativity, with particular emphasis on the gravitational N-body problem. The basic motivation for this research is a desire to identify some of the physical processes which can play a role in determining the structure and evolution of self-gravitating systems. The results described here will, for specificity, typically be formulated in the language of galactic dynamics. However, it should be evident that they also have potential implications in other settings as well, including, e.g., statistical quantum field theory in the early Universe. It should be stressed that the principal focus here is not on the detailed modeling of any specific class of astronomical object where, in particular, other nongravitational effects, such as dissipative hydrodynamics, can be important. However, the results reported here should find relatively direct applications to the study of systems like elliptical and lenticular galaxies, which are known to be gas poor, albeit not as gas poor as they were ten years ago. In approaching the study of self-gravitating systems, there are several different approaches which one might adopt. At the most fundamental level, one can attack the full N-body problem, either by performing and analysing numerical simulations or by proving (hopefully useful) mathematical theorems which provide insights into the qualitative character of the evolution. In either case, the focus here is not on solar system type problems, involving a relatively small number of masses, one or two of which are much larger than the others. Rather, the principal focus is on collections of large numbers N of objects, comparable in mass, in particular the problem of the N → ∞ limit. Conventional wisdom says that, in this large N limit, such a collection of compa- rable masses can be described by a collisionless Boltzmann equation, i.e., what the mathematician would term the Vlasov-Poisson or Vlasov-Einstein system. Such a description involves the assumption that, Newtonianly, particles follow trajectories in the self-consistent gravitational potential associated with the average mass distribu- tion. Relativistically, one supposes that the particles follow geodesics in a spacetime, the form of which is determined by the stress energy tensor associated with the av- erage mass distribution. In this general context, two things need to be done, namely (1) to determine precisely the conditions under which such a mean field description is justified and then (2) to understand the qualitative and quantitative implications of this description. 1 In this connection, it is important to stress that, although the Vlasov-Poisson system was first formulated by Jeans in the context of galactic dynamics[1] more than twenty years before it was formulated in plasma physics,[2] the gravitational equation is substantially less well motivated than is the plasma analogue. In particular, the absence of shielding prevents a systematic derivation except for the special case of cosmological systems, which are nearly homogeneous and of finite age.[3] The problem becomes especially acute for the case of relativistic systems, the point being that the derivation of the corresponding Vlasov-Maxwell system relies heavily on the linearity of Maxwell's equations,[4] whereas Einstein's equation is nonlinear. A yet simpler tact involves the consideration of orbits in a fixed potential. The idea here is to specify one's favourite potential, chosen to represent the bulk potential of some physical object, to study in detail the form of orbits in this potential, and only at the end of the day to incorporate the fact that the potential must be determined self-consistently by the mass distribution associated with the orbits themselves. If one chooses to focus simply on an average potential, in the context of a Vlasov description or otherwise, there remains the important question of determining pre- cisely, both qualitatively and quantitatively, what sorts of effects have been ignored. In other words, there remains the problem of structural stability. Only to the extent that these additional effects have been appropriately identified and understood can one say that one has a satisfactory understanding of so-called "collisionless stellar dynamics." The aim of this talk is to illustrate each of the preceding aspects by explaining several new results that have been derived during the past three or four years. Sec- tion 2 focuses on the full Newtonian N-body problem. Section 3 then turns to the collisionless Boltzmann equation. Section 4 addresses several issues related to the problem of orbits in a fixed potential, and Section 5 concludes with a discussion of some aspects of the problem of structural stability. 2 The gravitational N -body problem Detailed numerical simulations over the past thirty years have established that, given generic initial conditions corresponding to a bound configuration, a self-gravita-ting system of point masses will typically exhibit a rapid approach, on a characteristic crossing time tcr, towards a statistical quasi-equilibrium where, in a time-averaged sense, the system only shows subsequent systematic variability on substantially longer timescales. Moreover, there has evolved a substantial and detailed conventional wis- dom which serves, at least roughly, to determine what kinds of initial data give rise to 2 what kinds of final quasi-equilibria. However, despite these impressive successes, one does not completely understand this process. Indeed, at a truly fundamental level there is no real understanding of why this happens. To obtain some insights into this basic question, it is natural to identify various microscopic and mesoscopic phenomena which could perhaps conspire to produce the qualitative macroscopic evolution that is observed in numerical experiments. The aim here is to identify two such phenomena. 1. Viewed microscopically in the many particle phase space, Newtonian N-body sim- ulations exhibit an exponentially sensitive dependence on the specific choice of initial conditions. Specify unperturbed initial data, {ru A(0)}, (A = 1, ..., N), and per- form a simulation. Then specify perturbed initial data, {rp A(0)}, and repeat. What one discovers thereby is that, generically, quantities like the total N-particle configuration space perturbation A(0) pp A(0) pu δrA(t)2 ≡ N XA=1 N XA=1 rp A(t) − ru A(t)2 (1) will typically grow exponentially in time t on a relatively short time scale, even though the unperturbed and perturbed evolution are essentially identical at the macroscopic level. This fact was first observed by S. Ulam in the early 1960's, and the classic paper on the subject is by Miller.[5] However, only in the last several years, with the advent of improved computers, has this instability been studied systematically in complete and gory detail.[6, 7, 8, 9, 10] The net result of these investigations is that this instability is an exceedingly ro- bust phenomenon, which proceeds generically on a characteristic crossing time tcr, independent of many/most details. In particular, the timescale associated with this instability is independent of the detailed choice of initial data and the detailed choice of the initial perturbations, as well as the specific diagnostics used to quantify the growth of the perturbation -- e.g., configuration or momentum space perturbation, the total N-particle perturbation or the perturbation of "typical" or "representa- tive" particles. The timescale is also insensitive to a possible distribution of masses, provided that everything is not dominated by a few particularly massive particles. More significantly, the simulations also suggest strongly that the rate is insensi- tive to the total particle number N, provided at least that N ≫ 2. Thus, e.g., for 200 ≤ N ≤ 4000 one observes no appreciable changes if everything is scaled in terms of an N-dependent characteristic time tcr. In particular, there is no sense in which the instability appears to "turn off" in the limit of large N. Indeed, Goodman et al[9] have predicted that the instability should accelerate for large N, with the character- 3 istic timescale t∗ scaling as t∗ ∼ tcr/ln N. Interestingly, tcr is the same timescale on which generic initial data evolve towards a macroscopic quasi-equilibrium. 2. Gravitational N-body simulations evidence a considerable "memory." Viewed mi- croscopically or mesoscopically, the quasi-equilibrium does not correspond to a par- ticularly well-shuffled state. Many aspects of the initial data for individual particles are forgotten, but other aspects are in fact remembered. The strongest correlation between initial and final conditions, from which all others may perhaps derive, is between the initial and final values of the binding energy. Both for isolated sys- tems approaching a quasi-equilibrium and for collisions between pairs of objects, e.g., spherical polytropes and other axisymmetric or triaxial near-equilibria, there is a strong correlation between the initial and final values of the binding energy. Specifically, one observes that particles initially with high binding energy tend to end up with high binding energy, low with low, and intermediate with intermediate, even for an evolution that involves rapid, violent changes in the bulk potential, so that there is no obvious sense in which the binding energy should behave as an adiabatic invariant.[11, 12, 13, 14] This phenomenon can be quantified at a coarse-grained mesoscopic level, through various binnings of the orbital data.[13, 14] Alternatively, it can be quantified at the completely microscopic level through the computation of a rank correlation between the initial and final orderings of particles in terms of their binding energies.[14] Such a computation shows that there exist strong, albeit not complete, correlations between initial and final conditions. The absence of a complete correlation is at least partly "collisional" in origin, but appears to persist even in the limit of large N, where, according to conventional wisdom, the system should be essentially collisionless in character. To summarise: In the gravitational N-body problem one is confronted with a sys- tem that exhibits a rapid macroscopic evolution towards a statistical quasi-equilibrium. Viewed microscopically, this evolution is characterised by an exponentially sensitive dependence on the specific choice of initial conditions. However, despite this sensi- tive dependence, the quasi-equilibrium does not correspond, either microscopically or mesoscopically, to a particularly well-shuffled state. 3 The collisionless Boltzmann equation The principal message of this section is that the collisionless Boltzmann equation, i.e., a mean field Vlasov description, is a constrained Hamiltonian system. This fact was first established by Morrison[15] for the electrostatic Vlasov-Poisson system, and is equally true for the analogous gravitational Vlasov-Poisson system. In this setting, the 4 fundamental dynamical variable is the one-particle distribution function, f (x, p, m), evaluated at a fixed time t, which is interpreted as a number density of particles of mass m at the spatial point x with conjugate momentum p. The gravitational po- tential Φ(t) is viewed as a functional of f (t), constructed in terms of an appropriate Green function. The phase space is then the infinite-dimensional space of distribu- tion functions. The Hamiltonian character of the evolution is manifest through the identification of a Hamiltonian function H[f ] and a cosymplectic structure { . , . }, in terms of which the Vlasov equation takes the form ∂f /∂t = {f, H}. A generalisation to the spherically symmetric Vlasov-Einstein system is completely straightforward.[16] The crucial point is that, for spherical systems, the gravitational degrees of freedom are not triggered: given appropriate boundary conditions, one can view the spacetime metric gab at any given time t as a functional of the distribution function f at that same t. The full Vlasov-Einstein system is substantially more complicated, since, in the general case, one must incorporate the field degrees of freedom. However, the analysis still turns out to be straightforward, at least in principle.[17] The basic formulation is analogous to the Hamiltonian formulation of the Vlasov-Maxwell system,[18, 19] the nonlinearity of the Einstein equation not playing a significant role. Working in the context of the ADM formulation of general relativity, there are now three different dynamical variables, each defined on t = const hypersurfaces, namely (1) the distribution function, f (x, p, m), (2) the spatial three-metric, hab(x), and (3) the conjugate field momentum, Πab(x). The natural phase space is the infinite- dimensional space coordinatised by these three variables. In this case the cosymplectic structure given as the sum of two pieces, namely (1) the functional Poisson bracket of vacuum gravity and (2) the matter bracket appropriate for the spherically symmetric Vlasov-Einstein system (which coincides also with the bracket for the Vlasov-Poisson system). Explicitly, for two functions F [f, hab, Πab] and G[f, hab, Πab], hF, Gi = 16πZ d3x δF δhab δG δΠab − δG δhab δF δΠab! +Z d3xd3pdm f "δF δf , δG δf #, (2) where δ/δX denotes a functional derivative with respect to the variable X and [f, g] = ∂f ∂xi ∂g ∂pi − ∂g ∂xi ∂f ∂pi (3) denotes an ordinary three-dimensional Poisson bracket. To give meaning to variations δX, one requires a rule identifying particle coordi- a} in two nearby cotangent bundles. In the context of this nates {xa, pa} and {x′a, p′ 5 HM = Z dΓ f E = Z Nh1/2d3x T t t. (6) 3+1 formulation, it is natural to identify spatial coordinates and conjugate momenta, as well as time t and mass m, i.e., x′ = x p′ = p t′ = t m′ = m. (4) However, other choices are also possible.[20, 21] The Hamiltonian H = HG + HM = R d3xHG +R d3xHM is also given as the sum of two pieces, namely (1) the ADM Hamiltonian HG of vacuum gravity, i.e., 1 HG = 16π h1/2(Nh −(3) R + h−1(ΠabΠab − −2Nb[Da(h−1/2Πab)] + 2Da(h−1/2NbΠab)) 1 2 Π2)i (5) and (2) a matter Hamiltonian HM constructed as the integral of the local energy density, i.e., Here dΓ = d3xd3p dm denotes the covariant seven-dimensional volume element on a t = const hypersurface, Da a covariant derivative on the hypersurface, and E(x, p, m) = pt the particle energy. N and Na correspond respectively to the lapse function and shift vector. This formulation reproduces the Vlasov-Einstein system in the sense that the equation ∂F/∂t = hF, Hi for arbitrary functions F [f, hab, Πab] is equivalent to the Vlasov-Einstein system in its usual 3 + 1 form: The distribution function f satisfies ∂f /∂t = [E, f ], and ∂hab/∂t and ∂Πab/∂t both satisfy the appropriate 3 + 1 Einstein equations sourced by T a b[f ]. For the spherically symmetric case, with the metric viewed as a functional of f , the first term in the bracket hF, Gi disappears and the Hamiltonian H reduces to the ADM mass, HADM , realised as a volume integral in the natural fashion facilitated by Schwarzschild coordinates. Such a Hamiltonian formulation of the collisionless Boltzmann equation is in fact an example of a much more general result. Specifically, consider any Hamiltonian the- ory for a system with multiple degrees of freedom, and then construct the associated mean field description appropriate in the limit that correlations amongst the degrees of freedom are negligible (i.e., a statistical description in which the full N-"particle" distribution function is approximated as factorising into a product of reduced one- "particle" distribution functions). There is then a precise mathematical sense in which the mean field description of this Hamiltonian system is itself Hamiltonian.[22] The fact that the collisionless Boltzmann equation, or any other mean field theory, is Hamiltonian is significant in that a Hamiltonian evolution is much more restricted 6 than a non-Hamiltonian evolution. However, of more pragmatic importance per- haps is the fact that this Hamiltonian formulation permits, for the first time, a clear geometric approach to the problem of stability for general equilibrium solutions to the Vlasov-Einstein system. Here an "equilibrium solution" {f0, h0 0 } entails a stationary matter distribution, corresponding to a spacetime that admits a timelike Killing field. ab, Πab The crucial fact facilitating this geometric approach is that every such equilib- rium is an energy extremal, so that the first variation δ(1)H vanishes identically for perturbations {δf, δhab, δΠab} which satisfy the constraints. The field constraints are enforced by restricting attention to perturbed initial data for which δH/δN = δH/δN a = 0. The matter constraints, a reflection of Liouville's Theorem, im- ply that the perturbed distribution function must be generated from the unper- turbed f0 via a canonical transformation in terms of some generating function h, i.e., f0 + δf = exp([h, . ])f0 = f0 + [h, f0] + 1 The fact that the equilibrium is an energy extremal implies that stability hinges on the sign of the second variation δ(2)H. Indeed, modulo infinite-dimensional technical- ities the situation is analogous to the problem of stability for mechanical Hamiltonian systems. If δ(2)H > 0 for all infinitesimal perturbations, linear stability is guaran- teed. Alternatively, if δ(2)H is indeterminate, one cannot necessarily infer a linear instability, but one does expect at least nonlinear instability and/or instability in the presence of dissipation.[23] 2hh, [h, f0]i + ..... Indeed, one can actually prove that generic rotating axisymmetric equilibria are always unstable towards dissipation, as provided, e.g., by the emission of gravitational radiation.[24] This is the collisionless analogue of the theorem[25] that all rotating perfect fluid stars are unstable. Neither for collisionless nor collisional systems is there any guarantee that the timescale associated with this instability is sufficiently short to be of interest astronomically. However, it is significant that general relativity triggers a generic instability which, apparently, is insensitive to the form of the self- gravitating matter. The astronomical implications of this instability are currently under investigation. This general approach to stability can also be adapted to the consideration of steady-state equilibria, such as an homogeneous and isotropic Friedman cosmology, where it provides an interesting derivation of the Jeans instability.[26] Viewed New- tonianly, such an expanding Universe corresponds in the "inertial" frame to a sys- tem characterised by a time-independent Hamiltonian which finds itself in a time- dependent steady state. This explicit time-dependence can be removed by effecting a time-dependent canonical transformation into the average "comoving" frame. This 7 transformation leads to a new time-dependent Hamiltonian H(t). However, the first variation δ(1)H vanishes identically, and the second variation δ(2)H can be shown to satisfy dδ(2)H/dt ≤ 0. A simple Liapounov argument therefore guarantees that the existence of negative energy perturbations δ(2)H < 0 for sufficiently long wavelengths must imply an instability: If the system be perturbed in such a fashion that δ(2)H < 0, the energy can only become more negative, so that the "distance" from equilibrium, as probed by the magnitude of δ(2)H, can only increase. 4 Transient Ensemble Dynamics In studying the properties of orbits in a fixed potential, it is natural to apply the technology of nonlinear dynamics, as has been developed over the past several decades. However, in many settings involving gravitational systems, the utilisation of this technology may require a new twist. The standard formulation of nonlinear dynamics typically involves a theory of asymptotic orbital dynamics, which focuses primarily, if not exclusively, on the long time behaviour of individual orbits. However, for many astronomical systems this may not be appropriate. For example, in terms of their natural timescale, galaxies are relatively young objects, only ∼ 100 − 200 crossing times tcr in age, so that it is not at all obvious that an asymptotic t → ∞ limit is well motivated physically. Thus, e.g., standard estimates of Liapounov exponents typically require integrations for times t ≥ 104tcr, a period that is orders of magnitude longer than the age of the Universe, tH. Moreover, it is arguably true that, in many astronomical systems, individual orbits are not the fundamental objects of interest. It is, for example, obvious that one cannot track individual orbits of stars within a galaxy. All that one can detect are properties like the overall brightness distribution which reflect the contributions of many stars. Similarly, it is evident that one must focus on collections of orbits if he or she wishes ultimately to address the problem of self-consistency. For these reasons, it would seem more natural to consider instead a theory of transient ensemble dynamics, which focuses on the statistical properties of ensembles of orbits, restricting attention exclusively to short timescales, t < tH, and recognising that much of the observed behaviour may be intrinsically transient in character. This distinction may not be all that important for integrable, or near-integrable potentials, which contain only regular orbits. However, there is no reason to assume that the bulk potential associated with a self-gravitating system is integrable, or even near-integrable, and there are indications from numerical experiments that objects like rotating elliptical galaxies and barred spiral galaxies may admit large numbers 8 of stochastic, or chaotic, orbits.[27, 28] Roughly, regular orbits correspond to orbits that have regular shapes and are characterised by simple topologies, e.g., box orbits in two dimensions with trajectories that resemble a Lissajous figure or tube orbits in three dimensions that are restricted to a region with the topology of a torus. By contrast, stochastic orbits are manifestly irregular in shape and appear to "run all over" phase space. Unlike regular orbits, stochastic orbits exhibit an exponentially sensitive dependence on initial conditions, as manifest by the fact that such orbits are characterised by a positive Liapounov exponent.[29] The crucial point to be illustrated below is that, at least when considering stochas- tic orbits, the transient ensemble perspective can prove extremely important. Consider, for example, the scattering of photons incident on a multi-black hole system; or similarly, consider a star moving in a nonspherical potential, supposing that the star is unbound but that, owing to the shape of the potential, it can only escape in certain directions. For each of these problems, one discovers that, in certain phase space regions, the direction and time of escape to infinity exhibits a very sensitive dependence on initial conditions, in fact a fractal dependence, this being an example of what the nonlinear dynamicist would call chaotic scattering.[30] Naively, one might anticipate that this complex microscopic behaviour would lead to an equally complex description when considering the evolution of ensembles of orbits. However, this is not always the case. At least in certain cases, one observes instead striking regularities, which lead to a simple scaling behaviour,[31] that may actually be universal.[32] The very fact that the microscopic evolution is complex appears to be responsible for the fact that the macroscopic evolution is very simple. 1 As a concrete example, consider orbits in the two-dimensional potential V (x, y) = 2(x2 + y2) − ǫx2y2, holding fixed the value of the energy E and studying as a function of ǫ the evolution of orbits initially localised in a small phase space region. For ǫ below a critical value ǫ1 = 1/(4E), escape is impossible energetically. For values of ǫ slightly above ǫ1, escape is possible energetically, but only very few particles escape on short timescales and the time of escape exhibits no striking regularities, except that the escape probability eventually appears to decay towards zero. However, for ǫ above another critical value ǫ2 (only determined numerically), one sees the onset of striking scaling behaviour: 1) After the decay of initial transients, the escape probability per unit time approaches a constant value P∞(ǫ), which is independent of initial conditions, i.e., the location of the phase space region from which the initial ensemble was chosen. Moreover, this rate scales as P∞(ǫ) ∼ (ǫ − ǫ2)α for a critical exponent α. 9 2) For fixed size of the initial phase space region probed by the ensemble, the time T∞ required to converge to P∞ also depends on ǫ and satisfies T∞(ǫ) ∼ (ǫ − ǫ2)−β. 3) For fixed ǫ, the convergence time T∞ depends on the linear size R of the phase space region that was probed initially, satisfying T∞(R) ∼ R−δ Moreover, to within statistical errors α − β − δ ≈ 0. In a certain sense, the qualitative change in behaviour at ǫ = ǫ2 is like a phase transition, complete with a critical "slowing down" as the "order parameter" ǫ − ǫ2 → 0. As another example in which the transient ensemble perspective is important, consider the behaviour of ensembles of orbits of fixed energy E, evolving in a two- dimensional time-independent potential V (x, y), which admits both regular and sto- chastic orbits. If V (x, y) is bounded from below and diverges at infinity, the constant energy hypersurfaces will be compact, so that the notion of "equilibrium" is well defined. One might therefore anticipate that generic ensembles of initial conditions will evolve towards an invariant distribution corresponding to a statistical equilibrium. To test this hypothesis, one can select localised ensembles of initial conditions of fixed E, corresponding to stochastic orbits initially located far from any regular phase space regions, and then evolve these initial data into the future. At least for certain potentials,[33, 34] one then observes that the orbits do indeed disperse in such a fashion as to exhibit a coarse-grained evolution towards a quasi-equilibrium, which is at least approximately time-independent. This approach is, moreover, exponential in time and characterised by a rate Λ(E) which is independent of the specific choice of initial data. The characteristic timescale t∗ = Λ−1 is typically ≪ 100tcr, so that, in "physical units" for a galaxy, t∗ ≪ tH . One also observes that the rate Λ(E) is comparable in magnitude to the value of the Liapounov exponent χ(E), which[35] probes the average rate of instability exhibited by orbits of energy E. There is, moreover, a direct correlation between Λ and χ in the sense that, e.g., both have similar curvatures when viewed as functions of E. This is particularly tantalising in view of the fact that, for the N-body problem, the timescale associated with the approach towards a statistical quasi-equilibrium is comparable in magnitude to the timescale associated with the instability towards small changes in initial conditions. Despite these regularities, there is no guarantee that this apparent equilibrium coincides with the true invariant distribution, and, in general, it will not! Astronomers are well acquainted with the fact that collisionless equilibria do not constitute true N- body equilibria, which must incorporate discreteness effects that become important on sufficiently long timescales. However, the key point here is different, and more fundamental: Even motion in a smooth two-dimensional potential may not yield a 10 uniform approach towards a true statistical equilibrium.[34, 36] The explanation of the discrepancy between the true and approximate equilibria is simple. Viewed over sufficiently long timescales, there are only two different classes of orbits, namely regular orbits, with vanishing Liapounov exponent χ, and stochastic orbits, for which χ > 0. The distinction between these two classes is, moreover, absolute, since members of the two different classes are separated by invariant KAM tori. If, e.g., one were to compute a surface of section, plotting coordinates x and px for successive intersections of the y = 0 hyperplane, he or she would generically find islands of regularity embedded in a surrounding stochastic sea. However, this is not the whole story. Lurking in the shallows of the stochastic sea, slightly away from the shore, are cantori,[37] these corresponding to fractured KAM tori, associated with the breakdown of integrability, which contain a cantor set of holes. The point then is that these cantori serve as partial barriers that divide the stochastic orbits into two subclasses, namely confined, or sticky, stochastic orbits which are trapped near the regular islands, and unconfined, or filling, stochastic orbits which travel unimpeded throughout the rest of the stochastic sea. Because of the holes in the cantori, these barriers are not absolute, so that orbits can in fact change from one class to another via so-called intrinsic diffusion. However, this process is a slow one, requiring orbits to wend their way through a maze (cf. the "turnstile model" of MacKay, Meiss, and Percival,[38] so that the characteristic timescale is typically ≫ 100tcr, i.e., much longer than the age of the Universe. What this implies is that, on short times, ensembles of orbits initially outside the cantori will evolve towards a near-invariant distribution which uniformly populates the filling regions, but avoids the confined regions. The situation is analogous to the classical effusion problem. Consider two evacuated cavities connected one with another by an extremely narrow conduit, and suppose that gas is inserted into one of the cavities. If the conduit be sufficiently narrow, the timescale on which gas effuses from one cavity to the other will be much longer than the timescale on which the gas spreads to fill the original cavity. This implies, however, that, even though the true equilibrium corresponds to a uniform density concentration throughout both cavities, one can speak meaningful of a shorter time quasi-equilibrium, in which the original cavity is populated uniformly and the other is essentially empty. Significantly, these two different populations of stochastic orbits are fundamentally dissimilar in terms of their stability properties as well as where they are located in phase space. Although both sticky and filling stochastic orbits are exponentially unstable, there is a precise sense in which the sticky orbits are less unstable overall than are the filling orbits. Specifically, if one computes local Liapounov exponents,[39] 11 χ(∆t), for different ensembles of stochastic orbits, integrating for some relatively short interval ∆t, he or she will find[34, 36] that the typical χ(∆t) for a sticky orbit is substantially smaller than the typical χ(∆t) for a filling orbit. Indeed, the composite distribution of local Liapounov exponents (i.e., distribution of instability timescales) generated from a sampling of the true invariant measure appears to be given, at least approximately, as a sum of two different near-Gaussian distributions with unequal means. It should perhaps be noted explicitly that the general conclusions recounted in this Section have been observed for several different potentials, with rather different symmetries, including (1) the dihedral D4 potential of Armbruster, Guckenheimer, and Kim,[40] (2) the sixth order truncation of the three-particle Toda[41] lattice potential, and (3) a generalised anisotropic Kepler potential of the form 1 m (7) V (x, y) = − (cid:16)1 + x2 + y2(cid:17)1/2 − (cid:16)1 + x2 + ay2(cid:17)1/2 , with constant m and a, for E < 0. The fact that these diverse potentials, which are fundamentally different in appearance, yield similar conclusions, both qualitatively and semi-quantitatively, would suggest strongly that these conclusions are robust, depending only on such topological features as the existence of KAM tori and cantori. The existence of confined stochastic orbits is of potential importance astronomi- cally because such orbits can help (the theorist) support various sorts of structures, e.g., bars in a spiral galaxy. It is natural to assume that, in systems like galaxies, regular orbits serve to provide the skeleton to support various structures. However, because of resonance overlap one may find that, near corotation and other resonances, the desired regular orbits do not exist, even though sticky stochastic orbits are present. Finally, it should be stressed that one can observe similar short time "zones of avoidance" in higher dimensional systems as well. The key point physically is that just because a region of phase space is connected, so that orbits can pass throughout the entire region, does not mean that all of the region will be accessed on comparable timescales. 5 Structural stability of the smooth potential ap- proximation The collisionless Boltzmann equation is a Hamiltonian system which neglects various realistic non-Hamiltonian irregularities that must be present in any self-gravitating system. One obvious point is that such a Vlasov description neglects entirely all 12 discreteness effects, i.e., "collisions," by idealising the system as a continuum, rather than a collection of nearly point mass objects. Viewed in the N-particle phase space, the statistical description of an isolated N-body evolution is of course Hamiltonian. However, when projected into the reduced one-particle phase space, any allowance for particle-particle correlations that transcend a mean field description necessar- ily breaks the Hamiltonian constraints.[22] Another point, perhaps less obvious but equally important, is that a Vlasov description also neglects any couplings to an ex- ternal environment. In the past, astronomers have been wont oftentimes to pretend that galaxies exist in splendid isolation but, over the past several decades, it has become increasing evident that such an approximation may not be justified.[42] Detailed modeling of these sorts of perturbing influences may prove extremely complex. In particular, an external environment can give rise to a variety of different effects characterised by a broad range of timescales. Those influences proceeding on timescales ∼ tcr will be particularly complicated, in that the details of their effects may depend very sensitively on the details of the environment. However, there is a well-established paradigm in statistical physics,[43, 44] dating back to the begin- ning of the century,[45] which would suggest that irregularities proceeding on shorter timescales, ≪ tcr, can oftentimes be modelled as friction and noise, related via a fluctuation-dissipation theorem. This idea underlies, for example, Chandrasekhar's original formulation[46] of so-called "collisional stellar dynamics." It is therefore natural to investigate the structural stability of Hamiltonian tra- jectories towards the effects of friction and noise. This was done[47, 48] by effecting large numbers of Langevin simulations, in which the deterministic equations of motion were perturbed by allowing for (1) a dynamical friction −ηp, which serves system- atically to remove energy from the orbits and (2) random kicks, modeled as white noise with temperature, or mean squared velocity, Θ, which serve systematically to pump energy back into the orbits. As a first simple test, η was assumed to be con- stant, in which case the fluctuation-dissipation theorem implies that the noise must be additive, rather than multiplicative.[50, 51] Thus, in units with particle mass m = 1, one is led explicitly to equations of motion of the form dr dt = p and dp dt = −∇V (r) − ηp + F, (8) where F is characterised completely by its statistical properties. Here, e.g., component by component, hFi(t)i = 0 and hFi(t1)Fj(t2)i = 2ΘηδijδD(t1 − t2), (9) 13 where the angular brackets denote a statistical average. The idea is to effect large numbers of different realisations of the same initial conditions, and to analyse these realisations to extract statistical properties. It is well known that even very weak friction and noise will eventually become important on sufficiently long timescales. In particular, one knows that, on the natural timescale tR ∼ η−1, these effects will try to force the system to evolve towards a thermal state. The question of relevance here is quite different: Can the friction and noise have substantial effects already on much shorter timescales ≪ tR? The conventional wisdom in astronomy is that the answer to this is: no! For example, the standard assumption that "collisionality" is irrelevant in a galaxy relies completely on the observation that the natural timescale associated with binary encounters is much longer than the age of the Universe.[46] The Langevin simulations were effected for total times t ≤ 200tcr, and involved friction and noise corresponding to a broad range of characteristic timescales, 103 ≤ tR/tcr ≤ 1012. The most significant conclusions derived from these simulations are the following: When viewed in terms of the collisionless invariants, i.e., the quantities that are conserved in the absence of the friction and noise, these perturbing influences only serve to induce a classical diffusion process, with the unperturbed and perturbed orbits diverging significantly only on a timescale tR ∼ η−1. Thus, in particular, δE2 rms ≡ hEunp − Eper2i = A(E)EΘηt, (10) where A(E) is a slowly varying function of E with magnitude of order unity. In this sense, the conventional wisdom is confirmed. However, when viewed in configuration space or momentum space, the effects are more complicated, and actually depend on orbit class. Unperturbed and perturbed regular orbits only diverge as a power law in time, i.e., δrrms, δprms ∼ tp, so that, once again, one only gets macroscopic deviations after a time tR ∼ η−1. However, unperturbed and perturbed stochastic orbits diverge exponentially at a rate set by the Liapounov exponent χ, so that, even for very weak friction and noise, one gets macroscopic deviations within a few crossing times. In particular, when considering ensembles of stochastic initial conditions, one observes a simple scaling δrrms, δprms ∼ (Θη)1/2exp[+X(E)t], (11) where X is comparable to, but slightly larger than, the Liapounov exponent χ. This exponential divergence is easy to understand[52] and the functional depen- dence on Θ, η, and χ can actually be derived theoretically.[53] The obvious point is 14 that the unperturbed deterministic trajectory is an unstable stochastic orbit, so that even the tiniest perturbing influences will tend to grow exponentially. The average rate of instability is given by χ and, as such, moments like δrrms should grow at a rate X ∼ χ. That X is slightly larger than χ is a reflection of the fact that, for different noisy realisations, one sees somewhat different local Liapounov exponents, and that the total δrrms will be dominated by those noisy realisations for which the rate of instability is above average. This argument might suggest that, although the unperturbed and perturbed tra- jectories exhibit a rapid pointwise divergence, their statistical properties should be virtually identical. Specifically, one might anticipate that, on short times, the only ef- fect of the friction and noise is to continually displace the trajectory from one stochas- tic orbit to another with essentially the same statistical properties. This, however, is false. Under certain circumstances, even very weak friction and noise can also alter the statistical properties of ensembles of stochastic orbits on relatively short times ≪ 100tcr. Specifically, one observes that such perturbing influences can dramatically accelerate the rate of penetration through cantori by providing an additional source of extrinsic diffusion. Provided that the friction and noise are sufficiently weak, on short timescales the energy E is almost conserved, so that one can speak meaningfully of an evolution restricted to an "almost constant energy hypersurface." Suppose now that, for some energy E, this hypersurface contains large measures of both sticky and filling stochas- tic orbits. If, for this energy, the near-invariant distribution described in Section 4 is evolved into the future, allowing for even very weak friction and noise, one then observes a rapid (t ≪ 100tcr) systematic evolution towards a new noisy near-invariant distribution which is (1) quite different from the deterministic near-invariant distribu- tion and (2) much closer to the true deterministic invariant distribution. In this sense, it appears that, on timescales short compared the timescale tR on which the system would evolve towards a thermal state, the principal effect of the friction and noise is to accelerate the approach towards a deterministic invariant distribution which, in the absence of these perturbing influences, would only have been realised on much longer timescales. The key point in all of this is that friction and noise can induce changes in orbit class, from filling to confined stochastic, and vice versa. Moreover, when the deter- ministic invariant distribution contains large measures of both sticky and filling orbits, such transitions can happen within a time t < 100tcr, even for very weak friction and noise. Visual inspection of ∼ 2.5 × 104 orbits in several different potentials leads to the following conclusions. 15 Typically, for a relaxation time tR as long as 1012tcr, not many such changes are observed within a time t ∼ 100tcr. However, if tR be reduced to a value ∼ 109tcr, transitions begin to become more frequent, and, even for tR as large as ∼ 106tcr , transitions are quite common, occuring for > 50% of orbits within a time t ∼ 100tcr. If the amplitude of the friction and noise are further increased, one finds that, for tR ∼ 103tcr , transitions are so common that the distinction between filling and con- fined becomes essentially meaningless. The distinction between regular and stochastic is more robust. Only for tR as small as ∼ 103tcr are significant numbers of transitions between regular and stochastic orbits observed within a time as short as t ∼ 100tcr. The fact that even very weak friction and noise, with tR ∼ 106tcr − 109tcr, can significantly alter the statistical properties of ensembles of orbits on timescales t < tH ∼ 100tcr has direct astronomical implications since, e.g., for galaxies, the timescale tR for discreteness effects, i.e., "collisionality" is ∼ 106tcr! The natural timescale associated with external perturbations is less easily estimated, but may well be even shorter. To summarise, it is evident that even very weak friction and noise can alter both the pointwise and the statistical properties of stochastic orbits in a nonintegrable potential on relative short timescales ≪ tR. In particular, such effects may be man- ifest already on time scales much shorter than the time on which numerical errors in a simulation can accumulate. This fact has direct and immediate implications for the problem of "shadowing" for numerical orbits.[54, 55] Physicists, mathemati- cians, and astronomers are often worried[56, 57] about whether numerical simulations performed on a computer, which incorporate roundoff and/or truncation error, can correctly shadow the evolution of some model system described by a simple set of de- terministic differential equations. However, it would also seem relevant[58] to worry about whether the "real world," replete with other sorts of irregularities, can shadow either the model system or its numerical realisations. In this regard, one final remark is in order: Rather than being an impediment to realistic modeling, in certain cases numerical noise may actually be a good thing, in that it may capture, at least quali- tatively, some of the effects of small perturbing influences to which real systems are always subjected. Acknowledgments It is a pleasure to acknowledge useful collaborations with Robert A. Abernathy, Bren- dan O. Bradley, George Contopoulos, Salman Habib, Eric O'Neill, Haywood Smith, Jr., Christos V. Siopis, David E. Willmes, and, especially, M. Elaine Mahon. This 16 research was supported in part by the NSF through PHY92-03333 and by NASA through the Florida Space Grant Consortium. The simulations reported herein were facilitated by time made available by the Advanced Computing Laboratory at Los Alamos (CM5), The Parallel Computing Research Laboratory at the University of Florida (KSR), and the Northeast Regional Data Center (Florida) by the IBM Corp. References [1] Jeans, J. H. Mon. Not. R. Astr. Soc. 76 (1915) 70. [2] Vlasov, A. A. Zh. Eksp. Teor. Fiz. 8 (1938) 291. [3] Bisnovatyi-Kogan, G. S. and Shukhman, I. G. Zh. Eksp. Teor. Fiz. 82 (1982) 3. [4] Klimontovich, Yu. L. 1983, The Kinetic Theory of Electromagnetic Processes (Springer, Berlin, 1983). [5] Miller, R. H. Astrophys. J. 140 (1964) 250. [6] Kandrup, H. E. and Smith, H. Astrophys. J. 374 (1991) 255. [7] Kandrup, H. E. and Smith, H. Astrophys. J. 386 (1992) 635. [8] Kandrup, H. E., Smith, H., and Willmes, D. E. Astrophys. J. 399 (1992) 627. [9] Goodman, J., Heggie, D., and Hut, P.Astrophys. J. 415 (1994) 715. [10] Kandrup, H. E., Mahon, M. E., and Smith, H. Astrophys. J. 428 (1994) 458. [11] van Albada, T. S. Mon. Not. R. Astr. Soc. 201 (1982) 939. [12] Quinn, P. J. and Zurek, W. H. Astrophys. J. 332 (1988) 619. [13] Funato, Y., Makino, J., and Ebisuzaki, T. Pub. Astron. Soc. Japan 44 (1992) 291. [14] Kandrup, H. E., Mahon, M. E., and Smith, H. Astron. Astrophys. 271 (1993) 440. [15] Morrison, P. J. Phys. Lett. A80 (1980) 383. [16] Kandrup, H. E. and Morrison, P. J. Ann. Phys. (NY) 225 (1993) 114. 17 [17] Kandrup, H. E. and O'Neill, E. Phys. Rev. D49 (1994) 5115 . [18] Marsden, J. E. and Winstein, A. Physica D4 (1982) 394. [19] Kandrup, H. E. and O'Neill, E. Phys. Rev. D48 (1993) 4534. [20] Ipser, J. R. and Thorne, K. S. Astrophys. J. 154 (1968) 251. [21] Israel, W. and Kandrup, H. E. Ann. Phys. (NY) 152 (1985) 30. [22] Kandrup, H. E. Phys. Rev. D 50 (1994) 2425. [23] Bloch, A., Krishnaprasad, P. S., Marsden, J. E., and Ratiu, T. S. Ann. Inst. Henri Poincar´e: analyse non lin´eaire 11 (1994) 37. [24] Kandrup, H. E. Astrophys. J. 380 (1991) 511. [25] Friedman, J. L. and Schutz, B. F. 1978, Astrophys. J. 222 (1978) 281. [26] Kandrup, H. E. and O'Neill, E. Phys. Rev. D47 (1993) 3229. [27] Sparke, L. S. and Sellwood, J. A. Mon. Not. R. Astr. Soc. 225 (1987) 663. [28] Pfenniger, D. and Friedli, D. Astron. Astrophys. 252 (1991) 75. [29] Chirikov, B. Phys. Repts. 52 (1979) 265. [30] Smilansky, U. in Lectures at Les Houches, Chaos and Quantum Physics, ed. M.-J. Giannoni, A. Voros, and J. Zinn-Justin (Elsevier, Amsterdam, 1990). [31] Contopoulos, G., Kandrup, H. E., and Kaufmann, D. Physica D64 (1993) 310. [32] Siopis, C. V., Contopoulos, G., and Kandrup, H. E. Ann. NY Acad. Sci. (1994) in press. [33] Kandrup, H. E. and Mahon, M. E. Phys. Rev. E49 (1994) 3735. [34] Mahon, M. E., Abernathy, R. A., Bradley, B. O., and Kandrup, H. E., Mon. Not. R. Astr. Soc. (1994) submitted. [35] Bennetin, G., Galgani, L., and Strelcyn, J.-M. Phys. Rev. A14 (1976) 2338. [36] Kandrup, H. E. and Mahon, M. E. Astron. Astrophys. (1994) in press. [37] Mather, J. N. Topology 21 (1982) 457. 18 [38] MacKay, R. S. Meiss, J. D. and Percival, I. C. Phys. Rev. Lett. 52 (1984) 697. [39] Grassberger, P., Badii, R., Politi, A. J. Stat. Phys. 51 (1988) 135. [40] Armbruster, D., Guckenheimer, J., and Kim, S. Phy. Lett. A 140 (1989) 416. [41] Toda, M. J. Phys. Soc. Japan 22 (1967) 431. [42] Zepf, S. E. and Whitmore, B. C. Astrophys. J. 418 (1993) 72. [43] Chandrasekhar, S. Rev. Mod. Phys. 15 (1943) 1. [44] Kubo, R., Toda, M., and Hashitsume, N. 1991, Statistical Physics II: Nonequi- librium Statistical Mechanics (Springer, Berlin, 1991). [45] Einstein, A. 1905, Ann. d. Physik 17 (1905) 549. [46] Chandrasekhar, S. Principles of Stellar Dynamics (University of Chicago, Chicago, 1942). [47] Kandrup, H. E. and Mahon, M. E. Ann. NY Acad. Sci. (1994) in press.. [48] Habib, S., Kandrup, H. E., Mahon, M. E. Phys. Rev. Lett. (1994) submitted. [49] Habib, S., Kandrup, H. E., Mahon, M. E. Mon. Not. R. Astr. Soc. (1994) submitted. [50] Lindenberg, K. and Seshadri, V. Physica A109 (1981) 481. [51] Habib, S. and Kandrup, H. E. Phys. Rev. D46 (1992) 5303. [52] Pfenniger, D. Astron. Astrophys. 165 (1986) 74. [53] Kandrup, H. E. and Willmes, D. E. Astron. Astrophys. 283 (1994) 59. [54] Anosov, D. V., Tr. Mat. Inst. Steklov 90 (1982) 210. [55] Bowen, R. J. Diff. Eq. 18 (1975) 333. [56] Quinlan, G. D. and Tremaine, S. Mon. Not. R. Astr. Soc. 259 (1992) 505. [57] Willmes, D. E. Ann. NY Acad. Sci. (1994) in press. [58] Eubank, S and Farmer, D., in 1989 Lectures in Complex Systems, ed. E. Jen (Addison-Wesley, Redwood City, California, 1990). 19
astro-ph/0204365
1
0204
2002-04-22T13:57:44
Optical and Radio monitoring of S5 1803+74
[ "astro-ph" ]
The optical (BVRI) and radio (8.4 GHz) light curves of S5 1803+784 on a time span of nearly 6 years are presented and discussed. The optical light curve showed an overall variation greater than 3 mag, and the largest changes occured in three strong flares. No periodicity was found in the light curve on time scales up to a year. The variability in the radio band is very different, and shows moderate oscillations around an average constant flux density rather than relevant flares, with a maximum amplitude of $\sim$30%, without a simultaneous correspondence between optical and radio luminosity. The optical spectral energy distribution was always well fitted by a power law. The spectral index shows small variations and there is indication of a positive correlation with the source luminosity. Possible explanations of the source behaviour are discussed in the framework of current models.
astro-ph
astro-ph
Optical and Radio monitoring of S5 1803+784 R. Nesci, E. Massaro, M. Maesano, F. Montagni, S. Sclavi Dipartimento di Fisica, Universit´a La Sapienza, P.le A. Moro 2, 00185, Roma, ITALY [email protected] T. Venturi, D. Dallacasa Istituto di Radioastronomia CNR via P. Gobetti 101, I-40129 Bologna, Italy [email protected] and F. D'Alessio Osservatorio Astronomico di Roma, via di Frascati 33, Monteporzio Catone I-00040, Italy [email protected] ABSTRACT The optical (BVRI) and radio (8.4 GHz) light curves of S5 1803+784 on a time span of nearly 6 years are presented and discussed. The optical light curve showed an overall variation greater than 3 mag, and the largest changes occured in three strong flares. No periodicity was found in the light curve on time scales up to a year. The variability in the radio band is very different, and shows moderate oscillations around an average constant flux density rather than relevant flares, with a maximum amplitude of ∼30%, without a simultaneous correspondence between optical and radio luminosity. The optical spectral energy distribution was always well fitted by a power law. The spectral index shows small variations and there is indication of a positive correlation with the source luminosity. Possible explanations of the source behaviour are discussed in the framework of current models. Subject headings: BL Lacertae objects: individual (S5 1803+784), Galaxies: Ac- tive, Galaxies: Photometry – 2 – 1. Introduction The radio source S5 1803+784 was classified as a BL Lac object by Biermann et al. (1981). A redshift estimate (0.680) was derived from a weak MgII line by Lawrence et al. (1996) and recently confirmed by Rector & Stocke (2001). From the available flux measurements, although non simultaneous, the overall energy distribution from radio to the X rays (in the Log (νFν) vs Log (ν) plot) peaks in the IRAS band (Giommi et al. 1995), so that the source is classified as a Low Energy Peaked BL Lac (LBL) object in the scheme proposed by Padovani & Giommi (1995). Like other sources of this class, one would expect in the optical band a remarkable variability with a correlation between the spectral slope and the luminosity (see e.g. Massaro et al. 1999; Nesci et al. 1998). The published photometric data, however, are very few: Wagner et al. (1990) and Heidt & Wagner (1996) reported only uncalibrated luminosity variations relative to a few weeks, and practically nothing is known about the long term time behaviour of the source. Only in the radio band S5 1803+784 was extensively monitored, both from the group of the University of Michigan (Aller et al. 1998) and from the Metsahovi observatory (Teraesranta et al. 1998), while several VLBI images have been obtained at different epochs since 1979 (Biermann et al. 1981), because it is normally used as a reference radio source for geodynamics studies (see Britzen et al. 1999; Gabuzda 1999, 2000). In April 1996 we started an optical monitoring program to define the long term light curve and color variations of this source. A parallel radio monitoring program at 5 GHz and 8.4 GHz was carried out starting from January 1996 with the two 32-m antennas located in Medicina (Bologna, Italy) and Noto (Siracusa, Italy), as part of a larger project to provide radio lightcurves of several bright BL Lac objects, possible targets of X-ray observations (Venturi et al. 2001). In this paper we present the observational results of this monitoring and discuss some implications of our measures on the nature of the luminosity changes. The observational data are described in Section 2; the light-curve temporal analysis is discussed in Section 3; the optical spectral distribution is discussed in Section 4; the general discussion of the results is made in Section 5. 2. Observations 2.1. Optical observations The large majority of our observations were made with a 50 cm f/4.5 telescope at the Vallinfreda astronomical station near Rome (850 m a.s.l.), equipped with a CCD camera based on the Texas TC241 chip. A minor number of data were collected with a 32 cm f/4.5 – 3 – telescope located at Greve (Tuscany, 650 m a.s.l.) and a 70 cm f/8.3 telescope at Monte Porzio (Roma, 380 m a.s.l.) both equipped with the same model of CCD camera, based on the SITe SIA501A back-illuminated chip. The filter sets of all these telescopes are identical and match the standard B,V (Johnson) and R,I (Cousins) bandpasses. The typical exposure times were 300 s for the V,R,I bands and 600 s for the B band. When the source was faint (for R>16 mag) exposure times were doubled. Systematic differences between the data sets of the three observatories were checked during several observational campaigns of BL Lac objects using simultaneous, or nearly simultaneous observations, and found to be negligible. For S5 1803+784, in particular, we have two nights in common between Vallinfreda and MontePorzio, two nights between Vallinfreda and Greve, and one night between Greve and Vallinfreda. Bias, dark and flat field corrections of the frames were performed with IRAF tasks, and aperture photometry was made with IRAF-apphot, using an aperture of 5 arcsec radius. Local background level was estimated from an annular region concentric to each star. Magnitudes of S5 1803+784 were estimated differentially with respect to five nearby reference stars. The final results, listed in Table 3, are the average of the magnitude differ- ences with respect with the reference stars, and the quoted error is the standard deviation of the single measures. The intercalibration of these five stars, in each filter, was made using the best 35 nights. Their magnitude differences agreed with a standard deviation of 0.01 magnitudes in each filter. No evidence of variability of these stars was detected in all our database. The zero point calibration for each photometric band was made linking the brightest reference star (Star A in Table 1) to the photometric sequences of five other BL Lac objects, namely PKS 0422+004 (Miller et al. 1983); AO 0235+164 (Smith et al. 1985, McGimsey et al. 1976, Rieke et al. 1976; 3C66A (Fiorucci & Tosti 1996); S5 0716+714 (Ghisellini et al. 1997); BL Lacertae (Bertaud et al. 1969, Fiorucci & Tosti 1996). For this purpose the best photometric nights were used, with air masses ranging from 1.0 to 1.5. The number of nights used for each filter is listed (in round bracketts) in Table 1. For each filter and night a linear fit of the extinction law was derived and used to convert the instrumental magnitudes into the apparent ones. The final result is the average over the different nights, and the error is the standard deviation of the mean. Given the large errors found in deriving the calibration of the reference stars in the B band, a new calibration of star A was made using four Bright Stars (HR 6598, 6636, 6637 and 6811). Their B magnitudes were taken from the Bright Star Catalogue, 5th Revised Edition, as given at the CDS (Hoffleit & Warren 1991). The results are reported in Table 1: the errors quoted for star A include the zero point uncertainty, while those for the other stars are the – 4 – intercalibration standard errors of the mean, derived from the average of 35 images for each colour. The finding chart is shown in Fig. 1. From the digitized Palomar sky survey plates and using our photometric sequence, we also derived a few approximate historical magnitudes for S5 1803+784, which are collected in Table 2. 2.2. Optical light curve In total we collected data for 207 days, 85 with full BVRI photometry and further 49 with VRI photometry, since April 1996 until January 2002. The whole data set is given in Table 3 and the R light curve is shown in Fig. 2: typical errors are 0.02-0.03 mag when the source was bright and 0.06 when very faint. Until JD 2451350 (Sep. 1999), the source behavior can be simply described as a mono- tonic increase of the luminosity, with small (0.3 mag) amplitude and relatively fast variations above or below the average level, with three strong bursts at JD 711, 1127 and 1344 (here- after we report the JD dates minus 2450000) of comparable amplitude and time scale. In the last event the source reached its highest recorded level (R=14.0). Then a decay phase started, with still some oscillations overimposed, bringing the flux of S5 1803+784 at the lowest recorded level (R∼17.4, July 2001). After that a new rising phase started. The overall magnitude interval is therefore about 3.4 mag, confirming a large optical variability, typical of this class of BL Lac objects. The three largest bursts indicated above are reported enlarged in Fig. 3, 4 and 5. For ease of comparison, the same scales were used for all three Figs. The V−I colour index is also plotted, with the scale marked on the right hand side. The first episode (Fig. 3) was observed only in the decreasing branch and showed a monotonic decline of 1.4 mag in 20 days, with a mean rate of 0.07 mag/day. Because of the sampling only a lower limit of 0.036 mag/day can be set to the rate of the rising branch. In the second one (Fig. 4), one can see a rise by 1.5 mag in 21 days (0.071 mag/day). The subsequent decrease was initially fast, with approximately the same time scale of the increase; after, our data indicate a change of the slope. In the third event (Fig. 5) the source reached its brightest state and likely it remained so bright for about one month, then a decline phase started of about 1.7 mag in 78 days. The greatest variation rate, derived from this inspection of the main events, is therefore 0.07 mag/day, observed both in rising and in decaying segments. Given the uneven sampling – 5 – of the light curve we cannot exclude rates greater than this estimate, but such rates might have lasted only on very short (day) time scale. We think however that very fast variations, if present, do not represent the normal behavior of the source and the above estimate can be used to give constraints to theoretical models of variability mechanisms. 2.3. Near Infrared observations Observations in the Johnson JHK bands were performed in August 2001 with the AZT24 telescope at Campo Imperatore (110 cm f/7.9, Cassegrain) and a PICNIC type HgCdTe 256x256 camera, with a scale of 1 arcsec/pixel. Each image was the sum of 6 individual frames of 30s exposure, to avoid saturation. Preliminary data reduction was made with the PREPROCESS task developed at the Roma Observatory, and photometry was made with IRAF/daophot using a 4 arcsec radius aperture. Primary photometric standard stars (AS02, AS06, AS32, AS37-1) were taken from Hunt et al. (1998). All the observations were made at airmass smaller than 1.2; the standard error of our JHK observations is ∼0.05 mag. The results are collected in Table 4. Simultaneous optical photometry was obtained with the Vallinfreda telescope. 2.4. Radio observations The 4-year radio monitoring of S5 1803+784 was carried out at 5 GHz and 8.4 GHz with the 32-m antennas located in Medicina (Bologna, Italy) and Noto (Siracusa, Italy) on monthly basis, starting from January 1996. The observations were performed by means of ON-OFF measurements on the target source. The systematic calibration uncertainty of the absolute radio flux is of the order of 4%, while the typical inner error of the measurements is 10 - 20 mJy. Details on the observations and data reduction can be found in Venturi et al. (2001). The 8.4 GHz data (the more complete dataset in our radio monitoring) are shown in Fig. 6, upper panel; for ease of comparison the optical data are shown in the lower panel of the same figure. The radio light curve shows a significant variability around an average flux density of 3 Jy. The minimum measured flux was 2.46 ±0.02 Jy on JD 657 and the maximum was 3.46 ±0.01 on JD 1376. The sampling is much coarser than the optical one, so that a cross correlation analysis would not be very meaningful: a simple eye-comparison of the two light curves does not suggest anyway a simultaneus correspondence between them. We note however that simultaneous radio-optical measurements on July 12 2001 (JD 2103), when the – 6 – source was near its minimum optical brightness, gave radio flux densities of F8.4 = 2.53 Jy (and F5.0 = 2.50 Jy at 5 GHz on July 17), one of the lowest recorded radio levels in our monitoring. The amplitude of the radio variability seems to increase from 1997 to 2000: the flux density difference between maxima and minima going from ∼ 0.2 Jy (i.e. ∼ 7%) at the beginning of 1997 to 0.8 Jy (i.e. ∼ 28%) in 1999-2000. 3. Time scale analysis The search for possible time scales in the optical light curve was performed by means of the Discrete Fourier Transform (DFT) for unevenly spaced data (Deeming 1975). We also used the Structure Function (SF) analysis (Simonetti et al. 1985) and the Jurkevich (1971) method. The DFT power spectrum is plotted in Fig. 7: the upper panel shows the light curve spectrum, and the lower panel that of the sampling window. The former spectrum is clearly dominated by a very prominent maximum at the lowest frequency due to the long term trend; a few other peaks are also present, two of them corresponding to the periods of about 218 and 105 days. The first period is nearly the distance between the second and the third flare. The latter could be just the first harmonic of 218, given the achievable resolution step. Note that in the window power spectrum a feature is present at 105 days, and therefore this frequency in the source power spectrum could be enhanced by the convolution effect. None of these features in the power spectrum, however, is so prominent to suggest a stable periodicity. The window power spectrum shows a marked periodicity at ∼29 days, corresponding to the moon phase period. Application of the Jurkevich test for the search of periodicities was made starting with a trial period of 5 days, and increasing it with a 5 days step, binning into 29 intervals the phase-reduced light curve. Given our average sampling of about 1 point every 10 days, shorter time scales cannot be explored with our database, save that in a few short intervals of denser sampling. For a better detection of possible time scales, we subtracted from the R light curve the average trend, derived with the running mean over 5 consecutive values. The result of the Jurkevich test is shown in Fig. 8, where the f parameter is plotted against the trial period. We recall that f is defined as 1-VB/VT , where VB is the sum of the variances of the phase-reduced binned data and VT is the variance of the whole sample. Minima in this plot may be considered as indicators of possible time scales: a value of f smaller than ∼0.5 is often quoted in the literature as indicator of a 'true' time scale (e.g. Kidger, Takalo & – 7 – Sillanpaa 1992). Several minima appear in this plot, the shortest one around 620 days (close to the lag between the first and the third flare), but all of them are far from the significancy threshold. Changing the number of bins and the step of the trial periods does not change substantially the positions of the minima, nor their relative depth. Finally the SF plot, averaged over a logaritmic time window of 0.1, is shown in Fig. 9. It has been calculated on the flux density values in mJy and no normalization has been applied. The SF has an average slope of 0.62±0.04 in the time lag range between 10 and 1000 days. A plateau region is marginally apparent between 20 and 80 days. No clear periodicity, which would reveal as a local minimum, is present. On short time scales (days) the SF analysis for S5 1803+784 was previosly studied by Heidt & Wagner (1996) for two different observing runs, and for time lags greater than 1 day they report a slope of 0.58±0.13 practically coincident with our result. In Fig. 9 we also report the SF for the radio light curve at 8.4 GHz, calculated on the flux densities in Jy, with the same binning of the optical one, again without any normalization. The points are more scatted than the optical ones but the mean slope is not significantly different (0.51±0.18). No indication of periodicity is evident also in the radio data. 4. Optical energy distribution Besides the R band, most of the times also B, V and I band observations were secured, to allow the determination of the broad band shape of the energy distribution. Plots of the color indices B−V, V−R and R−I as a function of time are reported in Fig. 10. Their average values are B−V=0.59±0.05, V−R=0.49±0.04 and R−I=0.63±0.03, where the quoted errors are 1 σ deviations of the data set. Large systematic changes of the source colour are not evident, in particular during the three large luminosity variations episodes (see Figs. 3 to 5). Several BL Lac objects have optical spectra well reproduced by a power law (Fν = Aνα). To derive the spectral index it is necessary to correct the observed fluxes for the foreground absorption, which is expected to be not very large, given the relatively low (b=29 degrees) galactic latitude of this source. We evaluated the absorption from literature NH values (3.90 × 1020, Ciliegi et al. 1995; 3.70 × 1020, Murphy et al. 1996), and adopting the ratio NH/E(B−V) = 5.2 × 1021 cm−2mag−1 from Shull & Van Steenberg (1985). The resulting colour excess E(B−V)=0.075 was derived, and the corresponding extinctions in the B, V, R, I bands computed assuming the curve by Schlegel et al. (1998). Conversion from magnitudes to fluxes was made according to Mead et al. (1990). Any contribution from the host galaxy is – 8 – likely negligible, given that the CaII break is undetectable in the published spectra (Lawrence et al. 1996; Rector & Stocke 2001). We have 85 4-band simultaneous observations of this source, useful to derive the spectral index. A fit of log(Fν) vs log(ν) with a straight line of slope α gives a satisfactory result against a χ2 test for nearly all our simultaneous 4-bands observations, supporting the power law modelling of the spectral energy distribution. We report in Table 4 the JD, R, σ(R), α, σ(α) and χ2 for each day. Consistent results are obtained using the larger set (134) of good quality V, R, I simultaneous observations. Our average value (-1.54, with a dispersion of 0.12) is in good agreement with the result of the only spectrophotometric observation of S5 1803+784 available in the literature (-1.48; Lawrence et al., 1996). A plot of the spectral index vs the R magnitude is shown in Fig. 11. The linear correlation coefficient was found equal to 0.42, which means a probability less than 10−3 that the correlation is merely due to chance. From this plot it is apparent however that at relatively large changes of the luminosity do not correspond large variations of the spectral slope, at variance with other BL Lac objects, like ON 231 (Massaro et al. 1999) or BL Lacertae (Nesci et al. 1998). 5. Discussion In this paper we presented the first flux calibrated, long term variability study of the BL Lac object S5 1803+784 in the optical frequency range, spanning more than five years since Spring 1996. Furthermore, we made a comparison with the behaviour of this source in the same period in the radio band at 8.4 GHz. The main results can be summarized as follows: a) the source showed an optical luminosity variation with an amplitude of a factor ∼ 20; b) the general behaviour was characterized by an increase of the mean flux level for at least 1150 days during which we observed with three large flares, followed by a decrease in about 400 days down to a much lower state; c) there is no evidence of periodicity in the light curve on the time scales (several months) well sampled by our dataset; d) the three major flares were characterized by comparable rise and decay times of about 20 days, corresponding to an (unbeamed) dimension of the emitting region R ≃ c∆t ≃ 5 × 1016 cm; – 9 – e) changes of the source luminosity are not related to strong systematic trends of the spectral slope α. A correlation test provided evidence for an increase of the slope in fainter states: however the α values corresponding to the highest and lowest luminosity differ only by ∼0.1 while larger differences were occasionally found in intermediate states; f) the source showed variability in the radio band, with maximum to minimum difference of 1 Jy and a mean flux density value of ∼3 Jy at 8.4 GHz. The slopes of the optical spectrum, derived in Section 3, indicate that the synchrotron emission peak of the SED should be in the IR range. The only JHK photometric data available in the literature are those by Heckman et al. (1983) taken in 1980-81 and give a much flatter slope (about -0.8). The IRAS data (Impey & Neugebauer 1988) give a substantially flat energy distribution suggesting that the SED should peak between 10 and 100 µm. To derive a more complete SED we retrieved from the ISOCAM archive the four available calibrated images, taken in April 1996 but still unpublished, and made a simple aperture photometry. In the X ray band, the source was lately observed by BeppoSAX on 28-Sep- 1998 (Padovani et al. in preparation). The 2-10 keV spectrum had an energy spectral slope αx=-0.45, significantly flatter than the optical one (-1.7), indicating that the X-ray emission was completely dominated by the Inverse Compton component. Our nearest optical data are on 20-Sep and 11-Oct giving the source respectively at R=15.3 and R=15.7 (see Table 3). Using these data, although not simultaneous, we constructed the SED, given in Fig. 12, which improves on that given by Giommi et et. (1996). For our optical and radio monitoring we plotted only the minimum and maximum values, to give a feeling of the variability of the source. For the JHK bands we plotted our data, together with the simultaneous optical ones. The R value form our monitoring simultaneous to the ISOCAM observation is also plotted. The general impression of a peak of the SED in the infrared range is confirmed from Fig. 12. It is interesting to compare the luminosity and the behaviour of S5 1803+784 with that of other BL Lac sources at their maximum luminosity. At the highest level (R=14), adopt- ing H0=65 km s−1 Mpc−1 and q0=0.5, the absolute magnitude of the source is MR=−28.7, corresponding to an integrated optical luminosity of 3.6 1046 erg s−1. ON 231, which reached the highest recorded flux (R=12.2) in Spring 1998 (Massaro et al. 1999), at a redshift z = 0.10, reached an absolute luminosity M=−26.2, i.e. 2.5 mag weaker than S5 1803+784. BL Lac, during the large flare in 1997 (Bloom et al. 1997) reached R=12.3, corresponding to M=−25.2, i.e. 3.5 mag fainter than S5 1803+784. These different absolute luminosities – 10 – may be due to an intrinsic different power or to a different beaming factor. Lahteenmaki & Valtaoja (1999) estimated Doppler boosting factors δ from variations of brightness tem- perature at 22 and 33 GHz and found values of 1.6 for ON 231, 3.9 for BL Lac and 6.5 for S5 1803+784. The higher beaming could be the reason of the apparent higher luminosity of our source: however, in this case, the higher time contraction would imply large amplitude flux variations on quite short time scales, which have not been detected in our observation, at variance with ON 231 and BL Lac (Nesci et al. 1998). We can also estimate the unbeamed luminosity using the factor δ given above, and from the relation Lobs = δ(3−α)Lunb, (1) we obtain an intrinsic optical luminosity of 7.8 1042 erg/s, which is quite reasonable (Cavaliere & Malquori 1999). A simple radiative cooling mechanism of freshly accelerated electrons cannot explain the color behaviour of the bright flares of S5 1803+784, in particular that on JD 711: in fact one would expect a significant steepening of the spectral slope in the dimming phase, like that clearly detected in the flares of ON 231 (Massaro et al. 1999) which is not the case. A possible explanation would be that the observed spectral shape is the result of an equilibrium condition between acceleration and radiation processes, while the total number of particles involved, which determine the flux level, changes with time. In this case however one should invoke that the injection and leakage time scales must be similar. An alternative possibility is that the flares are originated by a change of δ. Being the flux proportional to δ(3−α) a variation of ∆m magnitudes would require a variation of δ according to the relation: log(δ2/δ1) = 0.4(m1 − m2)/(3 − α), (2) which for α ∼ −1.5 implies a variation of δ by a factor 1.5 for ∆m = 2, which looks rather substantial. ACKNOWLEDGEMENTS The criticisms of an anonimous referee has helped to improve the first version of this paper. Part of this work was supported by the Italian Ministry for University and Scientific Research (MURST) with grant Cofin-98-02-32 and Cofin 2001/028773. 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 – 11 – National Aeronautics and Space Administration. REFERENCES Aller M.F., Aller H.D., Hughes P.A. and Latimer G.E., 1998, ApJ, 512, 601 Bertaud C., Dumortier B., Veron P. 1969 A&A, 3, 436 Biermann P., Duerbeck H., Eckart A., Fricke K., Johnston K.J. et al. 1981, ApJ, 247, L53 Bloom S.D, Bertsch D.L., Hartman R.C. et al. 1997 ApJ, 490, L145 Britzen S., Witzel A., Krichbaum T.P., Muxlow T.W.B. 1999, New Astr. Rev. 43,751 Cavaliere A. and Malquori D, 1999 ApJ, 516, L9 Ciliegi P., Bassani L., Caroli E., 1995 ApJ, 439, 80 Deeming T.J., 1975, Ap&SS, 36, 137 Fiorucci M. and Tosti G. 1996, A&AS, 116, 403 Gabuzda D.C., 1999, New Astron. Rev., 43, 691 Gabudza D.C. 2000, in Astrophys. Phenomena Revealed by Space VLBI, H. Hirabayashi, P.G. Edwards and D.W. Murphy eds. Ghisellini G., Villata M., Raiteri C. et al. 1997 A&A, 327, 61 Giommi P., Ansari S.G., Micol A. 1995, A&AS, 109, 267 Heidt J. and Wagner S.J. 1996, A&A, 305, 42 Heckman T.M., Lebofsky M.J. Rieke G.H. van Breguel W. 1983, ApJ, 272, 400 Hoffleit E.D., Warren jr. W.H. 1991, Bright Star Catalogue 5th revised edition, Centre de Donnee astronomiques de Strasbourg. Hunt L.K., Mannucci F., Testi L. et al. 1998, AJ, 115, 2594 Jurkevich I., 1971, Ap&SS, 13, 154 Impey C.D. and Neugebauer G., 1988, AJ, 95, 307 Kidger M., Takalo L. and Sillanpaa A., 1992, A&A, 264, 32 – 12 – Lahteenmaki A. & Valtaoja E. 1999, ApJ, 521, 493 Lawrence C. R., Zucker J. R., Readhead A. C. S. et al. 1996, ApJS, 107, 541 Massaro E., Nesci R., Maesano M., Montagni F., Trevese D. et al A&A, 314, 87 Massaro E., Maesano M., Montagni F., Nesci R., Tosti G. et al. 1999, A&A, 342, L49 Mc Gimsey G.Q., Miller H.R., Williamon R.M. 1976 ApJ, 81, 750 Mead A.R.G., Ballard K.R., Brand P.W.J.L. et al. 1990, A&AS83, 183 Miller H.R., Mullikin T.L., McGimpsey B.Q. 1983, ApJ, Murphy E.M., Lockman F.J., Laor A., Elvis M., 1996, ApJS, 105, 369 Nesci R., Maesano M., Massaro E., Montagni F., Tosti G., 1998, A&A, 332, L1 Padovani P. and Giommi P. 1995, ApJ, 444, 567 Padovani P. et al. in preparation Rector T.A. and Stocke J.T. 2001, AJ, 122, 565 Rieke G.H., Grasdalen G.L., Kinman T.D., Hintzen P., Wills P.J., Wills D., 1976, Nature 260, 754 Schlegel D.J., Finkbeiner D.P. and Davis M., 1998, ApJ, 500, 525 Simonetti J.H., Cordes J.M., Heeschen D.S., 1985, ApJ, 296, 46 Shull, J. M., van Steenberg, M. E. 1985, ApJ, 294, 599 Smith P.S., Balonek T.J., Heckert P.A., Elston R., Schmidt G.D. 1985 AJ, 90, 1184 Teraesranta H., Tornikosky M., Mujunen A., et al. 1998, A&AS, 132, 305 Venturi T., Dallacasa D., Orfei A., Bondi M., Fanti R. et al. 2001, A&A, 379, 755 Wagner S.J., Sanchez-Pons F., Quirrenbach A., and Witzel A., 1990, A&A, 235, L4 This preprint was prepared with the AAS LATEX macros v5.0. – 13 – Fig. 1.- The finding chart of S5 1803+784 from the POSS-I 103aE plate. Reference stars are marked according to Table 1; the blazar is marked with S. Fig. 2.- The RC light curve of S5 1803+784 from Apr 1996 to Jan 2002. Fig. 3.- First flare. The R light curve (filled boxes) and the V−I colour index (crosses with error bars). The colour index scale is on the right side, the R magnitude scale is on the left. No appreciable variation of the V−I colour index is present. Fig. 4.- Second flare. The R light curve (filled boxes) and the V−I colour index (crosses with error bars). Scales as in Fig.3 Fig. 5.- Third flare. The R light curve (filled boxes) and the V−I colour index (crosses with error bars). Scales as in Fig. 3. Marginal evidence for a redder colour in the decreasing branch is present. Fig. 6.- The radio light curve of S5 1803+784 at 8.4 GHz since 1996 (upper panel), and the R light curve in linear scale (mJy) not corrected for reddening (lower panel). Fig. 7.- Power spectra of the light curve (upper panel) and of the sampling window (lower panel). The peaks corresponding to 218 and 105 days are indicated. Fig. 8.- The Jurkevich test for 5 days step periods. Fig. 9.- The structure function for the optical data, averaged over log(∆tdays)=0.1 (filled squares) and of the radio data averaged in the same way (open circles). No normalization has ben applied to the data. Fig. 10.- B−V, V−R and R−I colour indices plotted as a function of time. Some data with large errors have been omitted. Fig. 11.- The 4-band spectral index as a function of the R magnitude. The linear correlation coefficient is 0.42. – 14 – Fig. 12.- The spectral energy distribution of S5 1803+784 derived from literature and our measurements: NED data (radio and IRAS), open squares; radio data (max and min, this paper), filled squares; ISO and simultaneous optical data, filled triangles; simultaneous NIR and optical data (this paper), open triangles; optical data (max and min, this paper), filled circles; BeppoSAX x-ray data, crosses. – 15 – Table 1. Magnitudes of comparison stars for S5 1803+784. Stara B V R I A B C D E 15.39±.05 (7) 15.97±.01 16.37±.01 14.54±.03(8) 15.30±.01 15.73±.01 16.58±.01 16.49±.01 14.06±.03(10) 14.92±.01 15.36±.01 16.08±.01 16.10±.01 13.62±.03(8) 14.53±.01 14.97±.01 15.57±.01 15.70±.01 aErrors for star A include the zero point uncertainty, while for the others only the intercalibration uncertainty is given – 16 – Table 2. Photographic archive magnitudes of S5 1803+784. Band mag date Survey emulsion R V R B 14.5 15.5 15.5 16.6 POSS-I 11-08-1953 16-05-1983 Quick Blue 21-08-1993 12-09-1993 POSS-II POSS-II 103aE IIaD IIIaF IIIaJ – 17 – Table 3. B,V,R,I magnitudes of S5 1803+784. JDa B σB V σV R σR I σI Tel 193.4382 222.4194 246.4236 272.4507 277.5292 354.4250 357.3854 375.3771 377.3847 378.4056 381.2625 390.3125 391.2313 415.3014 421.2299 464.2500 476.3542 487.5382 488.6021 517.4750 521.4507 549.4910 572.4729 598.5042 628.4424 639.4271 669.3965 711.3493 712.2597 717.2931 719.2528 721.3368 723.2437 723.4479 725.3299 727.3264 731.3243 739.2917 742.2708 744.2792 747.2549 748.3292 756.2889 781.2681 791.2174 823.2160 825.2132 832.2194 838.2549 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 16.67 0.00 0.00 0.00 0.00 0.00 16.93 0.00 0.00 0.00 0.00 0.00 15.44 15.61 15.65 15.84 15.91 16.02 15.97 0.00 16.77 16.93 0.00 0.00 16.64 16.46 16.46 0.00 16.99 0.00 16.12 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.00 0.00 0.00 0.00 0.00 0.06 0.00 0.00 0.00 0.00 0.00 0.05 0.05 0.03 0.03 0.04 0.05 0.04 0.00 0.03 0.04 0.00 0.00 0.04 0.03 0.03 0.00 0.09 0.00 0.05 0.00 0.00 0.00 16.53 0.00 0.00 16.87 17.03 16.28 0.00 16.67 16.67 16.64 16.60 16.61 16.60 16.27 16.57 16.16 15.86 15.98 16.01 16.40 0.00 16.34 15.57 15.98 0.00 16.49 16.42 14.90 15.09 15.03 15.30 15.38 15.57 15.51 0.00 16.04 16.26 16.30 0.00 16.06 15.90 15.93 15.92 16.28 0.00 15.62 15.55 15.89 16.13 0.05 0.00 0.00 0.03 0.05 0.06 0.00 0.03 0.03 0.03 0.03 0.05 0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.00 0.03 0.03 0.03 0.00 0.03 0.03 0.03 0.03 0.04 0.03 0.03 0.03 0.03 0.00 0.03 0.04 0.04 0.00 0.03 0.03 0.03 0.03 0.02 0.00 0.03 0.03 0.04 0.04 15.96 15.93 15.76 0.00 16.55 15.82 15.87 16.21 16.24 16.19 16.19 16.16 16.18 15.82 16.11 15.71 15.43 15.48 15.52 15.94 15.64 15.85 15.02 15.50 15.67 16.00 15.94 14.43 14.54 14.54 14.77 14.82 15.06 14.99 15.15 15.51 15.78 15.82 15.64 15.51 15.40 15.39 15.44 15.70 15.27 15.11 15.04 15.35 15.55 0.03 0.03 0.02 0.00 0.05 0.03 0.02 0.03 0.03 0.03 0.03 0.05 0.05 0.04 0.03 0.03 0.03 0.03 0.04 0.04 0.04 0.03 0.03 0.03 0.03 0.03 0.03 0.02 0.03 0.03 0.03 0.03 0.03 0.03 0.02 0.03 0.03 0.03 0.03 0.03 0.04 0.03 0.03 0.03 0.03 0.03 0.03 0.02 0.02 15.38 0.00 0.00 15.73 15.90 0.00 0.00 0.00 15.55 15.53 0.00 0.00 0.00 15.18 0.00 15.04 14.86 14.83 14.90 15.24 15.12 15.23 14.38 0.00 0.00 15.29 0.00 13.76 13.87 13.92 14.11 14.19 14.43 14.37 0.00 14.88 15.13 15.18 0.00 14.88 14.78 14.75 14.82 15.09 0.00 14.45 14.37 14.75 14.92 VA 0.03 0.00 VA 0.00 MP VA 0.03 VA 0.03 0.00 VA VA 0.00 VA 0.00 VA 0.05 0.03 VA VA 0.00 VA 0.00 VA 0.00 VA 0.03 0.00 VA VA 0.03 VA 0.03 VA 0.03 VA 0.03 0.03 VA VA 0.03 VA 0.03 VA 0.03 VA 0.00 0.00 VA VA 0.03 VA 0.00 VA 0.03 VA 0.03 0.03 VA VA 0.03 VA 0.03 VA 0.03 0.03 VA VA 0.00 0.03 VA 0.03 MP VA 0.03 0.00 VA VA 0.03 VA 0.03 VA 0.03 0.03 VA 0.03 MP VA 0.00 VA 0.03 VA 0.03 VA 0.02 0.02 VA – 18 – Table 3-Continued JDa B σB V σV R σR I σI Tel 840.2556 842.2257 843.2250 858.2444 860.2382 862.2472 863.2597 865.5174 871.5194 872.4646 891.4333 895.5132 900.3889 901.4056 907.4604 924.5722 928.4799 942.3563 947.4076 950.4188 953.4444 955.4368 956.4264 966.3799 970.4146 983.4451 985.4597 988.4215 993.4562 997.5181 1001.4118 1004.3958 1005.3806 1008.3944 1011.4000 1013.3979 1015.3826 1016.4271 1018.3799 1019.4174 1021.3382 1025.4611 1026.3556 1037.4639 1039.4326 1040.3285 1041.4229 1042.3438 1043.3354 0.00 0.00 0.00 16.65 0.00 0.00 16.58 16.55 16.28 16.37 16.40 0.00 0.00 16.59 0.00 0.00 15.91 0.00 0.00 0.00 0.00 15.86 15.70 0.00 0.00 0.00 16.53 0.00 16.03 0.00 15.64 15.79 16.01 16.07 0.00 15.98 0.00 0.00 15.91 15.89 16.02 15.89 15.89 16.09 0.00 16.05 0.00 15.94 15.87 0.00 0.00 0.00 0.08 0.00 0.00 0.06 0.05 0.03 0.02 0.04 0.00 0.00 0.05 0.00 0.00 0.07 0.00 0.00 0.00 0.00 0.03 0.04 0.00 0.00 0.00 0.02 0.00 0.01 0.00 0.02 0.02 0.04 0.02 0.00 0.03 0.00 0.00 0.02 0.05 0.02 0.04 0.02 0.10 0.00 0.03 0.00 0.04 0.05 16.15 16.14 16.20 0.00 16.02 16.21 16.09 15.98 15.71 15.75 15.84 15.89 15.89 15.89 0.00 15.37 15.45 0.00 15.46 0.00 15.48 15.31 15.12 15.15 15.86 15.97 15.91 15.86 15.48 15.17 15.06 15.28 15.42 15.52 15.50 15.39 15.38 15.35 15.38 15.37 15.41 15.35 15.23 15.55 15.62 15.42 15.45 15.37 15.31 0.03 0.03 0.02 0.00 0.02 0.02 0.03 0.02 0.02 0.01 0.02 0.02 0.02 0.03 0.00 0.02 0.02 0.00 0.04 0.00 0.02 0.03 0.05 0.01 0.04 0.03 0.01 0.05 0.01 0.03 0.02 0.02 0.02 0.04 0.02 0.02 0.02 0.04 0.02 0.03 0.01 0.03 0.01 0.08 0.09 0.02 0.05 0.03 0.03 15.69 15.63 15.63 15.55 15.53 15.58 15.53 15.40 15.11 15.15 15.28 15.36 15.37 15.46 15.33 14.83 14.90 15.10 14.93 14.67 14.90 14.82 14.54 14.64 15.26 15.45 15.38 15.39 14.94 14.71 14.64 14.78 14.97 15.11 15.04 14.94 14.89 14.87 14.91 14.91 14.96 14.88 14.79 14.97 15.01 14.95 14.98 14.90 14.85 0.01 0.02 0.01 0.02 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.02 0.02 0.01 0.01 0.04 0.03 0.01 0.02 0.01 0.01 0.02 0.03 0.01 0.02 0.02 0.01 0.01 0.02 0.02 0.02 0.02 0.01 0.01 0.02 0.03 0.02 0.03 0.01 0.03 0.01 0.04 0.03 0.02 0.03 0.02 0.01 15.02 14.96 15.02 0.00 14.85 14.87 14.83 14.74 14.57 14.59 14.62 14.68 0.00 14.71 0.00 14.20 14.26 0.00 0.00 14.02 0.00 14.10 13.96 14.06 14.57 14.77 14.73 14.70 14.34 14.11 14.03 14.23 14.38 14.47 14.44 14.41 0.00 14.25 14.32 14.29 14.39 14.28 14.24 14.26 14.34 14.34 14.27 14.23 14.20 0.02 0.03 0.04 0.00 0.03 0.01 0.01 0.01 0.01 0.05 0.01 0.01 0.00 0.02 0.00 0.01 0.01 0.00 0.00 0.04 0.00 0.01 0.01 0.01 0.02 0.02 0.01 0.02 0.01 0.01 0.02 0.02 0.02 0.03 0.02 0.02 0.00 0.03 0.03 0.03 0.02 0.04 0.01 0.05 0.05 0.02 0.03 0.01 0.02 VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA VA GR VA GR VA GR VA GR GR VA GR VA VA – 19 – Table 3-Continued JDa B σB V σV R σR I σI Tel 1048.3194 1050.3146 1052.3271 1054.4174 1057.5313 1058.3424 1067.2944 1072.2847 1077.3354 1098.2444 1100.2979 1101.2889 1108.2667 1109.2410 1111.3681 1114.2347 1125.2090 1126.2056 1127.2854 1133.2986 1135.2083 1135.2715 1138.3042 1151.2493 1154.5431 1156.2472 1160.2583 1164.2410 1294.4354 1320.5000 1327.3444 1338.4424 1338.4979 1339.3847 1344.3701 1350.4347 1353.3701 1373.4035 1405.3611 1414.3722 1422.3549 1424.3854 1428.3278 1433.4410 1434.3847 1435.4042 1445.2660 1456.3312 1457.3306 15.92 15.97 15.91 0.00 15.90 15.86 16.25 16.37 16.45 16.73 0.00 16.86 0.00 16.72 0.00 16.28 15.39 15.26 15.15 15.21 15.45 15.44 15.64 0.00 15.86 0.00 15.82 15.76 0.00 0.00 15.74 15.26 15.23 15.22 15.11 15.21 15.06 15.27 16.42 16.30 16.70 16.53 16.66 16.56 16.53 16.71 17.04 16.97 16.89 0.03 0.02 0.03 0.00 0.02 0.02 0.05 0.01 0.02 0.03 0.00 0.04 0.00 0.01 0.00 0.02 0.03 0.01 0.03 0.02 0.08 0.01 0.02 0.00 0.01 0.00 0.03 0.02 0.00 0.00 0.03 0.03 0.05 0.03 0.03 0.03 0.03 0.03 0.08 0.08 0.04 0.06 0.04 0.05 0.06 0.05 0.01 0.09 0.01 15.32 15.34 15.27 0.00 15.22 15.25 15.58 15.71 15.80 16.14 16.27 16.25 16.18 16.18 15.82 15.76 14.82 14.73 14.66 14.66 14.88 14.83 15.01 0.00 15.19 15.24 15.14 15.14 15.72 0.00 15.12 14.65 14.65 14.55 14.43 14.58 0.00 14.71 0.00 15.73 16.03 15.90 15.99 0.00 15.90 16.12 16.44 16.47 16.32 0.01 0.01 0.02 0.00 0.01 0.02 0.00 0.02 0.02 0.05 0.01 0.03 0.03 0.04 0.01 0.02 0.02 0.01 0.02 0.03 0.05 0.03 0.01 0.00 0.03 0.03 0.03 0.02 0.07 0.00 0.02 0.03 0.04 0.03 0.03 0.03 0.00 0.03 0.00 0.05 0.04 0.03 0.01 0.00 0.04 0.04 0.02 0.04 0.02 14.86 14.87 14.81 14.72 14.75 14.73 15.10 15.20 15.31 15.69 15.77 15.76 15.70 15.63 15.40 15.30 14.38 14.30 14.18 14.20 14.37 14.40 14.58 14.57 14.70 14.73 14.70 14.69 0.00 14.62 14.62 14.16 14.17 14.09 14.04 14.10 14.05 14.19 0.00 15.21 0.00 15.40 15.46 0.00 0.00 0.00 15.90 15.88 15.79 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.02 0.02 0.01 0.02 0.03 0.02 0.01 0.02 0.01 0.01 0.02 0.01 0.03 0.01 0.04 0.02 0.01 0.01 0.03 0.02 0.00 0.04 0.03 0.03 0.02 0.03 0.02 0.03 0.03 0.03 0.00 0.03 0.00 0.03 0.01 0.00 0.00 0.00 0.02 0.03 0.02 14.23 14.25 14.18 14.07 14.10 14.10 14.47 14.57 14.68 15.06 15.15 15.13 15.08 15.03 14.78 14.70 13.80 13.70 13.63 13.61 13.76 13.80 13.89 0.00 14.10 14.15 14.07 14.07 0.00 0.00 14.01 13.56 13.56 13.47 13.33 13.53 13.33 13.58 14.62 14.48 14.76 14.76 14.82 14.81 14.75 14.88 15.25 15.24 15.11 VA 0.01 VA 0.01 VA 0.01 VA 0.03 VA 0.01 VA 0.01 VA 0.01 VA 0.01 VA 0.02 VA 0.01 VA 0.01 VA 0.01 VA 0.03 VA 0.01 VA 0.01 VA 0.02 GR 0.01 0.01 VA 0.02 MP 0.01 VA 0.03 MP 0.01 VA 0.03 MP VA 0.00 VA 0.01 0.01 VA VA 0.01 VA 0.02 GR 0.00 0.00 GR VA 0.03 0.03 GR 0.02 MP GR 0.03 0.02 GR GR 0.03 GR 0.03 GR 0.03 GR 0.06 0.03 GR GR 0.04 GR 0.03 VA 0.01 GR 0.03 0.03 GR GR 0.03 VA 0.02 GR 0.04 0.01 VA – 20 – Table 3-Continued JDa B σB V σV R σR I σI Tel 1480.2382 1484.2361 1491.2243 1508.2389 1510.3312 1520.2917 1546.2326 1549.4319 1556.2347 1569.2410 1578.3493 1606.4215 1613.4931 1655.3590 1688.5451 1695.3903 1706.3597 1717.5007 1718.3965 1721.4389 1722.5472 1724.4493 1747.3424 1749.3236 1751.3542 1752.3479 1754.3458 1758.3458 1765.3257 1768.3118 1789.3687 1814.5125 1838.2986 1841.2667 1910.2062 2012.4583 2083.5090 2087.3868 2093.4472 2098.4278 2102.4535 2103.3951 2112.4097 2113.3764 2115.4042 2116.4111 2117.4153 2119.4056 2120.3701 0.00 0.00 17.29 16.71 0.00 16.71 0.00 0.00 0.00 16.58 0.00 16.53 0.00 0.00 0.00 17.71 0.00 17.41 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 17.38 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.06 0.02 0.00 0.02 0.00 0.00 0.00 0.05 0.00 0.02 0.00 0.00 0.00 0.05 0.00 0.03 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.04 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 16.56 16.67 16.66 16.11 16.06 16.07 16.16 16.12 16.01 15.95 0.00 15.95 15.86 0.00 17.19 0.00 0.00 16.70 16.74 16.71 16.79 16.86 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 16.87 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 17.65 0.00 0.00 0.00 0.00 0.00 0.03 0.09 0.06 0.02 0.02 0.02 0.04 0.02 0.05 0.03 0.00 0.02 0.02 0.00 0.03 0.00 0.00 0.02 0.05 0.05 0.02 0.05 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.08 0.00 0.00 0.00 0.00 0.00 16.04 16.03 16.08 15.55 15.68 15.63 15.64 15.63 15.56 15.57 15.45 15.42 15.31 16.82 16.80 16.64 16.57 16.19 16.23 16.16 16.20 16.24 16.25 16.23 16.27 16.28 16.31 16.15 16.46 16.45 16.24 16.78 16.05 16.04 16.47 17.00 16.98 17.02 17.30 17.20 17.22 17.23 17.24 17.19 17.35 0.00 17.18 17.22 17.22 0.03 0.05 0.03 0.02 0.02 0.02 0.04 0.02 0.04 0.03 0.02 0.02 0.02 0.30 0.03 0.02 0.02 0.02 0.04 0.05 0.02 0.02 0.04 0.02 0.03 0.03 0.02 0.04 0.03 0.04 0.02 0.03 0.05 0.03 0.08 0.05 0.05 0.05 0.08 0.08 0.08 0.08 0.03 0.08 0.08 0.00 0.08 0.06 0.08 15.43 15.35 15.41 14.93 15.12 15.06 14.96 14.98 14.89 0.00 0.00 14.76 14.62 0.00 16.01 15.97 0.00 15.52 0.00 0.00 15.48 15.51 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 15.64 0.00 0.00 15.36 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 16.43 16.52 16.65 16.43 16.59 16.65 0.00 VA 0.02 GR 0.05 0.04 GR 0.03 MP 0.02 MP 0.01 VA GR 0.04 VA 0.02 GR 0.04 GR 0.00 0.00 VA VA 0.02 VA 0.03 GR 0.00 VA 0.04 0.02 VA VA 0.00 VA 0.02 GR 0.00 0.00 GR 0.02 MP 0.02 MP VA 0.00 VA 0.00 VA 0.00 0.00 VA VA 0.00 VA 0.00 VA 0.00 0.00 VA 0.03 MP 0.00 MP VA 0.00 VA 0.02 0.00 GR GR 0.00 VA 0.00 VA 0.00 VA 0.00 0.00 VA VA 0.00 VA 0.00 GR 0.02 VA 0.05 0.08 VA GR 0.08 VA 0.06 VA 0.06 0.00 VA – 21 – Table 3-Continued JDa B σB V σV R σR I σI Tel 2129.3722 2141.4139 2148.3146 2159.3069 2164.3736 2165.3438 2172.3306 2174.3521 2178.3076 2181.3042 2194.2715 2195.3174 2236.2097 2286.2333 2287.2556 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 17.39 0.00 17.69 17.01 17.08 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.08 0.00 0.03 0.08 0.08 0.00 0.00 0.00 0.00 0.00 17.24 0.00 0.00 0.00 0.00 16.76 0.00 0.00 16.38 16.44 0.00 0.00 0.00 0.00 0.00 0.08 0.00 0.00 0.00 0.00 0.05 0.00 0.00 0.03 0.03 17.12 16.86 16.78 16.85 16.74 16.70 16.83 16.79 16.69 16.65 16.29 16.29 16.55 15.91 15.93 0.08 0.05 0.05 0.09 0.02 0.05 0.05 0.05 0.02 0.05 0.03 0.03 0.03 0.03 0.03 16.52 0.00 0.00 0.00 0.00 16.07 0.00 16.18 16.02 0.00 15.67 0.00 15.88 15.24 15.27 0.05 0.00 0.00 0.00 0.00 0.03 0.00 0.03 0.03 0.00 0.03 0.00 0.03 0.02 0.02 VA VA VA GR VA VA VA VA VA VA VA VA TE VA VA aJD-2,450,000 bThe observation on JD 2236 was obtained with the 70 cm telescope of the Collurania-Teramo Observatory by one of us (RN) – 22 – Table 4. Near Infrared observations. dd-mm-yyyy J H K αnir 23-08-2001 24-08-2001 26-08-2001 14.99 15.06 15.00 14.09 14.04 14.00 13.26 13.27 1.26±0.11 1.36±0.11 – 23 – Table 5. Optical spectral slope fits. JD-2400000 R α σ(α) χ2 464.2 549.5 711.3 712.3 717.3 719.3 721.3 723.2 723.4 727.3 731.3 744.3 747.3 748.3 781.3 823.2 863.3 865.5 871.5 872.5 891.4 901.4 928.5 955.4 956.4 985.5 993.5 1001.4 1004.4 1005.4 1008.4 15.71 15.85 14.43 14.54 14.54 14.77 14.82 15.06 14.99 15.51 15.78 15.51 15.40 15.39 15.70 15.11 15.53 15.40 15.11 15.15 15.28 15.46 14.90 14.82 14.54 15.38 14.94 14.64 14.78 14.97 15.11 -1.378 -1.509 -1.508 -1.647 -1.538 -1.553 -1.578 -1.427 -1.402 -1.785 -1.632 -1.619 -1.469 -1.527 -1.740 -1.530 -1.730 -1.736 -1.550 -1.704 -1.664 -1.730 -1.594 -1.581 -1.570 -1.654 -1.528 -1.338 -1.295 -1.375 -1.320 0.063 0.084 0.079 0.079 0.064 0.063 0.072 0.079 0.072 0.063 0.074 0.072 0.064 0.063 0.093 0.079 0.084 0.079 0.063 0.079 0.072 0.079 0.088 0.063 0.075 0.063 0.079 0.063 0.063 0.072 0.064 3.037 0.060 1.356 2.529 0.031 2.169 2.340 3.299 3.556 1.751 0.154 0.998 0.534 2.307 1.830 2.322 3.715 1.805 4.092 2.906 1.813 2.407 2.859 3.455 1.581 0.254 1.151 0.327 1.034 0.033 1.259 – 24 – Table 5-Continued JD-2400000 R α σ(α) χ2 1013.4 1018.4 1019.4 1021.3 1025.5 1026.4 1037.5 1040.3 1042.3 1043.3 1048.3 1050.3 1052.3 1057.5 1058.3 1067.3 1072.3 1077.3 1098.2 1101.3 1109.2 1114.2 1125.2 1126.2 1127.3 1133.3 1135.2 1135.3 1138.3 1154.5 1160.3 14.94 14.91 14.91 14.96 14.88 14.79 14.97 14.95 14.90 14.85 14.86 14.87 14.81 14.75 14.73 15.10 15.20 15.31 15.69 15.76 15.63 15.30 14.38 14.30 14.18 14.20 14.37 14.40 14.58 14.70 14.70 -1.251 -1.341 -1.378 -1.365 -1.361 -1.373 -1.782 -1.502 -1.521 -1.471 -1.469 -1.516 -1.527 -1.632 -1.593 -1.581 -1.631 -1.581 -1.448 -1.541 -1.517 -1.327 -1.317 -1.269 -1.230 -1.339 -1.501 -1.385 -1.552 -1.562 -1.549 0.084 0.072 0.079 0.072 0.083 0.072 0.152 0.063 0.072 0.079 0.063 0.063 0.063 0.063 0.063 0.079 0.079 0.079 0.084 0.072 0.082 0.072 0.063 0.063 0.063 0.063 0.103 0.063 0.064 0.072 0.063 0.476 0.364 0.737 0.330 0.296 1.831 0.636 0.165 0.931 0.631 0.090 0.102 0.216 0.508 0.222 0.301 0.206 0.101 0.172 0.016 1.100 0.564 0.017 0.484 0.961 0.128 0.172 0.317 0.785 0.789 1.161 – 25 – Table 5-Continued JD-2400000 R α σ(α) χ2 1164.2 1327.3 1338.4 1338.5 1339.4 1344.4 1350.4 1373.4 1414.4 1424.4 1428.3 1445.3 1456.3 1457.3 1491.2 1508.2 1520.3 1606.4 1717.5 1789.4 2194.3 2286.2 2287.3 14.69 14.62 14.16 14.17 14.09 14.04 14.10 14.19 15.21 15.40 15.46 15.90 15.88 15.79 16.08 15.55 15.63 15.42 16.19 16.24 16.29 15.91 15.93 -1.465 -1.542 -1.495 -1.453 -1.556 -1.573 -1.462 -1.491 -1.787 -1.600 -1.705 -1.698 -1.698 -1.727 -1.795 -1.645 -1.363 -1.661 -1.750 -1.710 -1.600 -1.671 -1.740 0.063 0.063 0.063 0.082 0.063 0.063 0.063 0.063 0.103 0.084 0.072 0.095 0.116 0.093 0.120 0.072 0.079 0.084 0.089 0.088 0.103 0.091 0.091 0.165 0.139 0.120 0.040 0.756 3.000 0.750 0.760 1.155 0.010 0.308 0.334 1.912 0.512 0.350 1.028 0.726 0.568 0.197 4.749 1.084 0.989 0.837 14 15 16 17 S5 1803+78 1996-2001 500 1000 1500 2000 JD-2450000 4 3.5 3 2.5 2 8 6 4 2 0 500 1000 1500 2000 JD-2450000 S5 1803+78 218 105 window 1 1 0 0 -1 -1 -2 -2 -3 -3 0 0 1 1 2 2 log(lag(days)) 3 3 1 0.8 0.6 0.4 0.8 0.6 0.4 1 0.8 0.6 0.4 500 1000 1500 2000 500 1000 1500 2000 500 1000 1500 2000 JD - 2450000
astro-ph/0311139
1
0311
2003-11-05T22:09:13
Spatially Resolved Circumnuclear Dust in Centaurus A
[ "astro-ph" ]
In this paper we present results from our exploratory mid-IR study of Centaurus A circumnuclear environment using high-angular resolution imaging at the Magellan 6.5m telescope with the MIRAC/BLINC camera. We detected emission from a compact region surrounding the nuclear source, and obtained photometry at 8.8 microns and in the N band. Our analysis suggests that the nuclear region is resolved with a size of approximately 3 pc. The mid-IR emission from this region is likely associated with cool dust with an estimated temperature of \~160 K, surrounding the central ``hidden'' AGN. We discuss the characteristics of this emission in relation to other mid-IR observations and the implications on models of dust formation in AGNs.
astro-ph
astro-ph
Spatially Resolved Circumnuclear Dust in Centaurus A Margarita Karovska, Massimo Marengo, Martin Elvis, Giovanni G. Fazio, Joseph L. Hora Harvard-Smithsonian Center for Astrophysics, 60 Garden St., Cambridge, MA 02138 [email protected] Philip M. Hinz, William F. Hoffmann, Michael Meyer, Eric Mamajek Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721-0065 ABSTRACT In this paper we present results from our exploratory mid-IR study of Centau- rus A circumnuclear environment using high-angular resolution imaging at the Magellan 6.5m telescope with the MIRAC/BLINC camera. We detected emission from a compact region surrounding the nuclear source, and obtained photometry at 8.8 µm and in the N band. Our analysis suggests that the nuclear region is resolved with a size of ≈3 pc. The mid-IR emission from this region is likely asso- ciated with cool dust with an estimated temperature of ∼160 K, surrounding the central "hidden" AGN. We discuss the characteristics of this emission in relation to other mid-IR observations and the implications on models of dust formation in AGNs. Subject headings: galaxies: active -- galaxies: nuclei -- galaxies: (Centaurus A, NGC5128) -- techniques: high angular resolution individual 3 0 0 2 v o N 5 1 v 9 3 1 1 1 3 0 / h p - o r t s a : v i X r a 1. Introduction The nearest radio-bright AGN is located at a distance of "only" ∼ 3.5 Mpc in the giant elliptical galaxy NGC5128 (Cen A) (vis review by Israel 1998). The central engine powering this galaxy is thought to be a supermassive black hole enshrouded in a region containing dust and gas heated by the central source (Marconi et al. 2001). Multi-wavelength observations of Cen A reveal many complex multi-scale structures, at scales ranging from a sub-pc to hundreds of kpc (eg. Karovska et al. 2002). A radio and X-ray jet powered by the central source extends across the galaxy to a distance of several hundreds of kpc from the nucleus. -- 2 -- Optical and near-IR images of Cen A show large spatial scale (several kpc in size) dark bands stretching across the middle of the galaxy and obscuring the central region, probably due to absorption by dust and other cool material. These dust lanes are thought to be a remnant of a merger with a smaller spiral galaxy (Schiminovich et al. 1994). Using mid-infrared and sub-millimeter wavelength observations of the central region of Cen A, Mirabel et al. (1999) detected dust emission from a large scale bisymmetric structure (5kpc in size) resembling a barred spiral galaxy. Because of its proximity, Cen A provides a unique possibility for high-spatial resolution studies of the circumnuclear environment of AGNs using ground- and space-based large- aperture telescopes (0.1" corresponds to ∼ 1.5 pc at the Cen A distance of ∼3.5 Mpc). However, exploring the nuclear region of Cen A at optical wavelengths presents a major challenge because of the high extinction (eg. AV ∼ 10 mags; Alexander et al. 1999) due to heavy obscuration by the dust in the dust lanes. Although the diffraction limited telescope resolution is significantly lower in the mid- IR when compared to the optical (∼ 10 times), the spectral range from 8 to 20 µm is, nevertheless, particularly suitable for studying the emission from the region surrounding the nucleus because of at least ten times lower opacity. The lower opacity of astronomical dust at these wavelengths (Draine & Lee 1984) permits penetration of the optically opaque dust lanes. Most of the thermal radiation from dust is in fact emitted at these wavelengths, as seen in the mid-IR ISOCAM spectra of Cen A which show the presence of the characteristic silicate broad absorption feature at 9.8 µm, plus a possible PAH emission line at 11.3 µm. Alexander et al. 1999 suggested that the presence of a very deep 10 µm silicate absorption feature in the ISOCAM spectrum could be due to self-absorption in a dusty circumnuclear region, in addition to the extinction by the dust in the dust lanes (optical depth τν ≃ 1 at 10 µm). We report here results from our mid-IR imaging of Cen A with the MIRAC/BLINC mid-IR camera at Magellan, showing that the nuclear region is resolved at a sub-arcsecond scale. The observations and the techniques for data acquisition and reduction are described in the next section. Results and analysis are presented in section 3, and discussed in relation to other available observations and models in section 4. 2. Observations and Data Reduction During an exploratory observation of Cen A on May 1 and 2, 2002 at the Magellan I (Baade) telescope we detected a compact mid-IR source in the images of the nuclear region at -- 3 -- 8.8 µm and in the N band. The Baade Telescope, with a primary aperture of 6.5m equipped with active optics, provides diffraction limited sub-arcsecond images in the 8.8 µm spectral window (Shectman & Johns 2003). The images of Cen A and several reference sources were recorded using the mid-IR camera MIRAC/BLINC. MIRAC/BLINC uses a Boeing HF16 128×128 Si:As blocked impurity band detector (Hoffmann et al. 1998). We obtained images using the MIRAC/BLINC 8.8 µm 10% passband filter and the wide N band filter (from 8.1 to 13.1 µm) centered at 10.3 µm. The total on-source integration time was 1400 s at 8.8 µm and 900 s in the N band. The reference stars γ Cru and IRC+10220, observed while transiting at a similar airmass as the source, provide flux and Point Spread Function (PSF) calibration information. To insure similar observing conditions, including similar telescope orientation, airmass and seeing conditions, the observations of Cen A were made close in time (within couple of hours) to the observations of the corresponding reference stars. In addition, we used the Magellan telescope active optics (looking at standard stars in the field of view) to monitor the seeing stability, by checking that the size and the shape of the PSF at the beginning of each observation of Cen A is consistent with those of the corresponding reference. On the Magellan, MIRAC/BLINC has a plate scale of 0.12 arcsec pix−1, providing a total field of view of 17′′ × 17′′. This pixel scale ensures Nyquist sampling of the diffraction- limited PSF. As described below, we obtained further sub-sampling of the PSF by using dithering techniques. We used a standard nodding and chopping technique to remove the background signal, dithering the source on the array to obtain sub-sampling of the PSF. Chopping is carried out through rotation of an internal mirror to mitigate noise due to sky brightness fluctuations. The chop frequency was set to 10 Hz, with a throw of 8′′ in the North-South direction. The nod throw was also set to 8′′, but in the East-West direction, in order to have all four chop-nod beams inside the field of view of the array. Each individual nod cycle required 15 s (8.8 µm) or 10 s (N band) on-source integration, and the procedure was repeated for as many cycles as needed to obtain the total integration time at each wavelength. The data were analyzed by first subtracting the chop-on from the chop-off frames for both nodding beams. The two images thus obtained were then subtracted one from the other, in order to get a single frame in which the source appears in all four beams (two negative and two positive). We then applied a gain matrix, derived from images of the dome (high intensity uniform background) and the sky (low intensity uniform background), to flat field the chop-nodded image. This procedure was repeated for each of the nodding cycles for which the source was observed. A final high signal-to-noise ratio cumulative image was then obtained by co-adding together all the beam frames, each re-centered and -- 4 -- shifted on the source centroid. This last co-adding step was performed on a sub-pixel grid, providing a final pixel scale of 0.03 arcsec pix−1. A mask file to block out the effects of bad pixels and field vignetting was also created and applied at each individual frame before their shifting, preventing individual rejected pixels from contributing to the final image. To ensure a uniform treatment of the source and the standard the same observing and reduction procedure was also used for the reference star. 3. Results and Analysis 3.1. Imaging of the Circumnuclear region The mid-IR images of the nuclear region of Cen A obtained at 8.8 µm and in the N band are systematically larger then the unresolved reference sources (or the "observed" PSFs) recorded under similar observing conditions and using the same filters. In Figure 1 we show the images of Cen A and the reference star γ Cru obtained at 8.8 µm. The comparison of image sizes clearly shows that the Cen A image of the nuclear region is larger than the image of the unresolved point-like reference source at this wavelength. We note that the 8.8 µm images of Cen A and the reference have better resolution and are of higher signal-to-noise then the images obtained in the wide N band centered at 10 µm. The difference in size between the point source and the Cen A nuclear region is shown in Figure 2, which displays the radial profiles of the source and the reference at 8.8 µm normalized at the same peak value. This suggests that the Cen A nuclear region is resolved, with the size near the resolution limit of the Magellan 6.5 m telescope. We estimated the angular size of the Cen A nuclear region by convolving the reference star radial profile with a gaussian, and then fitting to the Cen A data at the corresponding bandpasses. The best fit at 8.8 µm gives a FWHM of 0.17 ±0.02′′. The N band size estimate is consistent, with larger error bars because of the lower S/N in these data. At the distance of Cen A, the measured FWHM angular size corresponds to a linear size of ≈ 3 pc, or ∼10 Ly. This is in agreement with the predicted size of the dust emitting region based on model fitting of ISOCAM spectra with emission from the dust lanes and from a separate toroidal dust region around the nucleus with a diameter of ∼ 3.6 pc (Alexander et al. 1999). -- 5 -- 3.2. Photometry We obtained photometry of the nuclear region using Cen A images in the two observed bandpasses by selecting a small aperture centered on the nuclear region. We used an aperture with a diameter of 20 MIRAC/BLINC pixels, corresponding to ∼ 2.4 arcsec, which is the minimum aperture size which includes most of the PSF flux. The photometric reference for the 8.8 µm image was γ Crux, which has a known flux of 1090 Jy at that wavelength, with an estimated uncertainty of 5% (Gezari et al. 1987). For the N band we used the source IRC+10220 as a reference. We estimated for this star an N band magnitude of ∼ 1.5 Jy, with a 30% uncertainty, based on interpolation of the K magnitude of 2.84 (∼ 45 Jy) and the IRAS 12 µm flux of ∼ 4 Jy from Gezari et al. (1987) (Table 1). The photometry of the Cen A nuclear region was calculated separately for the four nod/chop beams at each wavelength, which were then averaged together. This procedure ensured that the variation in the photometry of the four beams was within our error estimate. We also evaluated different sizes of the aperture and of the sky region for residual background subtraction, from 1.5 arcsec to 5 arcsec, finding similar values of the photometry, within our error bars (±1σ). We obtained an average flux of 0.9 Jy at 8.8 µm, with an uncertainty of ±0.1 Jy, and a flux of 1.4 Jy in the N band, with an uncertainty of ±0.5 Jy (Table 1). The significant uncertainties are due to the limited photometric quality of the sky during these observations. Larger aperture photometry (several arcseconds aperture) does not show significant differences within the error bars with the smaller aperture results that we are presenting in this paper. Our photometric results are in agreement with the mid-IR observations obtained by Whysong & Antonucci (2002) using the Keck I telescope. They detected emission from a compact unresolved nuclear source (with an upper limit on the size of ∼ 0.3′′) and measured fluxes of 1.6 Jy and 2.3 Jy at 11.7 µm and 17.75 µm, respectively. Our results are also in agreement within ≈ 3σ errors with the ISOCAM CVF spectrum obtained by Mirabel et al. (1999), which gives ∼ 0.7 Jy at 8.8 µm. In Figure 3 we plot our photometric results on the ISOCAM CVF spectrum (see Fig. 2 in Mirabel et al. 1999). The spectrum shows a broad silicate absorption feature around 9.8 µm, resulting in a lower flux at 10.3 µm with respect to 8.8 µm. Given the limited photometric and spectral accuracy of our mid-IR data we cannot discriminate between thermal emission from dust and the emission from late-type stars photospheres, because we essentially have only one color (the 8.8 micron filter is inside the N filter waveband). It is also likely that the eventual stellar radiation in the nuclear region is below our sensitivity limit. We therefore assume that the emission in the circumnuclear region is thermal, originating from dust surrounding the inner hot region. -- 6 -- 3.3. Dust Temperature Using the estimated size of the circumnuclear region at 8.8µm and the flux measurements based on our mid-IR observations, we derive an estimate of the temperature of the emitting dust. We assume that the emission is a black body thermal emission, and that there is an absorbing component associated with the dust lanes in front of the central source. For a simple spherical geometry of the emitting region we estimate the dust temperature at 8.8 µm using the following expression derived from the Planck law: T = 1440 (cid:20)ln(cid:18)1 + 1.1 · 105 R2 e−τν D2 Fν (cid:19)(cid:21)−1 (1) where T is the dust temperature in K, R is the radius of the region in pc, Fν is the flux in Jy, and τν is the optical depth of the dust lanes. Using a radius of ≈ 1.5 pc for the circumnuclear dust region, flux of 0.9 Jy, and τν ∼ 1 for the dust lanes optical depth at ∼ 10 µm, we estimate the dust temperature at ≈ 160 K. We compare the temperature obtained using this simplified model with the temper- ature derived from the ISOCAM spectra. The multicomponent spectral fit (see Figure 3 in Alexander et al. 1999) shows that the mid-IR emission from the circumnuclear region peaks at 20 µm, which corresponds to a dust temperature of T ≃ 150 K. This is in agree- ment with the estimated dust temperature of the circumnuclear region based on our mid-IR measurements. 4. Conclusions The results of our mid-IR imaging and photometry of the Cen A nuclear region suggest that the emission originates from a resolved source ≈3 pc in diameter. This emission is likely to be thermal radiation from dust at ≈150 K heated by UV/optical radiation from a "hidden" central AGN source (e.g. Peterson 1997; Whysong & Antonucci 2002). It is possible that we are detecting the external cooler region, possibly surrounding hotter inner dust (T ∼ 1000 K), as suggested by the deep silicate absorption feature in the ISO spectra around 10 µm. -- 7 -- Resolving the source of mid-IR emission in the nuclear region of Cen A provides an important key for understanding dust formation processes and the structure of the nuclear region of AGNs and quasars. For example, the current "unified" model of AGNs predicts a dusty torus with a size of ∼ 10 Ly surrounding the accretion disk and a black hole central en- gine (Antonucci and Miller 1985; Krolik and Begelman 1988; Peterson 1997). The predicted size of the torus is of the order of the measured size of the Cen A circumnuclear region using our 'mid-IR imaging. We note that the estimated size of the Cen A circumnuclear region is also similar to the size of the dusty region (∼ 3 pc) surrounding the central source of NGC 1068, recently resolved AGN using the VLTI observations (ESA Press release 17/03). The measured size of the mid-IR emitting region is also comparable to the distance from the central source at which dust will condense in an outflowing AGN wind. As recently demonstrated by Elvis et al. (2002), dust can be created from cooling Broad Emission Line clouds in AGN winds, and can survive in the quasar environment due to self shielding. This implies that Cen A could effectively be recycling dust from the host galaxy interstellar medium into IGM while changing the dust size and distribution and the dust-to-gas ratio. The possibility that AGN or quasars can provide an additional path for dust formation has important cosmological implications, since it could explain in a natural way the observed heavy obscuration of distant quasars (Omont et al. 2001; Elvis et al. 2002), and provide a new means of forming dust at early cosmological times. The current observations do not have enough accuracy to determine the true geometry of the emitting region or to distinguish between different geometries suggested by AGN models. The observations were challenging because of the limited telescope resolution. Further multi- wavelength observations of Cen A with higher spatial resolution are crucial to explore the geometry of the extended nuclear region and derive the physical characteristics and the origin of the emission. Multi band mid-IR observations combined with spectroscopy will allow determination of the characteristics of the circumnuclear dust, including its temperature, and an estimate of the mass and mass loss rate. The results may further constrain the unified AGN model, and models of dust formation in quasar outflows. We thank the Magellan staff for their outstanding support. We are grateful to Dr. Scott Wolk for valuable suggestions and aid in preparing the manuscript, and to the anonymous referee for useful comments and suggestions. M. K. and M. E. are members of the Chandra X- ray Center, which is operated by the Smithsonian Astrophysical Observatory under contract to NASA NAS8-39073. This research was supported in part by NASA through the American Astronomical Society's Small Research Grant Program. MIRAC is supported through SAO and NSF grant AST 96-18850. BLINC is supported through the NASA Navigator program. -- 8 -- REFERENCES Alexander, D.M., Efstathiou, A., Hough, J.H., Aitken, D., Lutz, D., Roche, P., Sturm, E. 1999, MNRAS, 310, 78. Antonucci, R.R.J., and Miller, J.S., 1985, ApJ, 297, 621 Packham, C., Hough, J.H., Young, S., Chrysostomou, A., Bailey, J.A., Axon, D.J., Ward, M.J. 1996, MNRAS, 278, 406 Draine, B.T., Lee, H.M. 1994, ApJ, 285, 89 Elvis, M., Marengo, M., and Karovska, M., 2002, ApJ Lett., 567, L107 Foltz, C.B, Williams, J.T., West, S.C., Fabricant, D.G., Martin, H.M. 1999, in proc. "16th IEEE Instrumentation and Measurement Technology Conference", eds. V. Piuri and M. Savino, p. 6333 Gezari, D.Y., Schmitz, M., Mead, J.M. 1987, "Catalog of Infrared Observations" Part I, NASA Reference Publication 1196 Hoffmann, W.F., Hora, J.L., Fazio, G.G., Deutsch, L.K., Dayal, A. 1998, Proc. SPIE 3354, 647 Israel, F. P. 1998, AA Rev., 8, 237 Karovska, M., et al. 2002, ApJ 577, 114 Krolik, J.H., and Begelman, M.C., 1988, ApJ, 329, 702 Marconi, A., Capetti, A., Axon, D. J., Koekemoer, A., Macchetto, D, and Schreier, E. J. 2001, ApJ, 549, 915 Mirabel I.F., et al. 1999, A&A, 341, 667 Omont, A., et al. 2001, A&A, 374, 371 Peterson, B. 1997, "An Introduction to Active Galactic Nuclei", Cambridge University Press. Schiminovich, D., van Gorkom, J. H., van der Hulst, J. M., and Kasow, S. 1994, ApJ, 423, L101 Shectman, S. & Johns, M. 2003, SPIE 4837, 910 Whysong, D., Antonucci, R. 2002, astro-ph/0207385. -- 9 -- This preprint was prepared with the AAS LATEX macros v5.0. -- 10 -- TABLE 1 MID-IR PHOTOMETRY Source Wavelength Photometry Cen A Cen A γ Crux IRC+10220 8.8 µm 10.3 µm 8.8 µm 10.3 µm 0.9 (± 0.1) Jy 1.4 (± 0.5) Jy 1009 (± 50) Jy 1.5 (± 0.4) Jy -- 11 -- Fig. 1. -- MIRAC/BLINC Images of Cen A (left) and the reference star γ Crux (right) at 8.8 µm. North is rotated 126 degrees to the left (from the vertical) Fig. 2. -- Radial profiles of Cen A image (solid line) and of the reference image (dashed line) at 8.8 µm. The radial profile of Cen A appears more extended when compared to the radial profile of the reference star, indicating therefore that Cen A is resolved. Fig. 3. -- ISOCAM CVF spectrum of the nuclear region (4" radius) of Cen A (Mirabel at al. 1999 showing several emission lines and a deep silicate absorption feature at ∼9 µm. Also plotted are the results from our 8.8µm and N band photometry of the 2.4" region centered on the nucleus. Cen A g Cru -- 1 2 -- 1" -- 13 -- -- 14 --
astro-ph/0409464
1
0409
2004-09-20T10:46:56
Using SKA to observe relativistic jets from X-ray binary systems
[ "astro-ph" ]
I briefly outline our current observational understanding of the relativistic jets observed from X-ray binary systems, and how their study may shed light on analogous phenomena in Active Galactic Nuclei and Gamma Ray Bursts. How SKA may impact on this field is sketched, including the routine tracking of relativistic ejections to large distances from the binaries, detecting and monitoring the radio counterparts to 'quiescent' black holes, and detecting the radio counterparts of the brightest X-ray binaries throughout the Local Group of galaxies.
astro-ph
astro-ph
4 Using SKA to observe relativistic jets from X-ray binary systems 0 0 2 a Astronomical Institute 'Anton Pannekoek', University of Amsterdam p e S 0 2 1 v 4 6 4 9 0 4 0 / h p - o r t s a : v i X r a R. Fendera I briefly outline our current observational understanding of the relativistic jets observed from X-ray binary systems, and how their study may shed light on analogous phenomena in Active Galactic Nuclei and Gamma Ray Bursts. How SKA may impact on this field is sketched, including the routine tracking of relativistic ejections to large distances from the binaries, detecting and monitoring the radio counterparts to 'quiescent' black holes, and detecting the radio counterparts of the brightest X-ray binaries throughout the Local Group of galaxies. 1. Introduction 1.1. Jets from X-ray binaries X-ray binary systems (Lewin & van der Klis 2004; Lewin, van Paradijs & van den Heuvel 1995) are the sites of the most dramatic, ongoing, high-energy astrophysical phenomena on non- cosmlogical scales. Compact, relativistic stars -- neutron stars or 'stellar mass' black holes -- ac- crete material from a binary companion in a rel- atively short (hours to weeks) period double sys- tem. An earlier brief discussion on the potential of SKA for the study of these objects was pre- sented in Fender (1999). The key signature of the relativistic accre- tion process in the radio band are the jets. As in Active Galactic Nuclei (AGN), and probably Gamma-Ray Bursts (GRBs) these jets seem to correspond to highly relativistic outflows of mat- ter, probably at least in part baryonic, from very close to the accreting central object (in some cases the launch point may be as close as a handful of gravitational radii, GM/c2). As with AGN and GRBs the emission mechanism is almost certainly synchrotron from relativistic electrons spiralling in magnetic fields. These synchrotron-emitting clouds of electrons have themselves relativistic bulk motions, along a (more or less) fixed axis, and this is the jet. Fig 1 provides a sketch of the probable geometry of a jet-producing X-ray binary. We know of currently 200-300 X-ray binaries (Liu, van Paradijs & van den Heuvel 2000,2001), probably corresponding to an underlying popu- 1 radio infrared optical soft-X hard-X gamma-ray COMPANION ACCRETION DISC JET CORONA ? Mass- donating companion star (IR-optical) Mass-flow Accreting neutron star or black hole Γ > 1 Jet (radio - ?) Accretion disc (optical - soft X-rays) 'Corona' (hard X-rays) Figure 1. A schematic diagram of a (low-mass) X- ray binary system, sketching the various physical components and regions of the electromagnetic spectrum in which they emit. The jet is unique in having an extremely broad spectral component, from being the only major contributor to the ra- dio emission to one of probably several emitting components in the X-ray band. When we detect an X-ray binary in the radio band we are almost certainly detecting the jet (see further arguments in Fender 2004). 2 lation of some tens of thousands of compact ob- jects in binary systems in our galaxy. The distri- bution of these X-ray binaries within our galaxy has been addressed most recently by Grimm, Gil- fanov & Sunyaev (2001) and Jonker & Nelemans (2004). We see the brightest X-ray binaries essen- tially to the other side of the galaxy. The distri- bution of systems from Grimm et al. is presented in Fig 2. High-mass and low-mass X-ray bina- ries (where the prefix applies to the mass of the companion star, not the accreting object) are dis- tributed differently, with the generally younger high-mass systems concentrated near regions of star formation, i.e. the spiral arms. Note how- ever, that there does not appear to be a strong effect on the disc-jet coupling whatever the mass of the companion star. This is natural since the inner regions of a bright accretion disc do not feel the influence of the companion star in any strong way. Nevertheless, in some cases the pho- ton field of a bright companion star might act as a source of photons for external comptonisation of the extended jet structure (e.g. Georganopoulos, Aharonian & Kirk 2002). Over the past decade, key observations with ATCA, MERLIN, The Ryle Telescope, VL(B)A, EVN and WSRT have provided remarkable in- sights into the variable process of jet formation in these systems. When coupled with simultane- ous X-ray measurements these observations have probed the dynamical coupling between accretion and outflow around accreting relativistic objects in a way which is not possible for AGN (which vary too slowly, in general) or GRBs (in which, in a sense, its all over in an instant). Based upon studies such as these, clear pat- terns have emerged in the coupling between ac- cretion and outflow in these sources, which we shall outline below. 1.1.1. Black hole X-ray binaries Fig 3 presents the radio vs. X-ray plane for nearly all the radio-loud black hole X-ray bina- ries (a notable exception is SS 433, which evades easy classification, but is essentially as radio-loud as Cyg X-3 at a significantly lower apparent X-ray luminosity), scaled to a distance of 1 kpc in order to compare luminosities (from Gallo, Fender & Pooley 2003). Examples are also shown of re- solved jets from different regions of the plane. Note that scaling by the estimated two orders of magnitude sensitivity improvement expected from SKA, much more of the plane will be acces- sible to direct imaging of the jets. The 'hard' (or 'low/hard') X-ray spectral state is ubiquitous at X-ray luminosites below about 1% of the Eddington luminosity (typically con- sidering a ∼ 10M⊙ black hole), and in some sources may persist to as bright as 10% of this in the rising phase of an outburst. The state gets its name from its hard X-ray spectra which shows only a weak blackbody component which may be ascribed to an accretion disc, but is in- stead dominated by a component which peaks in most sources around 100 keV and is gener- ally interpreted as arising in thermal Compton- isation (see e.g. McClintock & Remillard 2004 and references therein for this interpretation, and Markoff, Falcke & Fender 2001 for a 'synchrotron- based alternative'). In this state there is a 'compact' self-absorbed jet which manifests it- self as a 'flat' (spectral index α ∼ 0 where α = ∆ log Sν /∆ log ν) or 'inverted' (α ≥ 0) spectral component in the radio, millimetre and (proba- bly) infrared bands (e.g. Fender 2001, Corbel & Fender 2002). The radio luminosity of these jets shows a strong, non-linear correlation with X-ray luminosity (Corbel et al. 2003; Gallo, Fender & Pooley 2003) -- see Fig 3 up to a scaled X-ray flux of about 1 Crab. The correlation takes the form Lradio ∝ Lb X where b ∼ 0.7. This is consistent with the 'Funda- mental Plane' (see below, and Fig 6) for sources of all approximately the same mass. The steady jets which we infer to exist have only been directly spatially resolved in the case of Cyg X-1 (Stirling et al. 2001), although the 'plateau' jet of GRS 1915+105 is phenomenologically similar and has also been resolved (Dhawan et al. 2000; Fuchs et al. 2003). The suggestion that such steady, compact jets are produced even at very low accre- tion rates (Gallo, Fender & Pooley 2003; Fender, Gallo & Jonker 2003) has recently received sup- port in the flat radio spectrum observed from the 'quiescent' transient V404 Cyg at an average X- ray luminosity LX ∼ 10−6LEdd (Gallo, Fender & Hynes 2004). 3 the 'soft' Brighter sources may enter (or 'high/soft' or 'thermal dominated') X-ray state at higher luminosities (typically around 10 -- 50% Eddington). In this state the X-ray spectrum is dominated by a ∼blackbody component probably corresponding to an optically thick and geomet- rically thin accretion disc extending to the inner- most stable circular orbit (McClintock & Remil- lard 2004 review the properties in more detail). In this state the radio emission, and probably therefore jet production, is strongly suppressed (Tanabaum et al. 1972; Fender et al. 1999b; Gallo, Fender & Pooley 2003). This effect in which softer X-ray states reduce the radio emis- sion can be seen clearly in the interval 1 -- 10 Crab in Fig 3. Fig 4 (from Tigelaar 2004) presents the broad- band spectrum of the black hole X-ray binary Cygnus X-1 in hard (labelled 'Low/Hard') and soft(-er) (labelled 'High/soft or Intermediate') states. The change in the X-ray spectrum from hard to soft states, resulting in a reduction in emission around 100 keV (∼ 1019 Hz) and an in- crease in the temperature and luminosity of the accretion disc (around 1 keV ≡ 1017 Hz), are clearly associated with a dramatic 'quenching' of the radio component. Additionally there are bright events associated with transient outbursts and state transitions (of which more later), which are often directly re- solved into components displaying relativistic mo- tions away from the binary core (e.g. Mirabel & Rodriguez 1994; Hjellming & Rupen 1995; Fender et al. 1999,2002b) not only in the radio but also -- at least once -- in the X-ray band (Corbel et al. 2002). These events typically display opti- cally thin (synchrotron) radio spectra (α ≤ −0.5). Both kinds of jets are clearly very powerful and coupled to the accretion process. See Mirabel & Rodriguez (1999) and Fender (2004) for a more thorough review of the observational properties of X-ray binary jets. Fig 5 summarises our current best model for the phenomenology of the disc -- jet coupling in black hole X-ray binaries (from Fender, Belloni Figure 2. The distribution of known X-ray binary systems within the Milky Way. From Grimm, Gilfanov & Sunyaev (2002). Some of the dis- tances may be significantly underestimated (e.g. Jonker & Nelemans 2004), in which case we have probably observed sources up to 20+ kpc dis- tant. The Sun is at (0,8.5) kpc. Sold circles indi- cate high-mass X-ray binaries (those with massive companions) while open symbols indicate low- mass X-ray binaries (typically with companions of a solar mass or less). 4 C E S C R A i l l i M 25 20 15 10 5 0 -5 (i) 4 2 0 -2 2.0 cm (ii) C E S C R A i l l i M 8 6 4 2 0 -2 -4 -6 -8 3.6 cm C E S C R A i l l i M 8 6 4 2 0 -2 -4 -6 -8 -8 -10 -12 -14 8 6 4 2 0 -2 -4 -6 -8 MilliARC SEC 8 6 4 2 0 -2 -4 -6 -8 MilliARC SEC (iii) -4 -6 MilliARC SEC Figure 3. The radio : X-ray plane for galactic black hole X-ray binary systems, with all sources scaled to a distance of 1 kpc. The brightest transient sources have peak radio flux densities in excess of 1 Jy at GHz frequencies. The faintest sources currently detected at about the ∼ 0.1 mJy level. SKA will allow us to detect sources confidently down to the µJy level. From Gallo, Fender & Pooley (2003). The insets, clockwise from top left correspond to (i) the steady jet in the 'low/hard' state from Cyg X-1 (Stirling et al. 2001) (ii) the very powerful steady jet from GRS 1915+105 in the 'plateau' hard X-ray state (Fuchs et al. 2003) and (iii) relativistic ejections from GRS 1915+105 with spatially resolved linear polarisation (Fender et al. 2002b). 1e+06 10000 High/soft or Intermediate state Low/hard state 100 Radio/mm IR/optical X-ray Gamma-ray ) y J m ( y t i s n e d x u F l 1 0.01 0.0001 1e-06 1e-08 1e+08 1e+10 1e+12 1e+14 1e+16 1e+18 1e+20 1e+22 Frequency (Hz) Figure 4. The change in the broadband spectrum of the black hole X-ray binary Cygnus X-1 be- tween two X-ray 'states'. In the 'low/hard' state the accretion disc has a lower temperature, with a strong 'excess' around 100 keV, and the radio emission is 'on', with a relatively flat spectrum. This probably corresponds to the generation of a self-absorbed ∼conical jet during phases when we do not see a bright accretion disc extending to the innermost stable circular orbit. In the softer state the X-ray spectrum is dominated by a hotter accretion disc, probably closer to the black hole, and the radio emission (→ jet) is 'quenched'. & Gallo 2004). In essence, we suggest that un- til the disc comes very close to the black hole (around the time we observe it dominating the X-ray spectrum) a ∼steady jet is formed. During transitions to soft states as the disc is making its final 'collapse' inwards, the jet velocity increases sharply. This results in a relativisitically mov- ing, optically thin, internal shock, followed by a suppression of the 'core' radio emission while the source is in a soft X-ray state. The seemingly ubiquitous presence of jets in hard X-ray states has been taken as strong evidence for the magne- tohydrodynamic (MHD) production of jets (see e.g. Meier 2001). 5 1.1.2. Neutron star X-ray binaries The study of the disc-jet coupling in neutron star (NS) X-ray binaries has lagged behind that of the black holes in recent years. The reason for this is that the NS X-ray binaries are con- siderably less 'radio loud' than the black holes (Fender & Kuulkers 2001; Migliari et al. 2003 and in prep). Even when considering the mass term in the 'Fundamental plane' (see below), they appear to be less efficient at producing radio emission for a given X-ray flux. Despite complex patterns of behaviour in X- rays on relatively short (hours) timescales, the radio:X-ray plane for NS X-ray binaries appears similar to Fig 3, but with the neutron stars occu- pying a region a factor of ∼ 30 below the black holes (Migliari et al. in prep). This strongly im- plies that the study of these NS jet sources can tell us something about the black holes, since • The apparent similarity in the patterns tells us that the global properties of the disc-jet coupling in black holes are as a result of the accretion flow and not something unique to black holes themselves • The clear difference in radio power indi- cates that some difference between NS and black hole X-ray binaries (surface, radia- tion, magnetic field) is enough to affect the observed jet power Furthermore, we already have evidence that the most relativistic flows from binaries are in fact from the neutron star systems (see below). The detailed study of jets from accreting neutron stars is likely, for the reasons given above, to be crucial in our understanding of the jet formation process in black holes. 1.2. X-ray binaries as tools to understand AGN Much has been made in the past decade of the apparent analogy between the relativistic jets produced by supermassive (106M⊙ ≤ MBH ≤ 1010M⊙) black holes (AGN) and those produced in BH XRBs (3M⊙ ≤ MBH ≤ 20M⊙) -- hence the popular name 'microquasars' for this latter class of object. However, for most of this period the 6 iii iv VHS/IS Soft Hard LS HS Γ > 2 y t i s n e n i t X−ray hardness e n i l t e j r o t c a f z t n e r o L t e J no jet i s u d a r r e n n i c s D i Γ < 2 ii i iv ii iii i Figure 5. A schematic of our simplified model for the jet-disc coupling in black hole binaries. The central box panel represents an X-ray hardness-intensity diagram (HID); 'HS' indicates the 'high/soft state', 'VHS/IS' indicates the 'very high/intermediate state' and 'LS' the 'low/hard state'. In this diagram, X- ray hardness increases to the right and intensity upwards. The lower panel indicates the variation of the bulk Lorentz factor of the outflow with hardness -- in the LS and hard-VHS/IS the jet is steady with an almost constant bulk Lorentz factor Γ < 2, progressing from state i to state ii as the luminosity increases. At some point -- usually corresponding to the peak of the VHS/IS -- Γ increases rapidly producing an internal shock in the outflow (iii) followed in general by cessation of jet production in a disc-dominated HS (iv). At this stage fading optically thin radio emission is only associated with a jet/shock which is now physically decoupled from the central engine. As a result the solid arrows indicate the track of a simple X-ray transient outburst with a single optically thin jet production episode. The dashed loop and dotted track indicate the paths that GRS 1915+105 and some other transients take in repeatedly hardening and then crossing zone iii -- the 'jet line' -- from left to right, producing further optically thin r adio outbursts. Sketches around the outside illustrate our concept of the relative contributions of jet (blue), 'corona' (yellow) and accretion disc (red) at these different stages. comparisons remained largely phenomenological. However, in the past year all this has changed as quantitative scalings between jets, and between X-ray and radio power in black holes of all masses and all accretion rates have emerged (Heinz & Sunyaev 2003; Merloni, Heinz & di Matteo 2003; Falcke, Kording and Markoff 2004). Specifically, Merloni, Heinz & di Matteo (2003) and Falcke. Kording & Markoff (2004) have demonstrated the existence of a 'fundamental plane' of black hole activity in the 3D parameter space of black hole mass (MBH), X-ray luminosity (LX) and radio (Lradio) luminosity (Fig 6). This fundamental plane spans a range in > 108 in M , albeit cur- rently with a lot of scatter. Maccarone, Gallo & Fender (2003) have recently rebinned the data set of Merloni, Heinz & di Matteo to demonstrate the 'quenching' of jets in the same fractional Edding- ton range as known to occur in XRBs, suggesting the disc-jet phenomenology may also be the same. Detailed quantitative comparisons are only just beginning to be made; and will no doubt be the subject of many future research papers. At the very roughest level, it is tempting to asso- ciate the (disputed) 'radio loud' and 'radio quiet' dichotomy observed in AGN with jet-producing (hard and transient) and non-jet-producing (soft) states in X-ray binaries (see e.g. Maccarone, Gallo & Fender 2003), including of course the ef- fect of the mass term. Furthermore perhaps FRI jet sources can be associated with the low/hard state and FRIIs with transients. Meier (1999; 2001) has considered jet production mechanisms in both classes of object, and drawn interesting parallels. Gallo, Fender & Pooley (2003; amongst others!) have made a qualitative comparison be- tween FRIs and low/hard state black hole X-ray binaries and FRIIs and transients. It is interesting to note that the short timescale disc-jet coupling observed in GRS 1915+105 (Pooley & Fender 1997; Eikenberry et al. 1998; Mirabel et al. 1998; Klein-Wolt et al. 2001), in its most basic sense -- that radio events are preceded by a 'dip' and associated spectral hardening in the X-ray light curve -- may also have an analog in AGN: Marscher et al. (2002) have reported qualitatively similar behaviour in 3C 120. 7 2. Studying X-ray binaries with SKA The potential offered by the SKA to study the formation of jets in X-ray binaries, and their rela- tion to the accretion process (by means of simul- taneous X-ray observations) is immense. With a two orders of magnitude sensitivity leap and VLBI-scale angular resolution, we expect to reg- ularly resolve relativistic events from a large num- ber of X-ray binary systems, as well as detecting unresolved radio emission from radio cores at very low luminosities. 2.1. Monitoring bright sources A handful of X-ray binaries remain semi- continuously in very bright and variable X-ray and/or radio states ('outburst'). These systems include SS 433, Cygnus X-3, GRS 1915+105 (see Fig three inset (iii))and Circinus X-1 (see Fig 8). For such sources, with typical flux densities 1 -- 1000 mJy at GHz wavelengths, and previously resolved jet-like structures, we may expect low- duty-cycle radio monitoring to result in movies capturing the jet formation and propagation in exquisite detail. Furthermore, bright (close to Eddington at peak) new X-ray transients are expected to pro- duce (sequences of) strong radio outbursts (prob- ably corresponding to internal shocks; see Fig 5) with a frequency typically around ∼ 1/year. Such sources when resolved typically reveal ejections with projected velocities in the range 1 -- 15c, de- celerating at larger distances from the core (see e.g. Fender 2004 and references therein; Fender, Belloni & Gallo 2004; Kaaret et al. 2003). These jets typically fade rather rapidly (with e-folding times of hours to weeks) and become unobserv- able due to the sensitivity of the radio array be- fore we have observed them doing anything 'in- teresting' such as interacting with the ISM. SKA has the potential to track these sources much fur- ther from the core, observing such interactions and decelerations. Furthermore, the short inte- gration time required to make 'snapshot' monitor- ing observations of recent transients will undoubt- edly facilitate the discovery of more 'rebrighten- ing' events such as those observed in the case of XTE J1550-564 years after the initial outburst 8 Figure 6. The fundamental plane of black hole activity (Merloni, Heinz & di Matteo 2003; see also Falcke, Kording & Markoff 2004). Combined data from both X-ray binaries and AGN indicates the existence of an approximate plane in the Lradio:LX:MBH parameter space. This underlines the scale-invariance of the accretion:outflow coupling (Heinz & Sunyaev 2003) and demonstrates that results from X-ray binaries and AGN may be compared quantitatively. (Corbel et al. 2002). The jets from these sources are also known to display both linear and circular polarisation (e.g. Fender et al. 1999,2000a,b; Macquart & Fender 2004). Linear polarisation observations provide unique information on the degree or ordering and orientation of the magnetic field within the ejecta. The situation is clearly not straightforward -- spa- tially resolved linear polarisation images of GRS 1915+105 made with MERLIN (Fender et al. 1999) revealed a field which was different each day. Other, unresolved, events from the same source observed with ATCA have revealed both constant polarisation angles and 'rotator' events (Fender et al. 2002a,b -- see inset (iii) in Fig 3). Circular polarisation (Macquart & Fender 2004) offers a complex but potentially very powerful in- sight into the composition of relativistic jets. For example, the detection of a circular polarisation spectrum of the form V /I ∝ ν−0.5 from a homoge- nous, optically thin, synchrotron source would be strong evidence for a baryonic component in the jet plasma, something which (with the exception of SS 433) has eluded observations so far. 2.2. Resolving 'steady' jets in hard X-ray states As outlined above (see e.g. Fender 2004 for a more detailed discussion), black holes in 'hard' X-ray states seem to be ubiquitously associated with steady jet production (see insets (i) and (ii) in Fig 3). The power and physical conditions in these jets are central to our understanding of how accretion proceeds near a black hole. The over- whelming majority of black holes of all masses are likely to exist in this state. Observations of synchrotron time lags (e.g. Mirabel et al. 1998; Fender et al. 2002a) as well as direct imaging (Fuchs et al. 2003) in- dicate that the compact jet in GRS 1915+105 has a physical size of around 1014−15 cm at GHz wavelengths for a flux density in the range 40 -- 120 mJy. If we assume that this physical size corre- sponds to the distance from the black hole to the τ ∼ 1 surface (where τ is the optical depth) in a self-absorbed jet (e.g. Blandford & Konigl 1979; see also chapter by Falcke, Kording & Nagar), and that all compact jets have approximately the 9 same brightness temperature (as seems to be the case for AGN -- e.g. Ghisellini et al. 1993), then we can estimate the physical size of such a jet as rGHz ∼ 5 × 1014(cid:18) Sν 40 mJy(cid:19)(cid:18) d 11 kpc(cid:19)2 cm This corresponds to an angular size of α ∼ 35(cid:18) Sν 40 mJy(cid:19)(cid:18) d 11 kpc(cid:19) mas which means that for a given radio flux density, a distant source will be more easy to resolve than a nearby one, because it is more powerful. This function is plotted in Fig 7. A typical hard state X-ray transient will have a flux density of ≥ a few mJy and lie at a distance of 2 -- 20 kpc. Such sources are potentially resolveable with high sen- sitivity VLBI. Note that Gallo, Fender & Hynes (2004) have recently measured the radio spectrum of the black hole X-ray binary V404 Cyg in 'qui- escence' and have found it to be similar to that of Cyg X-1, supporting the interpretation of low- level radio emission originating in a compact jet. The handful of measurements of linear polari- sation from hard state sources indicate a typical level of ∼ 2% (Fender 2001). Furthermore, the polarisation angle seems to be aligned with the jet axis, for at least one source (Gallo et al. 2004). Careful monitoring of variations with time of the linear polarisation vector could, potentially, re- veal the precession of steady jets whose angular extent will never allow them to be directly re- solved (the effect can be quite large: in SS 433 the jets precess with a half-angle of ∼ 20◦). 2.3. Black holes in 'quiescence' -- probing advection The power-law correlation found for low/hard state black hole binaries (Corbel et al. 2003; Gallo et al. 2003), combined with X-ray mea- surements, allows us to predict the radio lumi- nosities of 'quiescent' black hole systems (Gallo et al. 2003). These systems are believed to be strongest examples of 'advection dominated ac- cretion flows' (ADAFs) and typically have soft X-ray luminosities around LX ∼ 10−8 Edding- ton. The predicted quiescent radio flux densities are in the range 1 -- 30 µJy (Gallo et al. 2003). Compact jets in low/hard state 30 mJy 3 mJy 0.3 mJy 30 microJy 3 microJy 10 ) s a m ( e z i s r a l u g n a 100 10 1 0.1 0.01 0.001 0.0001 1e-05 0.1 1 10 distance (kpc) Figure 7. Estimates (GHz) size of compact 'core' jets from black holes for GHz radio flux densities of 30, 3 and 0.3 mJy. A typical hard state has a flux density of ≥ few mJy at a distance of 2 -- 20 kpc and so is potentially resolveable with high sensitivity VLBI. Fender, Gallo & Jonker (2003) demonstrated that the same power-law relation, if maintained to such low luminosities, would be very strong ev- idence for the existence of 'jet dominated states' in which the dominant power output channel for the liberated accretion energy is a radiatively- inefficient jet, and not X-ray emission. Measure- ments of the broadband spectra of these 'qui- escent' black holes in the way done for Cygnus X-1 at higher luminosity (Fig 4) is strongly lim- ited by our radio sensitivity (and also γ-ray, but this component is rather unexplained anyhow!). We have recently obtained a four-frequency radio spectrum of a black hole X-ray binary, V404 Cyg, with a mean X-ray luminosity of ∼ 10−6 Edding- ton (Gallo, Fender & Hynes 2004). The radio spectrum is flat, supporting a jet interpretation, and the mean radio flux density is ∼ 0.3 mJy. The SKA will allow us to measure the broad- band radio spectra of the truly 'quiescent' sources, which are expected to be one to two or- ders of magnitude fainter than V404 Cyg (Gallo, Fender & Pooley 2003). which may be combined with IR/optical/X-ray measurements to allow us to estimate the jet power at such low accretion rates. Furthermore, a galactic plane survey with SKA may well discover a population of unknown quiescent accreting objects which stand out due to their compact size and flat spectra, but would not stand out in X-ray surveys. Inspection of Fig 7 shows that its rather unlikely we will be able to directly resolve the jets from such weak sources, with anticipated micro-arcsecond angular sizes. 2.4. Intermediate mass black holes in glob- ular clusters? It has been argued that globular clusters are likely to contain in their centres 'intermediate mass' black holes, formed via stellar mergers. The masses of these central black holes may be as high as 10−3 of the cluster mass. Maccarone (2004) has recently demonstrated that the fundamental plane of black hole activity implies that radio emission is a much more strin- gent test for the existence of these intermediate mass black holes in the centres of globular clus- ters than X-ray observations. For accretion from the ambient medium in these clusters, radio flux densities in the range 1 -- 100 µJy are expected. The SKA will be able to achieve such sensitiv- ities with relative ease, and strong tests for the existence of the postulated central black holes in globular clusters may be undertaken. 2.5. Extremes and unexpected phenomena Just as we are preparing to settle into a pat- tern of observation designed to confirm and con- solidate our ideas, along come, thankfully, some unexpected phenomena, There is no doubt that SKA will discover more such phenomena, but the examples outlined below both illustrate the sur- prises and present areas in which SKA will be able to make significant progress. 2.5.1. Ultrarelativistic flows from neutron stars Fomalont et al. (2001a,b) and Fender et al. (2004; see Fig 8) have reported evidence for the existence of highly relativistic, but essentially 'un- seen' outflows from neutron stars accreting at high rates. In the case of Sco X-1 (Fomalont et al. 2001a,b) the outflow is observed to move with a bulk Lorentz factor Γ ≥ 3; in the case of Cir X-1 (Fender et al. 2004) the lower limit is much greater Γ ≥ 15. This is probably due to the small angle of the Cir X-1 jet to the line of sight and suggests that at least all six neutron star 'Z' sources (of which Sco X-1 is the archetype) are producing such 'ultrarelativistic' flows. SS 433 may also be producing such a flow, resulting in transient energisation of the well-known knots which move at ∼ 0.26c (Migliari et al. 2004). These flows are rather unexpected in the sim- ple, widely-accepted framework in which jets re- flect the escape velocity in the region in which they are formed, since this should not be more than ∼ 0.4c for a neutron star (e.g. Livio 1999). The nature of these flows also remains a mystery; for example they may be bright but so fast they they're always beamed out of our line of sight, or they may be 'cold' or highly radiatively ineffi- cient and only ever observed via their interactions with other components. Their importance lies in the fact that they clearly demonstrate that what- ever it is that is necessary for the formation of highly relativistic flows, it is not something which 11 is unique to black holes. The very rapid variabil- ity of the jet in Cir X-1, with a projected velocity ≥ 15c and hints in the radio maps of a very tran- sient structure (Fender et al. 2004) highlight the need for high sensitivity snapshots such as will be provided by SKA. 2.5.2. Jets from obscured X-ray binaries as γ-ray sources Paredes et al. (2000) reported the discovery of a powerful, seemingly persistent, radio jet from the massive X-ray binary LS 5039. This system had however a strange broadband spectrum, be- ing weak in X-rays but potentially associated with an unidentified EGRET gamma-ray source. Furthermore, INTEGRAL observations have revealed what might be a population of X-ray bi- naries undergoing such intense local absorption (by gas, and maybe also dust, local to the binary system), that they have remained undetected by previous soft X-ray surveys. In such sources, whatever the source of the X-ray absorption, the jet is likely to extend beyond this region, and they will remain detectable as radio sources. A deep, multi-frequency radio survey of the galactic plane with SKA may well discover many such sources. 2.6. Beyond the Milky Way No X-ray binary system has clearly been unam- biguously detected in the radio band beyond the Milky Way, despite the observations of hundreds of such sources in external galaxies (e.g. Fabbiano & White 2004). The radio:X-ray plane for galactic black hole binaries presented in Fig 3 may be simply scaled to the distances of local galaxies to see how SKA may help our cause in this area. The top panel of Fig 9 shows the plane scaled to the distance of the LMC (and, effectively, the SMC). Bright transients in these two satellite galaxies are al- ready potentially visible with e.g. ATCA, and would be easily accessible to SKA. Perhaps more importantly, so would the steady 'low/hard' state sources, and the brightest neutron stars. Since we know of at least one of each class of object in the LMC (LMC X-3 and X-2 respectively), we could test the possible dependence of the disc-jet in both black holes and neutron stars as a func- 12 Oct 2000 core B A X−ray flaring May 2001 Dec 2002 8 arcsec (0.6 light years) 4.8 8.6 4.8 8.6 4.8 8.6 GHz Figure 8. An ultrarelativistic outflow: sequences of radio observations of Circinus X-1 in October 2000, May 2001 and December 2002. At each epoch, observations were made simultaneously at 4.8 and 8.6 GHz. White tickmarks indicate time steps of one day; the blue bar indicates the time of the X-ray flaring as observed by the Rossi X-ray Timing Explorer All-Sky Monitor. In October 2000 and May 2001 the observations were spaced every two days; in December 2002 they are daily. At each epoch the u-v coverage of the radio observations is identical for each image; maps in October 2000 and May 2001 are 'uniformly weighted', those in December 2002 are 'naturally weighted'. The crosses indicate the location of the binary 'core' from [?], and their size is proportional to the 'core' radio flux density. The observations reveal that following the X-ray flaring the extended radio structure brightens on timescales of days. The apparent velocities associated with this expansion are ≥ 15c, indicating an underlying ultrarelativistic flow. (a) (b) Figure 9. Top panel (a): As Fig 3, but scaled for a distance of 55 kpc, corresponding to the LMC, and also approximately appropriate for the SMC and Sagittarius, Sculptor, Sextans, Draco and Ursa Minor dwarf spheroidals. Clearly transient radio sources in these nearby galxies will be easily detectable with SKA, as will the brightest of the 'steady' (hard X-ray state) sources. Lower panel (b): as (a) but scaled to M31 / M33. Even at this distance, corresponding effectively to the ra- dius of the Local Group of (∼ 40) galaxies, bright radio transients should be readily detectable with SKA. The field of M31 / M33 in particular, as it will no doubt be regularly monitored in X-rays, will provide fertile ground for comparing the disc- jet coupling with that in our own galaxy. 13 Figure 10. The local group of galaxies. The SKA will detect persistent (hard X-ray state) binaries in the nearest galaxies (LMC, SMC and other nearby dwarf galaxies). Transient sources will be detectable throughout the local group, covering approximately 40 galaxies and ∼ 1011 stars. tion of metallicity. Extending our view across the Local Group of galaxies to M31 (Fig 9, lower panel), we can see that relatively little effort still is required to de- tect bright transients at this distance. Even the brightest steady sources would be detectable with sufficiently long (∼ 1 day) integrations. This is a very exciting prospect, allowing us to cal- ibrate the radio and X-ray correlation without the strong distance uncertainties which plague us within the Milky Way. Fig 10 illustrates in summary the 'reach' of SKA for the study of X-ray binary systems. The inner dark circle indicates those regions for which the detection of radio emission from both tran- sients and steady sources would be more or less trivial. For sources at the distance of the lighter, larger, ring (encompassing nearly all the mass of the Local Group), transients would still be rela- tively easy to detect, and steady sources would be accessible via long integrations. 14 3. Requirements for the SKA Acknowledgements The exact specifications for the SKA are slowly being assembled. Such a large step in sensitivity will have enormous benefits for this field of re- search, whatever the precise design. Nevertheless, it is useful to lay out what we have anticipated for the design, and what will be important for the study of jets from X-ray binaries. The author would like to thank Elena Gallo for producing Fig 9. Robert Braun, Heino Fal- cke, Ben Stappers and many others have con- tributed indirectly to this work via many useful discussions about the detection of transient radio sources with the next generation of radio tele- scopes. • The discussions in this document have been based upon a SKA which has, a sensitivity at GHz frequencies which is about two or- ders of magnitude better than the current VLA / ATCA / WSRT. This corresponds to, for example, a 10-min sensitivity at 5 GHz of ∼ 1µJy. • The angular resolution of the SKA required for the various studies outlined in this pa- per should be comparable to that obtained currently at a few GHz with VLBI. This is not only necessary for the direct imaging of jets for sources within our own galaxy but also for clearly resolving individual sources from the background at larger distances. • The frequency range of the proposed SKA is also important for the context of X-ray bi- nary studies. A broad frequency range will allow the spectrum of the radio emission, which we have found to be dependent on the X-ray state and -- possibly -- to the X- ray luminosity within that state, to be well determined. • Simultaneous multi-frequency capability will be important. The observations of time delays in the synchrotron emission from the powerful jet source GRS 1915+105 (e.g. Mirabel et al. 1998; Fender et al. 2002a) give us a direct insight into the scale of the jet unobtainable by other means. • Good polaristion sensitivity ('purity'). Both linear and circular polarisation pro- vide important clues about the structure and composition of the synchrotron emit- ting components which cannot be otained in any other way. REFERENCES 1. Corbel S., Fender R.P., 2002, ApJ, 573, L35 2. Corbel S., Fender R.P., Tzioumis A.K., Tom- sick J.A., Orosz J.A., Miller J.M., Wijnands R., Kaaret P., 2002, Science, 298, 196 3. Corbel S., Nowak M.A., Fender R.P., Tzioumis A.K., Markoff S., 2003, A&A, 400, 1007 4. Dhawan V., Mirabel I.F., Rodr´ıguez L.F., 2000, ApJ, 543, 373 5. Fabbiano G., White N.E., 2004, 2004, in 'Compact Stellar X-ray Sources', eds. W.H.G. Lewin and M. van der Klis, in press, (astro-ph/0307077) 6. Falcke H., Kording E., Markoff S., 2004, A&A, 414, 895 7. Fender R.P., 1999, In 'Perspectives on Ra- dio Astronomy: Science with Large Antenna Arrays', M.P. van Haarlem (Ed), ASTRON, Dwingeloo, The Netherlands 8. Fender R.P., 2001, MNRAS, 322, 31 9. Fender R.P., 2004, in 'Compact Stellar X-ray Sources', eds. W.H.G. Lewin and M. van der Klis, in press, (astro-ph/0303339) 10. Fender R.P., Gallo E., Jonker P., 2003, MN- RAS, 343, L99 11. Fender R.P., Belloni T., Gallo E., 2004, MN- RAS, submitted 12. Fender R.P., Garrington S.T., McKay D.J., Muxlow T.W.B., Pooley G.G., Spencer R.E., Stirling A.M., Waltman E.B., 1999a, MN- RAS, 304, 865 13. Fender R. et al., 1999b, ApJ, 519, L165 14. Fender R.P., Rayner D., Trushkin S.A., O'Brien K., Sault R.J., Pooley G.G., Norris R.P., 2002a, MNRAS, 330, 212 15. Fender RP, Rayner D, McCormick DG, 15 Muxlow TMB, Pooley GG, et al. 2002b. MN- RAS. 336:39 -- 46 16. Fender R., Wu K., Johnston H., Tzioumis T., Jonker P., Spencer R., van der Klis M., 2004, Nature, 427, 222 17. Fomalont E.B., Geldzahler B.J., Bradshaw C.F., 2001a, ApJ, 553, L27 36. Maccarone T., Gallo E., Fender R.P., 2003, 345, L19 37. Macquart J.-P., Fender R.P. (Eds), 2004, Circular Polarisation from Relativistic Jet Sources', Kluwer Academic Publishers, Dor- drecht, NL 38. Markoff S., Falcke H., Fender R.P., 2001, 18. Fomalont E.B., Geldzahler B.J., Bradshaw A&A, 372, L25 C.F., 2001b, ApJ, 558, 283 19. Fuchs Y. et al., 2003, A&A, 409, L35 20. Gallo E., Fender R.P., Pooley G.G., 2003, MNRAS, 344, 60 21. Gallo E., Fender R.P., Hynes R.I., 2004, MN- RAS, submitted [] Gallo E., Corbel S., Fender R.P., Maccarone T.J., Tzioumis A.K., 2004, MNRAS, 347, L52 22. Georganopoulos M., Aharonian F.A., Kirk J.G., 2002, A&A, 388, L25 39. Meier D.L., 1999, ApJ, 522, 753 40. Meier D.L., 2001, ApJ, 548, L9 41. Merloni A., Heinz S., di Matteo T., 2003, MN- RAS, 345, 1057 (MHdM03) 42. Migliari S., Fender R.P., Rupen M., Jonker P.G., Klein-Wolt M., Hjellming R.M., van der Klis M., 2003, MNRAS, 342, L67 43. Migliari S., Fender R.P., Blundell K., Mendez M., van der Klis M., 2004, MNRAS, submit- ted 23. Ghisellini G., Padovani P., Celotti A., 44. Mirabel, I.F., Rodr´ıguez, L.F, 1994, Nature, Maraschi L., 1993, ApJ, 407, 65 24. Grimm H.J., Gilfanov M., Sunyaev R., 2002, MNRAS, 391, 923 25. Heinz S., Sunyaev R.A., 2003, MNRAS, 343, L59 371, 46 45. Mirabel I.F., Dhawan V., Chaty S., Rodr´ıguez L.F., Mart´ı J., Robinson C.R., Swank J., Geballe T., 1998, A&A, 330, L9 46. Paredes J.M., Marti J., Ribo M., Massi M., 26. Hjellming R.M., Rupen M.P., 1995, Nature, 2000, Science, 288, 2340 47. Stirling A.M., Spencer R.E., de la Force C.J., Garrett M.A., Fender R.P., Ogley R.N., 2001, MNRAS, 327, 1273 48. Tanabaum H., Gursky H., Kellogg E., Giac- coni R., Jones C., 1972, ApJ, 177, L5 49. Tigelaar S., 2004, MSc thesis, University of Amsterdam 375, 464 27. Jonker P.G., Nelemans G., 2004, MNRAS, submitted 28. Kaaret P., Corbel S., Tomsick J.A., Fender R., Miller J.M., Orosz J.A., Tzioumis T., Wi- jnands R., 2003, ApJ, 582, 945 29. Lewin W.H.G., van der Klis M. (Eds), 2004, 'Compact Stellar X-ray Sources', Cambridge University Press, in press 30. Lewin W.H.G., van Paradijs J., van den Heuvel E.P.J., 1995, 'X-ray binaries', Cam- bridge University Press, Cambridge Astro- physics Series 26. 31. Liu Q.Z., van Paradijs J., van den Heuvel E.P.J., 2000, A&AS, 147, 25 32. Liu Q.Z., van Paradijs J., van den Heuvel E.P.J., 2001, A&A, 368, 1021 33. Livio M., 1999, Physics Reports, 311, 225 34. McClintock J.E., Remillard R.A., 2004, in 'Compact Stellar X-ray Sources', eds. W.H.G. Lewin and M. van der Klis, in press, (astro-ph/0306213) 35. Maccarone T., 2003, A&A, 409, 697
astro-ph/0506556
1
0506
2005-06-23T09:45:48
Molecular hydrogen in the diffuse interstellar medium at high redshift
[ "astro-ph" ]
Physical conditions within DLAs can reveal the star formation history, determine the chemical composition of the associated ISM, and hence document the first steps in the formation of present day galaxies. Here we present calculations that self-consistently determine the gas ionization, level populations (atomic fine-structure levels and rotational levels of H_2), grain physics, and chemistry. We show that for a low-density gas (n< 0.1 cm^-3) the meta-galactic UV background due to quasars is sufficient to maintain H_2 column densities below the detection limit (i.e N(H_2)< 10^14 cm^-2) irrespective of the metallicity and dust content in the gas. Such a gas will have a 21 cm spin temperature in excess of 7000 K and very low C I and C II* column densities for H I column densities typically observed in DLAs. We show that the observed properties of the ~15% per cent of the DLAs that do show detectable H_2 absorption cannot be reproduced with only the quasar dominated meta-galactic UV radiation field. Gas with higher densities (n>10 cm^-3), a moderate radiation field (flux density one to ten times that of the background radiation of the Galactic ISM), the observed range of metallicity and dust-to-gas ratio reproduce all the observed properties of the DLAs that show H_2 absorption lines. This favors the presence of ongoing star formation in DLAs with H_2. The absence of detectable H_2 and C I absorption in a large fraction of DLAs can be explained if they originate either in a low-density gas or in a high-density gas with a large ambient radiation field. The absence of 21 cm absorption and C II* absorption will be consistent with the first possibility. The presence of 21 cm absorption and strong C II* without H_2 and C I absorption will suggest the second alternative. (Abridged)
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 000 -- 000 (0000) Printed 1 October 2018 (MN LATEX style file v2.2) Molecular hydrogen in the diffuse interstellar medium at high redshift R. Srianand1, G. Shaw2, G. J. Ferland2, P. Petitjean3,4, C. Ledoux5 1 IUCAA, Postbag 4, Ganeshkhind, Pune 411007, India; [email protected] 2 Department of Physics and Astronomy, University of Kentucky, 177 Chem.-Phys. Building, Lexington, KY 40506, USA; [email protected], [email protected] 3 Institut d'Astrophysique de Paris -- CNRS, 98bis Boulevard Arago, F-75014 Paris, France; [email protected] 4 LERMA, Observatoire de Paris, 61 Avenue de l'Observatoire, F-75014, Paris, France 5 European Southern Observatory, Alonso de C´ordova 3107, Casilla 19001, Vitacura, Santiago, Chile; [email protected] Typeset 1 October 2018; Received / Accepted ABSTRACT The physical conditions within damped Lyα systems (DLAs) can reveal the star for- mation history, determine the chemical composition of the associated ISM, and hence document the first steps in the formation of present day galaxies. Here we present cal- culations that self-consistently determine the gas ionization, level populations (atomic fine-structure levels and rotational levels of H2), grain physics, and chemistry. We show that for a low-density gas (nH 6 0.1 cm−3) the meta-galactic UV background due to quasars is sufficient to maintain H2 column densities below the detection limit (i.e N(H2)6 1014 cm−2) irrespective of the metallicity and dust content in the gas. Such a gas will have a 21 cm spin temperature in excess of 7000 K and very low C i and C ii ∗ column densities for H i column densities typically observed 50 per cent in DLAs. We show that the observed properties of the ∼ 15 per cent of the DLAs that do show detectable H2 absorption cannot be reproduced with only the quasar dominated meta-galactic UV radiation field. Gas with higher densities (nH > 10 cm−3), a moder- ate radiation field (flux density one to ten times that of the background radiation of the Galactic ISM), the observed range of metallicity and dust-to-gas ratio reproduce all the observed properties of the DLAs that show H2 absorption lines. This favors the presence of ongoing star formation in DLAs with H2. The absence of detectable H2 and C i absorption in a large fraction of DLAs can be explained if they originate either in a low-density gas or in a high-density gas with a large ambient radiation field. The absence of 21 cm absorption and C ii ∗ absorption will be consistent with the first possibility. The presence of 21 cm absorption and strong C ii ∗ without H2 and C i absorption will suggest the second alternative. The N(Al ii)/N(Al iii) ratio can be used to understand the physical properties when only C ii ∗ (without H2) is less than that typically inferred from the components with H2 absorption. We also calculate the column density of various atoms in the excited fine-structure levels. The expected column densities of O i ∗ in a high-density cold gas is in the −1012 cm−2 for log N(H i)> 20 and the observed range of metallicities. It range of 1011 will be possible to confirm whether DLAs that do not show H2 originate predominantly in a high-density gas by detecting these lines in very high S/N ratio spectra. ∗ absorption is present. We find nH in components that show C ii ∗, O i ∗∗, and Si ii Key words: Galaxies: formation Quasars: absorption lines-ISM: molecules-Intergalactic medium- 1 INTRODUCTION Damped Lyα systems seen in QSO spectra are characterized by very large H i column densities, N(H i)& 1020 cm−2, that are similar to those observed through gas-rich spiral galax- ies. The importance of DLAs in the paradigm of hierarchical structure formation can be assessed from the fact that the mass density of baryonic matter in DLAs at zabs∼ 3 is sim- ilar to that of stars at present epochs (Wolfe 1995). Studies c(cid:13) 0000 RAS 2 R. Srianand et al. of Lyα and UV continuum emission from galaxies associated with DLAs usually reveal star formation rates(SFR) (or up- per limits) of a few M⊙ yr−1 (Fynbo et al. 1999; Bunker et al. 1999; Kulkarni et al. 2001; Moller et al. 2002 and 2004; Weatherley et al. 2005). The metallicity and depletion fac- tor in DLAs are usually estimated from N(Zn ii)/N(H i) and N(Fe ii)/N(Zn ii) respectively (Lu et al. 1996; Pettini et al. 1997; Prochaska & Wolfe 2002; Ledoux et al. 2002a and Khare et al 2005). The inferred metallicities typically vary between log Z = −2.5 to 0 for 2 6 zabs6 3 with a median of ≃ −1.3 (Ledoux et al. 2003). The measured depletions range between 0 to −1.6 with a median value of −0.3. If dust content is defined as κ = 10[Zn/H] (1 − 10[Fe/Zn]), then the median dust content in a typical DLA is κ = 0.07. This is less than 10 per cent of what is seen in the Galac- tic ISM for a similar neutral hydrogen column density. If the physical conditions in DLAs are similar to those in our Galaxy or the Magellanic clouds then H2 molecules should be detectable. There have been very few detections of H2 in DLAs (with 3 × 1014 6 N (H2)(cm−2) 6 3 × 1019) despite extensive searches (Ge & Bechtold 1999; Srianand, Petit- jean & Ledoux 2000; Petitjean, Srianand & Ledoux 2000; Ledoux, Srianand & Petitjean 2002b; Levshakov et al. 2002; Ledoux, Petitjean, & Srianand 2003; Reimers et al. 2003). Roughly 80 per cent of DLAs do not have detectable H2 (with N (H2) 6 1014 cm−2). The physical conditions in the H i gas can be probed by using the fine-structure absorption lines produced by the excited atomic species and the 21 cm transition. Apart from a few rare cases (for example Srianand & Petitjean 2001), C i is detected only in systems that show H2. The derived total hydrogen density (nH ) based on the fine-structure level populations of the heavy elements are usually large (> 20 cm−3) (Ledoux et al. 2002b; Srianand et al. 2005; Wolfe et ∗ absorption is detected in every case al. 2004). Like C i, C ii where one detects H2. However, it has also been seen with lower column densities in a considerable fraction of DLAs without H2 (Wolfe et al. 2003a; Srianand et al. 2005, Wolfe et al. 2004). The search for 21 cm absorption in DLAs at zabs> 2 have mostly resulted in null detections with typical spin tem- peratures > 103 K (Table 3 of Kanekar & Chengalur (2003) and Table 1 of Curran et al. (2005) for a summary of all the available observations). There are 8 DLAs at zabs> 1.9 for which redshifted 21 cm observations are available. Red- shifted 21 cm absorption is detected for only two systems (zabs= 1.944 toward PKS 1157+014 (Wolfe et al. 1981) and zabs= 2.0394 toward PKS 0458−02 (Briggs et al. 1989)). The measured spin temperatures are 865±190 K and 384±100 K. However, none of these systems show detectable H2 (Ledoux et al. 2003; Ge & Bechtold 1997). The DLA toward PKS 1157+014 is special as the QSO shows broad absorption lines and the zabs of the DLA is close to the zem of the QSO. The physical conditions in this system may not be a good representative of the general DLA populations. Inter- estingly, H2 is seen at zabs= 2.8110 toward PKS 0528-2505, while no 21 cm absorption is detected in this system (Carilli et al. 1996; Srianand & Petitjean 1998). The lower limit on the spin temperature derived for this system is 710 K. How- ever, the excitation temperature derived from H2 rotational levels is 6 200 K (Srianand & Petitjean 1998; Srianand et al. 2005). This system is also special since zabs>zem . The radiation field of the QSO has a much stronger influence on the physical conditions in this DLA (Srianand & Petitjean 1998). The upper limits on the spin temperature derived for the rest of the systems are higher than 1000 K. The H2 content of these systems is not known. Even though various properties of DLAs, listed above, have been investigated in detail (Matteucci et al. 1997; Prantzos & Boissier 2000; Howk & Sembach 1999; Izotov et al. 2001; Vladilo, 2001; Liszt 2002; Hirashita et al. 2003; Wolfe et al. 2003a,b; Calura et al. 2003; Wolfe et al. 2004), very few attempts have been made to investigate all of them in a single unified calculation. That is the main motivation of this paper. 2 CALCULATIONS The main aim of our study is to investigate the physical conditions in high-z DLAs. In particular, our goal is to un- derstand the equilibrium abundance of H2, the excitations of H2 and atomic fine-structure levels, the ionization state of the gas, and the 21 cm optical depth, self-consistently. In the Galactic interstellar medium (ISM) H2 is usually detected either in a diffuse medium irradiated by the UV background radiation field or in high-density photo-dissociation regions (PDRs) near OB stars. One can anticipate this to also be the case in DLAs. At high redshift the diffuse UV background from QSOs will be an additional source of UV radiation. Recently there have been three attempts to model H2 in DLAs. Liszt (2002) uses two phase models similar to Wolfire et al. (1995) for this purpose. The models consider dust-free gas, so only the slow gas-phase H− + H → H2 + e forma- tion process is important. The second attempt by Hirashita et al. (2003) models DLAs as large protogalactic disks. The density and temperature in the gas are determined by "non- linear hydrodynamic" effects. The radiation field is assumed to have a negligible contribution to the temperature of the gas and is used only for destroying the H2 molecules. Their model provides insight into the spatial distribution of H2. The third attempt by Hirashita & Ferrara (2005) uses sim- ple H2 equilibrium formation models to study H2 in DLAs. Unlike Liszt (2002), no attempt is made to model ionization conditions of the gas and excitations of the fine-structure lines in the later two studies. The main aim of our paper is to use full self-consistent numerical calculations to under- stand (i) physical conditions in DLAs with H2 detections, (ii) the reason for the lack of H2 in most of the DLAs (iii) ∗ absorption frequently seen in DLAs and the origin of C ii (iii) the absence of detectable 21 cm absorption at high red- shifts (i.e z> 2). The availability of good quality observational data al- lows us to estimate the metallicity, depletion, H2 abundance, H2 excitation, and populations of fine-structure levels in C i and C ii (Ledoux et al. 2003; Srianand et al. 2005). One can hope to build more realistic models with all these in hand. This forms the prime motivation of this work. We study the ionization state, chemical history, and temperature of the gas using version 96 of Cloudy, last described by Ferland et al.(1998), and available at http://www.nublado.org. The details of the new H2 model are given in Shaw et al. (2005; hereafter S05). A comparison between predictions of our code and several independent calculations of PDRs can c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 be found at http://hera.ph1.uni-koeln.de/∼roellig. A direct application of a PDR is given in Abel et al. (2004; hereafter A04). The calculations presented here take into account various heating (e.g. dust photo-electric heating cos- mic ray heating etc) and cooling processes (for details see Ferland et al. 1998 and S05). 2.1 The micro-physics of grains : We use the improved grain physics and molecular network as described in van Hoof et al. (2004), Ferland et al. (1994; 2002), and A04. The grain size distribution is resolved, and the formalism described by Weingartner & Draine (2001a; van Hoof et al. 2004) is used to determine the grain charge, temperature, photoelectric heating, and drift velocity self- consistently with the local radiation field and gas tempera- ture. The extinction curves of grains in DLAs are not well known. We use a grain size distribution that fits the extinc- tion curve of the diffuse interstellar medium with RV = 3.1 (Table 1 of Weingartner & Draine (2001b)). We emphasize that the physical treatment of grains, and their effects on the surrounding gas, is fully self-consistent, and does not rely on general fitting formulae, such as those in Bakes & Tielens (1994). The grain charge and temperature are determined by the local gas conditions (mainly the electron density and temperature) and radiation field (including the attenuated incident continuum and emission by the surrounding gas, mainly Lyα ). The result is a grain temperature and charge that depends on grain size and material. The temperature then determines the rate of H2 formation on grain surfaces - we adopt the temperature-material-dependent rates given by Cazaux & Tielens (2002). The charge establishes the floating potential of the grain, which then sets the grain photoelectric heating rate. 2.2 Molecular hydrogen : The detailed treatment of the micro-physics of H2 is de- scribed in S05. Here we will briefly mention some of those processes. Generally, H2 forms via grain catalysis in a dusty cold- neutral gas. We use the total formation rate given by Cazaux & Tielens (2002) along with the size and temperature re- solved grain distribution described in van Hoof et al.(2004). This is an exothermic process and H2 is formed in exited vi- brotational levels, a process referred to as formation pump- ing. The results presented below use the state-specific for- mation distribution function given by Takahashi (2001) and Takahashi & Uehara (2001). By contrast, in an equiparti- tion distribution function 1/3 of the binding energy is sta- tistically distributed as internal excitation (Black & van Dishoeck 1987). Both produce an ortho-para-ratio (OPR) that is nearly 3. H2 is formed by associative detachment from H− in a dust-free gas. This process is important in the clouds with lower dust content considered below. This is also an exother- mic process and we use the state-specific formation distri- bution function given by Launay et al. (1991). H2 is destroyed mainly via the Solomon process, in which the H2 molecule is irradiated by far UV (912A< λ < 1200A) radiation and is excited to higher electronic states. c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 H2 in DLAs 3 Approximately 10 per cent of these electronically excited H2 decay to the ground state continuum and are dissoci- ated. The other 90 per cent populate the higher vibrota- tional levels of the ground electronic state. These cascade to lower vibrotational levels producing infrared emission lines, referred to as Solomon pumping. Formation pumping on grains is only 10 percent as effective as Solomon pumping when Solomon destruction is balanced by grain formation. Thus, the H2 populations are non-thermal if the electronic lines are optically thin and the Solomon process is domi- nant (hereafter referred to as the optically thin case), while the H2 level populations may be in LTE at the local gas kinetic temperature if the electronic lines are optically thick and the Solomon process is slow (hereafter referred to as the optically thick case). Radiative decays between ortho and para states are not possible because of the different nuclear spin. However, ex- +, and interactions on grain change collisions with H, H+, H3 surfaces (below a certain critical temperature) can cause an ortho-para conversion. The column density ratio of J=1 and J=0 traces the kinetic temperature in a collisionally domi- nated gas but may fail to do so in a Solomon-process domi- nated region (Abgrall et al. 1992; Sternberg & Neufeld 1999; Roy, Chengalur & Srianand (2005)). The formation rate on dust has the largest uncertainty among the many processes considered in our calculations. There are significant variations in this rate even in the case of the Galactic ISM (Browning et al. 2003). There are also substantial differences between collisional rates of H2 com- puted by different groups at low temperatures (S05). These uncertainties should be kept in mind while comparing our results with observations. 2.3 H i spin temperature: It is commonly assumed that the N(H i) spin temperature, Ts, is equal to the gas kinetic temperature. The optical depth of the 21 cm transition is proportional to N(H0) / Ts, so is sensitive to the inverse of the gas kinetic temperature. The mean value of Ts we report here is TK with weighting by N(H0) / T. A separate paper (Shaw et al. 2005) discusses our treatment of Ts, and relationships between T, Ts, and H2 temperature indicators, in detail. In DLAs, Ts is usually estimated using the integrated optical depth τv in the 21 cm absorption line and N(H i) measured from Lyα using Ts = N (H i)fc 1.823 × 1018τv . (1) Here, fc is the fraction of the background radio source cov- ered by the absorbing gas. Thus, low 21 cm optical depths could either mean high Ts or low fc. Even for fc = 1, the derived temperature will be high if the mean N(H i) cov- ering the extended radio source is lower than the measured one along the line-of-sight toward the optical point source (Wolfe et al. 2003a). Thus, observations will constrain either the physical conditions or the projected H i surface density distribution of the absorbing gas. 2.4 Fine-structure level population: The ionization potential of C0 is close to the energy of the photons responsible for the H2 electronic band transitions. 4 R. Srianand et al. So, C i lines may be sensitive to the conditions in the H i − H2 transition region. The fine-structure level populations of C i are sensitive to the gas pressure and the IR radiation field. Thus, populations of the excited fine-structure levels of C i allow an independent probe of quantities that control the physical conditions in the H i − H2 transition region and the temperature of the cosmic microwave background (CMBR) (Srianand et al. 2000; Silva & Viegus 2002). The column densities of excited levels within the ground term of C i, C ii , Si ii, and O i are all calculated as part of the gas cooling function. All excitation and deexcitation processes, collisions, line trapping, destruction by background opaci- ties, and pumping by the external continuum, are included. At high z the IR pumping is predominantly by CMBR pump- ing, although the diffuse IR radiation from grains also con- tributes. 2.5 Cloud geometry: We envision the region where the absorption lines form as a layer or cloud of ISM gas exposed to several radiation fields. In keeping with much of the PDR literature, we as- sume a plane parallel geometry (Tielens & Hollenbach 1985; Draine & Bertoldi 1996; Wolfire et al. 1995, 2003; Wolfe et al. 2003a,b). We further assume that the gas has constant density for simplicity. In the absence of ongoing star formation, the meta- galactic background UV radiation field, dominated by QSOs (Haardt & Madau 1996), will determine the physical condi- tions and abundance of H2. If there is ongoing star formation then a locally-produced stellar radiation field will also con- tribute. OB stars are very short lived, and so do not move far from their birthplace before dying. So, newly formed OB stars will be close to their parent molecular cloud through- out most of their lives. This geometry is assumed in the PDR references cited above. Our main goal is to understand the physical conditions in the components with H2 and C i. Only the total H i column density is measurable in DLAs and the H i column density in the H2 component is generally unknown. So, we consider clouds with three values of N(H i); 1019, 1020, and 1021 cm−2. We assume the gas metallicity to be 0.1 Z⊙ and vary the dust-to-metal ratio in the range 0.001 to 0.1 (this corresponds to a range in κ (as defined in Section 1) of 10−4 to 10−2) of the galactic ISM. We consider three ionizing continua; the meta-galactic radiation field at z = 2, the direct radiation field from an O star, and an O star continuum that has been attenuated by intervening absorption. The first mimics the case in which there is no in situ star formation. The second is observed in galactic star-forming regions - the OB stars are close to the molecular cloud and an H II region lies between them. This will be called the stellar case below. The third would be similar to a diffuse ISM exposed to the galactic background starlight, and will be called the diffuse case from now on. Following the general practice in the PDR literature, we define the intensity of the incident UV radiation field using a dimensionless constant χ (as defined by Draine & Bertoldi 1996), R 1110A 912A h−1λuλ dλ 1.22 × 107 χ = (2) Here, λuλ is the energy density (ergs cm−3) of photons and χ = 1 for the Galactic UV field defined by Habing (1968). Thus χ provides the UV field strength in the units of Galactic mean UV field. We use the observed metallicity, depletion, H2 abun- dance, and fine-structure excitations of C i, C ii, and N(C i)/N(C ii) to constrain either the particle density or the intensity of the radiation field. The 21 cm spin-temperature and the level populations of H2 are used for consistency checks. 2.5.1 Ionization and thermal structure : In this sub-section we demonstrate the need for a composite self-consistent simulation of the gas in order to deduce the correct physical conditions using a pedagogical example. In Fig. 2.5 we show the ionization and thermal structure of a cloud irradiated by stellar and diffuse continua. Panel (a) plots the temperature of graphite and silicate grains (for the range of sizes considered in our calculations) as a function of depth from the illuminated side. All the calculations we present in this work use the self-consistently estimated grain temperature for a range of grain sizes which are important for different processes such as photoelectric heating and formation of H2 on the grain surfaces. The kinetic (TK ) and the electron density (ne) are plot- ted in panel (b). In this work we present the H i weighted harmonic mean kinetic temperature as spin temperature (TS). Detail investigations of relation- ship between TS and TK under different physical conditions is described in Shaw et al.(2005). Panel (c) plots the densities of H2, C0, C+, and C++ as a function of cloud depth. The ratio of carbon fine-structure levels are shown in panel (d). The electron temperature is high (∼104 K) and the electron density is nearly equal to the H+ density at the illuminated side of the gas for the stellar continua (panel b, solid line). A hydrogen ionization front occurs at a depth of 2.5×1018 cm, where T and ne fall. Across the PDR, T ranges between 300 − 800K and electrons are mainly donated by C+. The short-dashed lines show the results for the diffuse case. There is no H ii region, and so the entire cloud is a PDR. The physical conditions are nearly constant across the cloud, which does not have enough grain opacity to attenuate the incident continuum significantly. The behavior of C0 in the case of the stellar case is as follows. In photoionization equilibrium, n(C0) is ∝ ne n(C+) αrec ∝ ne n(C+) T −0.6 where αrec is the recombination co- efficient. Here ne decreases by three dex across the ionization front, however the electron temperature changes by less than two dex. This leads to two orders of magnitude decrease in n(C0) across the ionization front (see upper panel). As the ionization potential of C0 is very close to the energy of photons that are responsible for the excitations of electronic states in H2, one expects both C i and H2 to originate from the same part of the cloud. This happens for the diffuse case. But in the case of the stellar case, a considerable fraction of C i originates in warmer gas that does not possess H2. The predicted ratio of fine-structure level populations depends on the nature of the radiation field. The ratios are constant across the cloud in the case of the diffuse case. c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 H2 in DLAs 5 Figure 1. Pedagogical example: Various physical quantities are shown as a function of depth in a cloud irradiated by the stellar and diffuse continua with log N(H i) = 20.7. The metallicity is 0.1 Z⊙, nH = 50 cm−3, and log κ is −1.39 (corresponds to a dust to metal ratio 0.4 times that seen in the Galactic ISM). In these panels the short-dashed lines are for the diffuse case. In panel (a) thick and thin curves are for Silicate and Graphite grains respectively. The short-dashed curves are for models with diffuse continua. The assumed radiation field is more or less 4 times that of Galactic mean UV field. In panel (b) the thick curves represent electron density. Panels (c) and (d) show the ionization and fine-structure excitations as a function of depth in the cloud. However, they strongly depend on the radiation field for the stellar case. UV field (i.e χ = 1.44 × 10−2 for the Bgr at z = 2). Some of the results from our calculations are presented in Figs. 2 and 3. 3 IONIZATION BY THE META-GALACTIC UV BACKGROUND: First we consider the case in which the only available ra- diation field is the meta-galactic UV background. We use the QSO dominated meta-galactic UV radiation field (Bgr) computed by Haardt & Madau (1996) and the cosmic mi- crowave background radiation (CMBR) at z = 2 (assumed to be a black body with T = 8.1 K). The UV flux density in the Bgr at energies below 1 Ryd at z = 2 is roughly two orders of magnitude lower than the current mean Galactic 3.1 Gas pressure: Panel (a) of Fig. 2 plots the mean pressure of the H i gas as a function of the total hydrogen density (nH ) for three different values of N(H i). Although ionization and thermal gradients exist between the H ii and H i regions, the H i gas is fairly isothermal. Thus, we use the neutral hydro- gen weighted mean temperature to estimate the pressure. The continuous, short-dashed and long-dashed curves in all these panels are for dust to metallicity 0.1, 0.01, and 0.001 (i.e κ = 0.01, 0.001 and 0.0001) respectively . The regions of c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 6 R. Srianand et al. Figure 2. The results for various constant density clouds ionized by the meta-galactic UV background given by Haardt & Madau (1996). The results with continuous, short-dashed and long-dashed are for κ = 0.01, 0.001 and 0.0001 respectively. Panel (a) plots the mean gas pressure as a function of hydrogen density. The density-ranges of the cold and warm neutral medium are marked in this panel. The numbers near each line give the assumed log N(H i). This panel is useful to identify the warm and cold components of the stable two-phase medium under pressure equilibrium. The shaded histogram gives the observed distribution in the systems with H2 detections. The non-shaded histogram in panel (d) gives the results for the systems without H2. In panel (f) the non-shaded histogram provides the distribution of upper limits. The observational data used are mainly from Ledoux et al. (2003) and Srianand et al. (2005). The horizontal short-dashed lines in panels (b), (e) and (f) are typical detection limits obtained in echelle spectra. The horizontal short-dashed line in panel (c) gives the expected value of the ratio N(C i∗)/N(C i) when CMBR at z = 2 is the only source of excitation. thermal stability occur for d(logP )/d(logn)> 0. The warm neutral medium (WNM) and cold neutral medium (CNM) of the two-phase medium are shown for reference. As an example, gas with 0.3 6 nH (cm−3) 6 1 is thermally unsta- ble for N(H i)≃ 1021 cm−2. The gas will be in the stable WNM phase for 0.036 nH (cm−3) 6 0.3 and in the stable CNM phase for 16 nH (cm−3) 6 30. For reference, Fig. 6 of Wolfire et al. (1995) shows a similar phase diagram for various metallicities and dust content. The allowed mini- mum and maximum pressure in the two-phase medium are higher in our case than in the typical galactic ISM for a given column density, mainly because of the low metallicity and low dust-to-gas ratio (see also Petitjean et al. (1992), Lizst (2002), Wolfe et al. (2003a,b); Wolfe et al. (2004)). The main motivation for plotting the phase diagram from our calcu- lations is to have a rough idea of the nature of the gas at different densities and to compare our work with published models. It is worth keeping in mind the fact that ISM is more complex than different phases in pressure equilibrium with one another. For example magnetic field, if present in DLAs, can provide confinement even if there is no thermal pressure equilibrium between different phases. Thus, we do not make any serious attempt to model the DLAs as two- phase systems. c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 3.2 H2 abundance: In this section, we compare the predicted and measured N(H2) values (Ledoux et al. 2003) to determine the physical conditions in clouds both with and without observed H2. 3.2.1 Systems without H2 detections: First we consider the cases where H2 is not detected. The predicted H2 column densities as a function of hydrogen density (nH ) are shown in panel (b) in Fig. 2. The horizontal short-dashed line gives the typical detection limit achieved in H2 surveys (i.e., N(H2)= 1014 cm−2). The observed N(H2) is distributed uniformly between 1014 and 1019 cm−2 (The histogram in the left hand side in panel (b)). From this figure it is clear that for a given N(H i), the column density of H2 is independent of κ when the density is low and the gas is mainly the WNM. This is mainly be- cause in the low-density, high-temperature gas, H2 is formed predominantly through the H− process due to the low dust- to-gas ratio. It is also clear that the hydrogen density has to be higher than 0.1, 1.0, 30 cm−3 for N(H i) = 1021, 1020, and 1019 cm−2 respectively in order to detect H2. In the pres- ence of an additional local radiation field (perhaps generated due to in situ star formation) these critical densities will be larger. Thus, if the gas in DLAs is mainly a stable WNM in ionization equilibrium with the Bgr, then the equilibrium H2 column density will be below the detection limit. This in- ference is independent of κ since the H− process dominates the H2 formation at low densities. Fig. 2 suggests that if DLAs have a thermally stable CNM then the equilibrium abundance of H2 is high enough for the molecule to be easily detectable whenever κ is greater than 0.0001 or dust to metal ratio grater than 0.001 (The long-dashed curves in panel (b)). The H2 formation time- scale, which can be long, does not affect this result. A typ- ical time-scale for forming H2 with molecular fraction fH2 is ∼fH2 /2Rn(H0). Here n(H0) denotes the atomic hydrogen density. According to Jura (1975) R≃ 3 × 10−17 cm+3 s−1 in the Galactic interstellar medium. Scaling this value by κ we have, t = 5.025 × 108 fH2 κ nH yrs. (3) Assuming that all the hydrogen is H0, we find a typical H2 formation time-scale of ∼ 5 × 105 yrs with κ = 0.0001 and fH2 = 10−6 for nH =10 cm−3. The age of the cloud has to be less than 105 yrs for us not to detect H2 with N(H2)6 1014 cm−2 and κ = 0.0001. The hydrodynamical time-scales or pressure readjustment time-scales in the cold gas are usually larger than this value (Hirashita et al. 2003) due to the low sound speeds. Hence the typical age of the clouds is expected to be larger. Thus H2 should be detectable in a CNM with κ more than 0.0001 in the absence of any additional local radiation field. 3.2.2 Systems with H2 detections: Now we focus on the systems with detectable H2. Ledoux et al. (2003) have shown that these systems usually have a high metallicity and dust content, i.e; Z> 0.1Z⊙ and log κ > −2. If the gas originates from a stable CNM, our calculations c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 H2 in DLAs 7 predict N (H2) > 1019 cm−2 (Panel (b) of Fig. 2) for N(H i) > 1020 cm−2. The observed N(H2) is always smaller. Inter- estingly, the observed N(H2) values with κ > 0.01 are only reproduced in a very narrow density range, one that is usu- ally thermally unstable in the standard two-phase models. This signifies that, for a uniformly distributed range of den- sities, a cloud with a random choice of nH , with N(H i) in the range of 1020 to 1021 cm−2, and with κ > 0.01, will have either no detectable H2 or N(H2) > 1019 cm−2. The probability of detecting H2 in the observed column density range of 1016 to 1019 cm−2 is very low. Thus, in order to un- derstand the relatively low column densities of H2 in DLAs with clear detections, we need either the presence of an ad- ditional radiation field or for the cloud to be too young to produce H2. The existence of a local radiation field with in- tensity of the order or higher then the Galactic mean field has been suggested by various authors while discussing H2 in individual DLAs (Black et al. 1987; Ge & Bechtold 1997; Srianand & Petitjean 1998; Petitjean et al. (2000;2002); Ge, Bechtold & Kulkarni 2001; Ledoux et al. 2002b, Levshakov et al. 2002; Reimers et al. 2003). 3.3 Fine-structure excitations of C i and C ii: In this section, we compare the column densities predicted for C i and C ii fine-structure excitations with the obser- vations. Many of these results will not agree leading us to conclude that an additional source of radiation must exist in these systems. The time-scales to establish the ionization and populations within the fine-structure levels of neutral atoms are faster than the H2 formation time-scale, so this provides an indicator which should be in steady state. 3.3.1 C i absorption: detectability and degree of ionization The N(C i)/N(C ii) ratio is a good tracer of the flux of photons driving the Solomon process in the cold neutral gas where H2 forms since the ionization potential of C0 overlaps with the H2 electronic bands. However, N(C ii) cannot be accurately determined in DLAs since the C iiλ1334 line is usually saturated. Unlike C ii, the column densities of Si ii and S ii (which usually trace the same region as C ii (see Fig. 2.5)) are accurately measured using transitions with low oscillator strengths. Si and S are usually not highly de- pleted in DLAs (but see Petitjean et al. 2002 for a unique counter example). N(Si ii) can be used as a proxy to esti- mate N(C ii) (Srianand et al. 2000) by assuming the solar abundance of Si/C. Now we compare the predicted N(C i) and N(C i)/N(Si ii) with the observations. We plot N(C i) and N(C i)/N(Si ii) as a function of nH respectively in pan- els (e) and (f) of Fig. 2. The predicted values of N(C i) are much closer to the detection limit in a very low-density gas (nH 6 0.01 cm−3). As previously noted, our calculations are performed with Z = 0.1Z⊙, and the systems that do not show C i absorption tend to have metallicity in the range of 0.01Z⊙ 6 Z 6 0.1Z⊙ (Srianand et al. 2005). Thus, the ab- sence of C i is consistent with a cloud having nH 6 0.01 cm−3, for the metallicity and N(H i) typically measured in these systems. Such a cloud will have log N(C i)/N(Si ii) 6 −1.5 (Panel (f) of Fig. 2). This is consistent with the measured upper limits, shown as the non-shaded histogram 8 R. Srianand et al. in panel (f), in most of the systems without detected H2. But N(C i) is more than an order of magnitude larger than the typical detection limit for nH > 1 cm−3. Thus, C i should be detectable in a high-density gas, even for low metallicity, in the absence of any internal radiation field. The predicted values of N(C i)/N(Si ii) in the high-density gas is usually higher than the observed upper limits (see Liszt 2002 and Wolfe et al. 2003a). Next we concentrate on systems with detectable C i absorption. As noted in section 1, apart from only one case (zabs= 2.139 toward Tol 1037−270), all the C i detections are from DLAs that also have H2. These systems have metal- licities higher than we assume. Based on the detection of H2, we expect these systems to also have a higher density. Dense clouds (nH > 0.1 cm−3) produce a higher value of N(C i) than observed. The measured ratio of N(C i)/N(Si ii) in these components is much less than our predictions for a high-density gas. Clearly higher radiation field is needed to produce N(C i)/N(Si ii) as measured in these systems. The fine-structure populations of C i and C ii can also test the high density requirement for components with H2 detections. This will be discussed in the next subsection. 3.3.2 C i fine-structure excitation: ∗)/N(C i) ratio to constrain Here we use the observed N(C i nH in the components with H2. This ratio is regularly used to trace the pressure in a neutral gas (Jenkins & Tripp 2001; ∗)/N(C i) as a function Srianand et al. 2005). We plot N(C i of hydrogen density in panel (c) of Fig. 2. The dotted line gives the expected value of the ratio if CMBR pumping at z = 2 is the only source of C i fine-structure excitation. The ∗)/N(C i) (the histogram in the left observed values of N(C i hand side) are much higher than this, suggesting that col- lisions and UV pumping also contribute to the excitation. Most of the observed ratios are consistent with the predic- tions for nH > 10 cm−3 for the considered range of N(H i) and κ. As noted above, for such a high-density gas, our cal- culations predict N(H2) and N(C i)/N(Si ii) higher than observed. We show below that the presence of an additional radiation field can reduce both of these. 3.3.3 Excitations of C ii fine-structure level: ∗ is always detected whenever H2 is present Like C i, C ii in DLAs. However, it is also seen in a considerable frac- tion of DLAs without C i and H2 (Wolfe et al. 2003a; Sri- ∗ anand et al. 2005). The observed column density of C ii can directly give cooling rate (when the optical depth of [C ii]λ158 line is negligible) which can be used to constrain the star-formation rate once the physical conditions in th gas is known (Wolfe et al. 2003a; 2003b; 2004). C ii can originate from the CNM, WNM, and an ionized gas (i.e., H ii regions). ∗, while Collisions with atoms are important for exciting C i ∗. electron collisions are important for the excitation of C ii ∗ is expected to be detectable in systems with a Thus, C ii warm and/or ionized gas even if the high-density CNM is absent (see Lehner et al (2004) and Srianand et al. (2005)). ∗ is invariably detected in all the systems with log N(H i)> 21 and log Z> −0.03Z⊙ irrespective of the absorp- tion redshift. The observed column density is in the range C ii Figure 3. The calculated H i spin temperature is shown as a function of density with the meta-galactic UV background dom- inated by the QSOs at z ≃ 2. The results are presented for three different column densities of N(H i)(1021, 1019, 1020, 1021 cm−2) and three different values of κ (continuous, short-dashed and long- dashed curves are for κ = 0.01, 0.001, and 0.0001 respectively). ∗)614.0 for systems at 1.56zabs62.5. The 12.76log N(C ii typical upper limit is N(C ii) 6 1013 cm−2 (Srianand et ∗)/N(Si ii) ratio is shown al. 2005). The calculated N(C ii in panels (d) of Fig. 2. The shaded and non-shaded ob- served histograms in these panels are for the systems with and without H2 detection respectively. The observed range ∗)/N(Si ii) in systems with H2 detections suggests of N(C ii they originate in a high-density gas. This is consistent with our conclusion based on H2 and N(C i ∗)/N(C i). The observed N(C ii ∗)/N(Si ii) ratio tends to be smaller in systems without H2. As we have mentioned above, most of these systems have total N(H i) higher than 1021 cm−2. ∗ will originate in clouds Thus, the systems that show C ii with nH >0.1 cm−3. Thus, we will require a radiation field in excess of the Bgr to suppress C i and H2 in a high-density gas. 3.4 H i spin temperature: The thermal state of H i gas can be probed with the 21 cm spin-temperature (Ts). The harmonic weighted mean tem- perature, a proxy for Ts, is shown in panel (a) of Fig. 3. It is clear that if DLAs originate in a low-density WNM gas, then the spin temperature will be ≃ 8000 K. Thus, systems with ∗, and 21 cm absorption are consistent with no H2, C i, C ii a low-density WNM in radiative equilibrium with the Bgr. The predicted spin temperatures are usually less than 100 K for clouds with N(H i)> 1020 cm−2 in the CNM. Thus, we expect all DLAs to show detectable 21 cm absorption if they originate in a high-density CNM gas which covers the background radio source. Unlike C i or H2, the presence of an additional radiation field with hν 6 13.6 eV may not c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 H2 in DLAs 9 cm−2. We would need b ≫ 3 km s−1 to produce a significant effect on N(H2). ∗)/N(C i) when N(C i [C i] 610µ is optically thick at high column densities and the fine-structure level populations are influenced by line trapping. Turbulence increases the photon mean free path and reduces this trapping. The effect is clearly seen for the ∗)> 3 × 1012 cm−2, which ratio N(C i occurs when nH =10 cm−3 (Panel (c) of Fig. 4) for N(H i) = 1021cm−2. The effect is not seen significantly in the case of N(H i) = 1020 cm−2 since large column densities of N(C i ∗) only occur for densities above 100 cm−3. As pointed out be- fore, our calculations produce higher N(C i) than observed. N(C i) is invariably not saturated in most DLAs and λ610µ ∗)/N(C i) ratio. line trapping should not control the N(C i Thus the inclusion of micro-turbulence can not rectify the problems of the Bgr models in simultaneously reproducing the H2 and C i observations. The column density of C ii ∗ is also affected by the pres- ∗) > 3 × 1013 cm−2 due ence of turbulent motions for N(C ii ∗) to optical depth in the [C ii] 158µ line. The observed N(C ii is always higher than 3×1013 cm−2 (Table 1 of Wolfe et al. 2003a; Srianand et al. 2005) in DLAs with N(H i)> 1021 cm−2. Thus, line trapping effects may be important in pro- ducing the observed excitation of the C ii fine-structure level. Cosmic rays add heat to a highly ionized gas and pro- duce secondary ionizations in a neutral gas. H+ produced by cosmic ray ionization can cause ortho-para conversion and thermalize this ratio (Flower et al. 1994). We consider a cosmic ray ionization rate equal to the Galactic background ionization rate (∼ 2.5 × 10−17 s−1; Williams et al. 1998). These results are presented with long-dashed lines in Fig 4. The enhancement in the gas temperature increases the col- ∗ in the high-density gas. umn densities of C i The pressure of the neutral gas increases due to cosmic ray heating as expected. We also see a decrease in N(H2) at a given nH mainly because the increase in the gas temperature reduces the efficiencies in forming H2. ∗ and C ii Background cosmic ray ionization does not produce drastic changes for the low-density gas where the Bgr dom- inates (Fig. 4). A much larger cosmic ray ionization rate is needed to have an desired effect. 3.6 Summary: The main results for a cloud irradiated by the meta-galactic UV background radiation are: • The presence of the QSO dominated meta-galactic ra- diation field can maintain a H2 abundance lower than the detection threshold for nH 6 0.1 cm−3, irrespective of the dust content and N(H i). The presence of any extra radi- ation field in addition to the meta-galactic radiation field, or a slower H2 grain formation rate, increases this critical density. Thus the absence of H2 in 85 per cent of DLAs is consistent with the low density models. • The detection of C i absorption is inevitable whenever our line-of-sight passes through the CNM. However, the den- ∗)/N(C i) ra- sity range that produces the observed N(C i tio also produces N(C i) and N(H2) higher than the ob- served values. An additional source of radiation with ener- Figure 4. The effects of micro-turbulence & cosmic ray ioniza- tion. The solid curves give the result without turbulence and cos- mic ray ionization when Z = 0.1 Z⊙ and κ = 0.01. The short- dashed curves show the effects of 3 km s−1 turbulence. The long- dashed curves shows the effect of cosmic ray ionization (with a cosmic ray ionization rate of hydrogen 2.5 ×10−17 s−1). The re- sults are presented for two values of N(H i)(1020, 1021 cm−2). reduce the 21 cm optical depth because it can not ionize hydrogen. The absence of 21 cm absorption in most DLAs suggests that they originate in a low-density warm medium. ∗ should Our calculations also suggest that H i, H2, and C ii be found in systems with 21 cm absorption. The absence of C i and H2 in the few systems with 21 cm absorption is inconsistent with this prediction, again suggesting a local source of star light. 3.5 Effects of micro-turbulence and cosmic rays We have neglected cosmic rays and turbulent motions in the models considered until now. In this section we investigate whether the inclusion of these processes help bring the Bgr models into agreement with observations. The presence of micro-turbulence increases the mean free path for line photons and reduces line centre optical depth. We consider a turbulent velocity corresponding to a Doppler b parameter of 3 km s−1. This shows the greatest possible effect since the typical measured b parameters of H2 components are usually 6 3 km s−1(Srianand et al. 2005; Ledoux et al. 2003; Petitjean et al. 2002; Srianand et al. 2000). We consider two column densities, N(H i) = 1021 cm−2 and N(H i) = 1020 cm−2, along with κ = 0.01, and Z = 0.1 Z⊙. Some results are presented in Fig. 4. N(H2) changes very little for N(H i) = 1021 cm−2. However, the molecular column density is slightly lower in the case of N(H i) = 1020 c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 10 R. Srianand et al. Figure 5. The effect of density: The results of calculations of a cloud in the radiation field of a star with a surface temperature 40,000 K (stellar case). The cloud has log N(H i) = 20.7, Z = 0.1 Z⊙. The long-dashed, continuous and short-dashed curves are for log nH = 0.0, 1.0, and 1.7 respectively. The labels 1,2,3, and 4 in the short-dashed curves are for log(κ) = −1.4, − 1.6, − 1.8, and −2.0 respectively. We mark these numbers only for the short-dashed curves. The dependences on κ has the same sense for other values of nH as well. The results are presented as a function of χ (see Eq.2). In each panel, the observed distributions are given as histograms (see caption of Fig. 2 for details). gies 6 13.6 eV is needed to account for the low values of N(H2) and N(C i) in these systems. • Like H2, the absence of 21 cm absorption in most of the high-z DLAs can be naturally explained if DLAs originate mostly in the WNM. A low-density gas, corresponding to a WNM, has a very large spin temperate (Ts > 7000 K). If DLAs are dominated by such a gas then 21 cm absorption will not be detectable. • A high-density gas has Ts 6 100 K. This produces strong 21 cm absorption along with high values of N(H2) and N(C i). The fact that N(H2) and N(C i) are not seen in the few systems that do show 21 cm absorption suggests that an additional radiation field is present in these systems as well. This is consistent with results of detail investiga- tions of individual systems available in the literature (see references given in Section 3.3.1). • The too-large N(H2) and N(C i) column densities pre- dicted at higher densities cannot be explained by micro- turbulent motions (up to 3 km s−1) or cosmic ray heating. However, the non-thermal motions affect the fine-structure level populations when the infrared lines become optically ∗) is > 13.5 for N(H i)> 1021 thick. The observed log N(C ii cm−2. The effects of line trapping of [C ii] λ158 becomes very important in these systems whenever the turbulent motions are small. In the following section we explore the possibility of using in situ star formation to prevent the formation of too-large N(H2) in high-density systems. We wish to point-out that in- clusion of radiation from the Lyman Break Galaxies (LBGs) can increase the flux of Lyman Warner band photons in the UV background by upto a factor 10 (see Haardt & Madau (2001)). This will make the Bgr radiation roughly 5 times less than the Galactic mean field. However, Section 4 shows that the required radiation field is much higher than this c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 enhanced Bgr. In addition, this enhanced UV field will also produce bimodel distribution of N(H2) contrary to what has been observed. 4 IONIZATION BY YOUNG STARS: We add a stellar radiation field using a 40,000 K Kurucz' model atmosphere in addition to the metagalactic radiation field discussed above. We consider direct stellar radiation and stellar continuum attenuated by N(H i) = 1020 cm−2. These two cases are denoted by "stellar" and "diffuse", as discussed in Section 2.5. We showed that the meta-galactic UV background is sufficient to suppress the formation of H2 in low-density gas (i.e nH 6 0.1 cm−3). Thus, in this section we mainly concentrate on the high-density gas needed to ∗)/N(C i) ratio. We consider account for the observed N(C i clouds with three values of N(H i) (1020, 1020.7, and 1021 cm−2), three values of nH (1, 10, and 50 cm−3), and four values of κ (10−2, 10−1.8, 10−1.6, and 10−1.4). In all of these calculations we assume Z = 0.1 Z⊙, turbulent velocity b = 3 km s−1. Cosmic-ray heating is not considered in the models described below. We present the results of model calculations as a func- tion of χ (as defined in Eq. 2). The calculations for stellar case will have more high-energy photons than the diffuse one for the same value of χ. Therefore, for a given N(H i), the gas will be hotter and more ionized at the illuminated side of the cloud for the stellar case (see Fig. 2.5). We present the results of our calculations for the stellar and diffuse case for N(H i) = 5 ×1020 cm−2 in Figs. 5 and 6 respectively. The effects of changing N(H i) are shown in Figs. 7 and 8. 4.1 H2 abundance: Panel (a) of Fig. 5 shows the predicted column density of H2 as a function of χ for the stellar case, log N(H i)=20.7, and 3 values of nH (long-dashed, continuous and short-dashed curves are for nH = 1, 10 and 50 cm−3 respectively). For each values of nH the results are presented for 4 different κ. For a given nH and χ, a higher dust content κ produces a lower N(H2). Naively we would expect N(H2) to increase with in- creasing κ due to an enhanced probability of H striking a grain, and increased shielding. However, the gas temperature increases for higher κ due to additional grain photo-electric heating (see the next section). This reduces the H2 forma- tion rate as shown in Fig. 1 of Cazaux and Tielens (2002), so, N(H2) decreases. The calculations reproduce the observed range of N(H2) for 16 χ 6100, nH ∼ 10−50 cm−3, and log N(H i) = 20.7 for the stellar case. The range becomes 16 χ 6300 for the diffuse case (Panel (a) of Fig. 6). We find that with nH =1 cm−3, H2 should be detectable when χ 6 3 and χ 6 10 for the stellar and diffuse case respectively. Clearly, for a mod- erate local radiation field, χ ≃ 1 − 10, H2 will be detectable for nH >1 cm−3 and N(H i) > 5 × 1020 cm−2. Higher nH is needed to produce detectable N(H2) for a lower N(H i) (Pan- els (a) in Figs. 7 and 8). The range of χ that is consistent with the observed range of H2 is summarised in Table 1 for all the scenarios discussed in this work. Observations of atomic fine-structure lines will further narrow down this range. c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 H2 in DLAs 11 4.2 Spin temperature and 21 cm optical depth: Panels (b) of Figs. 5 and 6 show the predicted Ts for log N(H i) = 20.7, 3 values of nH (Long-dashed, continu- ous and short-dashed curves are for nH =1,10 and 50 cm−3 respectively) and 4 values of κ. It is clear from all these panels that for a given nH (curves with similar line style), Ts increases with increasing κ mainly due to photoelectric heating by dust grains (labels 1, 2, 3 and 4 on the short- dashed curve show models with κ in the increasing order). We also notice that for a given κ (say top-most curve with a given line-style) and χ the models with higher nH have lower Ts. This effect is very prominent in the stellar case. This is mainly because Ts is very large at the illuminated side of the gas in the stellar case. Panels (b) of Figs. 7 and 8 show the results with nH = 50 cm−3 for 3 different values of N(H i) (long-dashed, continuous and short-dashed curves are respectively for log N(H i)= 20, 20.7 and 21) for the range of κ. It is clear that for a given κ (say top most curves for differ- ent line-styles), clouds with lower N(H i) will have higher Ts and hence lower τv(21 cm). In the diffuse case (from Figs. 6 and 8) we notice that for a given nH and κ, Ts gradually increases with χ and saturates to a constant for large values of χ. The increase in Ts with lower value of χ is the effect of photo-electric heating. At larger χ the grains become highly charged and total heating rate will become independent of χ (Bakes & Tielens, 1994; Weingartner & Draine (2001a)). Thus Ts becomes independent of χ at large χ. The range of predicted Ts in the range of χ constrained by the H2 observations is summarised in Table. 1. This ta- ble also gives the expected 21 cm optical depth obtained using Eq. 1. It is clear from the table that for a high- density gas (i.e nH >10 cm−3) and log N(H i)> 20.7, 21 cm absorption should be detectable with an optical depth of τv(21 cm)/f > 0.27. Especially for high κ, the 21 cm optical depth becomes as low as 0.05, even when nH =50 cm−3, for log N(H i) = 20. Thus, the absence of 21 cm ab- sorption in high-density (or low temperature) systems will either mean that the average N(H i) along the radio source is much less than N(H i) seen along the optical sight-line (Wolfe et al. 2003a) or that the actual N(H i) in the high- density component is lower (Kanekar & Chengalur, 2003). Absence of H2 in the systems that show 21 cm absorption with low Ts will indicate a radiation field much higher (i.e χ >> 1). Our calculations also suggest that it is possible to detect H2 with low τv(21cm) for moderate nH (see Models A & AE in Table. 1). For example, when log N(H i) = 20.7 and nH ≃ 1 cm−3, the expected spin temperature is high (2300-5600 K) and 0.05 6 τv(21cm)/f 6 0.12. Thus, 21 cm absorption will either be weak or undetected in cases where H2 is detectable. Thus, the fine-structure excitation of C i or C ii with systems with detectable 21 cm absorption will lead to a better understanding of physical conditions in the gas. This is detailed in the following sections. 4.3 C i absorption: detectability and level of ionization: Panels (e) and (f) of Figs. 5 and 6 plot N(C i) and N(C i)/N(Si ii) as a function of χ for stellar and diffuse cases respectively with log N(H i)=20.7. It is clear from the figure that the predictions for different κ (curves with 12 R. Srianand et al. Figure 6. The results of calculations of a cloud in the radiation field of a star with a surface temperature 40,000 K and attenuated by N(H i) = 20 cm−2 (diffuse case). Rest are same as in Fig. 5 Table 1. Results for clouds irradiated by starlight. A-E are the stellar case and AE-EE are the diffuse case Parameters log N(H i) nH (cm−3 ) log N(H i(ext)) χ Ts (K) τv (21 cm)/f log N(C i)/N(Si ii) log N(C i log N(C ii log N(C i(tot)) log N(O i log N(O i log N(Si ii ∗) ∗∗) ∗) ∗)/ N(C i) ∗)/N(Si ii) Models A 20.7 1 .... 6 3 3150,5600 0.09,0.05 −2.6, −2.0 −0.76,0.0 −1.8, −1.3 12.0,13.4 11.2,11.3 11.2,11,3 6 10.2 B 20.7 10 .... 1,30 100,1000 2.74,0.27 C 20.7 50 .... 3,100 56,316 4.90,0.87 −2.6, −2.0 −2.5, −1.8 −0.3,0.3 −0.2,0.4 −0.5, −1.0 −0.2, −0.8 11.9,13.0 11.3,11.9 11.3,11.9 10.5,11.0 11.8,12.8 11.5,12.2 11.6,12.2 10.8,11.7 D 20.0 50 .... 0.5,35 104,1260 0.53,0.04 −1.8, −1.0 −0.1, 0.44 −0.5, 0.0 11.0,13.4 11.0,11.3 11.0,11.3 ∼ 9.5 E 21.0 50 .... 4,400 48,275 11.42,2.00 −3.0, −2.5 0.0, 0.40 −0.9, −0.1 11.8,13.0 10.5,11.7 10.5,11.5 9.6,10.2 AE 20.7 1 20.0 6 10 2290,5248 0.12,0.05 −3.8, −2.5 ∼ −0.76 ∼ −1.9 11.0,13.0 11.0,11.2 11.0,11.3 9.0,10.3 BE CE 20.7 10 20.0 1,100 90,900 3.04,0.30 −4.0, −2.04 −0.6, −0.5 −1.3, −1.1 11.0,13.0 10.0,11.5 11.0,11.53 9.9,10.0 20.7 50. 20.0 3,200 54,339 5.07,0.81 −3.7, −2.0 −0.4, −0.1 −0.9, −0.6 11.0,13.0 10.0,12.0 110.0,11.5 9.0,10.5 DE 20.0 50 20.0 0.3,22 40,560 1.37,0.10 −2.5, −1.0 −0.4, −0.1 −1.5, −0.5 11.2,13.0 10.0,11.5 0.0,11.3 9.0,10.6 EE 21.0 50 20.0 10,475 66,195 0.83,0.28 −3.7, −2.0 −0.26, −0.44 −1.08, −0.66 10.6,12.0 10.5,12.0 10.0,11.7 9.0,10.5 same line-style) are very much identical. Both N(C i) and N(C i)/N(Si ii) are lower than those produced by the Bgr (Fig. 2) due to the presence of additional ionizing photons. A minor reduction in N(C i) and N(C i)/N(Si ii) for a given nH and κ is noted for the diffuse case compared to the stellar case. This can be easily understood using Fig. 2.5. N(Si ii) is nearly identical for both the radiation fields. Whereas N(C i) is less in the diffuse case as ne and T are lower in these mod- els compared to stellar case (see panel (b) of Fig. 2.5). Here we concentrate on systems with H2 detections. The predicted N(C i) is well below the detection limit for log N(H i) = 20.7 and χ > 10 for the range of nH and the two stellar continua considered here. Thus, these models can ex- plain the weak or non-detection of C i absorption in some of DLAs that show strong H2 absorption (zabs= 3.025 toward c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 H2 in DLAs 13 Figure 7. Effects of column density: The results of calculations of a cloud in the radiation field of a star with surface temperature 40,000 K (stellar case). The cloud has nH =50 cm−3, Z = 0.1 Z⊙. The long-dashed, continuous and short-dashed curves are for log N(H i) = 20.0, 20.7 and 21.0 respectively. The labels 1,2,3, and 4 in the short-dashed curves are for log(κ) = −1.4, − 1.6, − 1.8, and −2.0 respectively. We mark these numbers only for the short-dashed curves. The dependences on κ has the same sense for other values of N(H i) as well. The histograms are as explained in Fig. 2. Q 0347−383 with log N(H i)= 20.56; zabs= 2.595 toward Q 0405−443 with log N(H i)= 20.90; and zabs= 2.811 to- ward Q 0528−250 with log N(H i)= 21.10). N(C i)/N(Si ii) is lower than the measured ratio (in DLAs that show both H2 and C i absorption) for the range in χ allowed by H2 for both the diffuse and stellar case, and log N(H i) = 20.7. Based on the trend seen in Figs. 5 and 6 we may need nH > 50 cm−3 and χ 6 10 to explain the observed range in N(C i)/N(Si ii) for H2 components that show detectable C i absorption. However, such a model will over produce N(H2). This inconsistency can be solved by using a lower value of N(H i) and a higher nH (see Panel (f) in Figs. 7 and 8). In these plots long-dashed, continuous and short-dashed curves are the results for nH =50 cm−3 with log N(H i) = 20.0, 20.7, and 21.0 respectively. The total N(H i) measured for zabs= 1.968 toward Q 0013−004 and zabs= 2.087 toward 1444+014 are consistent with log N(H i)620 in the H2 (and C i) com- ponents. The zabs= 1.962 system toward Q 0551-366 that c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 shows three H2 components has a total log N(H i)=20.5. The zabs=1.973 system toward Q 0013-004 has 15 well de- tached C i components with a total N(H i)=20.8. Clearly a low N(H i) is probable in the components that show H2 and C i absorption. A cloud with a moderate radiation field (i.e χ 6 10), log N(H i)=20.0, and nH = 50 cm−3 reproduces the observed range of N(C i)/N(Si ii), N(H2) and N(C i). 4.4 C i fine-structure excitation: In this section we consider the fine-structure excitation of ∗)/N(C i) as C i. Panel (c) of Figs. 5 and 6 shows N(C i a function of χ for log N(H i)=20.7. This ratio should be independent of χ if UV pumping is negligible because Ts depends mainly on density and is roughly independent of χ. This happens for the diffuse case (Panel (c) of 6). Also the effect of κ is clearly evident in this case. However, in the stel- ∗)/N(C i) increases lar case, we find that the predicted N(C i 14 R. Srianand et al. Figure 8. The results of calculations of a cloud in the radiation field of a star with surface temperature 40,000 K attenuated by N(H i) = 20 cm−2, our diffuse case. Rest are same as in Fig. 7 with increasing χ. We also notice that for a given nH and χ, the stellar case (Panel c in Fig. 7) with a lower N(H i) ∗)/N(C i). However, the produces a higher value of N(C i dependence on N(H i) is very weak in clouds irradiated by the diffuse radiation field (see panel c in Fig.8). This implies ∗)/N(C i) will depend only on density that the ratio N(C i ∗)/N(C i) will for the diffuse case. However, the ratio N(C i depend on the strength of the radiation field (also see panel (d) in Fig. 2.5) in the stellar case. Now we focus on systems that show detectable C i and ∗)/N(C i) is more sensi- H2. The predicted value of N(C i tive to nH and weakly depends on N(H i) and χ for the ∗)/N(C i) is consistent with diffuse case. The observed N(C i 106nH 6 100 cm−3. In the stellar case N(C i ∗)/N(C i) de- pends on nH , N(H i) and χ. Clouds with log N(H i) = ∗)/N(C i) for 20.7 reproduce the observed range in N(C i 16nH 650 cm−3 and the values of χ constrained by the H2 observations. As noted before, however, these models fail to reproduce the observed N(C i)/N(Si ii). Thus we require low N(H i) (≃ 1020 cm−3), high nH (≃ 50 cm−3), and low χ (6 10) components in order to be consistent with the observed N(C i) and N(C i)/N(Si ii) ratio. 4.5 C ii fine-structure excitation: ∗ in detail. Panel (d) of Here we discuss the predicted C ii ∗)/N (Si ii) for log N(H i) = 20.7 Figs. 5 and 6 shows N(C ii as a function of χ. The shaded histogram is the observed distribution of the systems with H2 components and the non-shaded histogram represents those systems that do not show detectable H2 and C i. In the stellar case N(C ii ∗)/N (Si ii) is higher for higher χ (see panel (d) in Fig. 5), and for a given χ the excitation is more for lower density (long-dashed, continuous and short- dashed lines are for nH = 1, 10, and 50 cm−3 respectively). The ratio depends only weakly on κ (different curves with different line-style) in the stellar case. All these trends are mainly because, for a fixed N(H i), a considerable fraction of C ii will originate from regions where hydrogen is ionized. The fraction of C ii originating from a hot ionized gas is higher in the case of lower nH and so the ratio is higher. c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 H2 in DLAs 15 Figure 9. The ratio of column densities of H2 in different rotational levels are shown as a function of total H2 column density for the stellar case. The points in the figures give the observed data (Ledoux et al. 2003). In all these calculations we assume the metallicity to be 0.1 Z⊙, log N(H i)=20.7, and log κ is varied between −2.0 and −1.4 (different curves with same line-style). The long-dashed, continuous and short-dashed curves are for nH = 1, 10 and 50 cm−3 respectively. This happens for the lower N(H i) case also (see panel (d) in Fig. 7 where long-dashed, continuous and short-dashed curves are for log N(H i) = 20., 20.7 and 21 respectively). Results for the diffuse case are summarised in panels (d) of Figs. 6,8. It is clear that for a given nH , N(H i), and κ the ratio increases with increasing χ when the χ is small. How- ever, at larger χ the ratio becomes independent of χ. This is due to the grain heating saturation for highly charged grains at high χ (see Weingartner & Draine, 2001a). This is also the reason for the lack of dependence of spin tem- perature on χ (see previous section). Thus, at high χ the ratios mainly depend on nH in the diffuse case. The mod- els presented by Liszt (2002) use the fitting function given by Bakes & Tielens (1994) and show a monotonic increase ∗)/N(C ii) with an increase in χ. Weingartner & in N(C ii Draine (2001a) (see figure 15 in their paper) show that the fitting function given by Bakes & Tielens (1994) over pro- c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 duce the photo-electric heating at high χ and because of this ∗)/N(Si ii) with increasing there will be an increase in N(C ii χ even at large values of χ. As our treatment is very close to that of Weingartner & Draine (2001a) we clearly see the effect of saturation of photo-electric heating by dust grains in our models. First we will concentrate on the systems with H2 detec- tions. In the diffuse case the range in nH that is consistent ∗)/N(C i) also reproduces the observed range in with N(C i ∗)/N(S ii). In the stellar case with log N(H i) = 20.7 N(C ii ∗)/N(Si ii) is consistent the observed distribution of N(C ii with nH in the range of 10-50 cm−3. The models with lower ∗)/N(Si ii) for a given N(H i) tend to produce higher N(C ii nH and κ. A cloud with the low N(H i) and high nH that are required to reproduce N(C i)/N(Si ii) will over produce ∗)/N(Si ii) seems N(C ii ∗)/N(Si ii). Thus the observed N(C ii 16 R. Srianand et al. Figure 10. Same as Fig. 9 but diffuse ionizing radiation is used in these calculations. to favor the diffuse radiation field. This produces a model that reproduces all the other observations. In the diffuse case the ratio N(C ii Now we concentrate on systems without H2 detections ∗ absorption. There are two possibilities for but showing C ii the absence of H2 in these systems: Either (i) the gas has lower density and so is partially ionized with a higher tem- perature or (ii) the gas is at a high density in a strong UV ∗)/N(Si ii) mea- field. sured in systems without H2 are consistent with nH in the range of 1-10 cm−3 (see panel (d) of Figs. 6). Srianand et al. (2005) pointed out that Al iii absorption seen in these sys- tems could be a useful indicator of the ionization of the gas. We will return to this issue while discussing the predicted N(Al iii)/N(Al ii) ratio (see Section 4.7). 4.6 Rotational excitation of H2: Here we focus on the H2 rotational excitation predicted in our calculations. 4.6.1 The ortho-para ratio (OPR): The OPR indicates the kinetic temperature when the H2 electronic bands are optically thick (i.e., log(N(H2))> 16; Tumlinson et al. 2002). Srianand et al. (2005) have shown that the OPR observed in DLAs are higher than those mea- sured in Galactic ISM, LMC, and SMC sight lines. Here, we probe the reason for this difference. Panel (a) of Figs. 9 and 10 plots the OPR as a function of N(H2) for the stel- lar and diffuse cases respectively with log N(H i) = 20.7. For optically thin H2, when the Solomon process dominates excitations of H2, (i.e log N(H2)6 16), the predicted OPR is close to 3 for clouds with log N(H i)=20.7. However, the OPR is greater than 3 for log N(H2) in the range of 16 to 18. At N(H2) > 1018 cm−2 the OPR traces the kinetic tem- perature of the gas. Sternberg & Neufeld (1999) show that the high value of the OPR seen for intermediate N(H2) is because the elec- tronic absorption lines of ortho-H2 become self-shielded at smaller column densities than para-H2. Thus, ortho-H2 ex- c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 H2 in DLAs 17 Figure 11. The ratio of column densities of H2 in different rotational levels are shown as a function of total H2 column density for diffuse case. The points in the figures give the observed data (Ledoux et al. 2003). In all these calculations, we assume the metallicity to be 0.1 Z⊙, nH = 50 cm−3, and log κ is varied between −2.0 and −1.4. The labels 1, 2, 3 and 4 in long-dashed curves are for log(κ) = −1.4, −1.6, −1.8 and −2.0 respectively. The long-dashed, continuous and short-dashed curves are for log N(H i) = 20., 20.7, and 21 respectively. ists while para-H2 is destroyed. Thus, we expect the OPR to be larger in the case of a higher radiation field when 166log N(H2)618. For a given N(H2) (in the intermediate range), the predicted OPR is higher for higher nH in mod- els with log N(H i)=20.7. We also notice that, for a given N(H2) and nH , the models with lower N(H i) produce a lower value of the OPR (panel (a) in Fig. 11). Thus, the observed OPR with 166log N(H2)618 will require a lower N(H i) and higher nH . This is consistent with what we in- ferred based on the N(C i)/N(Si ii) ratio. A detail observa- tion of an individual system confirms that the components with low OPR are consistent with a low value of N(H i) (see Table. 1 of Srianand et al. 2005). As an example, zabs= 1.96822 toward Q 0013-004 has the lowest OPR value mea- sured in DLAs (0.64±0.09) and has log N(H2) = 16.77 and log N(H i)6 19.43. We notice that the kinetic temperature c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 is in the range of 40 − 560 K for the consistent models. This is slightly higher than the kinetic temperature range (60−300K) derived using the OPR and assuming LTE as- sumptions (Srianand et al. 2005). We notice that the OPR does not track the kinetic temperature well in the intermedi- ate N(H2) range. A careful investigation of this is presented elsewhere (Shaw et al. 2005). All of our calculations are in qualitative agreement with the OPR seen for log N(H2)>18 components. 4.6.2 N(J=4)/N(J=2) and the radiation field: Panel (d) of Figs. 9 and 10 plots N(J=4)/N(J=2) as a func- tion of N(H2) for various log N(H i) and κ. The Solomon process controls these populations since the energy separa- tion between these energy levels is far too large and colli- 18 R. Srianand et al. sional excitation is inefficient. This ratio indicates χ when the H2 column density is low (Jura 1975). For a given N(H2) with log N(H2)6 16.0, the N(J=4)/N(J=2) ratio is larger for larger (i) nH , (ii)κ and (iii)N(H i) (panel d in Fig. 11). Apart from the two systems in Ledoux et al. (2003), ab- sorption from the J=4 level of H2 is not detected. These two measurements and the upper limits for the optically thin systems are consistent with a radiation field as high as χ = 30. There is very little difference between the diffuse and stellar continua since the excitation is mainly by elec- tronic line absorption. In the optically thick cases, N(H2) in the J=4 level is populated mainly by formation pumping. No clear trend is present since formation pumping depends on various quantities. 4.6.3 N(J=2)/N(J=0) and N(J=3)/N(J=1): The N(J=2)/N(J=0) ratio is more sensitive to collisional excitation than the population ratios of higher rotational levels. The observed and predicted N(J=2)/N(J=0) and N(J=3)/N(J=1) are plotted as a function of N(H2) in panels (b) and (c) of Figs. 9, 10, and 11 respectively. For the inter- mediate range of N(H2) the models that reproduce the OPR also roughly reproduce these two ratios. However, they do not explain the observed distribution for log N(H2)> 18. It is important to note that the J=2 and J=3 levels are mainly populated by cascades from high J levels following formation and UV pumping. The grain formation distribution func- tion and grain surface interactions can affect the excitation of these high J levels. Although fitting the observed results may shed light on a gas with a different metallicity and dust composition than the Milky Way, such an exercise will divert us from our main theme and is left to future work. 4.7 N(Al iii)/N(Al ii): Here we focus on N(Al iii)/N(Al ii) produced with a stellar radiation field on top of the Bgr(Fig. 12). Al ii is mainly ionized by the high energy photons from the Bgr. In the diffuse case N(Al iii)/N(Al ii) ratio depends more on the density than on χ. This is also the case for stellar case when χ is small. However, the ratio increases with increase in χ for large values of χ (See thin curves in Fig. 12). Higher κ produces higher N(Al iii)/N(Al ii), especially in the case of a high-density gas. The ratio N(Al iii)/N(Al ii) is higher for lower N(H i) for a given κ and nH . Our calculations predict log N(Al iii)/N(Al ii) to be less than −1.5 for the ranges of χ, nH , and κ that reproduce the observed properties of the H2 components. Srianand et al. (2005) have shown that most DLAs with ∗ absorption even when H2 and log N(H i) > 21 show C ii C i are clearly absent. All these systems also show Al iii ab- sorption with log N(Al iii)/N(Al ii) higher than−1.8 (Table 6 of Srianand et al. 2005). Our model calculations produce log N(Al iii)/N(Al ii) higher than −1.6 for log N(H i)> 20.7 when nH 6 10 cm−3. Clearly, the systems with only C ii ∗ ab- sorption without H2 and C i absorption need nH 610 cm−3 in order to have a N(Al iii)/N(Al ii) ratio at the detected level, since log N(H i) >21 in most of these components. This ∗)/N(Si ii) range in nH will also explain the observed N(C ii in these systems (see Section. 4.5). This implies that C ii ∗ Figure 12. The ratio N(Al iii)/N(Al ii) as a function of χ for clouds with log N(H i) = 20.7. The long-dashed, continuous and short-dashed curves are for nH = 1, 10, and 50 cm−3 respectively. As in the other figures, 1, 2, 3, and 4 marks the models with κ −1.4, −1.6,−1.8 and −2.0 respectively. The thin and thick curves represent the results for the stellar and diffuse case respectively. absorption originates in a region of lower density and higher ionization compared to the components that produce H2 and C i absorption. Lehner et al. (2004) show that a significant fraction of C ii∗ absorption detected toward high latitude lines-of-sight in our galaxy originate from warm ionized medium (WIM). Using the profile coincidence between Al ii, Al iii, and pho- toionization models (computed using Cloudy) Wolfe et al. ∗ in DLAs are unlikely to originate (2004) argue that C ii from WIM gas. They argue that a considerable fraction of Al iii can be produced from the gas that is by and large neu- tral (much like the models considered here). For the zabs= ∗ without H2 1.919 system toward Q 2206-19 that show C ii and C i absorption lines, Wolfe et al. (2004) derive a density of 1.6 cm−3. Srianand et al. (2005) discuss the profiles of absorption lines from different ionization states (including 21 cm absorption line) in the case of zabs= 1.944 system to- ward Q 1157+014 and conclude that a considerable fraction ∗ absorption originate from gas at lower densities (i.e of C ii nH ≃ 1 cm−3). All these are consistent with our conclusion that density is lower (i.e nH 6 10 cm−3) in the systems that show C ii∗ absorption without H2 and C i compared to the ones that show H2 absorption (i.e nH = 10-100 cm−3). 4.8 Other fine-structure lines: ∗ and O i We predict the fine-structure level populations of O i and Si ii in addition to C i and C ii. The column densities of both ∗∗ are in the range 2×1011 − 1012 cm−2, for the O i range of χ suggested by the C i and H2 observations (see ∗∗ lines are Table.1). The oscillator strengths of O i low (∼ 4×10−2) and these lines are never detected in DLAs. ∗) < 2×1011cm−2, con- Our calculations also predict N(Si ii ∗ and O i c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 H2 in DLAs 19 ∗ absorption is not detected sistent with the fact that Si ii in DLAs. It is most likely that one may not detect these lines directly even if the absorbing gas has a higher density due to their weakness. However, it may be possible to detect them by co-adding a large number of DLA spectra. Thus, one possible way to confirm the idea that most DLAs (with or without H2) originate in a high-density gas with star for- mation is to detect excited fine-structure lines of O I and Si II by co-adding many spectra as one does for metal lines in the Lyα forest. Galactic ISM) and indicate ongoing star formation in these systems. The mean radiation field is determined by both the SFR and radiative transport. As the mean dust optical depth in DLAs will be smaller than that of Galactic ISM, the typical SFR in DLAs with H2 absorption can not be much larger than that seen in our Galaxy. Even if such a moderate star formation exists in most DLAs, they will still contribute appreciably to the global star formation rate den- sity at higher redshifts (see Wolfe et al. 2003a,b; Srianand et al. 2005; Hirashita & Ferrara 2005). 4.9 Summary The main results for a high-density cloud in a stellar radia- tion field are: ∗)/N(C i) and N(C ii • The observed properties of DLAs with H2 are consis- tently reproduced by models with a local radiation field in addition to the QSO dominated BGR. Most of the ob- servations of the H2 components (such as N(C i)/N(Si ii), ∗)/N(Si ii)) are consistent with N(C i lower N(H i) (i,e log N (H i)≃20 cm−2) and higher densities (10 6 nH (cm−3) 6100). The median N(H i) in DLAs with H2 is ∼ 1020.8 cm−2, so in these systems only a fraction of the total N(H i) originates in regions with H2. The typical kinetic temperature ranges between 40 and 560 K. • We reproduce the observed range of the OPR in DLAs. The systems that are optically thin in the H2 electronic bands have a lower OPR, suggesting log N(H i)≃20, con- sistent with constraints from atomic species. The OPR > 3, seen in some of the components with intermediate H2 elec- tronic line optical depths, are produced by the different level of self-shielding in ortho and para H2. The absence of C i and H2 in J > 4 levels in the case of a few optically thick H2 components are consistent with a higher χ in these clouds. • Our predictions, the measurements, and the upper lim- its on the N(J=4)/N(J=2) ratio in the optically thin H2 components are consistent with a radiation field as high as χ = 30. This is consistent with the limits on the radiation field from the atomic species. The absence of N(J=4) lines in the optically thick H2 components are consistent with a low rate of formation pumping in these systems. • H2 and C i are not detectable if the radiation field is much higher irrespective of the model parameters. However, such clouds will be easily detectable in 21 cm absorption with spin temperature in the range of 100 to 1000 K. These ∗ absorption. However, clouds will also show very strong C ii ∗ will strongly depend on the the column density of C ii amount of ionized gas along the line of sight. Also these systems will show very strong Al iii absorption. 5 DISCUSSION AND CONCLUSIONS: 5.1 Nature of the radiation field: Ledoux et al. (2003) show that detectable H2 absorption (N(H2)> 1014 cm−2) is seen in 15-20 per cent of DLAs. We show that the observed properties of these systems are in- consistent with a gas irradiated by the meta-galactic UV radiation field. Our calculations suggest that these systems originate from a high-density gas (> 10 cm−3) irradiated by a moderate diffuse UV radiation field (1 to 30 times that of c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 5.2 Physical state of the H2 gas: Our calculations with a diffuse radiation field suggest high densities in the H2 gas (i.e 10 6 nH (cm−3) 6 100). The typ- ical temperature of the clouds that are consistent with the observations are in the range 40 to 560 K. Our calculations simultaneously explain H2 abundance and fine-structure ex- citations of atomic species without opting for enhanced H2 formation rate on dust grains as required by analytical mod- els of Hirashita & Ferrara (2005). The inferred range in temperature and densities are consistent with the physi- cal conditions in the CNM gas. We show that if the cloud is irradiated by a diffuse interstellar UV background then ∗)/N(C i) can directly probe the density of the gas. N(C i ∗)/N(Si ii) For radiation field with χ > 10 the ratio N(C ii will also trace the density of the gas as photo-heating satu- rates at higher values of χ. However, if the cloud is close to ∗)/N(C i) depends also on χ. The the ionizing source N(C i ∗)/N(Si ii) in the stellar case with predicted values of N(C ii low N(H i) (as required by other observations) are much higher than the observed values. Thus, our calculations re- quire that the H2 components in DLAs are ionized by a diffuse radiation field. Most of the observations are consis- tently reproduced with N(H i) = 1020 cm−2. This suggests that only a fraction of the total measured N(H i) is present in the H2 components. 5.3 DLAs without H2: The observations show that systems without detectable H2 (i.e ∼ 80 − 85 per cent of the DLAs) do not show C i ab- sorption and also have very small values of κ. Roughly 50 ∗ absorption. Our per cent of the DLAs show detectable C ii ∗ and 21 calculations suggest that the absence of H2, C i, C ii cm absorption in a considerable fraction of DLAs could just be a consequence of a low-gas density in a moderate radia- tion field (irrespective of dust content of the gas). This more or less agrees with the Wolfe et al. (2004)' conclusions that ∗ absorption originate the systems with upper limits on C ii in a warm neutral medium (WNM)(also see Liszt 2002). The observed N(C ii ∗)/N(Si ii) in systems that show ∗ absorption without H2 and C i is consistent detectable C ii with nH > 0.1 cm−3 (see Section 3.3.1). The absence of C i and H2 in these systems can be explained as a consequence of higher radiation field. Wolfe et al. (2003a; 2003b) in the framework of stable two phase medium argue that most of ∗ absorption should originate from the CNM gas in the C ii order to have reasonable global star-formation rate density. We note nH <10 cm−3 in systems that show C ii ∗ without ∗)/N(Si ii) as well H2 and C i so that the observed N(C ii 20 R. Srianand et al. as ionization state of Al can be consistently reproduced (see Section 4.7). This range is consistent with the one measured by Wolfe et al. (2004) for zabs= 1.919 system toward 2206- 10. Interestingly, the inferred density in these systems is less than that typically required to explain the property of H2 detected components (i.e nH > 10 cm−3). Thus, it appears ∗ seem to originate from that the systems that show only C ii lower density gas compared to the ones that also show H2 and C i absorption. Unlike H2 and C i, an additional radiation field (with hν 6 13.6 eV) can not suppress 21 cm absorption in the high-density gas. Our calculations with nH > 1 cm−3 pre- dict a spin temperature in the range of 100 − 1000 K for a range of κ, χ, and N(H i) typically seen in DLAs. Thus, 21 cm absorption is definitely detectable. Over the redshift range that is similar to the range used by Ledoux et al. (2003) for H2 searches (1.96zabs6 3.5), only two out of 8 DLAs show detectable 21 cm absorption (Kanekar & Chen- galur 2003). The rest of these systems have a lower limit on the spin temperature in the range of 700-9000 K. Both of the ∗ ab- systems with 21 cm absorption also show detectable C ii sorption. Detail investigation of one of these systems (zabs= ∗ originate 1.944 system toward Q 1157+014) shows that C ii not only from the gas responsible for 21 cm absorption but also from other components (Fig. 19 Srianand et al 2005). ∗ traces a wider range of physical conditions. Clearly C ii There are few systems (e.g zabs= 3.387 toward Q 0201+11 and zabs= 3.063 toward Q 0336-01) that show detectable ∗ absorption without 21 cm absorption. The derived up- C ii per limits on spin temperatures in these systems will mean very low CNM fraction along the sight lines if the gas covers the background radio source completely (Kanekar & Chen- galur, 2003). In the absence of VLBI observations, interpre- tation of these system will be very subjective (see Wolfe et al. 2003b for details). A careful analysis of Al iii/Al ii and ∗)/N(Si ii) in individual components is needed to get N(C ii the contribution of ionized gas to the excitations of C ii ∗. Alternatively, one can use the fine-structure state pop- ulations of O i and Si ii that trace C ii very closely. Our cal- culations also compute expected fine-structure excitations of ∗∗ O i and Si ii. The expected column densities of O i ∗ are in the range 1011 − 1012 cm−2 for the range of and Si ii parameters considered in our calculations. It may be possi- ble to detect these lines using pixel optical depth techniques that are used to detect metals in the diffuse low density IGM. Detection of such lines will put stringent constraints on density in these systems. ∗, O i 5.4 Conclusions: In this article, we present calculations that self-consistently determine the gas ionization, level populations (atomic fine- structure levels and rotational levels of H2), grain physics, and chemistry. We show that for a low-density gas (nH 6 0.1 cm−3) the meta-galactic UV background due to quasars is sufficient to maintain H2 column densities below the detec- tion limit (i.e N(H2)6 1014 cm−2) irrespective of the metal- licity and dust content in the gas. Such a gas will have a 21 cm spin temperature in excess of 7000 K and very low C i ∗ column densities for H i column densities typically and C ii observed in DLAs. Calculations with a high-density gas in the presence of a local radiation field reproduce most of the observations of H2 components in DLAs. Thus our study clearly confirms the presence of CNM at least in 15-20% of the DLAs. We also show only fraction of total N(H i) is in the H2 components. Unlike the components with H2, interpretation of sys- ∗ without additional constraints is tems that show only C ii not clear. This is because presence of free electrons can be more efficient in populating the fine-structure level of C ii. This can lead to a high value of inferred nH if the electron contribution is neglected. Using Al iii absorption we show ∗ in systems without H2 has that a gas that produces C ii lower density than the ones with H2 absorption. 6 ACKNOWLEDGEMENTS GJF acknowledges the support of the NSF through AST 03-07720 and NASA with grant NAG5-65692. GJF and RS acknowledge the support from the DST/INT/US(NSF-RP0- 115)/2002. GS would like to thank CCS, University of Ken- tucky for their two years of support. The hospitality of IU- CAA is gratefully acknowledged. RS and PPJ gratefully ac- knowledge support from the Indo-French Centre for the Pro- motion of Advanced Research (Centre Franco-Indien pour la Promotion de la Recherche Avanc´ee) under contract No. 3004-3. REFERENCES Abel N. P., Brogan C. L., Ferland G. J., O'Dell C. R., Shaw G., Troland T. H., 2004, ApJ, 609, 247 (A04) Abgrall, H., Le Bourlot, J., Pineau Des Forets, G., Roueff, E, Flower, D. R., & Heck, L. 1992, A&A, 253 525 Bakes, E. L. O., Tielens, A. G. G. M. 1994, ApJ, 427, 822 Black, J. H., Chaffee, F. H., Foltz, C. B. 1987, ApJ, 317, 442 Black, J. H., van Dishoeck, E, F., 1987, ApJ, 322, 412 Briggs F. H., Wolfe A. M., Liszt H. S., Davis M. M., Turner K. L., 1989, 341, 650 Browning M. K., Tumlinson J., Shull J. M., 2003, ApJ, 582, 810 Bunker A. J., Warren S. J., Clements D.L., Williger G.M., Hewet P. C., 1999, MNRAS, 309, 875 Calura F., Matteucci F., Vladilo G. 2003, MNRAS, 340, 59 Carilli C.L., Lane W., de Bruyn A.G., Braun R., Miley G.K., 1996, AJ 111, 1830 Cazaux S., Tielens A.G.G.M., 2002, ApJ, 575, L29 Curran, S. J., Murphy, M. T., Pihlstrom, Y. M., Webb, J. K., Purcel, C. R. 2005, MNRAS, 356, 1509 Draine B. T., Bertoldi F. 1996, ApJ, 468, 269 Ferland G. J., Fabian A. C., Johnstone R.M. 1994, MNRAS, 266, 399 Ferland, G. J.; Korista, K. T.; Verner, D. A.; Ferguson, J. W.; Kingdon, J. B.; & Verner, E. M. 1998, PASP, 110, 761 Ferland G. J., Fabian A. C., Johnstone R.M. 2002, MNRAS, 333, 876 Flower, D.R., Watt, G. D. 1984, MNRAS, 209,25 Fynbo J. U., Moller P., Warren S. J., 1999, MNRAS, 305, 849 Gardner J. P., Katz N., Hernquist L., Weinberg D. H., 1997, ApJ, 484, 31 Ge, J., Bechtold, J. 1997, ApJ, 477, 73. Ge J., Bechtold J., 1999 in Highly redshifted radio lines, Carilli, C. L., Redford S. J. E., Menten K. M., Longton G. I., eds. , ASP conf. Ser 156, 121. Ge, J., Bechtold, J., Kulkarni, V. P., 2001, ApJ, 547, 1 Haardt F., Madau P., 1996., ApJ, 461, 20 Haardt F., Madau P., 2001, /astro-ph/0106018 c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000 H2 in DLAs 21 Wolfe A. M., 1995. in QSO Absorption Lines, Proc. ESO Work- shop, ed. G. Meylan (Berlin:Springer), 13 Wolfe A. M., Briggs F. H., Jauncey D. L., 1981, ApJ, 248, 460 Wolfe A, M. ,Prochaska J. X., 2000, ApJ, 545, 603 Wolfe A, M., Prochaska J. X., Gawiser E., 2003a, ApJ, 593,215 Wolfe, A, M. & Gawiser, E., Prochaska, J, X., 2003b,ApJ, 593,235 Wolfe, A. M., Howk, J. C., Gawiser, E., Prochaska, J. X., Lopez, S. 2004, ApJ, 615, 525 Wolfire, M. G., Hollenbach, D., McKee, C. F., Tielens, A. G. G. M. 1995, ApJ, 443, 152 Wolfire, M. G., McKee, C. F., Hollenbach, D., Tielens, A. G. G. M. 2003, ApJ, 587, Habing H., 1968, BAN, 19, 421 Hirashita H., Ferrara A., Wada K., Richer P., 2003, MNRAS, 341,18 Hirashita H., Ferrara A., 2005, MNRAS, 356, 1529 Hollenbach D., McKee C.F., 1989, ApJ, 342, 306-336 Howk, J. C & Sembach, K. R. 1999, ApJ, 523, 141 Izotov, Y, I., Schaerer, D., Charbonnel, C. 2001, ApJ, 549, 878. Jenkins, E. B., & Tripp, T. M. 2001, ApJ, 137, 297 Jura M., 1975, ApJ, 197, 575J Kanekar, N., Chengalur, J. N., 2003, A&A, 399, 857 Khare, P., Kulkarni, V. P., et al. 2004, ApJ, 616, 86 Kulkarni, V. P., Hill, J. M., Schneider, G., Weymann, R. J., Storrie-Lombardi, L. J., Rieke, M. J., Thompson, R. I., Jan- nuzi, B. T. 2001, ApJ, 551, 37. Launay, J. M., Le Dourneuf, M., Zeippen, C. J. 1991, A&A, 252, 842 Ledoux C., Bergeron J., Petitjean P., 2002a, A&A, 385, 802 Ledoux C., Petitjean P., Srianand R., 2003, MNRAS, 346, 209 Ledoux C., Srianand R., Petitjean P., 2002b, A&A, 392, 781 Lehner, N., Wakker, B., Savage, B. D. 2004, ApJ, 615, 767 Levshakov S. A., Dessauges-Zavadsky M., ´DOdorico S., Molaro P., 2002, ApJ, 565, 696 Liszt H., 2002, A& A., 389, 393. Lu L., Sargent W. L. W., Barlow T. A., Churchill C. W., Vogt S. S., 1996, ApJ, 107, 475 Matteucci F., Molaro P., Vladilo G., 1997, A&A, 321, 45 Moller, P., Warren, S. J., Fall, M., Fynbo, J. U., Jakobsen, P. 2002, ApJ, 574, 51 Moller, P., Fynbo, J. P. U., Fall, M., 2004, A&A, 422, 32 Petitjean P., Bergeron, J., Puget, J. L. 1992, A&A, 265, 375 Petitjean P., Srianand R., Ledoux C., 2000,A&A, 364, 26 Petitjean P., Srianand R., Ledoux C., 2002, MNRAS, 332, 383 Pettini M., Smith L. J., King D. L., Hunstead R. W., 1997, ApJ, 486, 665. Prantzos N., Boissier, S., 2000, MNRAS, 315, 82 Prochaska J., Wolfe A. M., 2002, ApJ, 566, 68 Reimers D., Baade, R., Quast, R., & Levshakov, S.A. 2003, A&A, 410,785 Roy, N., Chengalur, J. N., Srianand, R., 2005, preprint Shaw G., Ferland G. J., Abel N. P., Stancil P. C., van Hoof P. A. M, 2005, ApJ, 624, 794 (S05), /astro-ph/0501485 Shaw G., Ferland G. J., Srianand R., 2005, submitted to ApJ. Silva, A. I., Viegas, S. M., 2002, MNRAS, 329, 135 Srianand R., Petitjean P., 1998, A&A, 335, 33. Srianand R., Petitjean P., 2001, A&A, 373, 816. Srianand R., Petitjean P., Ledoux C., 2000, Nature, 408, 931 Srianand R., Petitjean P., Ledoux C., Ferland G., Shaw G. 2004, (Preprint) Sternberg A., Neufeld D.A., 1999, ApJ, 516, 371 Takahashi J., 2001, ApJ, 561, 254-263 Takahashi J., Uehara H., 2001, ApJ, 561, 843 Tielens A. G. G. M., Hollenbach D., 1985, ApJ, 291, 722 Tumlinson, J., Shull, J. M., Rachford, B. L., et al. 2002, ApJ, 566, 857 van Hoof P.A.M., Weingartner J.C., Martin P.G., Volk K., Fer- land G.J., 2001, in Challenges of Photoionized Plasmas, (G Ferland & D. Savin, eds) ASP Conf Ser 247, 363-378 (astro- ph/0107183) van Hoof P.A.M., Weingartner J.C., Martin P.G., Volk K., Fer- land G.J., 2004, MNRAS, 350, 1330 Vladilo G., Centuri´on M., Bonifacio P., Howk C., 2001, ApJ., 557, 1007 Weatherley, S. J., Warren, S. J., Moller, P., Fall, M., Fynbo, J. U., Crisom, S. M. 2005, /astro-ph/0501422 Weingartner, J.C., & Draine, B.T., 2001a, ApJS, 134, 263 Weingartner, J.C., & Draine, B.T., 2001b, ApJ, 548, 296 Williams J. P., Bergin E. A., Caselli P., Myers P. G., Plume, R. 1998, ApJ, 503, 689 c(cid:13) 0000 RAS, MNRAS 000, 000 -- 000
astro-ph/0504071
1
0504
2005-04-04T14:52:30
On the Comment by Palle (astro-ph/0503562) on the vorticity and shear of the Universe
[ "astro-ph" ]
We show that a recent comment where Palle criticises works regarding CMBR fluctuations for a particular type of Bianchi models, is incorrect. In particular, in contrast to the claims of Palle, we point out that Bianchi models do allow for vorticity.
astro-ph
astro-ph
On the Comment by Palle (astro-ph/0503562) on the vorticity and shear of the Universe Sigbjørn Hervik Dalhousie University, Dept. of Mathematics and Statistics, Halifax, NS, Canada B3H 3J5 [email protected] Abstract We show that a recent comment where Palle criticises works regarding CMBR fluctuations for a particular type of Bianchi models, is incorrect. In particular, in contrast to the claims of Palle, we point out that Bianchi models do allow for vorticity. In a recent work Jaffe et al [1] attempt to fit WMAP data with CMBR calcu- lations for some spatially homogenenous Bianchi models. Soon after this paper appeared, a comment by Palle [2] criticises this work as well as an older paper by Barrow et al [3]. Part of Palle's criticism is based on an erroneous claim regarding the possibility of vorticity for these Bianchi models. For the spatially homogeneous Bianchi models there are two naturally de- fined time-like unit vector fields: (i) the unit vector field, uµ, normal to the group orbits, and (ii) the four-velocity, uµ, of the perfect fluid. The vector field uµ describes the evolution of the spatially homogeneous hypersurfaces and in the study of Bianchi models this vector field is usually taken to be the four- velocity of the fundamental observers [4, 5]. This vector field necessarily has zero vorticity, but in general gives rise to a non-zero shear tensor σab. Regarding µ is not the fluid four-velocity u aligned with uµ we call the model tilted and non-tilted otherwise. It is clear that the fluid can only have vorticity if the fluid is tilted (when we refer to vorticity in a cosmological model we usually refer to the fluid vorticity!). µ, this may or may not be aligned with u µ. If u In [2] Palle erroneously claims that "..it is very well known that standardly defined spacetime vorticity vanishes for these Bianchi models" and then refers to [6] which only considers non-tilted models (hence, zero vorticity by assumption). A more appropriate reference is [7] which considers tilted Bianchi models. This reference also discusses vorticity and gives conditions for when a tilted Bianchi model has vorticity. Clearly, based on these conditions Bianchi models can have vorticity and, in fact, non-zero vorticity is the general case. In particular, Bianchi type VIIh models can have two non-zero vorticity components as pointed out by Barrow et al [3]. Other papers that consider vortical motion in Bianchi cosmologies are, for example, [8, 9]. 1 Palle also states that the vorticity of [3] is "just certain space rotation propor- tional to the shear components". In principle, the vorticity of the fluid and the shear of the hypersurfaces are independent quantities. However, in cosmology we also impose the Einstein field equation which relates the stress energy tensor of the matter to the Ricci curvature of the universe. In the orthonormal frame formalism, the Einstein field equation imposes constaints which relate some of the matter variables with the variables for the geometry [4, 5]. In eqs. (4.3) and (4.4) of ref [3], two of these constraints have been solved to relate the vorticity of the matter to the shear of the spacetime for which it induces. Hence, Palle's "space rotation" is in fact some of the components of the Einstein field equation that has been solved to obtain expressions for the vorticity. Moreover, Palle expresses concern regarding the gauge invariance of some relations regarding the shear. In the background, which both [1] and [3] assume to be FRW, the shear vanishes, and hence, when perturbed, the shear will be gauge invariant [10]. Thus the corresponding expressions for the CMBR will also be gauge invariant. Based on this we can see that the criticism by Palle is based on claims that are erroneous, or at best, misleading. References [1] T.R. Jaffe et al, astro-ph/0503213 [2] D. Palle, astro-ph/0503562 [3] J.D. Barrow, R. Juszkiewicz and D.H. Sonoda, MNRAS 213 (1985) 917 [4] Dynamical Systems in Cosmology, eds: J. Wainwright and G.F.R. Ellis, Cambridge University Press (1997) [5] A.A. Coley, Dynamical Systems and Cosmology, Kluwer, Academic Pub- lishers (2003) [6] G.F.R. Ellis and M.A.H. MacCallum, Comm. Math. Phys. 12 (1969) 108 [7] A.R. King and G.F.R. Ellis, Commun. Math. Phys. 31 (1973) 20 [8] C.B. Collins and S.W. Hawking, Astrophys. J. 180 (1973) 317 [9] V.N. Lukash, Nuovo Cimento B35 (1976) 269 [10] J.M. Stewart and M. Walker, Proc. Roy. Soc. London A 341 (1974) 49 2
astro-ph/0003056
1
0003
2000-03-04T07:21:50
The Nuclear Activity of the Galaxies in the Hickson Compact Groups
[ "astro-ph" ]
In order to investigate the nuclear activity of galaxies residing in compact groups of galaxies, we present results of our optical spectroscopic program made at Okayama Astrophysical Observatory. We have performed optical spectroscopy of 69 galaxies which belong to 31 Hickson Compact Groups (HCGs) of Galaxies. Among them, three galaxies have discordant redshifts. Further, spectral quality is too poor to classify other three galaxies. Therefore, we describe our results for the remaining 63 galaxies. Our main results are summarized below. (1) We have found in our sample; 28 AGN, 16 HII nuclei, and 19 normal galaxies which show no emission line. We used this HCG sample for statistical analyses. (2) Comparing the frequency distributions of activity types between the HCGs and the field galaxies whose data are taken from Ho, Filippenko, & Sargent (382 field galaxies), we find that the frequency of HII nuclei in the HCGs is significantly less than that in the field. However, this difference may be due to selection bias that our HCG sample contains more early-type galaxies than the field, because it is known that HII nuclei are rarer in early-type galaxies than in later ones. (3) Applying correction this morphological bias to the HCG sample, we find that there is no statistically significant difference in the frequency of occurrence of emission-line galaxies between the HCGs and the field. This implies that the dense galaxy environment in the HCGs does not affect triggering both the AGN activity and the nuclear starburst. We discuss some implications on the nuclear activity in the HCG galaxies.
astro-ph
astro-ph
THE NUCLEAR ACTIVITY OF THE GALAXIES IN THE HICKSON COMPACT GROUPS Masashi Shimada, Youichi Ohyama, Shingo Nishiura, Takashi Murayama, & Yoshiaki Taniguchi Astronomical Institute, Graduate School of Science, Tohoku University, Aoba, Sendai 980-8578, Japan Received ; accepted 0 0 0 2 r a M 4 1 v 6 5 0 3 0 0 0 / h p - o r t s a : v i X r a -- 2 -- ABSTRACT In order to investigate the nuclear activity of galaxies residing in compact groups of galaxies, we present results of our optical spectroscopic program made at Okayama Astrophysical Observatory. We have performed optical spectroscopy of 69 galaxies which belong to 31 Hickson Compact Groups (HCGs) of Galaxies. Among them, three galaxies have discordant redshifts. Further, spectral quality is too poor to classify other three galaxies. Therefore, we describe our results for the remaining 63 galaxies. Our main results are summarized below. (1) We have found in our sample; 28 AGN, 16 H II nuclei, and 19 normal galaxies which show no emission line. We used this HCG sample for statistical analyses. (2) Comparing the frequency distributions of activity types between the HCGs and the field galaxies whose data are taken from Ho, Filippenko, & Sargent (382 field galaxies), we find that the frequency of H II nuclei in the HCGs is significantly less than that in the field. However, this difference may be due to selection bias that our HCG sample contains more early-type galaxies than the field, because it is known that H II nuclei are rarer in early-type galaxies than in later ones. (3) Applying correction this morphological bias to the HCG sample, we find that there is no statistically significant difference in the frequency of occurrence of emission-line galaxies between the HCGs and the field. This implies that the dense galaxy environment in the HCGs does not affect triggering both the AGN activity and the nuclear starburst. We discuss some implications on the nuclear activity in the HCG galaxies. Subject headings: galaxies: group of - galaxies: interaction - galaxies: nuclei - galaxies: Seyfert - galaxies: starburst -- 3 -- 1. INTRODUCTION It is known that compact groups of galaxies provide the densest galaxy environment rather than binary galaxies, loose groups of galaxies and clusters of galaxies (Hickson 1982; Hickson et al. 1992). Therefore, frequent galaxy collisions are expected to trigger either some nuclear activity or intense star formation in their member galaxies (Hickson et al. 1989; Zepf, Whitmore, & Levison 1991; Zepf & Whitmore 1991; Zepf 1993; Verdes-Montenegro et al. 1998). Further, compact groups would evolve into other populations in the universe because they would be able to merge into one stellar system within a timescale shorter than the Hubble time (Hickson et al. 1992; Barnes 1989; Weil & Hernquist 1996). Indeed previous studies have shown possible evidence that galaxy collisions may trigger either the nuclear activity or starbursts in the HCGs; e.g., HCG 16 (Ribeiro et al. 1996; de Carbalho & Coziol 1999), HCG 31 (Iglesias-P´aramo & V´ılchez 1997a), HCG 62 (Valluri & Anupama 1996), HCG 90 (Longo et al. 1995), and HCG 95 (Iglesias-P´aramo & V´ılchez 1997b). On the other hand, other statistical studies have shown that there may be no strong evidence for the unusually enhanced activity in the HCGs. Hickson et al. (1989) found that the far-infrared (FIR) emission is enhanced in the HCGs. However, later careful analysis of FIR data of HCGs showed that there is no firm evidence for the enhanced FIR emission in the HCGs (Sulentic & de Mello Rabaca 1993). Radio continuum properties of the HCG galaxies do not show evidence for the enhanced nuclear activity with respect to field spiral galaxies although the radio continuum emission from the nuclear region tends to be stronger than that from field spirals (Menon 1992, 1995). More recently, Coziol et al. (1998) have shown from a spectroscopic survey for 17 HCGs (de Carvalho et al. 1997) that active galactic nuclei (AGN) are preferentially located in the most early-type and luminous members in the HCGs. This result suggests possible relations among activity types, morphologies, and densities of galaxies in HCGs. V´ılchez -- 4 -- & Iglesias-P´aramo (1998a) made an Hα emission imaging survey for a sample of HCGs and found that over 85% of the early-type galaxies in their sample were detected in Hα (V´ılchez & Iglesias-P´aramo 1998b). However, they interpreted that the excess emission in Hα is attributed to photoionization by massive stars rather than AGN. Therefore, it is still uncertain what kind of activity is preferentially induced in the nuclear regions of HCG galaxies. In order to investigate nuclear emission-line activity of HCG galaxies in detail, our attention is again addressed to an investigation on how frequent galaxy collisions are related to the occurrence of both nuclear activity and star-formation activity in HCG galaxies. In this paper, we present results of our optical spectroscopic program for a sample of 69 galaxies belonging to 31 HCGs which are randomly selected in the list of HCG (Hickson 1982). In the original catalog of HCG (Hickson 1982), 100 compact groups with 493 galaxies are entried. However, eight groups are now dropped out from the original sample because they do not have more than two galaxies whose redshifts are accordant (Hickson et al. 1989; Hickson 1993; see also Sulentic 1997). Therefore, our sample is selected from the remaining 92 HCGs. 2. OBSERVATIONS We have performed optical spectroscopy of 69 galaxies in the 31 groups (see Table 1). The spectroscopic observations were made at Okayama Astrophysical Observatory (OAO) 188 cm telescope with the new Cassegrain spectrograph and an SITe 512×512 CCD camera during a period between 1996 February and 1997 January. The slit dimension was 1.8 arcsec (width) × 5 arcmin (length). Two-pixel binning was made along the slit and thus the spatial resolution was 1.75 arcsec per element. The 600 grooves mm−1 grating was used to cover 6300 -- 7050 A region with the spectral resolution of 3.4 A (≃ 157 km s−1 in velocity -- 5 -- at 6500 A). The observations were made under photometric conditions. The typical seeing during the runs was 2 arcsec. The data were analyzed using IRAF1. We also used a special data reduction package, SNGRED (Kosugi et al. 1995), developed for OAO Cassegrain spectrograph data. The reduction was made with a standard procedure; bias subtraction, flat fielding with the data of the dome flats, and cosmic ray removal. Flux calibration was obtained using standard stars available in IRAF. The nuclear spectra were extracted for individual galaxies with a 1.′′8 × 1.′′75 aperture. The extracted nuclear spectra are shown in Figure 1. A journal of the observations is given in Table 1. We also give morphological types of galaxies taken from Hickson (1993; see also Hickson, Kindle, & Huchra 1988; Mendes de Oliveira & Hickson 1994) and de Vaucouleurs et al. (1991) in Table 1. 3. RESULTS 3.1. Classification of Emission-line Activity In usual classification schemes for emission-line galaxies, some combinations of two emission-line intensity ratios (e.g., [O III]λ5007/Hβ versus [N II]λ6583/Hα) are often used (Veilleux & Osterbrock 1987). However, since our spectroscopic program was originally devoted to finding kinematical peculiarity of HCG galaxies (Nishiura et al. 1999), our nuclear spectra cover only a wavelength range between 6300 -- 7050 A. Therefore, emission lines available for the classification of nuclear activities are [O I]λ6300, [N II]λλ6548,6583, Hα, and [S II]λλ6717,6731. Among several combinations between a couple of the emission 1Image Reduction and Analysis Facility (IRAF) 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. -- 6 -- lines listed above, the most reliable indicator to classify nuclear activities seems the [N II]λ6583/Hα ratio (hereafter [N II]/Hα). In fact, Ho, Filippenko, & Sargent (1997) showed from the spectroscopic analysis of more than 300 nearby galaxies that this ratio is useful in distinguishing between AGN and H II nuclei; i.e., [N II]/Hα ≥ 0.6 for AGN while [N II]/Hα < 0.6 for H II nuclei. Therefore, applying this criterion, we classify the emission-line activity of our HCG galaxies. Galaxies without emission are referred as "Abs"; i.e., only stellar absorption features are seen in the optical spectra. For eight galaxies, we detected only [N II] line emission and did not detect Hα line emission (HCG 10a, 30b, 37a, 51b, 62a, 68a, 88a, and 93c). We classify them as AGN. The emission line flux data and the results of the classification are given in Table 2 and Table 3, respectively. . As shown in Figure 1, some nuclei show evidence for the Hα absorption. Since the Hα absorption leads to an underestimation of the Hα emission, it would be better to subtract a template spectrum whose absorption spectral features are nearly the same as those of the concerned spectrum from the target galaxy spectrum (see, for example, Ho et al. 1997). Since, however, we do not have such a template database, we used the observed [N II]/Hα ratios in our classification. In particular, in the case of very weak emission-line galaxies, the Hα emission may not be seen if the Hα absorption feature is strong. The most serious case may be poststarburst galaxies which show very strong Balmer absorption (e.g., Taniguchi et al. 1996 and references therein). Poststarburst galaxies have Hα absorption equivalent widths, EW (Hα) ≥ 3 A. However, the galaxies with Hα absorption in our sample have EW (Hα) ≤ 2 A; i.e., our sample contains no conspicuous poststarburst galaxy. We therefore expect that our emission-line classification is not affected by the effect of Hα absorption seriously. Recently, Coziol et al. (1998) studied the nuclear activity of southern HCG galaxies. They obtained optical spectra of 82 brightest galaxies in a sample of 17 HCGs (de Carvalho -- 7 -- et al. 1997). Among the 82 galaxies, 40 galaxies are original HCG members identified by Hickson (1982). Although their sample is taken from the HCGs located in the southern hemisphere, 13 galaxies in their sample were also observed by us. Since they used the template subtraction method in their classification of nuclear activity, their classification seems to be more reliable than ours. In order to examine how our classification based on the [N II]/Hα ratio without absorption correction is reliable, we compare our results with those of Coziol et al. (1998). The basic data of the 13 HCG galaxies commonly observed by both Coziol et al. (1998) and us are summarized in Table 4. We find that both studies give the same activity types for late-type galaxies. However, for early-type galaxies, though we have classified three galaxies (HCG 40a, 42a, and 87b) as absorption galaxies, they classified them AGNs (dwarf LINERs). These differences appear attributed to that we do not apply the template subtraction method while they did. However, it is noted that all the three galaxies are not typical Seyfert nuclei but dwarf LINER nuclei. Although our analysis may not miss typical Seyfert nuclei, it is safe to mention that about a half (e.g., 3/7 ≃ 43 %) of early type galaxies classified absorption galaxies in our study are AGNs. This point will be taken into account in later discussion. Finally, we classified 63 of 69 galaxies we observed as 28 AGNs, 16 H II nuclei, and 19 no line emissions. Three of the remaining six galaxies are redshift-discordant galaxies (HCG 73a, 87d, and 92a). For the other three, the signal-to-noise ratio of their spectrum are too low to classify (HCG 34a, 42b, and 52a). We exclude these six galaxies from the sample in later statistical analyses. 3.2. Nuclear Activity versus Group Properties Although the selection of HCGs was made homogeneously with the above criteria, it is known that the dynamical properties are different from HCG to HCG (Hickson et al. 1992). -- 8 -- Therefore, it is interesting to compare the nuclear activity in the member galaxies with the dynamical properties of the groups. As we mentioned previously, we adopt the [N II]/Hα intensity ratio as a measure of the nuclear activity. Since it is known that the nuclear activity type depends on the morphological types of host galaxies (e.g., Ho et al. 1997); i.e., AGN favors early-type galaxies while star-formation activity favors later-type ones, it is necessary to investigate relationships between the nuclear activity and the group properties for each morphological type. However, it is generally difficult to classify the morphology of galaxies which are interacting with their partner(s) (Mendes de Oliveira & Hickson 1994). Therefore, although we give detailed morphological types for the member galaxies in our sample in Table 1, we classify them broadly into the following three classes; 1) early-type galaxies (E/S0), 2) early-type spirals (S0a -- Sbc), and 3) late-type spirals (Sc or later). In Figures 2 -- 4, we show diagrams of [N II]/Hα against the number density of the groups ρN (Hickson et al. 1992), the radial velocity dispersion of the groups σr (Hickson et al. 1992), and the crossing time of the groups tc (Hickson et al. 1992), respectively. We adopt the null hypothesis that the [N II]/Hα ratio is correlated with each dynamical parameter and apply the Spearman-rank statistical test for all the correlations shown in Figures 2, 3, and 4. A summary of the statistical tests is given in Table 5. We find that there is no statistically significant correlation. Therefore, it is concluded that the nuclear activity of galaxies studied here has no physical relation to the dynamical properties of the groups. For disk galaxies in nearby HCGs, Iglesias-P´aramo & V´ilchez (1999) have found no clear correlations between the LHα/LB ratio and the dynamical properties of the groups. Our results are consistent with their results. -- 9 -- 3.3. Comparison of the Nuclear Activity between the HCG galaxies and Field Galaxies Our spectroscopic analysis shows that AGN is found in almost half of the HCG galaxies and star-forming activity is found in a quarter of the sample. An important question arises as whether or not these frequencies are unusual with respect to those in environment with less galaxy collisions. In order to examine this issue, we at first make a control sample which consists of so-called field galaxies and then compare the nuclear activity between the HCG galaxies and the field galaxies. Recently Ho et al. (1995, 1997) have made an extensive spectroscopic survey for nearby galaxies using the Palomar Observatory 5 m telescope. Their sample contains 486 galaxies with BT ≤ 12.5 and δ > 0◦ where BT is the apparent total B magnitude and δ is the declination. In order to make a sample of field galaxies, we have omitted the following galaxies from their sample; 1) galaxies belong to the Virgo cluster, 2) binary/interacting galaxies, 3) HCG galaxies (HCG 44a = NGC 3190, HCG 44b = NGC 3193, HCG 44c = NGC 3185, HCG 61a = NGC 4169, HCG 68a = NGC 5353, and HCG 68b = NGC 5354), 4) NGC 1003 whose activity type is uncertain, and 5) the Hubble type is uncertain for five galaxies (NGC 63, 812, 2342, 7798, and UGC 3714). Excluding the above galaxies, we obtain a sample of 382 field galaxies which consist of 167 AGNs, 174 H II nuclei, and 41 normal galaxies. This sample has no matching to the HCG sample in both apparent magnitude and morphology. Since the majority of the HCG galaxies are fainter than the field galaxies observed by Ho et al. (1997), it is difficult to obtain a magnitude-matched sample of field galaxies. However, when we compare the nuclear activity between the HCG galaxies and the field galaxies, we will take account of the morphological difference between the HCGs and the fields. In Figure 5, we show the frequency distributions of activity types for the HCGs (upper -- 10 -- panels) and for the field (lower panels). Applying the χ2 test, we examine whether or not the frequency distributions of the activity types for the HCGs are significantly different from those for the field galaxies for the morphological samples of E -- S0, S0a -- Sbc, Sc or later, and all the galaxies (the total sample). We adopt the null hypothesis that the HCG galaxies and field galaxies come from the same underlying distribution of the activity types. The results of our statistical test are summarized in Table 6. Although the difference in the frequency distribution is not statistically significant for each morphological type, the difference for the total sample is significant in that the HCGs have less H II nuclei while have more absorption galaxies than the field galaxies. The H II nuclei and the absorption galaxies are found in 26% and 31% of the HCG galaxies, respectively. On the other hand, in the field, the H II nuclei share 46% while the absorption galaxies share only 11% of the sample. Taking account that the nuclear activity type depends on the morphological types of host galaxies (e.g., Ho et al. 1997), we examine the difference in the morphological type distribution between the HCG galaxies and the field ones. In Figure 6, we show the frequency distributions of morphological types for each activity type and for the total sample. Applying the χ2 test, we examine whether or not the frequency distributions of the morphological types for our HCG galaxies are significantly different from those for the field galaxies for the nuclear activity type of AGN, H II, absorption, and the total sample. We adopt the null hypothesis that the HCG galaxies and field galaxies come from the same underlying distribution of the morphological types. The results of our statistical test are summarized in Table 7. Our HCG sample contain more E -- S0 galaxies while less late-type spirals than the field. This leads to the under population of H II nuclei in the HCG sample because H II nuclei favor such late-type spirals. However, the frequency of occurrence of AGN in the HCGs is nearly the same as that in the field. A remarkable difference may be that the H II nuclei are found in E -- S0 galaxies more frequently in the HCGs (≃ 13%) -- 11 -- than in the field (≃ 2%). This result appears consistent with the finding by Zepf et al. (1991); there are a number of early-type galaxies with unusually blue colors, suggesting the enhanced star formation in early type galaxies. We have found some interesting difference in the frequency distributions of the activity types between the HCGs and the field described above. However, since the frequency distribution of morphological types is different between the two samples, we cannot conclude that the differences are real. In order to check the effect of the difference in the morphological type distributions, we estimate the frequency of AGN, H II nuclei, and absorption galaxies in the HCGs if the morphological type distribution in the HCGs is the same as that in the field. For example, we can estimate the expected number of AGN in the HCGs as N exp AGN(HCG) = NE−S0 × PAGN,E−S0(Field) + NS0a−Sbc × PAGN,S0a−Sbc(Field) + NSc × PAGN,Sc(Field) where Px,y(Field) is the probability that galaxies with the morphological type y have the activity type x in the field sample. We can also estimate both N exp HII (HCG) and N exp Abs(HCG) in a similar way. We also adopt the null hypothesis that the observed distribution is the same as the expected distribution and apply the χ2 test. The results are given in Table 8. We find that there is no statistical difference in the activity-type distributions between the HCGs and the field. Hence, we conclude that the nuclear activity in the HCGs is not different from that in the field under the assumption that the morphology-activity relation is the same between the HCGs and the field. As mentioned in section 3.1, our spectral analysis may miss dwarf LINERs roughly in a half of early-type galaxies studied here. If we assume that a half of the early-type galaxies classified as "Abs" could be AGNs, our 63 HCG galaxies are classified into 36 AGNs, 16 H II nuclei, and 19 Abs. In this case, we obtain P (χ2) = 0.50. This means that the activity distribution of HCG galaxies is again indistinguishable from that of the field galaxies. -- 12 -- 4. DISCUSSION Our main results are summarized below. (1) We have described the results of our spectroscopic program for a sample of 63 galaxies in the 28 HCGs. We have found in our sample; 28 AGN, 16 H II nuclei, and 19 normal galaxies which show no emission line. We used this HCG sample for statistical analyses. (2) Comparing the frequency distributions of activity types between the HCGs and the field whose data are taken from Ho, Filippenko, & Sargent (382 field galaxies), we find that the frequency of occurrence of H II nuclei in the HCGs is significantly less than that in the field. However, our HCG sample contains more early-type galaxies than the field, the above difference for the H II nuclei may be due to this morphology bias because it is known that H II nuclei are rarer in early-type galaxies than in later ones. (3) Correcting this morphological bias to the HCG sample, we find that there is no significant difference in the frequency of occurrence of emission-line galaxies between the HCGs and the field. This implies that the dense galaxy environment in the HCGs does not affect triggering both AGNs and nuclear starbursts. (4) Since our classification of nuclear activities are judged by the raw optical spectra, we may miss some less-luminous AGNs, in particular in early-type galaxies. Even though this effect is taken into account, the distributions of activity types of HCG galaxies are indistinguishable from those of field galaxies. Our finding seems surprising because it is widely accepted that galaxy interactions lead to either nuclear activity such as AGN or nuclear starbursts or both (see for a review Shlosman, Begelman, & Frank 1990; Barnes & Hernquist 1992). Indeed, in 1980's, several systematic observational investigations of interacting or binary galaxies suggested that galaxy collisions may raise both nuclear activity and intense star formation (e.g., Kennicutt et al. 1984; Keel et al. 1985; Dahari 1985; Bushouse 1986, 1987) although the statistical significance was not so high; i.e., ≃ 90 -- 95% (see for recent papers; De Robertis, Yee, & -- 13 -- Hayhoe 1998; Taniguchi 1999). In addition, luminous and ultraluminous infrared galaxies are often detected in strongly interacting galaxies and merging galaxies (Sanders et al. 1988; see for a review Sanders & Mirabel 1996). Numerical simulations of interacting or merging galaxies have shown that gas fueling driven by galaxy interaction occurs efficiently (e.g., Noguchi 1988; Olson & Kwan 1990a, 1990b; Mihos & Hernquist 1994b). If tidal interactions lead to the formation of AGN and/or nuclear starbursts, we would observe a large number of such active galaxies in the HCGs because the member galaxies are expected to have experienced many tidal interactions during the course of their dynamical evolution. Galaxy interactions affect the star formation activity in galactic disks because the effect of tidal interactions is much stronger in the outer parts than in the nuclear regions (e.g., Noguchi & Ishibashi 1986; see also Kennicutt et al. 1987). Indeed, some radio studies have revealed that a large fraction of HCG spirals are H I deficient (Williams & Rood 1987; Huchtmeier 1997). If a HCG contains several gas-rich spiral galaxies, the average star formation rate would be more enhanced than that in the field galaxies (e.g., Young et al. 1986). However, such excess has not yet been confirmed by IRAS observations (Sulentic & De Mello Rabaca 1993 and references therein). Although deficient of the atomic hydrogen gas in HCG spirals implies that it is expected to occur intense star formation in HCG galaxies, Moles et al. (1994) have concluded that there are no strong starbursting galaxies in HCGs by optical and infrared observations. These results indicate that frequent galaxy collisions are not always able to increase the star formation rate intensely. Although no enhancement of far infrared emission may be partly attributed to that the HCGs prefer early-type spiral galaxies as well as elliptical ones, it should be noted that roughly half galaxies in the HCGs are late-type spirals and irregular galaxies (Hickson et al. 1988; Mendes de Oliveira & Hickson 1994). Therefore, it is suggested that off-nuclear star-formation activity is also not enhanced in the HCGs with respect to field galaxies. -- 14 -- Coziol et al. (1998) have shown from a spectroscopic survey for 17 HCGs (de Carvalho et al. 1997) that AGN are preferentially located in the most early-type and luminous members in the HCGs, suggesting a correlation between activity types, morphologies, and densities of galaxies in HCGs. They searched more possible member galaxies outside the original HCG members and then found the above interesting observational properties. However, our spectroscopic survey was made for only the original HCG members. Therefore, we do not think that our results are inconsistent with their results. An interesting point suggested by Coziol et al. (1998) is that AGN is preferentially found in luminous, early-type galaxies. Verdes-Montenegro et al. (1998) showed from their 12CO(J=1-0) emission survey for a large number of HCG galaxies that a number of early-type galaxies are detected in CO as well as in FIR. In addition, early-type galaxies with unusually blue colors are also found by Zepf et al. (1991). Although all these may be still circumstantial lines of evidence, it is suggested that the majority of early-type galaxies in the HCGs are affected by some environmental effect. One possible important effect is a merger between an early-type galaxy and a gas-rich galaxy such as a late-type spiral or a small satellite galaxy since galaxy mergers between unequal galaxies may lead to the formation of S0 galaxies (Bekki 1998). The above arguments suggest that mere tidal interactions between galaxies are not responsible for the triggering intense nuclear activities. Recently, instead of mere tidal interactions, minor mergers have been appreciated as a more important triggering mechanism both for nuclear starbursts (Mihos & Hernquist 1994a; Hernquist & Mihos 1995; Taniguchi & Wada 1996) and for Seyfert nuclei (De Robertis et al. 1998; Taniguchi 1999; see for an earlier indication Gaskell 1985). If this is the case, it is not surprising that the nuclear activities in the HCG galaxies are not significantly different from those in the field galaxies. Furthermore, if major mergers are more important to activate mode luminous starbursts and AGNs (e.g., Sanders et al. 1988), it is suggested that most of the HCGs -- 15 -- have not yet experienced such major mergers in the member galaxies. Since the dynamical relaxation timescale for the HCGs is shorter than the Hubble times, it is expected that each HCG will merge into one within a timescale of several Gyr (Hickson et al. 1992). Therefore, the HCG are expected to evolve either to luminous or ultraluminous infrared galaxies via multiple mergers (Xia et al. 1997; Taniguchi, Wada, & Murayama 1997; Taniguchi & Shioya 1998; L´ipari et al. 2000; Borne et al. 2000), or to quasars (Sanders et al. 1988; Taniguchi, Ikeuchi, & Shioya 1999) or to ordinary-looking elliptical galaxies (Barnes 1989; Weil & Hernquist 1996; Nishiura et al. 1997). We are grateful to the staff of OAO for kind help of the observations. We would like to thank an anonymous referee for useful comments and suggestions. YO and TM are JSPS Fellows. This work was partly supported by the Ministry of Education, Science, Culture, and Sports (Nos. 07044054, 10044052, and 10304013). -- 16 -- REFERENCES Barnes, J. 1989, Nature, 338, 123 Barnes, J. E., & Hernquist, L. , 1992, Nature, 360, 715 Bekki, K. 1998, ApJ, 502, L133 Borne, K., D., Bushouse, H., Lucas, R. A., & Colina, L. 2000, ApJ, in press (astro- ph/9912151) Bushouse, H. A. 1986, AJ, 91, 255 Bushouse, H. A. 1987, ApJ, 320, 49 Coziol, R., Ribeiro, A. L. B., de Carvalho, R. R., & Capelato, H. V. 1998, ApJ, 493, 563 Dahari, O. 1985, ApJS, 57, 643 de Carvalho, R. R., & Coziol, R. 1999, ApJ, in press (astro-ph/9901006) de Carvalho, R. R., Ribeiro, A. L. B., Capelato, H. V., & Zepf, S. E. 1997, ApJS, 110, 1 De Robertis, M. M., Yee, H. K. C., & Hayhoe, K. 1998, ApJ, 496, 93 de Vaucouleurs, G., de Vaucouleurs, A., Corwin, H. G. Jr., Buta, R. J., Paturel, G., & Fouqu´e, P. 1991, Third Reference Catalogue of Bright Galaxies (Springer-Verlag) (RC3) Gaskell, C. M. 1985, Nature, 315, 386 Hernquist, L., & Mihos, C. J. 1995, ApJ, 448, 41 Hickson, P. 1982, ApJ, 255, 382 Hickson, P. 1993, Astrophys. Lett. & Comm., 29, 1 Hickson, P., Kindle, E., & Huchra, J. P. 1988, ApJ, 331, 64 Hickson, P., Mendes de Oliveira, C., Huchra, J. P., & Palumbo, G. G. C. 1992, ApJ, 399, 353 -- 17 -- Hickson, P., Menon, T. K., Palumbo, G. G. C., & Pesic, M. 1989, ApJ, 341, 679 Ho, L. C., Filippenko, A. V., & Sargent, W. L. W. 1995, ApJS, 98, 477 Ho, L. C., Filippenko, A. V., & Sargent, W. L. W. 1997, ApJS, 112, 315 Huchtmeier, W. K. 1997, A&A, 325, 473 Iglesias-P´aramo, J., & V´ılchez J.M. 1997a, ApJ 479, 190 Iglesias-P´aramo, J., & V´ılchez J.M. 1997b, ApJ 489, L13 Iglesias-P´aramo, J., & V´ılchez J.M. 1999, ApJ 518, 94 Keel, W. C., Kennicutt, R. C. Jr., Hummel, E., & van der Hulst, J. M. 1985, AJ, 90, 708 Kennicutt, R. C. Jr., & Keel, W. C. 1984, ApJ, 279, L5 Kennicutt, R. C. Jr., Roettiger, K. A., Keel, W. C., van der Hulst, J., & Hummel, E. 1987, AJ, 93, 1011 Kosugi, G., Ohtani, H., Sasaki, T., Koyano, H., Shimizu, Y., Yoshida, M., Sasaki, M., Aoki, K., & Baba, A. 1995, PASP, 107, 474 L´ipari, S., Diaz, R., Taniguchi, Y., Terlevich, R., Dottori, H., & Carranza, G. 2000, ApJ, in press (astro-ph/9911019) Longo G., Grimaldi A., & Richter G. 1995, A&A 299, L45 Mendes de Oliveira, C., & Hickson, P. 1994, ApJ, 427, 684 Menon, T. K. 1992, MNRAS, 255, 41 Menon, T. K. 1995, MNRAS, 274, 845 Mihos, C. J., & Hernquist, L. 1994a, ApJ, 425, L13 Mihos, C. J., & Hernquist, L. 1994b, ApJ, 431, L9 Moles M., del Olmo A., Perea J., Masegosa J., M´arquez I., & Costa V. 1994, A&A 285, 404 -- 18 -- Nishiura, S., Sato, Y., Murayama, Y., & Taniguchi, Y. 1997, IAU Symp., 186, Galaxy Interactions at Low and High Redshift, eds. J. E. Barnes, and D. B. Sanders (Klewer: Dordrecht), 415 Nishiura, S., Shimada, M., Ohyama, Y., Murayama, T., & Taniguchi, Y. 1999, in preparation Noguchi, M., & Ishibashi, S. 1986, MNRAS, 219, 305 Noguchi, M. 1988, A & A, 203, 259 Olson, K. M., & Kwan, J. 1990a, ApJ, 349, 480 Olson, K. M., & Kwan, J. 1990b, ApJ, 361, 426 Ribeiro, A. L. B., de Carvalho, R. R., Coziol, R., Capelato, H. V., & Zepf, S. E. 1996, ApJ 463, L5 Sanders, D. B., & Mirabel, I. F. 1996, ARA & A, 34, 749 Sanders, D. B., et al. 1988, ApJ, 325, 74 Shlosman, I., Begelman, M. C., & Frank, J. 1990, Nature, 345, 679 Sulentic, J. W. 1997, ApJ, 482, 640 Sulentic, J. W., & de Mello Rabaca, D. F. 1993, ApJ, 410, 520 Taniguchi, Y. 1999, ApJ, 524, 65 Taniguchi, Y., Ikeuchi, S., & Shioya, Y. 1999, ApJ, 514, L9 Taniguchi, Y., Ohyama Y., Yoshida M., Yamada T., Mouri H. 1996, ApJ, 467, 215 Taniguchi, Y., & Shioya, Y. 1998, ApJ, 501, L167 Taniguchi, Y., & Wada, K. 1996, ApJ, 469, 581 Taniguchi, Y., Wada, K., & Murayama, T. 1997, RMxAC, 6, 240 Valluri, M., & Anupama, G. C. 1996, AJ 112, 1390 -- 19 -- Veilleux, S., & Osterbrock, D. E. 1987, ApJS, 63, 295 Verdes-Montenegro, L., Yun, M. S., Perra, J., del Olmo, A., & Ho, P. T. P. 1998, ApJ, 497, 89 V´ılchez, J. M., & Iglesias-P´aramo, J. 1998a, ApJS, 117, 1 V´ılchez, J. M., & Iglesias-P´aramo, J. 1998b, ApJ, 506, L101 Weil, M. L., & Hernquist, L. 1996, ApJ, 460, 101 Williams, B. A., & Rood, H. J. 1987, ApJS, 63, 265 Young, J. S., Kenney, J. D. Tacconi, L., Claussen, M. J., Huang, Y. -L., Tacconi-Garman, L., Xie, S., & Schloerb, F. P. 1986, ApJ, 311, L17 Xia, X. Y., Deng, Z. G., Wu, H., & Boller, T. 1997, IAU Symp., 186, Galaxy Interactions at Low and High Redshift, eds. D. B. Sanders, and J. Barnes (Klewer: Dordrecht), 421 Zepf, S. E. 1993, ApJ, 407, 448 Zepf, S. E., & Whitmore, B. C. 1991, ApJ, 383, 542 Zepf, S. E., Whitmore, B. C., & Levison, H. F. 1991, ApJ 383, 524 This manuscript was prepared with the AAS LATEX macros v4.0. -- 20 -- Fig. 1. -- Nuclear spectra of the HCG galaxies. Fig. 2. -- Correlations between the [N ii]/Hα ratio and the number density of galaxies in the HCGs ρN. The results are shown for all the sample (top left), E -- S0 galaxies (top right), S0a -- Sbc galaxies (bottom left), and Sc or later (bottom right). Fig. 3. -- Correlations between the [N ii]/Hα ratio and the radial velocity dispersion of galaxies in the HCGs σr. The results are shown for all the sample (top left), E -- S0 galaxies (top right), S0a -- Sbc galaxies (bottom left), and Sc or later (bottom right). Fig. 4. -- Correlations between the [N ii]/Hα ratio and the crossing time of galaxies in the HCGs tc. The results are shown for all the sample (top left), E -- S0 galaxies (top right), S0a -- Sbc galaxies (bottom left), and Sc or later (bottom right). Fig. 5. -- Comparison of frequency distributions of nuclear activity types between the HCGs and the field for all the sample, E -- S0 galaxies, S0a -- Sbc galaxies, and Sc or later. Fig. 6. -- Comparison of frequency distributions of morphological types between the HCGs and the field for all the sample, AGN, H ii nuclei, and absorption galaxies. 2 1 0 -- 1 2 1 0 -- 1 α H / ] I I N [ g o l All E -- S0 S0a -- Sbc 3 4 5 6 Sc -- 7 4 5 6 7 ρ log N / Mpc -- 3 Table 2. Properties of Line Emissions HCG a λc 7a 6650.66 0.09 Hα ∆λb 5.15 0.14 f luxc a λc ∆λb f luxc a λc ∆λb f luxc a λc ∆λb f luxc a λc ∆λb f luxc [NII]λ6548 [NII]λ6583 [SII]λ6717 [SII]λ6731 32.61 6635.91 2.01 0.09 5.15 0.14 5.42 0.48 6671.21 0.09 5.15 0.14 15.99 6806.19 1.41 0.37 4.33 0.57 3.40 1.15 6820.59 0.37 4.33 0.57 4.27 1.31 6698.58 13.46 14.76 0.95 1.62 3.69 10a 10b 30a 30b 31a 6651.73 0.02 31b 6653.34 0.05 31c 6650.67 0.02 34a 6762.39 1.37 37a 37b 6712.35 0.28 38a 6752.59 0.13 38b 6752.91 0.08 38c 6751.78 0.23 40a 40b 0 0 0 2 r a M 4 1 v 6 5 0 3 0 0 0 / h p - o r t s a : v i X r a 3.32 0.03 2.74 0.07 4.03 0.03 8.62 1.99 2.77 0.44 5.62 0.20 5.85 0.13 5.58 0.35 6647.68 0.77 84.76 6636.98 0.02 1.96 27.80 1.52 457.64 6635.92 0.02 6.72 1.13 3.32 0.03 2.79 1.02 4.97 0.38 6682.98 0.77 6672.28 0.02 4.03 0.03 12.87 6671.22 1.38 0.02 6692.22 0.53 8.27 0.83 9.79 2.10 7.45 3.50 2.58 3.58 1.21 13.51 6737.84 1.04 0.13 63.80 6738.16 3.10 0.08 13.89 6737.03 1.91 0.23 5.62 0.20 5.85 0.13 5.58 0.35 1.55 0.22 12.03 6773.46 0.77 2.48 0.46 0.08 6772.33 0.23 6.72 1.13 3.32 0.03 4.03 0.03 8.62 1.99 8.27 0.83 8.82 2.17 5.62 0.20 5.85 0.13 5.58 0.35 8.24 3.01 14.65 6807.60 1.11 0.07 6809.27 0.14 37.96 6806.30 4.08 6.62 3.45 28.90 6.19 4.53 2.41 4.58 0.64 0.09 6909.38 0.54 35.50 6911.67 2.26 7.31 1.36 0.39 6910.63 0.27 2.85 0.10 3.03 0.20 4.41 0.14 5.45 0.88 5.23 0.64 2.35 0.31 14.38 6822.00 1.14 5.92 0.91 0.07 6823.67 0.14 31.54 6820.70 2.26 0.09 2.72 0.97 4.08 1.24 2.59 0.78 6923.78 0.54 6926.07 0.39 6925.03 0.27 2.85 0.10 3.03 0.20 4.41 0.14 5.45 0.88 5.23 0.64 2.35 0.31 9.06 0.92 3.92 0.74 25.87 2.04 1.71 0.75 5.08 1.41 1.15 0.59 6782.94 1.37 6727.52 0.53 6731.07 1.43 6773.14 0.13 40c 6703.98 0.25 5.62 0.36 12.42 6689.23 1.82 0.25 5.62 0.36 42a 6630.75 12.45 2.37 3.80 2.47 0.46 3.87 2.52 6724.52 0.25 5.62 0.36 7.30 1.35 6666.05 12.45 11.41 2.37 3.80 7.42 44a 6592.53 3.86 6.44 6577.89 10.63 46.84 6613.19 10.63 138.18 6746.08 10.71 33.78 6760.48 10.71 37.79 HCG a λc Hα ∆λb f luxc a λc ∆λb f luxc a λc ∆λb f luxc a λc ∆λb f luxc a λc ∆λb f luxc [NII]λ6548 [NII]λ6583 [SII]λ6717 [SII]λ6731 0.98 1.40 5.15 0.20 0.30 2.85 0.20 0.30 8.39 0.50 0.53 4.79 0.50 0.53 4.99 Table 2 -- Continued 44b 44c 6589.46 0.05 44d 6597.17 0.23 4.99 0.08 2.94 0.35 47a 51a 51b 51f 126.59 6574.71 4.49 3.72 0.99 0.05 6582.42 0.23 6760.87 0.40 4.99 0.08 2.94 0.35 3.73 0.59 32.14 6610.01 1.28 0.41 0.20 0.88 0.30 0.05 6617.72 0.23 6796.17 0.40 4.99 0.08 2.94 0.35 3.73 0.59 94.80 6743.86 0.12 3.78 1.20 0.58 2.58 0.90 5.25 0.19 25.92 2.29 6758.26 0.12 5.25 0.19 27.71 2.38 6727.71 1.63 9.60 2.50 1.42 0.80 6763.01 1.63 9.60 2.50 4.19 2.35 53a 6700.64 0.89 57a 6755.24 0.42 61a 6644.47 0.47 61c 6650.69 0.16 8.93 1.30 6.18 0.67 2.41 0.91 7.73 0.24 3.53 1.48 9.19 2.63 5.39 4.32 28.93 2.05 6685.89 0.89 6740.49 0.42 6631.41 0.35 6635.94 0.16 8.93 1.30 6.18 0.67 9.50 0.52 7.73 0.24 1.83 0.62 4.33 1.06 6721.19 0.89 6775.79 0.42 18.88 6666.71 2.25 7.15 0.58 0.35 6671.24 0.16 8.93 1.30 6.18 0.67 9.50 0.52 7.73 0.24 5.40 1.82 12.79 3.13 55.69 6.62 21.09 1.72 6799.77 10.79 22.34 6814.17 10.79 21.74 0.55 6805.96 0.43 0.60 6.70 0.61 3.39 6.99 1.40 0.55 6820.36 0.43 0.60 6.70 0.61 3.29 2.68 0.95 61d 62a 62b 62c 6641.17 0.56 6.90 0.85 4.38 1.17 6676.47 0.56 6.90 0.85 12.92 3.44 67b 6613.54 0.51 68a 68b 6621.39 0.31 6.33 0.79 5.54 0.45 2.44 1.42 5.40 2.61 6598.79 0.51 6595.86 0.56 6606.64 0.31 6.33 0.79 8.76 0.84 5.54 0.45 2.88 0.78 6634.09 0.51 16.80 6631.16 3.50 8.25 1.46 0.56 6641.94 0.31 6.33 0.79 8.76 0.84 5.54 0.45 8.49 2.30 49.56 10.32 24.33 4.32 HCG a λc 68c 6613.24 0.05 Hα ∆λb 5.13 0.08 71a 6765.61 11.37 0.19 72a 6842.24 0.42 73a 6686.00 0.12 79a 6653.45 0.28 79b 6657.86 0.11 79c 79d 6663.08 0.10 80a 6759.43 0.11 82a 82b 6788.72 0.29 87a 6747.87 0.31 87c 6756.76 0.10 88a 88b 6700.00 0.35 88c 6693.22 0.12 88d 6696.14 0.10 89a 6755.68 0.15 92a 6579.86 0.24 3.38 0.61 3.39 0.17 2.86 0.44 5.79 0.16 3.12 0.13 5.77 0.18 6.68 0.47 3.86 0.46 3.18 0.15 2.75 0.62 3.29 0.17 3.40 0.15 4.38 0.22 2.92 f luxc a λc ∆λb f luxc a λc ∆λb f luxc a λc ∆λb f luxc a λc ∆λb f luxc [NII]λ6548 [NII]λ6583 [SII]λ6717 [SII]λ6731 Table 2 -- Continued 6598.49 0.05 5.13 0.08 6750.86 11.37 25.35 1.05 15.33 6633.79 0.05 5.13 0.08 6786.16 11.37 96.54 3.66 22.35 1.67 7.24 3.20 0.19 6827.49 0.42 15.31 6672.69 1.71 9.09 3.14 63.64 3.96 5.62 0.52 39.60 2.77 0.29 6638.70 0.28 6643.11 0.11 6648.33 0.10 6744.68 0.11 11.04 6773.97 1.70 2.64 0.78 6.88 0.72 2.45 1.16 5.84 0.69 7.37 0.71 8.94 1.00 3.21 0.29 6733.12 0.31 6742.01 0.10 6679.15 0.36 6683.84 0.32 6678.47 0.12 6681.39 0.10 6740.93 0.15 6565.11 0.24 3.38 0.61 3.55 0.43 2.86 0.44 5.79 0.16 3.12 0.13 5.77 0.18 6.68 0.47 3.86 0.46 3.18 0.15 7.99 0.54 9.15 0.49 3.29 0.17 3.40 0.15 4.38 0.22 2.92 0.75 2.26 1.02 2.16 0.58 1.87 0.78 9.63 0.91 0.29 0.10 9.80 0.79 2.14 0.43 0.87 0.25 1.00 0.16 7.66 1.11 5.37 0.62 0.83 0.15 1.19 0.16 1.47 0.23 0.46 0.19 6862.79 0.42 6707.99 0.29 6674.00 0.28 6678.40 0.11 6683.63 0.10 6779.98 0.11 6809.27 0.29 6768.42 0.31 6777.31 0.10 6714.44 0.36 6719.14 0.32 6713.77 0.12 6716.69 0.10 6776.23 0.15 6600.41 0.24 3.38 0.61 3.55 0.43 2.86 0.44 5.79 0.16 3.12 0.13 5.77 0.18 6.68 0.47 3.86 0.46 3.18 0.15 7.99 0.54 9.15 0.49 3.29 0.17 3.40 0.15 4.38 0.22 2.92 6768.02 0.17 6.56 0.26 18.23 1.72 6782.42 0.17 6.56 0.26 19.31 1.81 74.77 3.09 45.21 2.20 6.67 3.00 6.39 1.70 5.53 2.31 28.42 2.70 6814.39 0.31 6.47 0.47 8.36 1.49 6828.79 0.31 6.47 0.47 8.65 1.53 6818.77 0.21 3.69 0.32 1.72 0.34 6833.17 0.21 3.69 0.32 1.11 0.27 0.85 0.29 28.90 2.32 6.32 1.28 2.56 0.75 2.96 0.47 22.59 3.29 15.83 1.84 2.44 0.45 3.50 0.48 4.33 0.68 1.36 Table 2 -- Continued HCG a λc Hα ∆λb f luxc a λc ∆λb f luxc a λc ∆λb f luxc a λc ∆λb f luxc a λc ∆λb f luxc [NII]λ6548 [NII]λ6583 [SII]λ6717 [SII]λ6731 0.17 0.25 0.62 0.17 0.25 0.13 0.17 0.25 0.39 92b 92c 6710.93 0.05 8.38 0.10 117.07 6696.18 4.18 0.05 8.38 0.10 58.12 6731.48 1.53 0.05 8.38 0.10 171.46 6866.47 4.51 0.18 9.19 0.21 57.59 6880.87 3.58 0.18 9.19 0.21 56.19 3.54 92d 92e 93a 6673.92 10.08 0.32 93b 6666.56 0.07 0.46 4.16 0.11 93c 8.88 2.21 0.32 15.62 6651.81 0.93 0.07 6659.36 0.59 aThe center wavelength of line spectrum in the unit of A. bThe full width at half maxmum of line spectrum in the unit of A. cThe flux of line spectrum in the unit of 10−16 ergs s−1 cm−2. 96a 6753.33 0.18 8.95 0.39 52.54 6738.58 5.88 0.18 52.23 6909.15 11.20 11.43 6923.55 11.20 16.83 4.98 0.71 0.77 2.47 0.71 0.77 2.87 6659.17 10.08 11.27 6694.47 10.08 33.23 6830.10 0.46 4.16 0.11 6.92 0.88 8.95 0.39 1.13 3.48 0.25 3.92 1.08 0.32 6687.11 0.07 6694.66 0.59 17.71 6773.88 1.69 0.18 0.46 4.16 0.11 6.92 0.88 8.95 0.39 3.33 0.41 10.26 6821.91 0.33 0.72 11.55 3.18 8.87 0.51 4.17 0.49 13.94 6844.50 1.91 3.23 0.84 0.41 6836.31 0.33 8.87 0.51 4.17 0.49 8.66 1.62 1.52 0.59 α H / ] I I N [ g o l 2 1 0 -- 1 2 1 0 -- 1 -- 2 All E -- S0 S0a -- Sbc 1 2 Sc -- 1 2 3 3 σ log r / km s -- 1 α H / ] I I N [ g o l 2 1 0 -- 1 2 1 0 -- 1 -- 2 All E -- S0 S0a -- Sbc Sc -- -- 3 -- 2 -- 1 0 1 -- 3 -- 2 -- 1 0 1 log H0tc HCGs Fields ) % ( y c n e u q e r F 100 80 60 40 20 100 100 80 80 60 60 40 40 20 20 0 0 N G A I I H s b A I I H s b A N G A I I H s b A N G A N G A I I H s b A All E -- S0 S0a -- Sbc Sc -- Frequency(%) 0 0 2 2 0 0 4 4 0 0 6 6 0 0 8 8 0 0 1 1 0 0 0 0 2 0 4 0 6 0 8 0 1 0 0 i F e d s l H C G s A l l E -- S0 S0a -- Sbc Sc -- A G N E -- S0 S0a -- Sbc Sc -- H I I E -- S0 S0a -- Sbc Sc -- A b s E -- S0 S0a -- Sbc Sc -- Table 1. A journal of observations HCG Date Exp. (sec) PA (deg) Type Hickson Type RC3 Type Adopted 0 0 0 2 r a M 4 1 v 6 5 0 3 0 0 0 / h p - o r t s a : v i X r a 7a 10a 10b 30a 30b 31a 31b 31c 34a 37a 37b 38a 38b 38c 40a 40b 40c 42a 42b 42c 44a 44b 44c 44d 47a 51a 51b 51f 52a 53a 57a 61a 61c 61d 62a 62b 62c 67b 68a 68b 68c 71a 72a 73a 79a 79b 79c 19 Aug 1996 21 Feb 1996 07 Jan 1997 21 Feb 1996 23 Feb 1996 09 Jan 1997 09 Jan 1997 09 Jan 1997 23 Feb 1996 22 Feb 1996 21 Feb 1996 06 Jan 1997 23 Feb 1996 23 Feb 1996 23 Feb 1996 23 Feb 1996 23 Feb 1996 09 Jan 1997 09 Jan 1997 09 Jan 1997 21 Feb 1996 22 Feb 1996 21 Feb 1996 21 Feb 1996 21 Feb 1996 23 Feb 1996 23 Feb 1996 23 Feb 1996 20 Feb 1996 20 Feb 1996 25 Feb 1996 22 Feb 1996 21 Feb 1996 21 Feb 1996 22 Feb 1996 22 Feb 1996 22 Feb 1996 21 Feb 1996 22 Feb 1996 22 Feb 1996 21 Feb 1996 20 Feb 1996 25 Feb 1996 25 Feb 1996 22 Feb 1996 15 Aug 1996 18 Aug 1996 1800 1800 1800 1800 1800 1800 1800 860 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 240 1800 1800 1800 1800 1800 1800 1800 1800 1800 750 1800 1800 1800 1800 1800 1800 1800 1800 900 1800 1800 1800 1800 1800 1800 758 1800 1800 159 62 153 124 29 80 41 41 84 171 77 340 85 85 185 42 122 134 66 134 125 185 130 125 15 358 183 358 90 93 108 159 130 50 130 130 173 21 0 0 18 23 102 18 63 70 34 Sb SBb E1 SBa Sa Sdm Sm Im E2 E7 Sbc Sbc SBd Im E3 S0 Sbc E3 SB0 E2 Sa E2 SBc Sd SBb E1 SBbc S0 SBab SBbc Sb S0a Sbc S0 E3 S0 S0 Sc S0 E2 SBbc SBc Sa Scd E0 S0 S0 (R')SB(r)a: SB(r)b S0−: (R')SB(rs)0+: (R')SAB(rs)0+: Pec S? S? S0? Sb S? S? S? S? E SA(r)0−: pec SB(rs)b pec E3: SB(rs)00 E? SA(s)a pec E2 (R)SB(r)a SB(s)c pec SA(r): E S? S0? SB? SB? Sab pec S0 S S? S0+ S00 pec E+ Sb S0 S0 SB(r)b Scd: S0? SA(s)c Sa pec S0? S0 pec Sa Sb S0 S0 S P Sm Im S0 Sb Sbc Sbc Sd Im E S0 Sb E S0 E Sa E Sa Sc Sb E Sbc S0 Sab Sbc Sab S0 Sbc S0 S0 S0 E Sb S0 S0 Sb Scd S0 Sc Sa S0 S0 HCG Date Exp. (sec) PA (deg) Type Hickson Table 1. (continued) 79d 80a 82a 82b 87a 87b 87c 87d 88a 88b 88c 88d 89a 92a 92b 92c 92d 92e 93a 93b 93c 96a 18 Aug 1996 25 Feb 1996 15 Aug 1996 19 Aug 1996 18 Aug 1996 20 Aug 1996 19 Aug 1996 20 Aug 1996 15 Aug 1996 15 Aug 1996 15 Aug 1996 18 Aug 1996 19 Aug 1996 15 Aug 1996 15 Aug 1996 18 Aug 1996 15 Aug 1996 18 Aug 1996 19 Aug 1996 19 Aug 1996 19 Aug 1996 19 Aug 1996 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 1800 480 0 64 37 168 56 40 90 40 132 31 31 71 57 137 22 140 109 110 98 143 98 90 Sdm Sd E3 SBa Sbc S0 Sd Sd Sb SBb Sc Sc Sc Sd Sbc SBa SB0 Sa E1 SBd SBa Sc Type RC3 SB(s)c S? S0 SB0? S00 pec S0? S? Sb SB(r)a pec: SAB(r)bc? S? S? SA(s)d SB(s)bc pec SB(s)bc pec E2 pec SA0− SB(s)cd pec (R)SAB(s)0/a pec SA(r)bc pec Type Adopted Sc Sd S0 S0 S0 S0 Sd Sd Sb Sa Sbc Sc Sc Sd Sbc Sbc E Sa S0 Scd S0/a Sbc Table 3. Line Ratio and Dynamical Properties HCG [N II]/Hαa [S II]/Hαb [O I]/Hαc log ρN (Mpc−3) log σr (km s−1) log H0tc Activity Type 0.02 ± 0.01 07a 10a 10b 30a 30b 31a 31b 31c 34a 37a 37b 38a 38b 38c 40a 40b 40c 42a 42b 42c 44a 44b 44c 44d 47a 51a 51b 51f 52a 53a 57a 61a 61c 61d 62a 62b 62c 67b 68a 68b 68c 71a 72a 73a 79a 79b 79c 0.49 ± 0.07 0.24 ± 0.09 0.28 ± 0.03 0.35 ± 0.08 0.13 ± 0.01 0.33 ± 0.15 0.14 ± 0.05 0.27 ± 0.14 0.17 ± 0.02 0.08 ± 0.01 1.89 ± 2.39 1.26 ± 1.10 0.34 ± 0.07 0.56 ± 0.06 0.53 ± 0.17 0.59 ± 0.19 21.46 ± 18.45 11.11 ± 10.40 0.75 ± 0.06 0.32 ± 0.24 0.42 ± 0.05 1.53 ± 1.15 1.39 ± 0.74 10.33 ± 9.51 0.73 ± 0.11 8.17 ± 7.79 0.33 ± 0.10 3.48 ± 2.98 4.51 ± 2.98 0.77 ± 0.06 2.02 ± 0.25 0.92 ± 0.82 0.42 ± 0.16 0.61 ± 0.46 0.45 ± 0.07 0.39 ± 0.05 0.27 ± 0.06 4.00 3.07 3.07 3.85 3.85 6.13 6.13 6.13 5.41 4.70 4.70 3.55 3.55 3.55 5.54 5.54 5.54 4.03 4.03 4.03 4.24 4.24 4.24 4.24 4.30 3.77 3.77 3.77 3.04 3.58 3.64 4.48 4.48 4.48 4.69 4.69 4.69 3.91 4.52 4.52 4.52 3.76 4.33 6.49 6.49 6.49 1.95 2.32 2.32 1.86 1.86 1.75 1.75 1.75 2.50 2.60 2.60 1.11 1.11 1.11 2.17 2.17 2.17 2.33 2.33 2.33 2.13 2.13 2.13 2.13 1.63 2.38 2.38 2.38 2.26 1.91 2.43 1.94 1.94 1.94 2.46 2.46 2.46 2.32 2.19 2.19 2.19 2.62 2.42 2.14 2.14 2.14 -1.40 -1.48 -1.48 -1.22 -1.22 -1.80 -1.80 -1.80 -2.44 -2.27 -2.27 0.88 0.88 0.88 -2.12 -2.12 -2.12 -1.81 -1.81 -1.81 -1.65 -1.65 -1.65 -1.65 0.66 -1.73 -1.73 -1.73 -1.43 -1.14 -1.69 -1.60 -1.60 -1.60 -2.16 -2.16 -2.16 -1.76 -1.79 -1.79 -1.79 -2.04 -2.00 -2.43 -2.43 -2.43 HII AGN Abs Abs AGN HII HII HII AGN? AGN AGN HII HII HII Abs Abs HII Abs ? Abs AGN Abs AGN HII Abs Abs AGN Abs ? AGN AGN AGN AGN Abs AGN Abs Abs AGN AGN AGN AGN AGN AGN HII AGN HII Abs Table 3. (continued) HCG [N II]/Hαa [S II]/Hαb [O I]/Hαc log ρN (Mpc−3) log σr (km s−1) log H0tc Activity Type 79d 80a 82a 82b 87a 87b 87c 87d 88a 88b 88c 88d 89a 92a 92b 92c 92d 92e 93a 93b 93c 96a 0.50 ± 0.16 0.15 ± 0.07 0.73 ± 0.11 0.57 ± 0.20 0.97 ± 0.57 0.43 ± 0.11 6.47 ± 3.83 0.42 ± 0.13 0.47 ± 0.11 0.48 ± 0.13 0.42 ± 0.20 1.46 ± 0.09 0.97 ± 0.10 0.28 ±0.05 3.74 ± 1.31 0.66 ± 0.09 2.55 ± 1.03 0.30 ± 0.11 0.99 ± 0.21 0.54 ± 0.16 a[N II]λ6583/Hα. The error is one σ value. b[S II](λ6717+λ6731)/Hα. The error is one σ value. c[O I]λ6300/Hα. The error is one σ value. 6.49 4.78 3.43 3.43 4.39 4.39 4.39 3.49 3.49 3.49 3.49 3.67 4.63 4.63 4.63 4.63 3.43 3.43 3.43 4.54 2.14 2.43 2.79 2.79 2.08 2.08 2.08 1.43 1.43 1.43 1.43 1.74 2.59 2.59 2.59 2.59 2.32 2.32 2.32 2.12 -2.43 -2.16 -2.07 -2.07 -1.56 -1.56 -1.56 0.94 0.94 0.94 0.94 -0.84 -2.27 -2.27 -2.27 -2.27 -1.59 -1.59 -1.59 -1.76 HII AGN Abs HII AGN Abs HII ? AGN AGN HII HII HII HII Abs AGN Abs Abs AGN AGN AGN AGN Table 4. Comaparison of nuclear activities between our study and Coziol et al. (1998) HCG Type Coziol et al. (1998) Our study Adopted [N II]/Hα Activity Type [N II]/Hα Activity Type 40a 40b 42a 42c 62a 62b 62c 87a 87b 87c 88a 88b 88d E S0 E E S0 S0 E S0 S0 Sd Sb Sa Sc 0.87±0.06 1.78±0.04 0.60±0.01 0.83 1.29 0.44 0.89 1.91 0.32 AGN Abs AGN Abs AGN Abs Abs AGN AGN HII AGN AGN HII Abs Abs Abs Abs AGNa Abs Abs AGN Abs HII AGNa AGN HII 0.97±0.57 0.43±0.11 6.47±3.83 0.47±0.11 aWe have not detected Hα but have detected [N II]λ6583. Table 5. A summary of the Spearman rank test for the correlations between [N ii]/Hα ratio and the dynamical properties of the HCGsa All E−S0 S0a−Sbc Sc≤ log ρN 0.82 0.43 0.71 0.40 log([N II]λ6583/Hα) log σr 0.004 0.93 0.08 0.16 log H0tc 0.97 0.07 0.76 0.29 aThe probability that rejects the null hypothesis. Table 6. Frequency distributions of the nuclear activity types between the HCGs and the fielda Allb E-S0 S0a-Sbc Sc≤ HCGs Field HCGs Field HCGs Field HCGs Field AGN 44.4 (28) 25.4 (16) H II Abs 30.2 (19) P(χ2)c 43.7 (167) 45.5 (174) 10.7 (41) 28.0 (7) 8.0 (2) 64.0 (16) 60.8 (59) 4.1 (4) 35.1 (34) 70.8 (17) 16.7 (4) 12.5 (3) 57.2 (83) 40.0 (58) 2.8 (4) 28.6 (4) 71.4 (10) 0.0 (0) 17.9 (25) 80.0 (112) 2.1 (3) 3.93 × 10−5 0.01 0.01 0.55 aNumbers in parentheses are the actual numbers. b"All" means that the total sample of E−S0, S0a−Sbc, and Sc≤ galaxies. cThe possibility that rejects the null hypothesis. Table 7. Frequency distributions of the morphological types between the HCGs and the field as a function of the nuclear activitya Allb AGN H II Abs HCGs Field HCGs Field HCGs Field HCGs Field E-S0 S0a-Sbc Sc≤ P(χ2)c 39.7 (25) 38.1 (24) 22.2 (14) 25.4 (97) 38.0 (145) 36.6 (140) 25.0 (7) 60.7 (17) 14.3 (4) 35.3 (59) 49.7 (83) 15.0 (25) 12.5 (2) 25.0 (4) 62.5 (10) 2.3 (4) 33.3 (58) 64.4 (112) 84.2 (16) 15.8 (3) 0.0 (0) 82.9 (34) 9.8 (4) 7.3 (3) 0.03 0.51 0.08 0.41 aNumbers in parentheses are the actual numbers. b"All" means that the total sample of AGN, H II, and absorption galaxies. cThe possibility that rejects the null hypothesis. Table 8. Comparisons of expected numbers of the activity types with the observations for the HCG samplea Observations Corrected numbersb Expected numbers AGN H II Abs P(χ2)c 44.4 (28) 25.4 (16) 30.2 (19) 0.14 56.5 (35) 25.4 (16) 19.4 (12) 0.50 49.2 (31) 34.9 (22) 15.9 (10) aNumbers in parentheses are the actual numbers. bExpected numbers under assuming that we have taken AGN for Abs for a half (3/7) of early-type galaxies cThe possibility that rejects the null hypothesis.
astro-ph/0411415
1
0411
2004-11-15T17:20:09
Cosmological reionization after WMAP: perspectives from PLANCK and future CMB missions
[ "astro-ph" ]
The WMAP first year detection of a high redshift reionization through its imprints on CMB anisotropy T and TE mode angular power spectra calls for a better comprehension of the universe ionization and thermal history after the standard recombination. Different reionization mechanisms predict different signatures in the CMB, both in temperature and polarization anisotropies and in spectral distortions. The Planck capability to distinguish among different scenarios through its sensitivity to T, TE, and E mode angular power spectra is discussed. Perspectives open by future high sensitivity experiments on the CMB polarization anisotropy and spectrum are also presented.
astro-ph
astro-ph
JENAM 2004 -- The many scales in the Universe Cosmological reionization after WMAP: perspectives from PLANCK and future CMB missions C. Burigana1, L.A. Popa1,2, F. Finelli1, R. Salvaterra3, G. De Zotti4,3, and N. Mandolesi1 1INAF-CNR/IASF, Sezione di Bologna, via Gobetti 101, I-40129 Bologna (Italy) 2Institute for Space Sciences, Bucharest-Magurele R-76900 (Romania) 3SISSA, via Beirut 4, I-34014 Trieste (Italy) 4INAF - Osservatorio Astronomico di Padova, vicolo dell'Osservatorio 5, I-35122 Padova (Italy) Abstract The WMAP first year detection of a high redshift reionization through its imprints on CMB anisotropy T and TE mode angular power spectra calls for a better comprehension of the uni- verse ionization and thermal history after the standard recombination. Different reionization mechanisms predict different signatures in the CMB, both in temperature and polarization anisotropies and in spectral distortions. The Planck capability to distinguish among different scenarios through its sensitivity to T, TE, and E mode angular power spectra is discussed. Per- spectives open by future high sensitivity experiments on the CMB polarization anisotropy and spectrum are also presented. 1 Introduction The accurate understanding of the ionization history of the universe plays a fundamental role in modern cosmology. The classical theory of hydrogen recombination for pure baryonic cosmolog- 1 ical models [1, 2], subsequently extended to non-baryonic dark matter models [3, 4, 5, 34] can be modified in various ways to take into account additional sources of photon and energy production able to significantly increase the ionization fraction, xe, above the residual fraction (∼ 10−3) after the standard recombination epoch at zrec ≃ 103. These photon and energy production processes may leave imprints on the cosmic microwave background (CMB) providing a crucial "integrated" information on the so-called dark and dawn ages, i.e. the epochs before or at the beginning the formation of first stars and primeval galaxies, complementary to those obtained by the study of the diffuse backgrounds in other frequency bands and that are impossible or difficult to study with other direct astronomical observations. Among the extraordinary results recently achieved by WMAP [7] the detection of a cosmo- logical reionization at relevant redshifts [8] represents one of the new most remarkable discovery in the recent years. In spite of the uncertainty associated to the degeneracy [9] between the Thomson scattering optical depth, τ , and the spectral index of primordial perturbation, ns, in CMB anisotropy data, a cosmological reionization at substantial redshifts is supported by the decrease of the temperature angular power spectrum at high multipoles and by an excess in the TE cross-power spectrum on large angular scale (multipoles ℓ < ∼ 7) with respect to models with no-reionization or reionization with quite low (∼ 0.05) values of τ . According to WMAP 1-yr data, the favourite values of τ are > ∼ 0.1, with a current best fit of ≃ 0.17 based on the com- bination of WMAP with additional CMB anisotropy data at higher resolution and other kinds of astronomical observations (see, e.g., [10]). On the other hand, the details and the physical explanation of the cosmological reionization are still unclear. ∼ z(2) Since the WMAP detection of the reionization at substantial redshifts, many models have been proposed and/or reconsidered to account for the measure of τ (see, e.g., [11]). Certainly, a significant contribution (≃ 0.05 − 0.07) to the value of τ derives from the ionization, likely associated to photon production and hot gas ejection by primeval galaxies and quasars, at relatively low redshifts (z < reion ≃ 5− 7) where direct observations of the Gunn-Peterson effect in quasars probe a high ionization level of the intergalactic medium. The remaining contribution (≃ 0.05 − 0.12) to the observed value of τ should be provided by processes at higher redshifts. Although uncertain because of the possible non-uniformity of the ionization of the intergalactic medium, the indication of an increase of the Ly-α opacity for Gunn-Peterson tests toward some quasars at z ∼ 6 [12, 13, 14] combined to the large WMAP value of τ suggests the possibility of a twice reionization, the first one at relevant redshifts, z(1) reion ≃ 15, possibly associated to Population III stars [15], or even at higher redshifts, zreion ∼ some × 10, in models involving a relevant photon production by particle decay, matter-antimatter annihilation, or black-hole > evaporation and so on, followed by the subsequent reionization at z ≃ z(2) reion ≃ 6. The reionization imprints on the CMB can be divided in three categories: i) decreasing of the power of CMB temperature anisotropy at large multipoles because of photon diffusion, ii) increasing of the power of CMB polarization and temperature-polarization cross correlation anisotropy at all multipoles with relevant features possibly more remarkable at low and middle multipoles according to the reionization epoch because of the delay of the effective last scattering surface, iii) generation of CMB Comptonization and free-free spectral distortions associated to the heating of the electron temperature of the intergalactic medium during the reionization 2 epoch. The imprints on CMB anisotropies are mainly dependent on the ionization history while CMB spectral distortions strongly depend also on the thermal history. 2 Observational perspectives from next and future experiments The CMB anisotropy pattern is a single realization of a stochastic process and therefore it may be different from the average over the ensemble of all possible realizations of the given (true) cosmological model with given parameters. This translates into the fact that the aℓm coefficients are random variables (possibly following a Gaussian distribution), at a given ℓ, and therefore their variance, Cℓ, is χ2 distributed with 2ℓ + 1 degrees of freedom. The relative variance δCℓ on Cℓ is equal to p2/(2ℓ + 1) and is quite relevant at low ℓs. This is the so-called "cosmic variance" which limits the accuracy of the comparison of observations with theoretical predictions. Another similar variance in CMB anisotropy experiments is related to the sky coverage since the detailed CMB anisotropy statistical properties may depend on the considered sky patch. This variance depends on the observed sky fraction, fsky. At multipoles larger than f ew×102 the most relevant uncertainties are related to the experiment resolution and sensitivity. All these terms contribute to the final uncertainty on the Cℓ according to [16]: δCℓ Cℓ =s 2 fsky(2ℓ + 1)(cid:20)1 + Aσ2 N CℓWℓ(cid:21) , (1) where A is the size of the surveyed area, σ is the rms noise per pixel, N is the total number of observed pixel, and Wℓ is the beam window function. For a symmetric Gaussian beam Wℓ = exp(−ℓ(ℓ + 1)σ2 B) where σB = FWHM/√8ln2 defines the beam resolution. The two instruments at cryogenic temperatures on-board the ESA Planck satellite [17] (see also J. Tauber 2004, this Meeting), the Low Frequency Instrument (LFI; [18]; see also N. Mandolesi 2004, this Meeting) based on differential radiometers and the High Frequency Instrument (HFI; [19]) based on state-of-art bolometers at the focus of a 1.5 m aperture off-axis Gregorian telescope, will measure the CMB angular power spectrum with very high sensitivity up to multipoles ℓ ∼ 1000 − 2000 and an accurate control of the systematic effects. The two instruments will cover a frequency range from 30 to 857 GHz, necessary to accurately subtract the foreground contamination (see, e.g., G. De Zotti et al. 2004, this Meeting), with a (FWHM) resolution from ∼ 33′ to 5′ and a sensitivity from ∼ 15µK to the ∼ µK level in terms of antenna temperature on a square resolution element of with side of ∼ 10′. Fig. 1 compares Planck and WMAP final performances in terms of angular power spectrum recovery. Note that, at similar frequencies, the better sensitivity of Planck with respect to WMAP is mainly due to the lower level of the Planck instrumental noise achieved thanks to the better radiometer performance assured by the lower system temperature while at higher frequencies it derives from a combination of lower instrumental noise achieved thanks to high sensitivity bolometers and from the improving of resolution with the frequency for a given telescope size. The CMB angular power spectrum is reported without beam smoothing and by taking into account the beam window functions of several Planck frequency channels and of 3 Figure 1: Comparison between Planck and WMAP resolution and sensitivity. For each con- sidered frequency channel, the crossing between the CMB convolved angular power spectrum and the residual instrumental white noise angular power spectrum indicates the multipole value where the signal to noise ratio (ℓ by ℓ) is close to unity. Of course, binning the power spectrum on a suitable range of multipoles, as possible because of the smooth variation of Cℓ with ℓ, will allow to recover the CMB power spectrum with a good accuracy also at multipoles comparable that those corresponding to the crossings between the noise and CMB power spectra reported in the figure (for example for the Planck channels at 70 -- 100 GHz at ℓ ∼ 1500 a ∼ 3% binning over ℓ allows to reduce the uncertainty on the Cℓ recovery to ∼ 15%). 4 the highest WMAP frequency channel (which resolution is very close to that of the LFI 70 GHz channel). The corresponding angular power spectra of the residual nominal white noise (i.e. after the subtraction of the expectation value of the noise angular power spectrum) are also displayed. Similar considerations hold also for the measure of the polarization anisotropies (see, e.g., [20] for a summary of CMB polarization experiments). The sensitivity to the anisotropy measure of the Stokes parameters Q and U and of the polarization signal P (P =pQ2 + U 2) is √2 − 2 times worse than that for the temperature anisotropy, the exact value depending on ∼ to the detailed experimental strategy adopted to recover them from the combination of multi- beam data 1. The low (of few -- some %) polarization level predicted, and still only approximately measured [21], for the CMB anisotropy clearly calls for high sensitivity measurements which are less crucial for the temperature-polarization cross correlation (TE-mode) already measured by WMAP. A quite accurate measure of the E-mode polarization is expected from the next WMAP data release, at least at degree scales, while the Planck instruments will have a good sensitivity to the polarization E-mode angular power spectrum also at scales of some arcminutes, the main limitation being represented by the foreground contamination, as shown in Fig. 2. Thanks to the combination of temperature and polarization high quality data, Planck is expected to reduce the error bars in the recovery of the cosmological parameters at the level of f ew % or better and the final accuracy on a wide set of cosmological parameters will be largely independent [25] of the auxiliary information coming from other classes of astronomical observa- tions which will be necessary to just remove some degeneracies intrinsic in the CMB information [9]. As an example, Fig. 3 compares the Planck sensitivity in constraining the Thomson optical depth and the density contrast with the sensitivity of WMAP, possibly combined with the Ly-α forest information (from [10]). The Planck sensitivity and resolution will allow not only an extremely precise determination of the CMB angular power spectrum but also an accurate and multi-frequency imaging of the temperature anisotropy pattern on the whole sky and a quite accurate imaging of the polarization anisotropy pattern at least on some sky areas of particular sensitivity 2 by combining the multi- frequency information. This is required for the study of the high order statistical properties of the CMB anisotropy at high resolution, crucial to test some cosmological scenarios predicting localized features (topological defects, some inflationary models, ...). The detection of the polarization B-mode angular power spectrum is within the Planck capabilities at least over some suitable ranges of multipoles, but its accurate and exhaustive measure requires a new generation of dedicated experiments. As well known, the details of the reionization history strongly affect also the amplitude of the polarization B-mode angular power 1For example, a ratio √2 can be easily obtained by taking into account that for differential radiometers the information from four receivers coupled to two feeds can be combined to derive a measure of a single temperature anisotropy data or of two (Q and U ) polarization anisotropy data. 2Analogously to the WMAP survey, the scanning strategies foreseen for Planck will imply that the sky pixels on two areas (each of about 20 -- 30 squared degrees) close to the ecliptic poles will be observed for a time significantly longer than the average resulting into a sensitivity significantly better (by about 5 times) than the average. 5 Figure 2: Typical predicted ranges for the CMB polarization angular power spectra (E-mode, upper dotted regions; B-mode, lower dotted regions) including the lensing [22] compared to the foreground contamination (radiogalaxies (dots) contribution (see [23] for a recent detailed study); the small contribution expected from infra-red galaxies (dashes); Galactic dust (three dots-dashes) and synchrotron (dot-dashes) emission [24, 20]) at two representative multipoles, as function of the frequency. We report the final sensitivity of WMAP at 94 GHz (asterisk) and of some Planck frequency channels (crosses) binned over a quite large range of multipoles. spectrum, which depends on the amplitude of tensorial (gravitational waves) modes and, at small scales, on the lensing effect. The cosmic variance uncertainty and the Galactic foreground contamination, quite relevant at low multipoles, decreases at middle and high multipoles. The atmospheric emission is expected to be less crucial for CMB polarization measurements than for anisotropy and spectrum ones. Therefore, ground-based observations ([26]; see, e.g., the VSA project 3) appear of particular interest for obtaining precise information at intermediate and high multipoles, thanks to the high sensitivity that could be reached with very long integration times and/or the use of a very large number of receivers. On the other hand, the space is the favourite site for the accurate measure of the CMB polarization at low and intermediate multipoles (see, e.g., the NASA Inflation Probe 4 of the Beyond Einstein program). Fig. 4 reports the uncertainty (including cosmic variance and instrument sensitivity and resolution) in the recovery of the CMB polarization power spectra as achievable with a ∼ yr full sky space mission discussed in the context of the call of mission themes of the ESA Cosmic Vision 2015-2025, by using the state of the art radiometers (right panel) and bolometers (left panel) at the focus of a set of four ∼ 60 cm telescopes. The comparison with Fig. 5 shows that, as long as foreground polarization anisotropies (and, 3http://www.mrao.cam.ac.uk/telescopes/vsa/index.html 4http://universe.nasa.gov/program/inflation.html 6 Figure 3: Planck sensitivity (filled square) in constraining the Thomson optical depth and the density contrast σ8 compared with the sensitivity of WMAP 1-yr data (triangle), possibly combined with the Ly-α forest information (filled circle), and of WMAP 4-yr data (square). From [10]. in particular, the Galactic contamination) are accurately known and subtracted, the sensitivity levels presented above will allow to accurately measure also the B-mode or, depending on its intrinsic level, to set significant constraints on it. The possibility to significantly improve our knowledge of the CMB spectrum with respect to the current observational status, in which the COBE/FIRAS data play the major role in constraining the CMB spectral distortions, has been recently addressed. In particular, two space mission proposals, the DIMES experiment [27], designed to reach an accuracy close to that of FIRAS but at centimeter wavelengths, and the FIRAS II experiment [28] which will allow a sensitivity improvement by a factor ∼ 100 with respect to FIRAS, open the new perspective to detect CMB spectral distortion and not only to set constraints on them. DIMES [27] is a space mission submitted to the NASA in 1995, designed to measure very accurately the CMB spectrum at wavelengths in the range ≃ 0.33 − 15 cm [27]. DIMES will compare the spectrum of each ≃ 10 degree pixel on the sky to a precisely known blackbody to precision of ∼ 0.1 mK, close to that of FIRAS. The set of receivers is given from cryogenic radiometers with instrument emission cooled to 2.7 K operating at six frequency bands about 2, 4, 6, 10, 30 and 90 GHz using a single external blackbody calibration target common to all channels to minimize the calibration uncertainty. The DIMES design is driven by the need to reduce or eliminate systematic errors from instrumental artifacts. The DIMES sensitivity represents an improvement by a factor better than 300 with respect to previous measurements at centimeter wavelengths allowing a significant improvement of our knowledge on early dissipation processes and free-free distortions [29]. 7 Figure 4: Residual uncertainty in the knowledge of the polarization angular power spectrum (in terms of δT = (ℓ(ℓ+1)Cℓ/2π)0.5; thermodynamic temperatures are considered) including cosmic variance and detector sensitivity and resolution as in principle achievable by the next generation of CMB space missions. Right panel: the case of radiometer technology (typical sensitivity of ≃ 0.1 − 0.25µK on a pixel of 30′ side). Left panel: the case of bolometer technology (typical sensitivity of ≃ 0.04 − 0.12µK on a pixel of 30′ side). In each panel, the different lines refer to frequency channels at 32, 64, 94, 143, 217 GHz; at increasing frequency the error at high multipoles decreases because of the resolution increasing (from instance 1.09◦ to 0.16◦ (FWHM)) by using a given telescopes size. The cosmic variance has been computed by assuming, for simplicity, a constant polarization anisotropy level of 1, 0.1, and 0.01 µK in terms of δT (lines from the top to the bottom at low multipoles). Full sky coverage is assumed. Fixsen and Mather [28] described the fundamental guidelines to significantly improve CMB spectrum measures at λ < ∼ 1 cm. A great reduction of the residual noise of cosmic rays, dom- inating the noise of the FIRAS instrument, can be obtained by eliminating the data on-board co-add process or applying deglitching before co-adding, by reducing the size of the detectors and by using "spiderweb" bolometers or antenna-coupled micro-bolometers. They are expected to show a very low noise when cooled below 1 K while RuO sensors can reduce to 0.1 mK the read noise of thermometers. The Lagrangian point L2 of the Earth-Sun system is, of course, the favorite "site" for FIRAS II. Also, the calibration can be improved by order of magnitudes with respect to that of FIRAS by reducing the contribution to the calibrator reflectance of light from the diffraction at the junction between the calibrator and the horn. A complete symmetrical construction of the instrument is recommended. This allows the cross-check between calibrators and between calibrators and the sky and to realize "an end-to-end calibration and performance test before launch". According to the authors, FIRAS II can be designed to have a frequency coverage from 60 to 3600 GHz (i.e. from 5 mm to 83 µm) with a spectral resolution ν/∆ν < 200 8 Figure 5: CMB polarization angular power spectra (E-mode: black solid line; B-mode: black dashes; B-mode from lensing: black long dashes) compared with reasonable predictions for the overall (Galactic plus extragalatic) foreground polarized angular power spectrum at 30 (green line), 100 (blue line) and 217 GHz (red line). The few recent upper limits and measures are reported for comparison [31, 32, 33, 21]. and sensitivity in each channel about 100 times better than that of FIRAS. The combination of these two spectrum experiments open, at least in principle, the perspec- tives to detect and possibly measure spectral distortions imprinted by energy dissipations about 100 times smaller than those corresponding to the current upper limits set by FIRAS [30]. 3 Reionization phenomenological models The reionization process can be described phenomenologically in terms of injection of additional ionizing photons [34, 35, 36]. The ionization fraction of matter, xe = ne/n, can be obtained from the balance between the processes of recombination and ionization: dxe dt = −αrec(T )nbx2 e + εi(z)(1 − xe)H(z), (2) where nb(z) is and the mean baryonic density at the redshift z, T is the temperature of the plasma, αrec(T ) ≃ 4 × 10−13(cid:0)T /104K(cid:1)−0.6 s−1cm−3 is the recombination coefficient, and εi(z) is the efficiency of the ionizing photon production giving the ionizing photon production rate 9 dni/dt = εi(z)nb(z)H(z) ; H(z) = 1/texp, where texp = a/(da/dt) is the cosmic expansion time, a being here the cosmic scale factor. The choice of the function εi(z) allows to model the ionization history also in the presence of extra sources of ionizing photons (e.g. Ly-α and ionization photons from primavel stars, galaxies, and active galaxies, or from primordial black hole decays, electromagnetic cascades from particle release from topological defects or decay of super heavy dark matter, ...). Assuming equilibrium between the recombination and the ionization process, for any given history of the plasma temperature the evolution of the ionization fraction is given by the solution of the simple second order equation obtained by the equality dxe/dt = 0. 3.1 Late processes For late processes, a simple expression describing a first reionization at relevant redshifts followed by a second reionization, possibly mimicking the model by Cen [15], has been proposed [37] in the form: εi(z) = ε0 exp"− reion)2 (z − z(1) (∆z1)2 # + ε1(1 + z)−mΘ(z(1) reion − z) ; (3) reion, and ∆z1 ≪ z(1) here ε0, z(1) reion are free parameters describing the history of the first epoch of reionization which significantly decreases at z > z(1) reion are free parameters describing the history of the second epoch of reionization resulting into a increasing of εi(z) with the time, being Θ(x) the step function. reion; ε1, m, and (again) z(1) Although quite weakly in this modelization, because of the dependence on the temperature as a power of ≃ −0.6 of the recombination coefficient, the thermal history influences the ionization fraction evolution. By using the matter temperature evolution T (z) ≃ 270 (1 + z/100)2 K and modifying the ionization history in the cmbfast code, the CMB angular power spectrum, in both temperature and polarization, has been computed [37] for some representative choices of the above free parameters. As shown by the authors for the case of the E-mode polarization power spectrum, the differences between different late reionization scenarios associated to very different parameters are of particular interest at relatively low multipoles. Provided that the contribution from the Galactic foregrounds, particularly relevant at these multipoles, could be accurately modeled, the difference between a scenario involving a single reionization at z ≃ z(2) reion and scenarios involving a twice reionization could be detected by the Planck polarization accuracy. The relevance of the detection of spectral distortions has been partially renewed [38] by the WMAP satellite discovery of a reionization phase at relevant redshifts. The computation of the CMB spectral distortions associated to the reionization parametric model presented here is strictly related to the detailed description of the thermal history. For these kinds of late processes, we assume here a matter thermal history similar to that reported by Cen [15]: a matter temperature T in approximate thermal equilibrium with the radiation field (i.e. equal to T0(1 + z)) at z > ∼ 27; a linear dependence of logT on z at subsequent 10 Figure 6: Contour plots of the optical depth and of the fractional injected energy associated to the Comptonization parameter as functions of β and m in the case of late reionization processes. Left panel: τ evaluated by integrating up to z = 0. Middle and right panels: ∆ǫ/ǫi ≃ 4u (in units of 10−6) computed by integrating up to z = 0 and up to z = 5, respectively. times to reach a temperature T (1) (assumed here 1.5×104 K for numerical estimates) at z = z(1) reion, kept then constant up to z = z(2) reion when it rapidly increases up to a temperature T (2) (assumed here 2 × 104 K for numerical estimates) kept then constant up to low redshifts. We have implemented this thermal history in the equilibrium evolution of the ionization fraction and then used these thermal and ionization histories to compute the Comptonization and free-free distortions as described in [39] by generalizing the code to include also the case of a ΛCDM model. In Fig. 6 we report our contour plot results for the optical depth τ and the fractional en- ergy injection associated to the Comptonization distortion as functions of the free parameters β = ε1/109 and m, by separately showing for ∆ǫ/ǫi the results of the integration up to z = 0 and up to z = 5 (of course, the difference between these results gives the fractional energy injected in the plasma at z ≤ 5). For simplicity, we report here the results for an interval of β and m producing values of τ close to the current WMAP best fit [8]. Although a detailed computation requires to take into account the dependence of the ionization history on the matter tempera- ture evolution, these results can be approximately rescaled to different values of T (1) and T (2) by considering the linear dependence of the Comptonization parameter on T . These predicted Comptonization distortions could detected and possibly measured by an experiment with a sen- 11 sitivity like that of FIRAS II. Analogously to the case of the polarization anisotropy signatures, the difference between the Comptonization distortions by models involving a twice reionization reion) and models with a single reionization at reion is comparable or above the FIRAS II sensitivity. The free-free distortion parameter (or a continuous reionization starting at z ≃ z(1) z ≃ z(2) yB results to be too small, well below the DIMES sensitivity, and is not reported here 5. 3.2 High redshift processes Models involving a substantial reionization at redshifts, z ≃ zreion, much higher than z(1) can not be excluded by current data. A Gaussian parametric form reion ∼ 15 εi(z) = ξ exp(cid:20)− (z − zreion)2 (∆z)2 (cid:21) , (4) quite similar to that assumed to describe the first reionization epoch for late processes, can be exploited in this context [37]; again ξ, zreion, and ∆z are free parameters describing the history of this high redshift reionization scenario. Assuming ∆z ≪ zreion implies the choice of a peak-like model. Again, it is necessary to add the evolution of the matter temperature which should peak at typical values of Tp ≈ (1 − 2) × 104 K. In [37] it is assumed an extremely short heating phase (δ-like) about zreion. After the end of the heating phase the matter temperature is determined by the usual equation for the electron temperature evolution, dominated, at high redshifts, by the Compton cooling term. Therefore a set of two differential equations, one for the ionization fraction and the other for the matter temperature, describe the problem 6. This modified ionization history has been included in the cmbfast code to computed the CMB angular power spectrum, in both temperature and polarization, for some representative choices of the above free parameters. As shown by the authors for the E-mode polarization angular power spectrum, the differences between different early reionization scenarios associated to different parameters and between them and a scenario involving a single late reionization at z ≃ z(2) reion are relevant at low and also at middle multipoles, ℓ ∼ f ew × (10 − 100). This is of particular interest in the light of the Planck polarization accuracy in the multipole region of the CMB acoustic peaks where the contribution from the Galactic foregrounds is expected to significantly decrease with respect to the level it has at low multipoles. For the study of the spectral distortions associated to early peak-like reionization processes, we have numerically integrated the differential equations for the ionization fraction and the matter temperature as described above. Differently from the δ-like assumption for the matter temperature during the heating phase about zreion (that is not particularly critical for polariza- 5On the contrary, note that several specific physical models predict free-free distortion levels larger than those found in this simple modelization and clearly observable by a DIMES-like experiments (see, e.g., [30] and references therein). 6Simple approximate analytical solution can be used in the limit ξ ≪ 1. We find also that the the system can be easily solved through a numerical integration. A simple backward differential scheme with a very small time step compared to the other relevant timescales does not introduce relevant errors in this context. 12 Figure 7: Contour plots of the optical depth, τ , (left panel) and of the fractional energy injection associated to the Comptonization parameter, ∆ǫ/ǫi ≃ 4u, (right panel) as functions of ξ and zreion in the case of early peak-like reionization processes (logarithmic scale). tion anisotropy considerations) we adopt here a Gaussian parametric form T (z) = Tp exp(cid:20)− (z − zreion)2 (∆z)2 (cid:21) (5) for the matter temperature during the heating phase (within a redshift interval ≃ ±3∆z about zreion) and then assume the usual temperature evolution equation, dominated by the Compton cooling term, at later epochs (Tp = 1.5 × 104 K and ∆z = 0.025zreion are adopted here for It is in fact quite reasonable to assume a similar time dependence for numerical estimates). both the efficiency of the ionizing photon production, εi(z), and the matter temperature during the active phase. We assume as initial conditions at the beginning of the heating/ionization phase the thermal equilibrium between matter and radiation and the residual ionization fraction obtained at the considered redshift from the standard recombination. Finally, we have implemented the thermal and ionization history as described above in the code [39] for the computation of the Comptonization and free-free distortions. In Fig. 7 we report our results in terms of contour plots for the optical depth τ and the fractional energy injection associated to the Comptonization distortion as functions of the free parameters ξ and zreion. These results can be rescaled to different choices of Tp and ∆z by using the approximate proportionality relations between τ and ∆z and between u and Tp∆z 7. Fig. 7 shows that a large region of the ξ and zreion parameter space can be clearly ruled out 7In reality, for the lowest considered values of ξ (ξ ≈ 1) we find an increasing of the Comptonization distortion with ∆z larger by a factor ≈ 1.5 − 3 than that suggested by the above proportionality and related to the different initial conditions (the process starts at earlier times) and to the different coupling between the ionization fraction and the matter temperature. We find that this holds also for the free-free distortion parameter yB. 13 7000 6000 5000 4000 3000 2000 1000 0 1 7000 6000 5000 4000 3000 2000 1000 0 10 2 10 3 10 1 10 2 10 3 10 Figure 8: The CMB anisotropy power spectra for different neutrino masses considering (bottom panel) or not (top panel) the inhomogeneities of the reionization process. by current WMAP data because of the violation of the limits on the optical depth. It is also remarkable that in the permitted region of ξ and zreion (corresponding to τ ∼ 0.1) the model predicts values of fractional injected energy ∆ǫ/ǫi ≃ 4u ∼ 10−6, again in principle measurable with the FIRAS II sensitivity. We find again values of free-free distortion parameter yB too small with respect to the DIMES sensitivity 8. To summarize, since it is plausible to assume peak matter temperatures > ∼ (1 − 2) × 104 K to have an efficient (late or early) reionization, values of ∆ǫ/ǫi ≃ 4u ∼ (1 − 2) × 10−6 should be considered as typical (lower limit) predictions for the Comptonization distortion associated to reionization scenarios compatible with the WMAP results [40]. 4 Inhomogeneous reionization in the presence of massive neu- trinos As a specific example, we have investigated the role of a HDM component in the form of three massive neutrino flavors in the context of reionization scenarios. Assuming a flat background cosmology described by the best fit power law ΛCDM model with WMAP data (Ωbh2 = 0.024, 8By considering higher matter temperature values (up to 6× 104 K), we find only a weak dependence of yB on Tp because the typical decreasing of yB with Tp is approximately compensated by the increasing of the ionization fraction because of the decreasing of the recombination efficiency. 14 Figure 9: CMB temperature (left panel) and polarization (E-mode, right panel) anisotropy power spectra obtained for the above reionization model (including inhomogeneities) by considering or not the neutrino contribution for a specific value (25) of the reionization redshift. Note the sensitivity of polarization to the neutrino mass. Ωmh2 = 0.14, h = 0.72), we analyze the role of the neutrino mass for the properties of the gas in the intergalactic medium (IGM). We find that the temporal evolution of the hydrogen and helium ionization fractions is sensitive to the neutrino mass, with relevant implications for the CMB anisotropy and polarization angular power spectra. The reionization is assumed to be caused by the ionizing photons produced in star-forming galaxies and quasars. This process depends on the evolution of the background density field that determines the formation rate of the bounded objects, the gas properties in the IGM and their feedback relation. At the redshift of collapse, the fraction of the mass of the gas in the virilized halos can be obtained if the probability distribution function of the gas overdensity is known. The temperature-density relation and the virial mass-temperature relation determines the connection between the gas density and the matter density at the corresponding scales [10]. During and after recombination, neutrinos with masses in eV range can have significant interactions with photons, baryons and cold dark matter particles via gravity only. The neutrino phase-space density is constrained by the Tremaine-Gunn criterion [41] that puts limits on the neutrino energy-density inside bounded objects. Neutrinos cannot cluster via gravitational instability on scales below 15 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Figure 10: The probability density function of the neutrino masses (69% CL) as could be estimated by the Planck mission, using the CMB polarization measurements, for different reionization redshifts (fiducial model with mν = 0.23 eV). the free-streaming scale. The free-streaming distance determines the neutrino energy-density distribution inside the gravitationally collapsed objects [42], the temperature-density relation of the gas in the IGM and the hydrogen and helium ionization fractions. In Fig. 8 we show the WMAP data compared with the CMB anisotropy power spectra obtained for different neutrino masses with (bottom) and without (top) considering the inho- mogeneities of the reionization process. Fig. 9 compares the CMB temperature and polarization (E-mode) anisotropy power spectra obtained for the above reionization model (including inhomo- geneities) by considering or not the neutrino contribution for a specific value of the reionization redshift. Clearly, polarization information greatly helps to distinguish between the two cases. Finally, in Fig. 10 we show how in this specific reionization model the neutrino mass could in principle be constrained by Planck depending on the assumed reionization redshift. Note the improvement of the sensitivity to neutrino mass with the decreasing of the reionization redshift, as intuitive is because of the corresponding increase of the neutrino mass role. 5 Conclusion The WMAP detection of a high redshift reionization calls for a better understanding of the physics of the reionization process. Different models predicts different imprints on the CMB in both temperature and polarization anisotropies while the presence of late spectral distortions (Comptonization like, with ∆ǫ/ǫi ∼ some × 10−6, and also free-free distortions with amplitudes 16 strongly dependent on the specific considered models) seems unavoidable for reasonable thermal histories in the context of reionization scenarios compatible with WMAP results. Accurate measures of the CMB properties, such as those expected by the forthcoming Planck satellite and by future experiments will offer the opportunity to constrain the ion- ization and thermal history of the universe at moderate and high redshifts. Acknowledgements We thank L. Danese, L. La Porta, and L. Toffolatti for collaborations and discussions. We also thank L.-Y. Chiang, B. Ciardi, A. Ferrara, and P. Naselsky for useful conversations on the reionization history. The use of the cmbfast code is acknowledged. Some of the results in this paper have been derived applying the HEALPix 9 (Hierarchical Equal Area and IsoLatitude Pixelization of the Sphere) by G`orski et al. [43]) to the WMAP 1-yr data products. Some of the calculations presented here have been carried out on an alpha digital unix machine at the IFP/CNR in Milano by using some NAG integration codes. The staff of the LFI DPC in Trieste, where some simulations have been carried out, is acknowledged. References [1] P.J.E. Peebles, 1968, ApJ, 153, 1. [2] Ya. B. Zel'dovich, V. Kurt, R.A. Sunyaev, 1968, Zh. Eksp. Theor. Phys., 55, 278. [3] N.A. Zabotin and P.D. Naselsky, 1982, Sov. Astron., 26, 272. [4] B.J.T. Jones and R. Wyse, 1985, A&A, 149, 144. [5] S. Seager, D.D. Sasselov, D. Scott, 2000, ApJS, 128, 407. [6] P.J.E. Peebles, S. Seager, W. Hu, 2000, ApJ, 539, L1. [7] C.L. Bennett et al., 2003, ApJS, 148, 1. [8] A. Kogut et al., 2003, ApJS, 148, 161. [9] J.R. Bond, R. Crittenden, R.L. Davis, G. Efstathiou, P.J. Steinhardt, 1994, PRL, 72, 13; J.R. Bond, R.L. Davis, P.J. Steinhardt, 1995, Astroph. Lett. and Comm., 32, 53; G. Efstathiou and J.R. Bond, 1999, MNRAS, 304, 75. [10] L.A. Popa, C. Burigana, N. Mandolesi, 2004, NA, 9, 189. [11] R. Bean, A. Melchiorri, J. Silk, 2003, Phys. Rev. D 68, 083501. [12] R.H. Becker et al., 2001, AJ, 122, 2850. 9http://www.eso.org/science/healpix/ 17 [13] X. Fan et al., 2002, AJ, 123, 1247. [14] X. Fan et al., 2004, AJ, in press, astro-ph/0405138. [15] R. Cen, 2003, ApJ, 591, 12. [16] L. Knox, 1995, Phys. Rev. D 48, 3502. [17] J.A. Tauber, 2000, in The Extragalactic Infrared Background and its Cosmological Implica- tions, Proc. of the IAU Symposium, Vol. 204, eds. M. Harwit and M. Hauser. [18] N. Mandolesi et al., 1998, Planck LFI, A Proposal Submitted to the ESA. [19] J.L. Puget et al., 1998, HFI for the Planck Mission, A Proposal Submitted to the ESA. [20] G. De Zotti, 2002, in Astrophysical Polarized Background, eds. S. Cecchini et al., AIP Conf. Proc. 609, pg. 295. [21] J. Kovac et al., 2002, Nature, 420, 772. [22] M. Zaldarriaga and U. Seljak, 1998, Phys. Rev. D 58, 023003. [23] M. Tucci, E. Martinez-Gonzalez, L. Toffolatti, J. Gonzalez-Nuevo, G. De Zotti, 2004, MN- RAS, 349, 1267. [24] C. Burigana and L. La Porta, 2002, in Astrophysical Polarized Background, eds. S. Cecchini et al., AIP Conf. Proc. 609, pg. 54. [25] L.A. Popa, C. Burigana, N. Mandolesi, 2001, ApJ, 558, 10. [26] M. Hobson, High-resolution CMB interferometry, Talk at The Future of CMB, Meeting at TAC, Copenhagen (NL), November 2003. [27] A. Kogut, 1996, Diffuse Microwave Emission Survey, in XVI Moriond Astrophysics meeting Microwave Background Anisotropies, March 16-23, Les Arcs, France, astro-ph/9607100. [28] D.J. Fixsen and J.C. Mather, 2002, ApJ, 581, 817. [29] C. Burigana and R. Salvaterra, 2003, MNRAS, 342, 543. [30] C. Burigana and R. Salvaterra, 2003, astro-ph/0309509. [31] C.W. O'Dell, B.G. Keating, A. de Oliveira-Costa, M. Tegmark, P.T. Timbie, 2003, Phys. Rev. D. 68, 42002. [32] M.M. Hedman et al., 2002, ApJ, 573, L73. [33] P.C. Farese et al., 2004, ApJ, 610, 625. [34] P.J.E. Peebles, S. Seager, W. Hu, 2000, ApJ, 539, L1. 18 [35] A.G. Doroshkevich, P.D. Naselsky, 2002, Phys. Rev. D 65, 13517. [36] A.G. Doroshkevich, I.P. Naselsky, P.D. Naselsky, I.D. Novikov, 2003, ApJ, 586, 709. [37] P. Naselsky and L.-Y. Chiang, 2004, MNRAS, 347, 795. [38] A. Kogut, 2003, in the Proc. of The Cosmic Microwave Background and its Polarization, New Astronomy Reviews, eds. S. Hanany and K.A. Olive, Volume 47, Issues 11-12, pg. 945. [39] C. Burigana, G. De Zotti, L. Danese, 1995, A&A, 303, 323. [40] C. Burigana, F. Finelli, R. Salvaterra, L.A. Popa, N. Mandolesi, 2004, in Research Signspot, Recent. Res. Devel. Astronomy & Astrophys., Vol. 2, 59. [41] S. Tremaine and J.E. Gunn, 1979, PRL, 42, 407. [42] L.A. Popa, C. Burigana, N. Mandolesi, 2002, ApJ, 580, 16. [43] K.M. G´orski, E. Hivon, B.D. Wandelt, 1999, in Proc. of the MPA/ESO Conference on from Recombination to Garching, eds. A.J. Banday, Evolution of Large-Scale Structure: R.K. Sheth, L. Da Costa, pg. 37, astro-ph/9812350. 19
astro-ph/9901371
1
9901
1999-01-26T20:28:37
X-ray Variability from the Compact Source in the Supernova Remnant RCW 103
[ "astro-ph" ]
A new ASCA observation of 1E 161348-5055, the central compact X-ray source in the supernova remnant RCW 103, reveals an order-of-magnitude decrease in its 3 - 10 keV flux since the previous ASCA measurement four years earlier. This result is hard to reconcile with suggestions that the bulk of the emission is simple quasi-blackbody, cooling radiation from an isolated neutron star. Furthermore, archived EINSTEIN and ROSAT datasets spanning 18 years confirm that this source manifests long-term variability, to a lesser degree. This provides a natural explanation for difficulties encountered in reproducing the original EINSTEIN detection of 1E 161348-5055. Spectra from the new data are consistent with no significant spectral change despite the decline in luminosity. We find no evidence for a pulsed component in any of the data sets, with a best upper limit on the pulsed modulation of 13 percent. We discuss the phenomenology of this remarkable source.
astro-ph
astro-ph
DRAFT OF JAN 19 1999 Preprint typeset using LATEX style emulateapj X-RAY VARIABILITY FROM THE COMPACT SOURCE IN THE SUPERNOVA REMNANT RCW 103 E. V. GOTTHELF1, R. PETRE2 & G. VASISHT3 Draft of Jan 19 1999 ABSTRACT A new ASCA observation of 1E 161348- 5055, the central compact X-ray source in the supernova remnant RCW 103, reveals an order-of-magnitude decrease in its 3 - 10 keV flux since the previous ASCA measurement four years earlier. This result is hard to reconcile with suggestions that the bulk of the emission is simple quasi- blackbody, cooling radiation from an isolated neutron star. Furthermore, archived Einstein and ROSAT datasets spanning 18 years confirm that this source manifests long-term variability, to a lesser degree. This provides a natural explanation for difficulties encountered in reproducing the original Einstein detection of 1E 161348- 5055. Spectra from the new data are consistent with no significant spectral change despite the decline in luminosity. We find no evidence for a pulsed component in any of the data sets, with a best upper limit on the pulsed modulation of 13 percent. We discuss the phenomenology of this remarkable source. Subject headings: stars: individual (1E 161348- 5055) -- stars: neutron -- supernova remnants: individual (RCW 103) -- X-rays: stars 1. INTRODUCTION Recent spectro-imaging X-ray observations of central com- pact sources in supernova remnants (SNRs) challenge earlier notions that most young neutron stars (NSs) evolve in a man- ner similar to the prototypical Crab pulsar (Gotthelf 1998). In fact, the latest compilations shows that most of such associated objects manifest properties distinct from those of the Crab-like systems. Based on observational grounds alone, three classes of NSs in SNRs are known whose flux is dominated by their X-ray emission; these include the X-ray pulsars with anomalously slow rotation (with periods in the range of 5 - 12 s) and steep ∼ 3) power-law spectra (Gregory & Fahlman 1980; Gotthelf (Γ > & Vasisht 1998 and refs. therein), the soft gamma-ray repeaters (Cline et al. 1982; Kulkarni et al. 1994; Vasisht et al. 1994), and a population of radio-quiet NSs in remnants (Caraveo et al. 1996; Petre et al. 1996; Mereghetti et al. 1996). The above ob- jects are linked by their apparent radio-quiet nature, and taken collectively, may help further reconcile the NS birth rate with the observed SNR census. In this study we focus on the enig- matic X-ray source 1E 161348- 5055 in the SNR RCW 103, for which no clear interpretation yet exists within the above taxo- nomic framework. The Einstein X-ray source 1E 161348- 5055 lies near the pro- jected center of the bright, young (∼ 2 × 103 yrs; Carter et al. 1997) Galactic shell-type SNR RCW 103 (G332.4-0.4) and has been proposed as the first example of an isolated, cooling NS (Tuohy & Garmire 1980). It was discovered using the high res- olution imager (HRI) but went unseen by a prior Einstein IPC observation and a subsequent ROSAT PSPC one, supposedly due to the poorer spatial resolution of these instruments. Sur- prisingly, an initial observation with the ROSAT HRI also failed to detect the source; this was attributed to the reduced HRI sen- sitivity of the 10' off-axis pointing (Becker 1993). Finally, a 1993 ASCA observation re-discovered this elusive object (Got- thelf, Petre & Hwang 1997, GPH97 herein), but its spectral characteristics were found to be incompatible with a simple cooling NS model. This re-detection has been confirmed by more recent, on-axis, ROSAT HRI observations. Herein, we present the results of our follow-up (Sep 1997) ASCA observation of 1E 161348- 5055. In the same field lies the recently discovered 69 ms pulsar AX J161730-505505 (Torii et al. 1998), whose analysis is presented separately (Torii et al. 1999). While both sources are detected again, the flux from 1E 161348- 5055 has declined significantly since the pre- vious ASCA measurement. We discuss some implications of this large flux variability on the nature of 1E 161348- 5055. 2. OBSERVATIONS AND ANALYSIS A day-long follow-up observation with the ASCA Observa- tory (Tanaka et al. 1994) of RCW 103 was carried out on 1997 September 4. Data were acquired with both the solid-state (SISs) and gas scintillation spectrometers (GISs). The essen- tial properties of these instruments are qualified in GPH97. The SIS data were acquired in 1-CCD BRIGHT mode with 1E 161348- 5055 placed as close to the mean SIS telescope bore-sight as was practical, to minimize vignetting losses. The GIS data were collected in the highest time-resolution mode (0.5 ms or 0.064 ms, depending on the telemetry mode), with reduced spectral binning of ∼ 12 eV per PHA channel. The ef- fective, filtered observation time is 58(49) ks for each GIS(SIS) sensor. The new data were reduced and analyzed with the same methodology as in GPH97. 3. RESULTS We compared images of RCW 103 from our new observation with the ASCA images obtained four years earlier. The flux- corrected GIS images from the two epochs, restricted to the hard energy band-pass (3 - 10 keV), are displayed in figure 1 using an identical intensity scale. Both reveal a pair of distinct features, each having a spatial distribution consistent with that of a point source; one at the position of 1E 161348- 5055, and the other at the position of the 69 ms X-ray pulsar AX J161730- 1Columbia Astrophysics Laboratory, Columbia University, 550 West 120th Street, New York, NY 10027, USA; [email protected] 2NASA/Goddard Space Flight Center, Code 662, Greenbelt MD, 20771, USA; [email protected] 3 Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA, 91109, USA; [email protected] 1 2 X-ray Variability from RCW 103 505505 (due north); the flux of the latter has evidently remained constant (see Torii et al. 1999 for details). However, we es- timate that 1E 161348- 5055 has dimmed a factor ≃ 12 in the hard band, after the diffuse flux from the remnant has been taken into account using the following method. The soft X-ray flux (below 2 keV) is dominated by steady thermal emission from shock-heated gas in the remnant. The contribution of this component to the hard-band images is esti- mated using the soft-band images. The latter provides a good model for the spatial distribution of the surrounding shell on arcminute scales in the 3 - 10 keV range. The soft-band con- tribution from the shell was renormalized to the hard-band, and subtracted from the flux calibrated hard-band image to extract the flux contribution from the source alone. For the comparison, the new data were rebinned by a factor of 4 (∼ 1′ × 1′ pixels) to match the binning used with the earlier observation. The longer exposure of the second observation results in increased sensi- tivity to 1E 161348- 5055, however, this gain is offset by its location at a greater off-axis angle (as is evident by the asym- metrical PSF) relative to the first observation. An equivalent analysis of the SIS data reproduces the variability seen in the GIS hard band. 3.1. Spectroscopy We analyzed the spectrum from the new observation using the same approach presented in GPH97. To maximize the sen- sitivity, we simultaneously fit the spectra from all four ASCA detectors. We restricted our SIS spectral fits to > 1.2 keV as the calibration at the lower energies has become less reliable over time. The improved viewing geometry over the previous observation, coupled with the spectral stacking (four detectors), made it possible to measure the spectrum, despite the lower source flux (only one SIS data set was used in the earlier spec- tral analysis). The resulting fits to simple models (Table I) are consistent with those inferred for the first observation. Thus, while the flux has decreased by an order of magnitude, the spec- trum appears essentially unchanged. 3.2. The Long-term Light Curve To investigate its long-term flux behavior, we constructed a light curve of 1E 161348- 5055 which spans 18 years (1979 - 1997), using 10 available archival observations. For each ob- servation, we extracted background-subtracted countrates or 3σ upper limits. These rates are then used to estimate the flux in a given energy band. For lack of knowledge to the contrary, we assume that the spectral shape is invariant in time and modeled by a blackbody whose parameters are given in Table 1. The best-fit power-law model is unphysical at the softer X-ray ener- gies and thus the blackbody model is preferred for this compar- ison. We folded the latter model through the spectral response function of each instrument using the XSPEC spectral fitting package and inferred the source flux for each observation in a fiducial 0.5 - 2 keV energy band. The results, listed in Table 2 and plotted in Fig. 2, confirm a dramatic flux change between the ASCA observations, and suggest that 1E 161348- 5055 is variable, to a lesser extent, amongst the other observations. We present this source flux comparison among instruments with some caution, as these can be potentially unreliable. The different energy bands, flux calibrations, point response func- tions, and background contamination can produce large uncer- tainties in the derived fluxes. When extrapolating, the relative count rates are very sensitive to the instrumental energy band, along with the assumed NH and emission model. However, our fundamental result stands regardless of the aforementioned caveats: the ASCA data alone, and to a lesser extent the ROSAT data, establish the fact that 1E 161348- 5055 varied throughout the time it has been observed. 3.3. The Short-term Variability The excursions in flux noted from observations separated by months or years suggest that variability might be present on shorter time-scales. We searched the day-long ASCA observa- tions for hour-scale temporal variability. In addition, we exam- ined the behavior on a timescale of a few days using the August, 1995, ROSAT HRI observation, with a net exposure time of 50 ks spread over a week. In neither case did we find evidence of variability greater than the photon statistic limit of 10 percent. We searched the new ASCA GIS data for a coherent pulsed signal, as in GPH97. No significant periodicity was found in the period space between 10 ms - 1000 s. The upper limit on the pulsed fraction for this data set is ∼ 23%, compared with the 13% limit of GPH97. The data show no clear evidence for accretion noise such as redness in the power spectrum. The data were searched for X-ray bursts and other anomalies in the light curve, but none were found. 4. DISCUSSION A number of hypotheses have been advanced to explain the nature of 1E 161348- 5055. Upon discovery, this source was proposed as an isolated neutron star emitting blackbody radia- tion (Tuohy & Garmire 1980). Further optical and radio obser- vations (e.g. Tuohy et al. 1983; Dickel et al. 1996) have failed to identify a counterpart, thereby bolstering this interpretation. The observations of GPH97 showed that the point source could be described as a hot blackbody of kT ≃ 0.6 keV and a 0.5 - 10 keV flux of 6.5 × 10- 12 erg cm- 2 s- 1; therefore a luminosity of LX ≃ 1034 erg s- 1 (at 3.3 kpc) and an effective emitting area of 1 km2, or ∼ 0.1% the surface area of a NS. This corresponds to a rather small hot-spot on a stellar surface, which is in turn sur- prising since the source shows no rotational modulation (down to ∼ 13%). Also, the inferred temperature is too high for an object of age few ×103 yrs (but see below). Heyl & Hernquist (1998) recently attempted to salvage the cooling NS model by invoking an ultra-magnetized star (Bs ∼ 1015 G) with an accreted hydrogen atmosphere (Page 1997, for a review). This combination can enhance the cooling flux as well as shift the emission bluewards (Chabrier, Potekhin & Yakov 1997) so that it effectively mimics a hot blackbody in the ASCA spectrum. However, it is hard for cooling mod- els to address the issue of variability or the flux decimation observed between the ASCA epochs, unless there is a source for rejuvenating the heating of the NS interior. Such a source could be the super-strong magnetic field, implicit in the above model. Episodic rearrangement of the field (Thompson & Dun- can 1996) in the stellar interior could provide the energy to im- pulsively heat the core. The stellar surface would re-adjust to reflect the internal heat- ing on a short thermal timescale of a few months. Although heating of the NS is viable in this scenario, the rapid cooling on a timescale of a few years, observed between the ASCA epochs, cannot be explained without some very "exotic" cooling pro- cess . In addition, a factor of 10 variability in the hard (3 - 10 keV) band would result in a downward shift of the effective temperature by a factor 1.4, which should have been detected. On these grounds, we reject the hypothesis that the observed Gotthelf, Petre & Vasisht 3 X-ray emission from 1E 161348- 5055 is simple cooling radia- tion. < An ultra-magnetized NS is also a leading model for the anomalous X-ray pulsars (AXPs), an example of which is the ∼ 2,000 yr-old, 12-s X-ray pulsar in the remnant Kes 73 (Va- sisht & Gotthelf 1997). GPH97 compared 1E 161348- 5055 to the latter on the basis of similar spectral characteristics. And at least two AXPs are reported to vary significantly in flux, by as much as a factor of five (1E 1048.1-593; Oosterbroek et al. 1998). While 1E 161348- 5055 shows some properties that are tantalizingly similar to those of the AXPs, the lack of observed strong pulsations (∼ 30% modulation for the AXPs) is notably amiss, particularly as a large magnetic field should result in highly anisotropic surface emission. Gravitational defocusing and/or unfavorable viewing geometry, however, might account for the lack of observed pulsations. Alternatively, the variability can be indicative of an accreting compact object. Popov (1997) suggests that 1E 161348- 5055 is an old accreting NS with a low magnetic field and long spin period (∼ 103 s), the by-product of a disrupted binary and not of the same age as RCW 103. This proposition is bolstered by the discovery of the nearby 69 ms pulsar AX J161730-505505 (spindown age of 8100 yrs), located outside the remnant shell (Torii et al. 1998). Several arguments, including those based on the pulsar's implied velocity and lack of wind nebula, and the symmetry of the SNR, however, make an association unlikely (Kaspi et al. 1998, Torii et al. 1999). Finally, there exists the possibility that the source is an iso- lated stellar-mass black hole (BH), accreting from the surround- ing medium or from supernova ejecta fall-back. Brown & Bethe (1994) have discussed scenarios in which a massive progenitor explodes as a supernova and then evolves into a BH of several solar masses after accreting captured ejecta. Such a scenario is clearly applicable to the source in RCW 103. Temporal vari- ability and lack of pulsed emission are the natural consequences in such a model. An accretion process around a BH almost inevitably involves rotating gas flows. Popov (1997) has dismissed the possibility of a few stellar-mass BH in RCW 103 based on the small im- plied emitting area (∼ 1 km2) of an equivalent blackbody radi- ator (see §4). This argument would certainly apply for the case of the standard optically-thick, thin-disk model. However, low- efficiency solutions can exist for accretion flows (especially at low M) around BHs, in which most of the viscously generated thermal energy is advected into the BH. Below a critical mass accretion rate of ∼ 0.1 MEdd ( MEdd is the Eddington rate) ac- cretion flows turn advection dominated (or ADAF), and the ob- served luminosity in 1E 161348- 5055 would suggest an accre- tion rate of 10- (2.5- 3.0) MEdd (or ∼ 10- 10 M⊙ yr- 1) (cf ADAF models summarized in Narayan et al. 1998). At this rate, the BH would accrete ∼ 10- 7 M⊙ of matter, a small fraction of the mass of supernova ejecta, at the sustained present rate over its lifetime of ∼ 103 yr. Within the framework of the above argu- ments, it is possible that flow around 1E 161348- 5055 could be ∼ 22 after accounting for detected as a faint optical source (V > visual extinction) or a 10-100 µJy (1 GHz) radio source. 5. CONCLUSIONS The variability of the X-ray emission from the compact source in RCW 103 leaves little room for a conventional cooling NS origin. Instead, an accretion scenario may be considered, although the relatively low luminosity, lack of optical counter- part, and young age are inconsistent with a typical accreting NS binary. Accretion from a very low mass (≤0.1 M⊙) compan- ion (Mereghetti et al. 1996; Baykal & Swank 1996) or a fossil disk around a solitary NS (van Paradijs et al. 1995), however, is not ruled out. We suggest that within the context of ineffi- cient accretion (such as advection dominated flows), a stellar mass black hole is a viable possibility. We reiterate, however, that the spectral characteristics of 1E 161348- 5055 are remark- ably similar to those of the AXPs, which are suspected to be ultra-magnetized NSs and thought to be powered by magnetic field decay rather than rotational braking (Thompson & Duncan 1996; Vasisht & Gotthelf 1997). Independent of the above phenomenology, the properties of 1E 161348- 5055 add to the view that young collapsed stars can follow an evolutionary scenario quite distinct from those of Crab-like pulsars. The property of being radio quiet is common to all AXPs, SGRs and to 1E 161348- 5055 (Gotthelf 1998; for possible physical mechanisms see Baring & Harding 1998). It is likely that they all share a common heritage and may prove to be part of an evolutionary sequence. This work uses data made available from the HEASARC public archive at GSFC. G.V. thanks J Heyl for discussions. REFERENCES Baykal, A. & Swank, J. 1996, ApJ, 460, 470 Baring & Harding 1998, ApJ, 507, L55 Becker, W., Trümper, J., Hasinger, G., & Aschenbach, B. 1993 in "Isolated Pulsars", ed Van Riper, Epstein, & Ho, Cambridge Univ. Press (Cambridge), 116 Brown, G. E. & Bethe, H. A. 1994, ApJ, 423, 659 Caraveo, P. A., Bignami, G. F. & Trümper, J. E. 1996, A&AR, 7, 209 Carter, L. M., Dickel, J. R. & Bomans, D. J. 1997, PASP, 109, 990 Caswell, J. L., Murray, J. D., Roger, R. S., Cole, D. J., & Cooke, D. J. 1975, A&A, 45, 239 Chabrier, G, Potekhin, A. Y., & Yakovlev, D. G. 1997 ApJ, 477, L99 Carter, L. M., Dickel, J. R., Bomans, D. J. 1997, PASP, 109, 990 Cline, T. L. et al. 1982, ApJ, 255, L45 Dickel, J. R., Green, G., Ye, T., & Milne, D. K. 1996, AJ, 111, 340 Gaensler, B. M. & Johnston, S. 1995, MNRAS, 275, L43 Gotthelf, E.V. 1998, in "Workshop on the relationship between NS & SNRs," Memorie della Societa' Astronomica Italiana Gotthelf, E.V., Petre, R., & Hwang, U. 1997, ApJ, 487, L175 Gotthelf, E.V. & Vasisht, G. 1998, New Astronomy, 3, 293 Gregory, P. C. & Fahlman, G. G. 1980, Nature, 287, 805 Heyl, J. S. & Hernquist, L. 1998, MNRAS, 300, 599 Kaspi, V. M., Crawford, F., Manchester, R. N., Lyne, A. G., Camilo, F. D'Amico, N., & Gaensler, B. M. 1998, ApJ, in press Kulkarni, S. R., Frail, D. A., Kassim, N. E., Murakami, T., & Vasisht, G. 1994, Mereghetti, S. Bignami, G. F., & Caraveo, P. A. 1996, 464, 842 Oosterbroek, T., Parmar, A. N., Mereghetti, S., & Israel, G. L. 1998 A&A, 334, Nature, 368, 129 925 Narayan, R., Mahadevan, R., & Quataert, E. 1998, to appear in "The Theory of Black Hole Accretion Disks" eds. M. A. Abramowicz, G. Bjornsson, & J. E. Pringle (Cambridge Univ. Press) Petre, R., Becker, C. M., & Winkler, P. F. 1996, ApJ, 465, L43 Page, D. 1997, astro-ph/9706259 Popov, S.B. 1997, A&A, submitted Tanaka, Y., Inoue, H., & Holt, S.S. 1994, PASJ, 46, L37 Thompson, C., Duncan, R. C. 1996, ApJ, 473, 322 Torii, K, Kinugsa, K., Toneri, T., Asanuma, T., Tsunemi, H., Dotani, T., Mitsuda, K., Gotthelf E. V., & Petre, R. 1998, ApJ, 494, L207 Torii, K. et al. 1999, in prep. Tuohy, I. R., Garmire, G. P., Manchester, R. N., & Dopita, M. A. 1983, ApJ, 268, 778. Tuohy, I. & Garmire, G. 1980, ApJ, 239, L107 Vasisht, G, Kulkarni, S. R., Frail, D. A. & Greiner, J. 1994, ApJ, 431, L35 Vasisht, G. & Gotthelf, E. V. 1997, ApJ, 486, L129 4 X-ray Variability from RCW 103 TABLE 1 FITS TO ASCA SIS SPECTRUM kT or Γb . . . . . . . . . . . . . . . . . . . . . . 1997 . . . . . . . . . . . . . . . . . . . . . . Modela 2(DoF) χ Diffuse Surroundings NEI Source NEI NEI + BB NEI + PL NEI + BREM 390 (227) 417 (211) 260 (209) 238 (209) 247 (209) 0.3 - 0.630.71 0.57 3.94.1 3.6 1.431.76 1.20 kT or Γb 1993 0.3 - 0.560.59 0.53 3.23.4 3.0 1.61.8 1.4 aSee GPH97 for details. b Γ is the photon index; kT in units of keV. NH is fixed to the best fit values of 7.3 × 1021 cm2 (1997) and 6.8 × 1021 cm2 (1993). LOG OF X-RAY IMAGING OBSERVATIONS OF RCW 103 TABLE 2 Mission/Sensor Obs. Date (yy/mm/dd) Expo. (ks) Cnt ratea (cps) .5 -- 2 keV fluxb (10- 13 cgs) Einstein / IPC Einstein / HRI ROSAT / HRI ROSAT / PSPC ASCA / SIS ROSAT / HRI ROSAT / HRI ROSAT / HRI ROSAT / HRI ASCA / SIS 79/02/26 79/09/14 91/02/13 91/02/27 93/08/17 94/02/26 94/08/02 95/03/08 95/08/18 97/09/04 2.6 3.1 5.1 37.2 47.7 1.1 4.4 48.1 8.1 58.6 <40 3.8±0.7 <7.5 7.7±2.3 50±10c 13±4 9±2.6 8.4±0.7 7.4±1.6 4±1c 4.4±0.7 <3.2 1.1±0.3 55±11 6.0±1.7 3.8±1.2 3.6±0.4 3.1±0.6 <5.1 6±1 aSpectral fit band-pass: 0.2- 4.5 keV Einstein; 0.2- 2.4 keV ROSAT; 1.2- 10 keV ASCA. bFlux at earth in ergs cm- 2 s- 1, assuming best fit blackbody (kT=0.63 keV) and NH (7.3×1021 cm- 2) from the second ASCA observation. Upper limits are 3σ. Gotthelf, Petre & Vasisht 5 FIG. 1. -- ASCA GIS images of 1E 161348- 5055 (cross), the compact point source in RCW 103, taken 4 years apart (Left -- 1993; Right -- 1997). At the top of the picture is the 69 ms X-ray pulsar AX J161730-505505. These background subtracted hard energy band-pass (3 - 10 keV) images are flux corrected and plotted with the same scaling. The central source in RCW 103, 1E 161348-5055, has dimmed by a factor of ∼ 12, while the pulsar flux remained relatively constant. The soft nebula flux from the shell of RCW 103 has been subtracted from these images. Contours along the edge of GIS in the 1997 observation (upper right corner) are not significant. FIG. 2. -- The long term light curve of 1E 161348- 5055 in the 0.5 - 2.0 keV energy band-pass. Symbols represent: Einstein IPC (open circle); Einstein HRI (filled circle); ROSAT PSPC (open square), HRI (filled squares); and ASCA (star). Error bars are one sigma. The dashed vertical lines indicate the model dependent uncertainty in the ASCA flux when extrapolated to the 0.5 - 2 keV band-pass.
astro-ph/0604542
1
0604
2006-04-26T11:17:13
The multi-phase gaseous halos of star forming late-type galaxies - II. Statistical analysis of key parameters
[ "astro-ph" ]
In Paper I we showed that multi-phase gaseous halos of late-type spiral galaxies, detected in the radio continuum, in Halpha, and in X-rays, are remarkably well correlated regarding their morphology and spatial extent. In this work we present new results from a statistical analysis in order to specify and quantify these phenomenological relations. This is accomplished by investigating soft X-ray (0.3-2.0keV) luminosities, FIR, radio continuum, Halpha, B-band, and UV luminosities for a sample of 23 edge-on late-type spiral galaxies. Typical star formation indicators, such as SFRs, are determined and a statistical correlation analysis is carried out. We find strong linear correlations, covering at least two orders of magnitude, between star formation indicators and integrated (disk+halo) luminosities in all covered wavebands. In addition to the well established L_FIR/L_1.4GHz-relation, we show new and highly significant linear dependencies between integrated soft X-ray luminosities and FIR, radio continuum, Halpha, B-band, and UV luminosities. Moreover, integrated soft X-ray luminosities correlate well with SFRs and the energy input into the ISM by SNe. The same holds if these quantities are plotted against soft halo X-ray luminosities. Only a weak correlation exists between the dust mass of a galaxy and the corresponding X-ray luminosity. Among soft X-ray luminosities, baryonic, and HI-gas masses, no significant correlations are found. There seems to exist a critical input energy by SNe into the ISM or a SFR threshold for multi-phase halos to show up. It is still not clear whether this threshold is a physical or an instrument dependent sensitivity limit. These findings strongly support our previous results, but conflict with the concept of halos being due to infalling gas from the IGM.
astro-ph
astro-ph
Astronomy&Astrophysicsmanuscript no. aa6b September 23, 2018 c(cid:13) ESO 2018 6 0 0 2 r p A 6 2 1 v 2 4 5 4 0 6 0 / h p - o r t s a : v i X r a The multi-phase gaseous halos of star forming late-type galaxies⋆ II. Statistical analysis of key parameters R. Tullmann1, D. Breitschwerdt2, J. Rossa3, W. Pietsch4, and R.-J. Dettmar1 1 Astronomisches Institut, Ruhr-Universitat Bochum, D-44780 Bochum, Germany e-mail: [email protected],[email protected] 2 Institut fur Astronomie, Turkenschanzstrasse 17, A-1180 Wien, Austria e-mail: [email protected] 3 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, U.S.A. e-mail: [email protected] 4 Max-Planck Institut fur extraterrestrische Physik, Giessenbachstrasse, D-85748 Garching, Germany e-mail: [email protected] Received December 21, 2005; accepted April 26, 2006 ABSTRACT Context. In a previous paper (Paper I) we showed that multi-phase gaseous halos of late-type spiral galaxies, detected in the radio continuum, in Hα, and in X-rays, are remarkably well correlated regarding their morphology and spatial extent. Aims.In this work we present new results from a statistical analysis in order to specify and quantify these phenomenological relations. Methods. This is accomplished by investigating soft X-ray (0.3 -- 2.0 keV) luminosities, FIR, radio continuum (1.4 GHz), Hα, B-band, and UV (1550Å -- 1650Å) luminosities for a sample of 23 edge-on late-type spiral galaxies. Typical star formation indicators, such as star formation rates (SFRs), are determined and a statistical multi-parameter/frequency correlation analysis is carried out. Results. We find strong linear correlations, covering at least two orders of magnitude, between star formation indicators and integrated (disk+halo) luminosities in all covered wavebands. In addition to the well established LFIR/L1.4GHz relation, we show new and highly signifi- cant linear dependencies between integrated soft X-ray luminosities and FIR, radio continuum, Hα, B-band, and UV luminosities. Moreover, integrated soft X-ray luminosities correlate well with SFRs and the energy input into the ISM by SNe. The same holds if these quantities are plotted against soft halo X-ray luminosities. Only a weak correlation exists between the dust mass of a galaxy and the corresponding X-ray luminosity. Among soft X-ray luminosities, baryonic, and H  gas masses, no significant correlations are found. There seems to exist a critical input energy by SNe into the ISM or a SFR threshold for multi-phase halos to show up. It is still not clear whether this threshold is a physical one or represents an instrument dependent sensitivity limit. Conclusions. These findings strongly support our previous results that multi-phase gaseous galaxy halos in late-type spiral galaxies are created and maintained by outflowing gas produced in star formation processes in the disk plane. They conflict with the concept of halos being mainly due to infalling gas from the intergalactic medium. Key words. Galaxies: formation -- Galaxies: halos -- Galaxies: ISM -- Galaxies: spiral -- Galaxies: starburst -- X-rays: galaxies 1. Introduction In Paper I (Tullmann et al., 2006) we established the morpho- logical correlation and spatial coincidence between gaseous ha- los in late-type spiral galaxies, traced by their radio contin- uum, Hα, and soft diffuse X-ray emission. We further showed that the continuum emission radiated at UV wavelengths (∼ 210 nm) originated in the disks of the galaxies and is well as- ⋆ Based on observations obtained with XMM-Newton, an ESA sci- ence mission with instruments and contributions directly funded by ESA member states and NASA. sociated with diffuse ionized gas (DIG) in the halo. It is gen- erally accepted that the non-thermal radio continuum is a reli- able indicator for cosmic ray (CR) halos, that extraplanar soft X-rays are a good tracer of the hot ionized medium (HIM, e.g., produced by supernova remnants (SNRs), in superbub- bles or in superwinds), and that the presence of extraplanar DIG is indicative of the so called disk-halo interaction (see Dettmar, 2004, for a recent review). Therefore, we concluded that multi-phase gaseous halos are created by star formation (SF) related processes in the disk plane. Our results presented in Paper I did not yield convincing evidence that galaxy halos 2 R. Tullmann et al.: The multi-phase gaseous halos of star forming late-type galaxies form by infalling gas from an external reservoir as considered e.g. by Benson et al. (2000), Toft et al. (2002) or Pedersen et al. (2005). Apart from these phenomenological analogies there also should exist strong statistical correlations between total (disk+halo) soft X-ray, Hα (DIG), and radio continuum lumi- nosities. If this hypothesis is correct, these luminosities should also strongly correlate with other star formation indicators, such as the FIR, B-band, and UV luminosity (e.g., Condon, 1992; Read & Ponman, 2001), but also with direct tracers of massive star formation, such as star formation rates (SFRs, see Kennicutt, 1998), supernova (SN) energy input rates normal- ized to the area of active star formation Etot A (e.g., Dahlem et al., 1995; Rossa & Dettmar, 2003, simply called energy input rates, hereafter), and LFIR/D2 25 ratios (Lehnert & Heckman, 1996). In the following we investigate the interrelations among star formation related key parameters and luminosities in or- der to establish significant (statistical) correlations of the up to now purely morphological relations. Integrated luminosities are used to test whether the diffuse emission from the disk and halo is coupled to SF, whereas the pure extraplanar soft X-ray emission is put into relation to the above mentioned parameters in order to show that multi-phase halos are indeed tightly correlated with SF processes in the disk. Finally, we discuss the question of the existence of an en- ergy threshold provided by stellar feedback processes which needs to be exceeded in order to create multi-phase halos. From the above mentioned correlations first empirical estimates of the suspected energy threshold are derived. 2. Statistical analysis of key parameters 2.1. Keyparametersandsampleselection It is straightforward to include additional quantities which trace massive stars in the disk and the energy they provide for the interstellar medium (ISM) and to search for trends among all those parameters. Such supplemental key quantities considered here are the B-band and UV luminosites as well as SFRs (or the FIR SFR per unit area, expressed by LFIR/D2 25) and energy input rates by SNe, represented by Etot A . We also include variables which are not directly coupled to SF, but trace different neutral phases of the ISM, such as the dust mass and the H  mass of a galaxy. Moreover, the extent of the SN-driven gas and therefore the size of the multi-phase halo also depends on the gravitational potential of a galaxy (e.g., Mac Low & Ferrara, 1999). As a first order approximation of this quantity the baryonic mass of a galaxy (stars and H ) determined from the K-band Tully-Fisher relation (Bell & de Jong, 2001) is also included. As was already pointed out in Paper I, our initial sample is biased against starburst galaxies, that is towards higher SFRs or energy input rates Etot A . In order to investigate the possible correlations, non-starburst (but actively star forming) galaxies need to be included which have to cover the intermediate and lower energetic ends of these parameters. This obstacle was overcome by selecting edge-on (i > 70◦) late-type spiral galax- ies from the literature for which the above mentioned luminosi- ties have been published and the SF related parameters could be calculated. In order to reach the largest possible coverage of the parameter space no constraints were imposed on the inte- grated luminosities, the S 60/S 100 and LFIR/D2 25 ratios, and the energy input rates Etot A . Galaxies are considered to have multi-phase gaseous halos, if they show kpc-sized extraplanar radio continuum, DIG, and X-ray emission which surrounds a large fraction of the disk plane. Extended X-shaped DIG structures, as visible on Hα images of NGC 1482 (Strickland et al., 2004a) or NGC 5775 (Tullmann et al., 2006), trace most likely a limb brightened outflow cone and are also classified to possess a halo. Single plumes or filamentary structures are not considered to consti- tute a halo. It should be kept in mind that the sample is limited in sen- sitivity and is not complete. The enlarged sample consists now of 23 galaxies, nine from Tullmann et al. (2006), seven from Strickland et al. (2004a, three of their targets are also in our sample and their results are consistent with ours), and another seven galaxies taken from the literature. The number of targets is sufficiently large to test the individual correlations by means of a statistical anal- ysis. All physical parameters adopted throughout this study are listed for the whole sample in Table 1. Integrated X-ray luminosities for the first nine galaxies listed in Table 1 are from Paper I, the next seven are taken from Strickland et al. (2004a), whereas the last seven stem from other literature sources (see notes to table). 2.2. Statisticalanalysis We performed least-squares fitting (assuming Y = mX + b) and Spearman rank-order correlation analysis, to test the sig- nificance of the correlation between the investigated pairs of parameters. As these quantities, for example LX and LFIR, usu- ally vary only within a certain physical range, e.g., set by the number of stars in the disk, these variables are most likely not normally distributed and their interrelation needs to be investi- gated by means of rank-order correlation analysis. The slopes m, their intercepts b and b′ (for m = 1.0), the reduced χ2, the Spearman rank-order correlation coefficients rs, ts, the levels of significance of rs, and the corresponding p-values of rs are listed in Table 2. In the case that both quan- tities are not linearly correlated (null hypothesis), ts follows a Student's t-distribution with n − 2 degrees of freedom. The null hypothesis turns out to be wrong if ts > tn−2,1−α, where α is the level of significance, which we adopt to be 0.01. The quan- tile t21,0.99 of the t-distribution, as tabulated in relevant statistics textbooks (e.g., Hartung et al., 2002), is 2.518. For all ts-values which are larger, the null hypothesis needs to be discarded (with a 99% confidence level) which means that all those pairs of parameters are strongly correlated. Moreover, if the p-values listed in Table 2 for the integrated luminosities are less than the adopted α-level, the null hypoth- esis needs to be discarded, too. In the present case only H  and R. Tullmann et al.: The multi-phase gaseous halos of star forming late-type galaxies 3 Table 2. Fit parameters and correlation coefficients. Listed are the Spearman rank-order correlation coefficient rs, its correspond- ing significance ts, and the p-value of rs, together with results derived from linear regression (Y = mX + b) for the pairs of parameters plotted in Figs. 1 -- 3. Uncertainties of the gradient m and the intercept b are given on a 1σ level. b′ is the intercept if a slope of m = 1 is assumed. log(Y) log(X) m b b′ red. χ2 LFIR LX LX LX LX LX LX LX LX LX LX LX LX LX,h LX,h LX,h LX,h L1.4GHz L1.4GHz LFIR LHα LB LUV S FRFIR S FRHα LFIR/D2 25 EA Mdust MH i MBaryon S FRFIR S FRHα EA LFIR/D2 25 0.94±0.06 0.65±0.11 0.80±0.14 0.88±0.27 1.29±0.29 0.87±0.22 0.79±0.15 0.86±0.27 0.73±0.17 0.68±0.12 0.70±0.21 0.15±0.33 0.51±0.37 0.93±0.17 1.13±0.22 0.77±0.21 0.70±0.32 +0.85±0.04 −0.10±0.11 −0.87±0.13 −0.95±0.22 −1.12±0.19 +0.28±0.24 −0.58±0.11 −0.07±0.18 −1.07±0.19 −0.47±0.10 −0.27±0.13 −0.31±0.22 −0.66±0.25 −1.13±0.14 −0.31±0.15 −1.12±0.19 −1.60±0.45 +0.86 −0.01 −0.99 −1.03 −0.95 +0.40 −0.64 −0.02 −1.32 −0.51 −0.22 +0.09 −0.93 −1.11 −0.35 −1.14 −1.89 0.030 0.266 0.243 0.370 0.295 0.292 0.252 0.378 0.303 0.230 0.386 0.578 0.535 0.189 0.203 0.308 0.449 rs 0.97 0.74 0.76 0.67 0.71 0.73 0.76 0.64 0.69 0.78 0.58 0.09 0.29 0.72 0.71 0.69 0.61 ts 17.4 4.80 5.10 3.93 4.39 4.66 5.10 3.63 4.16 5.43 3.10 0.39 1.32 3.42 3.26 2.96 2.45 p-value <0.0001 0.0001 0.0001 0.0009 0.0003 0.0002 0.0001 0.0018 0.0005 <0.0001 0.0058 0.6980 0.2022 0.0057 0.0075 0.0144 0.0341 baryonic masses do not fulfill these criteria and are therefore assumed not to correlate with soft X-ray luminosities. The same analysis was carried out for the extraplanar soft X-ray emission (LX,h) and the derived parameters are also listed in Table 2. All correlations and their corresponding least-square fits are shown in Fig. 3. 3. Results and Discussion 3.1. Correlationswithintegratedluminosities Based upon statistical analysis of the sample strong linear cor- relations (rs ≥ 0.67) between star formation indicators and inte- grated (disk+halo) multi-frequency luminosities could be es- tablished. In addition to the generally accepted LFIR/L1.4GHz relation (e.g., de Jong et al., 1985; Condon, 1992, see also Table 2 of this work), highly significant linear dependencies between integrated soft X-ray luminosities (0.3 -- 2.0 keV) and integrated radio continuum (1.4 GHz), B-band, UV, Hα, and FIR luminosities have been found (see Fig. 1)1 Moreover, the integrated soft X-ray luminosities correlate well with star for- mation rates and SNe energy input rates (Figs. 1 and 2). These correlations unambiguously confirm that the diffuse radiation of the X-ray emission and the DIG, as well as the CRs detected in the disks and halos are indeed coupled to SF. Although Hα, B, and UV-band luminosities are generally a good tracer of the SF activity, these quantities can be seriously affected by absorption internal to the galaxy. The typical scatter of about 1 dex visible in our correlations might be indicative of 1 The latter two relations are not shown here, as the Hα and FIR luminosities are directly proportional to the corresponding SFRs (see Kennicutt, 1998). this effect. However, absorption appears to be unlikely to lower the luminosities in such a way that the correlations are seriously weakened or even destroyed. The strongly discrepant data point visible in the LX/LUV- diagram represents measurements for one of the most extreme starburst galaxies, M 82. In order for M 82 to follow our corre- lation, a much higher (lower) UV (X-ray) luminosity of about 1 dex is required. However, independent observations car- ried out in the UV (Code & Welch, 1982; Rifatto et al., 1995; Hoopes et al., 2005) and at X-ray energies (Fabbiano et al., 1992; Strickland et al., 2004a) yield very consistent results which imply that this "deviation" is an intrinsic feature of this outstanding galaxy. More than 60% of the UV emission of M 82 measured at about 1530Å is extraplanar in origin (Hoopes et al., 2005). These authors claim that extraplanar star formation or a com- bination of photoionization and shock ionization either appear unlikely or cannot account for the halo emission in the UV. Instead stellar continuum radiation from the disk plane scat- tered at dust particles in the halo shall be the main mechanism causing the UV emission. Given that M 82 is a low mass spi- ral galaxy (Mbar = 1.86 × 1010M⊙, Table 1) dust is assumed to be driven relatively easily by the superwind to high extra- planar distances. This independently supports the "scattering"- hypothesis and could also explain the lacking UV emission, provided the (LX, LUV)-relation is valid. Alternatively, the high X-ray and low UV luminosity of M 82 can be explained by gas within the starburst region which is thermalized beyond the UV excitation limit. In such a ex- treme case in which the diffuse soft X-ray emission is domi- nating the energy output, we expect a deviation from the linear correlation curve during phases of strong starburst activity. 4 R. Tullmann et al.: The multi-phase gaseous halos of star forming late-type galaxies Fig. 1. Diagnostic diagrams confirming the strong correlations between integrated radio continuum, B-band, UV, and soft X-ray (0.3 -- 2.0 keV) luminosities. X-ray luminosities also correlate well with the energy input rate by SNe ( EA) and with FIR and Hα SFRs. Circles address our sample, squares represent data from Strickland et al. (2004b), and triangles denote data, collected from the literature (cf. Table. 1). Filled (open) symbols refer to galaxies with (un)detected multi-phase gaseous halos. Solid lines are best fits from linear regression while dashed lines indicate the trend expected for a relationship of unit slope. rs is the Spearman rank-order correlation coefficient. Apparently only a weak correlation between the dust mass of a galaxy and the corresponding soft X-ray luminosity exists. Dust, as a byproduct of SF, is correlated with the FIR luminos- ity and should therefore also correlate (at least weakly) with other SF indicators, such as X-ray luminosities. The non-correlation between LX and MH i is also not sur- prising as the neutral gas component of the ISM is not directly involved in SF. Surprisingly only at a first glance, a correlation between galaxy potential and X-ray luminosities is not found. As the baryonic mass (Bell & de Jong, 2001) is calculated from ro- tational velocities of the stars and the H -gas in a galaxy, this parameter traces the dynamics of the system and is therefore in- sensitive to SF. Nevertheless, this finding bears two important implications. Firstly, the dynamics of the gas is apparently not related to the diffuse emission produced in SF events. Secondly, for a more accurate estimate of the baryonic mass a better mass- related approximation needs to be found. This relation should also include dark matter halos in order to study the effects this additional component might have on the established relations. If there is still no correlation with baryonic mass, then the tar- gets of our sample do not follow the LX ∼ v5 rot relation derived from the infall simulations by Toft et al. (2002). R. Tullmann et al.: The multi-phase gaseous halos of star forming late-type galaxies 5 Fig. 2. Functional dependence between soft X-ray luminosities and LFIR/D2 its H  mass (lower left), and the baryonic mass (lower right). 25 (upper left), the dust mass of a galaxy (upper right), 25 (upper left), EA Fig. 3. Functional dependence between LX,h, the soft (0.3 -- 2.0 keV) X-ray luminosity of the halo, and LFIR/D2 (upper right), S FRFIR (lower left), and S FRHα (lower right). Solid symbols represent our sample, open symbols denote data taken from Strickland et al. (2004a). Solid and dashed lines have the same meaning as described in Fig. 1. 6 R. Tullmann et al.: The multi-phase gaseous halos of star forming late-type galaxies It should be pointed out that Mbar represents the gravita- tional potential of a galaxy and should therefore be sensitive to the overall extent of the multi-phase halo. However, it is not sensitive to the existence/non-existence of such halos. In this regard the SFR and the energy input rate by SNe into the ISM turn out to be the most important parameters. 3.2. Correlationswithextraplanarluminosities In order to establish that multi-phase halos are indeed a product of SF, the pure extraplanar emission in the radio continuum, in Hα, and at soft X-ray energies also needs to correlate with SFRs and the energy input rate by SNe. At present this analysis can best be done in the X-ray regime. Other wavelengths, such as Hα and the radio contin- uum, are not usable, because the integral disk and halo lumi- nosities do not allow to reliably estimate the fraction of the lu- minosity originating in the halo. For the DIG, this fraction can vary from 12% to 60% (Rossa & Dettmar, 2000) and would therefore lead to unacceptable uncertainties. From the original sample, 13 galaxies are suitable for the "halo"-sample (those from Strickland et al. (2004a) and Tullmann et al. (2006)). The remaining galaxies either do not show extraplanar X-ray emission or the published data did not allow a disentangling of disk and halo luminosities. It should be pointed out that inclination issues, regarding the reliability of detecting multi-phase halos, are negligible. Either because extraplanar gas was detected in case of incli- nation angles ≤ 80◦ (see NGC 1482, Strickland et al., 2004a), or because the inclination-corrected extent of the "extraplanar" gas was indeed not large enough to be seen above the disk plane (see NGC 3877, Tullmann et al., 2006). Most importantly, LX and LX,h are not linearly correlated with the inclination angle i. This is proved by low Spearman rank-order correlation coef- ficients of rs = −0.26 and rs = 0.23, respectively, again on a 99% confidence level. In Fig. 3 we plot the soft (0.3 -- 2.0 keV) halo luminosity (LX,h) as a function of the different SF indicators (which are 25, and EA). All pairs of parameters S FRFIR, S FRHα, LFIR/D2 show strong correlations (rs ≥ 0.67) on a 99% confidence level, except LX,h vs. LFIR/D2 25 (rs = 0.61). A possible expla- nation could be that LFIR/D2 25, a measure of the SF rate inten- sity per unit disk area, underestimates the real SF intensity. In disk galaxies SF usually occurs within the Hα-disk and is not that widespread as the distribution of the old stellar population (measured by D25) implies. It turns out that the FIR and the Hα SFRs are the most im- portant parameters which decide on the formation of X-ray ha- los. Among these two quantities, S FRFIR seems to be the more reliable parameter, because it is very little affected by dust ab- sorption and is dominated by disk emission. On the other hand, the Hα emisson, and therefore S FRHα, is a more sensitive mea- sure of how much gas is converted into massive stars. However, this quantity is significantly affected by extinction and in some cases most likely dominated by extraplanar emission. Given the strong correlations between integrated X-ray, Hα, and radio continuum luminosities as well as between halo X-ray luminosities and SFRs or EA, it is plausible, to adopt similar relations for the extraplanar DIG and the CR halos. Therefore, we expect the SFR and the energy input rate by SNe to be the main driving-agents for multi-phase galaxy halos. Our statistical results are also fully consistent with the morpholog- ical and spatial coincidences presented in Paper I. For consis- tency reasons it is certainly worthwhile to re-examine existing radio and DIG data to allow for a clear separation of disk and halo luminosities. 3.3. Constrainingthestarformationthreshold In order to achieve a more comprehensive picture on the evolu- tion of galaxy halos in different wave bands, we need to answer the question of the existence of a threshold SFR (or equiva- lently the SN energy input rate EA) required to create multi- phase gaseous halos. The existence of such a threshold can be motivated as follows: It is well known that hot ionized gas is driven off-plane via superbubbles produced by SNe. From a theoretical study of breakout conditions of such superbubbles (Mac Low & Mc Cray, 1988) we know that the mechanical lu- minosity of SNe is a key quantity which determines whether a superbubble will breakout or collapse. In other words, the overpressured X-ray emitting gas will leave the disk and form a halo if the energy provided by SNe is above a certain limit. On the other hand, it can be argued that channels, through which the hot gas reaches the halo, were already blown into the ISM during previous periods of intense star formation. Hence, a substantially lower SFR is needed to elevate the gas into the halo, making it even more difficult to establish a threshold by means of observations. From Fig. 1 the existence of a critical SFR (or energy input rate) is hard to constrain, as the lower energy end of our corre- lations is statistically not well covered. It appears, however, that galaxies without halos (open symbols) also follow a linear rela- tion and that the region dividing galaxies with halos from those without is relatively narrow. For multi-phase halos to evolve, the data imply that a critical threshold of S FRFIR ≥ 1.0 M⊙/yr, S FRHα ≥ 0.1 M⊙/yr or EA ≥ 1 × 10−3 erg s−1 cm−2 needs to be exceeded. Interestingly, DIG seems to coexist with other gas components for Hα energy fluxes larger than (3.2 ± 0.5) × 1040 erg s−1 kpc−2 (Rossa & Dettmar, 2003). Unfortunately, with the present data a detection threshold, introducing systematic uncertainties by the limited sensitivity of the instrument, cannot be a priori ruled out. In order to dis- tinguish between a significant physical threshold and a sensi- tivity limit as well as to reach a better understanding on the formation of gaseous galaxy halos, deep observations of galax- ies with no or only little halo emission are required. Therefore, future studies should aim at increasing the sample to test the low energy end of our correlations and to investigate the above relations for galaxies of different Hubble-types, such as Sa and Irr, as the different SF histories of these galaxies might lead to different relations and thresholds. R. Tullmann et al.: The multi-phase gaseous halos of star forming late-type galaxies 7 de Avillez, M. A., & Breitschwerdt, D. 2005, A&A, 436, 585 de Jong, T., Klein, U., Wielebinski, R., & Wunderlich, E. 1985, A&A, 147, L6 Dettmar, R. -- J. 2004, Ap&SS, 289, 349 de Vaucouleurs, G., et al. 1991, Third Reference Catalogue of Bright Galaxies (RC3), Springer Verlag Fabbiano, G., Kim, D.-W., & Trinchieri, G. 1992, ApJS, 80, 531 Ferguson, A. M. N., Wyse, R. F. G., Gallagher, J. S. III, & Hunter, D. A. 1996, AJ, 112, 2567 Hartung, J., Elpelt, B., & Klosener, K.-H. 2002, "Statistik, Lehr- und Handbuch der angewandten Statistik", 13th ed., Oldenbourg (2002) Hildebrand, R. H. 1983, QJRAS, 24, 267 Hoopes, C. G., Heckman, T. M., Strickland, D. K., et al. 2005, ApJ, 619, L99 Huchtmeier, W. K., & Richter, O.-G. 1989, A General Catalog of H  observations of Galaxies, Springer Verlag Irwin, J. A., English, J., & Sorathia, B. 1999, AJ, 117, 2102 Kenney, J. D. P., & Koopmann, R. A. 1999, AJ, 117, 181 Kennicutt, R. C., Jr., 1998, ARA&A, 36, 189 Lehnert, M. D., & Heckman, T. M. 1996, ApJ, 472, 546 Mac Low, M.-M., & Mc Cray, R. 1988, ApJ, 324, 776 Mac Low, M.-M., & Ferrara, A. 1999, ApJ, 513, 142 Maoz, D., Filippenko, A. V., Ho, L. C., Macchetto, F. D., Rix, H.-W., & Schneider, D. P. 1996, ApJS, 107, 215 Pedersen, K., Rasmussen, J., Sommer-Larsen, J., Toft, S., Benson, A. J., & Bower, R. G. 2005, astro-ph/0511682 Read, A. M., & Ponman, T. J. 2001, MNRAS 328, 127 Rifatto, A., Longo, G., & Capaccioli, M. 1995, A&AS, 114, 527 Rossa, J., & Dettmar, R. -- J. 2000, A&A, 359, 433 Rossa, J., & Dettmar, R. -- J. 2003, A&A, 406, 493 Sanders, D. B., Mazzarella, J. M., Kim, D.-C., Surace, J. A., & Soifer, B. T. 2003, AJ, 126, 1607 Schlegel, D. J., Barrett, P., & Singh, K. P. 1997, AJ, 113, 1296 Strickland, D. K., Heckman, T. M., Colbert, E. J. M., Hoopes, C. G., & Weaver, K. A. 2004a, ApJS, 151, 193 Strickland, D. K., Heckman, T. M., Colbert, E. J. M., Hoopes, C. G., & Weaver, K. A. 2004b, ApJ, 606, 829 Toft, S., Rasmussen, J., Sommer-Larsen, J., & Pedersen, K. 2002, MNRAS, 335, 799 Tullmann, R., Pietsch, W., Rossa, J., Breitschwerdt, D., & Dettmar, R. -- J., 2006, A&A, 448, 43 (Paper I) Vogler, A., Pietsch, W., & Kahabka, P. 1996, A&A, 305, 74 Wang, Q. D., Chaves, T., & Irwin, J. A. 2003, ApJ, 598, 969 Young, J. S., Allen, L., Kenney, J. D. P., Lesser, A., & Rownd, B. 1996, AJ, 112, 1903 4. Summary and Conclusions With a sample of 23 actively star forming galaxies studied in the X-ray regime and by implementing additional wave bands (Hα and UV), we found remarkably strong linear correlations between integrated 1.4GHz radio continuum, FIR, Hα, B-band, UV, and soft X-ray luminosities. Strong correlations also exist if soft X-ray luminosities are plotted against SFRs, LFIR/D2 25, or the energy input rate by SNe per unit area, expressed by EA. Integrated X-ray luminosities neither correlate with the H  nor with the baryonic mass of a galaxy. Strong correlations are also found if the diffuse soft X-ray luminosity of the halo is plotted against the FIR and Hα SFR or the SN energy input rate EA. These quantities are considerd to be the most important parameters for the creation of multi- phase halos. If a critical energy threshold exists and an instrumental detection bias is negligible, the present data suggest that a threshold of S FRFIR ≥ 1.0 M⊙/yr, S FRHα ≥ 0.1 M⊙/yr or EA ≥ 1 × 10−3 erg s−1 cm−2 needs to be exceeded in order to create multi-phase galaxy halos. Our results are in agreement with previous findings of a morphological and spatial coincidence of gaseous multi-phase halos (see Paper I). They clearly imply that multi-phase halos are the consequence of stellar feedback processes in the disk plane (e.g., de Avillez & Breitschwerdt, 2004, 2005) but con- flict with the concept of halos being due to infalling gas from the intergalactic medium (e.g., Benson et al., 2000; Toft et al., 2002; Pedersen et al., 2005). Acknowledgements. RT acknowledges financial support by Deutsches Zentrum fur Luft -- und Raumfahrt (DLR) through grant 50 OR 0102. We appreciate the comments pointed out by the anonymous referee. This work has made use of the SIMBAD database and of HyperLeda (http://leda.univ-lyon1.fr/). References Bell, E. F. 2003, ApJ, 586, 794 Bell, E. F., & de Jong, R. S. 2001, ApJ, 550, 212 Benson, A. J., Bower, R. G., Frenk, C. S., & White, S. D. M. 2000, MNRAS, 314, 557 Burstein, D., & Heiles, C. 1982, AJ, 87, 1165 Code, A., & Welch, G. 1982, ApJ, 256, 1 Condon, J. J. 1992, ARA&A, 30, 575 Condon, J. J., Helou, G., Sanders, D. B., & Soifer, B. T. 1990, ApJS, 73, 359 Condon, J. J., Helou, G., Sanders, D. B., & Soifer, B. T. 1996, ApJS, 103, 81 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 Dahlem, M., Lisenfeld, U., & Golla, G. 1995, ApJ, 444, 119 Dahlem, M., Petr, M., Lehnert, M. D., Heckman, T. M., & Ehle, M. 1997, A&A, 320, 731 Dahlem, M., Lazendic, J. S., Haynes, R. F., Ehle, M., & Lisenfeld, U. 2001, A&A, 374, 42 Dahlem, M., Ehle, M., Jansen, F., Heckman, T. M., Weaver, K. A., & Strickland, D. K. 2003, A&A, 403, 547 de Avillez, M. A., & Breitschwerdt, D. 2004, A&A, 425, 899 Table 1. Physical properties of the sample. Galaxy d [Mpc] (1) NGC 0891 NGC 3044 NGC 3221 NGC 3628 NGC 3877 NGC 4631 NGC 4634 NGC 4666 NGC 5775 M 82 NGC 0253 NGC 1482 NGC 3079 NGC 4244 NGC 4945 NGC 6503 NGC 0055 NGC 1511 NGC 3556 NGC 4522 NGC 4565 NGC 4656 NGC 5907 (2) 9.5 17.2 54.8 10.0 12.1 7.5 19.1 20.2 26.7 3.6 2.6 22.1 17.1 3.6 3.7 5.2 1.6 17.5 14.1 16.0 9.7 7.5 14.9 i [◦] (3) 88 84 77 80 83 86 83 80 84 82 79 58 85 85 78 75 80 72 79 79 89 90 86 L1.4GHz LFIR LB0 T LUV [1038 erg s−1] [1043 erg s−1] (6) (7) LHα LX,soft [1040 erg s−1] (8) (9) [M⊙ yr−1] (10) S FRFIR S FRHα Etot A LFIR/D2 25 (4) 1.19a 0.62b 3.29b 0.93b 0.10b 1.22b 0.21c 2.64b 2.82b 2.10b 0.53b 3.04b 5.88b 0.002c 1.51c 0.03b 0.02a 0.82 f 1.71b 0.10c 0.22b 0.02b 0.35b (5) 9.62 4.06 36.0 7.66 1.92 6.63 2.53 23.0 25.7 20.9 7.93 20.1 21.3 0.11 12.5 0.46 0.95 10.1 9.92 0.60 1.60 0.48 4.32 0.11 3.54h 5.59 3.61 -- 8.83 j 8.28 -- -- 2.30k 6.54 0.12 0.12 -- 2.34 14.3k 1.15 7.00 6.40 j 0.14 1.46 57.5d 7.56 1.01 8.05 -- 23.4l 14.4h 16.7 4.90h 3.42 10.4 -- 9.60k 9.00k 6.94 0.42k 0.54 5.05 -- 2.40k 0.76k 0.69 0.76 0.01 0.06 0.11 0.11 0.09 2.00m 0.79 3.21 -- 3.63n 7.60i 8.05 1.30o 1.42 1.20p 7.47 1.41v 1.78 5.57q 2.74 0.62 0.10 0.07 0.02 0.01 0.70r 0.09r 1.07r 0.74r 0.02r 0.78r 0.15r 1.42r 1.55r 5.15k 0.55k 3.70k 3.20k 0.007k 0.34k 0.07k 0.31s 0.26n 2.00t 0.89o 0.26p 0.03p 0.05p 4.32 1.83 16.2 3.45 0.86 2.99 1.14 10.3 11.6 9.39 3.57 9.65 9.60 0.05 5.62 0.21 0.12 4.54 4.46 0.27 0.72 0.22 1.89 0.28 0.70 -- 0.18 -- 1.13 0.51 4.55 1.85 1.14 0.39 0.76 0.71 0.03 0.19 0.06 0.16 0.29 0.60 0.10 0.09 0.11 0.44 [10−3 erg s−1cm−2] [1040 erg s−1 kpc−2] (11) 3.90u 0.94g 2.20u 12.2u 0.77u 0.83g 2.62u 4.46g 3.26g 19.6 2.74 64.9 33.6 0.05g 4.50u 1.00 0.11 0.34n 2.85e 0.58 0.09g 0.05g 0.14g (12) 6.92 8.52 13.5 8.24 5.14 7.76 12.4 31.9 24.6 151 18.3 80.7 13.7 0.36 27.1 4.04 1.12 32.4 6.96 2.00 0.80 0.44 1.63 Mbar MH i [1010M⊙] (13) (14) 10.6 3.28 1.63 11.3 1.28 2.56 1.63 5.44 6.09 1.86 10.6 3.58 14.1 0.62 5.24 1.17 5.65 0.92 3.97 1.55 16.5 1.81 15.4 0.45 0.64 1.70 0.38 0.17 0.68 0.05 0.61 0.97 0.16 0.60 0.06 1.10 0.09 0.09 0.13 0.16 0.54 0.93 0.05 1.30 0.39 1.38 Mdust [106M⊙] (15) 2.02 0.49 6.55 0.85 0.52 0.69 0.58 3.77 4.39 0.54 0.41 1.16 2.74 0.06 1.69 0.13 0.05 0.77 1.70 0.21 1.19 0.04 2.56 Notes: Cols. (1) to (3): Name, distance, and inclination of the target. Col. (4): Radio continuum luminosities measured at 1.49 GHz. Col. (5): FIR-luminosities calculated from revised IRAS S 60 and S 100 flux densities (Sanders et al., 2003). Col. (6): Total B0 T luminosities calculated from RC3 (de Vaucouleurs, 1991) and corrected for redshift, Galactic and internal extinction, assuming reddening values from (Burstein & Heiles, 1982). Col. (7): De-reddened UV-continuum luminosities measured at 1650Å (Rifatto et al., 1995) and 1550Å (Code & Welch, 1982), respectively. The UV flux for NGC 4656 has been determined at 2270Å (Maoz et al., 1996). No correction for redshift has been applied. Col. (8): Hα + [N ii]-luminosities, uncorrected for extinction and [N ]-contamination. Col. (9): Integrated (disk+halo) diffuse X-ray luminosities obtained from spectral fitting between 0.3 and 2.0 keV. For NGC 3221 and NGC 3877 disk only. Col. (10): Total star formation rate, based on FIR and Hα luminosities according to (Kennicutt, 1998). Col. (11): SN energy input rates per unit area (Dahlem et al., 1995; Condon, 1992) for M > 5M⊙. Col. (12): The optical diameter of the 25th magnitude isophote (D25) was taken from RC3 (de Vaucouleurs, 1991). Col. (13): Baryonic masses were calculated according to Bell & de Jong (2001). Col. (14): To derive the H  gas mass, we used MH i = 4.78 × 105 d2 S (H i), where S (H i), the total H  flux in units of [Jy km/s], has been taken from Huchtmeier & Richter (1989). Col. (15): Dust masses were calculated following Hildebrand (1983). References: (a) Condon et al. (1996); (b) Condon et al. (1990); (c) Condon et al. (1998); (d) Dahlem et al. (1997); (e) Irwin et al. (1999); ( f ) Dahlem et al. (2001); (g) Dahlem et al. (1995); (h) Bell (2003); (i) Young et al. (1996); ( j) Rossa & Dettmar (2000); (k) Strickland et al. (2004a); (l) (Lehnert & Heckman, 1996); (m) (Ferguson et al., 1996); (n) Dahlem et al. (2003); (o) Kenney & Koopmann (1999); (p) Vogler et al. (1996); (q) Rand (priv. com.); (r) Tullmann et al. (2006); (s) Schlegel et al. (1997); (t) Wang et al. (2003); (u) Rossa & Dettmar (2003); (v) Aldering (priv. com.). 8 R . T u l l m a n n e t a l . : T h e m u l t i - p h a s e g a s e o u s h a l o s o f s t a r f o r m i n g l a t e - t y p e g a l a x i e s
astro-ph/0302236
2
0302
2003-04-03T14:11:43
The Synoptic Swift Synergy -- Catching Gamma-Ray Bursts Before They Fly
[ "astro-ph" ]
The advent of large panoramic photometric surveys of the sky offers the possibly of exploring the association of gamma-ray bursts (GRBs) with supernovae. To date, a few gamma-ray bursts have been connected possibly with supernovae: GRB 980425 -- SN 1998bw, GRB 011121 -- SN 2001ke, GRB 970228 and GRB 980326. A combination of a large detection rate of GRBs and rapid coverage of a large portion of the sky to faint magnitude limits offers the possibility of detecting a supernova preceding an associated GRB or at least placing limits on the rate of association between these two phenomena and the time delay between them. This would provide important constraints on theoretical models for gamma-ray bursts.
astro-ph
astro-ph
The Synoptic Swift Synergy -- Catching Gamma-Ray Bursts Before They Fly Jeremy S. Heyl1,2 ABSTRACT The advent of large panoramic photometric surveys of the sky offers the pos- sibly of exploring the association of gamma-ray bursts (GRBs) with supernovae. To date, a few gamma-ray bursts have been connected possibly with super- novae: GRB 980425 -- SN 1998bw, GRB 011121 -- SN 2001ke, GRB 970228 and GRB 980326. A combination of a large detection rate of GRBs and rapid coverage of a large portion of the sky to faint magnitude limits offers the possi- bility of detecting a supernova preceding an associated GRB or at least placing limits on the rate of association between these two phenomena and the time delay between them. This would provide important constraints on theoretical models for gamma-ray bursts. Subject headings: gamma rays: bursts 1. Introduction The gamma-ray burst GRB 021004 was detected by HETE II at 12:06 UT on the 4th of October 2002 (Shirasaki et al. 2002). Observations after about 9 minutes from the trigger revealed a fading optical transient (Fox 2002), which was densely sampled in several bands, especially at early times. The afterglow of GRB 021004 has shown several unusual features (Mirabal et al. 2002; Salamanca et al. 2002; Moller et al. 2002; Bersier et al. 2003). Perhaps its most unique feature was that the field had been observed shortly before the gamma-ray burst itself was detected. Wood-Vasey et al. (2002) give a limiting unfiltered magnitude of 21.4 for observations on the day before the burst and 22.3 integrated over the year before the burst. Astronomy is on the threshold of a new era where large portions of the sky are surveyed deeply and regularly. The question arises what is likelihood of getting photometry of a 1Harvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138; [email protected] 2Chandra Fellow -- 2 -- gamma-ray burst precursor, specifically if supernovae precede gamma-ray bursts as in the supranova model (Vietri & Stella 1998)? Although the flux upper limits for GRB 021004 are not strigent enough to constain theoretical models of gamma-ray bursts, the high-burst localization rate of Swift combined with the fast sky coverage of the Sloan Digital Sky Survey (SDSS) and later Pan-STARRS, LSST and the Supernova-Acceleration Probe (SNAP) could provide important constraints on gamma-ray burst precursors. During the first and second years of operation of the Swift mission, SDSS will scan approximately 3000 square degrees (or 7% of the sky) each year (SDSS Collaboration 2001). Over this area it will detect point sources down to R ≈ 23.2. If Swift or subsequent missions are operational in 2006, the Pan-STARRS program will observe 20,000 square degrees every four days (or 50% of the sky) to a limiting magnitude of R ≈ 24.2 (Kaiser et al. 2002). Finally, potentially beginning in 2010, SNAP will cover 15 square degrees every four days with each observation reaching a limiting magnitude of R ≈ 28 (Kim et al. 2002), coadding observations over a month would go one magnitude deeper. The SNAP lensing survey will cover 300 square degrees over five months to a similar limiting magnitude. Long gamma-ray bursts are thought to be associated with the collapse of a massive star, a supernova. Specifically, in the collapsar model, the formation of a black hole in the center of the star results in relativistic jets which pierce the envelope of the star (MacFadyen & Woosley 1999). Along the axis of the jets, the collapsing star appears as a gamma- ray burst, and the supernova reaches its peak a few weeks after the GRB. Vietri & Stella (1998) proposed an alternative model in which the gamma-ray burst accompanies the delayed collapse of a quickly spinning neutron star which is more massive than the maximum mass of a non-rotating neutron star. The neutron star may take several months or years after the supernova to spin down to the critical frequency and collapse. In this Letter, I will estimate the number of gamma-ray burst events with photometry which overlaps on the sky but shortly precedes in epoch from SDSS and other surveys and compare the flux limits with the expected flux from a supernova which may precede the gamma-ray burst. 2. Gamma-Ray Burst Overlap with Future Surveys To calculate how often sufficiently deep photometry will precede the observation of a gamma-ray burst on the sky, several ingredients are required: a model for the spectral-energy distribution as a function of time of a supernova associated with a GRB, an estimate of the luminosity-rate function of GRBs as a function of redshift ( φ(z, L)), a model for the field of -- 3 -- view of the gamma-ray burst detector (ΩGRB = 2 for Swift) and its detection threshold (P1) and the rate of sky coverage of the photometric program (Rphoto) and its detection threshold (Rlim). Porciani & Madau (2001) provide models for φ(z, L) ≡ RGRB(z)ψ(L). The rate of GRBs, RGRB(z), is taken to be proportional to the star-formation rate, and the luminosity function of GRBs, ψ(L), is constrained by the BATSE GRB number counts. The rate of overlapping photometry is given by the product of the rate of sky coverage with a integral over the assumed cosmological distribution of GRBs, dNoverlap dt dNtotal dt (P > P1, R < Rlim) = (P > P1, R < Rlim) = Rphoto∆t dNtotal 4π ΩGRB dt (P > P1, R < Rlim) 4π Z z:R(z)=Rlim 0 dzZ ∞ L(P1,z) dL dV (z) dz φ(z, L) 1 + z . (1) (2) For lack of a better model for the evolution of a supernova associated with a GRB, I will assume that SN2001ke (Garnavich et al. 2003) is a prototype for this class, and furthermore that a supernovae associated with a GRB maintains its peak brightness for a period ∆t = 14(1 + z) days in the observer's frame and otherwise it is undetectable (see Reichart 1999; Bloom et al. 1999, for other GRB-associated supernovae). It is reasonable to use the median value of z for GRBs whose associated supernova are brighter than the magnitude limit of the particular photometric survey. However, to be highly conservative, I will take ∆t = 14 days to calculate the rate of overlap. If a survey covers the same area of sky more often than once per interval ∆t such as the SNAP supernovae search and Pan-STARRS, the rate of sky coverage Rphoto should only account for the first visit in each period ∆t; for example Rphoto for Pan-STARRS is 2π per fourteen days. The additional visits during each fortnight do not increase Rphoto but they do allow the survey to probe deeper by coadding the successive images. According to the original supranova model (Vietri & Stella 1998), the supernova may reach its peak at any time up to several years before the GRB, so this calculation implicitly assumes that both the GRB survey and the photometric survey will be operating at the appropriate times. L(P1, z) is the luminosity of a GRB at a redshift z which is detected at a count-rate of P1, and R(z) is the R-band apparent magnitude of a GRB-associated supernovae at a redshift z. Both of these functions include the k−correction (Hogg 1999) and assume the cosmographic parameters, ΩM = 0.3 and ΩΛ = 0.7. Fig. 1 plots the number of GRBs detected by Swift per year whose associated supernova would be brighter at its peak than a particular R-band magnitude. The results shown in Fig. 1 assume the SF1 model of Porciani & Madau (2001). This model provides a conservative lower limit for the overlap. It predicts that Swift will localize about 110 bursts per year -- the more generous estimates range up to 300 bursts per year -- 4 -- Fig. 1. -- The number of GRB-associated supernova brighter than a given R-magnitude. The lines show the cumulative contribution of GRBs above given flux limits. The right panel shows the entire distribution while the left panel focusses on the bright end. From bottom to top, only the supernovae associated with GRBs whose peak flux is above 100.9, 100.6, 100.5, 100.4, 100.3, 100.1, 1, 100.2, 10−0.25, 10−0.45, 10−0.5, 10−0.7 and 10−0.75 pho- tons per square centimeter per second. See Porciani & Madau (2001) for further details. -- 5 -- (Myers 2002). Furthermore, this model predicts that the bursts detected will be a higher red- shifts than other models, so the accompanying supernovae will be fainter and more difficult to detect. Table 1 gives the overlap rate between various photometric surveys and the Swift GRB localization mission. What is striking is that the shallow but wide Pan-STARRS and LSST surveys will perform much better than any of the other surveys. Furthermore, if super- novae precede GRBs, Pan-STARRS and LSST each will detect nearly ten GRB-associated supernovae per year. If it finds none, it would place severe constraints on the supranova model for GRBs. It must be emphasized that this rate of overlap is extremely conservative. It assumes a low Swift burst localization rate and a distribution of GRBs skewed to high redshift (therefore, faint assocated supernovae). The actual rate of overlap will probably be higher if both programs operate simultaneously. Furthermore, Swift will generate a catalog of burst positions and redshifts. One will be able to cross-correlate a posteriori this catalog with earlier Pan-STARRS or LSST observations and exclude the appearance of transients to R ≈ 25 over a wide range of epochs preceding the burst yielding definitive constraints on GRB-progenitors independent of assumptions about the GRB luminosity function and its evolution. This calculation of the overlap rate assumes that either the GRB localization program or the photometric survey studies random portions of the sky. In fact both the Swift mission and all of the photometric surveys avoid studying the region of the sky near the sun. Although the average rate of overlap over a year in given by the formulae above and the values in the tables, the chance of detecting the supernova associated with GRB is somewhat higher than average if the supernova precedes the GRB by less than three months or between nine and fifteen months. If the supernova precedes the GRB by six to nine months, the chance of detecting it is somewhat lower than average. However, this seasonal variation is smaller than the uncertainities in the GRB luminosity function. 3. Discussion The philosophy employed for finding gamma-ray burst precursors is somewhat different that what is necessary for finding supernovae or microlensing events. Because the precursors will be sought after the gamma-ray burst is detected and localized, it is not necessary to have more than one epoch of data from the particular region of sky before the burst. Even a single epoch would yield important constraints. Furthermore, unless the cadence of the observations is sufficiently low (no more than biweekly), the repeated observations of the same patch of sky do not improve the chances of catching a precursor (unless one coadds -- 6 -- the data to probe deeper), because supernova typically evolve over the course of weeks. Consequently, although supernova and microlensing surveys have a large data rate of high quality photometry, because of their relative lack of sky coverage and depth they do not contribute much to the detection rate of precursors. From another point of view, only a small fraction (< 10−4) of supernovae result in GRBs directed toward us, so one would typically have to find at least 104(4π/ΩGRB) supernovae in a blind search before finding a single GRB-associated supernova. The best bets are the large deep wide surveys of the sky. SDSS is the prototype and Pan- STARRS and LSST should deliver results. There is a small possibility that SDSS will catch a supernova before a gamma-ray burst providing important evidence for the supranova model for gamma-ray bursts (it may have done so already). Pan-STARRS or LSST if it overlaps with a high-locatization-rate GRB mission such as Swift will be able to provide important constraints on gamma-ray-burst models. Specifically, it will be able to exclude the possibility that GRBs follow supernovae within a year. I would like to acknowledge useful discussions with Kris Stanek, Robert Lupton and Bob Kirshner and helpful suggestions from the anonymous referee. I was supported by the Chandra Postdoctoral Fellowship Award # PF0-10015 issued by the Chandra X-ray Observatory Center, which is operated by the Smithsonian Astrophysical Observatory for and on behalf of NASA under contract NAS8-39073. REFERENCES Bersier, D. et al. 2003, ApJ, 584, L43 Bloom, J. S. et al. 1999, Nature, 401, 453 Fox, D. 2002, GCN, 1564 Garnavich, P. M. et al. 2003, ApJ, 582, 924 Hogg, D. 1999, astro-ph/9905116 Kaiser, N., Boesgaard, H., Heasley, J., Hodapp, K., Jewitt, D., Kudritzki, R.-P., Kuhn, J., Luppino, G., Magnier, E., Onaka, P., Pickles, A., Simon, T., Szapudi, I., Tholen, D., & Tonry, J. 2002, POI: The Panoramic Optical Imager, Tech. rep., Institute for Astronomy -- 7 -- Kim, A. et al. 2002, Wide-Field Surveys with the SNAP Mission, Tech. rep., Lawrence Berkeley Laboratory MacFadyen, A. I. & Woosley, S. E. 1999, ApJ, 524, 262 Mirabal, N., Armstrong, E. K., Halpern, J. P., & Kemp, J. 2002, GCN, 1602 Moller, P. et al. 2002, A&A, in press (astro-ph/0210654) Myers, J. D. 2002, Swift: Catching Gamma-Ray Bursts on the Fly, Tech. rep., NASA, http://swift.nasa.gov Porciani, C. & Madau, P. 2001, ApJ, 548, 522 Reichart, D. E. 1999, ApJ, 521, L111 Salamanca, I., Rol, E., Wijers, R., Ellison, S., Kaper, L., & Tanvir, N. 2002, GCN, 1611 SDSS Collaboration. 2001, Five-Year Baseline Plan for SDSS Operations, Tech. rep., Sloan Digital Sky Survey Shirasaki, Y. et al. 2002, GCN, 1565 Vietri, M. & Stella, L. 1998, ApJ, 507, L45 Wood-Vasey, W. M., Aldering, G., Lee, B. C., Helin, E. F., Pravdo, S., Hicks, M., & Lawrence, K. 2002, GCN, 1572 This preprint was prepared with the AAS LATEX macros v5.0. -- 8 -- Table 1. Present and Future Large-Scale Photometric Surveys Survey Rlim zmax zmed Rphoto∆t dNtotal/dt[y−1] dNoverlap/dt[y−1] SDSS Pan-STARSS (single) Pan-STARSS (coadded) LSST (single) LSST (coadded) SNAP SN (single) SNAP SN (coadded) SNAP lensing 23.2 24.2 25.0 24.5 25.1 28.0 28.8 28.0 0.56 0.78 1.00 0.86 1.04 2.59 3.35 2.59 0.48 0.66 0.83 0.72 0.85 1.45 1.50 1.45 0.035 6.3 6.3 6.3 6.3 0.0046 0.0046 0.0091 1.9 7.1 17. 10. 19. 98. 110. 98. 0.0052 3.6 8.8 5.1 9.8 0.036 0.039 0.071
astro-ph/0701187
1
0701
2007-01-08T12:55:33
Relativistic Outflows in Gamma-Ray Bursts
[ "astro-ph" ]
The possibility that gamma-ray bursts (GRBs) were not isotropic emissions was devised theoretically as a way to ameliorate the huge energetic budget implied by the standard fireball model for these powerful phenomena. However, the mechanism by which after the quasy-isotropic release of a few $10^{50} $erg yields a collimated ejection of plasma could not be satisfactory explained analytically. The reason being that the collimation of an outflow by its progenitor system depends on a very complex and non-linear dynamics. That has made necessary the use of numerical simulations in order to shed some light on the viability of some likely progenitors of GRBs. In this contribution I will review the most relevant features shown by these numerical simulations and how they have been used to validate the collapsar model (for long GRBs) and the model involving the merger of compact binaries (for short GRBs).
astro-ph
astro-ph
To appear in "Circumstellar Media and Late Stages of Massive Stellar Evolution (2006)"RevMexAA(SC) RELATIVISTIC OUTFLOWS IN GAMMA-RAY BURSTS M. A. Aloy1,2, and M. Obergaulinger2 RESUMEN La posibilidad de que las erupciones de rayos gamma (GRBs) no sean emisiones isotr´opicas fue considerada te´oricamente para atenuar el problema asociado a la gran cantidad de energ´ıa implicada por el modelo de bola de fuego est´andar para estos potentes fen´omenos. Sin embargo, el mecanismo por el cual, tras la deposi´on cuasi- isotr´opica de unos pocos 1050 erg se origina una eyecci´on colimada de plasma no pudo ser explicada de forma satisfactoria anal´ıticamente. La raz´on de ello radica en que la colimaci´on de un flujo saliente por su sistema progenitor depende de una din´amica no lineal muy compleja. Ello ha hecho necesario el uso de simulaciones num´ericas para arrojar algo de luz sobre la viabilidad de algunos de los progenitores m´as probables de GRBs. En esta contribuci´on revisar´e los hechos m´as relevantes mostrados por tales simulaciones num´ericas y c´omo ´estas han sido utilizadas para validar el modelos de estrella colapsante (para GRBs largos) y el modelo que implica la fusi´on de un sistema binario de objetos compactos (para GRBs cortos). ABSTRACT The possibility that gamma-ray bursts (GRBs) were not isotropic emissions was devised theoretically as a way to ameliorate the huge energetic budget implied by the standard fireball model for these powerful phenomena. However, the mechanism by which after the quasy-isotropic release of a few 1050 erg yields a collimated ejection of plasma could not be satisfactory explained analytically. The reason being that the collimation of an outflow by its progenitor system depends on a very complex and non-linear dynamics. That has made necessary the use of numerical simulations in order to shed some light on the viability of some likely progenitors of GRBs. In this contribution I will review the most relevant features shown by these numerical simulations and how they have been used to validate the collapsar model (for long GRBs) and the model involving the merger of compact binaries (for short GRBs). Key Words: GAMMA-RAYS: BURSTS -- ISM: JETS AND OUTFLOWS -- MHD 1. INTRODUCTION Our current understanding is that gamma-ray bursts (GRBs) are produced in the course of the birth of a stellar-mass black hole (BH). In other as- trophysical systems where accreting BHs fuel col- limated beams of plasma (e.g., AGNs and BH -- X- ray binaries), there is a direct evidence for relativis- tic outflows and jet collimation which comes from the imaging of the system. Therefore, it seems reasonable to assume as starting point, that also GRBs result from relativistic, collimated outflows from accreting, stellar-mass BHs. We have inferred that outflows yielding GRBs are relativistic because of a couple of observational constraints, namely, the detection of radio scintillation of the interstel- lar medium (Frail et al., 1997) and the measure- ment of superluminal proper motions in imaged af- terglows (Taylor et al., 2004). The ultrarelativistic 1Departamento de Astronom´ıa y Astrof´ısica, Uni- Espana de Valencia, versidad ([email protected]). 46100, Burjassot, 2Max-Planck-Institut fur Astrophysik, Karl-Schwarschild- Str. 1, 85741 Garching, Germany expansion is also necessary to overcome the theoret- ical constrain imposed by the compactness problem (Cavallo & Rees, 1978). However, we only have an indirect evidence of collimation based on the obser- vational constraint posed by the achromatic break in the afterglow light curve of some GRBs (e.g., Harrison et al., 1999). From the theoretical point of view, if GRBs are collimated events, the true emit- ted energy Eγ is reduced by a factor fΩ ≃ θ2/2 (Rhoads, 1999; Sari et al., 1999), i.e., Eγ = fΩEγ,iso, where Eγ,iso is the detected equivalent isotropic en- ergy. Nonetheless, the mechanism by which after the quasi-isotropic release of an amount of energy in the range 1048 − 1051 erg results in a collimated ejection has not satisfactory been explained analyt- ically. The reason being that the collimation of an outflow by the progenitor system depends on a very complex and non-linear dynamics. That has made necessary the use of numerical simulations in order to understand the collimation mechanism as well as to shed some light on the viability of some systems proposed to be the progenitors of GRBs. A robust 1 2 ALOY & OBERGAULINGER result obtained in numerical simulations of genera- tion of GRBs is that the progenitor system yields collimated outflows under rather general conditions independent on whether the outflow is initiated ther- mally (e.g., Aloy et al., 2000, 2005) or it is magnet- ically driven (McKinney, 2006). In this contribution I will review the most rele- vant features shown by these numerical simulations and how they have been used to validate the col- lapsar model (for long GRBs; § 2) and the model involving the merger of compact binaries (for short GRBs; § 3). 2. OUTFLOWS EMERGING FROM PROGENITORS OF LONG GRBS Among the plethora of models devised to ex- plain the origin of long GRBs (lGRBs), the most widely accepted was put forward by Woosley. In the original collapsar model, also known as failed super- nova model (Woosley, 1993), the collapse of a mas- sive (MZAMS ∼ 30 M⊙) rotating star that does not form a successful supernova but collapses to a BH (MBH ∼ 3 M⊙) surrounded by a thick disk. The vis- cous accretion of the disk matter onto the BH yields a strong heating that, in its turn, produces a co- pious amount of thermal neutrinos and antineutri- nos, which annihilate preferentially around the ro- tation axis producing a fireball of e+e−pairs and high energy photons. Later it was noted that, per- haps ν -- powered fireballs might not be sufficiently energetic to fuel the most powerful GRB events and, thus, the collapsar model was extended to ac- count for alternative energy extraction mechanisms (MacFadyen et al., 2001). More explicitely, the ac- cretion energy of the torus could be tap by suffi- ciently strong magnetic fields (hydromagnetic gen- eration) by means of the Blandford-Payne process (Blandford & Payne, 1982), or a non-trivial fraction of the rotational energy of the BH may also be con- verted into a Poynting flux (Blandford & Znajek, 1977). The scape of the newly born fireball and its ter- minal Lorentz factor (Γ∞) depend on structural and on dynamical factors. The critical structural fac- tors are the environmental baryon density in the funnel around the rotation axis of the star and the ability of the progenitor star to loose its outer Hydrogen envelope. An under-dense funnel forms along with the accretion torus if the specific an- gular momentum of the core of the star lies in the range 3 × 1016 cm2 s−1 <∼ j <∼ 2 × 1017 cm2 s−1 (MacFadyen & Woosley, 1999). The existence of the funnel is key to collimate the fireball and to permit its propagation through the progenitor. The most favourable conditions for the generation and prop- agation of the fireball happen when the density of the funnel (ρf ) is much smaller than the density of <∼ 10−4 − 10−3 the torus (ρtorus), namely, ρf /ρtorus (MacFadyen & Woosley, 1999). The likelihood that the progenitor star had lost its Hydrogen envelope depends on a number of factors which occurrence is still a matter of debate like, e.g., the genera- tion of stellar winds, the interaction with a com- panion (Podsiadlowski et al., 2004), etc. The rele- vance of the lost of the hydrogen envelope resides on the fact that, unless the density of the funnel is extremely small (as proposed by M´esz´aros & Rees, 2001), only a mildly relativistic, poorly-collimated fireball would reach the outer edge of the hydro- gen envelop after very long times and with relatively small Lorentz factors (Γ ≃ 2) implying that the ob- servational signature would be an X -- ray/UV tran- sient with a duration of ∼ 100 − 1000 s but not a GRB (MacFadyen et al., 2001). Finally, the most important dynamical factor setting Γ∞ is the the amount of baryons entrained as the fireball propa- gates through the stellar mantle. Several analytic works have made estimates about the collimation angle and the Lorentz fac- tor when the fireball breaks out the surface of the star (e.g., M´esz´aros & Rees, 2001). Nevertheless, the complexity inherent to the non-linear (mag- neto)hydrodynamic interaction of the fireball plasma with the stellar environment makes it unavoidable the use of numerical simulations. With these simu- lations we have been able to give preliminary answers to the following questions: Collimation. The generated outflows are inertially or magnetically confined. In the first case, the colli- mator is the funnel within the progenitor while in the second case, the flow is self-collimated by its own magnetic field if it is strong enough. Rather independently on the initial conditions and on the inclusion of magnetic fields, the typical outflow half- opening angles, when the jet reaches the surface of <∼ 5o. These small the progenitor star, are θbreak half-opening angles result from the recollimation of the outflow within the progenitor and they are in- dependent on whether the boundary conditions are set to initiate the jet with much larger half-opening angles (e.g., θ0 = 20o; Zhang et al., 2003) or on whether the jet is generated by an energy release into a volume spanning a half-opening angle θd = 30o much larger than θbreak (Aloy et al., 2000). In the course of their propagation through the progenitor the jets develop a non-homogeneous structure, trans- RELATIVISTIC OUTFLOWS IN GRBS 3 verse to the direction of motion, whose main fea- tures are an internal ultrarelativistic spine (where the Lorentz factor may reach Γcore ∼ 30 − 50 at jet breakout) within a half-opening angle of < 2o later- ally endowed by a moderately relativistic, hot shear layer (Γshl ∼ 5 − 10) extending up to θshl < 20o − 30o (Aloy et al., 2002). We point out that the transverse structure of the jet is nearly Gaussian both in sim- ulations including magnetic fields (e.g., McKinney 2006) or not including them (e.g., Aloy et al., 2000). However, a more accurate fit of the transverse struc- ture of the jet, cannot be accommodated by a simple Gauss function (Aloy, 2001). Variability. All produced outflows are highly variable due to the generation of Kelvin-Helmholtz (Aloy et al., 2000; G´omez & Hardee, 2004), shear- driven (Aloy et al., 2002) or pinch magnetohydrody- namic (MHD) instabilities (McKinney, 2006). Such extrinsic variability is independent on the (intrin- sic) variability of the energy source and leads to the formation of irregularities in the flow which are the seeds of internal shocks in the outflow. Ex- cept in cases in which the source may produce quasi-periodic variability (perhaps induced by pre- cession or nutation modes of the accretion disc), the extrinsic variability might be indistinguishable form the intrinsic one. Numerical simulations of three-dimensional (3D) relativistic jets propagating through collapsar-like environments show that such jets are also stable (Zhang et al., 2004) but it still re- mains unknown whether 3D relativistic, magnetohy- drodynamic, collapsar-jets will also be stable along its whole trajectory. Jet breakout. The jets generated are much lighter than their baryon reach environments. Thus, they propagate through the collapsar at moderate speeds (∼ 0.3c) and fill up thick cigar-shaped cavities or co- coons of shocked matter that also propagates along with the beam of the jet and that, eventually may break the surface of the collapsar. Since in the cocoon a few 1050 erg may be stored as the jet drills its way through the star, it has been pro- posed that its eruption through the collapsar surface could yield a number of γ-ray/X-ray/UV-transients (Ramirez-Ruiz et al., 2002). it has been proposed that GRBs are but one observable phe- nomenon accompanying black hole birth and other possibilities may arise depending on the observer's viewing angle with respect to the propagation of the ultrarelativistic jet (Woosley, 2000). Thus, in a sort of unification theory for high-energy transients, one may see progressively softer events ranging from GRBs (when the jet emergence is seen almost head Indeed, on) to UV flashes (when the jet eruption is seen at relatively large polar angles) and accounting for X-ray rich GRBs (XRR-GRBs) and X-ray flashes (XRF) at intermediate angles. The jet emergence through the stellar surface and its interaction with the stellar wind (which likely happens during the late stages of the evolution of massive stars) could lead to some precursor activity (MacFadyen et al., 2001). Furthermore, ν -- powered jets are very hot at breakout (∼ 80% of the total energy is stored in the form of thermal energy) which implies that jets can still experience an additional acceleration by conver- sion of thermal into kinetic energy, even if the energy source has ceased its activity. 2.1. lGRBs produced in collapsars: MHD- or neutrino-powered jets? From a aesthetic point of view, it is beautiful if any invoked jet powering mechanism explains not only events with relatively small Lorentz factors Γ ∼ 100 but also the occurrence of events with very large inferred values of Γ ∼ 500 (Lithwick & Sari, 2001) or even Γ ∼ 1000 as suggested by models for some GRBs (Soderberg & Ramirez-Ruiz, 2003). However, there are no fundamental reasons to argue in favour of a unique and universal mechanism to extract the energy stored in the progenitor. Purely hydrodynamic, ν-powered jets in collap- sars of type-I seem to be able only to produce more moderate terminal values of the bulk Lorentz factor (Γ ∼ 100 − 400; e.g., Aloy et al., 2000; Zhang et al., 2003), even if there is a further acceleration of the forward shock as a result of an appropriate den- sity gradient in the medium surrounding the pro- genitor (Aloy et al., 2000), unless the density in the funnel around the rotation axis is very small (M´esz´aros & Rees, 2001). In collapsars of type-II the mass accretion rate ( M ∼ 10−3 − 10−2 M⊙ s−1) is insufficient to produce a ν-powered jet, but it may suffice to generate MHD-powered jets whose observational signature might be weak, poorly col- limated X-ray/UV transients or very long (∼ 100 − 1000 s) GRBs of low Eiso and small Γ if the pro- genitor star is able to loose its Hydrogen envelope (MacFadyen et al., 2001). From extrapolations of the numerical results of axisymmetric jets generated electromagnetically (McKinney, 2006) very large values of the asymp- totic Lorentz factor of the outflow (Γ∞ ∼ 1000) can be attained. Thus, it seems rather plausible that MHD mechanisms have to be employed to generate jets with Γ ∼ 500 − 1000 (McKinney, 2005, 2006) and, invoking the existence of a universal energy ex- traction mechanism, one may also argue that also 4 ALOY & OBERGAULINGER log ρ [cgs] 12.08 10.88 9.67 100 9 . 6 9 50 2 0.6 1 13.28 14.49 10.62 9.01 100 5 . 1 1 50 1 1 . 5 4 10.5 1 1 . 1 0 . 5 5 10.10 log ρ [cgs] 11.20 6 . 9 12.30 13.40 5 . 0 1 1 2 . 4 15.1 1 4 . 2 15.1 4 . 2 1 12.4 13.3 4 . 2 1 1 3 . 3 4.2 1 4 . 2 1 1 3 . 3 50 100 1 0 1.5 1 . 5 1.5 1 13.3 1 2 . 4 1 3 . 3 4 . 2 1 0 r [km] 1 2.4 6 12.46 13.38 1 3 . 3 8 1 4 .3 1 5 . 2 3 1 14.31 13.38 14.31 13.38 11.54 13.38 14.31 ] m k [ z 0 50 100 100 12.46 50 0 r [km] 50 100 11.5 ] m k [ z 0 2.4 1 15.1 15.1 14.2 50 1 3 . 3 100 100 14.2 50 12.46 11.54 Fig. 1. Logarithm of the density (gray-scale) with overimposed magnetic field lines (white lines) and total magnetic field strength (black contours) corresponding to models collapsing stellar cores with a small initial rotational energy and rotating almost rigidly (left) and with a larger initial rotational energy and differentially rotating (right). Each panel displays the state of the stellar core after the bounce for four different initial magnetic field strengths (B0 = 1010 G, 1011 G, 1012 G and 1013 G in the clockwise direction starting from the upper left corner). The figures correspond to models of Obergaulinger et al. (2006a). jets with much more moderate terminal Lorentz fac- tors are also MHD-generated. Nevertheless, these arguments have a number of issues: 1.- The estimates of the terminal Lorentz factor of MHD-generated jets are based on axisymmetric models. The 3D-stability of relativistic magnetized jets is still a matter of debate which needs of much more complex numerical simulations than the ones produced so far. Should 3D-MHD jets be unsta- ble, should the terminal Lorentz factor not reliably be calculated with the state-of-the-art axisymmetric, general relativistic, MHD (GRMHD) simulations up to date (McKinney, 2006). 2.- GRMHD initial models of accreting BH sys- tems consist on more or less realistic matter distri- butions over which an assumed poloidal field is im- posed, i.e., they are not the result of a consistent magneto-rotational core collapse (e.g., Mizuno et al., 2004a,b; McKinney, 2006). The use of poloidal fields is triggered by the fact that purely toroidal field con- figurations do not yield the production of bipolar jets (De Villiers et al., 2005). On the other hand, the initial magnetic field strengths are assumed, not consistently computed from the core collapse of mas- sive stars. Typically, initial field strengths as large as B ∼ 1015 − 1016 G are used. Such large val- ues of B in combination with maximal values of the dimensionless angular momentum of the BH (jBH ∼ 1) are necessary in order to efficiently ex- tract energy via Blandford-Znajek (BZ) mechanism, because the BZ-power scales rather sensitive with B as EBZ ∼ 1050j2 BH(MBH/3 M⊙)2(B/1015 G)2 erg s−1, (Lee et al., 2000). 3.- When numerical simulations of the magne- torotational core collapse of massive stars are per- formed (see, e.g., Obergaulinger et al., 2006b,a, and references therein), bipolar outflows as well as thick toroidal structures surrounding a central, low density funnel in the collapsed core are only generated for >∼ 1012 G and for initial magnetic field strengths B0 rather fast3 and differentially rotating stellar cores (Fig. 1 right). Such initial field strengths and angular velocities are well beyond the ones predicted by the state-of-the-art calculations of rotating massive stars (Heger et al., 2005). These initially strongly magne- tized models likely develop collapsed cores, with a mass of Mc ∼ 1 M⊙ and a specific angular momen- tum Jc ∼ 1016 cm2 s−1, and may form a rapidly ro- tating BH with jBH ∼ 1. Furthermore, the winding up of the initial poloidal field leads to maximal field strengths Bmax ∼ 1015 G, being the field predomi- nantly toroidal Btoroidal/Bpoloidal ∼ 1 − 10. On the 3The rotational energy being as large as ∼ 4% of the grav- itational energy. RELATIVISTIC OUTFLOWS IN GRBS 5 other hand, initially rigid, more moderately rotating cores do not yield tori around low-density funnels, do not produce bipolar jets and the maximum field strengths are ∼ 5 × 1014 G (Fig. 1 left). Considering that the collapsed cores are smaller (Mc ∼ 0.75 M⊙) and slowly rotating (Jc ∼ 2 × 1015 cm2 s−1) than in the previous case one may expect that the newly born BH resulting from the posterior evolution of these kind of cores will not be a maximally rotating but, instead, they will form BHs with more moder- ate jBH ∼ 0.6. Thereby, the conclusion that seems to emerge from detailed numerical simulations of Ober- gaulinger et al. is that if the initial magnetic field strength and rotational energy is as small as pre- dicted by the most detailed stellar evolution models, the collapsed core does not hold the appropriate con- ditions (B >∼ 1015 G, jBH ∼ 1) to efficiently extract energy via BZ-mechanism and, conversely, unrealis- tically large initial magnetic fields and rotational en- ergies need to be invoked to expect a BZ-like mech- anism to operate efficiently. However, we have to be cautious in order not to extract too far reaching conclusions from the previously mentioned numer- ical work. It remains true that the results of the simulations of Obergaulinger et al. are handicapped because they do not include general relativity, be- cause the numerical resolution is still small to cap- ture all the relevant magneto-rotationally unstable modes and, because they are restricted to axisym- metric models. From the above points, one may infer that the dynamical relevance of the magnetic field in the pro- cess of energy extraction from the central source will depend on fine details of the magnetorotational collapse of the collapsar core. On the other hand, the process of ν ¯ν-annihilation as the primary source of energy that fuels an ultrarelativistic fireball also needs of a more careful study in order to know how much energy such a process may release in the pro- genitor system and how such an amount of energy depends on the physical conditions of the progenitor. A step towards such goal is the work of Birkl et al. (2006), which contributes to better understand how the energy deposition rate due to the process of ν ¯ν-annihilation ( Eν ¯ν ) depends on general relativis- tic (GR) effects and on different neutrinosphere ge- ometries in hyperaccreting stellar-mass BH systems. Birkl et al. consider two families of neutrinospheres. On the one side, idealized geometries as thin disks, tori, and spheres. On the other side, more realistic models are constructed as non-selfgravitating equi- librium matter distributions for varied BH rotation. Independent of whether GR effects are included, con- sidering the same values of temperature and sur- face area for an isothermal neutrinosphere, thin disk models yield the highest energy deposition rates by ν ¯ν-annihilation, while spherical neutrinospheres lead to the lowest ones. Considering isothermal neutri- nospheres with the same temperature and surface area, it turns out that compared to Newtonian cal- culations, GR effects increase the annihilation rate measured by an observer at infinity by a factor of 2 when the neutrinosphere is a disk (in agreement with the previous works; Asano & Fukuyama, 2001). However, in case of a torus and a sphere the influ- ence of GR effects is globally only ∼ 25%, although locally, particularly in the vicinity of the rotation axis of the system, it can be significantly larger. Fo- cusing on the dependence of the energy deposition rate on the value of jBH, it is found that increas- ing it from 0 to 1 enhances the energy deposition rate measured by an observer at infinity by roughly a factor of 2 due to the change of the inner radius of the neutrinosphere. Furthermore, although the ab- solute values of the energy deposition rate have to be taken with care (because of the steady state approx- imation used and the need of more realistic mod- els for the accretion disk; see Birkl et al., 2006, for accretion disks of mass similar to the one expected <∼ 0.01 M⊙) typically, in the collapsar model (Mdisk Eν ¯ν ∼ 1050 − 1051erg s−1. Even if only an small fraction (∼ 10%) of that energy were used to boost a polar outflow, there is fair chance for neutrinos to be the dominant energy source of the fireball (at least, in some cases when the magnetic field is not too large). The most likely scenario that can be de- vised is that both mechanisms (MHD and neutrino energy release) might be operating simultaneously. Indeed, for MHD-produced jets, neutrinos will play a fundamental role in the pair-loading of the jet (e.g., Levinson & Eichler, 1993) while, for ν-powered jets, the magnetic field may be important to collimate the thermally generated outflow. Which of the two en- ergy deposition mechanisms dominates in every sin- gle GRB will depend on the exact conditions in the precollapse progenitor. 3. OUTFLOWS EMERGING FROM PROGENITORS OF SHORT GRBS Nowadays, it is commonly believed that short GRBs (sGRBs) are generated after the merger of a system compact binaries formed by either two neutron stars (NSs) or a neutron star and a BH (Paczynski, 1986; Goodman, 1986; Eichler et al., 1989; Mochkovitch et al., 1993). The remnant left by the merger consists of a newly born BH black 6 ALOY & OBERGAULINGER hole girded by a thick gas torus from which it swal- lows matter at a hypercritical rate. In such situa- tion radiation is advected inward with the accretion flow and the cooling is dominated by the emission of neutrinos (Popham et al., 1999). As in the case of progenitors of lGRBs, these neutrinos might ei- ther be the primary energy source blowing a fire- ball of e+e− pairs and photons or to act as media- tor in hydromagnetic or electromagnetic energy ex- traction mechanisms (see Sect. 2) to pair-load the Poynting dominated outflow. A fundamental dif- ference with respect to the collapsar model is that the accretion thick disk is cannot be continuously refilled from a surrounding matter reservoir (the stel- lar matter in case of a collapsar) and, therefore, the duration of the produced outflows is, in part (see Aloy et al., 2005) roughly limited by the time dur- ing needed by the black hole to engulf most of the matter of the accretion disk, namely, a few 100 ms. This limit on the time scale, set by the ON time tce of the source, of holds for both ν-powered jets and for MHD-generated outflows. In the first case, the neutrino luminosity fades as the mass of the disk de- creases and, thereby, there will be a critical torus mass below which a plasma outflow cannot be sus- tained. In the second case, for the same reason, there will not be sufficient neutrinos that pair-load the Poynting dominated jet after a sizable fraction of the torus has been accreted. Although it seems likely that releasing a few 1049 erg above the poles of a stellar mass BH in a region of nearly vacuum may yield an ul- trarelativistic outflow, numerical simulations are needed to attempt to answer questions about the collimation mechanism of the polar outflow, the opening angle of the ultrarelativistic ejecta, the asymptotic Lorentz factors that can be attained, the internal structure of the outflow, the dura- tion of a possible GRB event, and the isotropic equivalent energy which an observer would infer by assuming the source to expand isotropically. These questions may at most be guessed, but they cannot be reliably answered on grounds of merger models and a consideration of their energy release by neutrino emission and the subsequent conversion of some of this energy by ν ¯ν-annihilation to e+e- pairs (Ruffert & Janka, 1999; Janka et al., 1999; 2002; Rosswog et al., Rosswog & Ramirez-Ruiz, 2003; Birkl et al., 2006). Only self-consistently time-dependent (magneto)hydrodynamic modeling may give us some insight on the former questions. The reason being that the relativistic outflow develops in a complex interaction with the accretion funnel such that torus, cleaning its own axial later energy deposition encounters a much reduced baryon pollution. Some keys to answer the questions mentioned in the previous paragraph have been very recently re- vealed by time-dependent numerical simulations in which the main results are: Collimation. The generated outflows which may yield GRB signatures are either collimated by the accretion disk (Aloy et al., 2005) or self-collimated by the magnetic field (McKinney, 2006), depending on whether the jet is initiated thermally or magnet- ically, respectively. The typical outflow half-opening angles are ∼ 3o − 25o and, as in the case of jets pro- duced in collapsars, the baryon-poor outflows dis- play a transverse structure. This structure shows a central core which spans a half opening angle θcore < 3o − 12o where Γcore > 100 flanked later- ally by a layer, extending up to θshl ∼ 25o where the Lorentz factor smoothly decays to moderately rel- ativistic values and where a sizable fraction of the total energy is stored. This layer is rather hot in thermally initiated outflows and has the potential of accelerating even after the energy release by the cen- tral engine has ceased. Similar to the jets produced in collapsar environments the transverse structure of the Lorentz factor could be roughly fit by Gaus- sian profiles. However, somewhat more complicated functions are required to provide more accurate fits (Aloy et al., 2005). Variability. Even injecting energy close to the BH even horizon at constant rates, the produced out- flows are highly variable. The interaction of the newborn fireball with the accretion torus yields the growth of Kelvin-Helmholtz (Aloy et al., 2005) in- stabilities. The variability in case of MHD jets is im- printed by pinch instabilities (McKinney, 2006). Up to date only axisymmetric models have been com- puted. All of which seem to be stable or marginally stable. It is not yet numerically verified whether 3D jets emerging from hyperaccreting BHs are stable. Influence of the environment. Mergers of com- pact objects may take usually place in the intergalac- tic medium or in the outer skirts of their host galax- ies. Thus, the environment of the merger may have a very low density. However, in the course of the merger, after the first contact of the two compact objects, there is an ejection of matter, which is larger close to the orbital plane. Such matter forms a cool baryon reach environment with a mass Mhalo < a few 10−2 M⊙ (Ruffert & Janka, 2001; Oechslin & Janka, 2006). The exact amount of mass ejected sensitively depends on whether the two compact objects are RELATIVISTIC OUTFLOWS IN GRBS 7 both NSs or one of them is a BH, on the initial mass ratio between the two, etc. Thereby, when the central BH forms, the newly born system may, in some cases, be embedded into a relatively high density halo. Mergers in low density environments may fuel ultrarelativistic outflows with the poten- tial to produce normal sGRBs, while in case that the merger occurs in high density media, the obser- vational signature is not a sGRB but, most likely, a flash in the soft X-ray or UV bands (Aloy et al., 2005). In the latter case, the resulting event might only be observable if it happens very close to the Milky Way. The fact that depending on the envi- ronmental density an sGRB can be produced or not has the direct implication that not every merger may yield a sGRB. This fact has to be considered when making estimates about the true rates of sGRBs and compared with the rates of NS+NS mergers (e.g., Guetta & Piran, 2005). Asymptotic Lorentz factor. The saturation value of the outflow Lorentz factor, Γ∞, is difficult to es- timate on the basis of numerical simulations that cover only the initial fraction of a second in the evo- lution of an ultrarelativistic outflow generated in a hyperaccreting BH. Despite this difficulties rough es- timates can be made. For instance, in case that the outflows are thermally generated (Aloy et al., 2005), there is a clear trend to produce much higher val- >∼ 500 − 1000) than for ues of Γ∞ for sGRBs (Γ∞ lGRBs (Γ∞ ∼ 100). The reason for the difference re- sides on the much smaller density of the environment of the merger (even accounting for the mass ejected from the compact objects after the first contact; see above) as compared with the baryon-polluted envi- ronment that a relativistic jet finds inside a collaps- ing massive star. It has been speculated that this difference in Lorentz factor might be the reason for the paucity of soft sGRBs (Janka et al., 2006). The former trend seems not to be followed by MHD- generated outflows (McKinney, 2006). The most likely reason being that McKinney's simulations are set to be scale free, while there should be a big dif- ference between the environments of mergers of com- pact objects and the interior of collapsars. Proba- bly, such a difference cannot be accommodated easily with simple scale-free power-laws for the distribution of the physical variables. Duration of the events. As pointed out by Aloy et al. (2005) and Janka et al. (2006), the "shells" ejected by the central engine, accelerate much faster in the leading part of the outflow than the shells in its lagging part. The rear shells therefore need a longer time to reach velocities v ≃ c. This differential ac- celeration at early and late times of the relativistic jet leads to a stretching of the overall radial length of the outflow, ∆, relative to tce times the speed of light c, ∆ > ctce. This stretching has the important consequence that the overall observable duration of the GRB (in the source frame), T = t∆ = ∆/c, may be a factor of 10 or more longer than tce, even when the GRB is produced by internal shocks. 4. SUMMARY The numerical modelling of progenitors of GRBs has allowed us to gain some insight into a num- ber of important issues related with the nature of the outflows produced by these systems. First, it has allowed us to verify that some (but probably not all) collapsars can yield collimated relativistic outflows that turn into lGRBs at large distances from the source. Likewise, only a fraction of the mergers of compact objects may yield sGRBs. Sec- ond, the numerical modeling has given us informa- tion about the collimation mechanism and typical outflow opening angles. Third, it has shown that the outflow is heterogeneous both along the direc- tion of propagation and transverse to it, even if the central engine releases energy at a constant rate. Fourth, it seems rather plausible that some lGRBs with very high Lorentz factor (Γ >> 100) need of a MHD jet-formation mechanism. On the other hand, some other lGRBs with more moderate Lorentz fac- tors (Γ ∼ 100) can be explained by ν-powered jets, particularly if tce < 10 s. Fifth, MHD- and ν- mechanisms may work simultaneously. Therefore, it is likely that, in some cases MHD processes domi- nate the jet generation while in others neutrinos may be the dominant energy source. The numerical modeling done so far is still insu- ficient. In order to start from consistent initial mod- els, the state-of-the-art numerical codes must incor- porate, at least, the effects of strong gravity, mag- netic fields and a detailed neutrino transport. In the near future we may see how all these elements are in- cluded in more realistic numerical experiments that will deepen our understanding on how ultrarelativis- tic outflows are produced in progenitors of GRBs. Aloy, M. A. 2001, References in Highlights of Spanish As- trophysics II, eds. J. Zamorano, J. Gorgas, and J. Gallego, p. 33 Aloy, M. A., Muller, E., Ib´anez, J. M., Mart´ı, J. M., & MacFadyen, A. 2000, ApJL, 531, L119 Aloy, M. A., Ib´anez, J.-M., Miralles, J. A., & Urpin, V. 2002, A&A, 396, 693 8 ALOY & OBERGAULINGER Aloy, M. A., Janka, H.-Th., & Muller, E. 2005, ApJ, Mizuno, Y. , Yamada, S., Koide, S., & Shibata, K. 436, 273 2004a, ApJ, 606, 395 Asano K., & Fukuyama, T. 2001, ApJ, 546, 1019 Mizuno, Y. , Yamada, S., Koide, S., & Shibata, K. Birkl, R., Aloy, M. A., Janka, H. -Th., & Mueller, E. 2006, astro-ph/0608543 Blandford, R. D., & Znajek, R. L. 1977, MNRAS, 179, 433 Blandford, R. D., & Payne, D. G. 1982, MNRAS, 199, 883 Cavallo G., & Rees, M. J. 1978, MNRAS, 183, 359 2004b, ApJ, 615, 389 Mochkovitch, R., Hernanz, M., Isern, J., & Martin, X. 1993, Nat., 361, 236 Obergaulinger, M., Aloy, M. A., Dimmelmeier, H., & Muller, E. 2006a, ApJ, 457, 209 Obergaulinger, M., Aloy, M. A., & Muller, E. 2006b, ApJ, 450, 1107 Oechslin, R., & Janka, H.-Th. 2006, MNRAS, 368, De Villiers, et al. 2005, ApJ, 620, 878 1489 Eichler, D., Livio, M., Piran, T., & Schramm, D. N. Paczynski, B. 1986, ApJL, 308, L43 1989, Nat., 340, 126 Frail, D. A., Kulkarni, S. R., Nicastro, S. R., Feroci, M., & Taylor, G. B. 1997, Nat., 389, 261 G´omez, E. A., & Hardee, P. E. 2004, in AIP Conf. Proc. 727: Gamma-Ray Bursts: 30 Years of Dis- covery, eds. E. Fenimore & M. Galassi, p. 278 Goodman, J. 1986, ApJL, 308, L47 Guetta, D. & Piran, T. 2005, ApJ, 435, 421 Harrison, F. A., et al. 1999, ApJL, 523, L121 Podsiadlowski, P., et. al. 2004, ApJL, 607, L17 Popham, R., Woosley, S. E., & Fryer, C. 1999, ApJ, 518, 356 Ramirez-Ruiz, E., Celotti, A., & Rees, M. J. 2002, MNRAS, 337, 1349 Rhoads, J. E. 1999, ApJ, 525, 737 Rosswog, S. & Ramirez-Ruiz, E. 2002, MNRAS, 336, L7 Rosswog, S., Ramirez-Ruiz, E., & Davies, M. B. Heger, A., Woosley, S. E., & Spruit, H. C. 2005, 2003, MNRAS, 345, 1077 ApJ, 626, 350 Janka, H.-Th., Aloy, M. A., Mazzali, P. A., & Pian, E. 2006, ApJ, 645, 1305 Janka, H.-Th., Eberl, Th., Ruffert, M., & Fryer, C. L. 1999, ApJ, 527, L39 Lee, H. K., Wijers, R. A. M. J., & Brown, G. E. 2000, Phys. Rep., 325, 83 Levinson, A., & Eichler, D. 1993, ApJ, 418, 386 Ruffert, M., & Janka, H.-Th. 2001, ApJ, 380, 544 Ruffert, M., & Janka, H.-Th, 1999, ApJ, 344, 573 Sari, R., Piran, T., & Halpern, J. P. 1999, ApJL, 519, L17 Soderberg, A. M., & Ramirez-Ruiz, E. 2003, MN- RAS, 345, 854 Taylor, G. B., Frail, D. A., Berger, E., & Kulkarni, S. R. 2004, ApJL, 609, L1 Lithwick, Y., & Sari, R. 2001, ApJ, 555, 540 Woosley, S. E. 1993, ApJ, 405, 273 MacFadyen, A. I., & Woosley, S. E. 1999, ApJ, 524, 262 MacFadyen, A. I., Woosley, S. E., & Heger, A. 2001, ApJ, 550, 410 McKinney, J. C. 2005, ApJL, 630, L5 McKinney, J. C. 2006, MNRAS, 368, 1561 Woosley, S. E. 2000, in AIP Conf. Proc. 526: Gamma-ray Bursts, 5th Huntsville Symposium, eds. R. M. Kippen, R. S. Mallozzi, & G. J. Fish- man, p. 555 Zhang, W., Woosley, S. E., & MacFadyen, A. I. 2003, ApJ, 586, 356 Zhang, W., Woosley, S. E., & Heger, A. 2004, ApJ, M´esz´aros, P., & Rees, M. J. 2001, ApJL, 556, L37 608, 365
astro-ph/9903092
1
9903
1999-03-05T17:27:25
NGC 3576 and NGC 3603: Two Luminous Southern HII Regions Observed at High Resolution with the Australia Telescope Compact Array
[ "astro-ph" ]
NGC 3576 (G291.28-0.71; l=291.3o, b=-0.7o) and NGC 3603 (G291.58-0.43; l=291.6o, b=-0.5o) are optically visible, luminous HII regions located at distances of 3.0 kpc and 6.1 kpc, respectively. We present 3.4 cm Australian Telescope Compact Array (ATCA) observations of these two sources in the continuum and the H90a, He90a, C90a and H113b recombination lines with an angular resolution of 7" and a velocity resolution of 2.6 km/s. All four recombination lines are detected in the integrated profiles of the two sources. Broad radio recombination lines are detected in both NGC 3576 (DV_{FWHM}>= 50 km/s) and NGC 3603 (DV_{FWHM}>=70 km/s). In NGC 3576 a prominent N-S velocity gradient (~30 km/s/pc) is observed, and a clear temperature gradient (6000 K to 8000 K) is found from east to west, consistent with a known IR color gradient in the source. In NGC 3603, the H90a, He90a and the H113b lines are detected from 13 individual sources. The Y^+ (He/H) ratios in the two sources range from 0.08+/-0.04 to 0.26+/-0.10. We compare the morphology and kinematics of the ionized gas at 3.4 cm with the distribution of stars, 10 micron emission and H_2O, OH, and CH_3OH maser emission. These comparisons suggest that both NGC 3576 and NGC 3603 have undergone sequential star formation.
astro-ph
astro-ph
NGC 3576 and NGC 3603: Two Luminous Southern HII Regions Observed at High Resolution with the Australia Telescope Compact Array C. G. De Pree1, Melissa C. Nysewander1,2, W. M. Goss2 1Department of Physics & Astronomy, Agnes Scott College, 141 E. College Ave, Decatur, GA 30030 2National Radio Astronomy Observatory, P.O.Box 0, Socorro, NM 87801 ABSTRACT NGC 3576 (G291.28-0.71; l = 291.3o, b = -0.7o) and NGC 3603 (G291.58-0.43; l = 291.6o, b = -0.5o) are optically visible, luminous HII regions located at distances of 3.0 kpc and 6.1 kpc, respectively. We present 3.4 cm Australian Telescope Compact Array (ATCA) observations of these two sources in the continuum and the H90α, He90α, C90α and H113β recombination lines with an angular resolution of 7′′ and a velocity resolution of 2.6 km s−1. All four recombination lines are detected in the integrated profiles of the two sources. Broad radio recombination lines are detected in both NGC 3576 (∆VF W HM ≥ 50 km s−1) and NGC 3603 (∆VF W HM ≥ 70 km s−1). In NGC 3576 a prominent N-S velocity gradient (∼30 km s−1 pc−1) is observed, and a clear temperature gradient (6000 K to 8000 K) is found from east to west, consistent with a known IR color gradient in the source. In NGC 3603, the H90α, He90α and the H113β lines are detected from 13 individual sources. The Y+ (He/H) ratios in the two sources range from 0.08±0.04 to 0.26±0.10. The H113β/H90α ratio in NGC 3576 is close to the theoretical value, suggesting that local thermodynamic equilibrium (LTE) exists. This ratio is enhanced for most regions in NGC 3603; enhanced β/α ratios in other sources have been attributed to high optical depth or stimulated emission. We compare the morphology and kinematics of the ionized gas at 3.4 cm with the distribution of stars, 10µm emission and H2O, OH, and CH3OH maser emission. These comparisons suggest that both NGC 3576 and NGC 3603 have undergone sequential star formation. Subject headings: HII regions: kinematics and dynamics − ISM: abundances individual (NGC 3576, NGC 3603) − HII regions: 1. Introduction NGC 3576 (D = 3.0 ± 0.3 kpc) and NGC 3603 (D = 6.1 ± 0.6 kpc) are two of the highest luminosity optically visible HII regions in our Galaxy (Goss & Radhakrishnan, 1969). Indeed – 2 – NGC 3603 (with a bolometric luminosity of 107 L⊙) contains some 104 M⊙ of ionized gas, and aside from W49A, may be the most massive HII region in the Galaxy (Eisenhauer et al. 1998). Sources such as these are ideal laboratories to study the formation and evolution of massive stars and the initial mass function (IMF) of giant star forming regions. Similar regions that are visible with the Very Large Array (VLA) have been studied in detail. In particular, radio recombination line (RRL) studies of the Sgr B2 and W49A star forming regions with the VLA (Gaume et al. 1995; De Pree et al. 1995: Mehringer et al. 1995) have revealed small-scale outflows, very broad recombination lines (∆VF W HM ∼80 km s−1), expanding shells of ionized gas, and extremely compact structures. NGC 3576 and NGC 3603 appear to be similar regions in terms of luminosity, total mass and gas kinematics. Existing optical studies of NGC 3603, for example, have revealed broad emission lines and a stellar wind bubble, indicative of the early stages of a star forming region coming into equilibrium with its environment (Clayton 1986). Though close in projection (∼40′), the two regions are separated by about 30 km s−1 in radial velocity as indicated by HI absorption (Goss & Radhakrishnan 1969) and hydrogen recombination line measurements (Wilson et al. 1970). Both regions are partially obscured by foreground dust, and studies of the kinematics of the associated ionized gas have been limited to optical wavelengths (Clayton 1986; Clayton 1990; Balick et al. 1980). Thus, existing velocity fields of the ionized gas are incomplete. The low resolution (4′) 6 cm image of Goss & Shaver (1970) shows that the radio emission associated with the two regions extends well beyond the detected Hα emission. In both sources, however, it appears that the large scale ionized structures may be driven by energetic processes occuring near their cores. Evidence from a variety of wavelengths (including the radio line and continuum data from this study) indicate that both regions may have undergone sequential star formation, in NGC 3576 from east to west, and in NGC 3603 from north to south. NGC 3603 and NGC 3576 have not been previously imaged with the Australia Telescope Compact Array (ATCA). Radio observations have been limited to low resolution line and continuum observations such as the ∼1′ observations of Retallack & Goss (1980) at 21 cm. Thus, our radio frequency observations (θF W HM ∼7′′) improve by factors of ∼10 the best spatial resolution radio observations of the ionized gas associated with NGC 3576 and NGC 3603. For both NGC 3576 and NGC 3603, we present high resolution radio continuum images and radio recombination lines. New kinematic distances which are in reasonable agreement with the previously derived spectroscopic distances are derived from the observed line velocities and a revised Galactic radius of R0=8.5 kpc. We examine the ionized gas morphology and kinematics in the two sources, and possible star formation histories of these two regions. In addition, we derive electron temperatures, H113β/H90α ratios, helium abundances, and (where possible) carbon line strengths. – 3 – 2. Observations and Data Reduction Radio continuum and recombination line observations were made for NGC 3576 and NGC 3603 at the Australian Telescope Compact Array (ATCA) in the (750 m) B and (750 m) D configuration in two observing runs on the 13th and 19th of June in 1995 with a total of 24 hours of observing time. The observing frequency was 8.873 GHz, close to the the frequency of the H90α recombination line. The wide bandwidth (16 MHz) allowed the simultaneous observation of three additional recombination lines: He90α, C90α, and H113β. Detailed observing parameters are presented in Table 1. The initial calibrations were made using the AIPS (Astronomical Image Processing System) software distributed by NRAO. Each data set was flagged and the flagged continuum data were then calibrated using the observed phase and flux density calibrators. These flags and calibrations were copied to the line data, where a bandpass calibration was applied and the continuum was subtracted from the u-v plane using the AIPS task UVLIN. Continuum images (with angular resolution of ∼7′′) were made from the line free channels for each source using the AIPS task IMAGR. This spatial resolution corresponds to a linear size of approximately 0.09 pc for NGC 3576 (D= 3.0 kpc) and approximately 0.21 pc for the more distant NGC 3603 (D= 6.1 kpc). The noise in the continuum images is dominated by the dynamic range limitations; thus the rms noise is significantly reduced in the continuum-subtracted line data. After the initial data reduction in AIPS, the line and continuum data were analysed using GIPSY (the Groningen Image Processing System). Integrated line profiles were made for each pixel above a continuum cutoff level for both NGC 3576 (5σ = 90 mJy beam−1) and NGC 3603 (5σ = 55 mJy beam−1) using the GIPSY task PROFIL. Gaussian functions were then fitted to the spectral line data. The output from these fits are the line to continuum ratio (TL/TC), the integrated line strength, the central velocity (VLSR) and the velocity full width at half maximum (∆VF W HM ) of the line emission in each region. The HII region NGC 3576 consists of a single extended source (containing several local bright spots) with an angular size of ∼75′′ (∼0.7 pc). In NGC 3603, thirteen individual sources were observed above a 5σ level in the continuum image. 3. Results 3.1. 3.4 cm Continuum Emission Continuum images at 3.4 cm were made for both NGC 3576 and NGC 3603 using the AIPS task IMAGR. Figure 1 is the resulting 3.4 cm continuum image for NGC 3576. The crosses in the plot indicate the positions of the 10 µm sources tabulated by Frogel & Persson (1974). Table 2 lists the continuum parameters for NGC 3576. These parameters include peak flux density (Speak), integrated flux density (Stot), angular diameter in arcseconds (θ), and linear radius in pc (r). Using the Parkes 64-m telescope, McGee & Newton (1981) report a 14.7 GHz flux density of 97.8 Jy in NGC 3576. We detect about 73% of the single dish flux density, or 71±7 Jy. From the – 4 – 13 sources in NGC 3603, we detect a total flux density of 25 Jy, or 38% of the reported single dish flux density of 65.9 Jy. Figure 2 shows the 3.4 cm continuum image of NGC 3603. The crosses indicate the positions of the known 10 µm sources in NGC 3603 (Frogel et al. 1977). Each emission peak higher than the continuum 5σ cutoff (55 mJy beam−1) has been designated as a separate region. The regions are labeled A to M, corresponding to increasing right ascension. The AIPS task JMFIT fitted two dimensional Gaussians to the individual continuum sources in NGC 3603. Table 2 presents continuum parameters for each region (A-M) in NGC 3603. These parameters include peak flux density (Speak), integrated flux density (Stot), deconvolved geometrical mean diameter in arcseconds (θ), and the linear radius in pc (r). Table 3 presents derived physical parameters for the regions in NGC 3603. These calculated parameters include electron density (ne), emission measure (EM ), mass of ionized gas (MHII ), excitation parameter (U ), Lyman continuum photon flux (NLyC ) and continuum optical depth (τc). The continuum parameters were calculated assuming a uniform density, spherical, ionization bounded HII region, using the corrected formulae of Mezger & Henderson (1969). The calculated values are of course dependent on these assumptions, and in a complex region like NGC 3603, these assumptions may be overly simplistic. Figures 1 and 2 are plotted in B1950 coordinates. The last two figures in the paper (Figures 8 and 9) are plotted in J2000 coordinates to help in comparisons between the radio continuum emission and images at other frequencies, since images in the literature are shown in both epochs. 3.2. Radio Recombination Lines 3.2.1. NGC 3576 A continuum-weighted, integrated line profile was generated for NGC 3576 using the GIPSY task PROFIL. The integrated profile was generated for all pixels above a continuum 5σ cutoff level. In addition, integrated profiles were generated for each pixel above the continuum 5σ level in a 7′′×7′′ box centered on each of the positions of the Irs sources tabulated by Frogel & Persson (1974). The integrated source profile, and the four profiles line profiles were then fitted with Gaussian functions using the GIPSY task PROFIT. Figure 3a presents the integrated line profile and model fit for NGC 3576. Each of the four lines (H90α, He90α, C90α and H113β) is clearly detected, and labeled in Figure 3a. The integrated spectra for the Irs sources (Irs1/2, Irs3, Irs4 and Irs5; Frogel & Persson 1974) in NGC 3576 are shown in Figures 4a-d. The spectra in Fig. 3a and Figs. 4a-d show the data (points), Gaussian fits (solid line), and the residuals (dotted line) for each recombination line. The parameters of Gaussian fits to the lines (VLSR, ∆VF W HM , and Tl/Tc) are presented in Table 4. From the integrated emission from NGC 3576, we determine a helium abundance of Y +=0.09±0.01, and a H113β/H90α ratio close to the theoretically predicted value of 0.276 (β/α=0.25±0.01). The electron temperature in NGC 3576 (T∗ e) is calculated assuming local – 5 – thermodynamic equilibrium (LTE) from both the H90α and H113β recombination lines (using equation [22] from Roelfsema & Goss [1992]): e = (6943ν1.1 T ∗ Tc Tl∆VF W HM 1 1 + Y + )0.87K, (1) with line width (∆VF W HM ) given in km s−1and ν in GHz. These LTE temperatures are given in Table 4. An electron temperature (Te) may also be calculated with corrections made for non-LTE effects (using Eq. [23] from Roelfsema & Goss [1992]): Te = T ∗ e [bn(1 − βnτc 2 )]0.87K, (2) where bn and βn values are from the tables of Salem & Brocklehurst (1979). For NGC 3576, a turbulent velocity (vturb) and an electron temperature (Te) are calculated by solving two equations for line width (Eq. 3) using the hydrogen and helium (H90α and He90α) recombination line widths. The line width can be expressed as < v2 turb >= 3∆ V 2 H 8ln2 − 3kTe M (3) where the vturb is the turbulent velocity in km s−1, k is the Boltzmann constant, M is the mass of the atom, Te is the electron temperature, and ∆VH is the line FWHM in km s−1. Using this method for NGC 3576, we derive a turbulent velocity vturb=17±1.0 km s−1 and an electron temperature of (Te) of 6100±200 K. In order to examine the spatial variation in the H90α line parameters, single Gaussian fits were made to each pixel above the continuum 5σ level in NGC 3576. The results of these point-to-point fits to the H90α line are presented in Figures 5a and 5b. Figure 5a shows the spatial variation in the the central line velocity (VLSR), with continuum contours overlaid on the greyscale representation of the ionized gas velocity. Figure 5b shows the spatial variation in the linewidth (∆VF W HM ) of the H90α line. For NGC 3576, a map of the LTE electron temperature was also generated, using Equation 1 (above) for each point above 10σ in the continuum image. The resulting LTE temperature image is shown in Figure 5c, indicating a clear E-W temperature gradient. This temperature gradient has also been plotted using the AIPS task IRING, which averages emission in concentric semicircles. The semicircles (with a thickness of ∼7′′), are centered on the continuum peak of NGC 3576 and extend to the east. A plot of temperature as a function of distance from the peak (in arcsec) is presented as Figure 6. The apparent temperature gradient is discussed in §4.2.1. 3.2.2. NGC 3603 An integrated line profile was generated for the NGC 3603 region above a 5σ cutoff in the continuum emission. Figure 3b presents the integrated line profile and model fit for NGC 3603. – 6 – Because of the large variation in line velocities across the source, the carbon line (apparent in some of the individual source profiles in NGC 3603) was fitted with the line width fixed at 10 km s−1. Gaussian fits were made to the integrated profile, and the parameters of these fits are given in Table 5. Figures 7a-7m present the radio recombination line profiles and Gaussian fits for each of the thirteen subregions (A-M) in NGC 3603. For all 13 subregions, the signal-to-noise ratio is sufficiently high to fit the H90α, He90α, and H113β lines. The C90α recombination line was successfully fit only in regions NGC 3603 B and H with a fixed carbon line width. Sources in which parameters were fixed and the fixed values are indicated parenthetically in Table 5. In a few cases, the fitted linewidth (∆VF W HM ) of the He90α line may be broad due to blending with the C90α line in the sources where a C90α line was not separately fitted. The line parameters from the Gaussian fits to the integrated emission from NGC 3603 and its 13 subregions are presented in Table 5. As with NGC 3576, the table also includes the LTE electron temperature (T ∗ e ), the corrected electron temperature (Te), the ionized helium to hydrogen ratio (Y +), and the H113β to H90α ratio (β/α), and the turbulent velocity (vturb). For NGC 3603, the non-LTE electron temperature was calculated using Equation 2 (above). The turbulent velocity was calculated from the width of the H90α line and the non-LTE electron temperature using Eq. 3 (above). These temperatures and turbulent velocities are given in Table 5. 4. Discussion 4.1. Distances to NGC 3576 and NGC 3603 McGee & Newton (1981) detected recombination lines at 14.7 GHz from 25 high emission measure HII regions using the Parkes 64-m radio telescope. Both NGC 3576 (1109-610) and NGC 3603 (1112-610) were included in their survey. McGee & Newton detected the H76α line from NGC 3576 and NGC 3603 at -24.1±0.1 km s−1 and 7.7±0.1 km s−1 respectively. In addition to H76α, McGee & Newton detected the He76α line toward both sources, and the carbon line toward NGC 3576. For NGC 3576, the detected line width in the H90α line (∆VF W HM =28.7± 0.1) is in good agreement with the H76α line width published by McGee & Newton (∆VF W HM =29.1±0.1). The integrated H90α line width for NGC 3603 (∆VF W HM =36.9±0.25) is in reasonable agreement with the value for the H76α line given by McGee & Newton (∆VF W HM =41.9±0.2). The distance to NGC 3576 has been discussed at length in two papers, Goss & Radhakrishnan (1969), and Persi et al (1994). Goss & Radhakrishnan found a kinematic distance of 3.6 kpc, just beyond the tangential point of the Galaxy, using the Schmidt galactic model and a value of R0 = 10 kpc. This distance is derived from a line center velocity of -23 km s−1, and circular motion within the galaxy. This kinematic distance agrees well with the findings from Caswell & Haynes (1987). Our new observations, using a revised R0 = 8.5 kpc, and a line velocity of -20.0 ± 0.1 km s−1 suggest a distance of 3.0±0.3 kpc for NGC 3576. Because no ionizing cluster is optically visible, no reliable spectroscopic distance has been determined. Humphreys (1972), in a study of – 7 – the effect of streaming in the Carina arm, determined a distance of 3.2 kpc for NGC 3576, and detected both velocity variations (∼4 km s−1 ) and a large spread in the individual velocities of early type stars in this direction. Persi et al. (1994) derive a distance based on the star HD 97499 (position plotted in Figure 8). However, the high resolution radio continuum image of the source (Fig. 8) shows that there may not be a direct association between HD 97499 and NGC 3576. HD 97499 (RA = 11 12 10.1, Dec = -61 18 45.1 J2000) is located ∼2′ east of the peak in the 3.4 cm continuum emission. As a result, it is unlikely that HD 97499 contributes greatly to the ionization seen in NGC 3576. Finally, the stellar velocity of HD 97499 (-32 km s−1 ± 11 km s−1; Crampton, 1972) does not agree well with the observed velocities in the ionized gas (-23 km s−1 [McGee & Gardner, 1968] and -25 km s−1 [Caswell & Haynes, 1987]). Thus, distances associated with HD 97499 should not simply be applied to NGC 3576. The determination of the distance to NGC 3603 is aided by the presence of an optically visible ionizing cluster. Distance estimates made from stellar observations can be compared to kinematic distances derived from the ionized gas. The earliest spectroscopic distance estimate was made by Sher (1965) at 3.5 kpc. Without a large number of faint stars visible, the spectroscopic distance estimate was admittedly crude. Shortly thereafter, Goss & Radhakrishnan (1969) found a distance of 8.4 kpc based on the observed line velocity of 10 km s−1, using R0 = 10 kpc. Caswell & Haynes (1987) also found a kinematic distance using R0 = 10.5 kpc and V = +11 km s−1 of 8.6 kpc. McGee & Newton (1982) report an H76α line velocity of 7.7±0.1 km s−1. Our velocity for the integrated line in NGC 3603 is VLSR=9.1±0.1, consistent with previously reported values. The H90α velocities found for the thirteen individual regions range between −2 km s−1 and 19 km s−1. Using the velocity from the integrated line (VLSR=9.1 km s−1) and R0 = 8.5 kpc, we determine a kinematic distance of 6.1±0.6 kpc. UBV stellar photometry of NGC 3603 has resulted in distance estimates as low as 3.5 kpc (Sher, 1965) and as high as 8.1 kpc (Moffat, 1974). Melnick & Grosbφl (1982) derive a distance of 5.3 kpc. Moffat (1974), using the stars observed by Sher (1965) and a different reddening scale derive a distance of 8.1 ± 0.8 kpc. Finally, Melnick, Tapia & Terlevich (1989) use UBV photometry of the ionizing cluster of NGC 3603 to find a distance modulus of (m − M )0=14.3, which corresponds to a distance of ∼7.2 kpc. Their value is in good agreement with both Moffat (1974), and the kinematic distances of Goss & Radhakrishnan (1969), and Caswell & Haynes (1987), and in reasonable agreement with the distance derived from our integrated radio recombination line velocity. The smaller distances inferred from the measurements of Sher (1965) and Melnick & Grosbφl (1982) are likely the result of errors in photometry due to nebular contamination (Melnick, Tapia & Terlevich 1989). 4.2. NGC 3576 NGC 3576 has been extensively observed in the radio continuum (Goss & Shaver 1970) and in radio recombination lines (McGee & Gardner 1968, McGee & Newton 1981, Wilson et al. 1970). – 8 – Recombination line studies of the region show the line velocities to be centered at approximately −23 km s−1. The core (diameter∼1.5′) of NGC 3576 is the location of five identified infrared (10µm) sources (Frogel & Persson 1974), as well as CH3OH (Caswell et al. 1995) and H2O (Caswell et al. 1989) masers. The presence of bright 10µm emission, water masers and compact thermal radio emission are typical indications of the early stages of star formation and dense circumstellar environments. Persi et al. (1994) have identified 135 individual stars with IR excess (to distinguish cluster stars from red field stars) in an extremely young galactic cluster associated with the HII region NGC 3576, using JHK images and photometry. Approximately 40 of these IR sources are members of the cluster. In their JHK images, Persi et al. (1994) detect a color gradient in the near-IR, indicating increasing extinction toward the western edge of NGC 3576. This gradient suggests that star formation may have progressed in NGC 3576 from east to west, with the youngest stars currently located at the western edge of the source. In addition, Persi et al. (1994) provide a summary of existing studies of the region, and an overlay of the 21 cm continuum (Retallack & Goss 1980), the 10 µm continuum sources (Frogel & Persson 1974) and their K band mosaic of the region. The overlay (their Fig. 7) clearly shows that the radio continuum, near-IR continuum and 10µm continuum are all strongly peaked toward the western edge of the source. Our 3.4 cm radio continuum image (Fig. 1) improves significantly on the resolution of Retallack & Goss (1980), and shows that there is a coincidence between the location of the infrared sources (Irs1-Irs5) tabulated by Frogel & Persson (1974) and the extended ionized radio continuum emission. Figure 1 shows the 3.4 cm radio continuum with crosses and labels indicating the positions of the 10µm peaks (Frogel et al. 1977). All five sources are coincident with the ionized gas emission and are near local peaks in the radio continuum emission. 4.2.1. Characteristics of the Ionized Gas Our radio recombination line observations have allowed us to examine in detail the variation in electron temperature and helium abundance in NGC 3576. Our results in both cases seem to strengthen the case for sequential star formation in that source. The LTE electron temperatures show a clear gradient from east to west, with temperature increasing with distance (east) from the continuum peak. The gradient is apparent in Figures 5c and 6. Figure 5c shows the spatial variation in electron temperature across the source. Figure 6 is a plot of temperature as a function of distance from the 3.4 cm continuum peak. The LTE electron temperature increases with distance from 6000 K to a maximum of ∼8,000 K, and then falls to lower levels in the more diffuse gas to the east. If the ionizing stars are those IR sources coincident with the radio continuum, this rising temperature signature is expected (Hjellming 1966). Lower energy photons on average would be absorbed first, with the higher energy photons penetrating deeper into the diffuse gas. Thus, the temperature gradient confirms a scenario in which the youngest, hottest stars in the HII region are located near the sharp western edge of the radio continuum. – 9 – Under LTE conditions, the theoretical value of the H113β to H90α hydrogen ratio is 0.276 (Shaver & Wilson, 1979). Deviations from this theoretical ratio indicate local deviations from LTE, and the ratio can be used to correct the derived electron temperature. The H113β to H90α ratio in NGC 3576 varies from ∼0.25±0.05 to ∼0.35±0.05, with the lowest values associated with the bright peak in the radio continuum along the western edge of the HII region. Pressure broadening can lower the β/α ratio from the theoretical value. When the β/α ratio is lower than the theoretical value, electron temperatures derived from the β lines will of course be higher than those derived from α lines. In fact, for NGC 3576, the electron temperatures derived from the H113β line are ∼11 % higher than those derived from the H90α recombination line. A small number of Galactic objects are known to have helium abundances in excess of 20%. Such elevated helium abundances have been reported in the W3 core (Adler, Wood & Goss 1996, Roelfsema, Goss & Mallik 1992), and are thought to be associated with the environments of evolved stellar objects that are rapidly losing mass. A small but significant positive gradient in helium abundance (Y+) is detected from the western peak of the source (Y+ = 0.08 ± 0.01) to the more diffuse ionized gas to the east (Y+ = 0.14 ± 0.03). This range is consistent with the Y + value reported by McGee & Newton (1981) of N(He+)/N(H+)=0.091. The abundance gradient is consistent with the suggestion of Persi et al. (1994) who detect a color gradient in the near-IR indicating increasing extinction to the west of the source. Persi et al. (1994) propose that the youngest stars are located at the western edge of the ionized gas. The higher helium abundances found to the east of NGC 3576 may indicate the presence of a population of older stars and their associated mass loss. 4.2.2. Gas Kinematics in NGC 3576 Figures 5a and 5b show that both broad lines and velocity gradients are present in the ionized gas associated with NGC 3576. Double-peaked emission lines are located over a ∼15′′ region at the northern edge of the source. In Fig. 5a, the double-peaked lines have been fitted with a single Gaussian with a ∆VF W HM = 50 km s−1. Closer inspection (see Figure 1 inset) shows that locally the lines are possibly double-peaked. The double-peaked recombination lines in this region have a velocity separation of ∼25 km s−1, and may be due to either the expansion of an ionized shell of gas, or the outflow of ionized gas from a local source. The broad line region is located to the east of the Irs1/Irs2 peak (Frogel & Persson 1974). The region of the broadest radio recombination lines corresponds spatially to a distinct gap in the 10 µm emission between Irs1/Irs2 and Irs5. If the split lines are due to expansion, then the rate of expansion is ∼13 km s−1. Double-peaked lines with similar velocity separations have been seen in some of the compact HII regions in Sgr B2 North (De Pree et al. 1996). The spatial resolution of the current data is insufficient to determine whether an expanding shell of ionized gas is located at this position. Alternatively, the double-peaked lines may arise near the base of a large scale ionized outflow. – 10 – A large, approximately north-south velocity gradient of 30 km s−1 pc−1 is observed in the ionized gas, and is shown in Figure 5a. The velocity across the source varies from ∼0 km s−1 along the southern edge to ∼-35 km s−1 in the north. The gradient is not likely due to rotation in the ionized gas, since these rotational velocities would imply an approximate enclosed mass of M=2×105 M⊙. The gradient is more likely due to an ionized outflow, and may in fact be the base of a much larger scale outflow. A wide field of view optical image of the region (Fig. 8; Persi et al. 1994) shows several large loops and filaments of ionized gas extending several minutes of arc to the north of NGC 3576. 4.2.3. Sequential Star Formation in NGC 3576 Figure 8 shows the positions of cluster members with infrared excess (crosses; Persi et al. 1994) overlaid on the 3.4 cm radio continuum. Approximately 40 infrared excess sources have been identified. Eight of the stellar positions are coincident with the radio continuum emission, with the majority of the sources located to the north, east and south of the radio continuum. Only a single infrared excess source is located to the west of the radio continuum peak. This spatial distribution would be expected if star formation were proceeding from east to west into the molecular cloud. Water (H2O) and methanol (CH3OH) maser emission (plotted) are located coincident with the peak in the radio continuum emission. The location of molecular gas in NGC 3576 is also consistent with a triggered star formation scenario. The peak in the 1.0 mm dust continuum emission (Cheung et al. 1980) is located to the west of the bright radio continuum peak. The 1.0 mm emission traces the core of the molecular cloud located to the west of NGC 3576. Similar distributions of ionized and molecular gas are found near the Sgr B2 F HII regions (De Pree et al. 1996), G34.3 (Gaume et al. 1994), and others. In fact, the morphology of NGC 3576 is "cometary", and closely resembles a mirror image of the cometary source G34.3 (Gaume et al. 1994) in which a dense molecular cloud core is located at the eastern periphery of the ionized gas. Taken as a whole, the radio, infrared, optical and molecular imaging of NGC 3576 suggest that the oldest ionizing sources are located to the east of the region, and the youngest ionizing sources to the west. The distribution of stars, ionized gas, molecular gas, and water masers in NGC 3576 is reminiscent of simple scenarios of sequential star formation. In this picture, the star formation process slowly "eats away" at a dense cloud, so that the most evolved stars are located far from the cloud core. The ionized gas is found at the periphery of the cloud, and the youngest sources (protostars) are still embedded in the nascent molecular material, but reveal their presence with water maser emission. In many such regions in the Galaxy, the stellar population is not observable. The ability to observe stars, ionized gas and molecular gas in this distant source make it unique in the Galaxy. – 11 – 4.3. NGC 3603 NGC 3603 fills a gap in luminosity between a region like Orion, with ∼1 O-type star, and R136 in 30 Doradus, with ∼100 O-type stars (Eisenhauer et al. 1998). The ionizing core of NGC 3603 (HD 97950) consists of approximately 20 O-type and Wolf-Rayet stars (Moffat, 1983). From photometric measurements, Melnick, Tapia & Terlevich (1989) have identified 181 sources (50 with U , B, and V photometry) associated with the stellar cluster within NGC 3603. The authors found an average age of 2.5 ± 2 Myr and evidence that star formation propagated from north to south, ending at the bordering molecular cloud (Cheung 1980). Brandner et al. (1997) have identified a ring-shaped nebula and bipolar outflows from Sher 25, a cluster member of HD 97950. Sher 25 (a B1.5Iab supergiant) is located approximately ∼20" north of the core of the cluster. Brandner et al. present evidence for association of this star with HD 97950 and compare the star to the progenitor of SN 1987A. The initial mass function (IMF) of HD 97950, the stellar cluster associated with NGC 3603, has been discussed in recent articles by Moffat et al. (1994), Hofmann et al. (1995), Zinnecker (1995) and Eisenhauer et al. (1998). Because of the proximity of NGC 3603, high resolution, deep studies allow detailed counts of stellar mass distribution to be made. Moffat et al. (1994) observed the central cluster with the HST/PC, studying its dense stellar core and the population of Wolf-Rayet stars. Moffat et al. (1994) conclude that in terms of density and stellar population, NGC 3603 is a 'Galactic clone' of R136 in 30 Doradus. Hofmann et al. derive an IMF power law of Γ =−1.4 ± 0.6 (ξ ∝MΓ) for 30-60 M ⊙ which is similar to that of 30 Doradus. Using speckle masking techniques with the HST, Hofmann et al. (1995) found a similar slope for stellar masses of 15 to 50 M ⊙ of Γ =−1.6 ± 0.2. Both Eisenhauer et al. and Zinnecker use adaptive optics to extend the IMF to stellar masses less than 1 M ⊙. Some theoretical and observational studies have claimed that starburst systems are deficient in low mass stars, but Eisenhauer et al. (1998) find that the IMF follows a Salpeter power law with an index of ≤ −0.73, and that there does not appear to be a low mass truncation to the IMF down to the observational limit of 0.2 M ⊙. From such properties of NGC 3603, it may be possible to extrapolate properties of larger extragalactic starbursts. Sequential star formation has also been suggested in NGC 3603, proceeding from north to south (Melnick, Tapia & Terlevich 1989). The presence of evolved objects at the north end of the source and embedded IR sources and a molecular cloud to the south qualitatively confirms a sequential star formation scenario (Hofmann et al. 1995). Investigations of NGC 3603 have proceeded on two fronts: radio and optical observations of the ionized gas, and optical and infrared observations of the exciting stars. While the stars and gas in the giant HII region are visually observable, these high resolution radio observations present the first unobscured view of the ionized gas on small scales. Melnick, Tapia & Terlevich (1989) have tabulated the positions of 181 stars in NGC 3603 from their CCD UBV photometry. The positions of the 50 cluster member stars for which U, B, and V magnitudes are tabulated are shown in Figure 9. The figure shows the positions of these stars (crosses) plotted with the 3.4 cm radio continuum. Coordinates for – 12 – the stellar positions were derived from pixel coordinates using the known position of the B1.5Iab supergiant (Sher 25) as given by Brandner et al. (1997). Optical studies indicate both the presence of high velocity gas and of highly evolved stars. Clayton (1986) first investigated the velocity structure in the ionized gas located at the core of NGC 3603, and found both large scale motions (perhaps kinematic evidence of earlier supernovae) and two smaller shells, possibly wind-driven bubbles. The Hα and [NII] line profiles were made along two lines across the core of NGC 3603. One possible explanation of the wind-driven bubbles was suggested by Moffat et al. (1994) who detected three Wolf-Rayet (WR) stars in their HST/PC observations of the cluster core. Hofmann et al. (1995) confirm the presence of three, perhaps four WR stars in the cluster core. Brandner et al. (1997) provide fascinating evidence that one of the sources believed to be an older member of the cluster (Sher 25, a B1.5Ia supergiant) is at the center of a clumpy ring of optical emission and a possible bipolar outflow. They compare the source to the progenitor of SN1987A. Caswell & Haynes (1987) detected strong hydrogen radio recombination line emission from NGC 3603. The source was also included in the recombination line survey of McGee & Newton (1981), and both hydrogen and helium (at 14.7 GHz) were detected. Our integrated H90α line width (∆VF W HM =36.9±0.25) is in reasonable agreement with the value for the H76α line given by McGee & Newton (∆VF W HM =41.9±0.2). The present study of the ionized gas in NGC 3603 addresses a number of the outstanding issues in this source: 1. Are there any regions of enhanced helium abundance, as might be expected in the presence of 3-4 WR stars? 2. Is there any evidence in the ionized gas emission of the bipolar outflow associated with the B1.5 supergiant Sher 25 (Brandner et al. 1997)? 3. How does the location of ionized gas compare to the known positions of infrared sources (Frogel et al. 1977), WR stars (Moffat et al. 1994 and references therein) and maser emission (Caswell et al. 1989, Caswell et al. 1998)? 4.3.1. Helium Abundance and β/α Ratios McGee & Newton (1981) derived a helium abundance in NGC 3603 of Y+ =0.069 (no uncertainty given). Our calculated helium abundance from the integrated line profile is significantly higher (Y+ =0.13±0.01), but in their notes on individual sources, McGee & Newton (1981) indicate that their He76α line profile was rather noisy. Moffat et al. (1994) detect three prominent Wolf-Rayet stars in NGC 3603 in their narrowband HeII (4686 A) images. WR stars have been cited as the possible explanation of enhanced helium – 13 – abundances (Adler, Wood & Goss 1996). It is certainly reasonable to look for evidence of such enhanced abundances in the vicinity of known WR stars. In fact, the highest helium abundances in NGC 3603 are detected in source C, which has a value of Y+ = 0.26 ± 0.10. Interpretation of this high value is complicated by the large uncertainty. The ionized gas in NGC 3603 is located on the periphery of the WR wind-blown bubble, so it is possible that ionized helium (Y +) abundances are enhanced by the presence of an evolved star. The observed H113β/H90α ratios for the 13 sources in NGC 3603 range from 0.29 ± 0.04 to 0.53 ± 0.12, all above the theoretical value (β/α=0.276). Most detections are within 2σ of this theoretical value, but a few sources (NGC 3603 E, H, I, and K) exceed the theoretical value by more than 3σ. Anomalously high β/α ratios were found from H92α and H115β recombination line observations in two Galactic center regions (G0.18-0.04; the Sickle and the Pistol; Lang, et al., 1997). Thum et al. (1995) propose a model that can explain enhanced β/α ratios as the result of high continuum optical depth. However, NGC 3603 A-M appear to have high β/α ratios, and yet be low density (ne ≤104 cm−3) regions with τc ≤1. Cersosimo & Magnani (1990) suggest that under such conditions, high β/α ratios may result from stimulated emission caused by a background source. 4.3.2. Ionized Gas Kinematics No ionized gas is detected above a 5σ continuum level at the location of Sher 25. The optical study of Clayton (1986) indicates clearly that there are both large scale and small scale motions in the ionized gas. Plate 1 of Clayton shows an Hα image of the central 6′ of NGC 3603, and the positions of the two slit positions through the core. The bright ionized gas observed in our high resolution image (see Fig. 1b and Fig. 9) lies to the south of the cluster core. On scales probed by our high resolution observations, only one source (NGC 3603 F) has clearly double-peaked recombination line emission. Two Gaussians with velocities separated by ∼29 km s−1(VLSR =27.5 km s−1 and −1.1 km s−1) have been fitted to the H90α line in NGC 3603 F. The velocity separation is similar to that seen in other Galactic wind-driven bubbles and also in NGC 3576. In the radio continuum, source F is a relatively faint, marginally resolved source located near the western edge of the southern bright region. NGC 3603 F has a deconvolved diameter of θF W HM ∼25′′, giving the source a linear diameter of 0.5 pc (D = 6.1 kpc). Thus, source F may be an expanding shell of ionized gas. A small optically observable wind-driven "stellar bubble" with r ∼0.6 pc has been observed centered on the HD97950 star cluster in NGC 3603 (Balick, Boeshaar & Gull, 1980). In NGC 3603, the individual source velocities (as derived from the H90α line) range from −3.7±0.3 km s−1 in source A to to 15.9±0.2 km s−1 in source I. The broadest lines are detected in NGC 3603 F (∆VF W HM =52.7±2.2 km s−1). As indicated above, the broad lines in this source likely result from the double-peaked lines apparent in Fig. 7f. Broad lines can result from a variety of local conditions including pressure broadening or spatially unresolved velocity gradients. Given – 14 – the relatively high frequency of the observations and other evidence for velocity gradients in the source, it is likely that the broad lines in source M (located to the north of the cluster core) are related to gas kinematics. NGC 3603 M has vturb ≃ 26 ± 10 km s−1. Other localized regions in the source have very broad recombination line emission. Lines as broad as ∆V ∼80 km s−1 are detected in the low surface brightness gas between sources NGC 3603 D and F. And in the bright core of NGC 3603 D (∼0.07 pc in diameter) line widths approach ∆V ∼75 km s−1. In a sequential star formation scenario, sources D and F are located at the boundary of the molecular cloud and the stellar cluster, and would be in the early stages of their evolution, perhaps still coming into equilibrium with their environment or photoevaporating material from the molecular cloud interface. 4.3.3. Evidence for sequential star formation in NGC 3603 In Figure 2, the positions of ten 10µm sources (Frogel et al. 1977) are plotted over the 3.4 cm continuum. Figure 9 shows the positions of cluster stars and known maser emission. Error bars for the absolute positions of the IR sources are ±10′′, and the size of the crosses in Fig. 2 indicates the absolute error in the positions of the infrared sources. In several cases (e.g. Irs 1, 2, 9, 14), the infrared sources are coincident with radio continuum sources. Perhaps the clearest evidence for sequential star formation in NGC 3603 is the relative location of the radio emission from the ionized gas, the stellar positions, and the position of embedded infrared sources and maser emission. The distribution is very much like that observed in NGC 3576. The most evolved stars are located to the north, centered in what appears at many frequencies to be the wind-blown bubble resulting from the WR stars located there. The molecular cloud core is located farthest to the south, and the ionized gas emission (apparent at optical and radio frequencies) is located at the periphery between the stars and the cold gas. 5. Conclusions Both NGC 3576 and NGC 3603 provide rare opportunities to study massive star formation in a variety of spectral regions. Massive star forming regions like W49A and Sgr B2 are impressive at radio frequencies, but the stellar populations are inaccessible in the infrared. In these two sources, high resolution stellar photometry and radio continuum and recombination line observations provide a more complete picture of the physical conditions and complex gas kinematics. The following is a summary of our major conclusions: (1) In these high resolution radio continuum and recombination line observations, we have identified a number of previously unresolved radio continuum sources, particularly in NGC 3603. (2) The observed temperature gradient in NGC 3576 is in agreement with the known color – 15 – gradient in the source and supports a sequential star formation scenario that has been proposed for the region. (3) We have detected several compact sources in NGC 3603 with broad recombination lines. One source (NGC 3603 F) has double peaked profiles, with the peaks separated by ∼30 km s−1. (4) The brightest radio continuum emission in NGC 3603 is located to the south of the core of the stellar cluster HD 97950. There is no radio continuum or line emission apparent at the position of Sher 25. (5) Derived helium abundances and electron temperatures in NGC 3576 are in good agreement with values given in McGee & Newton (1981) while those in NGC 3603 are significantly higher. (6) Enhanced helium abundances ranging in value from Y +=0.16 to 0.26 are detected in NGC 3603 A-D and J. These regions may have had their helium abundance enriched by the 3 or 4 known WR sources located at the cluster core. (7) In both sources, the relative positions of stars, ionized gas, molecular gas and masers is suggestive of a sequential star formation scenario. In NGC 3576 the star formation appears to have proceeded from east to west, in NGC 3603 from north to south. C. G. De Pree gratefully acknowledges the support of an American Astronomical Society Small Research Grant. W. M. Goss would like to thank R. D. Ekers, J.B. Whiteoak, & D. McConnell for support during a sabbatical visit to the Australia Telescope Compact Array in 1996. The Australia Telescope is funded by the Commonwealth of Australia for operation as a National Facility operated by CSIRO. The National Radio Astronomy Observatory is a facility of the National Science Foundation, operated under cooperative agreement by Associated Universisites, Inc. The authors also thank H. Zinnecker, J. Caswell, W. Brandner, P. Palmer, D. Mehringer, L. Rodriguez, R. Gaume, M. Claussen and an anonymous referee for helpful discussions and comments. – 16 – REFERENCES Adler, D.S., Wood, D.O.S., Goss, W.M. 1996, ApJ, 471, 871 Balick, B., Boeshaar, G.O., Gull, T.R. 1980, ApJ, 242, 584 Brandner, W., Grebel, E.K., Chu Y., Weis, K. 1997, ApJ, 475, L45 Caswell, J.L., 1998, MNRAS, 297, 215 Caswell, J.L., Vaile, R.A., Ellingsen, S.P., Whiteoak, J.B., Norris, R.P. 1995, MNRAS, 272, 96 Caswell, J.L., Haynes, R.F. 1987, A&A, 171, 261 Caswell, J.L., Batchelor, R.A., Forster, J.R., Wellington, K.J. 1989, Austral. J. Phys., 42, 331 Caswell, J.L., Vaile, R.A., Ellingsen, S.P., Whiteoak, J.B., Norris, R.P., 1977, MNRAS, 272, 96 Cersosimo, J.C., Magnani, L. 1990, A&A, 239, 287 Cheung, L. H., Frogel, J. A., Hauser, M. G. & Gezari, D. Y. 1980, ApJ, 240, 74 Clayton, C.A. 1986, MNRAS, 219, 895 Clayton, C.A. 1990, MNRAS, 246, 712 Crampton, D. 1972, MNRAS, 158, 85 De Pree, C.G., Mehringer, D. M. & Goss, W. M. 1997, ApJ, 482, 307 De Pree, C.G., Gaume, R. A., Goss, W. M. & Claussen, M.J., 1996, ApJ, 464, 788 De Pree, C.G., Gaume, R.A., Goss, W.M. & Claussen, M.J., 1995, ApJ, 451, 284 Eisenhauer, F., Quirrenbach, A., Zinnecker, H., and Genzel, R. 1998, ApJ, 498, 278 Frogel, J.A., Persson, S.E., Aaronson, M. 1977, ApJ, 213, 723 Frogel, J.A., Persson, S.E. 1974, ApJ, 192, 351 Gaume, R.A., Fey, A.L. & Claussen, M.J. 1994, ApJ, 432, 648 Gaume, R.A., Claussen, M.J., De Pree, C.G., Goss, W.M. & Mehringer, D.M. 1995, ApJ, 449, 663 Goss, W.M., Radhakrishnan, V. 1969, Astrophys Letters, 4, 199 Goss & Shaver 1970, Austral. J. Phys. Astroph. Suppl., 14, 133 Hjellming, R. 1966, ApJ, 143, 420 Hofmann, K.H., Seggewiss, W., Weigelt, G. 1995, A&A, 300, 403 Humphreys, R.A. 1972, A&A, 20, 29 Lang, C., Goss, W.M., Wood, D.O.S. 1997, ApJ, 474, 275 Mehringer, D.M., De Pree, C.G., Gaume, R.A., Goss, W.M. & Claussen, M.J. 1995, ApJ, 442, L29 Mezger, P.G. & Henderson, A.P., ApJ, 147, 471 McGee, R.X., Gardner, F.F. 1968, Austral. J. Phys., 21, 149 – 17 – McGee, R.X., Newton, L.M. 1981, MNRAS, 196, 889 Melnick, J., Grosbφl, P. 1982, A&A. 107, 23 Melnick, J., Tapia, M., Terlevich, R. 1989, A&A, 213, 89 Moffat, A.F.J., 1974, A&A, 35, 315 Moffat, A.F.J., 1983, A&A, 124, 273 Moffat, A.F.J., Drissen, L., Shara, M.M. 1994, ApJ, 436, 183 Norris, R. P., Caswell, J. L., Gardner, F. F., & Wellington, K. J. 1987, ApJ, 321, 159 Persi, P., Roth, M., Tapia, M., Ferrari-Toniolo, M., Marenzi, A.R. 1994, A&A, 282, 474 Retallack, D.S., Goss, W.M., 1980, MNRAS, 193, 261 Roelfsema, P.R., Goss, W.M., 1992, A&A Review, 4, 161 Roelfsema, P.R., Goss, W.M., & Mallick, 1992 Salem, M., & Brocklehurst, M. 1979, ApJS, 39, 633 Shaver, P.A., Wilson, T.L., 1979, A&A 79, 312 Sher, D. 1965, MNRAS, 129, 17 Thum, C., Strelninski, V.S., Martin-Pintado, J., Matthews, H.E., Smith, H.A. 1995, A&A, 300, 643 Wilson, T. L., Mezger, P. G., Gardner, F. F., Milne, D. K. 1970, A&A, 6, 364 Zinnecker, H., 1995, 11th IAP Conference (Paris) This preprint was prepared with the AAS LATEX macros v4.0. – 18 – Table 1: Observational Parameters of NGC 3576 and NGC 3603 Parameter NGC 3576 NGC 3603 Date (configuration)......................... Total observing time (hr)................. R.A. of field center (J2000).............. Decl. of field center (J2000)............. Major axis (′′).................................. Minor axis (′′).................................. Position angle (o)............................. LSR central velocity (km s−1).......... Total bandwidth (MHz)................... Number of channels.......................... Channel Separation (kHz, km s−1)... Spectral resolution (kHz, km s−1)... Continuum noise (mJy beam−1)................ Line noise (mJy beam−1)........................... 13 June 1995 (750m D) 19 June 1995 (750m B) 13 June 1995 (750m D) 19 June 1995 (750m B) 12 11h 11m 51.88s -61o 18′ 31′′ 12 11h 15m 02.6s -61o 15′ 46′′ 7.0′′ 6.8′′ -9o -23 16 256 62.5 (2.1) 75.6 (2.6) 18 2.1 7.0′′ 6.9′′ -13o 10 16 256 62.5 (2.1) 75.6 (2.6) 11 2.2 – 19 – Table 2: Continuum Parameters for NGC 3576 and NGC 3603 Source NGC 3576 RA (J2000) 11h 11m 51.32s Dec (J2000) -61o 18′ 43.′′5 SP eak (Jy beam−1) 4.1±0.4 ST ot. (Jy) 71±7 θ (′′) ∼90 NGC 3603 A 11h 14m 56.02s 14m 56.51s 15m 00.90s 15m 01.99s 15m 02.62s 15m 06.13s 15m 08.76s 15m 09.45s 15m 10.36s 15m 14.24s 15m 18.75s 15m 24.20s 15m 31.59s B C D E F G H I J K L M -61o 13′ 55.′′8 13′ 17.′′0 13′ 32.′′7 16′ 33.′′0 15′ 53.′′8 16′ 40.′′5 16′ 55.′′7 16′ 41.′′4 16′ 17.′′4 17′ 34.′′0 16′ 58.′′1 12′ 54.′′0 13′ 16.′′7 0.10±0.01 0.08±0.01 0.06±0.01 0.11±0.01 0.40±0.02 0.12±0.01 0.30±0.01 0.23±0.01 0.25±0.02 0.11±0.01 0.08±0.01 0.06±0.01 0.06±0.01 1.3±0.3 2.5±0.4 0.85±0.3 1.2±0.3 1.9±0.2 1.7±0.3 4.4±0.3 4.6±0.3 1.3±0.2 2.2±0.3 1.2±0.3 0.94±0.3 1.1±0.3 24 38 24 22 13 25 26 30 14 30 25 29 26 r (pc) 1.4 0.52 0.80 0.52 0.46 0.28 0.52 0.54 0.64 0.30 0.65 0.53 0.61 0.55 – 20 – Table 3: Derived Continuum Parameters for NGC 3603 Source ne EM (103 cm−3) (106 pc cm−6) MHII (M⊙) U NLyC τc (pc cm−2) (1048 s−1) A B C D E F G H I J K L M 0.92 0.67 0.75 1.0 2.8 1.0 1.6 1.3 2.1 0.87 0.85 0.66 0.72 0.9 0.7 0.6 1.0 4.6 1.1 2.8 2.1 2.7 1.0 0.8 0.5 0.6 13 35 10 10 6 15 25 32 5 23 12 15 12 50 62 44 49 57 54 76 76 50 60 48 47 45 5.6 11 3.7 5.1 8.2 7.1 19 20 5.4 9.5 5.0 4.8 4.0 0.005 0.004 0.005 0.006 0.027 0.007 0.016 0.012 0.016 0.006 0.005 0.003 0.003 – 21 – Table 4: Recombination Line Parameters and Derived Quantities for or NGC 3576 V (km s−1) -20.0±0.1 -19.3±0.4 -21.5±0.9 -22.3±0.2 ∆V (km s−1) 28.7±0.1 24.8±0.8 8.75±1.8 33.6±0.4 T l/T c 0.105±0.001 0.011±0.001 0.003±0.001 0.022±0.001 ∗ Te (K) 6300±700 7100±800 -24.0±0.1 -17.2±0.1 -14.5±0.1 -18.8±0.2 29.1±0.1 26.5±0.1 32.0±0.3 36.2±0.4 0.100±0.001 0.112±0.001 0.081±0.001 0.098±0.002 6500±700 6400±700 7200±800 5500±600 H90α He90α C90α H113β H90α Irs1/Irs2 Irs3 Irs4 Irs5 – 22 – FIGURE CAPTIONS Fig. 1 The 3.4 cm continuum image of NGC 3576 is shown at a resolution of 7′′ (0.09 pc at 3.0 kpc). The first positive and negative continuum contours are at 3σ (54 mJy beam−1). Subsequent positive contours are 1.4, 2, 2.8, 4, 5.6, 8, 11.3, 16, 22.6, 32 and 45.3 times the 3σ level. The peak continuum flux density is 4.1 Jy beam−1 The positions of the 10 µm sources (Frogel & Person 1974) are indicated with crosses. Errors in the absolute IR positions are ±4′′(indicated by the cross size). The inset is an integrated profile made over a region of the source where broad recombination lines are detected. The best fit to the spectrum consists of two Gaussians separated by ∼ 25 km s−1. Epoch is B1950. Fig. 2 The 3.4 cm continuum image of NGC 3603 with the 13 named sources (A to M in order of increasing RA). The continuum data has a spatial resolution of 7′′ (0.21 pc at D = 6.1 ± 0.6 kpc). The first contour is at 5σ (55 mJy beam−1) with subsequent levels at 1.4, 2, 2.8, 4, 5.6, 8, 11.3, 16, 22.6, 32 and 45.3 times the 5σ level. The peak continuum flux density is 0.47 Jy beam−1. The positions of the 10 µm sources (Frogel et al. 1977) are indicated with crosses. Errors in the absolute IR positions from this later study are ±10′′(indicated by the cross size). Epoch is B1950. Fig. 3 (a) The continuum weighted, integrated line profile for NGC 3576. The recombination lines shown (left to right) are H113β, C90α, He90α, and H90α. The solid line is the Gaussian fit, the crosses are the data points, and the dashed line is the residual. The spectral resolution is 2.6 km s−1. (b) The continuum weighted, integrated line profile for the bright southern region in NGC 3603. The recombination lines shown (left to right) are H113β, C90α, He90α, and H90α. The solid line is the Gaussian fit, the crosses are the data points, and the dashed line is the residual. The spectral resolution is 2.6 km s−1. Fig. 4 The continuum weighted, integrated line profiles for the Irs sources in NGC 3576 (Frogel & Persson 1974), indicated in Figure 1. The solid line is the Gaussian fit, the crosses are the data points, and the dashed line is the residual. The spectral resolution is 2.6 km s−1. Figures show (a) Irs 1/2, (b) Irs 3, (c) Irs 4, (d) Irs 5. Fig. 5 Line parameters of the H90α line in NGC 3576 overlaid on the 3.4 cm continuum contours. Fits were made to each pixel above the 5σ level in the continuum. (a) The velocity (VLSR) of the H90α line in NGC 3576 for each pixel is presented in grey-scale with the NGC 3576 3.4 cm continuum contours overlayed. The contour levels are as indicated in Figure 1. The grey scale velocity range is from -35 km s−1 to 10 km s−1. (b) The full width at half maximum (∆VF W HM ) of the H90α line in NGC 3576 for each pixel is presented in false color. The line width range is from 20 km s−1 to 50 km s−1. (c) The electron temperature calculated for each pixel in NGC 3576 above 10σ in the continuum image. The temperature (grey scale) ranges from 5000 to 12000 K. The contours start at 4000 K and rise linearly to 12000 K in increments of 1000 K. Fig. 6 The integrated electron temperature in NGC 3576 for 18 concentric semicircles (width of 4 pixels) centered on the peak in the 3.4 cm continuum image. Temperatures range from 6000 K to 9000 K across the source. – 23 – Fig. 7 The integrated line profiles from the 13 individual HII regions in NGC 3603, as indicated in Figure 2. The recombination lines shown (left to right where detected) are H113β, C90α, He90α, and H90α. The plot shows the Gaussian fit (solid line), data (crosses), and the residual (dashed line). The spectral resolution is 2.6 km s−1. Figures 7a to 7m show the integrated profiles of sources NGC 3603 A-M respectively. Fig. 8 The 3.4 cm continuum contours for NGC 3576. The stellar positions of cluster member stars with infrared excess (Persi et al. 1994) are plotted as crosses. Also plotted are the positions of the star HD 97499 (diamond), and the H2O (triangle; Caswell 1989) and CH3OH (box; Caswell et al. 1995) maser positions. Continuum contours are as indicated in Figure 1. Epoch is J2000. Fig. 9 The 3.4 cm continuum contours in the vicinity of the known stellar cluster HD 97950. The positions of ∼50 cluster members with measured U, B, and V magnitudes (Melnick, Tapia & Terlevich 1989) are plotted as crosses. Coordinates of the 50 stars were derived from pixel offsets from the known position of Sher 25 (a B1.5Iab supergiant) as given by Brandner et al (1997). Also indicated is the position of HD 97950 (diamond), and H2O (triangles; Caswell 1989) and OH (box; Caswell 1998) maser positions. Continuum contours are as indicated in Figure 2. Epoch is J2000. – 24 – Table 5: Recombination Line Parameters and Derived Quantities for NGC 3603 ∆V (km s−1) 36.9±0.25 38.9±2.6 10.0 (fixed) 42.6±0.8 T l/T c 0.075±0.001 0.009±0.001 0.002±0.001 0.026±0.001 ∗ Te (K) 6600±500 4800±400 Te (K) Y+ β/α V turb 0.13±0.01 0.39±0.01 24±2 14±3 17±4 22±9 23±7 27±3 26.3±0.5 25.0±3.1 31.6±1.9 0.107±0.003 0.019±0.003 0.033±0.003 6300±1100 6700±1200 0.17 ±0.03 4900±900 5200±1000 0.31±0.03 28.3±0.7 28.8±4.8 10 (fixed) 29.0±2.6 33.2±1.7 44.1±10 49.1±10 35.9±1.1 32.8±5.6 31.4±2.6 0.103±0.004 0.016±0.004 0.015±0.003 0.029±0.004 0.104±0.007 0.020±0.006 0.023±0.006 0.078±0.003 0.014±0.003 0.029±0.003 6300±1600 6400±1600 0.16 ±0.04 6000±1500 6100±1500 0.29±0.04 4300±1700 4200±1700 0.26 ±0.10 4400±1700 4300±1700 0.32±0.11 6400±2000 6800±2100 0.16 ±0.05 5700±1800 6100±1900 0.33±0.05 1.9±0.2 -2.2±1.5 -2.3±0.6 41.9±0.5 36.9 (fixed) 46.5±1.4 0.066±0.001 0.008±0.001 0.022±0.001 6700±600 8500±800 0.11±0.01 5300±500 6700±600 0.36±0.02 11.5±0.8 27.6±2.7 5.9±2.4 15.5±0.4 20.0±1.6 8.3±1.2 11.8±0.1 12.4±0.9 8.2±1.6 8.5±0.4 15.9±0.2 7.9±3.1 13.4±0.7 7.3±0.3 2.9±3.6 5.3±1.2 5.5±0.3 7.9±1.9 3.8±0.7 6.4±0.6 5.7±4.5 2.7±1.3 4.6±0.5 8.5±6.3 4.3±2.1 52.7±2.2 22.7±6.9 59.9±6.8 25.6±1.0 17.2±3.8 31.2±3.1 0.056±0.003 0.011±0.005 0.020±0.003 0.115±0.006 0.024±0.001 0.040±0.006 32.2±0.3 29.3±2.2 10.0 (fixed) 37.9±0.9 0.082±0.001 0.011±0.001 0.003±0.001 0.028±0.001 34.8±0.6 44.9±8.0 39.6±1.8 39.9±0.9 59.0±9.7 47.8±2.9 40.8±0.7 31.2±4.7 43.8±1.9 0.068±0.002 0.006±0.001 0.023±0.002 0.085±0.003 0.011±0.003 0.031±0.003 0.093±0.002 0.012±0.003 0.037±0.002 38.2±1.5 35.0 (fixed) 32.9±3.1 0.142±0.008 0.018±0.004 0.063±0.008 37.3±2.1 33.8 (fixed) 44.9±5.4 0.115±0.009 0.015±0.005 0.051±0.009 6500±3300 7000±3600 36±19 4700±2400 5100±2600 0.40±0.08 0.08±0.04 6200±2200 7900±2800 0.14±0.05 4400±1500 5600±1900 0.43±0.07 6900±1200 8300±1400 0.12±0.02 5100±900 6300±1100 0.40±0.02 7900±2200 9300±2600 0.11±0.03 5900±1600 6800±1800 0.39±0.03 5700±1600 5800±1600 0.18±0.05 3600±1000 3700±1000 0.43±0.05 5100±1000 5000±1000 0.10±0.02 3600±700 3500±700 0.43±0.03 3800±1000 3700±1000 0.11±0.03 2900±800 2800±800 0.37±0.07 13±5 19±3 21±6 26±7 28±6 26±7 3600±1300 3500±1300 26±10 2600±1000 2500±1000 0.53±0.12 0.11±0.04 V (km s−1) 9.14±0.1 8.9±1.0 6.2±1.9 5.9±0.3 8.08±0.2 4.0±1.3 2.5±0.8 -3.65±0.3 0.9±2.0 -12.5±1.3 -7.4±1.1 3.8±0.7 9.1±4.3 -5.2±4.1 -0.80±0.4 1.1±2.2 -1.7±1.0 Source Total A B C D E F G H I J K L M H90α He90α C90α H113β H90α He90α H113β H90α He90α C90α H113β H90α He90α H113β H90α He90α H113β H90α He90α H113β H90α He90α H113β H90α He90α H113β H90α He90α C90α H113β H90α He90α H113β H90α He90α H113β H90α He90α H113β H90α He90α H113β H90α He90α H113β ) 0 5 9 1 B ( I N O T A N L C E D I 0 1 2 3 4 -61 01 00 NGC 3576 3.4 cm Continuum 30 02 00 30 03 00 30 04 00 11 10 00 09 55 Irs 1 Irs 2 Irs 5 Irs 4 Irs 3 Velocity (km/s) 50 40 RIGHT ASCENSION (B1950) 45 35 ) 0 5 9 1 B ( I N O T A N L C E D I -60 56 57 58 59 -61 00 01 02 0 100 200 300 400 Irs15 L Irs6 M NGC 3603 3.4 cm Continuum B A Irs5 C Irs4 E Irs14 Irs1 D Irs12 Irs13 K I H J Irs8 F Irs9 Irs2 G 11 13 15 RIGHT ASCENSION (B1950) 00 12 45 NGC 3576 (a) NGC 3603 (b) H90α ) c ( T / ) l ( T H90α H113 β He90 α α C90 H113 β He90 α α C90 Velocity (km/s) NGC 3576 Irs 1/2 (a) NGC 3576 Irs 3 ) C ( T / ) L ( T NGC 3576 Irs 4 (c) NGC 3576 NGC 3576 Irs 5 Irs 5 (b) (d) Velocity (km/s) -30 -20 -10 0 10 -61 17 45 18 00 NGC 3576 3.4 cm continuum V LSR ) 0 0 0 2 J ( I I N O T A N L C E D 15 30 45 19 00 15 30 45 20 00 11 12 05 00 11 55 RIGHT ASCENSION (J2000) 50 45 6 8 10 12 -61 18 00 NGC 3576 LTE Electron Temp. ) 0 0 0 2 J ( I N O T A N L C E D I 15 30 45 19 00 15 30 45 11 12 02 00 11 58 56 54 52 RIGHT ASCENSION (J2000) 6 7 5 3 C G N ) s / m k ( V M H V W F V ∆ 50 48 46 2 9 1 2 r e s U S P A I ) K 0 0 0 1 x ( e r u t a r e p m e T M A E B Y J o / l i K Distance (Arcseconds) Plot file version 3 created 09-JUL-1998 16:01:14 10 20 30 40 50 60 70 o o o o o o o o o o o o o o o o o o 9 9 8 8 7 7 6 6 5 4 3 2 1 0 0 10 20 30 40 Arcsec 50 60 70 NGC 3603 A (a) (a) NGC 3603 B (b) NGC 3603 C (c) 5% 5% 5% NGC 3603 D (d) NGC 3603 E (e) NGC 3603 F (f) 5% ) C ( T / ) L ( T 2% 2% NGC 3603 G (g) NGC 3603 H (h) NGC 3603 I (i) 5% 2% 2% Velocity (km/s) NGC 3603 J (j) NGC 3603 K (k) 5% 5% ) C ( T / ) L ( T NGC 3603 L (l) NGC 3603 M (m) 5% 5% Velocity (km/s) -61 16 30 NGC 3576 3.4 cm Continuum Stars and Maser Sources 17 00 30 18 00 30 19 00 30 20 00 30 21 00 HD 97499 CH OH maser H O maser Stars 3 2 11 12 20 15 10 11 55 RIGHT ASCENSION (J2000) 05 00 50 45 -61 14 NGC 3603 3.4 cm Continuum Stars and Maser Sources ) 0 0 0 2 J ( I N O T A N L C E D I 15 16 17 18 19 11 15 30 25 20 2 OH maser H O maser Stars HD 97950 00 14 55 15 RIGHT ASCENSION (J2000) 10 05
astro-ph/0112176
1
0112
2001-12-07T12:13:17
Detecting the progenitors of core collapse supernovae
[ "astro-ph" ]
The masses and the evolutionary states of the progenitors of core-collapse supernovae are not well constrained by direct observations. Stellar evolution theory generally predicts that massive stars with initial masses less than about 30M_sol should undergo core-collapse when they are cool M-type supergiants. However the only two detections of a SN progenitor before explosion are SN1987A and SN1993J, and neither of these was an M-type supergiant. Attempting to identify the progenitors of supernovae is a difficult task, as precisely predicting the time of explosion of a massive star is impossible for obvious reasons. There are several different types of supernovae which have different spectral and photometric evolution, and how exactly these are related to the evolutionary states of the progenitor stars is not currently known. I will describe a novel project which may allow the direct identification of core-collapse supernovae progenitors on pre-explosion images of resolved, nearby galaxies. This project is now possible with the excellent image archives maintained by several facilities and will be enhanced by the new initiatives to create Virtual Observatories, the earliest of which ASTROVIRTEL is already producing results.
astro-ph
astro-ph
Detecting the progenitors of core collapse supernovae Stephen J. Smartt Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge December 1, 2001 Abstract. The masses and the evolutionary states of the progenitors of core- collapse supernovae are not well constrained by direct observations. Stellar evolution theory generally predicts that massive stars with initial masses less than about 30M⊙ should undergo core-collapse when they are cool M-type supergiants. However the only two detections of a SN progenitor before explosion are SN1987A and SN1993J, and neither of these was an M-type supergiant. Attempting to identify the progeni- tors of supernovae is a difficult task, as precisely predicting the time of explosion of a massive star is impossible for obvious reasons. There are several different types of supernovae which have different spectral and photometric evolution, and how exactly these are related to the evolutionary states of the progenitor stars is not currently known. I will describe a novel project which may allow the direct identification of core-collapse supernovae progenitors on pre-explosion images of resolved, nearby galaxies. This project is now possible with the excellent image archives maintained by several facilities and will be enhanced by the new initiatives to create Virtual Observatories, the earliest of which (astrovirtel) is already producing results. Keywords: sample, LATEX 1. Which stars go supernovae ? Supernovae of Types II and Ib/Ic are thought to occur during core collapse in massive stars at the end of their lifetimes. However the only definite detection of a SN progenitor is that of SN1987A in the LMC (White & Malin, 1987), which was a blue supergiant (B3I; Walborn et al., 1989). The progenitor of SN1993J in M81 was possibly identified as a K0 Ia star (Aldering et al., 1994). Neither progenitor is consistent with the canonical stellar evolution picture, where core-collapse occurs while the massive star is an M-supergiant. We still don't understand the physical mechanisms which underpin the different supernovae types, and how these are related to the evolution of the progenitor star. There is an understandable lack of observational data to constrain the last moments of stellar evolution. This conference is dedicated to under- standing the basic building blocks of galaxy evolution, and supernova physics is a fundamental input parameter in determining the dynamical and chemical evolution of galaxies from the first stars in the Universe to present day gas-rich galaxies. Linking the observed supernova types to a stars initial mass, metallicity, binarity, environment and its subsequent c(cid:13) 2018 Kluwer Academic Publishers. Printed in the Netherlands. smartts_astroph.tex; 6/11/2018; 12:15; p.1 2 S.J. Smartt evolution is not only important for those of us working on massive stellar evolution, it will also impact on galaxy evolution as a whole. 2. Two recent supernovae with images before explosion The very well maintained data archives of the Hubble Space Telescope, the Canada-France-Hawaii Telescope, the Isaac Newton Group of Tele- scopes and those of ESO contain a vast array of multi-colour images of late-type galaxies within approximately 20 Mpc. At the spatial resolu- tion of ground-based telescopes, the most luminous individual massive stars can be resolved and their photometry accurately measured in galaxies within ∼8 Mpc. At the resolution of WFPC2 on board HST we can extend these measurement of individual massive stars to fainter intrinsic luminosities and distances out to ∼20 Mpc; the Cepheid Key Project is a clear demonstration of this (Freedman et al., 2001). Hence when a bright supernova is discovered in a spiral galaxy within ∼20 Mpc there is now a reasonable chance that images have been taken of this galaxy either with HST or a ground based facility - allowing the exciting prospect of directly identifying the star which has exploded. Towards the end of 1999 two bright supernovae were discovered in the spirals NGC1637 (7.5 Mpc) and NGC3184 (8 Mpc). These events (1999em and 1999gi) were both Type II-P, with very similar peak magnitudes (MV ≃ −16), very similar ∼100 day plateaus, and were both very faint X-ray and radio sources. By chance there are archive images of these galaxies taken several years before explosion by CFHT and HST. Similar resolution images taken after explosion have allowed the supernova position to be precisely determined on the pre-explosion frames. However in both cases there is no detection of a progenitor star at the SN position. Unfortunately the precursor objects are below the detection limits (see Fig. 1 for example of 1999gi). By measuring the sensitivity limits of the images, the bolometric luminosity limits of the progenitors (as a function of stellar effective temperature) have been determined. These can be plotted on an HR-diagram with stellar evolutionary tracks, which allows one to estimate the initial mass of the progenitor (see Fig. 2 for example). In Smartt et al. (2001a) and Smartt et al. (2002) upper mass limits of 9M⊙and 12M⊙for 1999gi and 1999em respectively were derived, with uncertainties of ±3M⊙. These SNe are very similar in their observed characteristics and have rather similar mass limits. In particular the low values of their X-ray and radio fluxes (Pooley et al., 2001 and Schlegel, 2001) suggest that the progenitor star had a relatively low mass-loss rate, which is consistent with the fairly low masses we derive. This is consistent with the progenitor stars smartts_astroph.tex; 6/11/2018; 12:15; p.2 Supernovae Progenitors 3 Figure 1. The WFPC2 F300W (a) and F606W (b) images before the explosion of SN1999gi and the F555W (c) image of the SN (from Smartt et al. 2001a). The two luminous stars OB1-1 and OB1-2 are clearly visible before and after explosion, and there is no visible object at the position of SN1999gi in either pre-explosion image having initial masses between 9-12M⊙, and having exploded as red- supergiants which have undergone normal mass-loss in the AGB phase. Smartt et al. (2002) have speculated that this type of homogeneous plateau event (which are generally X-ray and radio faint) could all come from moderate mass 8-12M⊙ progenitors, and that a Salpeter IMF would suggest that ∼50% of all core-collapse events should be similar to 1999gi and 1999em if this is true. To test this hypothesis we require better statistics on the relative numbers of the SN sub-types and crucially more direct information on progenitors. 3. Future prospects The two examples described show how having high-quality archive im- ages of SNe sites taken prior to explosion can allow quite stringent limits to be set on the nature of the progenitor stars. This will be improved in the future if we do get a real detection, rather than only a limiting mass. The chances of having suitable pre-explosion images available when a nearby core-collapse supernova is discovered are im- proving rapidly. By June 2002, a WFPC2 SNAP programme will be finished (9042, PI: Smartt) which will enhance the number of late- type galaxies with high-quality archive images. Combining these with data available in the ground-based archives Smartt et al. (2002) have estimated that on average ∼ 2.4±2 SNe per year will have pre-explosion information and hence a project lasting 3-5 years should significantly improve our knowledge. In particular this project has been greatly aided by the introduction of the astrovirtel1 initiative which is a 1 http://www.stecf.org/astrovirtel smartts_astroph.tex; 6/11/2018; 12:15; p.3 4 S.J. Smartt Figure 2. Mass limits for SN1999gi from Smartt et al. (2001a). The Schaller et al. (1992) tracks for 7−40 M⊙ are plotted with the positions of the progenitor of SN1987A and SN1993J indicated. The detection limits of the pre-explosion frames are used to set bolometric luminosity limits as a function of stellar effective temper- ature (plotted as the thick solid lines). The WFPC2 pre-explosion frames should be sensitive to all objects lying in the shaded region above these lines, with the F300W filter limit being the upper curve. This indicates an upper limit of 9+3 −2M⊙ first step at creating tools for future and more wide ranging Virtual Observatories. It is an example of exciting science that can easily be made possible with the European AVO. This is complemented with two HST GO approved proposals in Cycle 10 and Cycle 11 (9042 and 9353) that will provide the accurate astrometry of the supernova for positioning on the pre-explosion images. Already we have one more candidate that has pre-explosion HST data found by astrovirtel (2001du; Smartt et al., 2001b) with follow-up HST ToO images of the SN position just taken at the end of November 2001. References Aldering G., Humphreys R.M., Richmond M., AJ, 107, 662, 1994 Freedman W.L., et al., ApJ, 553, 47, 2001 Pooley D., et al., astro-ph/0103196, 2001 Schaller G., Schaerer D., Meynet G., Maeder A. A&AS, 96, 269 1992 Schlegel E.M., ApJ, 556, L25, 2001 Smartt S.J., et al., ApJ, 556, L29, 2001b Smartt, S. J., Kilkenny, D., Meikle, P. IAUC, 7704, 2001b smartts_astroph.tex; 6/11/2018; 12:15; p.4 Supernovae Progenitors 5 Smartt S.J., Gilmore G.F., Tout C.A., Hodgkin S.T. ApJ, in press, Feb., 2002 Walborn N. et al., 1989, A&A, 219, 229, 1989 White G.L., Malin D.F. Nat., 327, 36, 1987 smartts_astroph.tex; 6/11/2018; 12:15; p.5 smartts_astroph.tex; 6/11/2018; 12:15; p.6
0806.2795
3
0806
2009-01-14T11:08:22
Jupiter - friend or foe? I: the asteroids
[ "astro-ph" ]
The asteroids are the major source of potential impactors on the Earth today. It has long been assumed that the giant planet Jupiter acts as a shield, significantly lowering the impact rate on the Earth from both cometary and asteroidal bodies. Such shielding, it is claimed, enabled the development and evolution of life in a collisional environment which is not overly hostile. The reduced frequency of impacts, and of related mass extinctions, would have allowed life the time to thrive, where it would otherwise have been suppressed. However, in the past, little work has been carried out to examine the validity of this idea. In the first of several papers, we examine the degree to which the impact risk resulting from a population representative of the asteroids is enhanced or lessened by the presence of a giant planet, in an attempt to fully understand the impact regime under which life on Earth has developed. Our results show that the situation is far less clear cut that has previously been assumed - for example, the presence of a giant planet can act to enhance the impact rate of asteroids on the Earth significantly.
astro-ph
astro-ph
Jupiter – friend or foe? I: the asteroids J. Horner and B. W. JONES Astronomy Group, Physics & Astronomy, The Open Un iversity, Milton Keynes, MK7 6AA, UK e-mail: j.a.horner@open .ac.uk Phone: +44 1908 653229 Fax: +44 1908 654192 Received 12 March 2008, accepted 30 May 2008 (SHORT TITLE: Jup iter – friend or foe? I: the asteroids) This article appears in the International Journal of Astrobiology, 7 (3&4), 251-261 (2008) Asteroid s – IJA BWJ 08Jan2009 1 of 34 13/1/09 4:11 PM UT Abstract The asteroids are the major source of potential impactors on the Earth today. It has long been assumed that the giant planet Jupiter acts as a shield, significantly lowering the impact rate on the Earth from both cometary and asteroidal bodies. Such shielding, it is claimed, enabled the development and evolution of life in a collisional environment which is not overly hostile. The reduced frequency of impacts, and of related mass extinctions, would have allowed life the time to thrive, where it would otherwise have been suppressed. However, in the past, little work has been carried out to examine the validity of this idea. In the first of several papers, we examine the degree to which the impact risk resulting from a population representative of the asteroids is enhanced or lessened by the presence of a giant planet, in an attempt to fully understand the impact regime under which life on Earth has developed. Our results show that the situation is far less clear cut that has previously been assumed – for example, the presence of a giant planet can act to enhance significantly the impact rate of asteroids at the Earth. Key words: Solar System – general, comets – general, minor planets, asteroids, planets and satellites – general, Solar System – formation. Introduction Throughout Earth history, our planet has suffered impacts from asteroidal and cometary material. As well as disrupting the landscape, the larger of these impacts have had effects that led to climate changes, usually short-lived, that in turn have led to the extinction of a large proportion of species in the biosphere (Morris (1998)). Anyone who has watched popular science programmes which discuss the effect of impacts on the Earth, along with their implications for the survival of life, will have encountered the idea that Jupiter acts to lower significantly the flux of objects that hit the Earth. The inference, sometimes explicitly stated, is that Jupiter’s role in the evolution of life on our planet is surprisingly large (see, for example, Ward and Brownlee (2000), http://tinyurl.com/2g59ee, http://tinyurl.com/2yvk6x, and http://tinyurl.com/34msr8, for examples of the pervasiveness of this idea). It is claimed that, in preventing the great majority of threatening objects from encountering the Earth, Jupiter has significantly lowered the rate at which impact-driven mass extinctions happen, giving life time to get a foothold, and then evolve to its great present day diversity. Were Jupiter absent, so it is Asteroid s – IJA BWJ 08Jan2009 2 of 34 13/1/09 4:11 PM UT claimed, then the Earth could have suffered large impacts so frequently that it might not have acquired advanced life, or even be barren. These arguments are quite widely accepted in the academic world, but when one looks back through the literature, it seems that, until recently, very little work has been carried out to examine in detail the effects of the giant planet on the flux of cometary and asteroidal bodies through the inner Solar System. It has been suggested that, in systems containing only “failed Jupiters” (bodies which grew to the size of, say, Uranus and Neptune, but failed to develop beyond that stage), the impact flux experienced by any terrestrial planets would be a factor of a thousand greater than that seen in our system today (Wetherill, 1994). This is because of the less efficient ejection of material from the Solar System during its early days. However, very little work exists to support or argue against this conclusion, and it is unclear how it would be affected by the current understanding of planet formation. Laakso et al. (2006) approached the question from a different angle. Using numerical integration, they examined the effect of the position and mass of a Jovian planet on the rate of ejection of particles placed on eccentric orbits that initially crossed the habitable zone (being the range of distances from a star within which water at the surface of an “Earth” would be stable in the liquid phase, liquid water being essential for all forms of life on Earth). They used our Solar System as a test case for their method, and found the surprising result that Jupiter “in its current orbit, may provide a minimal amount of protection to the Earth”. Despite this, the idea that “Jupiters” automatically lower the impact rate in planetary systems is well entrenched in astronomical thinking, and the lack of planets analogous to Jupiter has been used to explain observations such as that of a significant dust excess around the star Tau Ceti (Greaves, 2006). However, questions about Jupiter’s effect on the terrestrial impact record have been raised in relation to the Late Heavy Bombardment. If the “Nice model” (Gomes et al., 2005) is considered, for example, then it is clear that removing Jupiter from our Solar System would greatly lessen or remove the effects of the Late Heavy Bombardment on our young planet. In our opinion, it seems that the idea of “Jupiter, the protector” dates back to the days when the main impact risk to the Earth was thought to arise from the population of long period comets (LPCs), falling inwards from the Oort cloud. The majority of such objects are expelled from the Solar System on their very first pass as a result of Jovian perturbations, hence lowering the chance of one of these cosmic bullets striking the Earth. Asteroid s – IJA BWJ 08Jan2009 3 of 34 13/1/09 4:11 PM UT In recent times, however, it has been estimated that among the near Earth objects (NEOs) i.e. asteroids and comets that make close approaches to the Earth, the comets contribute only a few percent of the population (Bottke et al. (2002), Chapman and Morrison (1994)). Among the comets most are short period comets (SPCs), so the LPCs contribute only slightly to the NEOs. (Near the Earth, comets generally move much faster that asteroids, and so the effect of an impact of a body of given mass, will be greater for a comet.) For the NEOs, the role of Jupiter as friend or foe is far less clear than in the case of the LPCs alone, as can be demonstrated by a thought experiment. Imagine our Solar System as it is today, and remove Jupiter entirely. At one fell swoop, you have removed the main driving force which transfers asteroidal bodies from the main belt between Mars and Jupiter (where the great majority lie) to the inner Solar System. Furthermore, you have lost the object which is the dynamical source and controller of the great majority of the short period comets (SPCs). On the other hand, you have also lost the object most efficient at removing debris from the inner Solar System, though if the detritus is not being put there in anywhere near the same quantity as with Jupiter present, then removal is less important. Overall, the situation is no longer clear cut. What Jupiter gives with one hand, it may take away with the other. In order to study the exact relationship between the giant planet and the impact rate on the Earth, we decided to run a series of n-body simulations to see how varying the mass of a giant planet in Jupiter’s orbit would change the impact rate on Earth. Since there are three distinct populations which provide the main impact threat to the Earth (the asteroids (sourced from the Main Belt (Morbidelli et al. (2002)), the SPCs (which come from the trans-Neptunian region, Horner and Evans (2006)), and the LPCs (which come from the Oort Cloud, Oort (1950)), we decided to split the problem three ways, and examine each population in turn. This paper details our results for the asteroids, an entirely different reservoir of bodies to that studied by Wetherill (who studied the LPCs), and generally accepted to be the most important population of potential Earth impactors. The SPC and the LPC components of the impact risk will be detailed in later work. Whereas the work here advances our understanding of what jovian characteristics have determined the bombardment suffered by the Earth, it also advances our understanding of the requirements for the habitability of “Earths” in exoplanetary systems. Simulating the impact flux Asteroid s – IJA BWJ 08Jan2009 4 of 34 13/1/09 4:11 PM UT Of the three parent populations that supply Earth impacting bodies, the most copious is the asteroids. However, in creating a swarm of test asteroids which might evolve on to Earth impacting orbits, we face huge uncertainties, particularly relating to N(a) at the start of a simulation (t = 0), where N(a) is the number of asteroidal bodies as a function of semimajor axis a. That Jupiter has been perturbing the orbits of the objects currently observed in the asteroid belt in our own Solar System since its formation means that using the current belt as the source would be misguided. It is therefore important to attempt to construct a far less perturbed initial population for the asteroid belt, if one wishes to observe the effect of changing Jupiter’s mass on the impact rate. However, this is more easily said than done. In Appendix 1, we discuss in some detail how we constructed such a test population for use in this work. We finally settled on a population distribution, N(a) at t = 0 given by (see Appendix 1) (1) )1 / 2 N 0 a( ) = k a " amin ( where k is a constant and amin is the inner boundary of the asteroid distribution. The value of amin was chosen to be 1.558 AU, equivalent to the orbital semi-major axis of the planet Marsi, plus three Hill radii, while the outer boundary, amax, was placed three Hill radii within the orbit of the giant planet i.e. interior to the 5.202 AU of Jupiter’s orbit. (To read footnotes, hover over the number.) Closer to the planets than these two distances, asteroidal bodies are unlikely to form. (The Hill radius gives the distance between two bodies, such as a planet and another body, at which their gravitational interaction is of the same order as the gravitational interaction of each body with the star they orbit. Three Hill radii of a planet is its “gravitational reach”. It is given ! by RH = ap where M is mass.) It is important to note that our main conclusions below 1 / 3 % " M planet ’ $ 3MSun & # concerning the variations of the impact rate on Earth as a function of giant planet mass are not sensitive to the precise form of N0(a). The placement of the inner and outer edges at 3RH beyond the orbit of the planets in question was chosen as a reasonable compromise between placing the edge of the belts far enough away from the planet so as not to experience significant perturbations in the early stages of the simulations, and placing the edge so distant that the belt itself would be unfairly constrained. To generate the values of a for our population of asteroids the cumulative probability distribution corresponding to N0(a) was sampled by a random number generator to generate 105 values of a between amin and amax. The other five orbital elements for each asteroid were randomly allocated, as follows. The orbital inclination, i, was randomly sampled from the range 0-10°, and the eccentricity, e, randomly allocated from the range 0.0-0.10. These ranges encompass the majority of Asteroid s – IJA BWJ 08Jan2009 5 of 34 13/1/09 4:11 PM UT ! the known asteroids today. In the distant past, at the start of our simulations, an even greater proportion would have been encompassed. They represent a disk of solid material that has received a moderate, but not excessive, amount of stirring during the formation of the planets (e,g, Ward 2002). The remaining three orbital elements – the longitude of the ascending node, the argument of perihelion, and the mean anomaly, were each randomly selected from the range 0-360°. (For a brief description of orbital elements, see, for example, Jones 2007(a).) We simulated these orbits for a period of 10 million years using the hybrid integrator contained within the MERCURY package (Chambers, 1999), along with the planets Earth, Mars, Jupiter, Saturn, Uranus and Neptune. We take t = 0 to be the moment when Jupiter became fully formed. The integration duration was chosen to provide a balance between the required computation time and the statistical significance of the results obtained. The Earth within our simulations was artificially physically inflated to have a radius of 1 million kilometres, in order to enhance the impact rate from objects on Earth crossing orbits. Simple initial integrations were carried out to confirm that this inflation did affect the impact rate as expected, with the rate scaling with the cross- sectional area of the planet (the effect of gravitational focussing on the impact rate was observed to be negligible). The asteroidal bodies interact gravitationally with the Sun and planets, but not with each other – they are treated as massless which is a good model as a typical asteroidal body is normally at least 1011 times less massive than Jupiter! The “Jupiter” used in our runs was modified so that we ran 12 separate masses. In multiples of Jupiter’s mass MJ these are: 0.01, 0.05, 0.10, 0.15, 0.20, 0.25, 0.33, 0.50, 0.75, 1.00, 1.50, and 2.00. Hereafter, we refer to these runs by the mass of the planet used, so that M1.00 refers to the run using a planet of 1.00 MJ, and M0.01 refers to the run using a planet of 0.01 MJ, and so on. The orbital elements for the “Jupiter” were identical in all cases to those of Jupiter today. Similarly, the elements taken for the other planets in the simulations were identical to those today – the only difference in the planetary setup between one run and the next was the change in Jovian mass – all other planetary variables were held constant. It is obvious that, in reality, were Jupiter a different mass, the architecture of the outer Solar System would likely be somewhat different. However, rather than try to quantify the uncertain effects of a change to the formation of our own Solar System, we felt it best to change solely the mass of the “Jupiter”, and therefore work with a known, albeit modified, system rather than an uncertain theoretical construct. In the case of the flux of objects moving inwards from the asteroid belt, this Asteroid s – IJA BWJ 08Jan2009 6 of 34 13/1/09 4:11 PM UT does not seem a particularly troublesome assumption, because Jupiter is by far the dominant influence on the asteroids. The complete suite of integrations ran for some six months of real time, spread over the cluster of computers sited at the Open University. This six months of real time equates to over twenty years of computation time, and resulted in measures of the impact flux for each of the twelve “Jupiters”. The eventual fate of each asteroidal body was also noted. Results In this section, we present the results of our simulations, leaving discussion for the next section. This should allow the reader to be familiar with the results, and perhaps reach their own conclusions, before we present a detailed discussion. Figures 1 to 3 show a variety of different results from our simulations (these Figures are at the end of the article). Figure 1 shows the evolution of our test populations as a function of time for M0.25 (Figure 1a) and M1.00 (Figure 1b). Five temporal snapshots are shown, detailing the distribution of asteroidal bodies at t = 0 Myr (the start of the simulation), 1 Myr, 2 Myr, 5 Myr and 10 Myr (the end of our simulation). In order to give a fair representation of the asteroid distributions, Figure 1 shows the number of objects located in rings of equal width (in semimajor axis), working outward from a semimajor axis of 1.5 AU to 5.5 AU. This space is broken up into 1000 equal width bins, so that the width of each bin is 0.004 AU. In effect, this means that the maximum initial population in any bin is less than 400 objects, and so the y-axis in all of the plots in Figure 1, extends from 0 to 400. The initial populations were, as described above, distributed according to equation 1, with inner and outer limits fixed as described. Note that the location of the outer edge of the belt changes between the two plots, in response to the larger Hill sphere of the more massive Jupiter in Figure 1b. Additionally, it is clear that the initial population is somewhat scattered, a result of the random number generator used to select a. As an aid for the reader, the points corresponding to each bin have been connected, which makes small details easier to see. The development of fine structure in the belts is clearly apparent as early as 1 Myr, and this structure continues to develop through the period of the simulations. Equivalent plots for all 12 “Jupiter” masses can be found in Appendix 2. Figure 1 The evolution of the asteroid populations as a function of time: (a) variation of population at M0.25 (b) variation of population at M1.00 Asteroid s – IJA BWJ 08Jan2009 7 of 34 13/1/09 4:11 PM UT Figure 1(a) shows the behaviour of asteroids in the case where the “Jupiter” has a mass 0.25 MJ, while Figure 1(b) shows the evolution of the objects in the case where the “Jupiter” has the same mass as ours (M0.25 and M1.00 cases respectively). The five time slices shown in each plot are, from top to bottom, t = 0, 1, 2, 5 and 10 Myr (the end of the simulations). Equivalent plots are given in Appendix 2 for all 12 “Jupiter” masses. The y-axis extends to about 400 in both cases, but is a function of bin width. Figure 2 shows the final populations (at 10 Myr) in the M0.25 and M1.00 cases. In order to allow easy comparison, the M0.25 results have been inverted, and placed below those for the M1.00 case. A number of differences are striking, and will be discussed in some detail. Between the two distributions, a number of + marks show the location of various of the Jovian mean motion resonancesii. Working from left to right, the resonances shown are 1:6, 1:5, 1:4, 2:7, 1:3, 3:8, 2:5, 3:7, 1:2, 4:7, 3:5, 5:8, 2:3, 5:7, 3:4, 4:5 and 1:1. It is clear that these resonances play an important role in the evolution of asteroid belts, and we will discuss them further in the following section. Figure 2 The final asteroid distributions for the two cases M0.25 (inverted, lower) and M1.00 (upper). This Figure allows the direct comparison of the final distributions between these two sample cases. Between the two distributions, a series of + marks show the location of a number of key mean motion resonances with the “Jupiter”. From left to right, the resonances shown are 1:6, 1:5, 1:4, 2:7, 1:3, 3:8, 2:5, 3:7, 1:2, 4:7, 3:5, 5:8, 2:3, 5:7, 3:4, 4:5 and 1:1. For a more detailed explanation of resonances, see the discussion section. Figure 3 shows the evolution with time of the number of collisions of asteroidal bodies with the inflated Earth as a function of “Jupiter” mass. The lines, in ascending order from the x-axis, show the total number of collisions versus mass that had occurred at 1, 2, 5 and 10 Myr. The form of these graphs will be discussed in detail in the next section, but note that the final two time slices (5 and 10 Myr) show that the form of the graphs has settled down. Figure 3 Plot showing the number of collisions with the inflated Earth as a function of “Jupiter” mass. The curves show the total number of collisions at a variety of times. Working upwards from the x-axis, the times are 1, 2, 5 and 10 Myr. The total numbers at 10 Myr are presented in Table 1. Discussion Asteroid s – IJA BWJ 08Jan2009 8 of 34 13/1/09 4:11 PM UT Figure 3, which illustrates our core result, is discussed first, then Figures 1 and 2. From Figure 3 it is clear that the notion that any “Jupiter” would provide more shielding than no “Jupiter” at all is incorrect, at least for impactors originating from the asteroid belt. It seems that the effect of a “Jupiter” on the impact flux on potentially habitable worlds is far more complex than was initially thought. With our current Jupiter (M = 1.0 MJ), potentially impacting objects seem to be ejected from the Solar System with such rapidity that they pose rather little risk for planets in the habitable zone (such as the Earth), and therefore, Jupiter offers a large degree of shielding, compared to “Jupiters” of smaller mass, down to about 0.1 MJ. You can see from Figure 3 that planets more massive than Jupiter offer little further improvement. At the other end of the scale, at very small “Jupiter” masses, fewer asteroidal objects are scattered onto orbits which cross the habitable zone, and so, once again, the impact rate is low. The more interesting and complicated situation occurs for intermediate masses, where the giant planet is massive enough to emplace asteroidal objects on threatening orbits, but small enough that ejection events are still infrequent. The situation which offers the greatest enhancement to the impact rate is one located around 0.20 MJ, in our simulations, at which point the planet is massive enough to efficiently inject objects to Earth-crossing orbits, but small enough that the time spent on these orbits is such that the impact rate is significantly enhanced. Had we used a different form of N0(a), the peak could well have been at a different intermediate mass (due to the shifting concentration of material in areas swept by secular resonances), but the broad picture in Figure 3 would be the same. (The double peak in Figure 3 is not a large feature and is possibly a statistical fluctuation, though time consuming further study would be needed to investigate whether this is so.) The effects of the other planets, particularly Saturn and Mars, are pretty much constant between the different runs. However, due to the reduction in the Jovian effect at the lower “Jupiter” masses (particularly below M < 0.2MJ), these planets play a more significant role in these cases, relatively, than at higher “Jupiter” masses, as the overwhelming and masking effects of the “Jupiter” are removed, allowing the effects of the smaller planets to be more clearly observed, and giving them longer to act. From this we can see that our Jupiter is approximately as effective a shield as a giant planet of about 0.05 MJ, which is 15.9 Earth masses (c.f. 14.5 and 17.1 Earth masses for Uranus and Neptune respectively). The M0.01 point (0.01 MJ) corresponds to a planet with a mass just 3.18 times that of Asteroid s – IJA BWJ 08Jan2009 9 of 34 13/1/09 4:11 PM UT the Earth. It is possible that in this case a planet would form in the asteroid region in the order of 10 million years, much depleting the asteroid population (e.g.Wetherill (1991)). In this case the reduction in asteroid numbers could well, in the long term, reduce the number of collisions subsequent to the planet’s formation below that at 10 Myr in Figure 3 (as a result of the planet acting to clear its immediate vicinity through the accretion and ejection of material). Clearly, this planetary system would be significantly different to our own. The discussion of such hypothetical systems is beyond the scope of this work (though we intend to study the complicated problem of alien planetary systems in future work). Let’s turn now to the number of collisions as a function of time. Figure 1 shows a rapid emergence of structures as time passes. In the M0.25 case (Figure 1(a)) the most obvious features are the depletion of asteroids in the outer area of the asteroid belt, and the sharp “spiky” distribution in this region, a result of the effect of mean motion resonances (MMRs), and a large depleted area around 2.5 AU, which is the result of strong secular resonancesiii involving Jupiter. These are discussed in more detail shortly. In the case of M1.00 (Figure 1(b)), a variety of similar features are visible. In fact, at first glance, the distributions appear strikingly similar. However, on closer inspection, a number of significant differences become apparent. First, in the M1.00 case, the asteroid belt is truncated at a smaller heliocentric distance (~4.0 AU vs. ~4.5 AU – it should be noted that in both cases this outer edge has been trimmed to be somewhat closer to the Sun than that of the initial population). Second, the severe depletion around 2.5 AU has shifted to just beyond 2 AU. This is evidence of how the location of the secular resonances in the asteroid belt is a function of the mass of the Jovian planet, whereas the locations of the MMRs are purely determined by the location of that planet alone (though the widths of these resonances, and their strengths are affected by the planet). In passing, we should note that it is well known that MMRs can cause depopulation, as at 3.28 AU (the 1:2 resonance) in both Figures, or help to enhance the population, as can be seen from the small “spikes” located at the orbit of the giant planet (the 1:1 resonance at 5.2 AU, showing objects captured as Jovian Trojans), again in both Figures. The latter corresponds to the temporary capture of objects in Jupiter-like orbits, in a manner similar to that shown for the Centaurs, the parent population of the SPCs (Horner & Evans 2006). Figure 2 allows the reader a better opportunity to see the detailed effects of MMRs on the belt. The + symbols mark the locations of a variety of such resonances (as detailed in the Figure caption), and it is clear that they have played an important role in shaping the young asteroid belts. Note again the Asteroid s – IJA BWJ 08Jan2009 10 of 34 13/1/09 4:11 PM UT 1:2 MMR at 3.28 AU clearly leading to depletion in both the M1.00 and M0.25 belts. What is also clearly visible with this resonance is the way that, as the mass of the “Jupiter” increases, the width of the MMR also increases – this is the case for all MMRs. On the other hand, the effects of secular resonances as the belt evolves show a different variation as a function of planetary mass. The location of these resonances moves with changing Jovian mass, and so they effectively “sweep” through the belt as the mass of the planet increases. It is well known that a resonance called ν6 marks the inner edge of the asteroid belt in our Solar System (the effect of this resonance can be seen in Figure 2 at around 2 AU in the M1.00 case). However, at lower Jovian masses, this resonance lies well within the belt, and results in a broad area of instability (clearly visible at around 2.5 AU in the M0.25 plot), which is probably the main route by which lower “Jovian” masses lead to enhanced impact fluxes. It is clear, therefore, from the examination of Figures 1 and 2, together with those shown in the Appendix 2, that the effects of secular and mean motion resonances play an important role in the removal of objects from the asteroid belt. While the MMRs are locked in semimajor axis, as the mass of the planet is increased, the secular resonances “sweep” through the belt, bringing instability to areas which would otherwise be stable on long timescales. This is clear from Figure 1 and the Figures in Appendix 2 – as the mass of “Jupiter” increases, a secular resonance steadily moves towards the inner edge of the asteroid belt. It also deepens and widens. This doubtless plays a major role in the size and variation of the impact flux on a terrestrial world in these simulations. In fact, it seems quite likely that the evolution of this resonance is the biggest single factor in the rise and fall of the impact rate visible in Figure 3. We believe it to be the ν6 resonance. Further study of this resonance (and perhaps others) in relation to our data is needed, but it will be time consuming. However, a detailed discussion of resonant behaviour is beyond the scope of this work, and indeed, such behaviour is already very well explained in the literature (e.g. Murray & Dermott 1999(a) and (b)), so we will leave our discussion of such resonant effects here. In Table 1, we present the numerical results of our twelve sets of simulations. The various columns detail the mass of the Jovian planet used, the number of impacts (collisions) experienced by the inflated Earth, the number of objects which impact other bodies in the Solar System, the number ejected (in our simulations, any object which reached a heliocentric distance of 1000 AU was considered ejected, and was removed from the calculations), and the number which remain somewhere within the Solar System at the end of the 10 Myr simulations. The variation in the Asteroid s – IJA BWJ 08Jan2009 11 of 34 13/1/09 4:11 PM UT number of objects ejected and remaining in the simulations is far lower than the variation in the impact rate on the Earth. In fact, the simulation in which the fewest asteroids survived is also that which showed the most impacts on the Earth – further evidence of the hugely destabilising effect of the planet in this case. It is interesting to note how the various bodies in our simulations fared, as a whole. Summed over the 12 different setups, the Earth was hit 157794 times (a result of its inflated size), while Mars received 1271 impacts, Jupiter 7783, Saturn 3424, Uranus 32 and Neptune 20. The Sun was hit a total of 558 times, although it should be stressed that, due to the time step chosen for our integrations, we would expect objects dropping to such low perihelion distances to be poorly dealt with in the integrator, so this number should be taken with a large pinch of salt! The effect of inflating the Earth is clearly visible, and given the small numbers of impacts on other bodies, fully justified. Table 1 The fate of asteroidal bodies . A t t = 0 there are 105 bodies. The f igures in the tab le are asteroid numbers n at t = 10 Myr. nremaining Nejected Nother impact nEarth-impact 84174 11166 1730 2930 71409 16104 1612 10875 66635 14083 1175 18107 19642 1109 13189 66060 67629 13753 986 17632 66675 14206 915 18294 16063 926 15611 67400 68757 16746 937 13560 69479 18088 986 11447 10233 935 16897 71935 74005 15316 838 9841 9169 930 18413 71488 M(Jupiter masses) 0.01 0.05 0.10 0.15 0.20 0.25 0.33 0.50 0.75 1.00 1.50 2.00 The evolution of the various asteroid belts considered above would doubtless continue beyond the end of our short simulations. Indeed, it is likely that the stirring of the belts due to mean motion and secular resonances would continue, and that the belts would slowly shed their less stable members. One factor which would prevent the belts studied from eventually evolving into analogues of that in our own Solar System, even in the M1.00 case, is that we do not take account of inter-asteroid interactions in this work (both collisional and gravitational). Nor do we take account of any non- gravitational perturbations, such as the Poynting-Robertson and Yarkovsky effects (Jones 2007(b) and (c) respectively). To incorporate all these features, and to run for the age of our Solar System, presents a huge and daunting technical challenge, and is far beyond the scope of this work. In the Asteroid s – IJA BWJ 08Jan2009 12 of 34 13/1/09 4:11 PM UT future, once computing power has developed enough to handle huge numbers of massive particles in a fully physical environment, such studies will doubtless be feasible and fascinating, but at the moment the incorporation of these features would mean that our simulations would have taken many orders of magnitude longer to run. Conclusions The idea that the planet Jupiter has acted as an impact shield through the Earth's history is one that is entrenched in standard scientific canon. However, when one looks beyond the general understanding of the impact flux on the Earth, it is clear that little work has been done to examine this idea. In the first of an ongoing series of studies, we have examined the question of Jovian shielding using a test population of particles on orbits representative of the asteroids, one of three reservoirs of potentially hazardous objects, the other two being the SPCs and the LPCs. The surprising result of this work is that the status of Jupiter as a shield is now under serious question. For an asteroidal population, it seems that our Jupiter is no better as a shield than a far less massive giant planet would be, were it placed on a similar orbit, and that intermediate mass giants enhance the number of collisions. Figure 3 shows that at intermediate mass the number of collisions at 5Myr and 10 Myr is about double that for our Jupiter. If the Earth had suffered double its actual impact rate there would doubtless have been more mass extinctions, though with what outcome for the biosphere today we can only speculate. Certainly, the risk of an impact large enough to wipe all plants and animals from the globe can only increase as the number of impacts increases. Figures 1, 2, and those in Appendix 2, show that mean motion resonances and at least one secular resonance sculpt the asteroid distribution and are thus responsible for sending impactors our way. We believe that the ν6 secular resonance plays a major role. Future work will continue the study of the role of Jupiter in limiting or enhancing the impact rate on the Earth by examining populations of bodies representative of the Centaurs and Trans-Neptunian objects (source of almost all of the SPCs) and the Oort cloud (source of the LPCs, and the population of potential impactors studied by Wetherill in 1994). We will also examine the effect of Jovian location on the impact fluxes engendered by the three populations, once studies of the effect of its mass are completed. Given the surprising outcome of the present work we hesitate to anticipate future results, though our integrations of the SPCs (which will follow in paper II) do show a comparable outcome to the work described here. Asteroid s – IJA BWJ 08Jan2009 13 of 34 13/1/09 4:11 PM UT Additionally, future work will also consider whether the absence of a Jupiter-like body would change the populations of objects which reside in the three reservoirs, a possible effect ignored in this work. Further into the future, we intend to study wholly different planetary systems, using both hypothetical versions of our youthful Solar System and other planetary systems based upon the rapidly expanding field of known exoplanets. The long term goal is to finally answer, once and for all, the question “Jupiter – friend or foe?”. Acknowledgements This work was carried out with funding from PPARC, and JH and BWJ gratefully acknowledge the financial support given by that body. References Bottke, W.F. et al. (2002). Debiased orbital and absolute magnitude distribution of the near-Earth objects. Icarus 156, 399-433. Chambers, J.E. (1999). A hybrid symplectic integrator that permits close encounters between massive bodies. MNRAS 304, 793-799. Chapman, C.R., Morrison, D. (1994). Impacts on the Earth by asteroids and comets: assessing the hazard. Nature 367, 33-40. Davis, S.S. (2005). The surface density distribution in the solar nebula. ApJ 627, L153-L155 Gomes, R., Levison, H.F., Tsiganis, K., Morbidelli, A. (2005). Origin of the cataclysmic Late Heavy Bombardment period of the terrestrial planets. Nature 435, 466-469. Greaves, J.S. (2006). Persistent hazardous environments around stars older than the Sun. International Journal of Astrobiology 5, 187-190. Horner, J. and Evans, N. W. (2006). The capture of Centaurs as Trojans. Monthly Notices of the Royal Astronomical Society, 367, 1, L20-L23. Jones, B.W. (2007). Discovering the Solar System, 2n d edition. John Wiley & Sons, (a) Chapter 1 (b) p79 (c) p84. Laakso, T., Rantala, J., Kaasalainen, M. (2006). Gravitational scattering by giant planets. A&A 456, 373-378. Morbidelli, A., Bottke, W.F., Froeschlé, Ch., Michel, P. (2002). Origin and Evolution of Near-Earth Objects. Asteroids III, 409-422. University of Arizona Press. Morris, S.C. (1998). The evolution of diversity in ancient ecosystems: a review. Philosophical Transactions of the Royal Society of London B 353, 327-345. Murray, C.D., Dermott, S.F. (1999). Solar System Dynamics. CUP, (a) Chapter 8 (b) Chapter 9. Asteroid s – IJA BWJ 08Jan2009 14 of 34 13/1/09 4:11 PM UT Oort, J.H. (1950). The structure of the cloud of comets surrounding the Solar System, and a hypothesis concerning its origin. Bulletin of the Astronomical Institutes of the Netherlands XI, 408, 91-110. Ward, W.R. (2005). Early clearing of the asteroid belt. Abstracts of the 36th Lunar and Planetary Science. Abstract 1491, 2 pages. Ward, W.R. and Brownlee, D. (2000). Rare Earth:Why Complex Life is Uncommon in the Universe. Copernicus, 238-239. Wetherill, G.W. (1991). Occurrence of Earth-like bodies in planetary systems. Science 253, 535-538. Wetherill, G.W. (1994). Possible consequences of absence of Jupiters in planetary systems. Astrophysics & Space Science 212, 23-32. Asteroid s – IJA BWJ 08Jan2009 15 of 34 13/1/09 4:11 PM UT Append ix 1: The Asteroid D i stribut ion In choosing the form of N0(a), the number of asteroidal bodies per unit interval of semimajor axis a at zero time in our simulations, we faced huge uncertainties. There is a range of models representing the distribution over a of dust and small bodies when the Solar System was young (e.g.Davis 2005). Another uncertainty is to what extent the abundant icy-rocky bodies that formed in the cooler conditions beyond the outer edge of the asteroid belt, mixed inwards. Whatever the details, it is crucial to remember that gravitational stirring by a giant planet orbiting beyond the asteroid belt prevented the formation of a planet between it and Mars. The asteroidal population must have been largely confined to this zone. We have used a form for N0(a) that is similar to the form implicitly favoured by Davis (2005), who derives the surface density of the early solar nebula by a cumulative mass model involving all the planets as they are today. Over the space between Mars and Jupiter )1 / 2 (1) N 0 a( ) = k a " amin ( fits his graph well enough, given the uncertainties. The value of amin has been set by us at three Martian Hill radii beyond the orbit of Mars, and thus at amin, N0(a) = 0. This is reasonable because Mars would have cleared bodies closer to its orbit than this, and the asteroid-asteroid collision speeds near amin would have been high, resulting in further depletion. The outer boundary amax at t = 0 is at three giant Hill radii interior to the giants orbit. The )1 / 2 dependence is within the ( a " amin range of possibilities, and gives us a greater number of asteroidal bodies at larger a than some other possible dependences. This is to our advantage because, with the giant planet being far more ! important than Mars at sending asteroids towards the Earth, it increases the number of collisions for a given t = 0 population. Remember that we are interested in the effect of the mass of the giant planet on the impact rate of asteroidal bodies on the Earth. The exact form of N0(a) is unlikely to affect our conclusion that Jupiter is no better as a shield than a far less massive giant planet, and that intermediate mass giants are poor shields. ! Asteroid s – IJA BWJ 08Jan2009 16 of 34 13/1/09 4:11 PM UT Append ix 2: Evolut ion of the various asteroid belts w ith time. The following Figures show the evolution of the asteroid belts as a function of time for each of our 12 “ J upiter ” simulations. In order, we show the cases from M0 .0 1 to M2 .0 0, sequentially by increasing mass. The five time slices shown are take at t = 0 Myr, 1 Myr, 2 Myr, 5 Myr and 10 Myr. The variations in the populations due to the changes in the mass of the giant planet are clear to see. The x- axis extends to about 400 in all cases, but is a function of bin width. Asteroid s – IJA BWJ 08Jan2009 17 of 34 13/1/09 4:11 PM UT M0 .0 1 Asteroid s – IJA BWJ 08Jan2009 18 of 34 13/1/09 4:11 PM UT M0 .0 5 Asteroid s – IJA BWJ 08Jan2009 19 of 34 13/1/09 4:11 PM UT M0 .1 0 Asteroid s – IJA BWJ 08Jan2009 20 of 34 13/1/09 4:11 PM UT M0 .1 5 Asteroid s – IJA BWJ 08Jan2009 21 of 34 13/1/09 4:11 PM UT M0 .2 0 Asteroid s – IJA BWJ 08Jan2009 22 of 34 13/1/09 4:11 PM UT M0 .2 5 Asteroid s – IJA BWJ 08Jan2009 23 of 34 13/1/09 4:11 PM UT M0 .3 3 Asteroid s – IJA BWJ 08Jan2009 24 of 34 13/1/09 4:11 PM UT M0 .5 0 Asteroid s – IJA BWJ 08Jan2009 25 of 34 13/1/09 4:11 PM UT M0 .7 5 Asteroid s – IJA BWJ 08Jan2009 26 of 34 13/1/09 4:11 PM UT M1 .0 0 Asteroid s – IJA BWJ 08Jan2009 27 of 34 13/1/09 4:11 PM UT M1 .5 0 Asteroid s – IJA BWJ 08Jan2009 28 of 34 13/1/09 4:11 PM UT M2 .0 0 Asteroid s – IJA BWJ 08Jan2009 29 of 34 13/1/09 4:11 PM UT Figures for inclus ion w ith the text Figure 1a – variation of popu lation at M0 .2 5 Asteroid s – IJA BWJ 08Jan2009 30 of 34 13/1/09 4:11 PM UT Figure 1b. Variat ion of popu lation at M1 .0 0 Asteroid s – IJA BWJ 08Jan2009 31 of 34 13/1/09 4:11 PM UT Figure 2. Fina l d i stribut ions obta ined at M0 .2 5 and M1 .0 0, show ing the locations of key Mean- Motion Resonances. Asteroid s – IJA BWJ 08Jan2009 32 of 34 13/1/09 4:11 PM UT Figure 3. Evolut ion of col l i s ion rate with Earth as a function of Jup iter mass Asteroid s – IJA BWJ 08Jan2009 33 of 34 13/1/09 4:11 PM UT i In addition, the mass of Mars was increased sligh tly from its actual mass of 0.107 Ear th masses, in order to account for any extra accretion wh ich would have occurred as a result of a lower mass “Jupiter”. The new Mars was g iven a somewhat arbitrary mass of 0.4 Ear th masses . Rather than attempt to recursively mod ify the Mars mass as Jupiter itself varied , we chose a value in termediate between the curren t mass of the planet and that of the Earth . The mass of Mars makes little difference to our simulations, since it is held constan t, and the p lanet is in ter ior to the inner boundary of the belt. Even though a yet more massive Mars wou ld have given a slightly larger per turbation to the inner asteroids , the small increase would have had no significan t effect on our results . ii Mean motion resonances are given in the form n :m, a simple integer ratio where, in the time it takes “Jupiter” to complete n orbits another object completes m orbits. For example, an asteroid located in the 3:7 mean motion resonance (n = 3, m = 7) wou ld complete 7 orbits in the time it takes “Jupiter” to complete 3. iii In much the same way as mean motion resonances result from a commensurability of the orb ital periods of a planet and a g iven ob ject, secu lar resonances occur as a resu lt of commensurability between the precession rates of the perihelion or the longitude of the ascending node (or bo th). For examp le, if the ascending node of Jupiter’s orbit precesses at the same rate as that of an asteroid , the two w ill be locked in a secu lar resonance, which can lead to significan t alteration of the astero ids orbit over time, as energy is transferred between the two bodies . A detailed discussion of secu lar resonances is beyond the scope of th is work, but we direct the in terested reader to e.g. Murray and Dermott (1999(b)) for more information. Asteroid s – IJA BWJ 08Jan2009 34 of 34 13/1/09 4:11 PM UT
0805.2197
1
0805
2008-05-15T01:56:41
Cosmological Radar Ranging in an Expanding Universe
[ "astro-ph" ]
While modern cosmology, founded in the language of general relativity, is almost a century old, the meaning of the expansion of space is still being debated. In this paper, the question of radar ranging in an expanding universe is examined, focusing upon light travel times during the ranging; it has recently been claimed that this proves that space physically expands. We generalize the problem into considering the return journey of an accelerating rocketeer, showing that while this agrees with expectations of special relativity for an empty universe, distinct differences occur when the universe contains matter. We conclude that this does not require the expansion of space to be a physical phenomenon, rather that we cannot neglect the influence of matter, seen through the laws of general relativity, when considering motions on cosmic scales.
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 000–000 (0000) Printed 15 April 2017 (MN LATEX style file v2.2) Cosmological Radar Ranging in an Expanding Universe⋆ Geraint F. Lewis1, Matthew J. Francis1, Luke A. Barnes2,1, Juliana Kwan1 & J. Berian James3,1 1Institute of Astronomy, School of Physics, A28, University of Sydney, NSW 2006, Australia 2Institute of Astronomy, University of Cambridge, Madingley Rd, Cambridge, CB3 0HA, UK 3Institute for Astronomy, Royal Observatory, Edinburgh, EH9 3HJ, UK 15 April 2017 ABSTRACT While modern cosmology, founded in the language of general relativity, is almost a century old, the meaning of the expansion of space is still being debated. In this paper, the question of radar ranging in an expanding universe is examined, focusing upon light travel times during the ranging; it has recently been claimed that this proves that space physically expands. We generalize the problem into considering the return journey of an accelerating rocketeer, showing that while this agrees with expectations of special relativity for an empty universe, distinct differences occur when the universe contains matter. We conclude that this does not require the expansion of space to be a physical phenomenon, rather that we cannot neglect the influence of matter, seen through the laws of general relativity, when considering motions on cosmic scales. Key words: cosmology: theory 1 INTRODUCTION The question "Is space really expanding?" has recently (re)surfaced in the literature, with varying views on whether the expansion of space is a phenomenon which can be di- rectly observed. Whiting (2004) entered the fray with a New- tonian analysis of particles detached from the Hubble flow, showing their motion does not agree to the simple viewpoint of expanding space-time as a form of force. Chodorowski (2006) also considered the physical implications of the ex- pansion of space, suggesting that the superluminal expan- sion of distant objects, often touted as the proof of expan- sion, can be removed with a transformation to conformal coordinates, and hence cannot be physical, although it has been shown that superluminal expansion does in fact re- main (Lewis et al. 2007). Francis et al. (2007) assessed the situation in detail, showing that the view of expanding cos- mologies as expanding space is a valid interpretation as long as the equations of relativity are used to guide common- sense. Recently, Abramowicz et al. (2007) considered radar ranging of a distant galaxy in expanding cosmologies and concluded that the fact that the radar and Hubble distance, from d = Hov, differ in all but an empty universe, that space must really expand1. In a counter argument, Chodorowski (2007) again considers radar ranging in open cosmological models. Instead of examining distances, he focuses upon the transit time of light in usual cosmological coordinates and its conformal representation. With this he reveals that in the former coordinates the paths are asymmetrical in tran- sit time, taking longer on the return journey, whereas in conformal coordinates, the light travel times to and from the distant galaxy are equal. Hence, he concludes that the expansion of space is a coordinate dependent effect which can be made to disappear with the correct coordinate trans- form, and therefore the expansion of space is not a physical phenomenon. In this contribution we examine the recent debate on cosmological radar ranging of objects, clarifying some of the issues discussed by other authors and demonstrate that while expanding space remains a useful concept, such ex- periments in no way require expanding space as a physical effect. The issue of radar ranging in Friedmann-Lemaitre- Robertson-Walker (FLRW) universes is addressed in Sec- tion 2, conformal coordinates and generalized to include the motion of an accelerating observer in Section 4. A compari- son of our results in light of previous studies is presented in ⋆ Research undertaken as part of the Commonwealth Cosmology Initiative (CCI: www.thecci.org), an international collaboration supported by the Australian Research Council 1 The title of their paper 'Eppur si espande' is a slight rewording of the mutterings of Galileo after his trail for heresy by the Vat- ican in 1633. It seems to demonstrate that these authors believe that space 'really' expands. 2 Lewis et al. Figure 1. Radar ranging in a fully conformal representation of an open universe. In both, a light ray emitted from the origin is represented by a dotted path, while a comoving observer is represented by the solid sloping line and another observer sits at the origin (r = 0). The green line shows the null geodesic representing the laser ranging beam originating at r = 0, which is reflected back by the distant comoving observer. In the left hand panel, the dashed lines represent curves of constant time in the FLRW metric, while on the right the dashed lines represent intervals of constant conformal time. The red and the blue paths represent the times measured by the observers for the outward and returning rays respectively. Section 5, where we also offer our conclusions on the issue of expanding space. tion 1) then it is easy to construct a conformally flat trans- formation such that 2 2 BACKGROUND 2.1 FLRW & Conformal Cosmology Modern cosmological models are described by the relativistic equations for a homogeneous and isotropic distribution of matter and energy. Many elementary textbooks show how, under these assumptions, the spacetime of the universe is described by the FLRW metric, whose invariant interval is given by ds2 = dt′2 −a2(t′) (cid:2)dx2 + R2 oS 2 k(x/Ro)(dθ2 + sin2θ dφ2)(cid:3) , (1) where Sk(x) = sin x, x, sinh x for spatial curvatures of k = +1 (closed), k = 0 (flat) and k = −1 (open) respectively, with the curvature given by R−2 o ; note, c = 1. The scale factor, a(t′), governs the dynamics of the expansion and is dependent upon the relative mix of matter and energy in the universe. Conformal transformations preserve angles at a point and are important in geometry. For the purposes of this study, the space-times of interest are those which are 'con- formally flat', such that the metric of a curved space time is related to that of the flat space-time of special relativity via g = Ω(x)gf lat. (2) ds2 = a2(η)(dη2 − dx2), (3) defining the conformal time to be related to the universal time through dt′ = a(η)dη. Throughout this paper, such a transformation will be referred to as a partial conformal transformation; clearly, in the (x, η) coordinates light rays (ds = 0) will trace out the undistorted light cones of special relativity. Typically, conformal representations of FLRW cosmolo- gies employ only the partial transform, although this flattens the radial part of the metric, and hence will not in general produce flat SR-like light cones in fully 4D space-time. For a general cosmology with spatial curvature, a full confor- mal transformation is required to make the metric confor- mally flat in all four dimensions, as explored in detail by Infeld & Schild (1945), in which the comoving radial coordi- nate must be transformed as well. For instance, in an open universe the fully conformal coordinates [see Chodorowski (2007) and Lewis et al. (2007) for the derivation and more details] are: r = Aeη sinh χ t = Aeη cosh χ (4) where A is a constant with χ = x/R0. We have referred to the fully conformal coordinates as (r, t), matching those employed in previous studies. It is now expected that light rays follow the classic special relativistic light cones in these coordinates. Clearly, since the spatial part of a flat (k = 0) FLRW metric is already flat, the partial and fully conformal where Ω(x) is an arbitrary function. If we take a particu- lar two-dimensional slice of the FLRW metric above (Equa- 2 In fact, any two dimensional subspace is conformally flat; see Appendix 11C of Hobson, Efstathiou, & Lasenby (2005) transformations are equivalent. In this study, both the par- tial and fully conformal transformations will be considered, to provide a comparison to previous studies and allow a cor- respondence with the flat space-time of special relativity as Ωo → 0. 3 RADAR RANGING The principle of radar ranging is simple; to calculate the distance to a distant object, fire a radar pulse at it and time the interval ∆τ until the beam returns; here τ is the proper time as measured by an observer sending out and receiving the radar beam. From this, it is straightforward to define a radar distance Drad = ∆τ 2 (5) assuming c = 1. For cosmological cases, it is usual to mea- sure the comoving distance to a galaxy whereas the time is measured by an observer at the origin. Clearly this is a well defined experiment as we are asking about the time ticked off along a world line, an observable quantity. Remember, how- ever, that Chodorowski (2007) asked a somewhat different question, namely how much time passes on the individual outward and return journeys. Figure 1 presents the radar ranging experiment for an open universe in fully conformal coordinates; here a radar pulse (green line) leaves the origin, and is reflected back from a distant, comoving (constant spatial FLRW coordinates) object, later being received back at the origin. Note that in this conformal picture, comoving observers move along paths which originate at the origin and move along a line of constant slope given by = tanh(χ), dr dt where χ is the comoving coordinate of the fundamental ob- server, with light rays traveling at 45o. Given this picture, it is simple to see that the differing travel times noted by Abramowicz et al. (2007) are simply an issue of synchronic- ity. (6) While the two panels in Figure 1 both represent a fully conformal representation of an Ω < 1 universe, each dis- play differing lines of simultaneity; in the left-hand panel the dashed line represents constant τ , or proper time as measured by fundamental observers. In standard FLRW uni- verses, these hyperbolae represent times (slices) of equal matter/energy density as seen by comoving observers. In the right-hand panel, the lines of simultaneity are represented by lines of constant comoving coordinate t. An examination of the left-hand panel, whose lines of synchronicity represent constant cosmological time in the FLRW metric, reveals that both observers agree that the duration of the outward light ray (red paths of the observers) is shorter than the return journey (shown in blue). Both observers agree on the length of time each leg of the journey took. However, an exami- nation of the right-hand panel, where lines of synchronicity are defined by slices of constant conformal coordinate time t, reveal a different picture. Now, both observers agree that the duration of the outward and return journey are equal, but they disagree on how much time the journey took in total. Cosmological Radar Ranging 3 How are we to interpret this picture? Clearly the is- sue lies with the fact that measuring the journey time for any individual leg depends upon the comparison of clocks in differing inertial frames, a message stated by Chodorowski (2007) who considered a Milne (special relativistic universe) and inertial observers. This paper shows that this is gen- erally true in the relativistic interpretation of any FLRW universe and the "different spacetime structure(s)" (Figures 3-4 of Abramowicz et al. (2007)) are purely due to the way they have chosen define synchronicity and slice up space- time. 4 CARRYING A CLOCK Of course, a major problem with considering the path of a photon is that it is null and the affine parameter that de- scribes its path has no physical significance. But what if the photon is replaced with an observer who can tick off their own proper time on the journey? To this end, we con- sider an observer who accelerates away from the origin in a rocket with a constant proper acceleration. The acceleration is continued for a fixed amount of proper time ∆τr (for the rocketeer), followed by a coasting period where the rocket is turned off. The rocket is then swung round as the rocketeer accelerates back towards the origin, with the rocket is fired for a time 2∆τr before again entering a coasting period. The rocket is turned round again so that the acceleration is away the origin and fired for a time ∆τr. Within the flat space- time of special relativity, such a symmetric path will bring the rocketeer to rest at the origin at the conclusion of their journey. A general discussion of the influence of expanding space on the motion of an accelerating observer will be presented in a future contribution (Kwan & Lewis in preparation), but here we consider three specific cases. The top row of Figure 2 presents the journey of two rocketeers in an expanding, open universe; in this case, Ωo = 0.001 and the resultant space- time structure should be akin to the Milne (empty) uni- verse. The first rocketeer leaves from the origin, and carries out the symmetric accelerations outlined above. The second undertakes the same journey, starting at the same cosmic time, t′, but at a different comoving spatial location. To emphasise the issue of synchronicity when making compar- isons between different coordinate systems, both journeys have been represented twice, in partially conformal coordi- nates on the left and in fully conformal coordinates on the right. As expected, in either representation the rocketeers return to rest at the the origin of their journey. However, an examination of the rocketeer's path in the partial confor- mal coordinates reveals that it is not symmetric, with more time η spent reaching the most distant point from the ori- gin, than the return journey. Furthermore, the mid-point of the journey as seen by the rocketeer (where the path colour switches from red to blue) also does not correspond to the most distant point reached in the journey. Moving to the fully conformal picture (right hand panel) we would expect this path to be virtually the same as that seen in special relativity (see Chodorowski 2007, for a dis- cussion of the behaviour of the conformal transformation for an open universe as Ωo → 0); this is precisely what is seen, with the path of the rocketeer starting at the origin 4 Lewis et al. Figure 2. Cosmological radar trips for rocketeers in partially conformal (left panels) and fully conformal (right panels) representations of three open universes with varying matter content. For each universe, we have considered the paths of two rocketeers, one leaving from the origin and another that starts out from an arbitary comoving location. The colour sections of the path describe the state of the rocket engine of the rocketeer. The first green region shows the initial outwards acceleration, followed by the red period corresponding to a coasting phase with no acceleration. The rocket is then swung around to accelerate towards the origin and this is reprented by the blue and then green region. The change from blue to green indicates the midway point in this acceleration phase as measured in the rocketeers time. Following this there is another coasting period, shown in red and finally the rocket is swung around again to accelerate away from the origin and this corresponds to the final blue region. The black dashed lines are the paths taken by a comoving observers starting from the origin, while the pink dashed lines are for comoving observers starting from the same comoving location as the second rocketeer. For the fully conformal cases, we have used A = 5.3 × 10−5, 8.3 × 10−2 and 28.3 for Ωo = 0.001, 0.500 and 0.999 respectively. being symmetric in the conformal time t. Remember that in this representation, observers at fixed comoving distances are now seen on sloping lines and the rocket still reaches its greatest comoving distance during its deceleration (dark blue), although the rocketeer reaches the maximum coordi- nate distance r at the mid-point of the journey. The path of the rocketeer who starts at the non-zero comoving coor- dinate is a little more complex, and quite different to that seen in the partial conformal coordinates, but it also returns to its origin after a symmetric flight. The second row of Figure 2 presents identical journeys in an open, matter dominated universe with Ωo = 0.500, again with the left hand panel presenting the partial con- formal coordinates and the right hand panel presenting the fully conformal representation. While several aspects of the paths of the rocketeer are similar to those seen in Fig- ure 2, there is a very important difference, namely that even though the paths are symmetric in terms of acceleration and coasting time for the rocketeer, they do not come to rest at the origin of their journey. In fact, in this open case, the rocketeer over shoots and even when their rocket is turned off, they are still moving away from their origin. Exactly the same behaviour is seen in the fully conformal picture. Al- though, it may seem that this asymmetry, absent in the case of a static universe (see Figure 1), is an indication that space is really expanding, this effect only occurs with the introduc- tion of matter content to act on the motion of the rocketeer. In any case, we might naively expect that if space is ex- panding, then the journey is longer on the way back than it was on the way forwards [c.f. figure 4 of Abramowicz et al. (2007)] and hence we might predict under-shooting, rather than over-shooting, the origin. It must be emphasised that this over shoot is no different in the Newtonian limit of the FLRW metric without expansion; using Gauss's law we can see that at a given radius R from the origin, the rocketeer experiences a gravitational acceleration towards the origin due to the mass contained by a sphere of radius R. This imaginary sphere changes size throughout the journey but the acceleration from gravity remains pointed towards the origin during the entire trip. Thus in addition to the thrust provided by the rocket, the rocketeer recieves at all times an additional push towards the origin. We should not, therefore, be surprised to find the rocketeer overshoots. Thinking very simply about the effects of gravity, rather than the more nebulous expansion of space, gives a much simpler intuitive view. This asymmetry is even more apparent in the bottom row of Figure 2 which again presents the rocketeers' paths in partial and fully conformal coordinates, except now the matter density is Ωo = 0.999; in this case, the universe is approaching the spatially flat Ωo = 1 universe and hence the slope of the comoving observer in the fully conformal picture is approaching that of the partially conformal case (as noted previously, for spatially flat models the partial and fully conformal transforms are equal). The key differ- ence between the three cases represented in Figure 2 is that the matter content of each universe increases, which we are free to interpret as the cause of the increasing asymmetry in the paths, rather than the asymmetry being caused by the expansion of space. Again this overshoot can be understood in the Newtonian limit of the FLRW metric without expan- sion; we would expect approximately the same behaviour to occur for a rocketeer travelling at non-relativistic speeds in a Newtonian potential to a destination close by. The implica- tions of this journey on the question of whether space really expands are presented in Section 5. 4.1 Synchronizing Clocks It is clear that the timing issues related to the radar ex- periment are related to the synchronization of clocks (as are many of the problems and apparent paradoxes in rel- Cosmological Radar Ranging 5 ativity). However, the universe itself provides a clock that can be employed to provide at least a working definition of synchronicity using the density of matter/energy; as the uni- verse expands, the density of matter falls and dashed lines in the right hand panel of Figure 1 correspond to slices of cosmic time in the FLRW metric along which the density is equal. Clearly, such synchronization is not possible in the Milne universe, as, being empty, there is no density yardstick with which to tick off cosmic time. However, the situation is the same in a universe containing only a cosmological con- stant term (with equation of state w = −1) as the energy density remains a constant. With either of these cases, or any other universe model, we can imagine a 'test CMB', a homogeneous and isotropic bath of photons of negligible energy fraction defining the Hubble rest frame and comoving coordinates. By measur- ing the temperature of the CMB all observers can calibrate their clocks to each other. Interestingly, even in the Milne universe which contains no gravitating energy, this test CMB provides a universal clock giving an excellent demonstration of how we can observe an apparent expansion of space in a universe known to be completely empty and equivalent to special relativity. Clearly the interpretation of expanding space 'stretching' photons causing them to redshift, while being a useful teaching aid, does not describe a casual phys- ical phenomenon if we can observe this effect in Minkowski space. 5 CONCLUSIONS: SO, IS SPACE REALLY EXPANDING? This work has grown out of a recent exchange in the litera- ture on the question on whether space 'really' expands. In a previous contribution (Francis et al. 2007), we argued that while space is 'completely and utterly empty' (to quote Steve Weinberg), it is perfectly valid to interpret the equations of relativity in terms of an expanding space. The mistake is to push analogies too far and imbue space with physical prop- erties that are not consistent with the equations of relativity. In their recent work, Abramowicz et al. (2007) showed that, in all but an empty universe, distances derived from the Hubble law and radar ranging differ and hence "one must conclude that space is expanding". But how is this differ- ence occurring? Is the expansion of space acting on a light ray (or even a rocketeer) as they travel through the universe? We can think of space as a rubber sheet that stretches to wash out peculiar motions and drives everything back into the Hubble flow [see Barnes et al. (2006)]. However, it is the presence of matter that necessitates the inclusion of gravi- tational forces upon the motion of the rocketeers and it is this - the changing gravitational influence of matter in the universe on the rocketeers - that causes the increasing asym- metry moving down the panels in Figure 2, not that space physically expands. In closing, we state that it is a fools errand to search for the truth of the existance of expanding space; not only because it is dependant upon a choice of coordinates, but also because general relativity is represented by Newtonian physics in the weak field limit and the global behaviour of the FLRW metric always reduces to Newtonian gravity in the limit of the local universe with no need for expanding 6 Lewis et al. space. While the expansion of space is a valid (but danger- ous picture when working with the equations of relativity, any attempts) to obtain observations to address the ques- tion of whether galaxies are moving through static space or are carried away by the expansion of space are doomed to failure. ACKNOWLEDGMENTS GFL acknowledges support from ARC Discovery Project DP0665574. MJF is supported by a scholarship partially funded by the Science Faculty at The University of Sydney. JK is supported by an Australian Postgraduate Award. REFERENCES Abramowicz M. A., Bajtlik S., Lasota J.-P., Moudens A., 2007, AcA, 57, 139 Barnes L. A., Francis M. J., James, J. B., Lewis G. F. 2006, MNRAS, 373, 382 Chodorowski M. J., 2005, PASA, 22, 287 Chodorowski M. J., 2006, astro, arXiv:astro-ph/0610590 Chodorowski M. J., 2007, MNRAS, 378, 239 Davis T. M., Lineweaver C. H., 2001, AIPC, 555, 348 Davis T. M., Lineweaver C. H., 2004, PASA, 21, 97 Davis T. M., Lineweaver C. H., Webb J. K., 2003, Am. J. Phys., 71, 358 Francis M. J., Barnes L. A., James J. B., Lewis G. F., 2007, PASA, 24, 95 Heyl J. S., 2005, PhRvD, 72, 107302 Hobson M. P., Efstathiou G. P., Lasenby A. N., 2005, General Relativity: An Introduction for Physicists, Cam- bridge, Cambridge University Press Infeld L. & Schild A., 1945, Phys. Rev., 68, 250 Lewis G. F., Francis M. J., Barnes L. A., James J. B., 2007, MNRAS, 381, L50 Peacock, J., 2006, www.roe.ac.uk/∼jap/book/additions.html Whiting A. B., 2004, Obs, 124, 174
astro-ph/0508599
3
0508
2005-10-11T00:35:04
Relationship between the rise width and the full width of gamma-ray burst pulses and its implications in terms of the fireball model
[ "astro-ph" ]
Kocevski et al. (2003) found that there is a linear relation between the rise width and the full width of gamma-ray burst pulses detected by the BATSE instrument based on their empirical functions. Motivated by this, we investigate the relationship based on Qin et al. (2004) model. Theoretical analysis shows that each of the two quantities, the rise width and the full width of observed pulse, is proportional to $\Gamma^{-2}\Delta\tau_{\theta,\rm FWHM}\frac{R_c}{c}$, where $\Gamma$ is the Lorentz factor for the bulk motion, $\Delta \tau_{\theta,\rm FWHM}$ is a local pulse's width, $R_{c}$ is the radius of fireballs and c is the velocity of light. We employ the observed pulses coming from four samples to study the relationship and find that: (1) Merely the curvature effect could produce the relationship with the same slope as those derived from Qin et al. (2004) model in the rise width vs. the full width panel. (2) Gamma-ray burst pulses, long or short ones (selected from the short and long GRBs), follow the same sequence in the rise width vs. the full width panel, with the shorter pulses at the end of this sequence. (3) All GRBs may intrinsically result from local Gaussian pulses. These features place constraints on the physical mechanism(s) for producing long and short GRBs.
astro-ph
astro-ph
Chinese Journal of Astronomy and Astrophysics manuscript no. (LATEX: ms.tex; printed on November 28, 2018; 12:56) Relationship between the rise width and the full width of gamma-ray burst pulses and its implications in terms of the fireball model Rui-Jing Lu1,2,3 ⋆, Yi-Ping Qin1,2 and Ting-Feng Yi2 1 National Astronomical Observatories/Yunnan Observatory, Chinese Academy of Sciences, P. O. Box 110, Kunming, Yunnan, 650011, P. R. China 2 Physics Department, Guangxi University, Nanning, Guangxi 530004, P. R. China 3 The Graduate School of the Chinese Academy of Sciences Received 2005 month day; accepted 2005 month day Abstract Kocevski et al. (2003) found that there is a linear relation between the rise width and the full width of gamma-ray burst pulses detected by the BATSE instrument based on their empirical functions. Motivated by this, we investigate the relationship based on Qin et al. (2004) model. Theoretical analysis shows that each of the two quantities, the rise width and the full width of observed pulse, is proportional to Γ−2∆τθ,FWHM Rc c , where Γ is the Lorentz factor for the bulk motion, ∆τθ,FWHM is a local pulse's width, Rc is the radius of fireballs and c is the velocity of light. We employ the observed pulses coming from four samples to study the relationship and find that: (1) Merely the curvature effect could produce the relationship with the same slope as those derived from Qin et al. (2004) model in the rise width vs. the full width panel. (2) Gamma-ray burst pulses, long or short ones (selected from the short and long GRBs), follow the same sequence in the rise width vs. the full width panel, with the shorter pulses at the end of this sequence. (3) All GRBs may intrinsically result from local Gaussian pulses. These features place constraints on the physical mechanism(s) for producing long and short GRBs. Key words: gamma rays: bursts -- gamma rays: theory -- methods: data analysis 2 R.-J. Lu, & Y.-P. Qin & T.-F. Yi 1 INTRODUCTION Although the mechanism underlying the gamma-ray bursts is still an unsolved puzzle, it is generally accepted that the large energies and the short timescales involved require the gamma-rays to be produced in a stage of fireball which expand relativistically (see, e.g., Goodman 1986; paczynski 1986). An individual shock episode gives rise to a pulse in the gamma-ray light curve, and superposition of many such pulses creates the observed diversity and complexity of light curves (Fishman et al. 1994). Therefore, the temporal characteristics of these pulses hold the key to the understanding of the prompt radiation of gamma-ray bursts. It is generally believed that some well-separated individual pulses represent the fundamental constituent of GRB time profiles (light curves) and appear as asymmetric pulses with a fast rise and an exponential decay (FRED), and many pulses have FRED-like shapes. What result in the observed light curves? According to Ryde & Petrosian (2002), the simplest scenario accounting for the observed GRB pulses is to assume an impulsive heating of the leptons and a subsequent cooling and emission. In this scenario, the rising phase of the pulse, which is referred to as the dynamic time, arises from the energizing of the shell, while the decay phase reflects the cooling and its timescale. However, in general, the cooling time for the relevant parameters is too short to explain the pulse durations and the resulting cooling spectra are not consistent with observation (Ghisellini et al. 2000). As shown by Ryde & Petrosian (2002), this problem could be solved when the curvature effect of the expanding fireball surface is taken into account. The diversities of gamma-ray light curves in morphology could be interpreted within the standard fireball model (Rees & M ´esz´aros 1992), and the observed FRED structure was found to be interpreted by the curvature effect as the observed plasma moves rel- ativistically towards us and appears to be locally isotropic (e.g., Fenimore et al. 1996, Ryde & Petrosian 2002; Kocevski et al. 2003, hereafter Paper I). Several investigations on modeling pulse profiles have previously been made (e.g., Norris et al. 1996; Lee et al. 2000a, 2000b; Ryde & Svensson 2000; Ryde & Petrosian 2002; Borgonovo & Ryde 2001; Paper I), they derived several flexible functions to describe the profiles of indi- vidual pulses based on empirical or semi-empirical relations. E.g., as derived in detail in Paper I, a FRED pulse can be well described by the equation (22) or (28). Using this model, they found that there is a linear relationship between the full width at half- maximum, often denoted FWHM, and the rise width of gamma-ray burst pulses detected by the BATSE instrument (see Fig. 10 in paper I), and the same result can be found in the gamma-ray burst pulses detected by the anti-coincidence shield of the spectrometer (SPI) of INTEGRAL (see Fig. 5a Ryde et al. 2003). ⋆ E-mail: [email protected]; [email protected] Relationship between the rise width and the full width and its implications 3 Qin (2002) has derived in detail the flux function based on the model of highly symmetric expanding fireball, where the Doppler effect of the expanding fireball surface is the key factor to be concerned, and then with this formula, Qin (2003) studied how emission and absorbtion lines are affect by the effect. Recently, Qin et al. (2004) presented the formula in terms of count rates. Based on this model, some relations, a power law relationship between the observed pulse width and energy (Qin et al. 2005a), an anti- correlation between the power law index and the local pulse width (Jia et al. 2005), and correlations between spectral lags and some physical parameters, such as Lorentz factor and the fireball radius (Lu et al. 2005a), have emerged. At the same time, some characteristics have been found, such as a reverse S-feature curve in the decay phase (Qin et al. 2005b) and an inflexion from concavity to convexity in the rising phase (Lu et al. 2005b) of the light curve determined by equation (21) in Paper II, The Combination of these knowledge is suggestive of a potential relationship between the rise width and the full width of the observed pulse, which motivates us to investigate the relationship found by Kocevski et al. based on Qin model and explore its implications in terms of the fireball model. Although the origins of short GRBs and long GRBs are not yet clear, it is generally suggested that short GRBs are likely to be produced by the merger of compact objects while the core collapse of massive stars is likely to give rise to long GRBs (see Zhang & M ´esz´aros 2004; Piran 2005). Many properties, such as luminosity, < V /Vmax >, the angular distribution, the energy dependence of the duration, and the hard-to-soft spectral evolution, even pulses profile of short GRBs , are also similar to those of long GRBs (e.g., Schmidt 2001; Ramirez-Ruiz & Fenimore 2000; Lamb et al. 2002; Ghirlanda et al. 2004; Cui et al. 2005), which indicate that the GRBs appear to have the same emission mechanism and possibly different progenitors for long and short bursts. Motivated by this, we also investigate the temporal structure of narrow pulses with durations shorter than 1 s (FWHM) from short GRBs and long GRBs. This paper is organized as follows. In section 2, we investigate the temporal char- acteristics of light curves of GRBs based on Qin model. In section 3, we examine the relationship between the rise width and the full width of observational pulses and ex- plore its possible implication in terms of the fireball model. Discussion and conclusions will be presented in the last section. 2 THE THEORETICAL ANALYSIS As derived in detail in paper II, the expected count rate of the fireball within frequency interval [ν1, ν2] can be calculated with 2πR3 eτθ,min eI(τθ)(1 + βτθ)2(1 − τ + τθ)dτθ R ν2 c Reτθ,max hcD2Γ3(1 − β)2(1 + kτ )2 ν1 g0,ν (ν0,θ ) ν dν . (1) C(τ ) = 4 R.-J. Lu, & Y.-P. Qin & T.-F. Yi In above formula, τθ is a dimensionless relative local time defined by τθ ≡ c(tθ − tc)/Rc, where tθ is the emission time in the observer frame, called local time, of photons emitted from the concerned differential surface dsθ of the fireball (θ is the angle to the line of sight), tc is a constant which could be assigned to any values of tθ, and Rc is the radius of the fireball measured at tθ = tc; Variable τ is a dimensionless relative time defined by τ ≡ [c(t − tc) − D + Rc]/Rc, where D is the distance of the fireball to the observer, and t is the observation time; eI(τθ) represents the development of the intensity magnitude in the observer frame, called as a local pulse function; g0,ν(ν0,θ) describes the rest frame radiation mechanisms; And k ≡ β/(1 − β). At the same time, the integral limits eτθ,min and eτθ,max are determined by eτθ,min = max{τθ,min, (τ − 1 + cos θmax)/(1 − β cos θmax)} and eτθ,max = min{τθ,max, (τ − 1 + cos θmin)/(1 − β cos θmin)}, where τθ,min and τθ,max are the lower and upper limits of τθ confining eI(τθ), and θmin and θmax are determined by the concerned area of the fireball surface, and then the radiation is observable within the range of (1−cos θmin)+(1−β cos θmin)τθ,min ≤ τ ≤ (1−cos θmax)+(1−β cos θmax)τθ,max. Formula (1) suggests that, light curves of sources depend mainly on Γ, eI(τθ) and g0,ν(ν0,θ). Observation suggests that the common radiation form of GRBs is the so-called Band spectrum function (Band et al. 1993) which was frequently, and rather successfully, employed to fit the spectra of the sources (see, e.g., Schaefer et al. 1994; Ford et al. 1995; Preece et al. 1998, 2000), therefore we take in this paper the Band function as the rest frame radiation form. And the rise width and the full width of light curves depend on the two factors of sources, Γ and eI(τθ). In this paper, we use τr and τFWHM refer to the rise width and the full width of light curves corresponding to variable τ , and tr and tFWHM to those corresponding to variable t, respectively. In the same way, we use ∆τθ,FWHM refer to the FWHM of a local pulse corresponding to variable τθ, and ∆tθ,FWHM to that corresponding to variable tθ. For the sake of simplicity, we first employ a local pulse with a power law rise and a power law decay to study this issue, which is written as eI(τθ) = I0{ ( τθ−τθ,min )µ τθ,0−τθ,min (1 − τθ−τθ,0 τθ,max−τθ,0 )µ (τθ,min ≤ τθ ≤ τθ,0) (τθ,0 < τθ ≤ τθ,max) . (2) Where τθ,0 and µ are constants. The FWHM of this local pulse would be ∆τθ,FWHM = (1 − 2(−1/µ))(τθ,max − τθ,min). The relationships between τr, τFWHM and Γ, and that between τr, τFWHM and ∆τθ,FWHM for the light curves determined by equation (1) are plotted in Fig. 1, and presented in Fig. 2 is the relationship between the τr and the τFWHM of the light curves. Figure 1 shows that the τr and τFWHM decrease with the Lorentz factor following τr ∝ Γ−2 and τFWHM ∝ Γ−2, and they increase with ∆τθ,FWHM following τr ∝ ∆τθ,FWHM for every value of ∆τθ,FWHM, and τFWHM ∝ ∆τθ,FWHM when ∆τθ,FWHM ≥ 1. Thus we Relationship between the rise width and the full width and its implications 5 obtain τr = k1Γ−2∆τθ,FWHM = k1p, τFWHM = kΓ−2∆τθ,FWHM = kp (∆τθ,FWHM ≥ 1), (3) (4) where p = Γ−2∆τθ,FWHM. k1 = 0.597 ± 0.006 and k = 1.335 ± 0.036 for this local pulse. Study reveals that each of the two quantities, τr and τFWHM, is proportional to p, but independent of Γ or τθ,FWHM. Considering the relation between τ and t, we get from (3) and (4) that tr = k1p Rc c , tFWHM = kp Rc c (∆τθ,FWHM ≥ 1). (5) (6) We find from the left panel in Fig. 2 that the τr increases linearly with the τFWHM when we take ∆τθ,FWHM = constant and Γ = variable. We thus perform a linear least square fit to the two quantities, and have log(τr) = A + Blog(τFWHM) for a certain value of ∆τθ,FWHM. There are almost the same slopes of "B" (i.e. B = 1.0) for different values of ∆τθ,FWHM, whereas the value of "A" changes from -1.981 to -0.526 when ∆τθ,FWHM changes correspondingly from 0.001 to 0.1, and there is a upper limit value of A≃-0.40 when ∆τθ,FWHM ≥ 1, thus a dead line can be found when ∆τθ,FWHM ≥ 1, i.e., log(τr) = -0.404 + 1.00×log(τFWHM). (for all situations of fitting, their correlation coefficient R >0.999 and the number of the data N=27.) According to equations (6) and (7) in Paper II, one could find that the photons that observer receives at different observation time τ emit from the different surface of fireball when ∆τθ,FWHM < 1, so that the profiles of the light curves determined by formulas (1) are affected by ∆τθ,FWHM. Whereas when ∆τθ,FWHM ≥ 1, the photons reaching the observer at different observation time τ come from the same whole surface of the fireball, in this situation, the profiles of the light curves don't change with ∆τθ,FWHM. This analysis is supported by the fact that the τr is sensitive to the ∆τθ,FWHM, but the τFWHM isn't significantly affected by the ∆τθ,FWHM, when ∆τθ,FWHM < 1, in fact in this case the τFWHM slightly decreases with the ∆τθ,FWHM, and when ∆τθ,FWHM → 0, the local pulse becomes a δ one, and the τFWHM would be determined by equation (44) in Paper II. However when ∆τθ,FWHM ≥ 1, each of the two quantities, τr and τFWHM, linearly increases with the ∆τθ,FWHM with the same slope (see the right panel in Fig. 1). Which naturally explains why one could find a dead line in the τr - τFWHM panel when ∆τθ,FWHM ≥ 1. When changing frequency interval from 100 ≤ ν/ν0,p≤ 300 to 25 ≤ ν/ν0,p≤ 50, or to other frequency interval, and repeating the same work as above, we find that the results don't change significantly, and the dead line is not sensitive to frequency interval. We study other forms of local pulses, such as µ = 1, 3 of equation (2), an exponential rise and exponential decay pulse, a Gaussian pulse, and a rectangle pulse, and so on, 6 R.-J. Lu, & Y.-P. Qin & T.-F. Yi and find that the equation (3) - (6) hold for all local pulses we investigate, and there are different values of k1 and k for different local pulse forms (see Table 1). For every local pulse form, a dead line can be found when ∆τθ,F W HM ≥ 1. The dead lines of the six local pulses, log(τr) = (-0.563 ± 0.005) +(1.015 ± 0.002)log(τFWHM) for local exponential pulse, log(τr) = (-0.520 ± 0.009) +(1.027 ± 0.003)log(τFWHM) for local Gaussian pulse, log(τr) = (-0.450 ± 0.003) +(1.005 ± 0.001)log(τFWHM) for µ = 3 of equation (2), log(τr) = (-0.404 ± 0.006) +(1.007 ± 0.001)log(τFWHM) for µ = 2 of equation (2), log(τr) = (-0.318 ± 0.005) +(1.011 ± 0.001)log(τFWHM) for µ = 1 of equation (2), and log(τr) = (-0.221 ± 0.011) +(1.027 ± 0.003)log(τFWHM) for local rectangle pulse, are presented in the right panel in Fig. 2. Studies show that, for all kinds of local pulse forms, the slopes of their dead lines are always equal to 1.0, but there are different intercepts for different local pulse forms in the τr - τFWHM panel. The intercept could therefore become an indicator of the local pulse form. We also note that the dead line of local rectangle form is the upper limit one for all local pulse forms (i.e., the intercept of a dead line for any local pulse forms would never exceed -0.20), which might be a criterion to check if the temporal behaviors of gamma-ray burst pulses do result from the contributions from the Doppler effect. 3 RELATIONSHIP BETWEEN THE RISE TIME AND THE WIDTH OF OBSERVATIONAL PULSES Kocevski et al. (2003) found that there is a linear correlation between the rise width vs. the full width of gamma-ray burst pulses provided by the BATSE instrument on board the CGRO (Compton Gamma Ray Observatory) spacecraft. For the sake of convenience of comparision with those obtained theoretically above, unlike Kocevski et al., we here investigate the temporal structures of the light curves of 2nd and 3rd channels based on their sample, respectively, which we call sample 1. As they pointed out that a power-law rise model can better describe the majority of the FRED pulses, so we measure the tr and the tFWHM of the pulses by fitted with the equation of (22) in their paper. The results are presented in Fig. 3. Fig. 3 shows that the tr increase linearly with the tFWHM of the observed pulses. We perform a linear least square fit at 1σ confidence level to the two quantities, and have log(tr)=(-0.531 ± 0.022) + (1.045 ± 0.029)log(tFWHM) with a linear correlation coefficient of 0.973 and a chance probability of p < 10−4 for the 2nd channel, and log(tr)=(-0.492 ± 0.029) + (1.031 ± 0.024)log(tFWHM) with a linear correlation coefficient of 0.980 and a chance probability of p < 10−4 for the 3rd channel. The results show that the two sequence in the tr − tFWHM panel have almost the same intercepts and slopes within their error. Intriguingly, one find that the slope is equal to the one obtained theoretically Relationship between the rise width and the full width and its implications 7 above, which indicates that the observed results are well consistent with those predicted by Qin model. We thus may come to the conclusions: (1) Merely the curvature effect can produce the relationship in the tr − tFWHM panel which is independent of any frequency interval. (2) The gamma-ray burst pulses most probably arise from local Gaussian form because the sequences in the tr − tFWHM panel is closed to the dead line of local Gaussian form. Note that the differences between the two panels, τr - τFWHM and tr − tFWHM, don't affect the comparison (See section (4)). To further demonstrate these conclusion, we choose another sample, call sample 2, based on gamma-ray burst pulses detected by HETE-2 instrument. Like Kocevski, we select pulses from the HETE-2 burst home page (http://space.mit.edu/HETE/Bursts/) with the simple criteria that pulses behave clean, well distinguished FRED-like form, thus 12 pulses could be available in our sample 2. We measure the tr and tFWHM of these pulses with the same methods as above. And the results are presented in Fig. 4. We perform a linear least square fit to the two quantities with the same methods adopted in Fig. 3, and obtain log(tr)=(-0.494 ± 0.065) + (1.012 ± 0.087)log(tFWHM) with a linear correlation coefficient of 0.960 and a chance probability of p < 10−4 for the band B, and log(tr)=(-0.504 ± 0.059) + (1.088 ± 0.078)log(tFWHM) with a linear correlation coefficient of 0.976 and a chance probability of p < 10−4 for the band C. The results are well consistent with those obtained from Fig. 3, which further testifies that the sequences in the tr − tFWHM plane is independent of any frequency interval. The first two smaller standard deviations of the four intercepts in the two sample from the ones of the six dead lines obtained above theoretically are 0.043 for local Gaussian pulse and 0.120 for local exponential pulse, which indicates that these four sequences are very closed to the dead line of the local Gaussian pulse and implicate that these observed pulses may arise from local Gaussian pulses. The conclusion is only a preliminary one which needs to be confirmed by larger samples in the future. As shown in Fig. 3 and 4, all pulses, which are selected from long bursts, are longer than 0.5 s. Norris et al. 2001 pointed out that short bursts with T90 < 2.6s have different temporal behaviors compared with long bursts. Whether or not short pulses and long ones follow the same sequence in the tr − tFWHM plane and behave the same temporal behavior? motivated by this, we select 6 short pulses, call sample 3, from the 64 ms count data of 532 short GRBs, and 23 short pulses (shorter than 1 s), call sample 4, from long GRBs with the same criteria adopted in sample 2. These short and long GRBs are detected by the BATSE instrument. We repeat the same work as above, and the results are plotted in the left panel in Fig. 5. We find from the left panel in Fig. 5 that, all pulses (selected from short and long GRBs) follow the same sequence in the tr − tFWHM plane, with the shorter pulses at the end of this sequence, which show that short pulses (or bursts) behave the same temporal 8 R.-J. Lu, & Y.-P. Qin & T.-F. Yi behaviors as long pulses, the only difference is that the quantity of p (see equation of (3) or (4)) of short pulse is smaller than the one of long pulse. 4 DISCUSSION AND CONCLUSIONS All analyses in this paper are based on formula (1), which is based on the case of a fireball expanding isotropically with a constant Lorentz factor, Γ > 1. In fact, the symmetry of expansion matters only over angles the order of a few times Γ−1, and beaming prevents us from observing other regions of the shell. The formula (1) is suitable for describing light curves of spherical fireballs or uniform jets. When considering a uniform jet and taking θmax = Γ−1 in the formula (1), we measure the τr and τFWHM of the resulting light curves. They show no difference from those of the spherical geometry, which indicate that the relationships in the τr - τFWHM plane are independent of gamma-ray burst pulses coming from either spherical fireballs or uniform jets. It is an ubiquitous trend that indexes of spectra of many GRBs are observed to vary with time (see Preece et al. 2000). We wonder how the relationship would be if the rest frame spectrum develops with time. As Preece et al. (2000) pointed out that the typical fitted value distributions for the low energy spectral index (α) is -1.5 ∼ - 0.3, the high energy spectral index β is -2 ∼ -3. Therefore here some sets of typical values of the indexes such as (-1.5, -2), (-0.3, -3) and (-0.8, -2.5) would be employed to investigate the relationship. Calculations show that the two quantities, τr and τFWHM, are not significantly affected by the rest frame radiation form. We find that, owing to the Doppler effect of the fireball surface (or the curvature effect), for any local pulse form, the width of the light curve would always be tFWHM = kΓ−2∆τθ,FWHM (There are different values of k for different local pulse forms, see Rc c Table 1). As derived in detail in Paper II, the width of the light curve of the local δ 2 Γ−2∆τ (see (48) in Paper II), where ∆τ is the interval of the observable time of the local δ function pulse, i.e., ∆τ = 1 + βτθ,0. function pulse would be τFWHM ≃ √2−1 When considering the relationship between τ and t, and taking τθ,0 = 0, we obtain tFWHM ≃ √2−1 102 )−2, which would be the lower limit of the width of light curves for any local pulse form. Because of the relativistic beaming of the moving c ≃ 2s( Rc 2 Γ−2 Rc 1015cm )( Γ radiating particles, only the emission from a narrow cone with an opening angle of Γ−1 is observed, Ryde & Petrosian (2002) obtained that the curvature timescale resulting from relativistic effects is τang = 1.7s( Rc 102 )−2 (See (5) in their paper). Even as they pointed out that this is a lower bound for the observed duration of a pulse. Thus it can 1015cm )( Γ be seen that the curvature timescale and the lower limit of the width of light curves are comparable. According to the equations (3) and (6), we know that the only difference between the two panels, tr−tFWHM and τr−τFWHM, is that the different observed pulses resulting from Relationship between the rise width and the full width and its implications 9 the same Γ and ∆τθ,FWHM but from different values of Rc would be only corresponding to one point in the τr − τFWHM plane; but in the tr − tFWHM plane these observed pulses would be corresponding to different points as they come from different values of Rc, and these different points must be on a line with the slope, B=1.0. However the intercepts of the light curves in the two panels dependent only on both forms and widths of their corresponding local pulses. It is widely accepted that gamma-ray bursts arise from the internal shocks at a distance of Rc ∼ 1013 − 1017cm in the scenario of standard fireball model (see Rees & M ´esz´aros 1992, 1994; M ´esz´aros & Rees 1993, 1994; M ´esz´aros 1995; Katz 1994; Paczynski & Xu 1994; Sari & Piran 1997; Piran 1999; Spada et al. 2000; Ryde & Petrosian 2002; and Piran 2005). To compare theoretic conclusions with the observed results in the tr − tFWHM plane, we merge the left panel in Fig. 5 into the right panel in Fig. 2 by applying equation (5) and (6) and taking Rc = 3 × 1015cm, and the results could be plotted in the right panel in Fig. 5. The results are in good agreement with those predicted by Qin model. We notice that the all observed pulses are under the dead line of local Gaussian form within their errors. As shown above, local pulses' widths, ∆τθ,FWHM ≥ 0.1, for most of observed pulses. Applying the relationship between τθ and tθ, we obtain that ∆tθ,FWHM≥ 0.1 Rc c . Because taking τθ,min = 0 (i.e., tθ,min = tc) in above analysis, we know that Rc at that time is the radius of fireball at which radiation begins to emit, which may correspond to the stage the fireball become optically thin and photons can escape freely. And the fact that ∆τθ,FWHM ≥ 0.1 indicates that local pulse's width is not less than 1 order of the time scale the fireball expands to become optical thin for most of observed pulses, and its implication is not clear now because we don't know what result in the local pulses yet. In all, we come to the following conclusions: (1)Merely the curvature effect could produce the observed relationship between tr and tFWHM. (2)Both long pulses and short ones follow the same sequence in the tr − tFWHM plane. If all observed pulses come from the same radius of fireballs (especially for the pulses in a burst), the shorter pulses have smaller value of p (Here p = Γ−2∆τθ,FWHM), which might implicate that short pulses come from larger Γ and narrower local pulses than long pulses, and short bursts come from largest Γ and narrowest local pulses than long bursts, this is a reasonable result if short GRBs are likely to be produced by the merger of compact objects while long GRBs result from the core collapse of massive stars. (3)All GRBs may arise intrinsically from local Gaussian pulses, and these local pulses' widths, ∆τθ,F W HM , would not be less than 0.1, which is in agreement with those found in Qin et al. (2005b). (4)The observed pulses deviated down from the the dead line of local Gaussian pulse would arise from the narrower local pulses (i.e., smaller than 1), and the shorter the local pulses, the more obvious the deviation feature. However we suspect that there will not be such a observed 10 R.-J. Lu, & Y.-P. Qin & T.-F. Yi pulse far deviated up from the dead line of the local Gaussian pulse in the tr − tFWHM plane if it arises from a local Gaussian pulse in terms of the fireball model. These features above may provide constraints on intrinsic emission mechanism re- sponsible for GRBs. Acknowledgements This work was supported by the Special Funds for Major State Basic Research Projects ("973") and National Natural Science Foundation of China (No. 10273019 and 10463001). References Band D., Matteson J., Ford L. et al., 1993, ApJ, 413, 281 Borgonovo L., Ryde F., 2001, ApJ, 548, 770 Cui X. H., Liang E. W., Lu R. J, 2005, Chin. J. Astron. Astrophys., 5, 151 Fenimore E. E., Madras C. D., Nayakshin, S., 1996, ApJ, 473, 998 Fishman G. J., Meegan C. A., Wilson R. B. et al., 1994, ApJS, 92, 229 Ford L. A., Band D. L., Mattesou J. L. et al., 1995, ApJ, 439, 307 Goodman J. 1986, ApJ, 308, L47 Ghirlanda G., Ghisellini G., Celotti A., 2004, A&A 422L, 55G Ghisellini G., Celotti A., Lazzati D., 2000, MNRAS, 313, L1 Jia L. W., Qin Y. P., ApJL, in press, astro-ph/0505309 Katz J. I., 1994, ApJ, 422, 248 Kocevski D., Ryde F., Liang E., 2003, ApJ, 596, 389 (Paper I) Lamb D. Q., Ricker G. R., Atteia J. L. et al. 2002, ApJ, submitted, astro-ph/0206151 Lee A., Bloom E. D., Petrosian V., 2000a, ApJS, 131, 1 Lee A., Bloom E. D., Petrosian V., 2000b, ApJS, 131, 21 Lu R. J., Qin Y. P., Zhang Z. B. et al., 2005a, MNRAS, Submitted, astro-ph/0509287 Lu R. J., Qin Y. P., 2005b, MNRAS, Submitted, astro-ph/0508537 M ´esz´aros P., Rees M. J., 1993, ApJ, 405, 278 M ´esz´aros P., Rees M. J., 1994, MNRAS, 269, L41 M ´esz´aros P., 1995, in Procs. 17th Texas Symp. Rel. Astroph. (NY Acad. Sci., New York), 759, 440 Norris J. P., Nemiroff R. J., Bonnell J. T. et al., 1996, ApJ, 459, 393 Paczynski B., 1986, ApJ, 308, L43 Paczynski B., Xu G., 1994, ApJ, 427, 708 Piran T., 1999, Phys. Rep., 314, 575 Piran T., 2005, Reviews of Modern Physics, 76, 1143 Preece R. D., Pendleton G. N., Briggs M. S. et al., 1998, ApJ, 496, 849 Preece R. D., Briggs M. S., Mallozzi R. S. et al., 2000, ApJS, 126, 19 Qin Y. P., 2002, A&A, 396, 705 Qin Y. P., 2003, A&A, 407, 393 Qin Y. P., Zhang Z. B., Zhang F. W. et al., 2004, ApJ, 617, 439 (Paper II) Qin Y. P., Dong Y. M., Lu R. J. et al., 2005a, ApJ, in press, astro-ph/0411365 Qin Y. P., Lu R. J., 2005b, MNRAS, 362, 1085Q Ramirez-Ruiz E., Fenimore E. E., 2000, ApJ, 539, 712 Rees M. J., M ´esz´aros P., 1992, MNRAS, 258, 41 Rees M. J., M ´esz´aros P., 1994, ApJ, 430, L93 Ryde F., Svensson R., 2000, ApJ, 529, L13 Ryde F., Petrosian V., 2002, ApJ, 578, 290 Ryde F., Borgonovo L., Larsson S. et al., 2003, A&A, 411, L331 Schaefer B. E., Teegaeden B. J., Fantasia S. F. et al., 1994, ApJS, 92, 285 Relationship between the rise width and the full width and its implications 11 Table 1 The Coefficients For Equation (3) And (4) Local pulse forms k1 k µ = 1 of equation(2) 0.412 ± 0.003 0.827 ± 0.022 µ = 2 of equation(2) 0.597 ± 0.006 1.335 ± 0.036 µ = 3 of equation(2) 0.717 ± 0.009 1.718 ± 0.041 An exponential rise and decay pulse 0.306 ± 0.001 2.006 ± 0.122 A Gaussian pulse 0.650 ± 0.007 2.782 ± 0.149 A rectangle pulse 0.149 ± 0.001 0.318 ± 0.013 ) M H W F ( g o l , ) r ( g o l 0 -1 -2 -3 -4 -5 -6 -7 0 -1 -2 -3 -4 -5 -6 -7 1 2 log( ) 3 -3 -2 -1 log( 0 1 FWHM) 2 3 Fig. 1 Relationships between τr, τFWHM and Γ (Left panel) and that between τr, τFWHM and ∆τθ,FWHM (Right panel) for the light curves determined by equation (1), where a band function rest frame radiation form with α0 = −1 and we take 2πR3 and β0 = −2.25, within the frequency range of 100 ≤ ν/ν0,p≤ 300, is adopted, c I0/hcD2 = 1, µ = 2, τθ,min = 0, θmin = 0, θmax = π/2, ∆τθ,FWHM=1 (Left panel), and Γ=100 (Right panel). The solid and the dash line represent the τFWHM and the τr of light curves in both panels, respectively. Schmidt M., 2001, APJ, 559, L79 Sari R., Piran T., 1997, MNRAS, 287, 110 Spada M., Panaitescu A., M ´esz´aros P., 2000, ApJ, 537, 824 Zhang B., M ´esz´aros P., 2004, IJMPA, 19, 2385Z This manuscript was prepared with the ChJAA LATEX macro v1.0. 12 R.-J. Lu, & Y.-P. Qin & T.-F. Yi ) r ( g o l 3 2 1 0 -1 -2 -3 -4 -5 -6 -7 -8 -9 0 -1 -2 -3 -4 -5 -6 -7 -7 -6 -5 -4 -3 -2 -1 0 1 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 log( FWHM ) Fig. 2 Relationships between τr and τFWHM for the light curves determined by equation (1). Left panel: We take Γ=2 to 1000 for different values of ∆τθ,FWHM. The solid lines from the bottom to the top represent ∆τθ,FWHM=0.001, 0.01, 0.1, 1, 10, 100, and 1000, respectively. (note: when ∆τθ,FWHM ≥ 1, their correspond- ing lines overlap each other.) Right panel: the dead lines of the six local pulse forms: the solid lines from the bottom to the top represent local exponential rise and exponential decay pulse, local Gaussian pulse, local power law pulse µ=3, 2, 1 of equation (2) and local rectangle pulse, respectively. Other parameters are the same as those adopted in Fig. 1. Relationship between the rise width and the full width and its implications 13 1.5 1.0 0.5 0.0 r ) t ( g o l -0.5 -1.0 -0.5 0.0 0.5 1.0 1.5 1.0 0.5 0.0 -0.5 -1.0 1.5 2.0 -0.5 log( tFWHM ) 0.0 0.5 1.0 1.5 2.0 Fig. 3 Relationships between tr and tFWHM for the observed pulses based on sample 1. The left and the right panel present the pulses of the 2nd channel and the 3rd channel, respectively. The two solid lines are the fit lines of their data. 1.5 1.0 0.5 0.0 r ) t ( g o l -0.5 -1.0 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 -1.0 -0.5 0.0 -1.5 2.0 log( tFWHM ) 0.5 1.0 1.5 2.0 Fig. 4 Relationships between tr and tFWHM for the observed pulses based on sample 2. The left panel and the right panel present the pulses of the band B (7-40 keV) and C (30-400 keV), respectively. The two solid lines are the fit lines of their data. 14 R.-J. Lu, & Y.-P. Qin & T.-F. Yi ) r t ( g o l 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 -1.0 -0.5 0.0 log( tFWHM ) 0.5 1.0 1.5 2.0 Fig. 5 Relationships between tr and tFWHM for the observed pulses. Left panel presents the observed pulses of the 3rd channel based on sample 1, 3 and 4. The open circle, the open rectangle and the cross present the pulses of sample 1, 3 and 4, respectively. Right panel is a combination of the left panel and the right panel in Fig. 2, where we take Rc = 3 × 1015cm.
0709.1895
1
0709
2007-09-12T15:15:06
Broadband X-ray spectrum of the newly discovered broad line radio galaxy IGR J21247+5058
[ "astro-ph" ]
In this paper we present radio and high energy observations of the INTEGRAL source IGR J21247+5058, a broad line emitting galaxy obscured by the Galactic plane. Archival VLA radio data indicate that IGR J21247+5058 can be classified as an FRII Broad Line Radio Galaxy. The spectrum between 610 MHz and 15 GHz is typical of synchrotron self-absorbed radiation with a peak at 8 GHz and a low energy turnover; the core fraction is 0.1 suggestive of a moderate Doppler boosting of the base of the jet. The high energy broad-band spectrum was obtained by combining XMM-Newton and Swift/XRT observation with INTEGRAL/IBIS data. The 0.4-100 keV spectrum is well described by a power law, with slope $\Gamma$=1.5, characterised by complex absorption due to two layers of material partially covering the source and a high energy cut-off around 70-80 keV. Features such as a narrow iron line and a Compton reflection component, if present, are weak, suggesting that reprocessing of the power law photons in the accretion disk plays a negligible role in the source.
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 1 -- 6 (2007) Printed 11 November 2018 (MN LATEX style file v2.2) Broad band X-ray spectrum of the newly discovered Broad Line Radio Galaxy IGR J21247+5058 M. Molina,1 M. Giroletti,2 A. Malizia,3 R. Landi,3 L. Bassani3, A.J. Bird,1 A.J. Dean,1 A. De Rosa,4 M. Fiocchi,4 F. Panessa5 1School of Physics and Astronomy, University of Southampton, SO17 1BJ, Southampton, U.K., 2IRA/INAF, via Gobetti 101, I-40129 Bologna, Italy, 3IASF/INAF, via Gobetti 101, I-40129 Bologna, Italy 4IASF/INAF, via Fosso del Cavaliere 100, I-00133 Rome, Italy 5Instituto de Fsica de Cantabria (CSIC-UC), Avda. de los Castros, 39005 Santander, Spain ABSTRACT In this paper we present radio and high energy observations of the INTEGRAL source IGR J21247+5058, a broad line emitting galaxy obscured by the Galactic plane. Archival VLA radio data indicate that IGR J21247+5058 can be classified as an FRII Broad Line Radio Galaxy. The spectrum between 610 MHz and 15 GHz is typical of synchrotron self-absorbed radiation with a peak at 8 GHz and a low energy turnover; the core fraction is 0.1 suggestive of a moderate Doppler boosting of the base of the jet. The high energy broad-band spectrum was obtained by combining XMM-Newton and Swift/XRT observation with INTEGRAL/IBIS data. The 0.4-100 keV spectrum is well described by a power law, with slope Γ=1.5, characterised by complex absorption due to two layers of material partially covering the source and a high energy cut-off around 70-80 keV. Features such as a narrow iron line and a Compton reflection component, if present, are weak, suggesting that reprocessing of the power law photons in the accretion disk plays a negligible role in the source. Key words: Galaxies -- AGN -- Radio -- X-rays. 1 INTRODUCTION IGR J21247+5058 was initially reported in the first INTEGRAL survey catalogue (Bird et al. 2004) and listed in subsequent sur- vey papers: in the most recent work the source is located at RA (J2000)=321.172 and Dec (J2000)=+50.972 with an associated 90% error circle of 1′ (Bird et al. 2007). Soon after its discovery, it was associated by Rib´o et al. (2004) with the radio source 4C 50.55, also known as GPSR 93.319+0.394, KR2, NRAO 659 or BG 2122+50: this object has the typical morphology of a radio galaxy showing a bright core and two lobes. The estimated position of the core from the NVSS map is RA (J2000)=21h24m39.25s and Dec (J2000)=+50◦58′23.80′′ (1′′ uncertainty). Confirmation of the AGN nature of IGR J21247+5058 came via optical observations obtained at the Loiano telescope (Masetti et al. 2004), despite the fact that the optical spectrum of the source looks very peculiar. It has in fact a broad, redshifted Hα complex superimposed onto a "normal" F/G-type Galactic star continuum. While most of the observed features (Na, Ca and Mg) are consistent with redshift z=0 and thus with a Galactic stellar origin, the Hα complex leads to a redshift z=0.02. This feature is very similar to the one observed in another bright radio galaxy, namely 3C390.3 (Dietrich et al. 1998). The Hα complex, together with the spatially coincident extended radio emission and the de- tection of strong hard X-ray radiation, strongly indicates the unfor- tunate situation of a chance alignment between a relatively nearby star and a background radio galaxy. Indeed the INTEGRAL/IBIS spectrum of this source is compatible with the canonical AGN spec- trum (Molina et al. 2006). Because the only optical line observed is broad, the source was tentatively classified as a Seyfert 1 or alter- natively as a broad line radio galaxy (BLRG). Unfortunately, given the confusion with the nearby star, it is impossible to gain more in- formation on the source optical characteristics with the currently available data (see Masetti et al. 2004); in particular no reliable measurement of the B magnitude can be used to estimate the source radio loudness using the relation RL=log[F(5GHZ)/F(B)]. At the observed redshift, the source luminosity in the 20-100 keV band is 8.5×1043 erg s−1, making IGR J21247+5058 one of the brightest AGN in the local Universe1. Here, we present a detailed radio analysis of the source based on archival VLA data. We also discuss archival XMM and Swift/XRT data in combination with a new INTEGRAL/IBIS spec- 1 Assuming H0=70 km s−1Mpc−1 and a flat Universe. 2 M. Molina et al. trum which capitalises on the larger exposure now available on this source. 2 RADIO OBSERVATIONS At the very low energy part of the electromagnetic spectrum, ra- dio observations provide valuable information on the nature of IGR J21247+5058. Radio images at various resolutions have been pre- sented in several works (e.g. Mantovani et al. 1982, Pandey et al. 2006). Here we show in Figure 1 a 1.4 GHz image of the field of IGR J21247+5058 obtained with data from the VLA archive2. The circle indicates the location and positional error of the INTEGRAL detection, which clearly points to the nucleus of IGR J21247+5058 as the source of the gamma-ray emission. The radio source has the typical edge-brightened morphology of an FRII radio galaxy, with a central compact core and two large lobes. The size of IGR J21247+5058 is ∼ 9.5′ and the total flux density at 1.4 GHz is 2.5 Jy. At the redshift proposed for this source, these data correspond to a total extent LS=230 kpc and a monochromatic radio power P1.4 = 1024.4W Hz−1. This makes IGR J21247+5058 a typical radio galaxy in size, with a radio power intermediate between FRI and FRIIs. In Figure 2 we show the spectrum of the core of IGR J21247+5058 between 610 MHz and 15 GHz. The data between 1.4 and 15 GHz were obtained from data in the VLA archive, while the 610 MHz point is taken from the GMRT data (Pandey et al. 2006). The spectrum is typical of synchrotron self-absorbed radia- tion, with a peak at about 8 GHz and a low frequency turnover. The core fraction at 1.4 GHz is about S c/S t = 0.1, which is suggestive of a moderate Doppler boosting of the base of the jet. In fact, the core is brighter than what would be expected on the basis of the correlation between core and total radio power (Giovannini et al. 1988). From the total flux density at low fre- quency (S 0.4 = 5.4 Jy, Mantovani et al. 1982), the core flux density should be only ∼ 50 mJy at 5 GHz, i.e. about a factor 10 less than observed. If we assume a typical Lorentz factor for the radio jet (γ=5, see e.g. Giovannini et al. 2001), we can use this constraint to estimate a viewing angle θ ∼ 35◦. This seems to be small enough to allow us to peer into the BLR and it also explains the broadening of the Hα line. Put altogether, the radio data seem to indicate that the counterpart of IGR J21247+5058 is an FRII broad line radio galaxy. It is however difficult to guess which is the approaching side of the source. The NW lobe is brighter, but a knot of enhanced bright- ness is visible in the SE jet at 90′′ from the core. If this brightness asymmetry is due to Doppler boosting, then the approaching side would be the SE one. A look at the parsec scale structure would be desirable to better study the properties of the inner jet and define this issue. It is also interesting to note that a weak feature is clearly de- tected in several VLA data sets at ∼ 2′ south of the core (RA = 21h24m39.97s, Dec = +50◦56′05.4′′). It has a flux of ∼ 4 mJy at 1.4 GHz and ∼ 2 mJy at 4.8 GHz. This source is probably unrelated to the INTEGRAL source, as it falls outside its 90% error circle and it is not detected in X-rays (see next section). 2 The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreeement by Associated Universities, Inc. ) 0 0 0 2 J ( I N O T A N L C E D I 51 02 01 00 50 59 58 57 56 55 54 21 25 00 24 45 RIGHT ASCENSION (J2000) 30 15 Figure 1. A 1.4 GHz image of the field of IGR J21247+5058 from VLA data (in B configuration). Contours are traced at (−1, 1, 2, 4, ...)×0.5 mJy beam−1 and the peak is 244 mJy beam−1. The restoring beam is 5.5×4.4 arcsec. The circle shows the ISGRI position and error box of IGR J21247+5058. Figure 2. Average spectrum of the core of IGR J21247+5058 with VLA data (between 1.4 and 15 GHz) and the GMRT (0.6 GHz). The error bars show the rms of the various measurements considered, as well as the abso- lute calibration uncertainty. 3 X-RAY OBSERVATIONS AND DATA ANALYSIS IGR J21247+5058 was observed by XMM-Newton on 2005 November 6 during orbit 1083, in the XMM-Newton Guest Ob- server Programme. Two other observations are present in the archive, but have not been used due to their poor statistical qual- ity. During this orbit, the EPIC PN (Struder et al. 2001) expo- sure was ∼25 ks, while the EPIC MOS1 and MOS2 (Turner et al. 2001) exposures were ∼27 ks. The EPIC PN camera was oper- ated in Large Window Mode with a thick filter applied, while the two MOS cameras were both operated in Small Window Mode. MOS and PN data were reprocessed using the XMM-Newton Stan- dard Analysis Software (SAS) version 7.0. Image analysis indi- cates that a bright source, localised at RA=21h24m39.36s and Dec (J2000)=+50◦58′23.86′′, compatible with the position of the ra- dio core, is detected with high significance, while no emission is Broad band X-ray spectrum of the newly discovered BLRG IGR J21247+5058 3 seen at the location of the radio lobes nor in correspondence with the radio source located south of the core of IGR J21247+5058. As for the Galactic star which is aligned with the radio galaxy by chance, it could emit X-rays by coronal activity as seen in many stars of similar late spectroscopic type. However, any contami- nation is likely to be negligible as the star is likely at 2.5 Kpc (Masetti et al. 2004), i.e. too far away to provide significant flux if the emission is from a stellar corona. MOS (PN) source counts were extracted from circular regions of 50′′ radius centered on the source, while background spectra were extracted from source free regions of 20′′ radius. Pile up is negligible for each of the instru- ments. In this analysis, only patterns corresponding to single, dou- ble, triple and quadruple events for the two MOS cameras were selected (PATTERN612), while for the PN only single and double events (PATTERN64) were taken into account; the standard selec- tion filter FLAG=0 was applied. Exposures have been filtered for periods of high background activity and the resulting values are listed in Table 1, together with other relevant information related to this XMM measurement. The ancillary response matrices (ARFs) and the detector re- sponse matrices (RMFs) were generated using the XMM-SAS tasks arfgen and rmfgen; spectral channels were rebinned in order to achieve a minimum of 20 counts per each bin. Here and in the following, spectral analysis was performed with XSPEC v.11.2.3 (Arnaud 1996) and errors are quoted at 90% confidence level for one parameter of interest (∆χ2=2.71). Since the source is lo- cated behind the Galactic plane, Galactic absorption is high, being 1.11×1022atoms cm−2 (Dickey & Lockman 1990); it has therefore been included in each fit so that the quoted absorbing column den- sities are always in excess to this Galactic value. Cross calibra- tion constants PN/MOS1 and MOS2/MOS1 were left free to vary and always found to be close to unity as expected. Data from the three EPIC cameras were initially fitted in the 0.4-10 keV range using a simple power law absorbed only by the Galactic column density (see Figure 3). This model does not yield an acceptable fit (χ2=4974.7 for 2624) as evident in Figure 3 and the resulting power law slope is very flat (Γ=0.94+0.01 −0.01). Since the data to model ratio are indicative of intrinsic absorption, the data were re-fitted adding this component (wag*wa*po in XSPEC); this new model provides a significant improvement in the fit (χ2=2526.2 for 2623 d.o.f.), a column density NH =0.62+0.02 −0.02×1022cm−2 but a still rather flat spectrum (Γ=1.28+0.01 −0.01). An even better fit (χ2=2446.1 for 2622 dof) is obtained using an ionized absorber instead of a cold one (wag*absori*po in XSPEC): the column density in this case is −0.09×1022cm−2 and the ionization state ξ (= L/nR2, Done et al. 1.12+0.10 1992) is 18.1+6.8 −5.9, i.e. the absorber is at most mildly ionized. The slope of the power law hardens but only marginally (Γ=1.33+0.02 −0.02). We then substituted the absorption (cold or ionized) with a partial covering component (*wag*pcfabs*po, see Table 2). This model provides a ∆χ2=103.8 for 1 d.o.f. compared to the cold absorption model and a better χ2 for the same d.o.f. compared to the ionized absorption model; the resulting spectrum has a similar photon index and column density as obtained for the ionized absorber model, but the absorption is now colder and covers around 80% of the source. At this stage we also checked the data for the presence of a cold iron line, adding to the partially covering model a narrow Gaussian component, having fixed the width to 10 eV. This model provides a fit improvement (χ2=2405.9 for 2620 d.o.f. or ∆χ2=16.5 for 2 d.o.f.), a line energy at 6.39+0.06 −0.08 keV and an equivalent width (EW) of 20.6+6.1 −9.6 eV (see Figure 4 for the confidence countours of the line energy versus line normalisation). The improvement is significa- tive at the 99.9% confidence level; however the EW is rather small, IGR J21247+5058 V e k / c e s / s t n u o c d e z i l a m r o n o i t a r 1 1 0 . 1 0 0 . 2 5 . 1 1 . 5 0 0.5 1 2 5 channel energy (keV) Figure 3. The XMM data fitted using a simple power law absorbed by Galactic column density. Confidence contours m r o n 5 − 0 1 × 2 5 − 0 1 × 5 1 . 5 − 0 1 6 − 0 1 × 5 6.2 6.25 6.3 6.35 6.4 6.45 6.5 E line (keV) Figure 4. The confidence contours of the line energy versus line normalisa- tion. i.e. too close to the capability limits of moderate resolution CCD instruments like XMM, calling for some caution in considering the line as a real feature or just local noise. Without going into more de- tails, the important point to stress here is that in IGR J21247+5058 a cold iron line is either very weak or not present. There is also a hint for a line at around 1.7 keV, but again its EW is very small and its energy suspiciously close to background features present in the PN and MOS camera to consider it as a real line. Despite the fact that a fit with just one absorber (cold, mildly ionized or partially covering the souce) is quite acceptable, the re- sulting spectrum is still rather flat. Hence we tried two different scenarios to steepen the XMM spectral data: reflection from neu- tral material (with the Compton reflection component described by the parameter R; pexrav model in XSPEC) or an extra layer of absorbing material (again cold, ionized or partially covering the source). In the first case the inclination angle was fixed at 30◦, im- plying a nearly face-on geometry. The reflection component result- ing from the fit was small R=0.59+0.45 −0.42 and the photon index still rather flat (Γ=1.43+0.06 −0.06); the low value of the reflection parame- ter, although not well constrained, is consistent with the weak iron line observed in the spectrum of IGR J21247+5058, suggesting that the reprocessing of the power law photons in the accretion disk plays a negligible role in the source. In the second case, the best fit is obtained with another layer of cold material partially cover- 4 M. Molina et al. Table 1 Observations Log Obs date Exposures (ksec) Filter Source counts (ct/s) XMM-MOS1 XMM-MOS2 XMM-PN XRT1 XRT2 INTEGRAL 06-11-2005 06-11-2005 06-11-2005 17-10-2006 25-01-2007 Nov. 2002 to Apr. 2006 24.2 23.6 22.0 7.2 3.7 768 Thick Thick Thick - - - 1.918±0.009 1.907±0.009 5.582±0.02 0.4953±0.01 0.3222±0.01 1.423±0.04 ing the source (wa*pcfabs*pcfabs*po, see Table 2). This model provides a power law slope of ∼1.6, close to the 1.7 value mea- sured in other BLRG (Sambruna et al. 1999; Grandi et al. 2006), although still flatter than the 1.9 value typically observed in ra- dio quite AGN (Perola et al. 2002). The absorbing column densi- ties are around 9×1022 cm−2 and 1022cm−2, covering 27% and 83% of the central source. Given the fact that the highest steepening of the spectrum is obtained with a double absorber model and that re- flection is either not required or small (also from the weakness of the iron line), we chose the first model as a better description of the XMM data. As some excess counts (mostly PN) are still visible at low energies, we have further added to this best fit model a soft thermal component (in the form of a black body or a mekal model) or a scattered component in the form of a power law model; in the latter case the two power laws have the same photon index but dif- ferent normalisations. In all cases, the excess emission remains and the fits return parameters which are unusual for AGN (a too high temperature and a scattered component with a too high normalisa- tion with respect to the primary continuum). Given the quality of the data at low energies and the small strength of this excess emis- sion, it is beyond the scope of the present paper to further enquire about this component and its analysis is postponed to when more detailed X-ray observations of IGR J21247+5058 are available. In order to check if the source could be classified as a radio- loud object, we exploited the X-ray flux measurement and used the RX =Lr(5GHz)/LX(2-10 keV) relation (e.g. Terashima & Wilson 2003 and Panessa et al. 2007); to do this we used the VLA 5 GHz measurement of the core component and then compared it to the 2-10 keV flux. Since LogRX =-3.3, IGR J21247+5058 could be de- fined a borderline object, as it is radio loud or quiet depending on our choice of the dividing line between these two classes: LogRX =- 4.5 (as in Terashima & Wilson 2003) or -2.8 (as in Panessa et al. 2007). However the LogRX value of IGR J21247+5058 is fully compatible with those of similar BLRG like 3C 111 (LogRX =- 3.7), 3C120 (LogRX =-2.1), 3C3 90.3 (LogRX =-3.1) and 3C 382 (LogRX =-4.0) (Grandi et al. 2006; Liu & Zhang 2002), implying that also this new source is most likely a radio loud AGN. IGR J21247+5058 was also observed with the XRT (X-ray Telescope, operating in the 0.2-10 keV range) on board the Swift satellite (Gehrels et al. 2004) for ∼7.2 ks on 2006 October 17 and for 3.7 ks on 2007 January 25. Data reduction was performed using the XRTDAS v1.8.0 standard data pipeline package (xrtpipeline v. 0.10.3) in order to produce screened event files. All data were col- lected in the Photon Counting (PC) mode (Hill et al. 2004), adopt- ing the standard grade filtering (0-12 for PC) according to the XRT nomenclature. Source data have been extracted using photons in a circular region of radius 20′′; background data have instead been taken from various uncontaminated regions near the X-ray source, using either a circular region of different radius or an annulus sur- rounding the source. The log of these two XRT observations is also reported in Table 1. Due to the lower quality of these data, we have employed a simple model, i.e. a power law passing through a single aborp- tion layer partially covering the central source. The first XRT ob- servation (hereafter XRT1) provides a power law spectrum with Γ=1.54+0.19 −0.18, moderate intrinsic absorption, a partial covering frac- tion f =0.74 and a 2-10 keV flux of 6.4×10−11 erg cm−2s−1 (see Ta- ble 1). To check if the best fit values could be improved, we added another partial covering absorber to the model, but the fit does not produce better results (χ2=132.2 for 138 d.o.f.). The second observation (hereafter XRT2) is of even poorer quality due to the lower exposure; it provides slightly different pa- rameters than the first observation (see Table 2) and a lower 2- 10 keV flux of (3.37±0.1)×10−11 erg cm−2 s−1. Despite the less precise modelling allowed by the XRT observations, the compari- son between the two XRT observations and with the XMM-Newton measurement suggests a change in the source absorption proper- ties related to a flux variation. Note that for consistency, also the XMM data are modelled here with a single layer of absorption. The change in the absorber properties is evident in Figure 5 where the contour plots of the column density versus covering fraction are displayed for the three available X-ray observations: in particular while XRT1 data are fully compatible with the XMM one, both dif- fer from the XRT2 observation parameters. Given that the model used is not the best fit found with the XMM data, as a second step we adopted the double absorber model and compared the 3 sets of data by means of an accurate parameter space expoloration. Due to the lower statistical quality of the XRT data, we have fixed to the XMM value the photon index and the two absorption parame- ters each at a time. The result of this procedure is that the likely change occuring is in the column density of the absorber covering 80% of the source while the other parameters are consistent within errors with each other. However, given the poor quality fo the XRT2 spectrum, i.e. the one that provides evidence for this change in ab- sorption, some caution is needed and further observations required to confirm this observational evidence. The INTEGRAL data reported here consist of several pointings of IGR J21247+5058 performed by the IBIS/ISGRI instrument be- tween revolution 12 and 429, i.e. the period from launch to the end of April 2006, and correspond to a total exposure of 768 ks. IS- GRI images for each available pointing were generated in various energy bands using the ISDC offline scientific analysis software OSA (Goldwurm et al. 2003) version 5.1. Count rates at the posi- tion of the source were extracted from individual images in order to provide light curves in various energy bands; from these light curves, average fluxes were then extracted and combined to pro- duce a source spectrum. Analysis was performed in the 17-100 keV band (the source count rate in this band is reported in Table 1). A Broad band X-ray spectrum of the newly discovered BLRG IGR J21247+5058 5 Confidence contours IGR J21247+5058 t c a r F r v C 9 9 9 . . . 0 0 0 8 8 8 . . . 0 0 0 7 7 7 . . . 0 0 0 6 6 6 . . . 0 0 0 5 5 5 . . . 0 0 0 0 0 0 2 2 2 4 4 4 nH 10^22 6 6 6 8 8 8 Figure 5. Constraints on the absorbing column densities and relative covering fractions for the XMM (solid line), XRT1 (dashed line) and XRT2 (dashed-dotted line) observations assuming a single absorption layer (wag*pcfabs*po). simple power law provides a good fit to the IBIS data (χ2=8.9 for 10 d.o.f.) and a photon index Γ=2.0±0.1 and a 17-100 keV flux of 1.15×10−10 erg cm−2 s−1. The INTEGRAL spectrum is clearly steeper than the XMM/XRT ones, implying that a cut-off is possi- bly present nd located in the IBIS energy band. Up to now IGR J21247+5058 has not been reported by Swift/BAT, and therefore the IBIS measurement is the only available above 10 keV. 3.1 The high energy broad-band spectrum X-ray and INTEGRAL data were then fitted together in order to obtain an average broad-band spectrum of the source. In the fit- ting procedure, a multiplicative constant, C, has been introduced to take into account possible cross-calibration mismatches between the X-ray and INTEGRAL data; this constant has been found to be close to 1 both for XMM and Swift/XRT using various source ty- pology (Landi et al. 2007; De Rosa et al. 2007; Masetti et al. 2007) so that significant deviation from this value can be confidently ascribed to source flux variability. Initially, we combined the XMM and the IBIS/ISGRI data together, employing the model (wa*pcfabs*pacfabs*po) used for the XMM data alone and con- sidered here as our baseline model (we have ignored the iron line at this stage but the results do not change significantly if this com- ponent is added to the fit). This model provides a good fit with a photon index of ∼1.7, and two absorption layers (∼ 1022 and ∼ 1023 cm−2 respectively, Table 3), covering 84% and 34% of the central source. To check for the presence of a high energy cut- off, we substituted the simple power law with a cut-off power law (wa*pcfabs*pcfabs*cutoffpl). The model yields our best fit (χ2=2354.0, 2628 d.o.f., Figure 6) and a power law slope around 1.5; a cut-off is indeed present and well constrained at around 75 keV (see Table 3 and Figure 7). In the above models the value of C is around 0.80, suggesting a good match and no major changes in flux between the XMM and INTEGRAL observations. Finally, we exploited the broader energy coverage check again for the pres- ence of reflection in the source spectrum by adding this compo- nent to the previous model (wa*pcfabs*pcfabs*pexrav, see Ta- ble 3). The fit shows an improvement (χ2=2350.9 for 2627 d.o.f., 99.9% confidence level) with respect to the simple double par- tial covering model (wa*pcfabs*pacfabs*po); however, when V e k / c e s / s t n u o c d e z i l a m r o n o i t a r 1 1 . 0 1 0 0 . 2 5 . 1 1 . 5 0 1 10 channel energy (keV) 100 Figure 6. XMM/ISGRI broad band (0.4-100 keV) spectrum of IGR J21247+5058: the model is a cut-off power-law absorbed both by Galactic column density and by two layers of absorbing material partially covering the source. Confidence contours 0 5 1 0 0 1 0 5 ) V e k ( c E 1.35 1.4 1.45 1.5 Photon Index 1.55 1.6 Figure 7. Constraints on the primary continuum (power-law photon index) versus the high energy cut-off of IGR J21247+5058. considering the best-fit model (wa*pcfabs*pcfabs*cutoffpl), there is only a marginal improvement in the fit, suggesting that the reflection component is not strongly required by the data. The model gives a power law slope of ∼1.5, the high energy cut-off is around 100 keV and the double absorption layer val- ues are well in agreement with those obtained with no reflection in the model (R=0). This reflection model yields a poorly con- strained value of R around 0.4 and a cross calibration constant ∼0.7. We have also tried to find evidence for a jet component in the high energy data introducing a second power law component (wa*pcfabs*pcfabs*(cutoffpl+po)) with the photon index of the primary continuum fixed at the canonical 1.7 value for BLRG. Again the fit doesn't show any improvement (χ2=2379.2, 2627 dof), the power law component used to model the jet emission has a slope of 1.50+0.16 −0.33 and the high energy cut-off is consistent with value found in the best fit (83.5+61.6 −29.2 keV). The values of the column densities and their covering fractions are in agreement with those found in the best fit. We can thus conlcude that the extra jet compo- nent is not required by the data and that the souce emission is not jet-dominated. Next best-fit model (wa*pcfabs*pcfabs*cutoffpl) was used to fit XRT and INTEGRAL data. In both XRT observations, 6 M. Molina et al. the fit is good (χ2=138.8 for 145 d.o.f. and χ2=47.8 for 54 d.o.f.), the power slopes are around 1.6 and 1.8 and the high energy cut- off values are not well constrained, although they could be placed above 50 and 80 keV respectively. The partial covering fractions for the XRT1-2/ISGRI broad-band spectrum are broadly compati- ble with those found for the XMM/ISGRI spectrum (see Table 3). It must be pointed out that due to the poor statistical quality of the XRT datasets, the values of the partial covering fractions are not well constrained and so errors are evaluated by freezing the param- eters related to each layer while calculating the uncertainties for the other. The constant between XRT1 and INTEGRAL spectra is around 0.8, suggesting again agreement between the two sets of data. The constant between the XRT2 observation and the INTE- GRAL data is instead higher (1.6), implying that the source could have undergone some minor changes in flux or have changed its absorption properties as discussed in the previous section. 4 DISCUSSION AND CONCLUSIONS The radio counterpart of IGR J21247+5058 shows characteristics typical of a Broad Line Radio Galaxy, probably of the FRII type. The spectrum between 610 MHz and 15 GHz, obtained with VLA and GMRT archival data, is typical of synchrotron self-absorption radiation. Moreover, there is evidence for a moderate Doppler boosting of the base of the jet, based on the correlation between core and total radio power. The 0.4-100 keV broad-band spectrum of IGR J21247+5058 is well modelled by a power-law continuum with a cut-off at around 70-80 keV, absorbed by two layers of cold material partially cover- ing the central source. A weak iron line is possibly present in the data, while the value of the reflection component is low (R∼0.4) and not well constrained. In many ways IGR J21247+5058 behaves like other BLRG which show weak reprocessing components and flatter X/gamma-ray power-law slopes than generally observed in radio quiet broad line galaxies like Seyferts (Grandi et al. 2006); both characteristics are generally interpreted as due to the presence of a beamed jet component in these bright radio sources. The ef- fect of this component is that of contaminating/diluting the Seyfert continuum which is likely present in the source. Indeed the high energy cut-off inferred by the INTEGRAL data is typical of Seyfert galaxies and not of beamed AGN like blazars, which tend to have much higher cut-off energies. We have not found strong evidence for a jet component in our broad-band data, although we cannot exclude its presence. A better way to look for beamed radiation in IGR J212471+5058, is by means of high resolution observations at radio frequencies, which would be extremely important and are therefore highly encouraged. On the other hand, complex absorp- tion is not a characteristic of classical BLRG and has been observed so far in only another object, i.e. 3C 445 (Sambruna et al. 2007). Both galaxies require cold absorbers partially covering the central source, even though in the case of 3C 445 three layers of obscur- ing materials are required while only two in IGR J21247+5058 are needed at low energies. However, while 3C 445 shows a strong iron line and a consequent strong reflection component, in the case of IGR J21247+5058 reflection is not required by the data and the iron line is weak or even absent. Several emission lines are also ob- served at low energies in the spectrum of 3C 445, making it some- what different from IGR J21247+5058. As for 3C445, the presence of absorption in IGR J21247+5058 is at odds with its Seyfert 1 clas- sification, but this discrepancy can be circumvented if the "clumpy torus" model recently proposed by Elitzur & Shlosman (2006) and successfully applied to 3C 445 (Sambruna et al. 2007) is also ap- plied to this newly discovered radio galaxy. In this model, the torus is not a continuous toroidal structure but is made of clouds with NH ∼ 1022-1023 atoms cm−2 distributed around the equatorial plane of the AGN. The BLR represents the inner segment of the torus and this is the region where the X-ray absorption is likely to oc- cur (Liu & Zhang 2002). Because of this clumpiness the difference between type 1 and 2 AGN is not only due to orientation but also to the probability of direct view of the active nucleus or, in other words, on how many clouds our line of sight intercepts. Indeed, the inclination to the line of sight inferred from the radio analy- sis suggests that we are peering directly into the broad line region; moreover, variations in the absorption properties are expected in this scenario, as the torus structure changes due to cloud motion. Overall we can conclude that IGR J21247+5058 is a new in- teresting member of the class of BLRG, with features that are in some respect typical of this class and others which are quite rare for its type: its brightness and proximity make it an ideal labora- tory in which to study radio galaxies properties. REFERENCES Arnaud K. A. 1996, Astronomical Data Analysis Software and Systems V, eds Jacoby G. and Barnes J., p.17, ASP Conf Series vol. 101 Bird A. J., Barlow E. J., Bassani L. et al. 2004, ApJ 607, 33 Bird A. J., Malizia A., Bazzano A. et al. 2007, ApJS, 170, 175 De Rosa A. et al., 2007 in prep. Dickey J.M. & Lockman F.J. 1990, ARA&A, 28, 215 Dietrich M., Peterson B.M., Albrecht P. et al. 1998, ApJ Suppl. Ser. 115, 185 Done C., Mulchaey J.S., Mushotzky R.F. et al., 1992, ApJ, 395, 275 Elitzur M. & Shlosman I. 2006, ApJL, 648, 101 Gehrels N., Chincarini G., Giommi P. et al. 2004, ApJ, 611, 1005 GeorgeI. M. & Fabian A. C. 1991, MNRAS, 249, 352 Giovannini G., Feretti L., Gregorini L. et al. 1988 A&A, 199, 73 Giovannini G., Cotton W.D., Feretti L. et al. 2001, ApJL, 522, 97 Goldwurm A., David P., Foschini L. et al. 2003, A&A, 411, 223 Grandi P., Guainazzi M., Haardt F. et al. 1999, A&A, 343, 33 Grandi P., Malaguti G., Fiocchi M. 2006, ApJ, 642, 113 Hill J.E., Burrows D. N. Nousek J.A. et al. 2004, SPIE, 5165, 217 Immler S., Brandt W. N., Vignali C. et al. 2003, ApJ, 126, 153 Landi R., De Rosa A. Dean A.J. et al. 2007, astro-ph0707.0832 Liu F.K., Zhang H. 2002, A&A, 381, 757 authoryearMalizia et al.2002]b41 Malizia A., Bassani L., Stephen J.B. et al. Mantovani F., Nanni M., Salter C. J. et al. 1982, A&A, 105, 176 Masetti N., Palazzi E., Bassani L. et al. 2004, A&A, 426, 41 Masetti, N., Landi, R., Pretorius, M.L. et al. 2007, A&A 470, 331 Molina M., Malizia A., Bassani L. et al. 2006, MNRAS, 371, 821 Pandey M., Manchanda R.K., Rao A.P. et al. 2006 A&A, 446, 471 Panessa F., Barcons X., Bassani L. et al. 2007, astro-ph/0701546 Perola G. C., Matt G., Cappi M. et al. 2002, A&A, 389, 802 Rib´o M., Combi J. A., Mirabel I. F. 2004, ATel 235 Sambruna R.M., Eracleous M., Mushotzky R.F. 1999, ApJ, 526, 60 Sambruna R.M., Reeves J.M., Braito V. 2007, astro-ph0704.3053 Struder L., Briel U., Dennerl K. et al. 2001, A&A, 365L, 18 Turner M. J. L., Abbey A., Arnaud, M. et al. 2001, A&A, 365L, 27 Terashima Y. & Wilson A.S. 2003, ApJ, 583, 145 Broad band X-ray spectrum of the newly discovered BLRG IGR J21247+5058 7 Table 2 Spectral Fits to XMM-Newton and Swift/XRT Γ N1 H (1022cm−2) f1 N2 H (1022cm−2) f2 F2−10keV (10−11 erg cm−2s−1) χ2 (dof) XMMa XMMb XRT1a XRT2a 1.35+0.02 −0.02 1.57+0.05 −0.05 1.54+0.19 −0.18 1.94+0.39 −0.37 1.11+0.10 −0.10 0.99+0.12 −0.12 2.08+0.88 −0.84 4.76+1.73 −1.67 0.76+0.02 −0.02 0.83+0.03 −0.02 0.74+0.07 −0.09 0.89+0.04 −0.07 - 9.14+1.93 −1.75 - - - 0.27+0.04 −0.04 - - 5.1 5.1 6.4 3.3 2422.4 (2622) 2345.8 (2620) 132.5 (140) 41.4 (48) Best fit parameters related to model (a) (wa*pcfabs*po) and model (b) (wa*pcfabs*pcfabs*po); Parameters are as following: Γ=Photon index; N1 H = Column density of first absorber; f1 = Covering fraction of first absorber; N2 H = Column density of second absorber; f2 = Covering fraction of second absorber; F2−10keV = 2-10 keV Flux and χ2 (dof)= Chi squares (degrees of freedom). Table 3 Spectral fits to broad-band XMM-Newton-INTEGRAL/IBIS and Swift/XRT-INTEGRAL/IBIS spectra Γ 1.67+0.05 −0.04 1.48+0.06 −0.06 1.54+0.08 −0.08 1.58+0.32 −0.25 1.75+0.15 −0.20 N1 H (1022cm−2) 10.96+1.81 −1.56 9.47+2.14 −1.98 8.19+2.27 −1.92 6.62+39.74 −6.57 4.95+1.18 −1.64 f1 N2 H (1022cm−2) f2 0.34+0.03 −0.03 0.25+0.04 −0.05 0.25+0.04 −0.05 0.13+0.08 −0.08 0.77+0.06 −0.07 1.11+0.12 −0.12 0.95+0.12 −0.13 0.97+0.12 −0.12 1.91+0.59 −0.49 0.56+1.95 −0.13 0.84+0.02 −0.02 0.82+0.03 −0.03 0.83+0.03 −0.03 0.75+0.08 −0.06 0.95+0.01 −0.961 Ec (keV) - 75.3+25.7 −15.9 99.1+54.5 −29.1 >50 >80 C 0.73+0.07 −0.06 0.85+0.08 −0.07 0.67+0.18 −0.12 0.78+0.32 −0.32 1.59+0.56 −0.31 R - - 0.42+0.47 −0.39 - - χ2 (dof) 2403.6 (2629) 2354.0 (2628) 2350.9 (2627) 138.8 (145) 47.8 (54) XMM/ISGRIa XMM/ISGRIb XMM/ISGRIc XRT1/ISGRId XRT2/ISGRId Best fit parameters related to model model (a) (wa*pcfabs*pcfabs*po), model (b) (wa*pcfabs*pcfabs*cutoffpl), model (c) (wa*pcfabs*pcfabs*pexrav) and model (d) (wa*pcfabs*pcfabs*cutoffpl); Parameters are as following: Γ=Photon index; N1 H = Column density of first absorber; f1= Covering fraction of first absorber; N2 H = Column density of second absorber; f2= Covering fraction of second absorber; Ec = Cut-off energy; C=Cross calibration constant; R= reflection Component; χ2 (dof)= Chi squares (degrees of freedom)
astro-ph/0703706
1
0703
2007-03-28T08:28:49
Pre-main-sequence stars in the young open cluster NGC 1893: II. Evidence for triggered massive star formation
[ "astro-ph" ]
(Abridged) The open cluster NGC 1893, illuminating the HII region IC410, contains a moderately large population of O-type stars and is one of the youngest clusters observable in the optical range. We have probed the stellar population of NGC 1893 in an attempt to determine its size and extent. We classify a large sample of cluster members with new intermediate resolution spectroscopy. We use H-alpha slitless spectroscopy of the field to search for emission line objects, identifying 18 emission-line PMS stars. We then combine existing optical photometry with 2MASS JHKs photometry to detect stars with infrared excesses, finding close to 20 more PMS candidates. While almost all stars earlier than B2 indicate standard reddening, all later cluster members show strong deviations from a standard reddening law, which we interpret in terms of infrared excess emission. Emission-line stars and IR-excess objects show the same spatial distribution, concentrating around two localised areas, the immediate vicinity of the pennant nebulae Sim 129 and Sim 130 and the area close to the cluster core where the rim of the molecular cloud associated with IC 410 is illuminated by the nearby O-type stars. In and around the emission nebula Sim 130 we find three Herbig Be stars with spectral types in the B1-4 range and several other fainter emission-line stars. We obtain a complete census of B-type stars by combining Stroemgren, Johnson and 2MASS photometry and find a deficit of intermediate mass stars compared to massive stars. We observe a relatively extended halo of massive stars surrounding the cluster without an accompanying population of intermediate-mass stars. The overall picture of star formation in NGC 1893 suggests a very complex process.
astro-ph
astro-ph
Astronomy&Astrophysicsmanuscript no. 6654 September 19, 2018 c(cid:13) ESO 2018 7 0 0 2 r a M 8 2 1 v 6 0 7 3 0 7 0 / h p - o r t s a : v i X r a Pre-main-sequence stars in the young open cluster NGC 1893,⋆ II. Evidence for triggered massive star formation I. Negueruela1,2,3, A. Marco1,3, G. L. Israel4, and G. Bernabeu1 1 Departamento de F´ısica, Ingenier´ıa de Sistemas y Teor´ıa de la Senal, Universidad de Alicante, Apdo. 99, E03080 Alicante, Spain e-mail: [email protected] 2 Observatoire de Strasbourg, 11 rue de l'Universit´e, F67000 Strasbourg, France 3 Department of Physics and Astronomy, The Open University, Walton Hall, Milton Keynes MK7 6AA, United Kingdom 4 Osservatorio Astronomico di Roma, Via Frascati 33, I00040 Monteporzio Catone, Italy Received ABSTRACT Context. The open cluster NGC 1893, illuminating the H ii region IC 410, contains a moderately large population of O-type stars and is one of the youngest clusters observable in the optical range. It is suspected to harbour a large population of pre-main-sequence (PMS) stars. Aims. We have probed the stellar population of NGC 1893 in an attempt to determine its size and extent. In particular, we look for signs of sequential star formation. Methods.We classify a large sample of cluster members with new intermediate resolution spectroscopy. We use Hα slitless spectroscopy of the field to search for emission line objects, identifying 18 emission-line PMS stars. We then combine existing optical photometry with 2MASS JHKS photometry to detect stars with infrared excesses, finding close to 20 more PMS candidates. Results. While almost all stars earlier than B2 indicate standard reddening, all later cluster members show strong deviations from a standard reddening law, which we interpret in terms of infrared excess emission. Emission-line stars and IR-excess objects show the same spatial distribution, concentrating around two localised areas, the immediate vicinity of the pennant nebulae Sim 129 and Sim 130 and the area close to the cluster core where the rim of the molecular cloud associated with IC 410 is illuminated by the nearby O-type stars. In and around the emission nebula Sim 130 we find three Herbig Be stars with spectral types in the B1 -- 4 range and several other fainter emission-line stars. We obtain a complete census of B-type stars by combining Stromgren, Johnson and 2MASS photometry and find a deficit of intermediate mass stars compared to massive stars. We observe a relatively extended halo of massive stars surrounding the cluster without an accompanying population of intermediate-mass stars. Conclusions. Stars in NGC 1893 show strong indications of being extremely young. The pennant nebula Sim 130 is an area of active massive star formation, displaying very good evidence for triggering by the presence of nearby massive stars. The overall picture of star formation in NGC 1893 suggests a very complex process. Key words. open clusters and associations: individual: NGC 1893 -- stars: pre-main sequence -- stars: emission line, Be -- stars: early-type 1. Introduction High mass stars are known to form preferentially in star clus- ters, but the exact details of how they are born and whether their formation has an impact on the formation of less massive stars are still the arguments of open discussions (see, e.g., ref- erences in Crowther 2002). As massive stars disrupt the molec- ular clouds from which they are born, it is generally assumed that the formation of massive stars closes a particular star for- mation episode by eliminating the material from which further Send offprint requests to: I. Negueruela ⋆ Partially based on observations obtained at the Nordic Optical Telescope and the Isaac Newton Telescope (La Palma, Spain) and Observatoire de Haute Provence (CNRS, France). stars may form (e.g., Franco et al. 1994). In this view, the for- mation of massive stars in an environment must take place over a short timescale. Numerous examples, however, seem to support the idea that the presence of massive stars triggers the formation of new stars in the areas immediately adjacent to their loca- tion (e.g., Walborn 2002). Such scenario would explain the formation of OB associations, extending over dozens of par- secs (Elmegreen & Lada 1977). Sequential star formation has been observed over both rather small spatial scales (e.g., Deharveng et al. 2003; Zavagno et al. 2006) and large, massive stellar complexes, such as 30 Dor (Walborn & Blades 1997), but in most cases doubts arise about the role of the first gen- eration of stars: does their impact on the surrounding medium 2 I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 actually trigger new star-formation episodes or simply blows away the clouds surrounding regions where star formation was already happening anyway? Investigations aimed at studying these questions can take a statistical approach (as in Massey et al. 1995) or concentrate on the detailed study of one particular open cluster where star for- mation is known to occur (e.g., the investigation of NGC 6611 by Hillenbrand et al. 1993). Unfortunately, there are not many open clusters with active star formation and a large population of OB stars easily accessible for these studies. One such cluster is NGC 1893, a rather massive cluster, with five catalogued O-type stars, which seems to have very re- cently emerged from its parental molecular cloud. The ionising flux of the O-type stars has generated the H ii region IC 410 on the edge of the molecular cloud (see Fig. 1). Though NGC 1893 is rather more distant than some other areas where star forma- tion can be studied, its very young age, moderately rich O star population and relatively low interstellar reddening make it a very interesting target for optical/infrared studies. Moreover, its large Galactocentric distance and projection onto a molecular cloud mean that there is very little contamination by a back- ground population. As with all very young open clusters, the age of NGC 1893 is uncertain. Several authors have estimated it around ∼ 4 Myr (Tapia et al. 1991; Vallenari et al. 1999), but the presence of luminosity class V early O-type stars would indicate an age < 3 Myr. Moreover, some emission-line B-type stars found in this area are likely to be PMS stars (Marco & Negueruela 2002, henceforth Paper II), and hence very young (as con- traction times for B-type stars are < 1 Myr). Even worse, Massey et al. (1995) classify some early-B stars in NGC 1893 as luminosity class III, implying ages ∼ 10 Myr. However, Massey et al. (1995) indicate that their spectral classifications are relatively rough, being directed to obtaining average spec- troscopic distances rather than discussing actual evolutionary stages. In Marco et al. (2001; henceforth Paper I), we used ubvy Hβ CCD photometry of ∼ 40 very likely main-sequence (MS) members to derive E(b − y) = 0.33 ± 0.03 and V0 − MV = 13.9±0.2 for NGC 1893. In Paper II, we identified several PMS candidates, based on their spectral type and observed colours, three of which were shown to be emission-line PMS stars. In this paper, we investigate the possible age spread in NGC 1893 and take a fresh look at the star formation process in this area by considering a rather larger field. The paper is structured as follows: in Section 2, we present the new obser- vations used in this study, which we discuss in Section 3, to- gether with existing optical photometry. In Section 4, we use 2MASS JHKS photometry to find stars with infrared excesses, which we identify as PMS candidates. We show that candidates concentrate around only two locations, notably in the vicinity of the bright emission nebula Sim 130. In Section 5, we present a spectroscopic study of stars in the area of this nebula, identi- fying several emission-line stars. Finally, in Section 6, we dis- cuss the interpretation of the results in terms of evidence for triggered star formation. Table 1. Log of new spectroscopic observations. The top panel shows low resolution spectroscopic observations. The bottom panel displays intermediate resolution observations. February 2001 observations are from the INT. October 2001 observations were taken with the OHP 1.93-m. December 2001 observations were taken with the NOT. Star Date of Dispersion λ Range S1R2N38 S2R2N43 S1R2N26 S1R2N26 S5003 S5003 E09 E10 E17 S1R2N35 S1R2N14 S1R2N40 S1R2N44 S1R2N55 S1R2N55 S1R2N56 S1R2N56 S1R3N35 S1R3N48 S2R3N35 S2R4N3 S3R1N5 S3R1N16 S3R2N15 S4R2N17 Hoag 7 HD 243035 HD 243070 observation 2001 Dec 5 2001 Dec 5 2001 Dec 6 2001 Dec 7 2001 Oct 25 2001 Dec 6 2001 Dec 6 2001 Dec 7 2001 Dec 7 2001 Feb 8 2001 Oct 24 2001 Oct 23 2001 Oct 24 2001 Oct 22 2001 Oct 23 2001 Oct 22 2001 Oct 23 2001 Oct 23 2001 Oct 24 2001 Oct 24 2001 Oct 24 2001 Feb 9 2001 Feb 9 2001 Feb 9 2001 Feb 9 2001 Oct 23 2001 Oct 24 2001 Oct 24 1.5Å/pixel 1.5Å/pixel 3.0Å/pixel 2.3Å/pixel 1.8Å/pixel 3.0Å/pixel 3.0Å/pixel 3.0Å/pixel 3.0Å/pixel 0.4Å/pixel 0.9Å/pixel 0.9Å/pixel 0.9Å/pixel 0.5Å/pixel 0.9Å/pixel 0.5Å/pixel 0.9Å/pixel 0.9Å/pixel 0.9Å/pixel 0.9Å/pixel 0.9Å/pixel 0.4Å/pixel 0.4Å/pixel 0.4Å/pixel 0.4Å/pixel 0.9Å/pixel 0.9Å/pixel 0.9Å/pixel 3830 -- 6830 Å 3830 -- 6830 Å 3100 -- 9100 Å 3100-6675 Å 3800 -- 6900 Å 3100 -- 9100 Å 3100 -- 9100 Å 3100 -- 9100 Å 3100 -- 9100 Å 3950 -- 5000 Å 3745 -- 5575 Å 3745 -- 5575 Å 3745 -- 5575 Å 6250 -- 7140 Å 3745 -- 5575 Å 6250 -- 7140 Å 3745 -- 5575 Å 3745 -- 5575 Å 3745 -- 5575 Å 3745 -- 5575 Å 3745 -- 5575 Å 3950 -- 5000 Å 3950 -- 5000 Å 3950 -- 5000 Å 3950 -- 5000 Å 3745 -- 5575 Å 3745 -- 5575 Å 3745 -- 5575 Å 2. Observations and data 2.1. Newdata We obtained imaging and slitless spectroscopy of NGC 1893 using the Andalucia Faint Object Spectrograph and Camera (ALFOSC) on the 2.6-m Nordic Optical Telescope (NOT) in La Palma, Spain, on the nights of December 5th-7th, 2001. The instrument was equipped with a thinned 2048 × 2048 pixel Loral/Lesser CCD, covering a field of view of 6.′4 × 6.′4. Standard Bessell UBVRI filters were mounted on the filter wheel, while a narrow-band Hα filter (filter #21, centred on λ = 6564Å and with FWHM= 33Å) was mounted on the FASU wheel. For the slitless spectroscopy, we made use of the Bessell R filter and grism #4. In total, 5 slightly overlapping images were taken, with 900-s exposure times. The weather was rel- atively poor, with some thin cloud veiling present on some of our exposures. The area covered by these observations is shown in Fig. 1. Spectroscopy of several emission-line stars in the field was taken on the same nights using the same instrument. I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 3 Fig. 1. The approximate boundaries of the five frames observed with slitless spectroscopy are indicated on the NGC 1893 field from a digitised DSS2-red plate. Note the position of the emission nebulae Sim 129 and Sim 130. The large white patch almost devoid of stars to the SW is the molecular cloud associated with IC 410. The five O-type stars in NGC 1893 are marked by circles and identified by name. The 8 early B-type stars in the periphery of NGC 1893 for which we present classification spectra are identified by squares. The O-type stars HD 242908 and HD 242926 are surrounded by bright nebulosity (darker patches). Grisms #3, #4 and #7 were used. A list of the objects observed is given in Table 1. We obtained intermediate-resolution spectra of stars in the region of the bright nebula Sim 130 and surrounding area dur- ing 22nd-25th October 2001, using the 1.93-m telescope at the Observatoire de Haute Provence, France. The telescope was equipped with the long-slit spectrograph Carelec and the EEV CCD. On the night of 22nd October, we used the 1200 ln/mm grating in first order, which gives a nominal disper- sion of ≈ 0.45Å/pixel over the range 6245 -- 7145Å. On the nights of 23rd & 24th October, the 600 ln/mm grating was used, giving a nominal dispersion of ≈ 0.9Å/pixel over the range 3745 -- 5575Å. Finally, on 25th October, the 300 ln/mm grat- ing was used, giving nominal dispersion of ≈ 1.8Å/pixel over the 3600 -- 6900Å range. Finally, intermediate-resolution spectroscopy of several bright stars in the field of NGC 1893 was obtained during a run at the 2.5-m Isaac Newton Telescope (INT) in La Palma (Spain), in February 2001. Details on the configurations used can be found in Paper II, while a list of all the observations presented here is given in Table 1. All the data have been reduced using the Starlink software packages ccdpack (Draper et al. 2000) and figaro (Shortridge et al. 1997) and analysed using figaro and dipso (Howarth et al. 1998). Sky subtraction was carried out by using the POLYSKY procedure, which fits a low-degree polynomial to points in two regions on each side of the spectrum. The ex- tent of these regions and their distance to the spectrum were selected in order to reduce the contamination due to nebular 4 I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 emission. Bright sky lines coming from diffuse nebular emis- sion are visible in almost all of our spectra. 2.2. Existingdata For the analysis, we combine the new observations with exist- ing photometric datasets. On one side, we use JHKS photom- etry from the 2MASS catalogue (Skrutskie et al. 2006). The completeness limit of this catalogue is set at KS = 14.3, which -- as we shall see -- roughly corresponds to the magnitude of an early A-type star in NGC 1893. In addition, we have two opti- cal photometric datasets. UBV data from Massey et al. (1995) cover a wide area around the cluster. Its magnitude limit is close to V ∼ 18, but errors become appreciable for V > 16. Stromgren photometry from Paper I only reaches V ∼ 16 and covers a more restricted central area. However, Stromgren photometry allows the determination of approximate spectral types under the single assumption of standard reddening. From Paper I, we know that V ∼ 16 roughly (depending on the red- dening) corresponds to the magnitude of an early A-type star in the cluster. Therefore, pretty much by chance, all three datasets have comparable limits. 3. Results 3.1. Emission-linestars In this paper, we address the search of emission-line PMS stars following a new approach: deep slitless spectroscopy of the whole field. This technique, based on the use of a low disper- sion grism coupled with a broad-band filter resulting in an "ob- jective prism-like" spectrogram of all the objects in the field, has been used by Bernabei & Polcaro (2001) for the search of emission line stars in open clusters. By combining a Johnson R filter and a low-resolution grism, we obtain bandpass imaging spectroscopy centred on Hα. The obvious advantage of this technique with respect to narrow-band photometry (as used in Paper I) is that it can reach very faint stars. The detection limit is difficult to define. In un- crowded regions, it is reached when the spectra are too faint to be seen against the background, but in crowded regions the overlap of adjacent spectra becomes important. Comparison to the photometry of Massey et al. (1995) shows that we have been able to detect emission lines in stars fainter than their limit at V >∼ 18, but we cannot claim completeness. We detect the 7 previously known emission-line objects (the 5 listed in Paper II and the two catalogued Hα emitters close to Sim 130). We fail to detect the candidate emission- line S5003 (Paper II), as its spectrum, given the orientation of our dispersion direction, is completely superimposed by that of its brighter neighbour S3R1N9. However, we have obtained a long-slit spectrum of this object and can confirm it as an emission-line star (see Table 3.1). In addition, we detect 10 new stars with Hα emission (see Table 3.1), and name them E09 -- E18. Only two of the new emission-line objects are bright enough to have been observed by previous authors, namely S1R2N55 and S1R2N26. We confirmed the emission-line na- ture of some of these objects through long-slit spectroscopy Table 2. Known emission-line stars in NGC 1893. E01 and E02 were already listed in the literature. E03 -- E07 were described in Paper II. E08 -- E18 are found or confirmed here. The spec- tral types of E09, E11 and E17 cannot be determined from our spectra. We did not take long-slit spectra of objects whose spec- tral type is marked as '−'. Note that the KS magnitudes (all from 2MASS) of some objects are affected by blending. Name RA Dec 05 23 09.2 E01 =S1R2N35 05 23 04.3 E02 =S1R2N38 05 22 43.0 E03 =S3R1N3 05 22 46.1 E04 =S3R1N4 05 22 48.2 E05 =S2R1N26 05 22 51.1 E06 =S2R1N16 05 22 52.1 E07 =S1R2N23 05 22 40.8 E08 = S5003 05 22 43.8 E09 05 22 49.6 E10 05 22 57.9 E11 = S1R2N26 05 23 00.0 E12 05 23 02.8 E13 05 23 04.4 E14 E15 05 23 06.3 E16 = S1R2N55N 05 23 08.3 05 23 08.9 E17 E18 05 23 09.9 +33 30 02 +33 28 46 +33 25 05 +33 24 57 +33 25 00 +33 25 47 +33 30 00 +33 24 39 +33 25 26 +33 30 00 +33 30 42 + 33 30 41 + 33 29 40 + 33 29 48 +33 31 02 +33 28 38 + 33 28 32 + 33 29 09 Spectral Type B1.5 Ve B4 Ve B0.5 IVe B1.5 Ve ∼G0 Ve ∼F7 Ve ∼F6 Ve Ke ? ? ∼A3 Ve ? − − − B1.5 Ve ? − KS 10.01 10.25 10.9 12.3? 11.9? 11.8 11.3 13.5 9.4 >15 12.6 13.0 13.9 13.2 13.2 > 10.3 11.5 13.2 (Table 1). Most of them are too faint for spectral classifica- tion. S1R2N26 is an A-type star with strong Hα emission. E12 does not show any photospheric features, but has all Balmer lines in emission and also shows strong Ca ii K lines and Ca ii 8498, 8542,8662 Å triplet emission. All the known emission- line stars in NGC 1893 are listed in Table 3.1. 3.2. O-typestars We have obtained accurate classifications for four O-type stars in the field of NGC 1893. The fifth O-type star, to the South of the cluster (HD 242926), was not observed here, but it has been observed as part of another programme and its spectral type is O7 V, in total agreement with Walborn (1973). This star is re- ported to show strong radial velocity changes by Jones (1972). The spectra of the four O-type stars observed are pre- sented in Fig. 2. The spectral types have been derived using the quantitative methodology of Mathys (1988), based on Conti's scheme. For S4R2N17 (HD 242908), the spectral type criteria fall on the border between O4 and O5 and we will adopt the O4 V((f)) classification given by Walborn (1973). In S3R2N15 (LS V +34◦15), the He i lines are stronger and the quantitative criteria indicate O5.5 V((f)). Jones (1972) reports two measure- ments of the radial velocity of this star, differing by more than 100 km s−1, strongly suggesting that there are two O-type stars in this system. In S3R1N16 (BD +33◦1025A), the condition He i λ4471Å ≃ He ii λ4541Å implies by definition O7, while the strength of He ii λ4686Å and very weak N iii emission in- dicate a MS classification. Finally, in S3R1N5 (HD 242935), I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 5 Table 3. O-type and early B-type stars around NGC 1893 for which we derive accurate spectral types, together with the derived reddening. Photometric data are from Massey et al. (1995). For HD 242935 there are no reliable photometric mea- surements due to heavy blending. Name V Spectral Type E(B − V) HD 242908 LS V +34◦15 Star Number S4R2N17 S3R2N15 S3R1N16 BD +33◦1025A 10.38 S3R1N5 S1R2N14 S1R3N35 S1R3N48 S2R3N35 S2R4N3 HD 242935 LS V +33◦23 LS V +33◦26 HD 243018 − LS V +33◦24 HD 243035 HD 243070 LS V +33◦27 Hoag 7 9.03 O4 V((f)) 10.17 O5.5 V((f)) O7 V O7.5 V((f)) B0.2 V B0.2 V B0 V B2.5 V B0.5 V B0.3 V B0.2 V B0.3 V − 11.22 11.17 10.94 12.43 11.02 10.90 10.83 10.74 0.57 0.79 0.59 − 0.45 0.54 0.42 0.47 0.66 0.50 0.51 0.49 Fig. 2. Classification spectra of the 4 O-type stars near the core of NGC 1893. The earliest spectral type is that of S4R2N17 (HD 242908), O4 V((f)). The other three objects have spec- tral types O5.5 V((f)) for S3R2N15, O7 V for S3R1N16 and O7.5 V((f)) for S3R1N5. He i λ4471Å is slightly stronger than He ii λ4541Å and the quantitative criteria indicate O7.5 V((f)), though He ii λ4686Å is rather weak and close to the limit for luminosity class III given by Mathys (1988). 3.3. B-typestars We have also derived new classifications for bright B-type stars to the East of NGC 1893, in order to check if they are connected to the cluster in spite of their relatively large angular distance to the cluster core (see their distribution in Fig. 1). These spectra are displayed in Fig. 3. At the resolution of our spectra, the traditional criterion for spectral classifica- tion around B0, namely the ratio between Si iv λ4089Å and Si iii λ4552Å (Walborn & Fitzpatrick 1990) is difficult to ap- ply, since Si iv λ4089Å is blended into the blue wing of Hδ. We have thus resorted to additional criteria. S1R3N48 is the earliest object in the sample, and the only one for which we give a B0 V classification, based on the conditions He ii λ4686Å ≃ C iii λ4650Å and He ii λ4686Å > He i λ4713Å. We have classified as B0.2 V those stars in which He ii λ4200Å is barely visible and He ii λ4686Å is compara- ble in strength to He i λ4713Å. We have taken as B0.3 V those stars in which He ii λ4200Å is not visible and He ii λ4686Å is clearly weaker than He i λ4713Å, and assigned B0.5 V to those objects in which He ii λ4686Å is visible, but very weak. All the stars analysed fall in the B0 -- B0.5 range (see Table 3), ex- cept for S2R3N35. For this object, we derive a spectral type B2.5 V, In this spectral range, the luminosity class is mainly in- dicated by the strength of the Si iii and Si iv lines compared to those of He i, while C iii and O ii lines can be used as subsidiary indicators (see Walborn & Fitzpatrick 1990, for a detailed dis- cussion). None of the spectra analysed suggests a luminosity class different from V. For all the B-type stars with accurate spectral types ei- ther here or in Paper I, we calculate the distance modu- lus, using UBV magnitudes from Massey et al. (1995), intrin- sic colours of Wegner (1994) and absolute magnitudes from Humphreys & McElroy (1984), under the assumption of stan- dard reddening (justified in the next section). The average value is 13.4 ± 0.3, in good agreement with the value found by Massey et al. (1995). 4. 2MASS data 4.1. InterstellarReddening The reddening along the face of NGC 1893 is known to be variable and this may have a bearing on the derivation of clus- ter parameters, as most methods are very sensitive to the in- terstellar reddening law assumed and treatment of the redden- ing. Deviations from the standard value are frequent in the optical, though the reddening law in the infrared has been proved to show very little variability along different lines of sight (Indebetouw et al. 2005). Because of this, we use the 2MASS JHKS photometry to check if the extinction law to- wards NGC 1893 is standard. For all B-type member stars with accurate Stromgren pho- tometry in Paper I, we calculate individual values of E(B − V) by using the simple relation E(B − V) = 1.4E(b − y). From the 2MASS magnitudes, we calculate E(J − KS) assuming the intrinsic colours from Ducati et al. (2001) and the spec- tral types derived from the Stromgren photometry. For a stan- dard reddening law (Rieke & Lebofsky 1985), we should have E(J − KS) ≃ 0.5E(B − V),1. We find that many stars have E(J − KS) ≈ 0.5E(B − V), but a significant fraction show rather larger E(J − KS) than expected from their E(B − V) . 1 The extinction law of Rieke & Lebofsky (1985) uses Johnson's K instead of KS, but the difference is negligible. Using the values calcu- lated by Hanson (2003), we would have E(J − KS) = 0.48E(B − V) 6 I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 Fig. 4. Plot of the infrared excess E(J − KS) against spectral type (represented by the corresponding Teff) for O and B-type MS stars in the cluster. The straight line E(J − KS) = 0.26 cor- responds to the cluster average E(B − V) = 0.53 (Massey et al. 1995) if a standard reddening is assumed. The error bars in E(J − KS) represent only the photometric errors. The uncer- tainty in the intrinsic colours has not been included, as it is difficult to quantify and unlikely to depend on spectral type. Fig. 3. Classification spectra of B-type stars in the area of NGC 1893. From top to bottom, stars are displayed from earlier to later spectral type. Photospheric features used for spectral clas- sification in this range are indicated. This deviation from the standard relationship is very un- likely due to a non-standard extinction law, because the amount of excess E(J − KS) is not strongly correlated with position within cluster. On the other hand, Fig. 4 shows the dependence of E(J − KS) with spectral type. For all O and B-type members, we plot E(J − KS) against the Teff corresponding to the spec- tral type derived according to the calibrations of Martins et al. (2005) for O-type stars and Humphreys & McElroy (1984) for B-type stars. Except for three stars, all stars earlier than B3 have about the same E(J − KS), compatible or very slightly above the value expected for a standard reddening law,2. All stars with spectral type B3 or later (except for two) have larger excesses, with a clear tendency to have larger excesses as we move to later spectral types. This dependence of the infrared excess on spectral type, while it shows no dependence on loca- tion, is clearly suggesting that the excesses are intrinsic to the stars and not related to the extinction law. To investigate this further, we use the chorizos code (Ma´ız-Apell´aniz 2004) to estimate the extinction law. This 2 Note that S3R2N15 has a high E(J − KS), but it also has E(B − V) much above other stars in Table 3. program tries to reproduce an observed energy spectral dis- tribution by fitting extinction laws from Cardelli et al. (1989) to the spectral distribution of a stellar model. As input, we used the UBV photometry from Massey et al. (1995) and the JHKS photometry from 2MASS, together with Kurucz models of main sequence stars with the Teff corresponding to our stars and log g = 4.0. We run the program for all the stars with spec- tral types derived from spectra or from Stromgren photometry. Out of 20 stars earlier than B3, 15 are best fit by reddening laws having 2.8 < R < 3.4, while 5 require R > 3.5. Out of 28 stars later than B3, 25 require R ≥ 3.5 and 3 did not converge to a solution. Again, we find a strong dependence of the red- dening law on the spectral type. Taking the average for all the stars with spectral type earlier than B3, we find R = 3.3 ± 0.2. Leaving out the 5 stars with R > 3.5, we find an average R = 3.16 ± 0.12. Our interpretation of this result is that the interstellar red- dening law to NGC 1893 is standard, but a substantial num- ber of objects show important individual (J − KS) excesses. These excesses are present in all stars later than B3 and in a few early stars (these early-type stars with anomalous values of R may perhaps have later-type companions with individ- ual excesses). This interpretation is further supported by the KS/(J − KS) diagram for cluster members (Fig. 5). All stars of mid and late B spectral type deviate strongly from the al- most vertical main sequence traced by early members, display- ing much larger (J − KS). This separation is in clear contrast I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 7 Fig. 5. Plot of KS against (J − KS) for all the 2MASS stars ful- filling the two conditions set in Section 4.2. Filled circles rep- resent O and B-type MS members from Marco et al. (2001) or with spectra in this paper. Open circles are known emission- line PMS stars. Stars represent the rest of the IR excess can- didates, three of which fall within the main sequence traced by known members (one has a known member superimposed). Note also the object with (J −KS) ≈ 3, our emission-line source E09. with the fact that stars in the B0 -- B8 range have almost iden- tical (J − K)0 (Ducati et al. 2001) and again confirms that all stars later than B3 show some E(J − KS) excess, with stronger excesses corresponding to later spectral types. Mid and late B stars fit the ZAMS rather well in differ- ent optical observational HR diagrams (e.g., Tapia et al. 1991; Massey et al. 1995; Marco et al. 2001), though they do not oc- cupy standard positions in the [c1]/[m1] diagram (Tapia et al. 1991; Marco et al. 2001). How do we then explain their E(J − KS) excesses? In Fig. 6, we plot an infrared HR diagram for the B-type members. For each star, we assume that the interstellar reddening E(J − KS)is = 0.5E(B −V) and the corresponding ex- tinction is AKS = 0.67E(J −KS)is, according to the standard red- dening law. We then shift their mKS = KS − AKS by DM= 13.5 (see Section 6.1) and plot them in the colour-magnitude dia- gram. We also plot the PMS isochrones from Siess et al. (2000) for 1 and 2 Myr, together with the youngest isochrone from Girardi et al. (2002), corresponding to 4 Myr, as the position of most B stars on this isochrone will not deviate in any measur- able way from the ZAMS. We observe that the smooth way in which the E(J − KS) excesses increase with decreasing mass does not fit at all the shape of the PMS isochrones. As a matter of fact, the posi- tions of the stars in the HR diagram cannot mean that they are moving towards the main sequence along PMS tracks, because the infrared excess, as it has been defined, is a measure of the discrepancy between optical and infrared colours. PMS stars Fig. 6. Observational HR diagram for B-type members of NGC 1893. The location of the ZAMS is represented by the 4- Myr isochrone from Girardi et al. (2002). The PMS isochrones for 1 and 2 Myr from Siess et al. (2000) are also shown. should have both optical and infrared colours appropriate for the position in the theoretical HR diagram. Indeed, if the devi- ation of B3 -- 4 stars from the ZAMS had to be attributed to their being on PMS tracks, the age of the cluster should be only a few 105 yr. In view of this and the fact that most B-type stars fit the ZAMS well in the optical, we are led to interpret the E(J − KS) excesses as due to the presence of material left over from the star formation process, most likely in the form of a remnant disk. If so, the PMS isochrones would suggest an age for the cluster of ∼ 2 Myr, with most stars later than B2 (i.e., stars with M∗ < 8M⊙) showing evidence for some remnant of a disk. This remnant cannot be very important, because stars fit the ZAMS well in the optical (both Johnson and Stromgren) HR diagrams and are not detected as emission line objects. High SNR spec- tra of the Hα line or in the H or K bands may be able to reveal some spectroscopic signature of such remnants and so confirm this hypothesis. 4.2. OBstarsandPMSselection The OB stars observed extend over a wide region to the East of the molecular cloud associated with IC 410. The core of NGC 1893 (defined as the region with the highest concentration of MS members) lies immediately to the West of the molecular cloud. There is an abrupt decrease in the number of optically visible stars as we move East from the O-type stars S3R1N16, HD 242935 and S3R2N15 (see Fig. 1). This suggests that an opaque part of the molecular cloud blocks our view (see also Leisawitz et al. 1989). As we lack three dimensional informa- tion, we cannot know if the cluster is spread over the wall of the dark cloud, but the fact that there is no bright luminosity 8 I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 in this area suggests that the dark cloud is partially between us and the cluster. As large areas of dark nebulosity block some sight-lines, we consider the possibility that there may be other OB stars in the field, obscured by gas and dust. The J, H and KS magnitudes from the 2MASS catalogue can be used to look for reddened early-type stars, under the assumption of a standard reddening law. Taking MK = −1.6 as the intrinsic magnitude for a B2 V star (Humphreys & McElroy 1984; Ducati et al. 2001) and DM = 13.9 to NGC 1893 (con- sidered an upper limit), all stars earlier than B2 located at the distance of the cluster and reddened according to the law of Rieke & Lebofsky (1985) will fulfil the following condition: KS − 1.78(H − KS) < 12.5 (1) Obviously many other stars will also fulfil this condition. However, if we define the reddening free parameter Q = (J − H) −1.70(H − KS ), OB stars will have Q ≃ 0.0. Combination of Condition 1 with Q ≃ 0.0 has been shown to be very efficient at identifying reddened OB stars (e.g., Comer´on & Pasquali 2005). However, not only OB stars fulfil both conditions. Foreground A-type stars and background red giants and super- giants will also fulfil the conditions. However, in our case, con- tamination by background red stars is unlikely, as the cluster lies at a large Galactocentric distance in the Anticentre direc- tion and is projected on to a dark cloud. We take all 2MASS objects within a 15′ radius of the centre of the cluster with magnitudes tagged as good and an error in the KS magnitude δKS ≤ 0.05 mag (larger errors would imply completely unreliable Q parameters) and select all stars ful- filling Condition 1 and having Q < 0.1. We discard foreground A-type stars using two criteria: (1) (B−KS) ≈ 0 indicates unred- dened stars (most stars have B magnitudes from Massey et al. (1995); for the rest we use USNO B1.0 magnitudes), and (2) (J − KS) lower than the average for the known OB members indicates that the stars are foreground to the cluster (only ob- jects substantially more reddened than this average would have escaped optical surveys). All known members earlier than B2 are selected by these two criteria. Moreover, because of their infrared excesses, sev- eral B-type members later than B2 are also selected. As a matter of fact, a substantial fraction of the emission-line stars listed in Table 3 (the brightest ones), including some of F type, have been selected. This is not surprising as these stars have strong infrared excesses and then : (a) are much brighter in KS than normal stars of the same spectral type (so much brighter that they pass our magnitude cut) and (b) the assumption of a standard interstellar law in the calculation of the Q parameter means that their (H − KS) excesses are "over-corrected", re- sulting in negative values of Q, values that no normal star can have. In view of this, we take stars fulfilling Condition 1, hav- ing Q < −0.05 and large values of (J − KS) as infrared ex- cess objects and therefore PMS star candidates. This procedure does not select all the objects with infrared excess in the field, but only those relatively bright and with strong (H − KS) ex- cesses. A search for a complete sample of infrared excess ob- jects in this field would need deeper photometry than provided by 2MASS and is beyond the scope of this paper. We find more than thirty stars fulfilling those criteria, among them, 11 of the known emission-line stars. Figure 5 shows the KS/(J − KS) diagram for all MS members (filled cir- cles) in NGC 1893 selected in Marco et al. (2001) and all the infrared excess stars. Known emission-line PMS stars (open circles) are located to the right of the main sequence, as ex- pected. We see that all the newly selected candidates occupy the same locations as the emission-line stars, but at fainter mag- nitudes, except for three objects falling close to the location of B-type MS members. Of these, one is a photometric member outside the area covered in Paper I, according to its UBV mag- nitudes. A second one is very far away from the cluster and could only be a member if it is much more reddened than any other member. Hence we reject it as a good photometric can- didate. The third object is S3R1N13, which was identified in Paper II as a candidate PMS star without emission lines. It has a spectral type B5 III-IV and its high reddening strongly sug- gests it is a PMS star in NGC 1893. The only alternative would be a foreground star with very unusual colours, but its spectrum does not show significant anomalies. Fig. 7 compares the distribution of these candidate infrared excess objects to that of MS members. The distribution is cer- tainly not random, as they strongly concentrate around two small areas, the vicinity of the pennant nebulae Sim 129 and Sim 130 and the rim of the molecular cloud closest to the clus- ter core. This distribution confirms beyond doubt that the stars selected are a population associated with the cluster rather than red background stars. Moreover, the spatial distribution of these objects coincides exactly with that of emission-line stars, giv- ing full support to the interpretation that most of them are also PMS stars. We cannot rule out the possibility that a few of these objects (especially the few ones at large distances from the cluster) are background red stars, but certainly the majority of these objects are PMS members with large infrared excesses. Once we account for known members, foreground stars, emission-line stars and infrared-excess candidate PMS stars, we have exhausted the list of stars selected according to our original criteria. This means that there are no obscured OB stars in this region, at least within the magnitude limit of 2MASS. Note, however, the position of E09 in Fig. 5. If this object be- longs to the cluster, it should be a young massive stellar ob- ject. Unfortunately, this object is so faint in the optical that our spectrum is extremely noisy. The only obvious feature is a very strong Hα emission line. Another possible emission fea- ture may correspond to the O i 8446Å line. 5. The area around Sim 130 Results in the previous sections clearly show that present-day star formation in NGC 1893 is strongly concentrated towards the pennant nebulae Sim 129 and Sim 130, with most of the PMS stars located around the latter. The "head" of Sim 130 contains a group of stars which were observed photometrically by Tapia et al. (1991) as if they were a single object (their Star 35). They derive the colours of an early-type star with emission lines. The head of Sim 130 has also been identified as the near-infrared counterpart of the IRAS source IRAS 05198+3325, considered a Young Stellar I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 9 Fig. 7. The spatial distribution of all known members of NGC 1893 is shown on this DSS2 image. Objects on the main sequence according to their optical colours are shown as squares (red for spectral types B2 and earlier, yellow for B3 and later). PMS candidates (emission line stars and infrared excess objects) are shown as blue circles. Main-sequence objects are spread over a much larger area than PMS candidates. Object (YSO) candidate (CPM16 in Campbell et al. 1989). In addition, in the immediate vicinity of Sim 130, we find the emission-line star S1R2N35, which based on low-resolution spectra, was considered to be a very good candidate to a Herbig Be star in Paper I. Not very far away lies the catalogued emission-line B star NX Aur = S1R2N38. Figure 8 identifies the brightest stars immersed in the bright nebulosity of Sim 130: S1R2N56 ([MJD95] J052307.57+332837.9 in the catalogue of Massey et al. 1995), S1R2N55 ([MJD95] J052308.30+332837.5) and a fainter star not observed by Cuffey (1973), which we will call S5004 ([MJD95] J052306.71+332840.3). There are many other fainter stars within the nebula (more clearly seen in R- band images), among them our emission-line objects E17 and E18. Also, partially immersed in the nebulosity, we find the bright star S1R2N44. 5.1. S1R2N44 S1R2N44 lies very close to Sim 130, just to the SW. Its spec- trum, displayed in Fig. 10, shows nebular emission lines on top of a normal absorption B-type stellar spectrum. A lower reso- lution spectrum does not show any signs of intrinsic emission in Hα. However, a cut of the spectrum in Hα clearly shows that the intensity of nebular Hα emission increases consider- ably around the source, strongly suggesting the association of S1R2N44 with the nebulosity. From the Full Width at Half Maximum (FWHM) of four He i lines, we estimate for S1R2N44 an apparent rotational ve- locity v sin i ≈ 320 km s−1, following the procedure described by Steele et al. (1999). We estimate its spectral type at B2.5 V. For this star, Fitzsimmons (1993) gives (b − y) = 0.273, imply- ing E(b−y) = 0.39, slightly above the average for cluster mem- bers. If the reddening is standard, V = 13.3 implies MV = −1.9, 10 I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 Fig. 8. An Hα image of the head of the cometary nebula Sim 130 (a portion of a 300-s exposure obtained on the night of 5th December 2001 with ALFOSC on the NOT, equipped with the narrow-band Hα filter #21). The stars discussed in Section 5 have been identified. Note the bow-shaped emission rim and the extended nebulosity surrounding the stars. Fig. 9. Hα spectra of the two stars in the "head" of Sim 130. While the emission features on the spectrum of S1R2N56 seem to be entirely due to the surrounding nebulosity, S1R2N55 is clearly an emission-line star (see the inset for details). This ob- ject is the counterpart to the IRAS source 05198+3325, identi- fied as a massive young stellar object. which is not surprising for the spectral type. This star must have reached the ZAMS, as its observed (J − KS) = 0.25 indicates that this object has no infrared excess. 5.2. S1R2N56 This star forms the "head" of Sim 130. Its spectrum is strongly contaminated by nebular emission, but this can be easily dis- tinguished from photospheric features at our resolution. There does not seem to be emission intrinsic to the star (see the Hα profile in Fig. 9, where also a weak He i λ6678Å absorption line can be seen). The blue spectrum of this object is displayed in Fig. 10. From the FWHMs of four He i lines, we estimate an apparent rotational velocity v sin i ≈ 210 km s−1. We estimate its spectral type at B1.5 V. For this object, Massey et al. (1995) measure (B − V) = 0.43, implying a reddening E(B − V) = 0.66, well above the average for cluster members. If the reddening law is standard, then the measured V = 13.54 implies MV = −2.0, which, though slightly too low for the spectral type, is not in strong disagreement. The 2MASS colour (J − KS) = 0.76 implies substantial reddening and this object is selected as an infrared excess candidate, but this could mainly be due to contamination of its photometry by the bright nebulosity. The available evidence suggests that S1R2N56 is a star settling on to the ZAMS. Observations by previous authors seem to indicate very large variability in V with a long-term dimming from V < 13, but since none of these authors mentions S1R2N55, it is pos- sible that both stars have been measured together (this is cer- Fig. 10. Blue spectra of S1R2N56 (top) and S1R2N44. S1R2N56 is immersed in the pennant nebula Sim 130 and its spectrum shows strong nebular emission. Nebular emis- sion is also present, though rather weaker, in the spectrum of S1R2N44, which lies just outside Sim 130. tainly the case in Tapia et al. 1991, whose star #35 corresponds to S1R2N56+S1R2N55). 5.3. S1R2N55 S1R2N55 is clearly immersed in the nebulosity associated with Sim 130. As a matter of fact, our NOT images clearly show S1R2N55 to be a close double, but the two components were not resolved during the OHP run, when spectra were taken. The red spectrum is shown in Fig. 9. In addition to strong nebular lines, a very broad Hα emission line can be seen: S1R2N55 is a Be star. The Hα line peaks at 15 times the con- tinuum intensity and has an Equivalent Width (EW) of −54 ± 3 I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 11 ber, then its intrinsic magnitude would be MV ≈ −1.7, clearly very bright for a normal main-sequence G-type star. S5004 may then be an intermediate mass star still on the contraction track, similar to θ1 Ori E (Herbig & Griffin 2006). A spectrum around the Li i 6707Å line could test this hypothesis. Alternatively, the photometry could be in error because of nebular contamination and this could be a foreground star. 5.5. E17 This star, too faint in the optical to have been observed by Massey et al. (1995), is very bright in the K-band (KS = 11.49 in 2MASS) and is selected as an infrared excess object. Its spectrum is completely featureless, except for a prominent Hα emission line EW= −40 ± 2Å. The emission line and infrared excess identify this object as a PMS star. 5.6. S1R2N35=E01 Classified by Kohoutek & Wehmeyer (1999) as an emission line star, this object in the immediate vicinity of Sim 130 was proposed as a Herbig Be candidate in Paper I, based on a low- resolution spectrum. A higher resolution spectrum is shown in Fig. 11. Classification is complicated by the strong emission lines, but, based on the presence of some weak O i lines and possible strength of Si iii lines, while the Si iv lines are not vis- ible, we adopt a B1 V, though it could be slightly later. Massey et al. (1995) give V = 12.33, E(B − V) = 0.52, indicating a very large colour excess E(B − V) ≃ 0.8. Again, this object is selected as an infrared excess candidate. From the intrinsic colour of a B1 V star, the implied excess is E(J −K)S = 0.57. This object is hence a Herbig Be star. 5.7. S1R2N38=E02 Already known as an emission-line and variable star (NX Aur), this object lies in the immediate vicinity of Sim 130, display- ing Hα and Hβ strongly in emission, EW(Hα)= −55 ± 2Å and EW(Hβ)= −3.7 ± 0.3Å. The upper Balmer lines display weak blue-shifted emission components. Prominent emission lines of Fe ii are also present. Though the object is clearly a B-type star, an exact spectral type is difficult to derive from our low- resolution spectrum. Based on the strength of the Mg i 4481Å line, we estimate it to be B4 V. Massey et al. (1995) give V = 14.41, E(B − V) = 0.57, indicating a E(B − V) ≈ 0.8. 2MASS gives KS = 10.25 and (J − KS) = 1.76, implying a huge infrared excess E(J − KS) ≈ 1.9. This object is hence also a Herbig Be star. 6. Discussion 6.1. Distance,reddeningandextent Our analysis shows that the optical/near-IR spectral energy dis- tributions of almost all stars earlier than B3 are best fit when a standard R = 3.1 reddening law is assumed. A standard inter- stellar law has also been found by Yadav & Sagar (2001), using less sophisticated techniques. Tapia et al. (1991) found a value Fig. 11. Blue spectra S2R1N35 (top) and S1R2N55, two early Herbig Be stars associated with Sim 130. S1R2N55 is actually a blend of two very close B-type stars, and likely only one of them is a Herbig Be star. Å, a fraction of which is attributable to the nebular component. Given its presence in the middle of bright nebulosity in an H ii region with active star formation, S1R2N55 is almost certainly a PMS Herbig Be star. In the R-band, the Northern component is slightly brighter than the Southern one, but in the Hα im- ages, it is much brighter. This clearly shows that the Northern component of S1R2N55 is the emission-line object. Ishii et al. (2001) identified S1R2N55 as the counterpart to the IRAS source IRAS 05198+3325 (YSO CPM16). In their K- band spectrum, they found strong Brγ emission from the source and H2 emission of nebular origin. The blue spectrum of S1R2N55 is shown in Fig. 11. Emission is present in Hβ and Hγ. The spectrum is likely dom- inated by the brighter Northern component, but must be a com- bination of the spectra of the two stars. From the apparent pres- ence of weak O ii 4076Å and Si iv 4089Å, the brighter compo- nent must be around B1 V. The fainter component cannot have a very different spectral type, around B2 V. This object is se- lected among our infrared excess candidates. At least the bright component is a Herbig Be star. 5.4. S5004 The faint star S5004 is located just on the bright rim of nebu- losity defining the head of the cometary nebula. In spite of this, we have been able to clean the spectrum of nebular emission. The spectrum shows a prominent G-band, comparable in inten- sity to Hγ, indicating a spectral type close to G0. The weakness of the Sr ii 4077Å and other luminosity indicators seems com- patible with a G0 V star. Massey et al. (1995) give V = 15.79, (B − V) = 1.27 for this object. This star is prominent in the K- band (KS = 11.55) and is selected as one of our infrared excess candidates. Indeed, comparison of the observed (J −KS) = 1.40 with the colours expected for this spectral type (Ducati et al. 2001) implies E(J − KS) = 1.13. This value is much larger than would correspond to its E(B − V), indicating a large infrared excess, typical of a PMS star. If this is object is a cluster mem- 12 I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 of R = 2.8 ± 0.1, slightly lower than our value. On the other hand, all stars later than B3 show evidence for an abnormal law, in the sense that E(J − KS) is larger than expected from the measured E(B − V). We interpret this result as showing that stars later than B3 have infrared excess emission, rather than non-standard reddening. This excess should be a consequence of the presence of some remnant of the disk from which the stars formed and points to a very young age for NGC 1893. The presence of this emission excesses likely accounts for the existence of some dispersion in the distance estimates for NGC 1893, ranging from DM= 13.1 (Tapia et al. 1991) to 13.9 (Paper I), as the different procedures used to deredden the data will treat this excess differently. In view of this difficulty, we prefer to assume the distance derived from spectroscopic par- allaxes using a standard R = 3.1 reddening law, i.e., a value d ∼ 5 kpc, as this is in line with the distances to tracers of the Outer Arm in the vicinity of NGC 1893 (Negueruela & Marco 2003). Since the area covered in this investigation is larger than the area covered by our photometry (Paper I), we can use the photometry of Massey et al. (1995) to identify B-type mem- bers outside the cluster core. In the B/(U − B) diagram, they form a very well defined sequence at bluer colours than any other star in the field. Comparison to the estimated spectral types of Paper I shows that B-type members are delimited by (U − B) < −0.1 and B < 15.5. This agrees well with the as- sumption of standard reddening, as a B9 star has (U − B)0 = −0.57 and therefore for the average reddening to the cluster E(U − B) = 0.72E(B − V) = 0.38, a B9 star should have (U − B) = −0.19. Three foreground stars can be easily iden- tified because of their bright B and position in the V/(B − V) di- agram, leaving ∼ 60 stars along the cluster sequence. Of them, 25 stars are spectroscopically confirmed to be B2 or earlier: 5 O-type stars and 20 in the B0-2 range (of which 4 show emis- sion lines). 6.2. Age Is there an age spread in NGC 1893? We do not find any ev- idence for a deviation from the ZAMS for almost any mem- ber. The only star that seems (slightly) evolved is the O7.5 V S3R1N5. However, there are other possible interpretations to its spectrum. For example, it could be the composite of two O- type stars with slightly different spectral types. We believe this to be rather possible, as this star lies in the densest region of the cluster and its spectroscopic distance modulus is rather shorter than the average for the cluster. Moreover, it is surrounded by PMS stars and shows a strong IR excess. Therefore we do not think that this is an evolved star, especially as nearby stars of earlier spectral type do not show any sign of evolution. The fact that an O4 V and an O5.5 V star are still close to the ZAMS places a strict upper limit on the age of the cluster at 3 Myr and supports an even younger age (Meynet & Maeder 2003). Comparison with PMS isochrones (Fig. 6) suggests an age of <∼ 2 Myr, in good agreement with the fact that stars as massive as B3 -- 4 V still show significant E(J − KS) excesses, likely due to the presence of remnants of the disks from which they formed. However, it is obvious that, while an important population of B-type stars, some of them as late as B8 -- A0, is already set- tling into the ZAMS, some massive stars, with spectral types in the B1 -- B2 range are still in the Herbig Be phase. This shows that there is some spread in the formation of stars. In this paper, we have selected the PMS stars with the strongest signatures of youth (emission lines and/or strong E(H − KS) excess) and found that their distribution is limited to two small regions within the relatively large area covered by MS mem- bers. The fact that recent star formation is confined to these two regions suggests that we are observing the star formation process spreading from the central cluster to the neighbouring dark cloud. The lack of three-dimensional information does not allow us to determine the relationship of the young PMS stars close to the cluster centre with the members already on the main se- quence. Images of the area suggest that part of the dark cloud is partially hiding the cluster and perhaps most of the star for- mation is taking place on the inner wall of the cloud, which we cannot see. However, it seems that the population of PMS stars to the East of the cluster is being formed on the illuminated surface of the molecular cloud around the two bright pennant nebulae Sim 129 and Sim 130. 6.3. TriggeredstarformationinSim130 As described in Section 5, the area surrounding the pennant nebula Sim 130 contains three Herbig Be stars and several other emission-line stars. Other emission-line stars cover the area between Sim 130 and Sim 129. In the vicinity of Sim 129, there are two early A-type PMS stars, S1R2N4 (Paper II) and S1R2N26 (Table 2). The impact of the ionising flux from the O-type stars on the nebulae is obvious. Their cometary aspect is due to the pres- ence of bright ionised fronts, taking a shape strongly resem- blant of a bow shock, combined with a "tail" that seems to run away from the centre of the cluster and is actually composed of bright filaments illuminated by the O-type stars. Moreover, similarly to other star forming regions in the vicinity of mas- sive clusters (e.g., M16; Hester et al. 1996), both nebulae show finger-like dust structures oriented towards the nearby O-type stars. In view of these properties, the area of recent star forma- tion around Sim 130 presents all the characteristics listed by Walborn (2002) as typical of regions of triggered star forma- tion: -- The younger (second generation) stars are associated with dust pillars oriented towards the O-type stars. -- The second generation is less massive than the first. In this case, we have 6-7 early and mid B-type stars, as compared to the ∼ 20 massive stars in the main cluster. -- The more massive stars in the second generation are less massive than the more massive stars in the first cluster (in this case, the earliest spectral type around Sim 130 is B1 V, as opposed to O4 V in the main cluster). I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 13 Walborn (2002) finds a characteristic age difference of ∼2 Myr between the first and second generation. This may indeed reflect the smallest age difference that we are able to detect, but the observed properties of the two populations in NGC 1893 are not incompatible with such an age difference. Obviously, it may be argued that the stars around Sim 130 may represent star formation that would have occurred regard- less of the effect of the nearby massive stars and that it is sim- ply being exposed now because of the erosion of the molecular cloud by the ionising flux of the O-type stars. Even though it is difficult to find strong arguments against this view, it should be noted that MSX images of the molecular cloud associated with IC 410 do not give any reason to suggest that star forma- tion is taking place anywhere except in the two areas marked by emission-line stars, just over the surface of the molecular cloud. It would be surprising if the only places in the molec- ular cloud in which star formation is spontaneously occurring happen to be, just by chance, in the process of being cleared up by the ionising flux of the nearby O-type stars. 6.4. ThePMSstars We have shown that, with the possible exception of the emission-line star E09, there are no obscured massive stars. We can consider the census of stars earlier than B3 complete. E09 is very bright in KS and extremely reddened. It is likely asso- ciated with the IRAS source 05194+3322,3 and therefore may be a massive young stellar object. We cannot define a strict detection limit for emission-line stars, but it is very unlikely that the slitless observations may have skipped any relatively bright (V < 15) stars. This means that, while there are 4 early Herbig Be stars in NGC 1893 (see Table 3.1), there is one mid Herbig Be star, no late Herbig Be stars and only one Herbig Ae star (also, as can be seen in Fig. 5, the K magnitudes of all the candidate infrared excess stars are too faint to expect any of them to be B-type stars). This fact is difficult to interpret. Obviously, any reasonable IMF should result in a rather larger number of intermediate-mass stars than massive stars. Moreover, the more massive the star is, the ear- lier it should reach the ZAMS. Finally, the UV flux of the early B-type stars should help to dissolve their disks much more quickly than those around lower mass stars. How can we then explain the excess of early Herbig Be stars with respect to other emission-line stars? The possibility that early Herbig Be stars retain their disks longer than later type Herbig ABe stars is counterintuitive. If the star formation process in the two active areas is very recent, it may be pos- sible that only the early Herbig Be stars have emerged from the parental cloud and less massive stars are too faint to be observed in the optical. In this view, the mid and late B stars that we see are associated with the first generation of massive stars, while the second generation is now emerging from the parental cloud, likely because the UV photons from the O-type stars are photodissociating the cloud around them. The later- 3 This source is identified in SIMBAD with the O7.5 V member HD 242935, but its coordinates coincide much better with E09. type emission line stars should then also be associated with the first generation of massive stars. 6.5. TheIMF The extreme youth of NGC 1893 offers a good prospect for determining the IMF of a population just emerging from the parental cloud. For this, deep infrared observations would be needed in order to probe the low-mass stellar populations. However, some difficulties stand out. First, as discussed above, it is possible that part of the clus- ter is obscured by parts of the dark cloud. Assuming typical masses for spectral types, the observed distribution of mem- bers is 5 stars with 25M⊙ ≤ M∗ ≤ 60M⊙ (O4 -- 07.5), 20 stars with 8M⊙ ≤ M∗ ≤ 16M⊙ (B0 -- B2) and ∼ 35 stars with 3M⊙ ≤ M∗ ≤ 8M⊙ (B2.5 -- B9). This distribution seems too biased towards early spectral types for a normal IMF. For a Salpeter IMF, which is valid in this mass range (Kroupa 2001), we would expect three times more intermediate mass (B2.5 -- B9) than massive(≤B2) stars. Ignoring any incompleteness due to multiplicity, the 25 massive stars with known spectral types result in a mass ∼ 400M⊙. Assuming for the sake of argument that the deficit in intermediate mass stars is due to observational effects and the IMF is standard (i.e., Kroupa 2001), this mass implies a total mass Mcl ∼ 2200M⊙ for NGC 1893. This estimate is a lower limit, as there are reasons to believe that a substantial fraction of the most massive stars in the cluster are, at least, binaries: a few radial velocity measurements by Jones (1972) show most of them to display large velocity variations. A second factor suggesting obscuration of part of the clus- ter is its shape. The distribution of members is traced in Fig. 7. If we take the intermediate mass stars as best tracers of the clus- ter extent, it is difficult to assign a morphological type or even define a centre. The main concentration of stars appears just on the edge of the molecular cloud. The conspicuous absence of any likely member to the West of the cluster core strongly sug- gests that NGC 1893 is located on the back side of the molec- ular cloud associated with IC 410. An even more striking difficulty is the fact, evident in Fig. 7, that there is a halo of high-mass stars surrounding the clus- ter in areas where there are essentially no intermediate-mass members. This is more obvious to the East of the cluster, be- yond Sim 130. If we consider the area lying between S2R3N35 (RA: 05h23m13s)) and the edge of Figure 7, it contains the 8 members identified in Fig. 1. These objects lie at distances of 8′ to 12′ from the cluster core (corresponding to 12 to 18 pc at d ∼ 5 kpc). As seen in Table 3, 7 of them have spec- tral types in the B0 -- 1 range and one is a mid B-type star. This area is fully covered by the photometry of Massey et al. (1995), which provides only three other photometric members. The brightest one, [MJD95] J052325.13+332609.8 was clas- sified as B1.5 by Massey et al. (1995), while the two other photometric members, [MJD95] J052336.60+332905.5 and J052339.25+333839.7 have colours and magnitudes appropri- ate for mid-B stars (note that J052339.25+333839.7 falls out- side the area covered in this investigation and lies outside 14 I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 Fig. 7, ∼ 4′ to the North of the Northernmost members dis- played). Therefore, this area includes 8 early B stars, 3 mid B stars and no late B-type photometric members. How do we arrive at such surprising mass distribution? An obvious candidate for an explanation is dynamical ejec- tion from the cluster core. As discussed by Leonard & Duncan (1990), massive clusters containing hard binaries with two components of similar mass may be quite effective at eject- ing stars via dynamical ejections. The majority of stars ejected will be B-type stars and their ejection velocity will be in- versely proportional to their mass. These factors could explain a concentration of early B-type stars in the vicinity of a young massive cluster, but the efficiency at ejecting stars estimated by Leonard & Duncan (1990) seems much lower than that re- quired to explain the population of objects around NGC 1893. However, more recent work by Pflamm-Altenburg & Kroupa (2006) suggests that, if massive stars are mainly born as part of multiple systems, the ejection rates can be much higher than estimated by Leonard & Duncan (1990). According to their re- sults, a cluster with a mass comparable to that of NGC 1893 could lose ∼ 75% of its high mass stars in 1 -- 2 Myr. In this respect, it is interesting to note that there is one further early-type star, LS V +33◦31, classified B0.5 V by Massey et al. (1995), lying about 24′ away from the cluster core, which falls along the sequence of cluster members. In the whole area covered by Massey et al. (1995), there are only three more objects that could be photometric members, all very distant from the cluster (d > 15′) and all compatible with be- ing late B-type stars [MJD95] J052425.46+331544.7,[MJD95] J052346.82+331440.6 and [MJD95] J052516.96+332403.9. Even more striking is the fact that HD 242908, nominally the most massive star in the cluster, lies at some distance from the main bulk of the cluster, in an area where very few other mem- bers are found. Intriguingly, if we do not count stars in this massive star halo, the ratio between massive and intermediate- mass stars is close to standard. Radial velocity measurements of the stars in the halo of NGC 1893 could provide a test on this hypothesis, as the sys- temic velocity of the cluster is close to zero (e.g., Jones 1972) and any measured components should be due to runaway ve- locities. One further issue to take into account is the complication arising from the presence of sequential star formation. If sites of triggered star formation (such as Sim 130) contribute only stars less massive than ∼ 12M∗ (B1 V), the total integrated (first generation + triggered generations) population will have a steeper IMF than the original first generation. If we observe this cluster in a few Myr, there will be no way of telling which stars have formed at which time. Of course, this has a bearing on how we can define an instantaneous IMF. 7. Conclusions 1. We have found a population of emission-line PMS stars in NGC 1893. The brightest among them cover the range from B1 to late F, with an obvious overpopulation of early B-type stars. Emission line stars appear only in two regions of the cluster. 2. We have identified a number of faint objects with high val- ues of (J − KS) that seem to show an infrared excess. These objects concentrate around the emission-line stars, indicat- ing that they are also PMS stars. 3. All the stars later than B3 show evidence for an infrared ex- cess, even though the main sequence is well traced down to A0 in the optical. This infrared excess increases as we move to later spectral types, strongly suggesting that it arises from the remnant of a disk. 4. The age of NGC 1893 is constrained to be < 3 Myr by the presence of main-sequence 04 and O5 stars and likely to be <∼ 2 Myr. This makes NGC 1893 one of the youngest clus- ters to be visible in the optical. It is very likely in the pro- cess of emerging from its parental cloud and perhaps more members lie hidden by dark portions of the cloud. If this is the case, they are quite faint and infrared observations reaching deeper than 2MASS are needed to detect them. 5. The area around the cometary nebulae Sim 129 and Sim 130 shows the highest number of emission-line and IR-excess PMS stars. Three B1-B4 Herbig Be stars cluster around Sim 130. This is likely to be a region of more recent star formation, triggered by the ionisation front generated by the O-type stars. 6. A second region containing emission-line stars and IR- excess PMS candidates lies on the interface between the cluster core and the molecular cloud. Here we could have another area of triggered star formation partially hidden by the molecular cloud. On the very edge of the cloud, we find the emission line object E09, which, with KS = 9.4 and (J − KS) = 3.0, could be a massive very young stellar ob- ject. 7. The picture of star formation emerging from our study of NGC 1893 is a rather complex one, with sequential star for- mation resulting in several slightly non-coeval populations and dynamical ejection depopulating the cluster of massive stars at a very young age. Acknowledgements. IN is a researcher of the programme Ram´on y Cajal, funded by the Spanish Ministerio de Ciencia y Tecnolog´ıa (currently Ministerio de Educaci´on y Ciencia) and the University of Alicante, with partial support from the Generalitat Valenciana and the European Regional Development Fund (ERDF/FEDER). This re- search is partially supported by the MEC under grant AYA2005-00095 and by the Generalitat Valenciana under grant GV04B/729. During part of this work IN and AM were visiting fellows at the Open University, whose kind hospitality is warmly acknowledged. IN was funded by the MEC under grant PR2006-0310. AM was funded by the Generalitat Valenciana under grant AEST06/077. The INT is operated on the island of La Palma by the Isaac Newton Group in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrof´ısica de Canarias. Based in part on observations made at Observatoire de Haute Provence (CNRS), France. IN would like to express his thanks to the staff at OHP for their kind help during the observing run. The Nordic Optical Telescope is operated on the island of La Palma jointly by Denmark, Finland, Iceland, Norway, and Sweden, in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias. Part of the data presented here have been taken using ALFOSC, which is owned by the Instituto de Astrof´ısica de Andaluc´ıa (IAA) and oper- I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 15 ated at the Nordic Optical Telescope under agreement between IAA and the NBIfAFG of the Astronomical Observatory of Copenhagen. This research has made use of the Simbad data base, operated at CDS, Strasbourg (France) and of the WEBDA database, operated at the Institute for Astronomy of the University of Vienna. This pub- lication 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 Foundation. We thank the anonymous referee for insightful comments that helped us clarify some topics. References Bernabei, S., & Polcaro, V.F. 2001, A&A, 371, 123 Campbell, B., Persson, S.E., & Matthews, K. 1989, AJ, 98,643 Cardelli, J.A., Clayton, G.C., Mathis, J.S. 1989, ApJ, 345, 245 Comer´on, F., & Pasquali, A. 2005, A&A, 430, 541 Crowther, P.A. (ed.) 2002, Hot Star Workshop III: The Earliest Stages of Massive Star Birth. ASP Conference Proceedings, vol. 267, San Francisco Cuffey, J. 1973, AJ, 78, 747 Deharveng, L., Zavagno, A., Salas, L., et al. 2003, A&A, 399, 1135 Draper, P.W., Taylor, M. , & Allan, A. 2000, Starlink User Note 139.12, R.A.L. Ducati, J. R., Bevilacqua, C. M., Rembold, S. B., & Ribeiro, D. Massey, P., Johnson, K.E., and DeGioia-Eastwood, K. 1995, ApJ, 454, 151 Mathys, G. 1988, A&AS, 76, 427 Meynet, G., Maeder, A. 2003, A&A, 404, 975 Negueruela, I. & Marco, A. 2003, A&A, 406, 119 Pflamm-Altenburg, J., & Kroupa, P. 2006, MNRAS, 373, 259 Rieke, G. H., & Lebofsky, M. J. 1985, ApJ, 288, 618 Siess, L., Dufour, E., & Forestini, M. 2000, A&A, 358, 593 Shortridge, K., Meyerdicks, H., Currie, M., et al. 1997, Starlink User Note 86.15, R.A.L Skrutskie, M.F., Cutri, R.M., Stiening, R. 2006, AJ, 131, 1163 Steele, I. A., Negueruela, I., & Clark, J. S. 1999, A&AS, 137, 147 Tapia, M., Costero, R., Echevarr´ıa, J., & Roth, M. 1991, MNRAS, 253, 649 Vallenari, A., Richichi, A., Carraro, G., & Girardi, L. 1999, A&A, 349, 825 Walborn, N.R. 1973, AJ, 78, 1067 Walborn, N.R. 2002, in: Crowther, P.A. (ed.) Hot Star Workshop III: The Earliest Stages of Massive Star Birth. ASP Conference Proceedings, vol. 267, San Francisco, p. 111 Walborn, N.R., & Blades, J.C. 1997, ApJS, 112, 457 Walborn, N.R., & Fitzpatrick, E.L. 1990, PASP, 102, 379 Wegner, W. 1994, MNRAS, 270, 22 Yadav, R.K.S., & Sagar, R. 2001, MNRAS, 328, 370 Zavagno, A., Deharveng, L., Comer´on, F., et al. 2006, A&A, 2001, ApJ, 558, 309 446, 171 Elmegreen, B.G., & Lada, C.J. 1977, ApJ, 214, 725 Fitzsimmons, A. 1993, A&AS, 99, 15 Franco, J., Shore, S.N., & Tenorio-Tagle, G. 1994, ApJ, 436, 795 Gaze, V.F., & Shajn, G.A. 1952, Izv. Krym. Astrofiz. Obs., 9, 52 Girardi, L., Bertelli, G., Bressan, A., et al. 2002, A&A, 391, 195 Hanson, M.M. 2003, ApJ, 597, 957 Herbig, G.H., & Griffin, R.F. 2006, AJ, 132, 1763 Hester, J.J., Scowen, P.A., Sankrit, R., et al. 1996, AJ, 111, 2439 Hillenbrand, L.A., Massey, P., Strom, S.E., & Merrill, K.M. 1993, AJ, 106, 1096 Howarth, I., Murray, J., Mills, D., & Berry, D.S. 1998, Starlink User Note 50.21, R.A.L. Humphreys, R.M., & McElroy, D.B. 1984, ApJ, 284, 565 Indebetouw, R., Mathis, J.S., Babler, B.L., et al. 2005, ApJ, 619, 931 Ishii, M., Nagata, T., Sato, S., et al. 2001, AJ, 121, 3191 Jones, F.S. 1972, PASP, 84, 459 Kroupa, P. 2001, MNRAS, 322, 231 Kohoutek, L., & Wehmeyer, R. 1999, A&AS, 134, 255 Leisawitz, D., Bash, F.N., & Thaddeus, P. 1989, ApJS, 70, 731 Leonard, P.J.T., & Duncan, M.J. 1990, AJ, 99, 608 Ma´ız-Apell´aniz, J. 2004, PASP, 116, 859 Marco, A., Negueruela, I. 2002, A&A, 393, 195 (Paper II) Marco, A., Bernabeu, G., & Negueruela, I. 2001, AJ, 121, 2075 (Paper I) Martins, F., Schaerer, D., & Hillier, D.J. 2005, A&A, 436, 1049 List of Objects 'NGC 1893' on page 1 'Sim 129' on page 1 'Sim 130' on page 1 'IC 410' on page 1 '30 Dor' on page 1 'NGC 6611' on page 2 'NGC 1893' on page 3 'S5003' on page 4 'S3R1N9' on page 4 'S1R2N55' on page 4 'S1R2N26' on page 4 'S3R1N3' on page 4 'S3R1N4' on page 4 'S2R1N26' on page 4 'S2R1N16' on page 4 'S1R2N26' on page 4 'HD 242926' on page 4 'S4R2N17' on page 4 'S3R2N15' on page 4 'S3R1N16' on page 4 'S3R1N5' on page 4 'HD 242908' on page 5 'S1R3N48' on page 5 'S2R3N35' on page 5 'S1R2N14' on page 5 'LS V +33◦23' on page 5 16 I. Negueruela et al.: Pre-main-sequence stars in the young open cluster NGC 1893 'S1R3N35' on page 5 'LS V +33◦26' on page 5 'HD 243018' on page 5 'S2R3N35' on page 5 'S2R4N3' on page 5 'LS V +33◦24' on page 5 'HD 243035' on page 5 'HD 243070' on page 5 'Hoag 7' on page 5 'LS V +33◦27' on page 5 'IRAS 05198+3325' on page 8 'S1R2N35' on page 9 'NX Aur' on page 9 'S1R2N38' on page 9 'S1R2N56' on page 9 '[MJD95] J052307.57+332837.9' on page 9 'S1R2N55' on page 9 '[MJD95] J052308.30+332837.5' on page 9 'S5004' on page 9 '[MJD95] J052306.71+332840.3' on page 9 'S1R2N44' on page 9 'Sim 130' on page 10 'S1R2N56' on page 10 'S1R2N55' on page 10 'IRAS 05198+3325' on page 11 'Sim 130' on page 11
0709.3072
1
0709
2007-09-19T17:14:25
The Distance of the First Overtone RR Lyrae Variables in the MACHO LMC Database: A New Method to Correct for the Effects of Crowding
[ "astro-ph" ]
Previous studies have indicated that many of the RR Lyrae variables in the LMC have properties similar to the ones in the Galactic globular cluster M3. Assuming that the M3 RR Lyrae variables follow the same relationships among period, temperature, amplitude and Fourier phase parameter phi31 as their LMC counterparts, we have used the M3 phi31-logP relation to identify the M3-like unevolved first overtone RR Lyrae variables in 16 fields near the LMC bar. The temperatures of these variables were calculated from the M3 logP-logTe relation so that the extinction could be derived for each star separately. Since blended stars have lower amplitudes for a given period, the period amplitude relation should be a useful tool for identifying which stars are affected by crowding. We find that the low amplitude stars are brighter. We remove them from the sample and derive an LMC distance modulus 18.49+/-0.11.
astro-ph
astro-ph
The Distance of the First Overtone RR Lyrae Variables in the MACHO LMC Database: A New Method to Correct for the Effects of Crowding C. M. Clement, X. Xu and A. V. Muzzin Department of Astronomy and Astrophysics, University of Toronto, Toronto, ON, Canada, M5S 3H8 ABSTRACT Previous studies have indicated that many of the RR Lyrae variables in the LMC have properties similar to the ones in the Galactic globular cluster M3. Assuming that the M3 RR Lyrae variables follow the same relationships among period, temperature, amplitude and Fourier phase parameter φ31 as their LMC counterparts, we have used the M3 φ31 −log P relation to identify the M3-like un- evolved first overtone RR Lyrae variables in 16 MACHO fields near the LMC bar. The temperatures of these variables were calculated from the M3 log P − log Tef f relation so that the extinction could be derived for each star separately. Since blended stars have lower amplitudes for a given period, the period-amplitude rela- tion should be a useful tool for determining which stars are affected by crowding. We find that the low amplitude LMC RR1 stars are brighter than the stars that fit the M3 period-amplitude relation and we estimate that at least 40% of the stars are blended. Simulated data for three of the crowded stars illustrate that an unresolved companion with V ∼ 20.5 could account for the observed amplitude and magnitude. We derive a corrected mean apparent magnitude hV0i = 19.01 ± 0.10 (extinction) ±0.02 (calibration) for the 51 uncrowded un- evolved M3-like RR1 variables. Assuming that the unevolved RR1 variables in M3 have a mean absolute magnitude MV = 0.52 ± 0.02 leads to an LMC distance modulus µ = 18.49 ± 0.11 Subject headings: galaxies: Magellanic clouds - distances and redshifts - stars: variables: other 1. Introduction Variable stars are useful standard candles for determining the distances to nearby galax- ies, but crowding can be a major source of uncertainty in the measurement of these distances. -- 2 -- If the image of a variable is blended with that of another star, its apparent magnitude will be too bright. As a result, its derived distance will be too small. It is not always possible to recognize which stars are affected by the crowding so the problem is usually addressed by taking a statistical approach. Another consequence of image blending is that the am- plitude of light variation is reduced. Thus, if one can determine which variables have low amplitudes for their periods, the blended stars can be flagged. In this paper, we propose a method, based on the period-amplitude relation, for identifying crowded stars in the LMC. We test our method on the 330 RR Lyrae stars in the MACHO database that we classified as bona fide RR11 variables in our previous study (Clement et al. 2002; Alcock et al. 2004, hereafter referred to as A04). Investigations of the LMC RR Lyrae variables by the MACHO collaboration have in- dicated that many of them have characteristics similar to the ones in the Galactic globular cluster M3. In a preliminary study that included 500 RR Lyrae stars, Alcock et al. (1996) calculated that the mean period of the RR0 variables with a V amplitude of 0.8 mag was 0d.552, compared with 0d.480 for M107, 0d.507 for M4, 0d.543 for M3 and M72 and 0d.617 for M15. In a subsequent paper, (Alcock et al. 2000a), a more rigorous selection that included only the least crowded RR0 stars2 in 16 fields near the bar was made. The ridgeline of the period-amplitude relation they plotted for the RR0 variables was similar to the M3 ridgeline based on the data of Kaluzny et al. (1998). The population of double-mode variables in the LMC is another feature that indicates a similarity to M3. More than 150 of the 181 RR01 stars discovered by Alcock et al. (1997, 2000b) have fundamental mode periods between 0d.46 and 0d.50. The only globular clusters known to have RR01 variables with periods in this range are M3 (Corwin et al. 1999, Clementini et al. 2004) and IC 4499 (Clement et al. 1985, Walker & Nemec 1996). In this study, we use Simon's (Simon & Lee 1981) Fourier decomposition technique. Previous investigations have pointed to a relationship among his Fourier phase parameter φ31, and the luminosity and metal abundance. Evidence for this can be seen in Figure 1 which shows φ31 − log P plots for the RR1 variables in four well studied globular clusters: M107, M5, M3 and M68. The diagram illustrates that the RR1 variables in a given cluster show a sequence of φ31 increasing with period and that the lower the cluster metallicity, the more the sequence is shifted to longer periods in the plot. Since the luminosities of RR Lyrae 1We use the system of notation that Alcock et al. (2000b) introduced for RR Lyrae variables: RR0 for fundamental mode, RR1 for first-overtone and RR01 for double-mode (fundamental and first-overtone), instead of the traditional RRab, RRc and RRd. 2The percentage flux inside the point-spread function box contributed by neighboring stars was estimated and only the ∼ 20% least crowded stars were included in their sample. -- 3 -- variables are known to be correlated with metal abundance, a plot like Figure 1 should be useful for identifying a group of RR1 stars with similar luminosities.3 The φ31 − log P plot that A04 made for the LMC indicated that most of the RR1 variables are similar to the ones in M2, M3 and M5. We therefore assume that the luminosities of the LMC RR1 variables are comparable to the ones in these three clusters. Since M3 is the most variable-rich globular cluster and also because CCC recently made a detailed multicolor and Fourier study of its RR Lyrae variables, we will compare the LMC stars with the ones in M3. Our modus operandi will be to use φ31 to identify the M3-like RR1 variables in the LMC, then use the period-amplitude relation to select the ones that are uncrowded and apply the M3 distance modulus to derive their absolute magnitudes. The φ31 parameter is effective for this analysis because it is not altered by crowding. Simon & Clement (1993) performed simulations that added constant light to RR Lyrae light curves and found that φ31 remained unchanged. By taking this approach, we avoid using the RR Lyrae MV − [Fe/H] relation and the problems that arise because of the uncertainty of its slope. In §2.1, we discuss our sample selection. Then in §2.2, we use the CCC period-amplitude relation to identify the crowded stars and perform simulations to ascertain the nature of their unresolved companions. In §2.3, we use CCC's M3 period-temperature relation to calculate the temperature for each LMC star so that the interstellar extinction and corrected magnitude V0 can be derived and in §2.4, we consider the effects of the LMC geometry on the apparent magnitude of the RR Lyrae variables. Finally in §3, we derive an LMC distance. 2. The Analysis 2.1. Identification of M3-like variables in the LMC In their seminal study of the M3 RR Lyrae variables, CCC published photometric and Fourier parameters for 23 variables that they classified as type RRc (RR1). We use their V data, but limit our study to the unevolved stars. Thus we exclude V70, V85, V129, V170 and V177. These five stars have longer periods for a given amplitude than the others and appear to have evolved off the ZAHB. CCC called them "longP /overluminous" stars. Another three 3Simon & Clement (1993) derived equations relating masses, luminosities and temperatures to pulsation period and φ31 based on hydrodynamic pulsation models. Later Catelan (2004) and Cacciari et al. (2005, hereafter referred to as CCC) both demonstrated that the equations should not be applied to individual stars, but CCC also noted that Fourier parameters could be used for estimating the average luminosity of a group of stars, after careful and proper calibration. In this investigation, we do not use the Fourier parameters to determine physical properties for individual stars. -- 4 -- stars, V105, V178 and V203, are excluded because they have short periods (P < 0.29 days) and small amplitudes. They do not fit into the same period-amplitude sequence as the other RR1 variables.4 This leaves 15 'unevolved' RR1 variables in our M3 reference sample. In Figure 2, we plot φ31, AV (the V amplitude), and hV i against log P for these stars. The diagram illustrates that φ31 and AV are both correlated with period, but hV i is not. The central lines in the φ31 and AV −log P plots are least squares fits to the data. The outer lines have the same slope and are envelope lines that encompass all of the data. It is important that none of these 15 stars is affected by crowding. To check this, we verified that there was no correlation between the observed V magnitude and ∆AV , the displacement from the central line in the AV − log P plot of Figure 2. Furthermore, the M3 finding chart published by Bailey (1913) indicates that all of these variables are located outside the cluster core. Therefore it seems reasonable to assume that crowding does not affect our reference sample. Our LMC data are the MACHO data for 330 RR Lyrae variables in 16 fields5 near the LMC bar. The observations were obtained between 1992 and 1999. The 330 stars that we analyse were all classified as bona fide RR1 variables by A04 who published their photometric and Fourier parameters. Figure 3 shows the φ31 −log P plot for these stars, with the envelope lines for M3 superimposed. LMC variables with log P > −0.45 (equivalent to P = 0d.355) have been excluded from the plot because M3 RR1 variables with periods greater than this appear to have evolved. In order to decide which of the LMC stars to include in our sample, we used a weighting scheme that Gladders & Yee (2000) devised for determining whether data points belong to a linear sequence, given their error. We constructed a Gaussian distribution for φ31 of each star, using a HWHM equal to the error listed by A04. Then we determined a 'weight' by measuring the fraction of the area under the Gaussian that fell between the envelope lines of Figure 3. Figure 4 shows the distribution of these weights. For our analysis, we will include only the stars with the highest probablility of fitting the M3 φ31 − log P plot, i.e. the ones with weight greater than 0.5. 4Some of these short period variables could be second overtone (RR2) pulsators. CCC plotted the Fourier parameters φ21 versus A21 for V105, V178 and V203 and concluded that at least V203 is an RR2 variable. 5A chart showing the locations of the MACHO fields is available at http://www.macho.mcmaster.ca -- 5 -- 2.2. The Crowded Stars 2.2.1. Correction for crowding The period-V amplitude relation for the M3-like RR1 variables in the LMC is shown in Figure 5 with the envelope lines from Figure 2 superimposed. Of the 147 stars plotted, 71 lie below the lines, 54 lie between them and 22 are above. If our hypothesis that low amplitudes are caused by blending is correct, we would expect the stars below the lines to be brighter. However, before we proceed to test this hypothesis, we need to consider the stars with the shortest periods (log P < −0.54, i.e. P < 0d.29). They all have low amplitudes and could therefore be the LMC counterparts of the M3 stars, V105, V178 and V203. Since their low amplitudes might be intrinsic, they are not suitable for our crowding test. We therefore exclude all LMC variables with P < 0d.29 from our analysis. This leaves 127 stars: 54 below the lines, 51 between and 22 above. The mean magnitudes for the stars in these three regimes are listed in Table 1. Each row represents a different threshold for the weights of the stars considered. We will base our discussion on all of the stars with weight > 0.5. The data of Table 1 demonstrate that the stars that lie below the M3 period-amplitude relation are brighter than the ones that lie between the lines. This is the result we expect if the low amplitude stars are blended. A t-test (for all the stars with weight > 0.5) indicates that the difference in hV i is highly significant, with a probability of only 0.0056 that the two groups of stars are drawn from the same population. As for the stars that lie above the M3 lines, the difference between their mean hV i and that of the stars between the lines is not significant. Two M3 counterparts to these stars, V85 and V177, are displaced with respect to the other RR1 variables in the period-amplitude relation, but their φ31 and hV i values do not set them apart. CCC classified them with their long period/high luminosity group. We therefore suggest that these high amplitude stars are evolved, even though they do not appear brighter than the others. Thus we can account for the relative mean magnitudes of the stars below, between and above the lines in Figure 5. In this discussion, we have not taken the effect of interstellar extinction into account and we know that it may vary from star to star. However, if the average extinction among the stars in each of the three regimes of Figure 5 is the same, the ranking of their mean hV i values should be correct. Table 1 indicates that the V magnitudes of 54 of 127 M3-like RR1 variables appear to be altered by crowding. This represents approximately 40% of the stars. -- 6 -- 2.2.2. Crowding Simulations If the RR1 stars that lie below the envelope lines in the period-amplitude diagram have unresolved companions, what are the apparent magnitudes of these companions? To answer this question, we performed simulations to ascertain how the presence of an unresolved companion would affect the observed magnitude and amplitude. A few examples of these simulations are presented in Table 2. The simulations show the change in V magnitude and amplitude when an RR1 variable with V = 19.4 or 19.7 mag is blended with a star with V = 20 − 22 mag. This range of V magnitudes was selected for the companions because the LMC color magnitude diagram plotted by Alcock et al. (2000a) shows a high density of main sequence stars with V > 20 mag and therefore the RR Lyrae are probably blended with stars like these. In Table 3, we show simulated data for three stars that are displaced from the central line in Figure 5 by more than 0.1 mag. According to the table, the 'true' magnitudes of these RR Lyrae stars could be V = 19.70, 19.56 and 19.63 while their unresolved companions have V magnitudes, 21.25, 20.28 and 20.50 respectively. These RR Lyrae V magnitudes are typical LMC RR1 magnitudes and the V magnitudes of the companions are consistent with LMC main sequence stars that belong to a younger population. RR Lyrae variables belong to an old population for which the main sequence turn-off is about 3.4 mag fainter than the horizontal branch. Thus older main sequence stars would be too faint to have much of an effect. The MACHO CM diagram also shows a high concentration of RR Lyrae variables and horizontal branch red clump stars with V ∼ 19.2 mag, but we do not expect the stars in our data set to be blended with stars this bright. LMC RR Lyrae variables that have unresolved companions with V ∼ 19.2 mag would be brighter than 19th magnitude. Such stars are known to exist in the MACHO database, but the 147 stars in our sample are fainter. (2005, hereafter referred to as DF05)6 identified five stars that could be blended with red clump stars among the RR Lyrae stars in their sample. These five stars had typical RR Lyrae periods, but small amplitudes and bright mean V (< 19) magnitudes. They concluded that one of these anomalous stars was an RR0 blended with a young main sequence star, but did not assign a definite classification to the other four. In an independent study of LMC RR Lyrae variables, Di Fabrizio et al. 6A major study of RR Lyrae variables near the LMC bar was made by Clementini et al. (2003, hereafter referred to as C03) and DF05. They made photometric observations with the 1.54 m Danish telescope at La Silla, Chile and spectroscopic observations with the 3.6 m ESO telescope and the VLT. They discovered approximately 135 RR Lyrae variables in two fields that overlap with parts of MACHO fields 6 and 13. -- 7 -- 2.3. The Extinction A serious difficulty in deriving the distance to LMC stars is that the amount of inter- stellar extinction is not constant. Schwering & Israel (1991) estimated that the foreground extinction due to dust in the Galaxy ranges from E(B − V ) = 0.07 to 0.17 over the LMC surface. Furthermore Harris et al. (1997) concluded that the distribution of dust within the LMC itself is clumpy. Therefore it is desirable to derive the extinction for the stars individually and we are in a position to do this. Since we have selected M3-like stars for our investigation, we can calculate the temperature of each star individually and then derive its reddening. We use CCC's M3 period-temperature relation: A = 13.353 − 1.19 log P0 − 4.058 log Tef f . (1) CCC found that a value of A = −1.82 ± 0.03 gave the best fit for the unevolved variables, but it predicted temperatures that were too low for stars that had evolved away from the ZAHB. Assuming that A = −1.82 and that log P0 = log P1 + 0.127,7 we derive the following equation for calculating the temperatures of the unevolved LMC stars: log Tef f = 3.702 − 0.293 log P1. (2) The unreddened color (V −R)0 can be computed from a relation derived by Kov´acs & Walker (1999) based on the models of Castelli, Gratton & Kurucz (1997a): log Tef f = 3.8997 − 0.4892(V − R)0 + 0.0113 log g + 0.013[M/H] (3) For this calculation, we assume that [M/H] = −1.3, the value adopted by CCC for their M3 study,8 and log g = 2.93, the mean of the log g values they calculated for the 15 stars in our reference sample. In order to calculate the corrected magnitudes V0, we assume a ratio of total to selective absorption: AV /E(V − R) = 5.35 (4) (Schlegel, Finkbeiner & Davis 1998). 7A typical ratio P1/P0 for the M3 RR01 variables (Clementini et al. 2004) is 0.746. This is equivalent to ∆ log P = 0.127. 8We also note that there is a small, but non-zero metallicity dependence subsumed into CCC's estimation of A. -- 8 -- The hV iF and hRiF magnitudes,9 the V extinction and corrected magnitudes V0 for the 51 uncrowded, unevolved stars are listed in Table 4. Figure 6 shows the distribution of E(V − R) for the stars that we consider to be uncrowded, unevolved M3-like RR1 variables. The mean is 0.073 ±0.0210 which corresponds to V band extinction of 0.39 ±0.10. This value can be compared with the LMC extinction that Zaritsky et al. (2004) derived using a different technique. They measured effective temperatures and line-of-sight extinction for millions of individual stars by comparing stellar atmosphere models with U, B, V, I photometry. Then they constructed extinction maps for stars in two temperature ranges where the model fitting between temperature and extinction was least degenerate: 5500K ≤ TE < 6500K (cool, older stars) and 12, 000K ≤ TE < 45, 000K (hot, younger stars). They derived a mean V absorption of 0.43 mag for the cool stars and 0.55 mag for the hotter stars. With temperatures in the range ∼ 6100 to 7300K, RR Lyrae variables are similar to their cool star group and the mean extinction we have derived agrees with theirs to within our quoted error. However, for the cool stars, Zaritsky et al. found a bimodal distribution which they attributed to the existence of a dust layer. This bimodal structure is not evident in our data, but our sample is several orders of magnitude smaller than theirs. A Shapiro-Wilk W test indicates that our data, plotted in Figure 6, do not deviate significantly from a normal distribution. We can also compare our extinction with the values derived by C03 for their RR Lyrae investigation since the two LMC fields they observed overlap with MACHO fields 6 and 13. They derived E(B − V ) = 0.116 ± 0.017 for their field A and 0.086 ± 0.017 for field B from the colors of the edges of the instability strip. The corresponding total V extinction values are 0.36 ± 0.05 and 0.27 ± 0.05. The mean extinction we calculate for the 9 stars in MACHO field 6 is 0.35 mag and for the 4 stars in field 13, it is 0.38. These are in good agreement with the extinction C03 derived for their field A and a bit high compared to their field B value, but still within the quoted errors. 9Our hV i and hRi values were derived by fitting a 6-order Fourier series of the form: mag = A0 + X Aj cos(jωt + φj ) j=1,6 (5) to the V and R magnitudes for each star, where ω is (2π/period). Thus our mean magnitudes are the A0 values from the fit of equation (5) to the observational data. 10A04 estimated that the error in E(V − R) due to uncertainties in the temperature-color transformation would be ∼ 0.01 mag. In addition, there is an error of ∼ 0.015 which arises from the uncertainty in the value of A in equation (1). Combining these in quadrature leads to an error of 0.018 mag in E(V − R) and 0.10 in the V extinction. -- 9 -- We have pointed out that equation (2) predicts temperatures that are too low for stars that have evolved away from the ZAHB. Thus if the stars above the lines in Figure 5 have evolved, we would expect the extinction derived from equations (2), (3) and (4) to be under- estimated and this would lead to faint V0 values. This is exactly what we see in Table 5 where we list the mean V0 these equations predict. Even though their mean hV i is comparable to the mean for the stars between the lines, their V0 is 0.08 mag fainter. We conclude that this supports our hypothesis that these stars have evolved. We will not include them in the sample of unevolved M3-like RR1 variables we use to determine the LMC distance. For the stars below the lines of Figure 5, equation (2) should be valid for computing temperatures, but if they are blended with other stars, their observed (V −R) colors are not their true ones. Thus the V0 values we derive for these stars individually will also be in error. However, if the mean extinction for these stars is comparable to the mean extinction for the stars between the lines, they will still appear brighter and this is what Table 5 illustrates. 2.4. Tilt Correction and Line-of-Sight Distribution (1983), Storm et al. The distribution of the RR Lyrae population in the LMC has not been well established. Kinematic studies by Freeman et al. (1991) and Schommer et al. (1992) all indicated that the oldest globular clusters belong to a flattened disk-like system with σRV ∼ 28kms−1. There was no evidence for the presence of a halo. It was therefore assumed that the RR Lyrae variables must belong to a disk population. However, van den Bergh (2004) pointed out that the observed radial velocities did not rule out the possibility that the globular clusters formed in a halo. Subsequently, radial velocities of LMC RR Lyrae variables derived by Minniti et al. (2003) and by Borissova et al. (2006) have indicated a larger velocity distribution (σRV = 50 ± 2kms−1) and these authors argue that there is an old and metal poor halo in the LMC. If the RR Lyrae variables belong to a halo population, they should be distributed spherically with respect to the LMC center. On the other hand, if they belong to a disk population, the tilt of its plane with respect to the plane of the sky is an effect that must be considered when deriving the distance. We made tilt corrections based on recent investigations of the LMC geometry by van der Marel & Cioni (2001) and by Nikolaev et al. (2004). Van der Marel & Cioni (2001) analysed the variations in brightness of asymptotic and red giant branch stars in near-IR color magnitude diagrams extracted from the DENIS and 2MASS surveys. They found a sinusoidal variation in apparent magnitude as a function of position angle, which they interpreted to be the result of distance variations because one side of the LMC plane is closer to us than the other. For their analysis, they assumed that the LMC center is located -- 10 -- · 25 and δ0 = −69 ◦ at α0 = 82 ◦ · 5 (van der Marel 2001) and included stars with ρ in the range 2 ◦ · 7 where ρ is the angular distance from the LMC center. They derived an inclination angle i = 34 ◦ · 3. Nikolaev et al. (2004) carried out a similar analysis based on more than 2000 MACHO Cepheids with ρ < 4◦. Assuming α0 = 79 ◦ · 1 and Θ = 151 ◦ · 2 and line-of-nodes position angle Θ = 122 ◦ · 7 ± 6 ◦ · 4 and δ0 = −69 ◦ · 03, they derived i = 30 ◦ · 7 ± 1 ◦ · 5 ± 8 ◦ · 5−6 ◦ · 0 ± 2 ◦ · 4. We used the equations listed in §2 of van der Marel & Cioni's paper to calculate corrected V0 magnitudes for both inclinations and they are listed in columns (7) and (8) of Table 4. Figure 7 shows the distribution of V0 based on the three different assumptions for the LMC tilt. All three show a peak at approximately 19.0 mag. Some of the dispersion in V0 is due to depth within the LMC, but we expect a dispersion in absolute magnitude as well because our M3 reference stars have a range of 0.2 in apparent magnitude.11 The normal parameter estimates are listed in Table 6. A Shapiro-Wilk W test indicates that none of the three deviates significantly from a normal distribution. We adopt the tilt corrections based on the viewing angles derived by Nikolaev et al. because the stars in their sample, like ours, are all within 4◦ of the LMC center. The mean V0 based on these viewing angles is exactly the same as the value obtained when no tilt correction is applied. Therefore our derived V0 does not depend on any assumption about the distribution of the LMC RR Lyrae variables. 3. The LMC Distance 3.1. The Apparent Magnitude of the RR1 Variables We have adopted a mean V0 = 19.02 for the 51 uncrowded, unevolved M3-like RR1 variables, based on the LMC viewing angles derived by Nikolaev et al. (2004). Since our mean magnitudes are the A0 values from equation (5), we need to convert them to intensity means before we compute an LMC distance. A04 showed that the intensity means are ∼ 0.01 mag brighter than A0 so we revise our adopted mean V0 to 19.01 mag. To determine the precision of the V0 values listed in Table 4, we need to consider the errors in reddening and the errors in the calibration of our photometry. We have already pointed out in §2.3 that the error in E(V − R) is 0.018 mag which corresponds to 0.10 mag 11The error in V0 due to the error in extinction is ±0.10 mag. However, this is a systematic error and should not affect the shape of the V0 distributions plotted in Figure 7. -- 11 -- in the V exinction. Our calibrated V and R magnitudes are from the study by A04, and were calculated from transformation equations derived by Alcock et al. (1999), designated calibration version 990318. They derived an internal precision of σV = 0.021 for their V magnitudes by comparing the results for stars in overlapping (MACHO) fields. Another way to test our calibration is to compare with the results from other inves- tigations. Two of the 51 uncrowded, unevolved M3-like RR1 stars that we have listed in Table 4 were observed by DF05. The mean (intensity) magnitudes derived from the two studies for these stars are listed in Table 7 and they agree to within 0.01 mag. In our earlier paper (A04), we listed mean magnitudes for five additional RR1 stars that were observed by both groups and the MACHO magnitudes were significantly brighter. However, in our new analysis, we have classified two of these stars as crowded. Their mean magnitudes are also listed in Table 7 and it is clear that the MACHO magnitudes are brighter than the ones derived by DF05. The remaining three are not M3-like so we can not ascertain whether or not they are crowded. DF05 also compared their V magnitudes with MACHO data provided by Alves and by Kov´acs (Alcock et al 2003) and found that the MACHO magnitudes were approximately 0.04 mag brighter on average. DF05 concluded that the difference occurred because the MACHO reduction procedure did not adequately compensate for crowding in stars with V > 18.25. The MACHO collaboration recognized the crowding problem so A04 made crowding corrections by adding artifical stars of known magnitude to the the image frames and measuring ∆V the difference between their input and recovered magnitudes. However, these corrections introduce a relatively large error of ∼ 0.10 mag into the adopted apparent magnitudes. In the present investigation, we have dealt with this problem by identifying crowded stars and removing them from our data set. We can not determine which of the MACHO stars in the Alves and Kov´acs samples were crowded; however, our sample of M3-like stars can provide some insight. According to the data listed in Table 1, approximately 40% of the M3-like variables are crowded and as a consequence, their average V magnitude is 0.09 mag brighter. If we assume that the mean MACHO magnitude for 40% of the stars in the data sets that Alves and Kov´acs provided to DF05 is 0.09 mag brighter (as a result of blending with main sequence stars), while the remaining 60% have V magnitudes similar to the ones that DF05 derived, we would expect the MACHO stars to be about 0.04 mag brighter on average. Thus if the crowded stars could be removed from the MACHO sample, the photometry derived in these two independent studies would be in good agreement. Since our V photometry for uncrowded stars appears to be in good agreement with that of DF05, we assume that the uncertainty in our calibrated V magnitudes is 0.02 mag. -- 12 -- Therefore we adopt V0 = 19.01 ± 0.10 (extinction) ±0.02 (calibration) for our sample of uncrowded, unevolved M3-like variables in the LMC. This is in good agreement with the result of C03: V0 = 19.064 ± 0.064 at [Fe/H]= −1.5 on the metallicity scale of Harris (1996). Taking into account the fact that the RR1 variables in our sample are M3-like, that [Fe/H] for M3 is −1.57 on the Harris scale and that C03 derived ∆MV /∆[Fe/H] = 0.214 for the RR Lyrae variables in the LMC, we estimate that V0 = 19.05 at [Fe/H]= −1.57 for the data of C03. 3.2. The Absolute Magnitude and Distance of the RR1 Variables A major challenge in determining the LMC distance is to identify a homogeneous group of stars for which the absolute magnitude is well established. We deal with this problem by using the M3 distance modulus to calculate the absolute V magnitude of our 15 M3 reference stars. Then we assume that the uncrowded stars in our LMC sample have the same mean MV . The M3 distance modulus has been derived in investigations of RR Lyrae variables by Marconi et al. (2003) and by Sollima et al. (2006). Taking a theoretical approach, Marconi et al. applied pulsation theory to the BV observations of M3 by Corwin & Carney (2001) and the K observations of Longmore et al. (1990). They compared the results from pulsation theory with the observed edges of the instability strip, the observed K band period-magnitude relation and the observed relations among period-magnitude-color and period-magnitude-amplitude. For their calculations, they used bolometric corrections and temperature-color transformations provided by Castelli, Gratton & Kurucz (1997a, 1997b) and adopted a mean RR Lyrae mass of 0.67M⊙, based on evolutionary models of Cassisi et al. (2004). They computed a mean distance modulus DM = 15.07 ± 0.05,12 but pointed out that if they had used the models of Vandenberg et al. (2000) instead, DM would have been 15.05. We adopt the distance modulus based on the Cassisi models, DM=15.07±0.05. Our 15 M3 reference stars have V magnitudes ranging from 15.48 to 15.68 with mean hV i = 15.60. If we apply CCC's extinction, E(B −V ) = 0.01±0.01, this corresponds to V0 = 15.57 which leads to MV = 0.50 ± 0.06. Sollima et al. derived an M3 distance modulus from the RR Lyrae period-metallicity-K band luminosity (P LK) relation that they calibrated from observations: MK = −2.38(±0.04) log PF + 0.08(±0.11)[Fe/H] − 1.05(±0.13) (6) 12The evolutionary distance modulus that Marconi et al. derived was about 0.08 ± 0.05 longer than the pulsational value. However, they stated that it would be shorter if element diffusion were properly taken into account because the luminosity of HB models would be about 0.03 − 0.04 mag fainter. -- 13 -- where MK is the absolute K magnitude, PF is the fundamentalized pulsation period and [Fe/H] refers to the metallicity scale of Carretta & Gratton (1997). They derived their coefficient of log PF from the slope of the K − log PF relation for 538 RR Lyrae variables in 16 globular clusters and their [Fe/H] coefficient from the slope of the (MK − 2.38 log P ) - [Fe/H] relation for the four globular clusters in their sample that had distance determinations based on Hipparcos trig parallaxes for local subdwarfs. They obtained their zero point from the K magnitude of RR Lyrae combined with the trig parallax that Benedict et al. (2002) measured for it from HST astrometry. From equation (6), Sollima et al. calculated an M3 distance modulus of 15.07. Thus MV = 0.50 ± 0.20. Unfortunately, the large error in the coefficient of [Fe/H] in equation (6) results in a large uncertainty in MK. Therefore it seems appropriate to calculate the M3 distance modulus directly from the absolute magnitude that Benedict et al. derived for RR Lyrae: MV = 0.61−0.11 +0.10. The [Fe/H] for RR Lyrae is −1.39 (Clementini et al. 1995), which is comparable to the M3 metal abundance, −1.34 on the scale of Carretta & Gratton. Furthermore, the pulsation period of RR Lyrae, 0d.567, and its maximum V amplitude13 which is 0.9 mag according to Smith et al. (2003) places it on the M3 period-amplitude relation. CCC studied five RR0 stars with periods within 0.01 days of 0d.567 (V10, V69, V135, V137 and V142). The mean V amplitude for these stars is 0.89 mag and their mean hV i is 15.65±0.02. Assuming that their mean absolute magnitude is the same as that of RR Lyrae, we derive a mean MV = 0.56 ± 0.11 for our 15 M3 reference stars. The Baade-Wesselink technique has not been applied to any RR Lyrae variables in M3. However, in a review of RR Lyrae luminosities, Cacciari (2003) reported the result of a B-W analysis for the star RR Cet which has a metal abundance comparable to that of M3. Cacciari et al. (2000) derived MV = 0.56 ± 0.15 for this star which, with a period of 0d.553 and V amplitude 0.98 mag (Simon & Teays 1982) fits the M3 P-A relation. Six stars analysed by CCC (V36, V40, V71, V89, V133 and V149) have periods within 0.01 days of 0d.553 and the mean V amplitude and mean hV i they derived for these stars were 0.99 and 15.65 respectively. By comparing the mean magnitudes of these six stars with those of our 15 M3 reference stars, we derive mean MV = 0.52 ± 0.15 for the reference stars. The mean of our four MV values is 0.52 with a standard deviation 0.02 mag. Combining this with the V0 = 19.01 ± 0.10 (extinction error) ±0.02 (calibration error) that we derived for the uncrowded, unevolved M3-like RR1 variables in our LMC sample and adding the estimated errors in quadrature, leads to distance modulus µLM C = 18.49 ± 0.11. 13RR Lyrae exhibits the Blazhko effect, but according to Szeidl (1988), the maximum light amplitude for a Blazhko star fits the period amplitude relation for singly periodic variables. -- 14 -- Our distance modulus is in good agreement with µ0 = 18.48 ± 0.08 derived by Borissova et al. (2004) from K-band photometry of 37 RR Lyrae variables in the inner regions of the LMC. In their investigation, they derived a mean hKi = 18.20 and assumed that the mean K band absorption AK = 0.05 mag. By following the procedure described by Bono et al. (2001, 2003), they calculated MK = −0.332 at log P = −0.30. Their adopted absolute K magnitude was 0.85 mag brighter than our adopted MV . This is consistent with the apparent magnitudes we have derived; their mean K0 (18.15) is 0.86 mag brighter than our V0, 19.01. Our adopted µ0 is also comparable to the value (18.48) obtained by McNamara et al. (2007) in their recent analysis of δ Scuti stars. By identifying the crowded stars and removing them from the sample, we have avoided using the crowding corrections that introduced an additional uncertainty of 0.10 mag to the distance modulus we derived in our earlier study (A04). The main source of error in this investigation is the error in estimating the V extinction. 4. Summary We have devised a new method for identifying crowded RR1 variable stars in the LMC, based on simulations that show that stars with unresolved companions have low amplitudes for their periods. Given that many LMC RR Lyrae variables have properties similar to the ones in the Galactic globular cluster M3, we used the M3 φ31 − log P relation to identify the M3-like unevolved RR1 variables in our LMC sample. The Fourier phase parameter φ31 is useful for selecting a homogeneous sample because it is not affected by crowding. When the M3-like variables were plotted on the M3 period-amplitude diagram, we found that the mean V magnitude of the LMC stars with low amplitudes was 0.09 mag brighter than the mean for the stars that fit the M3 period-amplitude relation. Four of the stars in our sample were observed in the study of LMC RR Lyrae variables by DF05. Comparing the photometry from the two studies, we found that our V magnitudes for the two stars considered to be uncrowded agreed to within 0.01 mag, while the MACHO V magnitudes for the two stars we considered to be crowded were more than 0.05 mag brighter. From this, we conclude that our method for identifying crowded RR Lyrae variables is effective. It could prove to be useful for identifying crowded RR Lyrae variables in other local group galaxies. We used the M3 period-temperature relation for unevolved RR Lyrae variables to de- termine the temperature and reddening for each of the uncrowded RR1 variables in our sample. After making corrections for the tilt of the LMC, we derived a mean V0 magnitude -- 15 -- of 19.01 ± 0.10 (extinction) ±0.02 (calibration). Then to estimate the absolute magnitude, we used the M3 distance modulus and the trig parallax of RR Lyrae to derive the mean absolute magnitude of the unevolved RR1 variables in our M3 reference sample. This turned out to be MV = 0.52 ± 0.02. Finally, we derived an LMC distance modulus µLM C = 18.49 ± 0.11 which is in good agreement with the results of other recent studies and with 18.5 mag, the value employed by the Hubble Space Telescope's key project for measuring the Hubble constant (Freedman et al. 2001). We thank David Clement and Doug Welch for their helpful comments during the prepa- ration of this manuscript. We also express our gratitude to our referee, Bruce Carney, who made several suggestions that have improved the paper. In addition, financial support from Science and Engineering Research Canada (NSERC) is gratefully acknowledged. -- 16 -- REFERENCES Alcock, C. et al. 1996, AJ, 111, 1146 Alcock, C. et al. 1997, ApJ, 482, 89 Alcock, C. et al. 1999, PASP, 111, 1539 Alcock, C. et al. 2000a, AJ, 119, 2194 Alcock, C. et al. 2000b, ApJ, 542, 257 Alcock, C. et al. 2003, ApJ, 598, 597 Alcock, C. et al. 2004, AJ, 127, 334 (A04) Bailey, S. I. 1913, Harvard Ann. 78, 1 Benedict, G. F. et al. 2002, AJ, 123, 473 Bono, G., Caputo, F., Castellani, V. Marconi, M. & Storm, J. 2001, MNRAS, 326, 1183 Bono, G., Caputo, F., Castellani, V. Marconi, M., Storm, J. & Degl'Innocenti, S. 2003, MNRAS, 344, 1097 Borissova, J., Minniti, D., Rejkuba, M., Alves, D., Cook, K. H., & Freeman, K. C. 2004 A&A, 423, 97 Borissova, J., Minniti, D., Rejkuba, M., Alves, D. 2006 A&A, 460, 459 Cacciari, C. 2003, in New Horizons in Globular Cluster Astronomy, ed. G. Piotto, G. Meylan, S. G. Djorgovsli & M. Riello (ASP Conf. Ser. 296) (San Francisco: ASP), 329 Cacciari, C., Clementini, G., Castelli, F. & Melandri, F. 2000, in IAU Colloq. 176, The Impact of Large-Scale Surveys on Pulsating Star Research, ed. L. Szabados & D. W. Kurtz (ASP Conf. Ser. 203) (San Francisco: ASP), 176 Cacciari, C., Corwin, T. M. & Carney, B. W. 2005, AJ, 129, 267 (CCC) Carretta, E. & Gratton, R. G. 1997, A&AS, 121, 95 Cassisi, S., Castellani, M., Caputo, F. & Castellani, V. 2004, A&A, 426, 641 Castelli, F., Gratton, R. G. & Kurucz, R. L. 1997a, A&A, 318, 841 Castelli, F., Gratton, R. G. & Kurucz, R. L. 1997b, A&A, 324, 432 -- 17 -- Catelan, M. 2004, in IAU Colloq. 193, Variable Stars in the Local Group, ed. D. W. Kurtz & K. R. Pollard (ASP Conf. Ser. 310) (San Francisco: ASP), 113 Clement, C. M., Muzzin, A. V., Rowe, J. F. & the MACHO Collaboration. 2002, BAAS, 34, 651 Clement, C. M., Nemec, J. M., Robert, N., Wells, T., Dickens, R. J. & Bingham, E. A. 1985, AJ, 92, 825 Clement, C. M. & Shelton, I. 1997, AJ, 113, 1711 Clementini, G., Carretta, E., Gratton, R. G., Merighi, R., Mould, J. R. & McCarthy, J.K. 1995, AJ, 110, 2319 Clementini, G., Corwin, T. M., Carney, B. W. & Sumerel, A. N. 2004, AJ, 127, 938 Clementini, G., Gratton, R. G., Bragaglia, A., Carretta, E., Di Fabrizio, L. & Maio, M. 2003, AJ, 125, 1309 (C03) Corwin, T. M. & Carney, B.W. 2001, AJ, 122, 3183 Corwin, T. M., Carney, B.W. & Allen, D. M. 1999, AJ, 117, 1332 Di Fabrizio, L., Clementini, G., Maio, M., Bragaglia, A., Carretta, E., Gratton, R., Monte- griffo, P. & Zoccali, M. 2005, A&A, 430, 603 (DF05) Freedman, W. L. et al. 2001, ApJ, 553, 47 Freeman, K. C., Illingworth, G., & Oemler, A. Jr. 1983, ApJ, 272, 488 Gladders, M. D. & Yee, H. K. C. 2000, AJ, 120, 2148 Harris, J., Zaritsky, D., & Thompson, I. 1997, AJ, 114, 1933 Harris, W. E. 1996, AJ, 112, 1487 Ka luzny, J., Hilditch, R. W., Clement, C., & Ruci´nski, S. M. 1998, MNRAS, 296, 347 Ka luzny, J., Olech, A., Thompson, I., Pych, W., Krzemi´nski, W., & Schwarzenberg-Czerny, A. 2000, A&AS, 143, 215 Kov´acs, G. & Walker, A. R. 1999, ApJ, 512, 271 Kraft, R. P. & Ivans, I.I. 2003, PASP, 115, 143 -- 18 -- Longmore, A. J., Dixon, R., Skillen, I., Jameson, R. F., & Fernley, J. A. 1990, MNRAS, 247, 684 McNamara, D. H., Clementini, G. & Marconi, M. 2007, AJ, 133, 2752 Marconi, M., Caputo, F., Di Criscienzo, M. & Castellani, M. 2003, ApJ, 596, 299 Minniti, D., Borissova, J., Rejkuba, M., Alves, D. R., Cook, K. H., & Freeman, K.C. 2003, Science, 301, 1508 Nikolaev, S., Drake, A. J., Keller, S. C., Cook, K. H., Dalal, N., Griest, K., Welch, D. L., & Kanbur, S. M. 2004, ApJ, 601, 260 Schlegel, D. J., Finkbeiner, D. P. & Davis, M. 1998, ApJ, 500, 525 Schommer, R. A., Olszewski, E. W., Suntzeff, N. B. & Harris, H. C. 1992, AJ, 103, 447 Schwering, P. B. W. & Israel, F. P. 1991, A&A, 246, 231 Simon, N. R. & Clement, C. M. 1993, ApJ, 410, 526 Simon, N. R. & Lee, A. S. 1981, ApJ, 248, 291 Simon, N. R. & Teays, T. J. 1982, ApJ, 261, 586 Sollima, A., Cacciari, C. & Valenti, E. 2006, MNRAS, 372, 1675 Smith, H. A. et al. 2003, PASP, 115, 43 Storm, J., Carney, B. W., Freedman, W. L., & Madore, B. F. 1991, PASP, 103, 261 Szeidl, B. 1988, in Multimode Stellar Pulsations, ed. G. Kov´acs, L. Szabados, & B. Szeidl (Budapest: Konkoly Obs.), 45 van den Bergh, S. 2004, AJ, 127, 1897 Vandenberg, D. A., Swenson, F. J., Iglesias, C. A., & Alexander, D. R. 2000, ApJ, 532, 430 van der Marel, R. P. 2001, AJ, 122, 1827 van der Marel, R. P. & Cioni, M. L. 2001, AJ, 122, 1807 Walker, A. R. 1994, AJ, 108, 555 Walker, A. R. & Nemec, J. M. 1996, AJ, 112, 2026 -- 19 -- Zaritsky, D., Harris, J., Thompson, I. B. & Grebel, E. K. 2004, AJ, 128, 1606 This preprint was prepared with the AAS LATEX macros v5.2. -- 20 -- Fig. 1. -- Plot of φ31 vs. log P for the RR1 variables in four well studied Galactic globular clusters: the Oosterhoff type I clusters M107 (open circles), M5 (solid circles) and M3 (open triangles) and the Oosterhoff type II cluster M68 (solid triangles). The φ31 values plotted here for these four clusters were determined by Clement & Shelton (1997), Kaluzny et al. (2000), Cacciari et al (2005) and Walker (1994) respectively. Their metal abundances are −1.10, −1.32, −1.50 and −2.43 on the FeII metallicity scale of Kraft & Ivans (2003). -- 21 -- Fig. 2. -- Plots of φ31, V amplitude and the (intensity) mean V magnitude versus log P for the 15 M3 unevolved RR1 variables that we include in our sample. The data are taken from the investigation by CCC. In the two upper panels, the central lines represent least squares fits to the data and the outer lines, plotted with the same slope, are the envelope lines that encompass all of the data. -- 22 -- Fig. 3. -- A plot of the Fourier phase difference φ31 vs log P for 330 bona fide RR1 variables in 16 MACHO fields in the LMC. The superimposed lines represent the φ31 − log P relation for the M3 RR1 variables studied by CCC. (See the upper panel of Figure 2.) -- 23 -- Fig. 4. -- The distribution of 'weights' for the LMC RR1 variables in our sample. The weight is a measure of the probability that a star lies between the envelope lines of Figure 3. Our procedure for determining these weights is described in §2.1. -- 24 -- Fig. 5. -- The period-V amplitude relation for the M3-like RR1 variables in 16 MACHO fields in the LMC. Only the 147 stars with weight greater than 0.5 in Figure 4 are included. The superimposed lines represent the period-amplitude relation for the M3 RR1 variables (see the centre panel of Figure 2). -- 25 -- Fig. 6. -- A histogram of E(V-R) for the 51 uncrowded, unevolved M3-like RR1 variables in 16 MACHO fields in the LMC. These are the stars that lie between the envelope lines of Figure 5. -- 26 -- Fig. 7. -- A histogram for the LMC variables that lie between the lines in Figure 5. These are the stars that we consider to be uncrowded M3-like unevolved RR1 variables. Three histograms are plotted: the first represents the data with no tilt correction and the other two include the tilt corrections of van der Marel & Cioni (2001) and Nikolaev et al. (2004) respectively. The curves are Gaussian fits to the data. -- 27 -- Table 1. Mean hV i and Location in the Period-Amplitude Diagram Weight Mean hV i N N Mean hV i N Mean hV i (between) (3) (4) (5) (1) (below) (2) (above) (6) Wt. ≥ 0.5 Wt. ≥ 0.6 Wt. ≥ 0.7 Wt. ≥ 0.8 Wt. ≥ 0.9 19.31 ± 0.02 19.30 ± 0.02 19.28 ± 0.03 19.32 ± 0.03 19.31 ± 0.05 54 43 31 17 9 19.40 ± 0.02 19.40 ± 0.02 19.38 ± 0.02 19.35 ± 0.03 19.39 ± 0.03 51 33 25 13 6 19.43 ± 0.04 19.45 ± 0.05 19.37 ± 0.05 19.45 ± 0.06 19.40 ± 0.06 (7) 22 16 8 3 2 Note. -- The terms 'below', 'between' and 'above' refer to the location relative to the lines in Figure 5. The weights represent the probability that the stars lie between the envelope lines in Figure 3. Stars with P <0d.29 are not included. -- 28 -- Table 2. Crowding Simulations V Amplitude V V (RR) (1) (RR) (2) 19.4 19.4 19.4 19.4 19.4 19.7 19.7 19.7 19.7 19.7 0.45 0.45 0.45 0.45 0.45 0.45 0.45 0.45 0.45 0.45 (Companion) (Observed) (3) 20.0 20.5 21.0 21.5 22.0 20.0 20.5 21.0 21.5 22.0 (4) 18.90 19.06 19.17 19.25 19.30 19.08 19.27 19.41 19.51 19.57 Amplitude (Observed) (5) 0.29 0.33 0.37 0.39 0.41 0.26 0.30 0.35 0.38 0.40 Note. -- The simulated magnitudes listed in columns (1) and (3) combine to produce the magnitudes listed in column (4). Thus if an RR Lyrae variable with V = 19.4 has an unresolved companion with V = 20.0, its observed magnitude will be V = 18.90. As a result of the blending, its original amplitude (0.45 mag) will be reduced to 0.29 mag. -- 29 -- Table 3. Sample Magnitudes for Three Blended Stars Star log P (1) (2) 80.6475.3548 80.6708.6879 81.8398.799 -0.486 -0.503 -0.486 Amplitude (observed) (3) 0.36 0.30 0.30 hV i hV i Amplitude (observed) (RR + companion) (4) 19.46 19.10 19.22 (5) 19.70+21.25 19.56+20.28 19.63+20.50 (true) (6) 0.45 0.46 0.44 Note. -- The simulated magnitudes listed in column (5) combine to produce the observed magnitudes of column (4). For example, our simulation implies that star 80.6475.3548 is really an RR Lyrae with V = 19.70 blended with an unresolved com- panion whose V = 21.25 mag. As a result of the blending, the original V amplitudes listed in column (6) are reduced to the observed amplitudes listed in column (3). Table 4. Parameters of the Uncrowded, Unevolved M3-like RR1 Variables Star (1) Period (2) hV iF (3) hRiF (4) 2.4789.946 2.5150.896 2.5151.982 2.5269.422 2.5511.772 2.5633.1369 0.326910 0.337123 0.344523 0.318123 0.302923 0.293763 19.23 19.21 19.33 19.52 19.33 19.30 19.05 19.01 19.10 19.29 19.17 19.13 Ext(V ) (5) 0.18 0.24 0.37 0.48 0.18 0.27 V0 (6) 19.05 18.97 18.96 19.04 19.15 19.03 V0(vdM) V0(Nik) Weight (7) (8) 19.05 18.97 18.96 19.03 19.15 19.03 19.04 18.96 18.95 19.03 19.15 19.03 (9) 0.89 0.59 0.63 0.90 0.93 0.71 Note. -- The V0 values are corrected for extinction; the V0(vdM) and V0(Nik) values are corrected for tilt as well, according to the LMC viewing angles derived by van der Marel & Cioni (2001) and by Nikolaev et al. (2004) respectively. The weight is the probability that the star lies between the envelope lines of Figure 3. Table 4 is presented in its entirety in the electronic edition of the Astronomical Journal. A portion is shown here for guidance regarding its form and content. -- 30 -- Table 5. Mean V0 for Stars with Wt. > 0.5 Location in Mean V0 Fig 5 Below the lines Between the lines Above the lines 18.88 ± 0.02 19.02 ± 0.02 19.10 ± 0.05 N 54 51 22 Note. -- The V0 values have been de- rived under the assumption that the stars' temperatures can be computed from equa- tion (2) and that their observed (V − R) colors are correct. These two assumptions are valid for the stars that lie between the lines, but not for the others. This is dis- cussed in the final paragraph of §2.3. Table 6. Fitted Normal Parameter Estimates for Fig 7 Tilt correction µ σ N None van der Marel Nikolaev 19.02 19.00 19.02 0.16 0.15 0.16 51 51 51 Table 7. Comparison with the DF05 Photometry ID (MACHO) Crowded? (MACHO) hV iint ID (MACHO) (DF05) hV iint (DF05) 6.7054.710 13.5838.667 6.6689.563 13.6079.604 no no yes yes 19.45 19.39 19.23 19.24 7864 7648 2249 4749 19.46 19.38 19.37 19.31
astro-ph/0610086
1
0610
2006-10-03T17:51:00
Detection of the Irradiated Donor in the LMXBs 4U 1636-536 (=V801 Ara) and 4U 1735-444 (=V926 Sco)
[ "astro-ph" ]
Phase-resolved VLT spectroscopy of the bursting Low Mass X-ray Binaries 4U 1636-536/V801 Ara and 4U 1735-444/V926 Sco is presented. Doppler images of the NIII 4640 Bowen transition reveal compact spots which we attribute to fluorescent emission from the donor star and enable us to define a new set of spectroscopic ephemerides. We measure Kem=277+-22 km/s and Kem=226+-22 km/s from the NIII spots in V801 Ara and V926 Sco respectively which represent strict lower limits to the radial velocity semi-amplitude of the donor stars. Our new ephemerides provide confirmation that lightcurve maxima in V801 Ara and likely V926 Sco occur at superior conjunction of the donor star and hence photometric modulation is caused by the visibility of the X-ray heated donor. The velocities of HeII 4686 and the broad Bowen blend are strongly modulated with the orbital period, with phasing supporting emission dominated by the disc bulge. In addition, a reanalysis of burst oscillations in V801 Ara, using our spectroscopic T0, leads to K1=90-113 km/s. We also estimate the K-corrections for all possible disc flaring angles and present the first dynamical constraints on the masses of these X-ray bursters. These are K2=360+-74 km/s, f(M)=0.76+-0.47 Msun and q=0.21-0.34 for V801 Ara and K2=298+-83 km/s, f(M)=0.53+-0.44 Msun and q=0.05-0.41 for V926 Sco. Disc flaring angles alpha>12 deg and q~0.26-0.34 are favoured for V801 Ara whereas the lack of K1 constraint for V926 Sco prevents tight constraints on this system. Although both binaries seem to have intermediate inclinations, the larger equivalent width of the narrrow NIII line in V801 Ara at phase 0.5 relative to phase 0 suggests that it has the higher inclination of the two.
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 000–000 (0000) Printed 29 January 2018 (MN LATEX style file v2.2) Detection of the Irradiated Donor in the LMXBs 4U 1636-536 (=V801 Ara) and 4U 1735-444 (=V926 Sco) J. Casares1⋆, R. Cornelisse1,2⋆, D. Steeghs3⋆, P.A. Charles2,4⋆, R.I. Hynes5⋆, K. O'Brien6⋆, T. E. Strohmayer7⋆ 1 Instituto de Astrof´ısica de Canarias, 38200 La Laguna, Tenerife, Spain 2 School of Physics & Astronomy, University of Southampton, Southampton, SO17 1BJ, UK 3 Harvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138, USA 4 South African Astronomical Observatory, P.O. Box 9. Observatory 7935, South Africa 5 Department of Physics and Astronomy, 202 Nicholson Hall, Lousiana State, Baton Rouge, L.A. 70803, USA 6 European Southern Observatory, Casilla 19001, Santiago 19, Chile 7 Laboratory for High Energy Astrophysics, NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA 29 January 2018 ABSTRACT Phase-resolved VLT spectroscopy of the bursting Low Mass X-ray Binaries 4U 1636- 536/V801 Ara and 4U 1735-444/V926 Sco is presented. Doppler images of the NIII λ4640 Bowen transition reveal compact spots which we attribute to fluorescent emission from the donor star and enable us to define a new set of spectroscopic ephemerides. We measure Kem = 277 ± 22 km s−1 and Kem = 226 ± 22 km s−1 from the NIII spots in V801 Ara and V926 Sco respectively which represent strict lower limits to the radial velocity semi- amplitude of the donor stars. Our new ephemerides provide confirmation that lightcurve max- ima in V801 Ara and likely V926 Sco occur at superior conjunction of the donor star and hence photometric modulation is caused by the visibility of the X-ray heated donor. The ve- locities of HeII λ4686 and the broad Bowen blend are strongly modulated with the orbital period, with phasing supporting emission dominated by the disc bulge. In addition, a reanaly- sis of burst oscillations in V801 Ara, using our spectroscopic T0, leads to K1 = 90 − 113 km s−1. We also estimate the K − corrections for all possible disc flaring angles and present the first dynamical constraints on the masses of these X-ray bursters. These are K2 = 360 ± 74 km s−1, f (M ) = 0.76 ± 0.47 M⊙ and q = 0.21 − 0.34 for V801 Ara and K2 = 298 ± 83 km s−1, f (M ) = 0.53 ± 0.44 M⊙ and q = 0.05 − 0.41 for V926 Sco. Disc flaring angles α ≥ 12◦ and q ≃ 0.26 − 0.34 are favoured for V801 Ara whereas the lack of K1 constraint for V926 Sco prevents tight constraints on this system. Although both binaries seem to have intermediate inclinations, the larger equivalent width of the narrrow NIII line in V801 Ara at phase 0.5 relative to phase 0 suggests that it has the higher inclination of the two. Key words: stars: accretion, accretion discs – binaries: close – stars: individual: (V801 Ara; V926 Sco)– X-rays: binaries – 6 0 0 2 t c O 3 1 v 6 8 0 0 1 6 0 / h p - o r t s a : v i X r a 1 INTRODUCTION Low mass X-ray binaries (LMXBs) are interacting binaries where a low mass donor transfers matter onto a neutron star or black hole. 4U 1636-536 (=V801 Ara) and 4U 1735-444 (=V926 Sco) are among the optically brighter members in the class of persistent LMXBs, characterized by Lx ≃ 1037−38 erg s−1 and blue spectra with weak high-excitation emission lines (mainly HeII λ4686 and [email protected] (JC); ⋆ E-mail: (RC); (PAC); [email protected] (TES) [email protected] [email protected] (RIH); [email protected] [email protected] (DS); (KOB); [email protected] the CIII/NIII Bowen blend at λ4640). They also share similar prop- erties: they are both atoll sources (as based on the pattern traced in X-ray color-color diagrams, see e.g.Hasinger & van der Klis 1989) with frequent burst activity and short orbital periods (3.80 and 4.65 hrs respectively) revealed through optical photometry (Corbet et al. 1986, Pedersen et al. 1981). Their lightcurves display shallow si- nusoidal modulations which have been interpreted as due to the geometrically varying visibility of the irradiated donor star (e.g. van Paradijs et al. 1988). Therefore, the photometric maxima sup- posedly define orbital phase 0.5 i.e. inferior conjunction of the compact object. Note, however, that this assumption requires con- firmation because photometric maxima can sometimes be associ- ated with asymmetries in the disc structure such as the visibility of 2 J. Casares et al. the irradiated inner disc bulge at phase ∼ 0.3 (e.g. 4U 1822-371 Hellier & Mason 1989) or superhump activity (see Haswell et al. 2001). Only a few spectroscopic studies have been presented on these two binaries up to now. For example, Smale & Corbet (1991) re- port Hα spectroscopy of V926 Sco showing that the line core is dominated by emission from the disc bulge or splash region where the gas stream interacts with the outer disc rim. On the other hand, Augusteijn et al. (1998) (A98 hereafter) present radial veloc- ity curves of HeII λ4686 and the Bowen blend at λ4640 for both V926 Sco and V801 Ara. They conclude that these high excitation lines are also tracing the motion of the disc bulge. V801 Ara is particularly remarkable since it is one of only 14 bursters where "burst oscillations" (i.e. nearly coherent high- frequency pulsations) have been detected during several thermonu- clear X-ray bursts (Giles et al. 2002, G02 hereafter). Furthermore, a train of burst oscillations was also discovered in a 13 min in- terval during a "superburst", showing a frequency drift which has been interpreted as the orbital motion of the neutron star (Strohmayer & Markwardt 2002, SM02 hereafter). By fitting the frequency evolution with a circular orbit SM02 constrain the radial velocity amplitude of the neutron star to the range 90 < K1 < 175 km s−1. These constraints on K1 will be readdressed and improved in this paper. Further constraints on the system parameters and the component masses, however, require dynamical information on the donor star which, unfortunately, is normally overwhelmed by the optical emission from the X-ray irradiated disc. A new indirect method to extract dynamical informa- tion from donor stars in persistent LMXBs was proposed by Steeghs & Casares (2002) (SC02 hereafter). They detected narrow high excitation emission lines from the irradiated surface of the donor star in Sco X-1 which led to the first determination of its radial velocity curve and mass function. These lines are strongest in the Bowen blend, a combination of C III and N III lines which are powered by photoionization and fluorescence (respectively) due to UV photons arising from the hot inner disk (McClintock et al. 1975). This technique has been amply confirmed by subsequent studies of the eclipsing ADC (Accretion Disc Corona) and X-ray pulsar 4U 1822-371 (Casares et al. 2003), the black hole soft X-ray transient GX 339-4 during its 2002 outburst (Hynes et al. 2003) and Aql X-1 during its 2004 outburst (Cornelisse et al. 2006). In this paper we apply this method to the X-ray bursters 4U 1636-536 (=V801 Ara) and 4U 1735-444 (=V926 Sco) and present the first detection of the donor stars in these two LMXBs. The pa- per is organized as follows: Section 2 summarizes the observation details and data reduction. The average spectra and main emission line parameters are presented in Sect. 3, with multigaussian decon- volution of the Bowen blend. In Section 4 we analyse the radial velocities and orbital variability of the strong Bowen blend and HeII λ4686 emission lines. Estimates of the systemic velocities are obtained through the Double Gaussian technique applied to the wings of the HeII line. Using these systemic velocities we compute Doppler tomograms of HeII λ4686 and the Bowen fluorescence NIII λ4640 which are presented in Section 5. The NIII maps dis- play evidence of irradiated emission from the donor star, which is used to refine the absolute phasing and systemic velocities. Finally, in Section 6 we provide an improved determination of K1 for V801 Ara and present our constraints on the masses in the two binaries. 7 6 5 4 3 4600 4800 5000 5200 5400 5600 Figure 1. Summed spectra of V801 Ara and V926 Sco with the principal emission lines indicated. 2 OBSERVATIONS AND DATA REDUCTION V801 Ara and V926 Sco were observed on the nights of 23 and 25 June 2003 using the FORS2 Spectrograph attached to the 8.2m Yepun Telescope (UT4) at the Observatorio Monte Paranal (ESO). A total of 42 spectra of 600s of V801 Ara and 102 exposures of 200s of V926 Sco were obtained with the R1400V holographic grating, covering a complete orbital cycle per night for each tar- get. We used a 0.7 arcsec slit width which rendered a wavelength coverage of λλ4514-5815 at 70 km s−1 (FWHM) resolution, as measured from Gaussian fits to the arc lines. The seeing was vari- able between 0.6" – 1.2" during our run. The flux standard Feige 110 was also observed with the same instrumental configuration to correct for the instrumental response of the detector. The images were de-biased and flat-fielded, and the spectra subsequently extracted using conventional optimal extraction tech- niques in order to optimize the signal-to-noise ratio of the out- put spectra (Horne 1986). A He+Ne+Hg+Cd comparison lamp im- age was obtained in daytime to provide the wavelength calibration scale. This was obtained by a 4th-order polynomial fit to 19 lines, resulting in a dispersion of 0.64 A pix−1 and rms scatter < 0.05 A. Instrumental flexure was monitored through cross-correlation between the sky spectra and was found to be very small, always within 14 km s−1 (0.4 pix.) on each night. These velocity drifts were removed from each individual spectrum, and the zero point of the final wavelength scale was established from the position of the strong OI λ5577.338 sky line. All the spectra were calibrated in flux using observations of the flux standard Feige 110. However, due to light loss caused by our narrow slit and variable seeing con- ditions, our flux calibration is only accurate to ∼ 50%. 3 AVERAGE SPECTRA AND EMISSION LINE PARAMETERS Figure 1 presents the average spectra of V801 Ara and V926 Sco in fλ flux units. They show a blue continuum with broad high exci- tation emission lines of HeII λ4686, λ5411, the Bowen blend at λλ4630-50 and Hβ, which are typical of X-ray active LMXBs. Possible HeI lines at λ4922 and λ5015 are also identified but these are significantly weaker. Table 1 summarizes the FWHM, EW and centroid λc of the main emission lines obtained through sim- ple Gaussian fits. We note that the average spectra (including the Irradiated donor in 4U 1636-536/V801 Ara and 4U 1735-444/V926 Sco 3 Table 1. Emission line Parameters Line V801 Ara Bowen HeII λ4686 Hβ V926 Sco Bowen HeII λ4686 Hβ FWHM (km s−1) EW ( A) Centroid ( A) ∆V (km s−1) 1848 ± 65 1216 ± 45 963 ± 56 4.25 ± 0.03 3.31 ± 0.03 1.67 ± 0.02 4643.2 ± 0.4 4685.6 ± 0.3 4860.9 ± 0.3 – -9 ± 11 -25 ± 21 1662 ± 65 657 ± 26 507 ± 39 3.99 ± 0.02 2.18 ± 0.02 0.43 ± 0.02 4641.5 ± 0.4 4683.9 ± 0.2 4859.2 ± 0.4 – -118 ± 12 -134 ± 25 1.3 1.2 1.1 1 4600 4650 4700 Figure 2. Combined fit to HeII λ4686 and the Bowen blend using three Gaussians at ≃4634 A (NIII), ≃4641 A (NIII) and ≃4651 A (CIII). See text for details. line strengths) look very similar to those presented by A98, which seems to imply that there has been no large, long-term variations between the two data epochs. Incidentally, the EWs and FWHMs of all lines do not show any significant nigh-to-night variability nor modulation with the orbital period. Aside from the intrinsically broad Bowen blend, which is a blend of at least three CIII/NIII transitions, we note that emission lines are a factor ∼2 narrower in V926 Sco than in V801 Ara. Given the similarities in their or- bital periods, this suggests a projection effect with a lower inclina- tion angle for V926 Sco. In addition, the emission line centroids in V926 Sco are significantly blueshifted, probably due to a larger (approaching) systemic velocity. This also applies to the Bowen complex since the difference in wavelength between V801 Ara and V926 Sco (-110 km/s) is consistent with the difference in velocities found for the other lines. The blue continuum is also steeper for V926 Sco, as expected because of its lower reddening (see A98). The Bowen blend mainly consists of emissions correspond- ing to the NIII transitions at λλ4634,4641,4642 and CIII at λλ4647,4651,4652 (Schachter et al. 1989). In an attempt to esti- mate the relative contribution of the different transitions we have performed a combined multi-Gaussian fit to the average emission profiles of HeII λ4686 and the Bowen blend for the two LMXBs. The fit consists of four Gaussians, one for the HeII line, two for the NIII emissions at λ4634 and λ4641-2 and another for the CIII emissions at λ4647-52. Free parameters are the line widths, which are set to be equal for all the lines, individual line centroids and intensities. The line widths are mainly driven by the fit to the un- blended HeII line and hence the latter is effectively used as a tem- plate to constrain the line profiles within the Bowen blend. Figure 2 presents the results of the fit. Because widths are set to be the same for all lines, flux ratios are given by simple peak ratios. Our best fit yields NIII ratio I(λ4634)/I(λλ4641 − 2) = 0.31 ± 0.10 and 0.59 ± 0.02 for V801 Ara and V926 Sco, respectively. This is a factor ≤ 2 lower than the theoretical NIII ratio of 0.71, computed by Nussbaumer (1971) for Bowen fluorescence, but consistent with the results of Schachter et al. (1989) for Sco X-1. Similarly, Hynes (private communication) finds a NIII ratio in the range 0.36-0.56 (depending on time) for GX 339-4 during its 2002 outburst. We can also calculate the CIII/NIII ratio I(λλ4647 − 52)/I(λ4634 + λλ4641 − 2) and find 0.38 ± 0.06 and 0.35 ± 0.01 for V801 Ara and V926 Sco respectively. For comparison, we have performed the same analysis on average spectra of Sco X-1 and 4U 1822-371, using data presented in SC02 and Casares et al. (2003), and find 0.44 ± 0.03 and 0.49 ± 0.01 respectively. The lower CIII/NIII ra- tio in V801 Ara and V926 Sco may indicate possible evidence of CNO processed material but we have to be cautious here because of the oversimplification in our fitting model. In particular, we cannot rule out other possible transitions (most likely OII lines at λ4641.8 and λ4649.1; see McClintock et al. 1975) contributing to our com- ponents at ∼4641 A and ∼4651 A, which so far we have assumed to be dominated by NIII and CIII transitions. Furthermore, since the CIII and OII lines are not powered by Bowen fluorescence but photoionization, different line ratios may stem from differences in efficiency of these two excitation mechanisms rather than true CNO abundance variations. 4 RADIAL VELOCITY STUDY AND ORBITAL VARIABILITY Figure 3 displays the radial velocities of the Bowen blend and HeII λ4686 line for V801 Ara and V926 Sco, obtained by cross- correlating the individual spectra with Gaussians of fixed FWHM as given in Table 1. We have assumed λc = 4643.0 for the central wavelength of the Bowen blend. The velocities are folded in orbital phase, using the ephemerides of G02 and A98 for V801 Ara and V926 Sco but shifted in phase by +0.5 to make phase zero coin- cide with the inferior conjunction of the donor star. These will be called the photometric ephemerides and will be used throughout this Section. The accumulated phase uncertainty at the time of our observations is ±0.06 for both systems. Note that this phase con- vention assumes that lightcurve maxima are driven by the visibility of the irradiated donor star. The radial velocity curves for V801 Ara show maxima at phase ∼ 1. This phasing suggests that the velocity variations contain a significant component arising in the disc bulge (which has its maximum visibility at phase ∼ 0.75), caused by the in- teraction of the gas stream with the outer edge of the disc, as is typically seen in persistent LMXBs (e.g. 4U 1822-371; Cowley, Crampton & Hutchings 1982). On the other hand, the ra- dial velocity curves for V926 Sco show maxima at phase ∼ 0.7. By comparing the HeII curves in the two binaries we clearly see evi- dence for a large systemic velocity of ∼ -150 km s−1 in V926 Sco, in good agreement with Smale & Corbet (1991). We also note that the CIII/NIII blend in V801 Ara is modulated with a much larger amplitude than in V926 Sco i.e. 280 km s−1 versus 70 km s−1, respectively. Figure 4 presents the trailed spectra of the Bowen blend and 4 J. Casares et al. 600 400 200 0 -200 200 0 -200 -400 0 0.5 1 1.5 2 Orbital Phase 600 400 200 0 -200 200 0 -200 -400 0 0.5 1 1.5 2 Orbital Phase Figure 3. Radial velocity curves of the Bowen blend (top) and HeII λ4686 in V801 Ara and V926 Sco. Figure 5. Diagnostic Diagram for HeII λ4686 in V801 Ara (solid circles) and V926 Sco (open circles). The dotted horizontal line marks φ0 = 0.5 or the expected inferior conjunction of the compact object, according to the photometric ephemerides. ity curves from the wings of the profile, which are expected to follow the motion of the compact star. We have used a double- Gaussian bandpass with F W HM = 100 km s−1 and relative Gaussian separations in the range a = 400 − 1400 km s−1 in steps of 100 km s−1. In order to improve statistics, we have co-added our spectra in 15 phase bins using the photometric ephemerides. The radial velocity curves obtained for different Gaussian separa- tions are subsequently fitted with sine-waves of the form V (φ) = γ + K sin 2π(φ − φ0), fixing the period to the orbital value. The best fitting parameters are displayed as a function of the Gaussian separation a in a Diagnostic Diagram (see Fig. 5). In both cases we see how the line cores are dominated by high- amplitude (K ∼ 200 km s−1) S-waves which fade as we move to the line wings, where lower K−amplitudes of a few tens of km s−1 are found. On the other hand, the γ velocities are very steady throughout the profiles, with average values of ∼ −30 (V801 Ara) and ∼ −130 km s−1 (V926 Sco). The blue-to-red crossing phase for V926 Sco displays a smooth decreasing trend from the line core to the wings, whereas it is rather constant for V801 Ara, with an av- erage value of φ0 ∼ 0.7. We estimate that the velocity points start to be corrupted by continuum noise for Gaussian separations larger than a ∼ 1000 as indicated by the diagnostic parameter σ(K)/K (see Shafter et al. 1986). Therefore, we decided to adopt the aver- age values for the parameters obtained from the last two separa- tions before σ(K)/K starts to rise i.e. a = 900 − 1000 km s−1 for V801 Ara and 1000-1100 km s−1 for V926 Sco. This yields K1 = 93 ± 25 km s−1, γ = −42 ± 4 km s−1, φ0 = 0.68 ± 0.03 Figure 4. Trailed spectra showing the orbital evolution of the Bowen blend and HeII λ4686 in 15 phase bins using the photometric ephemerides HeIIλ4686 for V801 Ara and V926 Sco, after co-adding our indi- vidual spectra in 15 phase bins to improve statistics. The core of the emission lines in V801 Ara show clear S-wave components. The S- wave in the Bowen blend is quite narrow and it may arise from the heated face of the donor star, as observed in Sco X-1 (SC02). The S-wave in HeII is rather extended and is likely produced in a region with a large velocity dispersion such as the disc bulge. On the other hand, the Bowen blend in V926 Sco is rather noisy and does not ex- hibit any clear components that are visible by eye. The HeII trailed spectra do show a clear orbital modulation with complex structure. Following our work on Sco X-1 (SC02) and 4U 1822-371 (Casares et al. 2003), we have applied the double-Gaussian tech- nique (Schneider & Young 1980) to the wings of the HeII λ4686 line in an attempt to estimate the radial velocity curve of the com- pact object. The trailed spectra presented in Fig. 4 demonstrate that the line cores are dominated by strong, complex, low-velocity components associated with asymmetric emission from the outer disc and/or the donor star, which we want to avoid. Therefore, by convolving the emission line with a double-Gaussian filter of sufficiently large Gaussian separation we can extract radial veloc- Irradiated donor in 4U 1636-536/V801 Ara and 4U 1735-444/V926 Sco 5 for V801 Ara and K1 = 96 ± 9 km s−1, γ = −121 ± 7 km s−1, φ0 = 0.20 ± 0.01 for V926 Sco. The errorbars have been adjusted to incorporate the scatter between different values within our preferred range. Note that φ0 of V801 Ara is delayed by ∼ 0.18 with respect to the inferior conjunction of the compact object (dot- ted line in Fig. 5), as predicted by the photometric ephemerides. This is classically observed in interacting binaries and interpreted as contamination of the line wings by residual emission from the disc-bulge/hot-spot (Marsh 1998). On the other hand, φ0 of V926 Sco leads the inferior conjunction of the compact star by ∼ 0.30, according to the photometric ephemerides. Such a large shift is un- expected and a full discussion is diverted to Sect. 6. Furthermore, we also note the very asymmetric distributions obtained in the HeII Doppler maps (see next Sect.) which may invalidate the double- Gaussian technique. Therefore, the constraints on zero-phase, γ- velocity and K1 (for V801 Ara) will be superseded in the following sections. 5 BOWEN FLUORESCENCE FROM THE IRRADIATED COMPANION Doppler tomography enables us to map the brightness distribution of a binary system in velocity space through combining orbital phase spectra (see details in Marsh 2001). This is particularly ef- fective when dealing with weak emission features which are barely detected or embedded by noise in individual spectra, such as our narrow Bowen emission lines. In order to compute Doppler im- ages of the NIII λ4640 fluorescence line in V801 Ara and V926 Sco we have rectified the individual spectra by subtracting a low order spline fit to the continuum regions and rebinned them into a uniform velocity scale of 37 km s−1 per pixel. Doppler images were subsequently computed using the photometric ephemerides and γ = −42 and −121 km s−1 for V801 Ara and V926 Sco, respectively. The maps show compact spots shifted in phase by +0.03 and −0.18 with respect to the expected location of the donor star in the velocity maps (i.e. along the vertical Vy axis)1. Assum- ing that these spots are produced on the irradiated hemisphere of the donor star, as has been shown to be the case in Sco X-1 (SC02) and 4U 1822-371 (Casares et al. 2003), we correct the previous photometric ephemerides and derive the following spectroscopic ephemerides, which will be used in the remainder of the paper: T0(HJD) = 2452813.531(2) + 0.15804693(16)E T0(HJD) = 2452813.495(3) + 0.19383351(32)E (1) (2) where the zero-phase error comes from the uncertainty in the centroid position of the spots. Equation (1) corresponds to the ephemerides of V801 Ara and equation (2) to V926 Sco. With these ephemerides, the HeII λ4686 maps of the two binaries show a cres- cent shape brightness distribution pointing towards emission from an extended disc bulge. Following SC02 we have used the HeII Doppler maps to re- fine the systemic velocities derived in the previous section. The χ2 value of the map was calculated for a range of γ's, and the best fit in terms of mimimal χ2 was achieved for γ ≃ −39 km s−1 (V801 Ara) and γ ≃ −132 km s−1 (V926 Sco). The use of an incorrect systemic velocity has the effect of blurring bright spots Figure 6. Doppler maps of HeII λ4686 (left panels) and NIII λ4640 (right panels) for V801 Ara and V926 Sco, computed using our spectroscopic ephemerides. Systemic velocities γ = −34 km s−1 and γ = −140 km s−1 were adopted for V801 Ara and V926 Sco, respectively. For compari- son we also plot the gas stream trajectory (in units of 0.1 RL1) and Roche lobe of the donor star for the case of K2 = 360 km s−1 (V801 Ara) and K2 = 298 km s−1 (V926 Sco). A mass ratio q = 0.27 has been assumed for the two binaries. into , as opposed to V801 Ara,elongated "defocused" features in Doppler images. Therefore, as a further test, we have also com- puted NIII λ4640 maps for a set of γ-velocities in the range -200 – +200 km s−1 and we looked for the best focused NIII spots by computing the skewness using box sizes of 5, 10 and 20 pixels. We find that the most symmetric and compact spots are found for γ val- ues in the range -29 – -39 km s−1 (V801 Ara) and 137-143 km s−1 (V926 Sco). Therefore, we decided to adopt γ = −34 ± 5 km s−1 (V801Ara) and γ = −140±3 km s−1 (V926 Sco) and these are the values that we have used in the final Doppler maps, which are pre- sented in Fig. 6. The NIII maps reveal compact bright spots which we interpret as irradiated emission from the inner hemisphere of the donor star. These spots are obviously located along the vertical axis because of our ephemerides definition. We have calculated the position of these spots by computing their centroids using the cen- tering algorithm CENTROID in IRAF2 and find Kem = 277 ± 22 km s−1 (V801 Ara) and 226 ±22 km s−1 (V926 Sco). We note that Kem is very weakly dependent on γ, with only a 10 km s−1 drift when γ varies by ±20 km s−1 around our central value. There- fore, our Kem determinations seem very robust and they represent strict lower limits to the true radial velocity semiamplitude K2 of the donor stars because they must arise from the irradiated hemi- sphere facing the neutron star. Further confirmation of our system 1 Note that the greyscale in the NIII maps have been set to enhance the contrast of the sharp components. This is the reason why these maps do not show any trace of emission from the underlying broad component. 2 IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the Association of Universities of Research in As- tronomy, Inc., under cooperative agreement with the Nacional Science Foundation. 6 J. Casares et al. 1.3 1.2 1.1 1 1.3 1.2 1.1 1 4620 4640 4660 4680 4700 4620 4640 4660 4680 4700 Figure 7. Average spectra of V801 Ara (left panel) and V926 Sco (right panel). Top spectra have been Doppler corrected in the rest frame of the donor using the spectroscopic ephemerides and the K-velocity amplitudes derived from the bright spots in the NIII λ4640 Doppler images. The nar- row NIII and CIII lines are clearly seen in the Bowen blend. These are smeared out and cannot be detected in the straight averages shown at bot- tom. Noise has been filtered using a 2 pixel boxcar smoothing. parameters is provided by the appearance of the sharp CIII/NIII transitions after coadding all the spectra in the rest frame of the NIII λ4640 emission line region (see Fig. 7). Note that we have not attempted to compute Doppler images of the Hβ lines because these are contaminated by phase-variable absorption components, as was also the case in 4U 1822-371 (e.g. Casares et al. 2003). Ab- sorption violates the principles of Doppler Tomography and makes it very difficult to interpret the corresponding Doppler images. 6 DISCUSSION The detection of NIII λ4640 fluorescence emission from the irra- diated donor in V801 Ara and V926 Sco provides new absolute (spectroscopic) ephemerides and opens the possibility to derive the first constraints on the dynamical masses of these two LMXBs. In the light of the new ephemerides, the photometric lightcurve max- ima of V801 Ara take place at phase 0.47 ± 0.06 (with the un- certainty dominated by error propagation of the G02 ephemerides) and, therefore, they are consistent with X-ray irradiation of the donor star. On the other hand, lightcurve maxima in V926 Sco are located at orbital phase 0.82 ± 0.06. This is in between the maxi- mum visibility of the irradiated donor and the disc-bulge and hence it would imply that lightcurve maxima arise by a combination of these two emitting regions. However, we note that the uncertainty in the photometric ephemerides of V926 Sco is likely to be higher than stated in A98 due to the limited number of arrival times fitted and their large scatter (Augusteijn, private communication). There- fore, it is possible that lightcurve maxima in V926 Sco are also consistent with superior conjunction of the irradiated donor star. This is the most likely scenario because it would be hard to under- stand an emission source peaking at phase ∼ 0.8 in an irradiation dominated environment without invoking some strange (and stable) disc configuration. We have also determined the systemic velocities for the two binaries i.e. −34 ± 5 km s−1 (V801 Ara) and −140 ± 3 km s−1 (V926 Sco). These can be compared with Vr, the radial velocity, Figure 8. Top panel: Dynamic Fourier power contours showing the excur- sion in oscillation frequency during the 2001 February 22 superburst from 4U 1636-53 (V801 Ara) with the best fitting orbit model overlayed. Time zero for the x-axi is MJD 51962.7102530038 (TDB) Bottom panel: Pulse phase residuals for the best fitting model. relative to the Local Standard of Rest, due to Galactic rotation at the location of the binaries. For the case of V801 Ara, we take l = 332.9◦, b = −4.8◦ and d = 6 − 7 kpc, based on peak fluxes of radius-expansion X-ray bursts for M1 = 1.4 − 2 M⊙ (Galloway et al. 2005). Assuming the Galactic rotation curve of Nakanishi & Sofue (2003) we find Vr in the range -103 to -111 km s−1. This is much larger than our systemic velocity which could be explained through the recoil velocity gained by the binary dur- ing the supernova explosion which formed the neutron star. Thus, we can set a lower limit to the kick velocity in V801 Ara of ≃ 70 km s−1. Regarding V926 Sco, we take l = 346.1◦, b = −7.0◦ and a distance d = 9.1 kpc (for a canonical 1.4 M⊙ neutron star, see A98). The Galactic rotation curve at the position of V926 Sco yields Vr = −111 km s−1, close but significantly lower than our systemic velocity. The difference might also be ascribed to a kick velocity received by the neutron star at birth or, alternatively, a neu- tron star mass of ≃1.7 M⊙. Note that the latter implies a larger distance of d = 10 kpc and hence Vr = −140 km s−1. 6.1 System Parameters for V801 Ara The NIII λ4640 spot in the Doppler map yields Kem = 277 ± 22 km s−1. It must arise on the inner hemisphere of the donor and hence it sets a lower limit to the velocity semi-amplitude of the companion's center of mass K2. In order to find the real K2 we need to calculate the K − correction or Kem/K2 for the case of emission lines in illuminated atmospheres, which depends mainly on the binary mass ratio q = M2/M1 (with M2, M1 the masses of the donor and compact star respectively) and the disc flaring angle α. The K − correction has been calculated by Munoz-Darias, Casares & Mart´ınez-Pais (2005) using an irra- diation binary code which includes shadowing by an axisymmet- ric flared disc. The results are tabulated as a function of q, α and the inclination angle i in Table 1 of their paper, although the de- pendence on i is very weak. The K − correction is constrained between α = 0 (i.e. we neglect disc shadowing) and the geomet- Irradiated donor in 4U 1636-536/V801 Ara and 4U 1735-444/V926 Sco 7 ric limit set by emission from the irradiated limb of the donor, i.e. Kem/K2 < 1 − 0.213q2/3(1 + q)1/3 (see Munoz-Darias et al. 2005). An upper limit on q is established by taking M1 ≥ 1.4M⊙ and M2 ≤ 0.48 M⊙, the largest possible zero-age main sequence (ZAMS) star fitting in a 3.79 hr period Roche lobe (Tout et al. 1996), which leads to q ≤ 0.34. As a first approach, one can as- sume the empirical Mass-Radius relation for low-mass stars in cat- aclysmic variables and LMXBs (see e.g. Smith & Dhillon 1998, Warner 1995) which yields M2 ≃ 0.32. This, combined with a canonical neutron star of M1 ≃ 1.4 M⊙, would lead to a plausible q ∼ 0.23. The K − correction for Kem = 277 ± 22 km s−1 and q ∼ 0.23 would yield 303 < K2 < 404 km s−1, where we have adopted the coefficients for the case i = 40◦ in Table 1 of Munoz-Darias et al. (2005) because V801 Ara is not eclipsing. A more refined K − correction requires a knowledge of q. This can be constrained by the rotational broadening Vrot sin i of the companion star which, for the case of synchronous rotation, is related to q through Vrot sin i ≃ 0.462K2q1/3(1 + q)2/3 (3) (Wade & Horne 1988). A lower limit to Vrot sin i can be estimated from the width of the sharp NIII fluorescence emission in the Doppler corrected average spectrum. This is because emission lines arise not from the entire Roche lobe but from the irradiated part only. A multigaussian fit to the Bowen profile presented in Fig. 7 gives F W HM (N IIIλ4640) = 140±21 km s−1, which includes the effect of our intrinsic instrumental resolution (70 km s−1). This has been accounted for by broadening a Gaussian template of F W HM = 70 km s−1 between 10 and 200 km s−1, in steps of 10 km s−1, using a Gray rotational profile (Gray 1992) without limb- darkening, because fluorescence lines arise in optically thin condi- tions. Our simulation indicates that Vrot sin i = 92 ± 16 km s−1 is equivalent to F W HM = 140 ± 21 km s−1. On the other hand, by substituting Vrot sin i ≥ 76 km s−1 and the K − correction for α > 0◦ into equation (3), one finds a secure lower limit q ≥ 0.08. An upper limit to the rotational broadening of the donor star is set by Vrot sin i = 2πR2 sin i/P . By assuming that the donor must be more evolved than a ZAMS star within a 3.79 hr Roche lobe, then R2 ≤ 0.44R⊙ (Tout et al. 1996). This, together with i ≤ 78◦ (lack of X-ray eclipses for q ≥ 0.08), yields Vrot sin i ≤ 138 km s−1. Further constraints are provided by the study of burst oscilla- tions which can set limits to the neutron star's projected velocity K1(= qK2). SM02 derived 90 ≤ K1 ≤ 175 km s−1 by fitting the frequency drift of highly coherent X-ray pulsations observed in an 800s interval during a superburst. They used a circular orbit model and fixed the zero phase to the ephemeris of G02. We have used our new spectroscopic ephemeris to reanalyse the superburst pulsation data in order to better constrain K1. Based on our spectroscopic ephemeris, the pulsation interval during the superburst from V801 Ara comes slightly earlier in orbital phase by 0.032 cycles as com- pared to the G02 ephemeris. We fit the pulsation data (see SM02 for details on the phase timing analysis) to a circular orbit model with the reference epoch fixed to our new T0, and we also fixed the orbital period (using the G02 value). This leaves two free parame- ters, the projected neutron star velocity, K1, and the rest-frame spin frequency, ν0. We find acceptable fits with a reference epoch within the ±1σ range for our spectroscopic T0. The inferred K1 velocity ranges from 90 - 113 km s−1 as T0 ranges over ±1σ. The best fit χ2 values range from 14.6 to 10.6 (with 7 dof) over this same range, that is, higher velocities are modestly favored, but given the short- Figure 9. Constraints on K2 and q for V801 Ara. The disc flaring angle must be between α = 0◦ and α = max, whereas the rotational broadening is constrained between the observed Vrot sin i ≃ 76 km s−1 and 138 km s−1, the maximum allowed for a ZAMS star in a 3.8hr period and for i ≤ 78◦. The shaded area represents the K1 solution obtained from a reanalysis of the superburst pulsations. These constrain the binary mass ratio to be q = 0.21 − 0.34 and K2 = 286 − 433 km s−1. For comparison we also mark (dashed line) the solution for a hypothetic disc flaring angle of α = 12◦. ness of the pulse train compared to the orbital period, we do not consider these differences as significant. With the reference epoch fixed, the statistical error on the velocity is much smaller than the range given above, so 90 - 113 km s−1 is a robust range for K1 at the ±1σ limits of T0. The best fit ν0 ranges from 582.04143 to 582.09548 Hz, and which may help constrain future searches for a persistent millisecond pulsar in V801 Ara. Figure 8 summarizes the results of the new timing analysis. Fig. 9 summarizes all our restrictions in the K2 − q parameter space i.e. 0◦ ≤ α ≤ max, Vrot sin i = 76 − 138 km s−1 and K1 = 90 − 113 km s−1. The shaded area indicates the region allowed by our constraints, which yields q = 0.21 − 0.34 and K2 = 286 − 433 km s−1. These imply a mass function f (M ) = M1 sin3 i/(1 + q)2 = 0.76 ± 0.47 M⊙ and hence M1 sin3 i = 1.24 ± 0.77 M⊙. Clearly the large uncertainties are dominated by the error in Kem and the uncertainty in the disc flaring angle. 6.2 System Parameters for V926 Sco The same reasoning can be applied to V926 Sco. The spot in the NIII λ4640 map yields Kem = 226 ± 22 km s−1 whereas the ZAMS Mass-Radius relation for a 4.65 hr Roche lobe leads to M2 ≤ 0.58M⊙ (Tout et al. 1996) and hence q ≤ 0.41. On the other hand, we measure F W HM (N IIIλ4640) = 115 ± 23 km s−1 which, after deconvolution of the instrumental resolution using Gray rotation profiles as above, yields Vrot sin i ≥ 71±21 km s−1. This lower limit, combined with the K − correction for α > 0◦ and equation (3), yields q ≥ 0.05 and, hence i ≤ 80◦. And, under the assumption that the donor star is more evolved than a ZAMS star, the Roche lobe geometry implies R2 ≤ 0.54R⊙ which, for i ≤ 80◦, leads to Vrot sin i ≤ 137 km s−1. All these restrictions translate into constraints on the K2 − q plane which are presented in Fig. 10. The allowed region results in q = 0.05 − 0.41 and K2 = 215 − 381 km s−1, depending on the value of α. These numbers imply f (M ) = 0.53 ± 0.44 M⊙ and, 8 J. Casares et al. Figure 10. Same as Fig. 9 but for V926 Sco, but with no pulsation con- straints yet. Table 2. System Parameters. T0 indicates zero phase or inferior conjunction of the donor star. Orbital periods Porb are from G02 and A98. Parameter Porb (days) T0 (HJD−2 452 000) γ (km s−1) Kem (km s−1) q (M2/M1) K1 (km s−1) K2 (km s−1) f (M ) (M⊙) i (deg) 4U 1636-536 V801 Ara 4U 1735-444 V926 Sco 0.15804693(16) 813.531 ± 0.002 0.19383351(32) 813.495 ± 0.003 -34 ± 5 277 ± 22 0.21-0.34 90-113 360 ± 74 0.76 ± 0.47 36-60 -140 ± 3 226 ± 22 0.05-0.41 - 298 ± 83 0.53 ± 0.44 27-60 thus, M1 sin3 i = 0.80 ± 0.71 M⊙. Unfortunately we do not have any contraints on K1 from burst oscillations and hence our mass restrictions are not as well constrained as for V801 Ara. Our best estimates of the system parameters for both objects are presented in Table 2. 6.3 Constraints on the inclination Further constraints on the stellar masses requires a knowledge of the binary inclination. While strict upper limits are set by the ab- sence of X-ray eclipses, lower limits can be established by com- bining M1 ≤ 3.1 M⊙ (the maximum mass allowed for a stable neutron star, Rhoades & Ruffini 1974) with our mass function and q restrictions. This leads to i = 36◦ − 74◦ and i = 27◦ − 80◦ for V801 Ara and V926 Sco, respectively. In addition, the physical model of Frank, King & Lasota (1987) indicates i ≤ 60◦ due to the lack of X-ray dips. On the other hand, we mentioned in Section 3 that the fac- tor ∼2 narrowness of the HeII λ4686 profile in V926 Sco with respect to V801 Ara indicates a lower inclination. To test this hy- pothesis we have produced Doppler corrected averages for V801 Ara and V926 Sco by coadding all the spectra in two bins cen- tered at orbital phases 0.0 and 0.5, when the visibility of the irra- diated face of the donor is minimum and maximum, respectively. These spectra are presented in Fig. 11 and show a marked differ- ence for the two binaries: the narrow CIII/NIII lines become signif- icantly enhanced around phase 0.5 (top spectra) for V801 Ara, but not much difference is seen in V926 Sco between phase 0.5 and phase 0. This clearly supports a higher inclination angle in V801 Ara. Furthermore, the relative contibution of the NIII λ4640 line with respect to the broad base, as estimated through a multigaus- sian fit, is a factor ∼ 2 larger for V801 Ara than for V926 Sco. This can be taken as the relative contributions of the heated donor and disc, and hence, strongly suggests that V801 Ara is seen at a higher inclination angle than V926 Sco. This seems at odds with the fact that the optical lightcurves in V801 Ara and V926 Sco display similar amplitudes, A ≃ 0.2 mags (van Amerongen et al. 1987, van Paradijs et al. 1990). However, de Jong et al. (1996) have shown that lightcurve amplitudes are more sensitive to α than i, which might simply imply a thicker disc in V926 Sco and hence lightcurve amplitudes cannot be taken as a simple indication of the inclination angle. 6.4 Mass Estimate and Evolutionary Scenarios Meyer & Meyer-Hofmeister (1982) analyzed the effects of X-ray heating in the vertical structure of discs and demonstrated that these are strongly correlated. Even in the absence of irradiation, they find a disc opening angle of ∼ 6◦. De Jong et al. (1996) modelled the effects of irradiation in optical lightcurves and, by comparing with observations of LMXBS, they infer a mean disc opening angle of 12◦. Most of their systems are short period LMXBs, just like V801 Ara and V926 Sco, and hence it seems justified to speculate what our system parameters would be in this particular case. Regarding V801 Ara, α = 12◦ leads to q ≃ 0.25 − 0.30, K2 ≃ 356 − 375 km s−1 (see dashed line in Fig. 9) and hence f (M ) ≃ 0.81 ± 0.07 M⊙ and M1 sin3 i ≃ 1.32 ± 0.13 M⊙. Asssuming that the donor star is a 0.48 M⊙ ZAMS, then M1 ≃ 1.6 − 1.9 M⊙ and i ≃ 60 − 71◦. On the other hand, a canonical 1.4 M⊙ would imply an evolved donor with M2 ≃ 0.35 − 0.42 M⊙ and i ≥ 71◦, which is difficult to reconcile with the absence of X- ray eclipses, dips and the low amplitude of the optical modulation (see Sect. 6.3). Therefore, a flaring angle of α ≃ 12◦ seems to sup- port scenarios with ZAMS donors and massive neutron stars. These masses can be reproduced by recent evolutionary scenarios where some LMXBs are suposed to descend from binaries with high mass donors (Schenker & King 2002). These are significantly evolved at the begining of the mass transfer phase and mass is transferred in a thermal timescale until mass ratio reverses, which can result in the accretion of several tenths solar masses by the neutron star. This would make V801 Ara similar to other LMXBs such as 4U 1822- 371 (Munoz-Darias et al. 2005). However, it is also possible to accommodate a canonical neu- tron star with a MS donor and a plausible inclination angle by push- ing the disc flaring angle to higher values. For instance, α = 16◦ leads to q ≃ 0.30 −0.34, f (M ) ≃ 0.51 M⊙ and M1 sin3 i ≃ 0.88 M⊙. Then, for M1 = 1.4 M⊙, M2 ≃ 0.45 M⊙ and i ≃ 57◦. These masses would be consistent with standard evolutionary pic- tures where donors in LMXBs descend from slightly evolved low- mass stars through dynamically stable mass transfer and result in slightly undermassive stars. Note also that α < 12◦ are virtually ruled out because this would imply higher K2 and f (M ) values which result in massive neutron stars and very high inclinations. For instance, α = 8◦ leads to f (M ) ≃ 0.96 M⊙ and M1 sin3 i ≃ 1.5 M⊙. Then, for i ≤ 76◦ (lack of X-ray eclipses), M1 ≥ 1.7 M⊙ whereas a plaussible i ≃ 57◦ yields M1 ≃ 2.6 M⊙. Irradiated donor in 4U 1636-536/V801 Ara and 4U 1735-444/V926 Sco 9 1.3 1.2 1.1 1 1.3 1.2 1.1 1 4620 4640 4660 4620 4640 4660 Figure 11. Doppler corrected averages of V801 Ara (left panel) and V926 Sco (right panel) in the rest frame of the donor star. Top spectra present av- erages between orbital phases 0.25-0.75 and bottom spectra between 0.75- 0.25. The sharp CIII/NIII components in V801 Ara are clearly more intense around phase 0.5 than 0, suggesting a higher inclination binary. Spectra have been smoothed by a 2 pixel boxcar. The case of V926 Sco is unfortunately unconstrained because of the lack of an X-ray mass function. For example, assuming α = 12◦ leads to q ≃ 0.09 − −0.38, f (M ) ≃ 0.22 − −0.68 M⊙ and M1 sin3 i ≃ 0.29 − −1.07 M⊙. The allowed range in M1 and M2 is too wide as to impose any useful restriction on possible evoultionary scenarios. More, higher resolution data are required to measure the NIII λ4640 flux and Vrot sin i as a function of orbital phase, from which tighter constraints on the inclination and disc flaring angle can be set. This, together with the determination of the X-ray mass function for V926 Sco through pulse delays of burst os- cillations and smaller errors in the Kem determinations, is expected to provide stronger limits on the stellar masses and the evolutionary models for these two LMXBs (see e.g. Munoz-Darias et al. 2005). 7 CONCLUSIONS The main results of the paper are summarized as follows: • We have presented the first detection of the donor stars in the bursters LMXBs V801 Ara (=4U 1636-536) and V926 Sco (=4U 1735-444) through NIII λ4640 fluorescent emission caused by ir- radiation. • The narrow NIII λ4640 spots in the Doppler maps define Kem = 277 ± 22 km s−1 (V801 Ara) and Kem = 226 ± 22 km s−1 (V926 Sco), and a new set of spectroscopic ephemerides which lend support to the assumption that photometric modula- tion is driven by the visibility of the irradiated donor stars. On the other hand, the phasing of the radial velocity curves of HeII λ4684 and the Bowen blend suggest that these lines are mainly associated with the disc bulge. Our spectroscopic ephemerides, combined with burst oscillations data for V801 Ara, enables us to refine the neu- tron star's projected velocity to K1 = 90 − 113 km s−1 for this particular system. • Following Munoz-Darias et al. (2005), we have computed the K − corrections for the two LMXBs and obtain K2 = 360 ± 74 km s−1, q = 0.21 − 0.34, f (M ) = 0.76 ± 0.47 M⊙ (V801 Ara) and K2 = 298±83 km s−1, q = 0.05−0.41, f (M ) = 0.53±0.44 M⊙ (V926 Sco). Both systems are seen at intermediate inclination angles in the range i ≃ 30 − 60◦, with V801 Ara having the higher inclination of the two. • Regarding V801 Ara, disc flaring angles α ≤ 8◦ seem to be ruled out because of the high inclinations implied. Opening angles α ≃ 12◦ support massive neutron stars and main-sequence donors which may descend from intermediate-mass X-ray binaries as pre- dicted by some evolutionary models (Schenker & King 2002). Al- ternatively, higher opening angles α ≃ 16◦ are consistent with canonical neutron stars and main-sequence (or slightly evolved) donors which have evolved from standard LMXBs through a dy- namically stable mass tranfer phase. • The lack of an X-ray mass function in V926 Sco prevents to set tighter constraints to the K − correction, mass estimates and evolutionary history in this LMXB. ACKNOWLEDGMENTS We thank the referee Thomas Augusteijn for helful comments to the manuscript. MOLLY and DOPPLER software developed by T.R. Marsh is gratefully acknowledged. JC acknowledges support from the Spanish Ministry of Science and Technology through the project AYA2002-03570. DS acknowledges a Smithsonian Astrophysical Observatory Clay Fellowship as well as support through NASA GO grants NNG04GG96G and NNG04G014G. Based on data collected at the European Southern Observatory, Monte Paranal, Chile. REFERENCES van Amerongen S., Pedersen H., van Paradijs J., 1987, A&A, 185, 147 Augusteijn T., van der Hooft F., de Jong J.A.., van Kerkwijk M.H., van Paradijs J., 1998, A&A, 332, 561 (A98) Casares J., Steeghs D., Hynes R.I., Charles P.A., O'Brien K., 2003, ApJ, 590, 1041 Corbet R.H.D., Thorstensen J.R., Charles P.A., Menzies J.W., Naylor T., Smale A.P., 1986, MNRAS, 222, 15P Cornelisse R., Steeghs D., Casares J., Barnes A.D., Charles P.A., Hynes R.I., O'Brien K., 2006, MNRAS, submitted Cowley A.P., Crampton D., Hutchings J.B., 1882, ApJ, 255, 596 Frank, J., King A.R., Lasota J.-P., 1987, A&A, 178, 137 Galloway D.K., Psaltis D., Muno M.P., Chakrabarty D., 2005, ApJ, in press, astro-ph/05113801 Giles A.B., Hill K.M., Strohmayer T.E., Cumming N., 2002, ApJ, 568, 279 (G02) Gray D.F., 1992, "The observation and Analysis of Stellar Photo- spheres", CUP 20 Hasinger G., van der Klis M., 1989, A&A, 225, 79 Haswell C.A., King A.R., Murray J.R., Charles P.A., 2001, MN- RAS, 321, 475 Hellier C., Mason K.O., 1989, MNRAS, 239, 715 Horne K., 1986, PASP, 98, 609 Hynes R.I., Steeghs D., Casares J., Charles P. A., O'Brien K., 2003, ApJ, 583, L95 de Jong J.A., van Paradijs J., Augusteijn T., 1996, A&A, 314, 484 Marsh T.R., 1998, in Wild Stars in the Old West, eds. S. Howell, E. Kuulkerss and C. Woodward (ASP Conf. Series) vol. 137, p.236 Marsh T.R., 2001, in Astrotomography, Indirect Imaging Methods in Observational Astronomy, eds. H.M.J. Boffin, D. Steeghs and J. Cuypers (Lecture Notes in Physics) vol. 573, p.1 10 J. Casares et al. Meyer F., Meyer-Hofmeister E., 1982, A&A, 106, 34 McClintock J.E., Canizares C., Tarter C.B., 1975, ApJ, 198, 641 Munoz-Darias T., Casares J., Mart´ınez-Pais, I.G., 2005, ApJ, 635, 502 Nakanishi H., Sofue Y., 2003, PASJ, 55, 191 Nussbaumer H., 1971, ApJ, 170, 93 Paczynski B., 1971, Ann. Rev. A&A, 9, 183 van Paradijs J., van der Klis M., Pedersen H., 1988, A&A, 76, 185 van Paradijs J. et al., 1990, A&A, 234, 181 van Paradijs J., McClintock J. E., 1995, in X-ray Bina- ries,Cambridge Astrophysics Series: Cambridge University Press. eds. W.H.G. Lewin, J. van Paradijs & E.P.J. van den Heuvel, p. 58 Pedersen H., van Paradijs J., Lewin W.H.G., 1981, Nature, 294, 725 Rhoades C.E., Ruffini R., 1974, Phys. Rev. Lett., 32, 324 Shafter A.W., Szkody P., Thorstensen J.R., 1986, ApJ, 308, 765 Schachter J., Filippenko A.V., Kahn S.M., 1989, ApJ, 340, 1049 Schenker K., King A.R., 2002, ASP Conf. Ser. 261: The Physics of Cataclysmic Variables and Related Objects, 261, 242 Schneider D.P., Young P., 1980, ApJ, 238, 946 Smith D.A., Dhillon V.S., 1998, MNRAS, 301, 767 Steeghs D., Casares J., 2002, ApJ, 568, 273 (SC02) Strohmayer T.E., Zhang W., Swank J.H., White N.E., Lapidus I., 1998, ApJ, 498, L135 Strohmayer T.E., Markwardt C.B., 2002, ApJ, 577, 337 (SM02) Strohmayer T.E., Bildsten L., 2004, in X-ray Binaries, Cambridge Astrophysics Series: Cambridge University Press. 2004. eds. W.H.G. Lewin & M. van der Klis. Smale A.P., Corbet R.H.D., 1991, ApJ, 383, 853 Tout C.A., Pols O.R., Eggleton P.P., Zhanwen H., 1996, MNRAS, 281, 257 Wade R.A., Horne, K., 1988, ApJ, 324, 411 Warner B., 1995, ApJ&SS, 232, 89 This paper has been typeset from a TEX/ LATEX file prepared by the author.
astro-ph/0009414
1
0009
2000-09-26T13:38:38
Hydrodynamic approach to the evolution of cosmological structures
[ "astro-ph" ]
A hydrodynamic formulation of the evolution of large-scale structure in the Universe is presented. It relies on the spatially coarse-grained description of the dynamical evolution of a many-body gravitating system. Because of the assumed irrelevance of short-range (``collisional'') interactions, the way to tackle the hydrodynamic equations is essentially different from the usual case. The main assumption is that the influence of the small scales over the large-scale evolution is weak: this idea is implemented in the form of a large-scale expansion for the coarse-grained equations. This expansion builds a framework in which to derive in a controlled manner the popular ``dust'' model (as the lowest-order term) and the ``adhesion'' model (as the first-order correction). It provides a clear physical interpretation of the assumptions involved in these models and also the possibility to improve over them.
astro-ph
astro-ph
Hydrodynamic approach to the evolution of cosmological structures Theoretische Physik, Ludwig -- Maximilians -- Universitat, Theresienstr. 37, D -- 80333 Munchen, Germany Laboratorio de Astrof´ısica Espacial y F´ısica Fundamental, Apartado 50727, E -- 28080 Madrid, Spain Alvaro Dom´ınguez∗ Phys. Rev. D; submitted: May 18, 2000; accepted: July 18, 2000 A hydrodynamic formulation of the evolution of large-scale structure in the Universe is presented. It relies on the spatially coarse-grained description of the dynamical evolution of a many-body gravitating system. Because of the assumed irrelevance of short-range ("collisional") interactions, the way to tackle the hydrodynamic equations is essentially different from the usual case. The main assumption is that the influence of the small scales over the large-scale evolution is weak: this idea is implemented in the form of a large-scale expansion for the coarse-grained equations. This expansion builds a framework in which to derive in a controlled manner the popular "dust" model (as the lowest-order term) and the "adhesion" model (as the first-order correction). It provides a clear physical interpretation of the assumptions involved in these models and also the possibility to improve over them. 98.80.-k, 98.80.Hw, 98.65.Dx, 04.40.-b I. INTRODUCTION The standard model to understand the large-scale features of the matter distribution in the Universe after decoupling from radiation can be hardly simpler: a collection of many identical point particles interacting with each other via the Newtonian gravitational force in an expanding spatial background [1,2]. Structure arises as a consequence of the gravitational instability of initially tiny density perturbations. This model neglects relativistic effects, which become important only at scales of the order of the horizon and beyond, or when dealing with relativistic velocities. The model also excludes nongravitational interactions, which are assumed to be relevant only at small enough scales. The general solution to the dynamical evolution of this model is unknown due to the mathematical difficulties. N-body simulations, which numerically solve the dynamical equations (see Eqs. (1) in the next section), have been very helpful in understanding this evolution. In this work I look for an analytical derivation of some of the relevant features in the formation of large-scale structures. One is not usually interested in following the detailed path of each particle, but rather in some general properties that typically depend on the behavior of a large number of particles and which in the end are the kind of data provided by observations, e.g., the smoothed density and velocity fields. This naturally leads to the use of a coarse-grained or smoothed description: the evolution of a few, "macroscopic" variables is isolated by making suitable approximations for the evolution of the neglected degrees of freedom and for their influence on the relevant ones. The idea of a spatial coarse-graining has a long history in the field of statistical mechanics, where it has proven quite successful as a powerful tool for extracting information about the dynamical evolution of many-body systems. In this work I explore its application in the context of cosmological structure formation. Although the procedure of smoothing a field is widely used in cosmology, this application has mainly had a descriptive purpose. Unlike this, I study systematically the dynamical evolution of the smoothed fields. As we shall see, this method has the merit of providing a common framework for different models (dust and adhesion) of structure formation: it offers a clear explanation of the approximations involved in each model and of their physical meaning, so that it also opens the way to systematically relax them and obtain improved models. Starting from the microscopic equations of motion of the particles, I derive an exact set of hydrodynamic-like equations for the (coarse-grained) mass-density and velocity fields (Sec. II). These equations do not form an autonomous set, but rather constitute the first ones of an infinite hierarchy that must be truncated to become a useful tool. I discuss in Sec. III how this can be achieved by resort to a large-scale expansion, based on the assumption that the dynamical influence of the small scales over the large-scale evolution is weak. The lowest-order term of the expansion yields the dust model (Sec. IV), and it corresponds to a complete neglect of the structure below the coarsening scale. This leads to an eventual failure of the model (in the ∗email: [email protected] 1 form of pancake-like singularities), which can be prevented by considering the first-order correction in the expansion (Sec. V): it accounts for the dynamical influence of the structure below the coarsening scale, and I show by means of boundary-layer techniques that it gives rise to a robust "adhesive" behavior of the same kind as the adhesion model. I end up in Sec. VI with a discussion of the results. II. COARSE-GRAINING THE BASIC EQUATIONS The basic model is a system of nonrelativistic, identical point particles which (i) are assumed to interact with each other via gravity only; (ii) look homogeneously distributed on sufficiently large scales, so that the evolution corresponds to an expanding Friedmann-Lemaıtre cosmological background; and (iii) deviations to homogeneity are relevant only on scales small enough that a Newtonian approximation is valid to follow their evolution. Let a(t) denote the expansion factor of the Friedmann-Lemaıtre cosmological background, H(t) = a/a the associated Hubble function, and b(t) the homogeneous (background) density on large scales. xi is the comoving spatial coordinate of the i-th particle, ui is its peculiar velocity, and m its mass. In terms of these variables the evolution is described by the following set of equations [1] (∇i denotes a partial derivative with respect to xi): xi = 1 a ui, ui = wi − Hui, ∇i · wi = −4πGa a3 Xj6=i  m δ(xi − xj ) − b  , ∇i × wi = 0, (1a) (1b) (1c) (1d) where wi is the peculiar gravitational acceleration acting on the i-th particle. Finally, Eqs. (1) must be subjected to periodic boundary conditions in order to yield a Newtonian description consistent with the Friedmann-Lemaıtre solution at large scales [3]. To implement the idea of a spatial coarse-graining one employs a smoothing window W (z): this smoothing window should define a bounded region of space and whose inner structure is smoothed out. In App. A I discuss the general properties I will require from a smoothing window. Let L denote the (comoving) coarse-graining length scale. A microscopic mass density field and a coarse-grained mass density field can be defined respectively as follows: mic(x, t) = δ(3)(x − xi(t)), m a(t)3 Xi L3 W(cid:18) x − y (x, t; L) =Z dy L (cid:19) mic(y, t). (2a) (2b) (3a) (3b) The physical interpretation of the field (x; L) follows straightforwardly from the properties of the smoothing window: it is proportional to the number of particles contained within the coarsening cell of size ≈ L centered at x. A microscopic peculiar-momentum density field and the corresponding coarse-grained field can be defined in the same way: jmic(x, t) = ui(t) δ(3)(x − xi(t)), m a(t)3 Xi L3 W(cid:18) x − y j(x, t; L) =Z dy L (cid:19) jmic(y, t). One can introduce peculiar velocity fields umic and u by definition as j = u and similarly for umic. The physical meaning of u(x; L) is also simple: it is the center-of-mass peculiar velocity of the subsystem defined by the particles 2 inside the coarsening cell of size ≈ L centered at x. Notice that u is not obtained by coarse-graining umic: from a dynamical point of view, it is more natural to coarse-grain the momentum rather than the velocity, since the former is an additive quantity for a system of particles. Finally, one can define peculiar gravitational acceleration fields wmic and w through a coarse-graining of the force: micwmic(x, t) = wi(t) δ(3)(x − xi(t)), m a(t)3 Xi L3 W(cid:18) x − y w(x, t; L) =Z dy L (cid:19) micwmic(y, t). (4a) (4b) The field w(x) has the physical meaning of the center-of-mass peculiar gravitational acceleration of the subsystem defined by the coarsening cell at x. From these definitions and Eqs. (1a) and (1b), it is straightforward to derive the evolution equations obeyed by the coarse-grained fields and u (from now on, ∂/∂t is taken at constant x, and ∇ means partial derivative with respect to x): ∂ ∂t + 3H = − 1 a ∇ · ( u), ∂( u) ∂t + 4H u = w − 1 a ∇ · ( u u + Π), where a new second-rank tensor field has been defined: Π(x, t; L) =Z dy L3 W(cid:18) x − y L (cid:19) mic(y, t) [umic(y, t) − u(x, t; L)][umic(y, t) − u(x, t; L)]. (5a) (5b) (6) The field Π(x) is due to the velocity dispersion, i.e., to the fact that the particles in the coarsening cell have in general a velocity different from that of the center of mass. The trace of Π is proportional to the internal kinetic energy of the coarsening cell, that is, the total kinetic energy of the particles in the reference frame of the center of mass. The physical meaning of Eqs. (5a) and (5b) is simple: they are just balance equations, stating mass conservation and momentum conservation, respectively. The term ∇ · Π represents a kinetic pressure due to the exchange of particles between neighboring coarsening cells (just like in the ideal gas) and it has the same physical origin as the convective term ∇ · ( u u), i.e., a nonlinear mode-mode coupling of the velocity field. The difference is that the convective term couples only modes on scales > L, while the velocity dispersion term codifies the dynamical effect of the coupling of the modes on scales > L with the modes on scales < L. The term w codifies the gravitational interaction between the coarsening cells and it is shown later that it can be split in a similar manner into a contribution due to the large scales and another due to the coupling of the large scales with the small ones. Although Eqs. (5) look similar to the ordinary hydrodynamic equations, there is the important difference that these equations are exact: as one changes the smoothing length, the fields , u, w, Π change but in such a way that the equations remain the same (for example, upon increase of the smoothing length, part of the dynamical effect described by ∇ · ( u u) is shifted towards ∇ · Π). This property is reflected in that the equations are not an autonomous system for and u. In fact, they are just the first ones of an infinite hierarchy, as can be checked by computing the evolution equations for the fields w and Π (see Eq. (B1), for example). To obtain a useful set of equations, it is necesary to truncate this hierarchy by looking for a functional dependence of w and Π on and u. This will be the task of the next section. III. THE LARGE-SCALE EXPANSION In this section I discuss the closure of the hydrodynamic hierarchy at the level of Eqs. (5). When the particle interaction is dominated by a fast-decaying ("collisional") force, as in normal fluids, this truncation is achieved by the assumption of local equilibrium (see, e.g., Refs. [4,5]). In this case, the interaction determines a privileged smoothing scale L and it drives the coarsening cells of size L towards an approximate internal thermal equilibrium, as if isolated from each other. The evolution of the large scales (≫ L) is then ruled by the interaction between neighboring coarsening cells. In this way, for example, Eq. (5b) becomes Navier-Stokes' equation. One cannot, however, apply this approach to the problem in hand: a system with a long-ranged, unshielded interaction such as gravity does not obey the usual thermodynamics (see, e.g., the brief review in Ref. [6] and the 3 more technical remarks in Ref. [5]). Moreover, this long range implies that the interaction is self-similar and does not pick up itself a favored coarsening length. One must therefore make use of a different approach to close the hydrodynamic hierarchy. For this purpose, I introduce the large-scale expansion, which relies on the assumption that, in the context of cosmological structure formation, the evolution of the large scales is weakly influenced by what is going on in the small scales. This assumption is further discussed in Sec. VI. Here I show how to formulate this idea in order to write the fields w and Π in Eqs. (5) in terms of and u. Let a tilde denote the Fourier transform of any field: dx eik·x φ(x), φ(k) =ZV V Xk 1 φ(x) = e−ik·x φ(k), (7a) (7b) where V denotes the volume of periodicity for Eqs. (1). Then the definition (2b) can be written as (k; L) = W (L k) mic(k). Making use of the Taylor-expansion (A3) and formally inverting it, one gets: mic(k) = [1 + 1 2 B (L k)2 + o(L k)4] (k; L), mic(x) = [1 − 1 2 B (L ∇)2 + o(L ∇)4] (x; L). Consider first the field w. Fourier-transforming its definition (4) and using Eqs. (1c) and (1d), one finds: g w (k; L) = − 4πGa V Xq6=0 iq q2 mic(q) mic(k − q) W (L k). (8a) (8b) (9) This expression collects the mode-mode coupling via gravity of the scales ∼ k−1 with every other scale. Take now a large scale k−1 ≫ L: the assumption of weak coupling to the small scales then means that in the summation in Eq. (9), the main contribution arises also from the large scales, q−1, k − q−1 ≫ L. One is then justified to insert the expansions (8a) and (A3) into Eq. (9). Transforming back into real space yields the result: w = wmf + BL2(∇ · ∇)wmf + o(L ∇)4, ∇ · wmf = −4πGa ( − b), ∇ × wmf = 0. (10a) (10b) (10c) Thus, the unknown field w has been written as a functional of the coarse-grained density field. The combination L ∇ (↔ L k) can be formally viewed as a parameter which measures the influence of the small scales over the large scales and which has been assumed to be small: hence the name large-scale expansion. The field wmf , which can be called the macroscopic gravitational field, represents the gravitational field created by the monopole moment of the matter distribution in the coarsening cells, i.e., as if they had no spatial extension (L = 0). It obeys the equations that one would have naively guessed and we see that this is not the whole story: there exist a correction due to the higher multipole moments that can then be properly called tidal correction and it represents the coupling of the large to the small scales induced by gravity. Hence, this decomposition into a macroscopic field and a tidal correction is analogous to the decomposition in Eq. (5b) into convection, u u, and velocity dispersion, Π. The same procedure can be now applied to the field Π. Now, jmic = micumic obeys an expansion like (8b), but umic does not, because it is not defined by a straightforward coarse-graining. What one has in this case is slightly more involved: umic = jmic mic = [1 − 1 [1 − 1 2 B (L ∇)2 + o(L ∇)4] j 2 B (L ∇)2 + o(L ∇)4] =(cid:20)1 − 1 2 B (L ∇)2 − BL2 ∇ · ∇ + o(L ∇)4(cid:21) u. (11) Because of this minor complication, it is simpler in this case to work directly with the representation in real space, Eq. (6), into which the expansion (11) and the one corresponding to jmic are now inserted. Because the smoothing window 4 effectively restricts x − y . L, the fields within the integral evaluated at point y can be consistently Taylor-expanded around point x to order (L ∇)4. The final result reads: Π = BL2 (∂iu)(∂iu) + o(L ∇)4, (12) where a summation over the index i is implied. Taking this result and (10) into the hydrodynamic hierarchy (5), and dropping the assumed small corrections o(L ∇)4, I finally obtain ∂ ∂t + 3H = − 1 a ∇ · ( u), ∂( u) ∂t + 4H u = wmf − ∇ · ( u u) + BL2 (cid:26)(∇ · ∇)wmf − 1 a ∇ · [(∂iu)(∂iu)](cid:27) , 1 a ∇ · wmf = −4πGa ( − b), ∇ × wmf = 0. (13a) (13b) (13c) (13d) This is an autonomous system of equations for the two coarse-grained fields and u. Compared to the hydrodynamic equations of a normal fluid, we see that, because of the long range of the interaction, its contribution w to the equation for momentum conservation cannot be written as the divergence of a tensor, i.e., as a nonideal correction to the kinetic contribution represented by Π. Also the expression to lowest-order for Π is another evident difference. In the next sections, I explore the dynamical evolution described by Eqs. (13). In this section I consider the lowest order in the large-scale expansion. This corresponds to formally setting L = 0 in Eq. (13b), thus yielding: IV. THE DUST MODEL ∂ ∂t + 3H + 1 a ∇ · ( u) = 0, ∂u ∂t + Hu + 1 a (u · ∇)u = wmf , ∇ · wmf = −4πGa ( − b), ∇ × wmf = 0. (14a) (14b) (14c) (14d) This is the popular and throughly studied dust model (see, e.g., Refs. [1,2,7]). The large-scale expansion provides a clear picture of the approximations leading from Eqs. (1) to this model: by setting L = 0 one assumes that the coarsening cells can be thought of as "big particles" lacking completely an internal structure (Fig. 1). This implies neglecting (i) the velocity dispersion Π compared to the convection term u u; and (ii) the spatial extension of the cells and thus the tidal correction compared to the macroscopic gravitational field wmf . It will be useful to briefly review some results for the dust model. In the linear regime of small fluctuations about the homogeneous cosmological background, Eqs. (14) can be linearized and the resulting set of linear differential equations solved. In particular, in the long-time limit (but still within the linear approximation), the peculiar velocity and the gravitational acceleration are related by the condition of parallelism: with the function wmf (x, t) = F (t) u(x, t), F (t) = 4πGb(t) b(t) b(t) > 0, 5 (15) (16) where b(t) is the growing mode of the (small) density contrast, i.e., the growing solution of the equation b + 2H b − 4πGbb = 0. The linear evolution eventually breaks down due to the growth of inhomogeneities by gravitational instability. The solution to the fully nonlinear equations (14) is not known, but a successful approximation in this regime is the Zel'dovich approximation (ZA hereafter) [7 -- 9], which turns out to be an exact solution of Eqs. (14) for some highly symmetric configurations. In the most general case, it can be understood as the extrapolation of the parallelism condition (15) into Eq. (14b) [10]. More properly, considering the way Eqs. (14) were derived, the actual approximation is the truncated ZA [11], since the density and velocity fields have been smoothed on a scale L. Comparison with N-body simulations [11] shows that the truncated ZA performs substantially better than the original ZA: this is no wonder within the present coarse-graining formalism, since the particles that must obey the "dust" evolution (and thus be moved according to the ZA) are not the N-body particles of the simulations, which follow Eqs. (1), but rather the "big particles" which represent the coarsening cells. If one introduces a rescaled velocity field v = u/ab, and uses b(t) as the new temporal variable, then Eq. (14b) becomes in the ZA: ∂v ∂b + (v · ∇)v = 0, ∇ × v = 0, (17a) (17b) together with the irrotationality constraint following from Eqs. (15) and (14d). Hence, the problem reduces to the curl-free evolution of a fluid under no forcing at all. The solution to this equation is then inserted into the continuity equation (14a) to yield the evolved density field. As is well-known, however, the ZA has the problem of giving rise to singularities in the fields, mainly sheet-like singularities or "pancakes" [9] at which ∇ · v → −∞ and → +∞. Beyond this moment, the ZA ceases to be valid. The reason lies in the nonlinear, convective term in Eq. (17a), which deforms the initial velocity field and generically leads to a multivalued velocity field (shell crossing in the cosmological literature). This feature is likely not exclusive to the ZA but a generic property of the fully nonlinear dust model, Eqs. (14), as checked, e.g., by the application of Lagrangian perturbation techniques (see [7] and refs. therein). The generation of singularities by the convective term is not prevented by the gravitational acceleration wmf , which, on the contrary, favors this process. The emergence of singularities signals the unsuitability of the dust model beyond that time: in the next section I investigate how the correction to dust following from the large-scale expansion regularizes the singularities. V. THE ADHESION MODEL The singularities of the dust model arise because the approximation of the coarsening cells as "big particles" is bad when the density of these "particles" is large (formally infinite): the interaction between them is no longer the simple macroscopic gravitational force wmf , but it becomes more and more dependent on the internal structure of the coarsening cells. Therefore, a first step to improve the dust model is to take into account the first-order correction following from the large-scale expansion and study Eqs. (13). Obviously, the nonlinearities represented by this correction make it even more difficult to solve these equations than in the case of the dust model. Hence, in order to understand the effects of the correction on the dust evolution, I introduce some further simplifying assumptions. The idea is to consider the limit BL2 → 0+ in Eqs. (13), so that the correction is irrelevant everywhere except at those places where the dust evolution would predict a singularity. Therefore, the evolution will follow the ZA almost everywhere and the parallelism approximation (15) will be good: this implies in particular, via Eqs. (13c) and (13d), that u is also curl-free and that ∇ · u ∝ −( − b). Since the correction to dust in Eq. (13b) will be effective only near the singularities and these correspond in the ZA predominantly to pancakes, the correction can be computed under the approximation of a local plane-parallel collapse: and u change only along the direction n of the local plane-parallel collapse and ∂iuj ≈ (∇ · u) ninj. Making use of these approximations, the correction term can be simplified as follows: 1 (∇ · ∇)wmf ≈ F (∇ · u)∇ ≈ F ∇2u (∝ −∇), 1 a ∇ · [(∂iu)(∂iu)] ≈ 1 a ∇[(∇ · u)2] ≈ 3 a (∇ · u)∇2u (∝ −∇), (18a) (18b) 6 where it has been taken into account that the correction will be effective only at those places where ∼ (−∇·u) → +∞. In this limit, the contribution (18b) due to the velocity dispersion will be dominant over that of the tidal correction (18a). Inserting this result in Eq. (13b) and in terms of the variables v and b previously introduced, I finally arrive at the following model: ∂v ∂b + (v · ∇)v = ν ∇ · v ∇2v (ν = 3BL2 → 0+), ∇ × v = 0, (19a) (19b) This is a closed set of equations for the velocity field: once solved, this field is used to compute the evolved density field from the continuity equation (13a), much in the same way as with the ZA. The coefficient ν was called the gravitational multistream viscosity in Ref. [12]. Because it multiplies the highest-order derivative in Eq. (19a), one cannot simply drop this term in the limit ν → 0+, but one must rather apply the techniques of the boundary-layer theory to study this limit. The mathematical handling is collected in the next subsection; here I simply quote the conclusion that, as expected, the correction regularizes the pancake-like singularities predicted by the dust model, which become robust structures, where more and more mass gets "adhered". Another conclusion is that this "adhesive" behavior is rather insensitive to the detailed functional dependence on ∇ · v of the factor in front of ∇2v: hence, the same behavior arises in particular if one simply drops the ∇ · v to obtain a linear correction. But then, one recovers the original adhesion model [7,9,13], which was introduced as a phenomelogical (and, when compared against N-body simulations [13], very successful) correction to the ZA to go beyond the epoch of formation of singularities. In the framework of the large-scale expansion, the physical explanation of the adhesive behavior is clear (Fig. 2). The corrections to dust (18) behave as a force in the opposite direction to the density gradient, so that in the neighborhood of a pancake there is a competition between the inflow of matter driven by the gravitational attraction and this "repulsion", which prevents the formation of a singularity. The dominant contribution as → +∞ is the correction (18b) due to the velocity dispersion and represents the conversion of "coherent" streaming kinetic energy into "disordered" internal kinetic energy within the coarsening cells [12,14,15] (the mechanism has the same origin as the pressure and viscous-like forces in a gas). The subdominant contribution (18a) from the tidal correction also exhibits this "repulsive" behavior, and means that the macroscopic field wmf overestimates the gravitational attraction between coarsening cells. A. Application of the boundary-layer theory In this subsection I apply the theory of boundary layers [16] to the model (19). The physical picture is that one can study the evolution almost everywhere setting ν = 0. However, there arise regions ("shocks") spatially well separated from each other and of vanishing volume (as ν → 0+) where v has a discontinuity, ∇v diverges and the effect of the term multiplied by ν must be taken into account. The fact that these shocks correspond mainly to plane-parallel structures greatly simplifies the technical handling, since it reduces the problem to the solution of an ordinary differential equation. The purpose is to study Eqs. (19) in the vicinity of a pancake, so that one takes a point at the pancake and introduces a new coordinate system moving rigidly with it and with one of the axis normal to the pancake. In this new noninertial reference frame Eq. (19a) must be appended with a term A which collects the acceleration due to the inertial forces, and the velocity v is now understood as relative to this new reference frame. New rescaled coordinates are defined as x′ = x/ε and the idea is to take ε → 0+ in such a way that the fields and their derivatives (and consequently also the thickness of the pancake) remain finite in terms of the new coordinates x′ even in the limit ν → 0+. (The prime of the rescaled coordinates will be dropped hereafter to simplify the notation). Let n, xk denote the coordinates normal and parallel to the plane of the pancake, respectively, and vn, vk the corresponding components of the velocity v. Let ∇k denote the gradient with respect to xk. In terms of these new coordinates, Eqs. (19) now read: ε ∂v ∂b − ε A − dε db (cid:18)n ∂ ∂n + xk · ∇k(cid:19) v +(cid:18)vn (cid:18)en ∂ ∂n ∂ ∂n + vk · ∇k(cid:19) v = ε1−γ ν(cid:12)(cid:12)(cid:12)(cid:12) + ∇k(cid:19) × (en vn + vk) = 0, ∂vn ∂n + ∇k · vk(cid:12)(cid:12)(cid:12)(cid:12) 7 γ−2 (cid:18) ∂2 ∂n2 + ∇2 k(cid:19) v, (20a) (20b) where I have generalized the model (19) by allowing the prefactor of ∇2v to be an arbitrary function of ∇ · v whose asymptotic behavior as ∇ · v → −∞ is characterized by the power γ (the case γ = 2 recovers the original adhesion model). It has also been considered the case that ν (and hence, the rescaling factor ε) may depend on time. The motivation for this generalization is twofold: (i) This kind of behavior corresponds to the models discussed in Refs. [12,14,17], where the hydrodynamic hierarchy (5) is closed by simply neglecting tidal corrections altogether and assuming Πij = κγδij, κ → 0+. (ii) The computation of the trace of Π from N-body simulations also yields this kind of behavior [18]. The factor ε must be chosen so as to absorb the explicit dependence on ν in Eq. (20a) and thus to render the coefficient of the highest-order derivative of order unity, namely ε = ν1/(γ−1) and requiring γ > 1, so that ε → 0+. This latter constraint states that the correction to dust must grow fast enough with ∇v so as to succesfully oppose the gravitational attraction and overcome the formation of a singularity. Eqs. (20) are now simplified by keeping only the dominant terms in the limit ε → 0+: first, one can set ∇k → 0, because in terms of the rescaled coordinates the pancake looks like an infinite plane and the spatial variations along it take place over an infinitely large lenght scale: only the derivative normal to the pancake is relevant. Also, the acceleration terms ∂v/∂b and A are finite in the neighborhood of pancakes. Finally, dε/db is of order ε at any finite time. Therefore, Eqs. (20) reduce to the following system of ordinary differential equations: vn ∂vn ∂n γ−2 ∂2vn ∂n2 , ∂vn ∂n (cid:12)(cid:12)(cid:12)(cid:12) = 0. =(cid:12)(cid:12)(cid:12)(cid:12) ∂vk ∂n (21a) (21b) The physical interpretation of these equations is straightforward. The tangential velocity, vk, is smooth at the pancakes and so constant in the rescaled coordinates. The normal velocity, which looks discontinuous in the physical coordinates, is determined in rescaled coordinates by the balance between the convective transport towards the pancake (governed in turn by the gravitational attraction of the pancake) and the outwards "pressure" due to the velocity dispersion. The boundary conditions to be imposed to these equations are: (i) vn(0) = 0, meaning that the rescaled coordinate system is centered at the pancake and moves with it; (ii) ∂vn/∂n ≤ 0 for any n, so that there is an inflow of matter towards the pancake (and ≥ 0, see Eq. (23)); (iii) ∂vn/∂n → 0 as n → ±∞, so that the derivative in physical coordinates is not divergent outside of the pancake. Integrating twice Eq. (21a) with these boundary conditions, one finds an implicit solution vn(n): Z vn(n)/V 0 ds (1 − s2) 1 1−γ = − ε n ∆ , ∆ = ε(cid:18) γ − 1 2 V 3−γ(cid:19) 1 1−γ , (22) where V > 0 is an integration constant. Knowning vn(n), one can get an approximation to the pancake density profile by applying the parallelism condition (15) together with Poisson equation (13c) in the limit ε → 0+: = − b b ε ∂vn ∂n , whose solution reads (n) = b bV ∆ (cid:20)1 − vn(n)2 V 2 (cid:21) 1 γ−1 . (23) (24) A detailed study of the solution (22) leads to the general picture shown in Fig. 3: the fields remain univalued at all times and no singularity arises. The parameter ∆ → 0+ measures the pancake thickness in the physical coordinates. This study also shows that for the cases γ > 2 the solution approaches so fast its asymptotic values at large distances that in fact ∂vn/∂n = 0 at two finite values of the normal coordinate, n+ and n− = −n+. In such cases, the solution (22) must be replaced beyond these points by: vn(n) = (−)V , n < n− (> n+). From the general solution (22), one can compute the behavior of the solution vn(n = z/ε) in the limit ε → 0+ for a fixed value z0 of the physical coordinate z = ε n: vn = V (if z < z0), 0 (if z = z0), − V (if z > z0), (25) independently of the value of γ and the temporal dependence of ν. The conclusion is therefore that the adhesive-like behavior is a robust property of the whole family of models (20) parametrized by γ: whenever the dust evolution (ν = 0) predicts a singularity, this must be replaced by the "adhesive prescription" (25), which leads, via the continuity equation (13a), to a steady accretion of mass. 8 VI. DISCUSSION I have derived and applied a hydrodynamic-like formulation for the process of cosmological structure formation: this is a nontrivial statement, because one could believe that the collisionless nature of the basic model (1) renders a "fluid" description impossible after shell-crossing, and that then one should employ a different approach, e.g. the BBGKY hierarchy. I have shown, however, that a "fluid" description is feasible after the breakdown of the dust model: what makes a difference with "down-to-Earth" fluids is how the closure of the hydrodynamic hierarchy (5) is achieved. For this pupose, I have introduced the large-scale expansion, which builds on the assumption that the large-scale evolution is insensitive to the small scales. This assumption lies behind many reasonings in the cosmological literature: for example, behind the idea that on large scales the evolution follows a Friedmann-Lemaıtre model, regardless of the small-scale inhomogeneities, and also behind the confidence on cosmological N-body simulations, where each N-body particle is so massive that it must correspond in the real world to a full structure in its own, composed of many smaller particles. The plausibility of the assumption can be argued on the basis of the long range of gravity: the evolution can be expected to be dominated by the large scales provided there is enough large-scale power initially. (One can then also expect that the validity of the large-scale expansion should depend on the initial and boundary conditions). The good agreement with N-body simulations of the truncated ZA and the adhesion model [11,13], in which small-scale structure is completely disregarded, can be viewed as a support of this large-scale dominance. From the derived hydrodynamic equations, I have shown how the dust model arises as the lowest-order term in the large-scale expansion, while the first-order correction gives rise to a model which can be reduced to the adhesion model by further approximations. This derivation provides a clear physical interpretation of the two models: the "particles" which are assumed to follow the dust model are not such but have an internal structure, and the "interaction" between these "particles" due to its internal structure explains the adhesive behavior. By applying boundary-layer techniques, I found in particular that this behavior is a robust property of the model, in the sense that it arises quite independently of the detailed dependence of the velocity dispersion and the tidal correction on the density and velocity fields. It is interesting to compare this work with previous, related works dealing also with a hydrodynamic-like formulation of large-scale structure formation. In Refs. [12,19], the case is studied in which the hydrodynamic hierarchy is truncated by a phenomenological ansatz which writes the corrections to the adhesion model as a stochastic term (a noise). The adhesion model, in turn, was justified in Refs. [12,14,17] (see Ref. [20] for a relativistic generalization) by disregarding the tidal correction altogether and by letting the velocity dispersion be given as Πij = κγδij . The robustness of the adhesive property already mentioned explains why this behavior also arises in the models that follow from this truncation. It is particularly instructive to compare with the work in Ref. [14], where a truncation relying on something else than phenomenology was studied. I have been able to close the hydrodynamic hierarchy with less restrictive assumptions and thus shown that the adhesive behavior can still be recovered when assumptions (A2) (isotropic Π) and (A4) and (A5) (further restrictions on the form of Π) in Sec. 4 of Ref. [14] are dropped. Assumption (A1) seems in practice equivalent to the large-scale expansion, while I also employed assumption (A3) (parallelism) in order to derive adhesion-like models. But there is also a major improvement compared to that work: I do not assume the mean-field approximation from the outset (and thus the Vlasov equation, which was the starting point in Ref. [14]). In fact, I could derive an expression for the tidal correction and show that it also behaves "adhesively". As explained in more detail in App. B, this correction is also the origin for the discrepancy between expression (12) and the relationship Πij = κ5/3δij derived in Ref. [14]. Finally, I have shown that the derivation of adhesion-like models requires in principle further assumptions than the simple large-scale expansion. Thus, one could improve on these models by relaxing those assumptions and rather studying Eqs. (13). An important difference between the adhesion-like models and Eqs. (13) is that the corrections to dust in the latter generate vorticity, even if it is initially absent: this may be a relevant feature when modelling galaxy formation. Another difference is that now there is no need in principle to retain the condition BL2 → 0+: pancakes and other singularities are no longer of vanishing volume, but have an inner structure whose evolution could be studied with Eqs. (13). For such purpose, the role of the length L must be better understood. I would like to thank C. Beisbart, T. Buchert and H. Wagner for their useful comments on the manuscript. ACKNOWLEDGMENTS 9 APPENDIX A: SMOOTHING WINDOWS A smoothing window W (z) [21] should behave as a window that defines a bounded region of space and as a smoothing filter that erases the structural details inside this region. These conditions are implemented by requiring W (z) and its Fourier transform W (q) to decay faster than any power as z → ∞ or q → ∞, respectively. The smoothing window acts as an integral kernel: if φ(x) is a given field, then the associated coarse-grained field φL(x) over the scale L and its Fourier transform φL(k) are given by φL(x) = 1 L3 Z dy W(cid:18) x − y L (cid:19) φ(y), φL(k) = W (L k) φ(k). (A1a) (A1b) Therefore, the coarse-grained field at point x is just the (weigthed) addition of the original field over a region of size ≈ L around that point. Two more conditions are also required: (i) W (z = 0) = 1, so that the contribution from the neighborhood of the center of the window is unweighted; (ii) W (q = 0) = 1, so that the coarse-grained field has the same large-scale structure as the original field. This latter condition implies the following normalization: Z dx W (x) = 1, (A2) so that limL→0 L3W (x/L) = δ(x), and limL→0 φL(x) = φ(x). The condition on the decay of W (q) for large q prevents the window W (z) from having sharp borders, which thus becomes a "fuzzy" window. Otherwise, the coarse-grained field φL(x) could change in a discontinuous manner as the center of the window, x, sweeps the system. In the same way, the fast decay of W (z) for large z (which implies that W (q) lacks sharp borders in Fourier space) implies the property of spatial locality for the coarse-grained field. This also guarantees the existence of the Taylor-expansion of W (q) at q = 0 in the form: W (q) = 1 − 1 2 B q2 + o(q4), where the constant B = 1 3Z dz z2 W (z) = 4π 3 Z +∞ 0 dz z4 W (z) (A3) (A4) is related to the quadrupole moment of the smoothing window. The required constraints on the smoothing window exclude a step function (either in real or in Fourier space). A useful function which satisfies the constraints is a Gaussian: W (z) = exp(−πz2) ⇒ W (q) = exp(−q2/4π). APPENDIX B: THE EVOLUTION EQUATION FOR Π In this appendix I study the evolution equation for the velocity dispersion and show that (12) is a solution of the equation. The purpose is to compare with the result in Ref. [14] for Π, also obtained from the evolution equation for this tensor. Starting from the definition (6) and Eqs. (1a) and (1b), one can obtain the following equation for the temporal evolution of the velocity dispersion: ∂Πij ∂t + 5H Πij = − 1 a ∂k(ukΠij) − 1 a Πik∂kuj − 1 a Πjk∂kui − 1 a ∂kLijk + Pij, (B1) where a summation over repeated indices is implied, and a second-rank and a third-rank tensor fields have been defined: P(x, t; L) =Z dy L3 W(cid:18) x − y L (cid:19) mic(y, t)n[umic(y, t) − u(x, t; L)][wmic(y, t) − w(x, t; L)] + +[wmic(y, t) − w(x, t; L)][umic(y, t) − u(x, t; L)]o, (B2a) 10 L(x, t; L) =Z dy L3 W(cid:18) x − y L (cid:19) mic(y, t)[umic(y, t) − u(x, t; L)][umic(y, t) − u(x, t; L)][umic(y, t) − u(x, t; L)]. (B2b) The term L accounts for the change of velocity dispersion due to the exchange of particles between the coarsening cells, while the term P represents the change due to the gravitational interaction. To get a physical picture, just consider the equation for the trace of Π (the internal kinetic energy): then L reduces to the equivalent of the kinetic contribution to the heat flux in the usual hydrodynamics, while P becomes the power performed by the gravitational interaction. Compared to Eq. (4c) of [14], Eq. (B1) contains the extra term P, and the reason is that this term drops if tidal corrections are neglected. Indeed, if one performs a large-scale expansion of P and L in the same way as explained in Sec. III, one obtains P = BL2 [(∂iu)(∂iwmf ) + (∂iwmf )(∂iu)] + o(L ∇)4, L = o(L ∇)4. (B3a) (B3b) When these expressions are inserted into Eq. (B1), one can check that the velocity dispersion tensor Π given by (12) is indeed a solution. In Ref. [14], Eq. (B1) (with P = 0, as explained) was solved for the trace of Π by imposing the constraint of it is then found that T r Π = κ5/3, where κ is determined by the initial velocity dispersion. This shear-free flow: must be viewed as a solution valid only for early times; when the tidal corrections grow, the term P becomes relevant: Π eventually forgets its initial condition, due to the term 5H Π in Eq. (B1), and becomes "slaved" to the density and velocity fields, as given by Eq. (12). This is a mechanism similar to the "slaving" represented by the parallelism condition (15) in the linearized dust evolution. [1] P. J. E. Peebles, The Large-Scale Structure of the Universe (Princeton University Press, Princeton, 1980). [2] T. Padmanabhan, Structure Formation in the Universe (Cambridge University Press, Cambridge, 1993). [3] T. Buchert and J. Ehlers, Astron. Astrophys. 320, 1 (1997). [4] J. H. Irving and J. G. Kirkwood, J. Chem. Phys. 18, 817 (1950). [5] R. Balescu, Equilibrium and Nonequilibrium Statistical Mechanics (John Wiley & Sons, New York, 1991). [6] P. Hut, Complexity 3, 38 (1997). [7] V. Sahni and P. Coles, Phys. Rep. 262, 1 (1995). [8] Y. B. Zel'dovich, Astron. Astrophys. 5, 84 (1970). [9] S. F. Shandarin and Y. B. Zel'dovich, Rev. Mod. Phys. 61, 185 (1989). [10] S. Bildhauer and T. Buchert, Prog. Theor. Phys. 86, 653 (1991); T. Buchert, Mon. Not. R. Astron. Soc. 254, 729 (1992); L. A. Kofman, in IUAP Proceedings on Nucleosynthesis in the Universe, edited by K. Sato (Kluwer:Dordrecht, 1991). [11] P. Coles, A. L. Melott, and S. F. Shandarin, Mon. Not. R. Astron. Soc. 260, 765 (1993); A. L. Melott, T. Pellman, and S. F. Shandarin, Mon. Not. R. Astron. Soc. 269, 626 (1994). [12] T. Buchert, A. Dom´ınguez, and J. P´erez-Mercader, Astron. Astrophys. 349, 343 (1999). [13] S. N. Gurbatov, A. I. Saichev, and S. F. Shandarin, Mon. Not. R. Astron. Soc. 236, 385 (1989); D. H. Weinberg, and J. E. Gunn, Mon. Not. R. Astron. Soc. 247, 260 (1990); L. Kofman, D. Pogosyan, S. F. Shandarin, and A. L. Melott, Astrophys. J. 393, 437 (1992); A. L. Melott, S. F. Shandarin, and D. H. Weinberg, Astrophys. J. 428, 28 (1994); B. S. Sathyaprakash et al., Mon. Not. R. Astron. Soc. 275, 463 (1995). [14] T. Buchert and A. Dom´ınguez, Astron. Astrophys. 335, 395 (1998). [15] T. Buchert, in From Stars to the Universe (Annals of Shanghai Observatory, Shangai, 1998). [16] C. M. Bender and S. A. Orszag, Advanced Mathematical Methods for Scientists and Engineers (McGraw-Hill, New York, 1978). [17] S. Adler and T. Buchert, Astron. Astrophys. 343, 317 (1999). [18] A. Dom´ınguez, submitted to Astron. Astrophys. [19] A. Dom´ınguez et al., Astron. Astrophys. 344, 27 (1999). [20] R. Maartens, J. Triginer and D. Matravers, Phys. Rev. D 60, 103503 (1999). [21] Restricting the dependence to z ≡ z means a spherically symmetric window, so that the process of coarse-graining does not introduce any favored direction. 11 [22] T. Buchert, in Proc. "International School of Physics Enrico Fermi", edited by S. Bonometto, J. Primack, A. Provenzale (IOP Press: Amsterdam, 1996). 12 FIG. 1. The dust model amounts to neglecting altogether the internal structure (velocity dispersion and spatial extension) it is approximated by a "big particle" located at the center of the cell, whose mass is that contained of the coarsening cell: within it and whose velocity is the center-of-mass velocity. FIG. 2. Sketch of the particle trajectories forming a pancake, based on N-body simulations [22]. The horizontal segment represents a coarsening cell. The dashed trajectory corresponds to the dust model and its extrapolation (by the ZA) beyond the singularity; it is unable to reproduce the stabilization of the pancake. The solid trajectory represents the adhesion model and is more realistic. There is no difference initially, but at the pancake the velocity dispersion becomes very large and produces an effective "adhesive" forcing that corrects the dust evolution. 13 FIG. 3. Qualitative aspect of the coarse-grained density and velocity fields near a pancake according to Eqs. (22) and (24). ∆ represents a measure of the pancake thickness. 14
astro-ph/0703303
2
0703
2007-06-18T10:13:34
Limits on coupling between dark components
[ "astro-ph" ]
DM--DE coupling can be a phenomenological indication of a common origin of the dark cosmic components. In this work we outline a new constraint to coupled--DE models: the coupling can partially or totally suppress the Meszaros effect, yielding transfered spectra with quite a soft bending above $k_{hor,eq}$. Models affected by this anomaly do not show major variation in the CMB anisotropy spectrum and it is herefore hard to reconcile them with both CMB and deep sample data, through the same value of the primeval spectral index.
astro-ph
astro-ph
Limits on coupling between dark components Roberto Mainini, Silvio Bonometto Department of Physics G. Occhialini -- Milano -- Bicocca University, Piazza della Scienza 3, 20126 Milano, Italy & I.N.F.N., Sezione di Milano Abstract. DM -- DE coupling can be a phenomenological indication of a common origin of the dark cosmic components. In this work we outline a new constraint to coupled -- DE models: the coupling can partially or totally suppress the Meszaros effect, yielding transfered spectra with quite a soft bending above khor,eq. Models affected by this anomaly do not show major variation in the CMB anisotropy spectrum and it is herefore hard to reconcile them with both CMB and deep sample data, through the same value of the primeval spectral index. PACS numbers: 98.80.-k, 98.65.-r 7 0 0 2 n u J 8 1 2 v 3 0 3 3 0 7 0 / h p - o r t s a : v i X r a Limits on coupling between dark components 2 1. Introduction There can be little doubts that a tenable cosmological model must include at least two dark components, cold Dark Matter (DM) and Dark Energy (DE); yet only hypotheses on their nature exist, most of them assuming that DM and DE are physically unrelated and that their similar densities, in today's world and just in it, are purely accidental. Attempts to overcome this conceptual deadlock were made by several authors suggesting, first of all, that DE has a dynamical nature [1] (for a review see [2] and references therein). An alternative idea is that DE is a phenomenological consequence of the emergence of nonlinearity; this appealing option was repeatedly considered (see, e.g., [3] and references therein), but is far from being shown and leaves however apart the question of DM nature. Interactions between DM and dynamical DE [4] (see also [5]) might partially cure the problem, keeping close values for their densities up to large redshift. This option could also be read as an approach to a deeper reality, whose physical features could emerge from phenomenological limits to coupling strength and shape. A longer step forward was attempted by [6], suggesting that DM and DE derive from a single complex scalar field, being its quantized phase and modulus, respectively. The complex field could be the one responsible for CP conservation in strong interactions, within a scheme similar to Peccei & Quinn framework [7] (see also [8]). At variance from previous suggestions, which introduce parameters and aim at limiting them through data fitting, this option -- dubbed dual -- axion model -- cuts the available degrees of freedom, including as many parameters as a standard -- CDM approach. It is then quite appealing that its reduced parameter budget is sufficient to fit quite a number of observational constraints [9], still allowing for a common nature of DM and DE and for a specific shape of interaction between them. In this paper, however, we keep on the phenomenological side and discuss generic constraints to DM -- DE interactions. This discussion will have a fallout also on the dual -- axion approach, which does face a problem, because of the feature of the DM -- DE coupling it causes. A coupling of baryons with DE is ruled out by observational consequences similar to modifying gravity. Limits are looser for DM -- DE coupling, whose consequences can be appreciated only over cosmological distances. It should also be outlined that forces acting within the dark sector could modify predictions on high concentration DM lumps. There can be little doubts that cold DM particles, feeling gravity only, give them NFW profiles. Yet, observational data do not lend much support to this shape, for any scale range, and direct interaction between DM particles is severely constrained also by recent data. This is a further reason to consider DM -- DE interactions, either as a fundamental theory or as an effective framework to approach deeper physics. It is then important to devise any observational limit to such interactions and, in this paper, we outline further constraints to its shape; they are consequences of the early behavior of density Limits on coupling between dark components 3 fluctuations, over scales destined to evolve into non -- linear structures. Fluctuations over such scales enter the horizon before matter -- radiation equality and their growth is initially inhibited by the overwhelming density of the radiative component, then still behaving as a single fluid together with baryons. While fluctuations in the fluid behave as sonic waves, self gravitation of DM is just a minor dynamical effect is respect to cosmic expansion. This freezing of fluctuation amplitudes until equality is known as Meszaros effect. The main point we wish to outline here is that DM -- DE coupling can damp Meszaros effect, so that the rate of fluctuation growth, between the entry in the horizon and equality, is significantly enhanced. As a matter of fact, fluctuation freezing is essential, in shaping the transfered spectrum, which peaks on the scale khor,eq entering the horizon at equality. At smaller mass scales (k > khor,eq) the spectrum declines because of the increasing duration of the freeze. The freezing or its damping have modest consequences on the evolution of fluctuations in baryons and radiation, evolving then in the sonic regime. What we shall therefore find are significant changes in the transfer funcions, while CMB spectra keep almost unaffected. Constraints to coupled DE models arise from both linear and non -- linear effects. It has been known since long that coupling may cause a φ -- MD epoch after matter -- radiation equality (see, e.g., [10]). This changes the (comoving) distance of the last scattering band. In order to fit data, the present value of the Hubble parameter Ho needs then to be increased. Limits on Ho turn then into limits to the coupling. Limits to the coupling, in the case of a Ratra Peebles [11] self -- interaction potential, where also found in [12], by studying halo concentration distibution. The feature outlined in this work affects the transfer function, leaving almost unaffected CMB anisotropies. Discussing how the transfer function is affected by DM -- DE coupling is the main aim of this technical paper. We shall also exhibit CMB angular spectra, to confirm that they suffer just marginal changes. No general data fitting, constraining parameters and/or showing specific model advantages, will be made here. In fact, what we wish to outline is a major effect, which allows to discard a class of models, a priori. This is why we keep to cosmological parameter values ensuing from WMAP3 best -- fit [13], although deduced by assuming a ΛCDM model. In particular, we shall take an overall density parameter Ω = 1; the present value of the cold DM (baryon) density parameter will be Ωo,c = 0.224 (Ωo,b = 0.044); the dimensionless Hubble parameter will be h = 0.704; the primeval spectral index, when not taken as a free parameter, will be n = 0.947 . Within this frame we shall consider a self -- interacting scalar field, causing cosmic acceleration when its pressure/density ratio w = pDE/ρDE falls in the range (−1, −1/3). Quite in general, it is ρDE = ρk,DE + ρp,DE ≡ φ2/2a2 + V (φ), pDE = ρk,DE − ρp,DE , (1) Limits on coupling between dark components 4 so that it is −1/3 ≫ w > −1 when dynamical equations yield ρk,DE/V ≪ 1/2. Here ds2 = gµνdxµdxν = a2(τ )(−dτ 2 + dxidxi) (i = 1, .., 3) (2) is the background metric and dots indicate differentiation with respect to τ (conformal time). The w ratio exhibits a time dependence set by the shape of V (φ). Much work has been done on dynamical DE (see, e.g., [2] and references therein), also aiming at restricting the range of acceptable w(τ )'s, so gaining an observational insight onto the physics responsible for the potential V (φ). Our analysis here will however be restricted to SUGRA potentials [14] V (φ) = (Λα+4/φα) exp(4πφ2/m2 p) (3) admitting tracker solutions. Here mp = G−1/2 is the Planck mass. This will enable us to focus on peculiarities caused by the coupling. Let us also remind that, once the DE density parameters ΩDE is assigned, either α or the energy scale Λ, in the potentials (3), can still be freely chosen. In this paper we show results for Λ = 102 GeV. The SUGRA potential, at least in the absence of coupling, yields an excellent fit of observational data [15]. We tested the effects of coupling for a number of values of the scale Λ, from 10 to 104, and also changing the shape of the potential into Ratra -- Peebles. In the latter one, quantitative changes can be significant. The overall behavior is however identical and this potential is known to yield a poor fit to CMB data, unless quite a small Λ scale is taken, so spoiling its physical appeal. In the former case we find just marginal shifts. Within this frame we aim at focusing problems and showing the quantitative consequences of different options. 2. Dynamical equations Let us then start from the background equations for DE and DM when the metric is (2), reading φ + 2 a a φ + a2V,φ = C(φ)a2ρc , ρc + 3 ρc = −C(φ) φρc a a where we set C(φ) = 4rπ 3 β mp φ mp!ǫ = 4r π 3 β mp (4) (5) A possible time dependence of the coupling strength was considered but not deepened since the early work of [19]. The dual -- axion model naturally predicts a coupling C = 1/φ, consistent with eq. (5), if ǫ = −1 and β ≃ 0.244. Quite in general, for dimensional reasons, C can be expressed through products of ma p and φb with a+b = −1. During most cosmic evolution φ is a monotonically increasing function, so that the sign of the exponent of φ tells us whether the coupling was stronger of weaker in the past. (An exception can be recent times, as the exponential term in the SUGRA potential can cause a re-bounce of φ when it approaches mp; our arguments here concern much earlier times, when it is safely φ < mp). Limits on coupling between dark components 5 An expression of C, made by a polynomial including terms with different powers of φ, could select a peculiar epoch to have then a weaker or stronger coupling. Such an option, however, is clearly ad -- hoc and does not seem to deserve further investigation. An explicit dependence of C upon time would be hard to reconcile with the Lorentz invariance of the Lagrangian mass term Lc = −B(φ)mχ ¯χχ (6) setting the coupling between the DE scalar field φ and a spinor field χ supposed to yield DM, as C(φ) = d(ln B)/dφ. (Notice that B(φ)mχ is the time -- dependent mass of DM quanta). A dependence of C on φ seems therefore the only way to instaure a time -- dependent coupling. Let us then describe fluctuation equations. In the period when Meszaros effect occurs, DM, photons and baryons can be treated as fluids; (massless) neutrinos, instead, are not a fluid. Fluctuations in a generic fluid with p/ρ = w and δp/δρ = c2 s fulfill the equations δ = −(1 + w)(kv + h/2) − 3(c2 s − w)(H − C φ)δ − (1 − 3w)(C ϕ + C ′ φϕ) c2 s w v = −(1 − 3w)(H − C φ)v + 1 − 3w 1 + w Here δ = δρ/ρ, v = iuiki/k, (ρ + p)σ = −(ki · kj − δij)(T ij − δijT k k /3) (ui are the space components of the velocity field in the fluid and T ij is its stress -- energy tensor) and H = a/a, while the DE field v − kσ − kC 1 + w 1 + w kδ − ϕ . (7) φ(τ, x) = φo(τ ) + ϕ(τ, x) (8) is split into a background component φo, coinciding with the φ field obeying the eq. (4), and the space -- dependent fluctuation ϕ; in eqs. (7), Fourier components of ϕ are considered which fulfill the equation: ϕ + 2H ϕ + k2ϕ + a2V ′′ (9) while fluctuation self -- gravity is fully accounted by h, obtained by integrating the equation φ ϕ + φ h/2 = Ca2ρcδc + C ′ φa2ρcϕ , s)δγb + Ωcδc] − 2a2V ′ (10) φϕ + φ ϕo h + H h = −8πGna2ρ[Ωγb(1 + 3c2 In the absence of coupling (C = 0), we can take v = 0 in eqs. (7) without loss of precision, and face the dynamical problem though a single first order equation. This is no longer true when DM particles can be pushed by DE forces. Then the DM equation, on scales below the horizon and on times before matter -- radiation equality, reads δc +" a a − 4 mpr π 3 β(φ) φ# δc − 4πGa2ρ(cid:26)(cid:20)1 + 4 3 β 2(φ)(cid:21) Ωcδc + Ωγbδγbh1 + 3c2 si(cid:27) = 0 , (11) Here, ρ is the overall density; Ωc and Ωγb are time dependent density parameters for DM and the baryon -- photon fluid. To show eq. (11) one needs a little algebra, which will be postponed to the end of the section. Limits on coupling between dark components 6 Still from eqs. (7), we can also work out the equations for the photon -- baryon fluid, by taking C = 0 and s ≡ δpγ/(δρb + δργ) = [3(1 + 3Ωb/4Ωγ)]−1 , c2 wγb(a) ≡ pγ/(ρb + ργ) = [3(1 + Ωb/Ωγ)]−1 . (12) (13) Here Ωb and Ωγ are the time dependent baryon and photon density parameters. The baryon component can be responsible for a shift of wγb from 1/3; although initially small, it can approach 20 -- 25 % at the eve of recombination. Notice then that a non -- vanishing factor 1 − 3w ∼ Ωb/Ωγ keeps a direct influence of ϕ on baryon -- photon fluctuations. This set of equations enables the reader to build a simplified numerical algorithm, directly testing the suppression of Meszaros' effect. Clearly, to study the later evolution, since the eve of baryon -- photon decoupling a full kinetic treatment of the radiation is needed. This will be used to confirm that CMB anisotropies are just marginally affected, but is unessential to focus on the greater changes occurring to the transfer function. Let us now summarize the procedure to obtain eq. (11); a reader unintersted in it can skip the rest of this section. The starting point are again the eqs. (7). Together with them, let us consider again eq. (9). There, any mass -- like term multiplying φ, in comparison with k2, is negligible; then, before equality, it yields ϕ + (2/τ ) ϕ + k2ϕ = Ca2ρcδc − φ h/2 (14) Here, two kinds of time dependence must be compared, over fluctuation and Hubble time scales. Accordingly, we can express ϕ as sum of rapidly and slowly varying terms by actually summing up the (rapidly varying) general integral of the equation obtainable by equating to zero the l.h.s. and a (slowly varying) integral obtained by equating the last term at the l.h.s. with the r.h.s. . If we then time -- average over fluctuation time scales, only the latter contribution survives and 1 hϕi ≃ k2 (cid:16)Ca2ρcδc − φ h/2(cid:17) (15) Let us recall that dropping the contribution of fluctuating terms is the standard procedure to obtain an analytical description of Meszaros' effect. The point here is that, while all ϕ derivatives can be dropped, there is a slowly varying contribution to ϕ which cannot be soon disregarded. However, if the expression (15) is used to replace ϕ in eqs. (7) for CDM, yielding δc + kvc + h 2 dτ  (kτ )2 mp!2ǫ Ωcδc  φ = −2β 2 d  + 4r π β k2 d dτ 3 φ h p ! (16) φ−ǫm1+ǫ it becomes clear that also the slowly varying ϕ contributions can be dropped. In fact, the first term at the r.h.s. contains a division by (kτ )2, which is the squared ratio between long and short timescales and, altogether, it is ∼ O[ δ/(kτ )2] ≪ δ. We must then acknowledge that a time derivative of a quantity Q, varying over the Hubble time Limits on coupling between dark components 7 scale, is ∼ O(Q/τ ). The second term at the r.h.s. is then ∼ O[(φ/mp)1+ǫ h/(kτ )2], while we took 1 + ǫ > 0 and, in the epoch considered, φ/mp ≪ 1. It must then be even smaller than the the first one. Setting then to zero the l.h.s. of eq. (16), we obtain the relations δc + k vc + h 2 ≃ 0 , kvc ≃ − δc − h 2 . In turn, the second eq. (7), using this latter equality, yields the relation (17) (18) k vv = (C φ − H)h− δc − h/2i − k2Cϕ , which can be replaced in the former eq. (17), together with eq. (10) and (15), obtaining δc +hH − C φi δc − 4πGna2ρ[Ωγb(1 + 3c2 s)δγb + Ωcδc] − 2a2V ′ φϕ + φ ϕo − C 2a2ρcδc = 0 Neglecting here ϕ fluctuations as source terms -- DE yields a minor contribution to the overall density, as shown in Fig. 6 -- and using eq. (5), eq. (11) is soon obtainable. 3. Overcoming the freeze Let us then focus our attention on eq. (11) and outline first that, in average, the term Ωγbδγb (1 + 3c2 s) almost vanishes, as sonic waves fluctuate. Then, in the absence of coupling ( β = 0), the self -- gravitation term ∝ δc is also damped by Ωc ≪ 1. If the photon -- baryon term is then neglected, the increasing mode (approximately) reads δc ∝ 1 + 3y/2 (19) with y = 3wγb a/aeq (see, e.g., [16]) and this yields a growth from horizon to equality never exceeding a factor 2.5 , that we shall approximate as δc ∝ a1/4, according to numerical outputs. Meanwhile, above the horizon, in a synchronous gauge, δc ∝ a2 . Hence, for k > khor,eq, the growth is slowed down by a factor ∝ a7/4 h (k), while the scale factor when k passes through the horizon, ah(k) ∝ k−1. Altogether, for k > khor,eq, we expect a transfered spectrum P (k) = A knT 2(k) ≃ Akn−3.5 (T (k) is the transfer function). It is then easy to see what can change because of the coupling. The coefficient of the friction term ∝ δc is certainly reduced and can even invert its sign. Even more significantly, the term [1+4 β 2(φ)/3]Ωc can attain or overcome unity, not only because of the greater size of β(φ), but also thanks to the modified time dependence of Ωc. It turns out, in fact, that a decreasing dependence of β on φ (ǫ < 0) yields higher Ωc values in the relevant redshift interval. In general, φ is smaller at earlier times and ǫ < 0 causes a stronger coupling in the past. Let us however debate first the constant coupling case (ǫ = 0). Available data set then a constraint β < 0.1 -- 0.2 [10], [12] (see also [17]; beware of the different coupling definition), limiting the acceptable discrepancy of DM and DE evolution, after recombination, from uncoupled models. The most direct effect, in this case, concerns large scales entering the horizon late, when such discrepancies occur. The low -- l plateau Limits on coupling between dark components 8 Figure 1. Scale dependence of the density parameters of the various components in coupled DE models with constant coupling. This plot shows also the displacement of zeq and, henceforth, of khor,eq as β increases: thicker (thinner) curves refer to β = 0.01 (0.0244). Figure 2. Best fits of SDSS data for constant β from 0 (solid line) to 0.25 (dotted line). Different lines correspond to a β increase by 0.05 . The vertical dotted line yields the scale of C10. Limits on coupling between dark components 9 Figure 3. If values of n at 1 -- or 2 -- σ's from best fits are taken, spectra are significantly modified. Here we show the effect in the case with C = 1/mp. Figure 4. n intervals for increasing (constant) coupling strength of the CMB anisotropy spectrum can then undergo a modified ISW effect, while some changes in the Cl behavior, up to the first peak at l ∼ 200, can be compensated by slightly modifying n, h and other parameters. But the most significant shifts occur on the transfer function. Its slope, at k > khor,eq and up to a scale k ∼ 0.1/hMpc−1, where non -- linearity effects become important, is slightly distorted as β increases. The main effect, however, is a progressive displacement of khor,eq itself. In Figure 1 we compare the Ω evolution in models with different constant coupling, showing the significant displacement of the crossing between Ωr and Ωc, when different β's are taken. We can show the impact of this displacement by fitting transfered spectra, over these scales, with the Luminous Red Galaxies sample data from the Sloan Digital Sky Survey (SDSS) [18] and allowing the primeval spectral index n, assumed to be constant, to act as a free parameter. Limits on coupling between dark components 10 In Figure 2 we show the result of this fit. Transfered spectra, when khor,eq vary, easily accommodate deep sample data in the linear range, as the khor,eq shift is compensated by a slightly smoother slope. All that however requires a non negligible decrease of n, and transfered spectra risk to become too high in the region where they should fit CMB data (the scale of the 10 -- pole is indicated in the Figure). An attempt to balance this spectral distortion can be made by varying other It is then significant to consider Figure 3, showing deep sample data In parameters. vs. transfered spectra computed with n's within 1 -- and 2 -- σ's from the best -- fit. the Figure we took β = q3/π/4 ≃ 0.244. Then, in Figure 4, we show the 1 -- and 2 -- σ range of n, for constant coupling strength. Let us also recall that likelihood analysis showed that models with constant β <∼ 0.2 do not exhibit severe disagreements with data. The effect on CMB data fitting arising from n values distant from unity can be also directly inspected in Figure 5. According to it, we can examine Fig. 4 assuming to be viable only those models which, at the 1 -- σ level, admit n >∼ 0.85 . With reference to this admittedly qualitative criterion, we shall now consider the variable coupling case. First of all, when coupling varies, the evolution of the density parameters exhibits significant discrepancies from the constant coupling behavior. The point is that, at high redshift, they further strengthen DM self -- gravity, coherently with higher β effects. Results of a numerical illustrating this issue, are shown in Figure 6, where we compare the redshift dependence of density parameters in constant (ǫ = 0) and strongly variable (ǫ = −1) coupling models, allowing to appreciate a substantial enhancement of Ωc at the eve of equality. Altogether [1 + 4 β 2(φ)/3]Ωc is significantly greater, and the freeze of δc, between horizon entry and equality, is almost canceled. integrations, This explains the behaviors shown in Figure 7, which exhibit one of the main findings of this work. These plots are obtained from a numerical evaluations of δc, in models with different β for ǫ = 0 or -1. In the former case, the high -- z effect of coupling is marginal. In the latter one, the fluctuation growth is substantially enhanced, more significantly for greater k values. Accordingly, we can expect modifications in the transfer function and in the fit of observational data. Our general conclusion is that a time dependence of DM -- DE coupling, making it stronger at higher redshift, can prevent DM fluctuations to have a stationarity period after their entry in the particle horizon, so causing large modification of the transfer function. 4. Results These expectations will be tested by using an algorithm solving the whole set of dynamical equations. To this aim we can use our extension of the public programs CAMB, or our own code, with identical outputs. Results for transfer functions are shown in Fig. 8 for β = 0.1 and 0.244, and a variety of values of ǫ. Limits on coupling between dark components 11 Figure 5. Anisotropy spectrum of the ΛCDM model yielding the best fit to WMAP3 data compared with the spectra for coupled models with β = 0.1 and ǫ = 0, for n = 1 and n = 0.7. Already in the former case some difference exists, but no major qualitative changes occur; by adjusting other model parameters one can expect a reasonable fit to data. Taking n = 0.7, any fitting to CMB anisotropy data is apparently excluded. The suppression of fluctuation freezing is obviously stronger for greater β (and increasingly negative ǫ values). For ǫ = −1, enclosing the case C = 1/φ when β = 0.244, the steepness of the transfer function, for k > khor,eq is much reduced. The effect is still significant also for ǫ = −0.5, namely when β = 0.244 . A further effect shown by these plots is a significant displacement of the scale where T begins its gradual descent. As a consequence, different coupling laws may cause displacements on the scale where transfered spectra peak. Notice that, while this occur, the CMB anisotropy spectum keeps quite a reasonable behavior, if n ≃ 1, as is shown in Figure 9, similar to Fig. 5 but for variable coupling; here we compare the WMAP3 data on the anisotropy spectrum with the spectra obtained Limits on coupling between dark components 12 Figure 6. Redshift dependence of density parameters Ω in coupled DE models with constant and variable coupling. The two panels refer to different values of β. In both panels we show Ω's for ǫ = 0 (C ∝ β/mp, thinner lines) and ǫ = −1 (C ∝ β/φ, thicker lines). Figures are rather intricate and their reading may begin from dashed lines, yielding Ωr = Ωγ + Ων, which are almost independent from the coupling law. A more relevant effect occurs on Ωc (dotted lines), whose values, in the case of variable coupling, exceed those of constant coupling until equality. The most relevant effect occurs for DE (solid lines), whose contribution to the overall density is enhanced by several order of magnitude by variable coupling. DE fluctuations will fade after the entry in the horizon; however, high ΩDE values increase their impact on other components before their disappearance. Limits on coupling between dark components 13 log a Figure 7. Evolution of DM fluctuations in a time interval enclosing the entry in the horizon and the matter -- radiation equality. In all cases, in the presence of DM -- DE coupling some modification occurs. They are however almost negligible for constant coupling, while, for coupling ∝ φ−1, Meszaros' freezing is almost completely canceled. Limits on coupling between dark components 14 Figure 8. Transfer functions for different behaviors of DM -- DE coupling with redshift and/or for different coupling normalization. The case ǫ = 0 corresponds to redshift independent coupling intensity. The case ǫ = −1 with β = 0.244 correspond to a coupling C = 1/φ . Besides of the different slopes, notice the dependence on the model of the bending scale and, in particular, its dependence on the coupling strength, also in constant coupling models (dash -- dotted lines). for β = 0.1, ǫ = −1 and two n values: n = 1 apparently allowing a reasonable fit, and n = 0.7, as needed to obtain a reasonable transfered spectrum. Figure 8, in fact, is a direct evidence that, in order to recover a fair slope of the transfered spectrum in the scale range where structures accumulate, small primeval n values are unavoidably required. In order to perform an evaluation of the effect, we actually built transfered spectra and compared them again with SDSS data. In this way we find n ∼ 0.5 -- 0.7. Fig. 9 shows that this spoils the fit with Cl data. Spectra with ordinary downward bending fit both deep sample and CMB data with n ≃ 1. On the contrary, spectra with low n's and standard σ8's cause greater Cl still in Limits on coupling between dark components 15 -9 -9.5 -10 -10.5 0 1 2 3 Figure 9. Anisotropy spectrum of the ΛCDM model yielding the best fit to WMAP3 data compared with the spectra for coupled models with β = 0.1 and ǫ = 1, for n = 1 and n = 0.7. As in Fig. 5, in the former case one can expect to recover a reasonable fit to data by adjusting other model parameters . Taking n = 0.7, a value just acceptable to fit deep sample data, any fitting to CMB anisotropy data is apparently excluded. the Sachs & Wolfe plateau, while reducing the relative height of the Cl peaks (see again Fig. 9. These effects could be partially compensated by an adjustment of other model parameters, whose search is out of the scopes of this work. Finding n values below 0.8 -- 0.9, however, clearly means that we are dealing with unlikely physical frameworks, hardly allowing to fit CMB and deep sample data simultaneously. In the case ǫ 6= 0, the discrepancy from unity of the spectral index n, assumed to be constant, is mostly a measure of the distortion caused by the suppression of Meszaros effect, which overwhelms the effects of the displacement of khor,eq, already considered in the β = const. case. Moreover, such discrepancy is a significant estimate of the distance of the model from uncoupled physics. Our fits, shown in Figures 10 and 11, Limits on coupling between dark components 16 Figure 10. Model comparison with SDSS digital survey data. Different curves refer to different values of ǫ (as in previous Figure) with the solid line (ǫ = 1) essentially coinciding with an uncoupled model. Constant coupling models (ǫ = 0) are described by dot -- dashed curves. Negative ǫ's yield a further decrease of n. The vertical dotted line is the approximate scale where the Sachs & Wolfe Cl plateau begins. Constant coupling causes a rise of C10 by a factor ∼ 1.8 . A further factor ∼ 2 arises from a coupling C = 1/φ. Figure 11. As fig. 10, for a smaller coupling intensity. The effect on the C10 scale is much reduced, for constant coupling; however, variable coupling yields a dramatic decrease of n and, for ǫ = −1, the level of β = 0.244 is almost attained. are complemented by 1 -- and 2 -- σ intervals around best -- fit n values (Figure 12); they are an indication of which models, by adjusting other parameters, might be susceptible to approach the observational scenario. The behavior of transfered spectra in respect to data, when taking n values at 1 -- and 2 -- σ from best fits is also shown in Figure 13 (similar to Fig. 3). If we take again n = 0.85 at 1 -- σ as a threshold to discard a model, no ǫ < 0 coupled model is allowed with β = 0.244, while ǫ < −0.16 are also inhibited with β = 0.1 . At Limits on coupling between dark components 17 Figure 12. 1 -- and 2 -- σ intervals of n, when ǫ varies, for β values. Figure 13. If values of n at 1 -- or 2 -- σ's from best fits are taken, spectra are significantly modified. Here we show the effect in the case with C = 0.4(π/3)1/2/pmpφ. Limits on coupling between dark components 18 2 -- σ's the situation is not much improved for β = 0.244, while lower values of ǫ are admitted for β = 0.1 . In particular, a model with C = 1/φ, as the dual -- axion model, lays outside of the range indicated. The analysis was extended here to models with positive ǫ, for which coupling rises while φ increases. A large deal of these models is apparently allowed. 5. Conclusions The quest for models fitting observational data and avoiding fine tuning and coincidence suggests to test cosmologies where DM is coupled to dynamical DE. It is then important to recognize that quite a few models, with constant or variable DM -- DE coupling, are consistent with observational data, although their likelihood might be slightly smaller than uncoupled models. Constraints on these models were known to arise from a number of linear and non -- linear effects. The main linear anomaly is the existence of a prolongated period, after matter radiation equality, when the φ -- field energy affects the expansion rate (the φ -- MD epoch). As a consequence, the comoving distance of the last scattering band can be different from ordinary models. In principle, this can be compensated by varying the present value of the Hubble parameter Ho. There are however severe limits on Ho, which turn into limits on the coupling strength, discussed in [9]. The same feature causes also a displacement of the wave number khor,eq, corresponding to the scale which enters the horizon at matter -- radiation equality. Here we discussed also this effect, which was however expected. Constraints on coupled DE models were also found by studying their non -- linear evolution. In [12] it is shown that coupling can affect the halo concentration. Although this effect could be softened by using suitable self -- interaction potentials, it yields a limit β <∼ 0.15 -- 0.20, for the coupling, when the self -- interaction potential is Ratra -- Peebles [11]. In this paper we outlined a new effect, that DM -- DE coupling may cause in a class of models. A prolongated period, between the horizon entry and the equivalence, when DM fluctuation growth stagnates, is essential in shaping transfered spectra Ptr(k). Because of this Meszaros's effect, Ptr(k) bends at k > khor,eq. In this paper we outline that this stagnation period can be partially or totally suppressed by the coupling of DM with the φ field. As a result, transfered spectra exhibit a much softer bending at k > khor,eq. We also outline that, while this occurs, the dynamics of sonic waves in the baryon -- photon fluid is only marginally affected. Accordingly, while transfered spectra suffer major changes, CMB anisotropies are almost invariant. Using a single primeval spectral index n, we can fit both CMB and deep sample data, in the presence of standard Meszaros effect. In the presence of the above softening, the values of n required to fit CMB and deep sample data become badly discrepant. In this work we discuss which class of coupled DE models are affected by this If the coupling is constant or proportional to a positive power of φ, the anomaly. Limits on coupling between dark components 19 anomaly is absent. On the contrary, models where the coupling is proportional to an inverse power of φ are at risk. We gauge the impact of this anomaly by evaluating the value of n required to fit just deep sample data with each model. When n <∼ 0.85, it is legitimate to believe that the model likelihood is highly suppressed. The search of such n value should therefore be preliminary to any attempt to reconcile a coupled model to the whole set of CMB, deep sample (and other) data. Admittedly, apart of conceptual reasons, within the present observational framework there lacks any specific phenomenological push to invoking a DM -- DE coupling. Yet, uncoupled models, even apart of their conceptual weakness, cause a number of questionable predictions, e.g. NFW profiles. Interactions within the dark side were often advocated to cure such difficulties. An impact on profiles was actually shown to exist, but new form of coupling need to be inspected. Furthermore, fresh data on the redshift dependence of ρc, ρb and ρDE, at z ∼ 1 -- 5, might soon be available, if experiments like DUNE [20] will become operational. The discovery of an anomalous scaling of ρc, for instance, would set a strong prior, completely biasing likelihood distributions, just as priors on the value of h (Hubble parameter) suppress the likelihood of SCDM models with h ∼ 0.4 , which would otherwise allow a reasonable fit of large sets of data [21]. In particular, models with rising coupling (ǫ > 0) could become an important option. Such coupling behavior could directly arise from suitable microphysics or be a phenomenological description of a complex underlying physics. Acknowledgments Luca Amendola and Loris Colombo are gratefully thanked for their comments on this work. Limits on coupling between dark components 20 References [1] Wetterich C. 1988, Nucl.Phys.B 302, 668; Ratra B. & Peebles P.J.E., 1988, Phys.Rev.D 37, 3406 [2] Peebles P.J.E. & Ratra B., 2003, Rev.Mod.Phys. 75, 559 [3] Kolb E.W., Matarrese S. & Riotto A., 2006, New J.Phys. 8, 322 [4] Ellis J., Kalara S., Olive K.A. & Wetterich C., 1989, Phys. Lett. B228, 264; Wetterich C., 1995, A&A 301, 321 Amendola L., 2000, Phys.Rev. D62, 043511 Gasperini M., Piazza F.& Veneziano G., 2002, Phys.Rev. D65, 023508 [5] Casas J.A., Garca -- Bellido J & Quiros M., 1992, Class.Quant.Grav. 9, 1371; Anderson G.W . & Carroll S.M., Procs. of "COSMO-97, First International Workshop on Particle Physics and the Early Universe", Ambleside, England, September 15-19, 1997, astro-ph/9711288; Bartolo N. & Pietroni M., 2000, Phys.Rev. D61, 023518; Pietroni M., 2003, Phys.Rev D67, 103523; Chimento L.P., Jakubi A.S., Pavon D. & Zimdahl W.,2003, Phys.Rev D67, 083513; Rhodes C.S., van de Bruck C, Brax P., & Davis A.C., 2003, Phys.Rev. D68, 083511; Farrar G.R. & Peebles P.J.E., 2004, ApJ 604, 1 Gromov A., Baryshev Y. & Teerikorpi P., 2004, A&A, 415, 813 [6] Mainini R. & Bonometto S.A., 2004, Phys.Rev.Lett. 93, 121301 [7] Peccei R.D. & Quinn H.R. 1977, Phys.Rev.Lett. 38, 1440; Weinberg S. 1978, Phys.Rev.Lett. 40, 223; Wilczek F. 1978, Phys.Rev.Lett. 40, 279; Kim J.E. 1979, Phys.Rev.Lett. 43, 103 [8] Preskill J. et al 1983, Phys.Lett B120, 225; Abbott L. & Sikivie P. 1983 Phys.Lett B120, 133; Dine M. & Fischler W. 1983 Phys.Lett B120, 137; Turner M.S. 1986 Phys.Rev.D 33, 889 [9] Mainini R., Colombo L. & Bonometto S.A., 2005, ApJ 632, 691 [10] Amendola L. & Quercellini C., 2003, Phys. Rev. D68, 023514 [11] Ratra B. & Peebles P.J., 1988, Phys.Rev. D37, 3406 [12] Maccio' A. V., Quercellini C., Mainini R., Amendola L., Bonometto S. A., 2004, Phys. Rev. D69, 123516 [13] Spergel D.N. et al. 2006, ApJ (in press), astro-ph/0603449 [14] Brax P. & Martin J., 1999, Phys.Lett., B468, 40; Brax P. & Martin J., 2001, Phys.Rev. D61, 10350; Brax P., Martin J., Riazuelo A., 2000, Phys.Rev. D62, 103505 [15] Colombo L.P.L. & Gervasi M., 2006, JCAP 0610, 001 [16] Coles M. & Lucchin F., 1995, Cosmology, John Wiley & Sons, Chichester [17] Lee S., Liu G. & Ng K., 2006, Phys.Rev. D73, 083516; Guo Z., Ohta N. & Tsujikawa S., 2007, astro-ph/0702015 [18] Tegmark M. et al., 2006, Phys.Rev. D74, 123507 [19] Amendola L., 2004, Phys.Rev.D69, 103524 [20] R´efr´egier A. et al., Procs. of SPIE symposium "Astronomical Telescopes and Instrumentation", Orlando, may 2006, astro-ph/0610062 [21] Blanchard A., Douspis M., Sarkar S., Rowan -- Robinson M., 2003, A&A 412, 35
astro-ph/9907298
1
9907
1999-07-21T23:02:26
The Blandford-Znajek mechanism and emission from isolated accreting black holes
[ "astro-ph" ]
In the presence of a magnetic field, rotational energy can be extracted from black holes via the Blandford-Znajek mechanism. We use self-similar advection dominated accretion (ADAF) models to estimate the efficiency of this mechanism for black holes accreting from geometrically thick disks, in the light of recent magnetohydrodynamic disk simulations, and show that the power from electromagnetic energy extraction exceeds the accretion luminosity for ADAFs at sufficiently low accretion rates. We consider the detectability of isolated stellar mass black holes accreting from the ISM, and show that for any rapidly rotating holes the efficiency of energy extraction could reach 0.01. The estimated total luminosity would be consistent with the tentative identification of some EGRET sources as accreting isolated black holes, if that energy is radiated primarily as gamma rays. We discuss the importance of emission from the Blandford-Znajek mechanism for the spectra of other advection dominated accretion flows, especially those in low luminosity galactic nuclei.
astro-ph
astro-ph
ApJL in press Preprint typeset using LATEX style emulateapj v. 04/03/99 THE BLANDFORD-ZNAJEK MECHANISM AND EMISSION FROM ISOLATED ACCRETING BLACK HOLES Philip J. Armitage and Priyamvada Natarajan Canadian Institute for Theoretical Astrophysics, McLennan Labs, 60 St George St, Toronto, M5S 3H8, Canada ApJL in press ABSTRACT In the presence of a magnetic field, rotational energy can be extracted from black holes via the Blandford-Znajek mechanism. We use self-similar advection dominated accretion (ADAF) models to estimate the efficiency of this mechanism for black holes accreting from geometrically thick disks, in the light of recent magnetohydrodynamic disk simulations, and show that the power from electromagnetic energy extraction exceeds the accretion luminosity for ADAFs at sufficiently low accretion rates. We consider the detectability of isolated stellar mass black holes accreting from the ISM, and show that for any rapidly rotating holes the efficiency of energy extraction, ǫBZ ≡ LBZ/ M c2, could reach 10−2. The estimated total luminosity would be consistent with the tentative identification of some EGRET sources as accreting isolated black holes, if that energy is radiated primarily as gamma rays. We discuss the importance of emission from the Blandford-Znajek mechanism for the spectra of other advection dominated accretion flows, especially those in low luminosity galactic nuclei. Subject headings: accretion, accretion disks - black hole physics - magnetic fields - galaxies: nuclei - gamma rays: observations - X-rays: galaxies 1. INTRODUCTION The formation of both stellar mass and supermassive black holes could well lead to rapidly rotating Kerr black holes. If so, this reservoir of rotational energy constitutes a potentially important energy source for driving jets and emission from black hole systems. Moreover, this energy can be extracted if the black hole is embedded in a mag- netic field (Blandford & Znajek 1977). Since all black holes of current astrophysical interest are probably ac- creting from magnetized disks, this has led to suggestions that the Blandford-Znajek process plays a vital role in Active Galactic Nuclei (AGN) and other accreting black hole systems (Begelman, Blandford & Rees 1984; Wilson & Colbert 1995; Moderski, Sikora & Lasota 1998; Paczyn- ski 1998; Lee, Wijers & Brown 1999). Recent papers (Ghosh & Abramowicz 1997; Livio, Ogilvie, & Pringle 1999; Li 1999) have re-evaluated the importance of the Blandford-Znajek mechanism for AGN in light of advances in the understanding of accretion disks – in particular, the demonstration that turbulence driven by magnetohydrodynamic instabilities (Balbus & Hawley 1991) leads to disk angular momentum transport (Hawley, Gammie & Balbus 1995; Brandenburg et al. 1995; Stone et al. 1996; for a review see e.g. Hawley & Balbus 1999). Two principal effects lead to substantial downward revi- sions in the estimated efficiency of the Blandford-Znajek mechanism for thin disks. First, in an accretion disk both the viscosity and the magnetic diffusivity are generated by the same turbulent processes, and so have comparable magnitudes. This implies that any external magnetic field can diffuse outward through the inflowing gas, and will not accumulate on the hole. Second, the strongest magnetic fields generated by a thin disk are expected to have char- acteristic scales of order the disk scale height H, with only weaker large scale fields threading the black hole and con- tributing to the extraction of the rotational energy. Both arguments are weakened for thick disks for which H ∼ R. In this Letter, we consider the Blandford-Znajek contri- bution to the luminosity of black hole systems at extremely low accretion rates. Motivated by the hard high energy spectra and low luminosity of such systems, Narayan & Yi (1994; 1995) have developed an advection dominated accretion flow (ADAF) model, in which radiatively ineffi- cient accretion occurs in a hot, geometrically thick geome- try (see also Rees et al. 1982; Ichimaru 1977). Strong disk winds may also contribute to the observed dearth of emis- sion (Blandford & Begelman 1999). As Rees et al. (1982) noted, the low radiative efficiency ǫADAF = LADAF/ M c2 of ADAFs means that even rather weak emission arising from the Blandford-Znajek mechanism could make an im- portant contribution to the overall luminosity. Here, we concentrate on the extreme case of isolated black holes ac- creting from the interstellar medium (ISM), and show that the power from the Blandford-Znajek process substantially improves the prospects for detecting any rapidly rotating holes. We also comment on the implications for other very low luminosity accreting black holes, particularly those in quiescent galactic nuclei. 2. EVALUATING THE BLANDFORD-ZNAJEK POWER For a black hole of mass M and angular momentum J, with a magnetic field B⊥ normal to the horizon at Rh, the power arising from the Blandford-Znajek mechanism is given by (e.g. Ghosh & Abramowicz 1997; Macdonald & Thorne 1982; Thorne, Price & Macdonald 1986), LBZ = 1 32 F B2 ω2 ⊥R2 hc(cid:18) J Jmax(cid:19)2 (1) where Jmax = GM 2/c is the maximal angular momentum of the hole. The factor ω2 h depends on the angular velocity of field lines ΩF relative to that of the hole, Ωh. Conventionally, we will assume that ωF = 1/2, F ≡ ΩF (Ωh−ΩF )/Ω2 1 2 Blandford-Znajek emission in low luminosity black hole systems which maximizes the power output (Macdonald & Thorne 1982; Phinney 1983; Thorne, Price & Macdonald 1986). To estimate LBZ we need to relate B⊥ to the conditions at the inner edge of an ADAF. First though, we briefly consider the possibility that B⊥ at the black hole hori- zon could exceed the disk field at the inner disk edge, if the disk is able to advect an organized external magnetic field inwards. Whether this is possible depends on the relative timescales for inward advection, controlled by the disk shear viscosity ν, versus outward diffusion, controlled by the magnetic diffusivity η. For a disk with magnetic Prantl number P ≡ ν/η, the ability of the disk to advect field lines inwards depends on the dimensionless param- eter D = (R/H)P, where H is the disk scale height at radius R. Lubow, Papaloizou & Pringle (1994) found that D ∼< 1 was a necessary condition for strong inward field dragging (see also Reyes-Ruiz & Stepinski 1996). Since it is generally believed that in a turbulent medium P ≃ 1 (e.g. Parker 1979), this result implies that for a thin disk D ∼ (R/H) ≫ 1, and external fields cannot accumulate at small radii. For thick disks, where H ∼ R, the situation is less clear, but for a lower limit on B⊥ we will assume that even in thick disks, advection of the field is ineffec- tive. The field at the black hole then reflects only the local disk magnetic field strength. To estimate the physical conditions in the inner part of the disk, we employ the vertically averaged, self-similar ADAF solution of Narayan & Yi (1994; Spruit et al. 1987). This solution assumes a Shakura-Sunyaev (1973) prescrip- tion for the shear viscosity, ν = αcsH, where α is a free parameter and cs is the local disk sound speed. The char- acter of the solution depends on the parameter ǫ′, where ǫ′ ≡ 5/3 − γ f (γ − 1) , (2) and γ is the ratio of specific heats. The factor f mea- sures the efficiency of radiative cooling, so that f = 1 corresponds to the limit of no cooling at all, while f = 0 resembles the case of a thin disk in which cooling is effi- cient. For α ≪ 1, as suggested by numerical simulations, the solution for the sound speed and the radial velocity vR is (Narayan & Yi 1994), vR ≃ − 3α 5 + 2ǫ′ vk, c2 s ≃ 2 5 + 2ǫ′ v2 k, (3) where vk = (GM/R)1/2 is the Keplerian velocity. Making use of H/R = cs/vk, and the continuity equation, ρ = − M /4πRHvR, (4) we find that the pressure P = ρc2 s is given by, M√2 12πα P = (5 + 2ǫ′)1/2(GM )1/2R−5/2. (5) Studies that have dropped the requirements of self- similarity and vertical averaging have found that this solution provides a good approximation to ADAF flows (Narayan & Yi 1995). Numerical simulations of thin, magnetized accretion disks (Hawley, Gammie & Balbus 1995; Brandenburg et al. 1995; Stone et al. 1996), have shown that angular momentum transport is dominated by Maxwell stresses. The magnetic contribution to α typically exceeds the fluid stresses by an order of magnitude, so that α ≃ αmagnetic. The dominant field component is toroidal, with satura- tion occurring when Pmag ≪ P . Initial results from global simulations (Armitage 1998) suggest that Narayan & Yi's assumption – that in thick disks the magnetic fields pos- sess most of their power in long wavelength modes of scale ∼ R – is reasonable. We assume that Pmag = B2/8π ∼ αP in the inner disk, and make the important assumption that B⊥ ≈ B. We further assume that the energy extracted from the hole is not all dumped into the internal energy of the accreting gas and lost invisibly across the horizon – the fate of the bulk of the accretion energy (Narayan, Garcia & McClin- tock 1997). This is reasonable provided that some fraction of the field lines threading the hole are open, for example as a result of a wind or a jet. Evaluating the Blandford- Znajek power for the case of a maximally rotating hole, where J = Jmax and Rh = GM/c2, we then obtain, √2 192 LBZ ≃ (5 + 2ǫ′)1/2 M c2. (6) This corresponds to a constant efficiency, ǫBZ ≡ LBZ/ M c2, given for the range of thick disk models by, √14 192 √10 192 ǫBZ ≃ ǫBZ ≃ for ǫ′ = 1 for ǫ′ = 0 (7) (8) This efficiency drops rapidly for more slowly spinning holes, due to both the (J/Jmax)2 factor in equation (1), and because the pressure in the disk, which should be evaluated at the marginally stable orbit, drops as R−5/2. Nonetheless, it suggests that the Blandford-Znajek mech- anism will be an important power source in the overall energy budget of ADAFs around rapidly rotating black holes, with efficiencies of up to ∼ 10−2. We note that if ǫ′ > 1 (corresponding to a thin disk), these scalings suggest still higher efficiencies. More realistically, the suppressing effects identified by Livio, Ogilvie & Pringle (1999) are ex- pected to be important in this regime. Of course, if a thin disk existed the holes would be readily detectable due to the high radiative efficiency (∼ 0.1) of thin disk accretion. Additional electromagnetic extraction of energy from the inner disk, which Livio, Ogilvie & Pringle (1999) argue is likely to exceed that from the hole itself, would also boost the efficiency. 3. ISOLATED BLACK HOLES The known neutron stars and stellar mass black holes in mass transfer binaries must be greatly outnumbered by the isolated populations of such remnants, accreting only at a very low rate from the ISM (e.g. Blaes & Madau 1993). To date, only a few isolated neutron stars have been identified (Walter & Matthews 1997; Neuhauser & Trumper 1999). Identification of the corresponding black hole population would provide both a record of massive star formation in the galaxy, and insight into the behavior of accretion flows at extremely low M . Armitage & Natarajan 3 The accretion rate onto a black hole accreting from the ISM is given by the Bondi-Hoyle rate (Hoyle & Lyttleton 1939; Bondi & Hoyle 1944). For a hole moving with veloc- ity v∞ through a uniform medium with sound speed c∞, the accretion radius is, Ra = 2GM v2 ∞ + c2 ∞ . (9) Gas falling within a cylinder of this radius is accreted, so that the accretion rate is ∞, where ρ∞ is the density far upstream of the accretor. MBondi = πR2 aρ∞pv2 ∞ + c2 The most luminous black holes are expected to be those accreting from nearby, dense, molecular clouds (Fujita et al. 1998). For gas at temperatures of T ∼ 102 K, v∞ ≫ c∞, and the accretion rate is, 10M⊙(cid:19)2 MBondi = 7.3 × 1015(cid:18) M 10 kms−1(cid:19)−3 (cid:18) 102 cm−3(cid:17) gs−1 (10) (cid:16) n∞ v∞ where we have taken a density n∞ (in particles / cm3) typical for a molecular cloud. The accretion rate will be much lower for holes accreting from hotter phases of the ISM. For gas at n∞ ∼ 1 cm−3 and T ∼ 104 K, the same hole accreting transonically (v∞ ≈ c∞ ≈ 15 kms−1) will have an accretion rate, Some black holes in mass transfer binaries are believed to be rotating rapidly (Zhang, Cui & Chen 1997), and we assume that some fraction of isolated black holes also have rapid (J ∼ Jmax) rotation. Taking the Blandford-Znajek efficiency calculated earlier for the ǫ′ = 1 ADAF model (appropriate for a fully advection dominated, γ = 4/3 gas), the estimated power is, MBondi = 8 × 1012 gs−1. LBZ ≃ 1.3 × 1035 ergs−1 (n∞ = 102 cm−3, v∞ = 10 kms−1) LBZ ≃ 1.4 × 1032 ergs−1 (n∞ = 1 cm−3, v∞ = c∞ = 15 kms−1). (11) These luminosities are comparable to those predicted on the basis of models of spherical accretion of magnetized gas (Heckler & Kolb 1996; Ipser & Price 1982, 1983). Spheri- cal models are unrealistic at least for black holes accreting at high Mach numbers. The luminosities in equation (11) are considerably higher than estimates based on the radia- tive efficiency of an ADAF (Fujita et al. 1998), which is M (Narayan & Yi 1995; Mahadevan & very low for small Quataert 1997). We note that at these accretion rates, the spindown time for the hole, even including the Blandford- Znajek losses, remains much greater than a Hubble time (e.g. King & Kolb 1999). Our estimates suggest that the prospects for detecting isolated black holes are substantially better than would be inferred on the basis of the low radiative efficiency of an ADAF, if a reasonable fraction of the holes are rapidly rotating. Indeed, these estimates are energetically con- sistent with associating isolated black holes with some unidentified EGRET gamma ray sources (Lamb & Ma- comb 1997; Dermer 1997; Yadigaroglu & Romani 1997). Dermer (1997) estimates that these sources have charac- teristic luminosities, in the 100 MeV - 5 GeV band, of ∼ 2.5× 1035 ergs−1 (for low latitude sources with b < 10◦), and ∼ 6 × 1032 ergs−1 (for sources with b > 10◦). This would be roughly consistent with the estimates for rotating 10M⊙ black holes, provided that the efficiency of gamma ray emission was very high. High energy emission in AGN occurs via similar processes (e.g. Levinson & Blandford 1996). We note that the rough proportionality of the lu- minosity of the unidentified sources on the ISM density (Dermer 1997) would be consistent with the approximately constant efficiency of the Blandford-Znajek mechanism. It would not be consistent with an ADAF model, whose ra- diative efficiency declines at low M (Narayan & Yi 1995). 4. ADVECTION DOMINATED FLOWS IN GALACTIC NUCLEI Similar considerations apply to other low accretion rate flows that are advection dominated. The radiative effi- ciency of an ADAF is given approximately by, ǫADAF = 0.1(cid:18) m α2(cid:19) , (12) where m ≡ M / MEdd, and the Eddington accretion rate for an electron scattering opacity κes, and fiducial efficiency ηEff = 0.1, is defined as MEdd = 4πGM/ηEff κesc (Narayan & Yi 1995). For an estimated α = 0.1, this implies that the Blandford-Znajek effect could be an important con- tributor to the overall power output for m ∼< 2 × 10−3. In addition to isolated black holes in the galactic disk, some supermassive black holes in quiescent galactic nuclei also fall into this regime (Fabian & Rees 1995). The weak emission from the nuclei of these galaxies provides both a test of the ADAF model (Di Matteo et al. 1999a), and a possible contributor to the X-ray background (Di Mat- teo & Fabian 1997; Di Matteo et al. 1999b). Taking our estimates at face value, the good agreement of computed ADAF spectra with observations in some systems – espe- cially the galactic center (Mahadevan 1998) – would be consistent with the black holes in those systems not being close to maximally rotating. 5. DISCUSSION In this Letter, we have discussed the importance of the Blandford-Znajek (1977) mechanism for ADAFs at low accretion rates. For these thick disks, which are similar to the tori originally considered by Rees et al. (1982), the arguments advanced for the likely irrelevance of the Blandford-Znajek mechanism (Ghosh & Abramow- icz 1997; Livio, Ogilvie & Pringle 1999) are considerably weaker than for thin disks. Even with pessimistic assump- tions, the total electromagnetic power liberated from the hole or inner disk will exceed the radiative luminosity for sufficiently low accretion rates. For maximally rotating black holes, and self-similar ADAF models, we obtain the criteria m = M / MEdd ∼< 2 × 10−3. Additional Blandford- Znajek power could then be important for the spectra of ADAFs, both for supermassive black holes in the appar- ently quiescent cores of elliptical galaxies (Allen, Di Mat- teo & Fabian 1999), and for stellar mass black holes that are similarly starved of fuel. 4 Blandford-Znajek emission in low luminosity black hole systems For the extreme case of isolated stellar mass black holes accreting from the local ISM, these results imply that the detectability of any rapidly rotating holes is substantially better than estimates based on pure ADAF models (Fujita et al. 1998). The inferred luminosities are similar to those suggested for some unidentified EGRET sources (Dermer 1997), if we assume (very optimistically) that most of the power from the hole is ultimately radiated as gamma-ray emission. With better signal to noise observations, vari- ability is likely to provide additional constraints on mod- els for these EGRET sources. Isolated stellar mass black holes, like those in mass transfer binaries, will display vari- ability on all timescales exceeding the shortest dynamical timescales (of the order of milliseconds) available in the system. For holes accreting at high Mach numbers there may also be additional quasi-periodic variations arising from 'flip-flop' type instabilities in Bondi-Hoyle accretion, though the persistence of these features in three dimen- sional geometry (Ruffert 1997) as opposed to axisymmet- ric calculations (Benensohn, Lamb & Taam 1997), is not certain. These instabilities have characteristic timescales of ∼ Ra/c∞, the sound crossing time at the Bondi accre- tion radius. These variability signatures apply both to the high energy emission, and to emission in the optical wave- bands that may be detectable via current wide area sky surveys (Heckler & Kolb 1996). We thank Brad Hansen, Norm Murray, Gordon Ogilvie and the referee for useful discussions on these topics. REFERENCES Allen, S. W., Di Matteo, T., & Fabian, A. C., 1999, MNRAS, submitted, astro-ph/9905053 Armitage, P. J., 1998, ApJ, 501, L189 Balbus, S. A., & Hawley, J. F. 1991, ApJ, 376, 214 Begelman, M. C., Blandford, R. D., & Rees, M. J., 1984, Reviews of Mod. Phys., Vol. 56, Part I, p. 255 Benensohn, J. S., Lamb, D. Q., & Taam, R. E., 1997, ApJ, 478, 723 Blaes, O. & Madau, P., 1993, ApJ, 403, 690 Blandford, R. D., & Begelman, M. C., 1999, MNRAS, 303, L1 Blandford, R. D., & Znajek, R. L., 1977, MNRAS, 179, 433 Bondi, H., & Hoyle, F., 1944, MNRAS, 104, 273 Brandenburg, A., Nordlund, A., Stein, R. F., & Torkelsson, U. 1995, ApJ, 446, 741 Dermer, C. D., 1997, Proc. of the Fourth Compton Symp., eds. Dermer, Strickman & Kurfess, AIP Conf. Proc. 410, p. 1275. Di Matteo, T., & Fabian, A. C., 1997, MNRAS, 286, 393 Di Matteo, T., Esin, A., Fabian, A. C., & Narayan, R., 1999b, MNRAS, 305, L1 Di Matteo, T., Fabian, A. C., Rees, M. J., Carilli, C. L., & Ivison, R. J., 1999a, MNRAS, 305, 492 Fabian, A. C., & Rees, M. J., 1995, MNRAS, 277, L55 Fujita, Y., Inoue, S., Manmoto, T., & Nakamura, K. E., 1998, ApJ, 495, L85 Ghosh, P., & Abramowicz, M., 1997, MNRAS, 292, 887 Hawley, J. F., & Balbus, S. A., 1999, Astrophysical Discs, eds. Sellwood & Goodman, ASP Conf. Series, 160, p. 108 Hawley, J. F., Gammie, C. F., & Balbus, S. A. 1995, ApJ, 440, 742 Heckler, A. F., & Kolb, E. W., 1996, ApJ, 472, L85 Hoyle, F., & Lyttleton, R. A., 1939, Proc. Cambridge Philos. Soc., 35, 405 Ichimaru, S., 1977, ApJ, 214, 840 Ipser, J. R., & Price, R. H., 1982, ApJ, 255, 654 Ipser, J. R., & Price, R. H., 1983, ApJ, 267, 371 King, A. R., & Kolb, U., 1999, MNRAS, 305, 654 Lamb, R. C., & Macomb, D. J., 1997, ApJ, 488, 872 Lee, H. K., Wijers, R. A. M. J., & Brown, G., 1999, Physics Reports, submitted, astro-ph/9906213 Levinson, A., & Blandford, R. D., 1996, ApJ, 456, L29 Li, L.-X., 1999, ApJL, submitted, astro-ph/9902352 Livio, M., Ogilvie, G., & Pringle, J. E., 1999, ApJ, 512, 100 Lubow, S. A., Papaloizou, J., & Pringle, J. E., 1994, MNRAS, 267, 235 Macdonald, D. D., & Thorne, K. S., 1982, MNRAS, 198, 345 Mahadevan, R., & Quataert, E., 1997, ApJ, 490, 605 Mahadevan, R., 1998, Nature, 394, 651 Moderski, R., & Sikora, M., & Lasota, J.-P., 1998, MNRAS, 301, 142 Narayan, R., Garcia, M. R., & McClintock, J. E., 1997, ApJ, 478, L79 Narayan, R., & Yi, I., 1994, ApJ, 428, L13 Narayan, R., & Yi, I., 1995, ApJ, 452, 710 Neuhauser, R., & Trumper, J. E., 1999, A&A, 343, 151 Paczynski, B., 1998, ApJ, 494, L45 Parker, E. N., 1979, Cosmical Magnetic Fields. Clarendon Press, Oxford Phinney, E. S., 1983, Ph.D. Thesis, University of Cambridge Rees, M. J., Phinney, E. S., Begelman, M. C., & Blandford, R. D., 1982, Nature, 295, 17 Reyes-Ruiz, M., & Stepinski, T. F., 1996, ApJ, 459, 653 Ruffert, M., 1997, A&A, 317, 793 Shakura, N. I., & Sunyaev, R. A. 1973, A&A, 24, 337 Spruit, H. C., Matsuda, T., Inoue, M., & Sawada, K., 1987, MNRAS, 229, 517 Stone, J. M., Hawley, J. F., Gammie, C. F., & Balbus, S. A., 1996, ApJ, 463, 656 Thorne, K. S., Price, R. H., & Macdonald, D. D., 1986, Black Holes: The Membrane Paradigm. Yale Univ. Press, New Haven CN, p. 132 Walter, F. M., & Mathews, L. D., 1997, Nature, 389, 358 Wilson, A. S., & Colbert, E. J. M., 1995, ApJ, 438, 62 Yadigaroglu, I.-A., & Romani, R. W., 1997, ApJ, 476, 347 Zhang, S. N., Cui, W., & Chen W., 1997, ApJ, 482, L155
0704.2288
1
0704
2007-04-18T08:38:33
Suppressed star formation in circumnuclear regions in Seyfert galaxies
[ "astro-ph" ]
Feedback from black hole activity is widely believed to play a key role in regulating star formation and black hole growth. A long-standing issue is the relation between the star formation and fueling the supermassive black holes in active galactic nuclei (AGNs). We compile a sample of 57 Seyfert galaxies to tackle this issue. We estimate the surface densities of gas and star formation rates in circumnuclear regions (CNRs). Comparing with the well-known Kennicutt-Schmidt (K-S) law, we find that the star formation rates in CNRs of most Seyfert galaxies are suppressed in this sample. Feedback is suggested to explain the suppressed star formation rates.
astro-ph
astro-ph
THE ASTROPHYSICAL JOURNAL LETTERS: IN PRESS Preprint typeset using LATEX style emulateapj v. 11/12/01 SUPPRESSED STAR FORMATION IN CIRCUMNUCLEAR REGIONS IN SEYFERT GALAXIES JIAN-MIN WANG, YAN-MEI CHEN, CHANG-SHUO YAN, CHEN HU AND WEI-HAO BIAN Key Laboratory for Particle Astrophysics, Institute of High Energy Physics, CAS, 19B Yuquan Road, Beijing 100049, China Received 2007 January 17; accepted 2007 March 30 ABSTRACT Feedback from black hole activity is widely believed to play a key role in regulating star formation and black hole growth. A long-standing issue is the relation between the star formation and fueling the supermassive black holes in active galactic nuclei (AGNs). We compile a sample of 57 Seyfert galaxies to tackle this issue. We estimate the surface densities of gas and star formation rates in circumnuclear regions (CNRs). Comparing with the well-known Kennicutt-Schmidt (K-S) law, we find that the star formation rates in CNRs of most Seyfert galaxies are suppressed in this sample. Feedback is suggested to explain the suppressed star formation rates. Subject headings: galaxies: active -- galaxies: Seyfert -- galaxies: feedback 1. INTRODUCTION The implications of the well-known relations between black hole masses and bulge magnitudes (Magorrian et al. 1998), and the velocity dispersions (Gebhardt et al. 2000; Frarreasse & Merrit 2000) show a coevolution of the black holes and their host galaxies. However, how do black holes know the evolu- tion stage of the galaxies and how to control the growth of the black holes are currently understood via the the feedback from the black hole (Silk & Rees 1998; Croton et al. 2006; Schaw- inski et al. 2006). Numerical simulations show two roles of feedback from the black hole activity: (1) modulating the star formation rates; (2) heating the medium and finally quenching the black hole activity (Di Matteo et al. 2005). The direct evi- dence for the presence of the feedback from active black holes has to be shown from observations, yet. The main goal of the present paper is to show one piece of evidence for the feedback role in active galaxies. We show the AGN feedback domain, where starburst should be sup- pressed. We find the star formation rates in Seyfert galaxies is significantly lower than the rates predicted by the Kennicut- Schmidt's law. We use the cosmological parameters H0 = 75km s- 1 Mpc- 1, ΩM = 0.3 and ΩΛ = 0.7 throughout calcula- tions. 2. AGN FEEDBACK DOMAIN - 1 M⊙ pc- 2, When the CNR medium is optically thick, namely, the opti- cal depth τ = κabsΣgas ≥ 1, where κabs is opacity and Σgas the gas surface density, the radiation from the black hole activity will continuously heat the medium and blow the gas away so as to lower the star formation rates. The condition of τ = 1 yields a critical density gas = 9.0 × 102(cid:0)κabs/5(cid:1) Σc1 (1) κabs has a mean value of 5 for the CNR medium (Semenov et al. 2003). This is feedback driven by AGN radiation. We note outflows from Seyfert active nucleus have much low kinetic luminosities, typically ∼ 10- (3- 6)LBol based on X-ray warmer absorbers (Blustin et al. 2005), where LBol is the bolometric luminosity. Feedback from outflows could be thus neglected. When Σgas > Σc1 gas, the AGN radiation-driven feedback will sup- press the star formation. On the other hand, AGN feedback reaches its maximum when an AGN radiates at the Eddington limit LAGN = LEdd = 1.3 × 1038(M•/M⊙)erg/s. In the case of LEdd ≤ LIR SFR, AGN have inefficient feedback to star formation. 1 Σc2 gas is gas = 8.2 × 105M0.7 With the help of SFR = 4.5(cid:0)LIR/1044erg s- 1(cid:1) M⊙yr- 1, Σc2 given by using the K-S law ΣSFR = AΣγ gas (Kennicutt 1998a), 9 R- 1.4 (2) where ΣSFR = SFR/πR2 is the surface density of the star forma- tion rate, A = 2.5 × 10- 4, γ = 1.4, M9 = M•/109M⊙ is the black hole mass and R200 = R/200pc the size of the circumnuclear star forming region. When Σgas ≥ Σc2 gas, the gas is so dense that the luminosity from star formation dominates over the AGN. We call 200 M⊙ pc- 2, Σc1 gas ≤ Σgas ≤ Σc2 gas, (3) the AGN feedback domain as shown in Fig. 1, in which the K-S law is broken. The strong radiation pressure from the black hole accretion disk at Eddington limit is PAGN ≈ 1.0 × 10- 7M8R- 2 200dyn cm- 2, where M8 = M•/108M⊙. The pressure from supernovae explo- sion is PSN ≈ ǫ ΣSFRc = 2.0 × 10- 8 ΣSFR,2ǫ- 3dyn cm- 2, where ΣSFR,2 = ΣSFR/102M⊙yr- 1kpc- 2 and ǫ- 3 = ǫ/10- 3 is the ef- ficiency converting the mass into radiation (Thompson et al 2005). We find PAGN ≥ 5PSN within CNRs of radius ∼ 200pc for typical values of the parameters of ǫ, M• and ΣSFR. This indicates that the radiation from AGN dominates over the local feedback from supernovae explosion. After an AGN switches on, the star formation is suppressed and then feedback from supernovae is further weakened. The timescale of the AGN feedback to the starburst regions can be estimated by tFB ∼ Egas/ fFBCLAGN, where LAGN is AGN luminosity, C = ∆Ω/4π is the covering factor, the thermal energy Egas ≈ kT Mgas/mp, k is the Boltzmann constant, mp is the proton mass, T is the gas temperature, Mgas = πR2Σgas is the gas mass and fFB is the feedback efficiency. We have tFB ∼ 2.6 × 105 f - 1 (4) 43 yr, where Σgas,4 = Σgas/104M⊙ pc- 2, T3 = T /2 × 103K, fFB,- 2 = fFB/10- 2, L43 = LAGN/1043erg s- 1 and C0.5 = C/0.5 is the cov- ering factor of the CNRs. Such a short timescale indicates that the AGN feedback is very efficient. This is supported by a large fraction of the post-starburst AGNs in a large Sloan Digital Sky Survey sample (Kauffmann et al. 2003). The physics behind K-S law is not sufficiently understood (Thompson et al. 2005; Krumholz & McKee 2005). It is beyond the scope of the present paper to give a quantitative description of the suppressed star formation rates. Comparing Seyfert galaxies with the K-S law, 200T3Σgas,4C FB,- 2R2 0.5L- 1 - 1 2 Wang et al. TABLE 1 THE SEYFERT GALAXY SAMPLE Seyfert 1 Redshift FWHM log λLλ Object (1) 3C120 IC4329 MCG-2-33-34 MCG-5-13-17 NOTE.- Table 1 is published in its entirety in the electronic edition. A portion is shown here for guidance regarding its form and content. log M• (6) 7.74a 6.99a 6.11 7.50 Ref. (5) 2 2 22, 5 20, 1 SPAH (9) 76 220 54 67 (2) 0.033 0.016 0.014 0.013 (4) 44.17 43.32 42.61 43.44 M• (7) 0.24 0.03 0.01 0.04 R (10) 0.48 0.24 0.21 0.19 (12) 1.32 1.35 1.21 1.24 (11) 0.85 1.31 0.70 0.79 (8) 4.18 3.70 3.13 3.92 1565 4000 (3) ... ... log Σgas log ΣSFR athe blackhole mass are directly taken from Peterson et al. (2004). brefers to [O III] FWHM. cbased on M• - Mbulge relation, F01475-0740: Mbulge = - 18.80; NGC 3660: Mbulge = - 18.38. Note.-(1) source name; (2) redshift; (3) FWHM of Hβ for Seyfert 1s or stellar velocity dispersion σ for Seyfert 2s (in km s- 1); - 0.5 or [O III]λ5007Å (in erg s- 1); (5) references for columns (4) luminosity of 5100Å deduced from extrapolation of Fν ∝ ν (3) and (4) are given below, respectively; (6) black hole mass (in M⊙); (7) accretion rate (in M⊙yr- 1); (8) gas surface density (in M⊙pc- 2); (9) surface brightness of the 3.3 µm PAH emission feature (in unit of 1039ergs s- 1kpc- 2); (10) the scale of the starburst regions (in kpc); (11) and (12) are the lower ( ΣL SFR) limits of surface density of star formation rates, respectively (in M⊙yr- 1kpc- 2). Reference.-(1) NED; (2) Peterson et al. (2004); (3) Spinogilio et al. (1995); (4) Doroshenko & Terebezh (1979); (5) Kinney et al. (1993); (6) Nelson & Whittle (1995); (7) Dahari & Robertis (1988); (8) Lipari et al. (1991); (9) Corral et al. (2005); (10) Kirhakos & Steiner (1990); (11) Visvanathan & Griersmith (1977); (12) Cid Fernandes et al. (2004); (13) Gu & Huang (2002); (14) Kailey & Lebofsky (1988); (15) Heraudeau & Simien (1998); (16) Bassani (1999); (17) Whittle et al. (1988); (18) Whittle (1992); (19) Garcia-Rissmann et al. (2005); (20) Crenshaw et al. (2003); (21) Marzini et al. (2003); (22) Veron-Cetty et al. (2001); (23) Postman & Lauer (1995). SFR) and upper ( ΣU a universal rule of cosmic star formation, we may get the un- dergoing physics in the CNRs. We have to point out here that the short feedback time does NOT mean the same timescale of the starburst. The present tFB means the starburst rates will be suppressed once AGN is triggered and make it possible for AGN and starburst coexist. 3. APPEARANCE OF FEEDBACK IN SEYFERT GALAXIES For the goal to test the above scenario, we compile 57 Seyfert galaxies (Imanishi 2002; Imanishi 2003; Imanishi & Wada 2004). The star formation rates in CNRs of Seyfert galaxies can be traced by several indicators, particularly, PAH features at 3.3, 6.2, 7.7, 8.6 and 11.2µm, which radiate from vibration of PAH grains containing about 50 carbon atoms. Among the fea- tures, 3.3µm emission is intrinsically strong and less affected by broad silicate dust absorption (Imanishi 2002). We choose 3.3µm emission as an indicator of the star formation rate. We convert the PAH emission into IR luminosity via LIR = 103LPAH relation with a scatter by a factor of 2-3 for pure star formation (Imanishi 2002). Since some PAH grains would be destroyed by EUV and X-ray photons from the central engine, we have the lower limit of the surface density of the star formation rates (5) by using the relation of the star formation rate and the in- frared luminosity (eq. 7) (Kennicutt 1998a), where LPAH,41 = LPAH/1041erg s- 1. On the other hand, the infrared emission from Seyfert galaxies covers the contribution from starburst and reprocessing radiation from AGNs, we have the upper limit of the surface density of the star formation rates SFR = 35.8LPAH,41R- 2 ΣL 200 (M⊙yr- 1kpc- 2), 200 (M⊙yr- 1kpc- 2), SFR = 35.8LFIR,44R- 2 ΣU (6) where LFIR,44 = LFIR/1044erg s- 1 is the observed far-IR luminos- ity. We take the geometric average ΣSFR = (cid:16) ΣL and the error bars correspond to ΣL SFR. We have to stress SFR and ΣU SFR(cid:17)1/2 ΣU SFR SFR and ΣL this average only represents the central value of logarithm of ΣU SFR and the upper and lower limits of ΣSFR are the most important. Table 1 gives the sample of Seyfert galaxies, which have been observed through IRTF SpeX or Subaru IRCS with spatial resolution of 0.9′′ - 1.6′′. For Seyfert 1 galaxies, we estimate LBol = 9L5100, where L5100 is the luminosity at 5100Å and then the accretion rate M• = LBol/ηc2, where η = 0.1 is the accretion efficiency. The black hole masses are estimated from the empirical relation of reverberation mapping (Kaspi et al. 2000), or directly taken from the mapping (Peterson et al. 2004). We assume that the potential of the total mass within the CNRs controls the massive disk fueling to the black hole, where star formations are taking place either. Assuming the Keplerian rotation, the surface den- sity of the disk is - 4/5 0.1 M3/5 •,1 f Σtot = 2.1 × 106α - 3/5 - 1/5 200 M⊙ pc- 2, • M1/5 (7) 8 R given by the disk model (King et al. 2002; Yi & Black- man 1994; Tan 2005), where the opacity κabs = 5 in this re- gion, f• is the ratio of the black hole mass to the total, M•,1 = M•/1.0M⊙yr- 1 and α0.1 = α/0.1 is the viscosity (Shakura & Sunyaev 1973). This estimation is the lower limit since we re- place the infalling mass rates in CNRs by black hole accretion rates. The gas surface density of the disk Σgas = fgΣtot Σgas = 1.0 × 105 fg,0.05α - 3/5 200 M⊙ pc- 2, (8) where the gas fraction to the to- fg,0.05 = fg/0.05 is tal. the Considering disk the bulge, fg = Mgas/Mdisk > Mgas/MBulge = we (cid:0)Mgas/Mdust(cid:1)(cid:0)Mdust/M•(cid:1)(cid:0)M•/MBulge(cid:1), where Mdisk is the to- tal mass of the disk, Mgas/Mdust is the gas-to-dust mass ra- tio and MBulge ≈ 103M• is the bulge mass (Kormendy & Gebhardt 2001; McLure & Dunlop 2002). It has been the dust mass in PG quasars is comparable found that with in Seyfert galaxies (Spinoglio et al. 2002; Haas et - 1/5 • M1/5 8 R M3/5 •,1 f located - 4/5 0.1 inside have is Suppressed Starburst in Seyfert Galaxies 3 c2 gas. The Compton FIG. 1. -- The plot of gas and star formation rate surface densities. The yellow region is the AGN feedback domain given by Σ thick region has Σgas ≥ 8.0 × 103M⊙ pc- 2 (i.e. NH ≥ 1024cm- 2). The red squares are starburst galaxies taken from Kennicutt (1998b). The cyan and blue-magenta stars are Seyfert 1 and 2 galaxies, respectively. The blue star is NGC 3227, in which the star formation rate is 0.05M⊙ yr- 1 and the gas mass Mgas = (2- 20) ×108M⊙ within 65pc taken from Davies et al. (2006). c1 gas ≤ Σgas ≤ Σ 2003). The mean value of gas-to-dust mass ratio is al. hMgas/Mdusti ∼ 250 (Haas et al. 2003). We estimated dust L(cid:2)exp(143.38/Tdust) - 1(cid:3)M⊙, mass from Mdust ∼ 4.78 f100µD2 temperature is estimated by Tdust = (1 + and the dust z)(cid:2)0.5 - 82/ ln(0.3 f60µ/ f100µ)(cid:3)K, where f100µ and f60µ are the fluxes at 100µm and 60µm in unit of Jy, respectively, DL is the luminosity distance in unit of Mpc (Evans et al. 2005). We find the mean value of hMdust/M•i = 0.2 ± 0.2 in our sample. So we have fg ≥ 0.05 as a lower limit in this paper. Thompson et al. (2005) used fg = 0.1. We note Σgas ∝ f , resulting in uncertainties of Σgas by a factor of 4 for f• = 10- 3 - 1. α = 0.1 is used for all Seyfert galaxies. - 1/5 • For Seyfert 2 galaxies, dusty tori obscure the active regions. We estimate the bolometric luminosity from LBol = 3500L[O III] with a mean uncertainty of 0.38 dex (Heckman et al. 2004), where L[O III] is the [O III]λ5007 luminosity, and hence the black hole accretion rates. The black hole masses are estimated through the M• - σ relation (Tremaine et al. 2002), where the dispersion velocity σ = FWHM([O III])/2.35 if the dispersion velocity is not available. Fig. 1 shows the Σgas - gas > Σgas > Σc1 ΣSFR plot of Seyfert CNRs. We find that CNR gas surface densities of Seyfert galaxies are lo- cated within the AGN feedback domain. There are clearly three branches in the figure, separating the Seyfert galaxies, when Σc2 gas. Seyfert galaxies marked in Zone I still sat- isfy the K-S law. Those (Mrk 273, Mrk 938, NGC 5135 and NGC 1068) marked in Zone II are located between the K-S law and Zone III. These are ultra-luminous infrared galaxies, or mixed with strong starbursts. The main energy sources in the CNRs are in a transition state from a starburst to an AGN in these galaxies. The fraction of the transiting galaxies is only 4/57 ∼ 1/10. Though the completeness of the present sample is uncertain, this fraction implies that the transition is quite short and indicated by the feedback timescale from equation (4). The Seyfert galaxies in Zone III are undergoing suppressed star for- mation strongly, being 1-2 orders lower than that predicted by the K-S law. The suppressed ΣSFR is obviously caused by the feedback. Galaxies obeying the K-S law are powered by nu- clear energy from stars, however gravitational energy released from accretion onto the black holes is powering AGNs if a tran- sition from starburts to active galaxies happens. With the dis- sipation of CNR gas due to star formation and accretion onto the black holes, Σgas is decreasing and the galaxies may return to the K-S law once AGNs switch off. Such a behavior likes evolution of stellar energy sources in the Hertzprung-Russell diagram. It has been found that black hole duty cycles follow the his- tory of star formation rate density (Wang et al. 2006). The above scenario then implies that both the black hole activities and starbursts are episodic (Davies et al. 2006). The multiple cycles of the black holes and starbursts make it impossible to measure the time delay between the two episodes. However the stellar synthesis may tell the star formation history and then give the black hole activity history. 4. CONCLUSIONS AND DISCUSSIONS We find direct evidence for the feedback from active black holes in Seyfert galaxies. Once a black hole is triggered, the feedback will significantly suppress the starbursts within a quite short timescale of a few 105years. The duty cycles of Seyfert galaxies strongly indicate there is an efficient way to frequently trigger black holes and quench starbursts. The data presented in this paper are only lower limits of the gas densities. Future VLT (Very Large Telescope) and ALMA (Atacama Large Millimiter Array) measurements of the star formation rates and gas densities will finally identify roles of the feedback from the black hole activities. 4 Wang et al. The helpful comments from the referee are acknowledged. The authors are grateful to R. C. Kennicutt, L. C. Ho, S. N. Zhang and X.-Y. Xia for useful discussions. We appreciate the stimulating discussions among the members of IHEP (Insti- tute of High Energy Physics) AGN group. J.-M.W. thanks the Natural Science Foundation of China for support via NSFC- 10325313 and 10521001, CAS key project via KJCX2-YW- T03. REFERENCES Bassani, L., et al. 1999 ApJS, 121, 473 Blustin, A. J., Page, M. J., Fuerst, S. V., Branduardi-Raymont, G., Ashton, C. E. 2005, A&A, 431, 111 King, A., Frank, J. & Raine, D. J., 2002, Accretion Power in Astrophysics, Cambridge University Press, p.90 Kinney, A. L., Bohlin, R. C., Calzetti, D., Panagia, N., & Wyse, R. F. G. 1993 Cid Fernandes, R., et al. Rodrigues Lacerda, R.; Joguet, B. 2004 MNRAS, 355, ApJS, 86, 5 273 431, 97 Crenshaw, D. M., Kraemer, S. B. & Gabel, J. R. 2003 ApJ, 126, 1690 Croton, D. J., Springel, V., White, S. D. M., De Lucia, G., Frenk, C. S., Gao, L., Jenkins, A., Kauffmann, G., Navarro, J. F., Yoshida, N. 2006, MNRAS, 365, 11 Corral, A., Barcons, X., Carrera, F. J., Ceballos, M. T., Mateos, S. 2005 A&A, Dahari, O. & Robertis, M. M. D. 1988 ApJS, 67, 249 Davies, R. I., et al. 2006, ApJ, 646, 754 Doroshenko, V. T. & Terebezh, V. Yu. 1979 SvAL, 5, 305 Di Matteo, T., Springel, V., Hernquist, L. 2005, Nature, 433, 604 Evans, A. S., Mazzarella, J. M., Surace, J. A., Frayer, D. T., Iwasawa, K. & Sanders, D. B. 2005, ApJS, 159, 197 Ferrarese, L. & Merritt, D., 2000, ApJ, 539, L9 Garcia-Rissmann, A., Vega, L. R.; Asari, N. V.; Cid Fernandes, R.; Schmitt, H.; Gonzoulez Delgado, R. M.; Storchi-Bergmann, T. 2005 MNRAS, 359, 765 Gebhardt, K. et al., 2000, ApJ, 543, L5 Gu, Q. & Huang, J. 2002 ApJ, 579, 205 Haas, M., et al. 2003, A&A, 402, 87 Heckman, T. M. et al. 2004, ApJ, 613, 109 Heraudeau, P. & Simien, F. 1998 A&AS, 133, 317 Imanishi, M., 2002, ApJ, 569, 44 Imanishi, M. 2003, ApJ, 599, 918 Imanishi, M. & Wada, K. 2004, ApJ, 617, 214 Kailey, W. F. & Lebofsky, M. J. 1988 ApJ, 326, 653 Kaspi, S., Smith, P. S., Netzer, H., Maoz, D., Jannuzi, B. T., Giveon, U., 2000, ApJ, 533, 631 Kauffmann, G., et al. 2003, MNRAS, 346, 1055 Kennicutt, R. C., Jr. 1998a, ARA&A, 36, 189 Kennicutt, R. C. Jr. 1998b, ApJ, 498, 541 Kirhakos, S. D. & Steiner, J. E. 1990 AJ, 99, 1722 Kormendy, J. & Gebhardt, K., 2001, in The 20th Texas Symposium on Relativistic Astrophysics, ed. H. Martel & J.C. Wheeler, AIP, (astro-ph/0105230) Krumholz, M. R. & McKee, C. F. 2005, ApJ, 630, 250 Lipari, S., Bonatto, C. & Pastoriza, M. 1991 MNRAS, 253, 19 Magorrian, J. et al., 1998, AJ, 115, 2285 Marzini, P., Sulentic, J. W., Zamanov, R., Calvani, M. & Dultzin-Hacyan, D., 2003 ApJS, 145, 199 McLure, R. J. & Dunlop, J. S. 2002, MNRAS, 331, 795 Nelson, C. H. & Whittle, M. 1995 ApJS, 99, 67 Peterson, B. M. et al. 2004, ApJ, 613, 682 Postman, M., Lauer, T. R. 1995 ApJ, 440, 28 Silk, J. & Rees, M. J. 1998, A&A, 331, L1 Schawinski, K., et al. 2006, Nature, 442, 888 Semenov, D., Henning, Th., Helling, Ch., Ilgner, M., Sedlmayr, E. 2003, A&A, Shakura, N. I. & Sunyaev, R. A. 1973, A&A, 24, 337 Spinoglio, L., Andreani, P., Malkan, M. A. 2002, ApJ, 572, 105 Spinoglio, L., Malkan, M. A., Rush, B., Carrasco, L. & Recillas-cruz, E. 1995 410, 611 ApJ, 453, 616 Tan, J. C. & Blackman, E. G. 2005, MNRAS, 362, 983 Thompson, T. A., Quataert, E., Murray, N. 2005, ApJ, 630, 167 Tremaine, S. et al. 2002, ApJ, 574, 740 Veron-Cetty, M.-P., Veron, P. & Gongalves, A.C. 2001 A&A, 372, 730 Visvanathan, N., Griersmith, D. 1977 A&A, 59, 317 Wang, J.-M., Chen, Y.-M. & Zhang, F. 2006, ApJ, 647, L17 Whittle, M. 1992 ApJS, 79, 49 Whittle, M., Pedlar, A., Meurs, E. J. A., Unger, S. W., Axon, D. J. & Ward, M. J. 1988 ApJ, 326, 125 Yi, I., Field, G. B. & Blackman, E. G. 1994, ApJ, 432, L31 Suppressed Starburst in Seyfert Galaxies Table 1 The Seyfert Galaxy Sample Seyfert 1 5 log ΣSFR (11) 0.85 1.31 0.70 0.79 0.82 0.52 0.76 0.79 0.61 0.47 0.86 0.64 0.48 0.75 0.33 0.66 1.33 0.75 0.58 1.56 (11) 0.66 0.74 0.66 1.02 0.66 0.94 0.97 1.54 1.39 0.88 1.10 0.66 1.72 0.44 1.04 0.57 1.26 0.57 0.27 0.92 0.44 1.09 0.66 0.92 0.66 0.66 1.04 1.52 0.87 1.30 0.92 0.80 0.44 0.27 1.06 1.04 0.44 (12) 1.32 1.35 1.21 1.24 1.31 0.68 1.24 1.15 1.59 0.76 1.42 1.70 0.95 1.55 1.58 0.66 2.10 1.14 1.10 3.11 (12) 0.99 1.51 1.37 2.21 1.87 1.20 1.37 2.66 2.19 1.67 1.49 1.16 2.28 0.82 1.19 1.47 2.55 0.85 1.92 1.44 2.10 2.00 1.43 1.30 2.12 2.43 1.45 2.59 1.30 2.23 1.31 1.87 1.14 2.09 2.16 1.88 0.76 log ΣSFR Object (1) 3C120 IC4329 MCG-2-33-34 MCG-5-13-17 Mrk79 Mrk335 Mrk509 Mrk530 Mrk618 Mrk704 Mrk817 Mrk1239 NGC863 NGC931 NGC2639 NGC4235 NGC4253 NGC5548 NGC5940 NGC7469 Object (1) F01475-0740 F04385-0828 F15480-0344 IC3639 MCG-3-34-64 Mrk34 Mrk78 Mrk273 Mrk334 Mrk463 Mrk477 Mrk573 Mrk938 Mrk993 NGC262 NGC513 NGC1068 NGC1194 NGC1241 NGC1320 NGC1667 NGC2992 NGC3660 NGC3786 NGC4388 NGC4501 NGC4968 NGC5135 NGC5252 NGC5256 NGC5347 NGC5674 NGC5695 NGC5929 NGC7172 NGC7674 NGC7682 Redshift (2) 0.033 0.016 0.014 0.013 0.022 0.025 0.035 0.029 0.035 0.030 0.031 0.019 0.027 0.016 0.011 0.008 0.013 0.017 0.034 0.016 Redshift (2) 0.017 0.015 0.030 0.011 0.017 0.015 0.037 0.038 0.022 0.051 0.038 0.017 0.019 0.015 0.015 0.020 0.004 0.013 0.014 0.010 0.015 0.008 0.012 0.009 0.008 0.008 0.010 0.014 0.023 0.028 0.008 0.025 0.014 0.008 0.009 0.029 0.017 FWHM (3) ... ... 1565 4000 ... ... ... 6560 3018 5684 1075 ... ... 1830 3100 7600 1630 5240 ... ... σ (3) ... 907b 664b 95 155 570b 172 211 250b 545b 370b 123 330b 392b 118 152 151 396b 136 116 173 158 ... 142 119 171 105 128 190 315b 93 129 144 121 154 144 123 log λLλ (4) 44.17 43.32 42.61 43.44 43.72 43.86 44.28 44.04 44.00 43.53 43.82 43.84 43.81 43.70 43.77 43.51 43.41 43.51 44.07 43.72 log L[O III] (4) 41.69 40.12 42.95 42.11 42.32 43.04 42.22 42.39 41.29 43.44 43.54 42.00 42.77 40.87 41.91 40.60 42.65 40.84 42.47 40.96 41.91 41.92 40.91 41.52 41.68 39.80 42.37 42.31 41.96 41.85 40.45 41.87 40.50 40.96 39.77 42.49 41.72 Ref. (5) 2 2 22, 5 20, 1 2 2 2 20, 1 21, 4 21, 1 2 22, 3 2 20, 1 20, 1 20,11 20, 4 2 20, 1 2 Seyfert 2 Ref. (5) 13 7,13 8,13 19,13 12,13 7, 7 6,17 19,16 7,13 18,18 18,13 6,13 7,13 9,7 6,13 6,13 6,13 10,14 19,13 6,13 6,13 6,16 13 6, 7 6,13 15,13 19,16 19,13 6,13 7,13 6,13 19,16 6,13 6,13 19,13 6,13 6,13 log M• (6) 7.74a 6.99a 6.11 7.50 7.72a 7.15a 8.15a 8.35 7.65 7.88 7.69a 6.65 7.68a 7.01 7.51 8.11 6.70 7.83a 8.18 7.08a log M• (6) 7.55c 8.77 8.22 6.83 7.69 7.96 7.87 8.22 6.52 7.88 7.20 7.28 7.00 7.30 7.21 7.65 7.64 7.32 7.46 7.18 7.88 7.72 7.33c 7.53 7.22 7.86 7.01 7.35 8.04 6.92 6.79 7.36 7.56 7.25 7.67 7.56 7.28 M• (7) 0.24 0.03 0.01 0.04 0.08 0.14 0.30 0.17 0.16 0.05 0.11 0.11 0.10 0.08 0.09 0.05 0.04 0.05 0.19 0.08 M• (7) 0.30 0.01 5.46 0.80 1.30 6.78 1.03 1.52 0.12 16.82 21.57 0.62 3.66 0.05 0.51 0.02 2.73 0.04 1.83 0.06 0.51 0.52 0.05 0.20 0.30 0.01 1.44 1.26 0.57 0.44 0.02 0.45 0.02 0.06 0.01 1.93 0.32 log Σgas (8) 4.18 3.70 3.13 3.92 4.00 3.94 4.31 4.25 4.04 3.84 3.97 4.02 4.00 3.93 4.17 4.22 3.87 3.96 4.20 4.14 log Σgas (8) 4.36 3.71 5.12 4.66 4.79 5.36 4.63 4.79 4.00 5.28 5.28 4.52 4.88 3.86 4.47 3.69 5.10 3.89 4.91 4.00 4.64 4.74 3.94 4.43 4.49 3.49 4.79 4.82 4.56 4.29 3.67 4.49 3.75 4.08 3.47 4.72 4.35 SPAH (9) 76 220 54 67 71 36 62 67 44 32 78 47 33 61 23 49 230 61 41 390 SPAH (9) 50 60 50 113 50 95 102 377 265 81 135 50 570 30 120 40 198 40 20 90 30 133 50 91 50 50 120 360 80 214 90 68 30 20 125 120 30 R (10) 0.48 0.24 0.21 0.19 0.32 0.37 0.51 0.42 0.51 0.44 0.45 0.17 0.40 0.24 0.16 0.12 0.12 0.26 0.49 0.12 R (10) 0.26 0.22 0.43 0.13 0.24 0.17 0.41 0.42 0.19 0.56 0.42 0.24 0.28 0.22 0.22 0.30 0.15 0.19 0.18 0.15 0.19 0.12 0.19 0.13 0.12 0.12 0.15 0.16 0.33 0.31 0.12 0.21 0.18 0.12 0.10 0.42 0.24 6 Wang et al. athe blackhole mass are directly taken from Peterson et al. (2004). brefers to [O III] FWHM. cbased on M• - Mbulge relation, F01475-0740: Mbulge = - 18.80; NGC 3660: Mbulge = - 18.38. Note.-(1) source name; (2) redshift; (3) FWHM of Hβ for Seyfert 1s or stellar velocity dispersion σ for Seyfert 2s (in km s- 1); - 0.5 or [O III]λ5007Å (in erg s- 1); (5) references for columns (4) luminosity of 5100Å deduced from extrapolation of Fν ∝ ν (3) and (4) are given below, respectively; (6) black hole mass (in M⊙); (7) accretion rate (in M⊙yr- 1); (8) gas surface density (in M⊙pc- 2); (9) surface brightness of the 3.3 µm PAH emission feature (in ×1039ergs s- 1kpc- 2); (10) the scale of the starburst regions (in kpc); (11) and (12) are the lower ( ΣL SFR) limits of surface density of star formation rates, respectively (in M⊙yr- 1kpc- 2). SFR) and upper ( ΣU Reference.-(1) NED; (2) Peterson et al. (2004); (3) Spinogilio et al. (1995); (4) Doroshenko & Terebezh (1979); (5) Kinney et al. (1993); (6) Nelson & Whittle (1995); (7) Dahari & Robertis (1988); (8) Lipari et al. (1991); (9) Corral et al. (2005); (10) Kirhakos & Steiner (1990); (11) Visvanathan & Griersmith (1977); (12) Cid Fernandes et al. (2004); (13) Gu & Huang (2002); (14) Kailey & Lebofsky (1988); (15) Heraudeau & Simien (1998); (16) Bassani (1999); (17) Whittle et al. (1988); (18) Whittle (1992); (19) Garcia-Rissmann et al. (2005); (20) Crenshaw et al. (2003); (21) Marzini et al. (2003); (22) Veron-Cetty et al. (2001); (23) Postman & Lauer (1995). 1. NASA/IPAC Extragalatic Database 2. Peterson, B. M., et al. 2004 ApJ, 613, 682 3. Spinoglio, L., Malkan, M. A., Rush, B., Carrasco, L. & Recillas-cruz, E. 1995 ApJ, 453, 616 4. Doroshenko, V. T. & Terebezh, V. Yu. 1979 SvAL, 5, 305 5. Kinney, A. L., Bohlin, R. C., Calzetti, D., Panagia, N., & Wyse, R. F. G. 1993 ApJS, 86, 5 6. Nelson, C. H. & Whittle, M. 1995 ApJS, 99, 67 7. Dahari, O. & Robertis, M. M. D. 1988 ApJS, 67, 249 8. Lipari, S., Bonatto, C. & Pastoriza, M. 1991 MNRAS, 253, 19 9. Corral, A., Barcons, X., Carrera, F. J., Ceballos, M. T., Mateos, S. 2005 A&A, 431, 97 10. Kirhakos, S. D. & Steiner, J. E. 1990 AJ, 99, 1722 11. Visvanathan, N., Griersmith, D.1977 A&A, 59,317 12. Cid Fernandes, R., et al. 2004 MNRAS, 355, 273 13. Gu, Q. & Huang, J. 2002 ApJ, 579, 205 14. Kailey, W. F. & Lebofsky, M. J. 1988 ApJ, 326, 653 15. Heraudeau, P. & Simien, F. 1998 A&AS, 133, 317 16. Bassani, L., et al. 1999 ApJS, 121, 473 17. Whittle, M., et al. 1988 ApJ, 326, 125 18. Whittle, M. 1992 ApJS, 79, 49 19. Garcia-Rissmann, A., et al. 2005 MNRAS, 359, 765 20. Crenshaw, D. M., Kraemer, S. B. & Gabel, J. R. 2003 ApJ, 126, 1690 21. Marzini, P., et al. 2003 ApJS, 145, 199 22. Veron-Cetty, M.-P., Veron, P. & Gongalves, A.C. 2001 A&A, 372, 730 23. Postman, M., & Lauer, T. R. 1995 ApJ, 440, 28
astro-ph/0310149
1
0310
2003-10-06T16:11:33
Extraction of cluster parameters from Sunyaev-Zeldovich effect observations with simulated annealing optimization
[ "astro-ph" ]
We present a user-friendly tool for the analysis of data from Sunyaev-Zeldovich effect observations. The tool is based on the stochastic method of simulated annealing, and allows the extraction of the central values and error-bars of the 3 SZ parameters, Comptonization parameter, y, peculiar velocity, v_p, and electron temperature, T_e. The f77-code SASZ will allow any number of observing frequencies and spectral band shapes. As an example we consider the SZ parameters for the COMA cluster.
astro-ph
astro-ph
Extraction of cluster parameters from Sunyaev-Zeldovich effect observations with simulated annealing optimization Steen H. Hansen University of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland Abstract We present a user-friendly tool for the analysis of data from Sunyaev-Zeldovich effect observations. The tool is based on the stochastic method of simulated annealing, and allows the extraction of the central values and error-bars of the 3 SZ parameters, Comptonization parameter, y, peculiar velocity, vp, and electron temperature, Te. The f77-code SASZ will allow any number of observing frequencies and spectral band shapes. As an example we consider the SZ parameters for the COMA cluster. 1 Introduction Galaxy clusters typically have temperatures of the order keV, Te = 1 − 15 keV, and a CMB photon which traverses the cluster and happens to Compton scatter off a hot electron will therefore get increased momentum. This up- scattering of CMB photons, which results in a small change in the intensity of the cosmic microwave background, is known as the Sunyaev-Zeldovich (SZ) effect, and was predicted just over 30 years ago (Sunyaev & Zeldovich 1972). The first radiometric observations came few years later (Gull & Northover 1976, Lake & Partridge 1977), and while recent years have seen an impressive improvement in observational techniques and sensitivity (Laroque et al. 2002, De Petris et al 2002), then the near future observations will see another boost in sensitivity by orders of magnitude. These include dedicated multi-frequency SZ observations like ACT 1 and SPT 2 . The SZ effect will thus soon provide us with an independent description of cluster properties, such as evolution and radial profiles. For recent excellent reviews see (Birkinshaw 1999, Carlstrom, Holder & White 2002). 1 http://www.hep.upenn.edu/∼angelica/act/act.html 2 http://astro.uchicago.edu/spt/ Preprint submitted to Elsevier Science 30 October 2018 The SZ effect is traditionally separated into two components according to the origin of the scattering electrons ∆I(x) I0 = ∆Ithermal (x, y, Te) + ∆Ikinetic (x, τ, vp, Te) (1) = y (g(x) + δT (x, Te)) − βτ (cid:16)h(x) + δkin(x, Te)(cid:17) , with x = hν/kTcmb and I0 = 2(kTcmb)3/(hc)2 where Tcmb = 2.725 K. The first term on the rhs of eq. (1) is the thermal distortion with the non-relativistic spectral shape g(x) = x4 ex (ex − 1)2 (cid:18)x ex + 1 ex − 1 − 4(cid:19) , and the magnitude is given by the Comptonization parameter y = σT me c2 Z dl ne kTe , (2) (3) where me and ne are masses and number density of the electrons, and σT is the Thomson cross section. For non-relativistic electrons one has δT (x, Te) = 0, but for hot clusters the relativistic electrons will slightly modify the thermal SZ effect (Wright 1979). These corrections are easily calculated (Rephaeli 1995, Itoh & Nozawa 2003, Ensslin & Kaiser 2000, Dolgov et al 2001), and can be used to measure the cluster temperature purely from SZ observations (Hansen, Pastor & Semikoz 2002). For the implementation below we will use an extension of the method developed in Aghanim et al. (2003), using a fit to the spectral shape of δT (x, Te), which everywhere in the range 20 − 900 GHz and Te < 24 keV is very accurate, δf it T − δT/(δT + g) < 0.005. In the range 24 keV < Te < 100 keV the accuracy is slightly lower. The kinetic distortions have the spectral shape h(x) = x4 ex (ex − 1)2 , (4) and the magnitude depends on β = vp/c, the average line-of-sight streaming velocity of the thermal gas (positive if the gas is approaching the observer), and the Thomson optical depth τ = σTeZ dl ne . 2 (5) Thus, when the intra-cluster gas can be assumed isothermal one has y = τ kTe/(mc2). For large electron temperatures there are also small corrections to the kinematic effect, δkin(x, Te) 6= 0 (Sazonov & Sunyaev 1998, Nozawa, Itoh, & Kohyama 1998), an effect which is negligible with present day sensitivity. Given the different spectral signatures of g(x), h(x) and δT (x, Te), it is straight forward to separate the physical variables y, vp and Te from sensitive multi- frequency observation. However, due to the complexity of the spectral shapes, in particular of δT (x, Te), the parameter space spanned by y, vp and Te may be non-trivial with multiple local minima in χ2. We therefore present a stochastic analysis tool SASZ based on simulated annealing, which allows a safe and fast parameter extraction even for such a complex parameter space. It is worth noting that whereas we here choose to use the set of cluster variables, (y, Te, vp), which are easily understood physically, then the anal- ysis could be simplified significantly by introducing a set of normal param- eters whose likelihood function is well-approximated by a normal distribu- tion (Kosowsky, Milosavljevic & Jimenez 2002, Chu, Kaplinghat & Knox 2003). This set of normal parameters could e.g. be (y, Te, K = τ vp), which directly enter in eq. (1). Another set could be (y, Te, K = vpT −0.85 ), where K enters because the cross-over frequency, ν0, is easily determined observa- tionally (due to a fast variation of ∆I(x)) and the fact that this cross-over frequency to a good approximation depends only on Te and K e ν0 = 217.4 (1 + 0.0114 T5) + 12 T −0.85 5 v500 GHz , (6) using T5 = Te/(5keV) and v500 = vp/(500km/sec). 2 Simulated annealing Let us now discuss the stochastic method used in SASZ . The idea behind the technique of simulated annealing is as follows. If a thermodynamic sys- tem is cooled down sufficiently slowly then thermodynamic equilibrium will be maintained during the cooling phase. When the temperature approaches zero, TA → 0, then the lowest energy state of the system, Emin, will be reached (Kirkpatrick, Gelatt, & Vecchi 1983). One can use this idea to search for the minimum in the space of allowed parameters, in which case the energy is replaced with E = χ2. The method is related to Monte Carlo methods, because one is basically jumping randomly around in parameter space, and if a given new point has lower energy (that is lower χ2 i−1), then this point is i accepted. To ensure that a system is not trapped in a local minimum there is in our case) than the previous point (χ2 3 a certain probability of keeping the new point even if its energy may be larger than the previous one (Metropolis et al. 1953) Paccept =   1 e−(Ei−Ei−1)/TA if Ei ≤ Ei−1 if Ei > Ei−1 and this probability, Paccept, will then be compared to a random number be- tween 0 and 1, to decide if the point will be kept or not. The temperature of the system, TA, is then slowly lowered to ensure that the global minimum χ2 is reached. We emphasize that the simulated annealing temperature, TA, is completely unrelated to the electron temperature, Te, of the galaxy cluster. In reality the jumping in parameter space is not completely random, instead the new points are drawn according to xi = xi−1 + Aβs TA T0 ran (7) where ran is a random number between −1/2 and 1/2. The size of the jump is such that initially all of parameter space is easily sampled, Aβ ≈ (xmax − xmin), whereas for a cooled system only very small jumps are allowed due to the √TA in eq. (7). For flat directions in parameter space (such as Te and vp) the coefficient in Aβ is a factor 10 larger than for the dominating Comptoniza- tion parameter y. In the first version of SASZ x is a 3 dimensional vector, ~x = (y, Te, vp), as energy we are using E = χ2, and we allow approximately 1000 random jumps at each temperature step. The cooling scheme is adaptive according to how many point are accepted or rejected, but can trivially be fixed to an exponential cooling scheme, TA(j) = c TA(j − 1), where c < 1. We use T0 = 1 and Tfinal = 10−12, and the parameter space allowed is presented in Table 1. Parameter Allowed range Units log(y) Te vp −7 0 −104 −2 100 104 keV km/sec We note that the technique of simulated annealing, which is often used for problems like the travelling salesman problem, has previously been used is cosmology, e.g. for parameter extraction from cosmic microwave background radiation data (Knox 1995, Hannestad 2000). 4 2.1 How to use SASZ in practice? The user has a set of observing frequencies (and possibly frequency bands) with corresponding SZ observations. The user specifies if some of the parameters have been observed with other methods; the temperature can be known to be e.g. Te = 12 ± 1 keV, or the peculiar velocity can be known e.g. vp = 300 ± 500 km/sec. The Fortran77 code SASZ will then stochastically analyse the parameter space (as described above) and presents the central value and 1σ error-bars for the cluster parameters (y, vp, Te). The method is much faster than e.g. the method of automatic refining grids (Aghanim et al. 2003). A user guide with examples can be downloaded together with the code 3 . This user guide also explains how to implement measured frequency band if available. 2.2 Error-bars There are two possible steps for determination of error-bars. The first is a very fast Gaussian approximation, and the second is a more accurate method which simultaneously provides data for nice figures. The Gaussian estimate of the error-bars comes for free. While searching for the global minimum SASZ will remember some intermediate points, ~xi, and their χ2 i . For each of the parameters, e.g. Te, one can consider the 2-dimensional figure (Te, ∆χ2), where all points lie within a quadratic curve. One can fit a quadratic curve through the best fit point and any of the other points. This method gives a fairly good estimate of the error-bar by performing such fit for all the points and then choosing the largest value. If upper and lower error- bars are different from each other, then the bigger is chosen for simplicity. The code selects a certain number of point for the fit, and presents automatically the error-bars for the 3 SZ parameters. The second method is to make a grid in the parameter we are interested in, and then minimize over the other parameters for each point on that grid. In this way one can get different upper and lower error-bars more accurately. This method is, however, slower because one will have to run another optimization for each chosen point in the grid. SASZ automatically volunteers to perform such a calculation in the 2σ range found with the Gaussian method or within a user-defined range. 3 http://krone.physik.unizh.ch/~hansen/sz/ 5 3 An example: COMA As an example of the use of SASZ we consider the nearby cluster COMA at redshift, z = 0.0231 ± 0.0017. This cluster has been measured at 6 different frequencies, 32 GHz (Herbig et al 1995), 143, 214 and 272 GHz (De Petris et al 2002), and at 61 and 94 GHz (Bennett et al. 2003). For a thorough discussion of consistency and a combined analysis see Battistelli et al. (2003), where the results are combined in their Table 1. Fig. 1. Using SASZ to determine the electron temperature or peculiar velocity for COMA. Figure 1a shows ∆χ2 as a function of the electron temperature, Te. Other parameters have been maximized over. The 1σ error-bar on Te (corresponding to ∆χ2 = 1) gives Te < 20 keV. Figure 1b shows ∆χ2 as a function of the peculiar velocity, vp. Other parameters have been maximized over. The 1σ error-bar on vp (corresponding to ∆χ2 = 1) gives vp = −300 ± 500 km/sec. Let us first consider the effect of bandwidth on the determination of y. We look at the case where the temperature is measured, Te = 8.25± 0.10 keV (Arnaud et al. 2001), and the peculiar velocity is known, vp = −29±299 km/sec (Colless et al 2001). We consider 3 cases, a) delta function for the spectral band shape, b) Gaussian shape, and c) flat (top-hat) shape. For the Gaussian and top- hat shapes we use the band widths from Table 1 in Battistelli et al. (2003). We find a) log(y) = −4.081+0.047 −0.055, and c) log(y) = −4.077+0.040 −0.054. We thus see, that the differences between the filters have almost no effect on the central value of the Comptonization parameter, and fairly small effect on the error-bars of log(y). This is good news (and in agreement with the findings of Church, Knox, & White (2003)), because the use of filters makes the computation much longer since one effectively must calculate at many more frequencies. Leaving vp as a free parameter to be maximized over has little effect, but leaving Te as a completely free parameter instead will increase the error-bars on log(y) by almost 50%. −0.048, b) log(y) = −4.080+0.039 6 Using Gaussian filters and assuming the peculiar velocity is measured, vp = −29±299 km/sec (Colless et al 2001), one finds that COMA has a temperature of Te = 6.6 ± 13 keV, as seen on figure 1a. This is in good agreement with the X-ray determination, but still with much larger error-bars. In comparison, the first cluster temperature measurement using purely the SZ observations gave Te = 26+34 −19 keV for A2163 (Hansen, Pastor & Semikoz 2002), where only observations at 4 frequencies were available. Similarly, using the observed temperature, Te = 8.25± 0.10 keV (Arnaud et al. 2001), one finds the peculiar velocity vp = −300 ± 500 km/sec, as seen on figure 1b, which is in good agreement with the findings of Colless et al (2001). 4 Conclusion We are presenting a user-friendly tool for parameter extraction from Sunyaev- Zeldovich effect observations. The tool SASZ is based on the stochastic method of simulated annealing, and is useful for any number of observing frequencies and any spectral band shape. The first version of the tool allows a determi- nation of the Comptonization parameter, y, the peculiar velocity, vp, and the electron temperature, Te. The f77 code SASZ can be readily downloaded to- gether with a user guide with examples from http://krone.physik.unizh.ch/~hansen/sz/ . Acknowledgements It is a pleasure to thank Sergio Pastor and Dima Semikoz for collaboration on the fitting formulae, and Nabila Aghanim for discussions and pointing me towards the data from COMA. References Aghanim, N., Hansen, S. H., Pastor, S., & Semikoz, D. V. 2003, Journal of Cosmology and Astro-Particle Physics, 5, 7 Arnaud, M. et al. 2001, A&A, 365, L67 Battistelli, E. S. et al. 2003, arXiv:astro-ph/0303587 Bennett, C. L. et al. 2003, ApJS, 148, 97 Birkinshaw, M. 1999, Phys. Rept., 310, 97 7 Carlstrom, J. E., Holder, G. P., & Reese, E. D. 2002, Ann. Rev. Astron. Astrophys., 40, 643 Chu, M. Kaplinghat, M., & Knox, L. 2003, arXiv:astro-ph/0212466 Church, S., Knox, L., & White, M. 2003, ApJL, 582, L63 Colless, M., Saglia, R. P., Burstein, D., Davies, R. L., McMahan, R. K., & Wegner, G. 2001, MNRAS, 321, 277 De Petris, M., D'Alba, L., Lamagna, L., Melchiorri, F., Orlando, A., Palladino, E., Rephaeli, Y., Colafrancesco, S., Kreysa, E. & Signore, M. 2002, astro- ph/0203303 Diego, J. M., Hansen, S. H., & Silk, J. 2003, MNRAS, 338, 796 Dolgov, A. D., Hansen, S. H., Pastor, S., & Semikoz, D. V. 2001, ApJ, 554, 74 Ensslin, T. A. & Kaiser, C. R. 2000, A&A, 360, 417 Gull, S. F. & Northover, K. J. E. 1976, Nature, 263, 572 Hannestad, S. 2000, PRD, 61, 23002 Hansen, S. H., Pastor, S., & Semikoz, D. V. 2002, ApJL, 573, L69 Herbig, T., Lawrence, C. R., Readhead, A. C. S., & Gulkis, S. 1995, ApJL, 449, L5 Itoh, N., & Nozawa, S. 2003, astro-ph/0307519 Kirkpatrick, S., Gelatt, C. D., & Vecchi, M. P. 1983, Science, 220, 671 Knox, L. 1995, PRD, 52, 4307 Kosowsky, A., Milosavljevic, M., & Jimenez, R. 2002, PRD, 66, 63007 Lake, G & Partridge, R. B. 1977, Nature, 270, 502 LaRoque, S.J. Reese, E.D., Holder, G.P., Carlstrom, J.E., Holzapfel, W.L., Joy, M. & Grego, L. 2002, astro-ph/0204134 Metropolis, N., Rosenbluth, A. W., Rosenbluth, M. N., Teller, E., & Teller, E. 1953, J. Chem. Phys., 21, 1087 Nozawa, S., Itoh, N., & Kohyama, Y. 1998, ApJ, 508, 17 Rephaeli, Y. 1995, ApJ, 445, 33 Sazonov, S. Y. & Sunyaev, R. A. 1998, ApJ, 508, 1 Sunyaev, R. A. & Zel'dovich, Ya. B. 1972, Comments Astrophys. Space Phys., 4 173 Wright, E. L. 1979, ApJ, 232, 348 8
0807.1758
1
0807
2008-07-10T23:39:41
Thermohaline mixing and gravitational settling in carbon-enhanced metal-poor stars
[ "astro-ph" ]
We investigate the formation of carbon-enhanced metal-poor (CEMP) stars via the scenario of mass transfer from a carbon-rich asymptotic giant branch (AGB) primary to a low-mass companion in a binary system. We explore the extent to which material accreted from a companion star becomes mixed with that of the recipient, focusing on the effects of thermohaline mixing and gravitational settling. We have created a new set of asymptotic giant branch models in order to determine what the composition of material being accreted in these systems will be. We then model a range of CEMP systems by evolving a grid of models of low-mass stars, varying the amount of material accreted by the star (to mimic systems with different separations) and also the composition of the accreted material (to mimic accretion from primaries of different mass). We find that with thermohaline mixing alone, the accreted material can become mixed with between 16 and 88 per cent of the pristine stellar material of the accretor, depending on the mass accreted and the composition of the material. If we include the effects of gravitational settling, we find that thermohaline mixing can be inhibited and, in the case that only a small quantity of material is accreted, can be suppressed almost completely.
astro-ph
astro-ph
Mon. Not. R. Astron. Soc. 000, 1 -- 12 (0000) Printed 13 November 2018 (MN LATEX style file v2.2) Thermohaline mixing and gravitational settling in carbon-enhanced metal-poor stars Richard J. Stancliffe1,2⋆ and Evert Glebbeek3 1Institute of Astronomy, The Observatories, Madingley Road, Cambridge CB3 0HA, U.K. 2Centre for Stellar and Planetary Astrophysics, Monash University, PO Box 28M, Clayton VIC 3800, Australia 3Sterrekundig Instituut Utrecht, Postbus 80000, 3508 TA Utrecht, The Netherlands. Accepted 0000 December 00. Received 0000 December 00; in original form 0000 October 00 ABSTRACT We investigate the formation of carbon-enhanced metal-poor (CEMP) stars via the scenario of mass transfer from a carbon-rich asymptotic giant branch (AGB) primary to a low-mass companion in a binary system. We explore the extent to which material accreted from a companion star becomes mixed with that of the recipient, focusing on the effects of thermohaline mixing and gravitational settling. We have created a new set of asymptotic giant branch models in order to determine what the composi- tion of material being accreted in these systems will be. We then model a range of CEMP systems by evolving a grid of models of low-mass stars, varying the amount of material accreted by the star (to mimic systems with different separations) and also the composition of the accreted material (to mimic accretion from primaries of different mass). We find that with thermohaline mixing alone, the accreted material can become mixed with between 16 and 88% of the pristine stellar material of the accretor, depending on the mass accreted and the composition of the material. If we include the effects of gravitational settling, we find that thermohaline mixing can be inhibited and, in the case that only a small quantity of material is accreted, can be suppressed almost completely. Key words: general stars: evolution, stars: AGB and post-AGB, stars: carbon, binaries: 1 INTRODUCTION Carbon-enhanced, metal-poor (CEMP) stars are defined as stars with [C/Fe]1>+1.0 (Beers & Christlieb 2005), with [Fe/H]< −2 in most cases. These objects appear with in- creasing frequency at low metallicity (Lucatello et al. 2006). The study of CEMP stars is being used to probe conditions in the early universe. For example, CEMP stars have been used to infer the initial mass function in the early Galaxy (e.g. Lucatello et al. 2005a). Chemical abundance studies have revealed that the majority of the CEMPs are rich in s- process elements like barium (Aoki et al. 2003), forming the so-called CEMP-s group. Recent survey work has detected a binary companion in around 68% of these CEMP-s stars and this is consistent with them all being in binary systems (Lucatello et al. 2005b). Binary systems provide a natural explanation for these objects, which are of too low a luminosity to have been able to produce their own carbon. The primary of the sys- tem was an asymptotic giant branch (AGB) star which be- came carbon-rich through the action of third dredge-up2 and transferred material on to the low-mass secondary (most likely via a stellar wind). The primary became a white dwarf and has long since faded from view, with the carbon-rich sec- ondary now being the only visible component of the system. It has commonly been assumed that the accreted material remains on the surface of the secondary until the star as- cends the giant branch, at which point the deepening of the convective envelope (referred to as first dredge-up because material that has experienced CN-cycling in the stellar in- terior is brought to the surface) mixes the material with the interior of the star. However, the transferred material should become mixed with the interior of the accreting star via the process of thermohaline mixing (see e.g. Stancliffe et al. 2007; Chen & Han 2004; Bitzaraki et al. 2003). This occurs when the mean molecular weight of the stellar gas increases toward the surface. A gas element displaced downwards and ⋆ E-mail: [email protected] 1 [A/B] = log(NA/NB) − log(NA/NB)⊙ 2 Third dredge-up occurs when the convective envelope of an AGB star deepens after a thermal pulse and material that has experienced nuclear burning is brought to the surface. 2 R. J. Stancliffe & E. Glebbeek compressed will be hotter than its surroundings. It will therefore lose heat, become denser and continue to sink. This leads to mixing on thermal timescales until the molecu- lar weight difference is eliminated (Kippenhahn et al. 1980). Stancliffe et al. (2007) showed that the inclusion of thermo- haline mixing could result in the accreted material being mixed throughout 90% of the star. Recent work has questioned the efficiency of thermo- haline mixing in carbon-enhanced metal-poor stars. Using a sample of barium-rich CEMP stars, Aoki et al. (2008) showed that the distribution of [C/H] values in turn-off stars (i.e. those stars that have reached the end of their main- sequence lives, are still of low-luminosity and have yet to become giants) was different from that in giants suggesting that significant mixing only happened at first dredge-up. A similar point was made by Denissenkov & Pinsonneault (2008b) using the data of Lucatello et al. (2006). These au- thors showed that the turn-off stars and giants were consis- tent with coming from the same distribution if first dredge- up resulted in the [C/H] value (and also the [N/H] value) being reduced by around 0.4 dex. They find that this re- sult is consistent with having an accreted layer of material mixed to an average depth of about 0.2 M⊙ (or alternatively having an accreted layer of 0.2 M⊙ that remains unmixed). Neither of these scenarios is consistent with the extensive mixing found by Stancliffe et al. (2007). A possible source for reduced thermohaline mixing effi- ciency has been suggested by Thompson et al. (2008). These authors suggest that the action of gravitational settling will alter the composition gradient of the accreting star near its surface. Helium will settle from the surface, reducing the mean molecular weight at the surface but leading to an in- crease in the layers beneath. This produces a small region in which the mean molecular weight, µ, decreases outwards toward the stellar surface -- a situation which is stable to thermohaline mixing. This stabilising composition gradient (a so-called 'µ-barrier') can inhibit the process of thermoha- line mixing. This paper extends the work of Stancliffe et al. (2007), examining the effect of varying the composition of the ac- creted material (mimicking accretion from different masses of companion) and the amount of material that is accreted. We also investigate the effect that gravitational settling has on the extent to which material is mixed. 2 THE STELLAR EVOLUTION CODE Calculations in this work have been carried out using a mod- ified version of the stars stellar evolution code originally developed by Eggleton (1971) and updated by many au- thors (e.g. Pols et al. 1995). The version used here includes the nucleosynthesis routines of Stancliffe et al. (2005) and Stancliffe (2005), which follow the nucleosynthesis of 40 iso- topes from D to 32S and important iron group elements. The code uses the opacity routines of Eldridge & Tout (2004), which employ interpolation in the OPAL tables (Iglesias & Rogers 1996) and which account for the variation in opacity as the C and O content of the material varies. At low temperatures and in the carbon-rich atmo- spheres of low-mass asymptotic giant branch (AGB) stars, molecular opacities become important (Marigo 2002). These Initial No. of mass TPs ( M⊙) Final mass ( M⊙) Core mass ( M⊙) [C/H] [N/H] µ 1 1.5 2 2.5 3 3.5 13 10 11 13 16 23 0.650 0.892 1.237 2.422 1.942 1.871 0.636 0.622 0.659 0.734 0.825 0.850 0.661 0.613 0.648 0.501 0.264 -0.030 -0.643 -1.527 -1.611 -1.710 -0.036 0.831 0.607 0.621 0.626 0.619 0.611 0.628 Table 1. Details of the final state of the AGB models. The columns are: initial mass in solar masses, number of thermal pulses the model was evolved through (note that not all the en- velope has been removed by the end of the run -- see the main text for more details), the final mass of the model in solar masses, the final H-exhausted core mass of the model in solar masses, the final [C/H] value, the final [N/H] value and the mean molecular weight, µ, of the ejected material, assuming it is fully ionised. The values of [C/H] and [N/H] for the average composition of the ejected material will be lower. are not included in the Eldridge & Tout (2004) opacity rou- tines. There are no tables of molecular opacities across a range of metallicities currently available (though this is be- ginning to change, see Cristallo et al. 2007, for example), so we adopt the procedures of Marigo (2002) to account for molecular opacity. Briefly, this involves the computation of the abundances of the molecules H2, H2O, OH, CO, CN and C2 in the equation of state and adding their contribution on to the opacity taken from the regular opacity tables. The form of the opacity for the molecules H2, H2O, OH and CO are taken from Keeley (1970), while for CN the polynomial fit of Scalo & Ulrich (1975) is used. The C2 opacity is as- sumed to follow that of the CN opacity. 3 AGB MODELS We have evolved a set of models with masses of 1, 1.5, 2, 2.5, 3 and 3.5 M⊙ from the pre-main-sequence without the use of convective overshooting. A mixing length of α = 2.0 has been employed throughout. This value is chosen based on calibration to a solar model. All the models were evolved using 999 mesh points. The metallicity was set to Z = 10−4 and the initial composition of the stellar material set to a solar-scaled composition (Anders & Grevesse 1989), giving the models an iron-to-hydrogen abundance of [Fe/H]= −2.3. Thermohaline mixing was included during the evolution and the mixing rate was enhanced by a factor of 100 above the Kippenhahn et al. (1980) value based on the results of Charbonnel & Zahn (2007), who showed that this enhanced rate could reproduce the observed behaviour of certain abun- dance trends on the first giant branch. On the asymptotic giant branch, the AGB specific mesh spacing function of Stancliffe et al. (2004) is employed, as is the mixing scheme used by these authors. Prior to the AGB, we employ the mass-loss rate of Reimers (1975) with η = 0.4 while on the AGB we use the Vassiliadis & Wood (1993) formula. All of the models experience third dredge-up (TDUP). The dredge-up efficiency λ (which is defined as the ratio of the amount of material dredged up to the growth of the core in the preceding interpulse period) is about 0.15 in the 1 M⊙ model, rising to about 0.68 in the 1.5 M⊙ model and is close to unity in the other models. The final outcome of the models is given in Table 1. Only the 1 M⊙ model was evolved to the white dwarf cooling track. The remaining models all suffered from numerical problems in the stellar envelope during the superwind phase, with the exception of the 2.5 M⊙ one which suffered an unrelated numerical prob- lem. In each case there was a rapid increase in the stellar radius prior to the failure of the model, typically just after the peak of a thermal pulse. Similar behaviour was reported by Karakas & Lattanzio (2007), who suggest that the prob- lem may be related to the input physics and particularly to the opacities. Further study of this phenomenon is clearly warranted. With the exception of the 2.5 M⊙ model, all of the mod- els have lost over half their stellar mass and they have all entered the superwind phase. We therefore estimate the con- tribution of the remaining envelope to the total yield by assuming that it is stripped from the star without further change to its composition. As mass loss is very rapid in the superwind phase we believe that we have missed only a few thermal pulses (and associated episodes of third dredge-up) at most. We record the final yield, the mass removed and the age of the final model for each of the masses considered. These details are needed to model the accretion on to the secondary as described below. The average composition of material ejected from our models is displayed in Figure 1. Full details of the total mass of each isotope ejected are presented in Table A1 in the appendix. The trends are as one might expect for AGB stars. Below 3 M⊙, the models show an increase in the 12C abundance due to the action of third dredge-up (TDUP). For the 3 and 3.5 M⊙ models there is also an increase in the 13C and 14N abundances as hot bottom burning sets in. Hot bottom burning happens because the convective en- velope is deep enough in the star to dip into the top of the H-burning shell, allowing CNO cycling to occur during the interpulse period. Similarly, there is evidence for the en- hanced occurrence of the p-burning Ne-Na and Mg-Al cycles in the more massive models. The greatest abundance of 22Ne is seen to occur in models of 1.5-2.5 M⊙ masses. This iso- tope is produced from α-captures on to 14N in the intershell, where it is also destroyed by the neutron-providing reaction 22Ne(α, n)25Mg. This latter reaction becomes active at tem- peratures of around 3×108 K and tends to favour destruction of 22Ne in the hotter, more massive cores of the higher mass models. For a more detailed discussion of AGB abundance patterns and nucleosynthesis, see Stancliffe & Jeffery (2007) and references therein. We now compare our models to others in the litera- ture. Both Karakas & Lattanzio (2007) and Herwig (2004) have produced models of the same metallicity. The mod- els of Karakas & Lattanzio (2007) -- hereafter KL07 -- have similar input physics to our own, as they use similar mass- loss prescriptions and do not employ the use of convective overshooting. Herwig's models employ the Blocker (1995) mass loss formula and convective overshooting is employed at both the bottom of the convective envelope and the pulse driven convection zone. KL07 have produced models of 2, 2.5, and 3 M⊙, along with several others that do not have the same mass as ours Mixing processes in CEMP stars 3 Figure 2. Comparison of the average composition for isotopes in three different evolution codes. The upper panel is for a 2 M⊙ model; the lower is for a 3 M⊙ model. Triangles indi- cated the results of this work, squares are from the work of Karakas & Lattanzio (2007) -- KL07 -- and circles are the models of Herwig (2004) designated H04. and hence are not considered further. Herwig (2004) -- here- after H04 -- has produced models of 2 and 3 M⊙, along with higher mass models which we do not consider here. For the two initial masses common to all three works, we have plot- ted the average composition of the ejecta for those isotopes common to all three codes. The results are displayed in Fig- ure 2. We note that there is generally good agreement be- tween the 2 M⊙ models, with few of the isotopes showing variations of over one order of magnitude. For many of the heavier isotopes, our average compositions tend to lie be- tween those of KL07 and H04. The maximum mass dredged up after a thermal pulse (TP) is similar to that of H04, but we dredge up nearly twice the amount of material as we have nearly twice as many pulses with TDUP. Compared to KL07 we also dredge up a similar amount of material, but their model has almost twice as many TPs as ours. The number of TPs a model experiences is primarily determined by the mass loss that the model experiences. If the superwind phase begins earlier, a model will experience fewer thermal pulses. The issue of mass loss on the AGB is a serious problem for stellar models (see Stancliffe & Jeffery 2007, for a detailed discussion). The comparison is less reassuring for the 3 M⊙ models (bottom panel of Figure 2), with a much greater spread in abundance predictions. For example, the 14N abundance dif- 4 R. J. Stancliffe & E. Glebbeek Figure 1. The average abundance in the ejecta of the models for selected isotopes from the nucleosynthesis network. The elements on the right are the isotopes that are used in the structure code. The isotope 'g' is a sink particle whose abundance relates to the efficiency of the s-process. fers by nearly 3 orders of magnitude! Again, our models tend to lie in between those of KL07 and H04. The patterns of the CNO isotopes suggests that the H04 model does not un- dergo significant hot bottom burning (HBB) as it has lower 13C and 14N, while those of this work and KL07 do. How- ever, the KL07 model has much more 14N (as well as 23Na from the Ne-Na cycle) than ours due to more mass being dredged-up per pulse and there being more thermal pulses with TDUP and HBB. The KL07 model typically dredges up around 7 × 10−3 M⊙ per pulse (although this value reaches 1.1 × 10−2 for one pulse), compared to around 5 × 10−3 M⊙ for our model. We also note the KL07 model has around 40 pulses with TDUP, compared to the 13 in our model. At the low mass end, we have no direct models to com- pare to KL07 as we have produced 1 and 1.5 M⊙ models, while they have made 1.25 and 1.75 M⊙ models. However, we note similar trends. In the lower mass models, the dredge- up efficiency drops noticeably and sets of models dredge-up similar masses for each episode of TDUP. We note that once again, the KL07 models have more episodes of TDUP than ours. 3.1 Cessation of third dredge-up The 1 M⊙ model displays some behaviour not found in the other models, namely that TDUP operates only between pulses 3 and 8 and ceases until the final pulse, where it oc- curs once more. We believe that the reason for the temporary cessation of TDUP is a consequence of the changing metallic- ity of the envelope. As dredge-up occurs, carbon is brought into the envelope making it more metal-rich. It is well es- tablished that stellar models of higher metallicity experience less efficient third dredge-up which tends to occur only at higher core masses and for more violent thermal pulses (see e.g. Karakas et al. 2002). In the 1 M⊙ model presented here, the envelope metallicity (in terms of the CNO elements) is raised up to a sufficiently high level that further thermal pulses are too weak to cause the occurrence of TDUP, un- til the last pulse which is significantly more violent than its predecessors. We have also evolved a 0.9 M⊙ model in the same way as described above. We find that it has an epsiode of third dredge-up on its 4th thermal pulse and then TDUP ceases until the 10th (and final) pulse when TDUP happens once more. We note that the pulse strengths for the last pulse is log Lmax He /L⊙ = 7.66 for the 4th pulse. Pulses 8 and 9 have pulse strengths greater than that for pulse 4, but no TDUP occurs. This seems to support the above hypothesis. He /L⊙ = 7.84 compared to log Lmax It should be noted that most authors do not find TDUP at the low envelope masses noted here (see e.g. Straniero et al. 2003; Karakas & Lattanzio 2007). However, Stancliffe & Jeffery (2007) have found dredge-up at low en- velope mass and advanced an argument why this is physi- cally reasonable. The efficiency and occurrence of TDUP is a contentious issue and has been for some time. However, Wood (1981) has shown that the minimum mass for dredge- up to occur is strongly dependent on the mixing length parameter, α. He shows that for models with metallicity Z = 0.001, the minimum stellar mass for TDUP to occur changes from 3.4 M⊙ when α = 0.67 to just 1.13 M⊙ when α = 1.5. We note that our α is larger than the value of 1.75 employed by KL07 and this could help to explain the dis- crepancy between computations from different codes. While this is not proof of the reason why we obtain TDUP at low envelope mass, it certainly presents an interesting avenue of further study. 4 CEMP MODELS We now turn to the modelling of the secondaries in these putative AGB binaries. We assume we are free to choose the primary mass, secondary mass and amount of material accreted to produce a secondary which has the appropriate mass (about 0.8 M⊙, see below). The primary mass deter- mines the composition of the accreted material and the age at which mass is transferred to the secondary. In the case of a 1 M⊙ primary, material will be transferred at an age of around 5.8 × 109 years, while for a 3.5 M⊙ primary matter transfer takes place at an age of around 1.9 × 108 years. Variations in the amount of mass the secondary accretes would presumably be caused by variations in the separation of the system. Following Stancliffe et al. (2007), we assume that matter is accreted from the AGB wind (rather than from Roche lobe overflow). In all cases we accrete matter at a rate of 10−6 M⊙ yr−1 which is approximately the rate at which the secondary would accrete matter via the Bondi- Hoyle process (Bondi & Hoyle 1944) from the AGB super- wind. We evolve a grid of models of stars of 0.7, 0.75, 0.78, 0.79, 0.795, 0.798 and 0.799 M⊙ each accreting enough ma- terial to make the final mass of the star equal to 0.8 M⊙, which is approximately the turn-off mass for the halo at the current age of the universe. We accrete material from each of the AGB models described above on to these stars, using the average composition of the ejecta and starting the mass transfer at the final age of the AGB model. We evolve two separate sets of models: one does not include thermohaline mixing, while the other does. Mixing processes in CEMP stars 5 pre-accretion post-accretion end of MS post 1st dredge-up 0.7 0.68 0.66 0.64 0.62 0.6 t h g i e w r a l u c e l o m n a e M 0.58 0.0001 0.001 0.01 dM/M 0.1 1 Figure 3. Mean molecular weight of the stellar material as a func- tion of the fractional mass (the mass above the considered layer) for the model accreting 0.1 M⊙ of material from a 1.5 M⊙ com- panion with thermohaline mixing included. The displayed mean molecular weight profiles are: just prior to the onset of accretion (solid line), just after accretion has ended (long-dashed line), at the end of the main sequence (short-dashed line) and just after the end of first dredge-up (dotted line). The total stellar mass has increased from 0.7 M⊙ to 0.8 M⊙ by the end of the accretion phase. At the metallicity considered here (i.e. Z = 10−4) a star of around 0.7 M⊙ has only a shallow convective envelope. Any material transferred to the star does not become mixed into the stellar interior via convection. If a deep convection zone did exist (as it does in such stars at higher metallicity), accreted material would rapidly be mixed into the stellar interior by convective motions. The impact of thermohaline mixing would then be reduced as the mean molecular weight of the accreted material would decrease as it becomes diluted with the pristine stellar material in the envelope. However, at low metallicity no such convective mixing happens and thermohaline mixing may efficiently mix the accreted matter with the stellar interior. To better understand the physical processes involved, it is instructive to consider the structural changes that occur during the evolution of one of these models. Figure 3 shows the mean molecular weight profile at varies points in the evolution of the model which accretes 0.1 M⊙ from a 1.5 M⊙ companion with thermohaline mixing taken into account. Initially, the star burns hydrogen in its core, raising the mean molecular weight in the central regions while the outer regions retail their initial composition (the solid line of Fig- ure 3). The star then accretes material from its companion and this material has a higher mean molecular weight than that of the original stellar material (long-dashed line). This situation is unstable to thermohaline mixing and the mate- rial mixes with the lower stellar layers, reducing the mean molecular weight of the surface layers and increasing that in the interior (short-dashed line). As this happens the star continues to burn hydrogen in the core with the mean molec- ular weight in this region continuing to rise. Eventually, the star evolves to the giant branch and a deep convective en- velope develops. This is first dredge-up and it results in the mean molecular weight of the surface being homogenised over the whole of the convective region. The mean molec- 6 R. J. Stancliffe & E. Glebbeek M1 ( M⊙) 0.1 0.05 Macc 0.02 ( M⊙) 0.01 1 1.5 2 2.5 3 3.5 0.37 0.20 0.18 0.12 0.15 0.07 0.45 0.28 0.22 0.22 0.23 0.17 0.54 0.40 0.33 0.35 0.35 0.27 0.57 0.47 0.41 0.43 0.42 0.35 0.005 0.002 0.001 0.61 0.51 0.47 0.47 0.50 0.43 0.65 0.58 0.55 0.57 0.57 0.54 0.67 0.63 0.61 0.62 0.62 0.60 Table 2. Mass coordinate of the greatest extent of thermohaline mixing (in solar masses), as a function of companion mass, M1, and the amount of material accreted, Macc. At first dredge-up, the base of the convective envelope reaches a mass co-ordinate of around 0.35 M⊙ at its maximum depth. ular weight of the envelope is raised because material that has experienced hydrogen burning is brought to the surface. Table 2 shows the depth to which thermohaline mixing is able to mix the accreted stellar material. The most ex- tensive mixing in the above grid mixes accreted matter with around 88% of the accreting star. This is consistent with the work of Stancliffe et al. (2007), whose model involved accre- tion from a highly C-enriched 2 M⊙ companion. In this work, the closest corresponding model does not mix as deeply be- cause the companion is not as He- and C-enriched as their model was. Only 16% of the star is mixed in the case with the shallowest mixing. The depth of mixing increases with the amount of material that is accreted. This is to be expected as the more accreted material there is, the more pristine matter must be mixed with it to reduce its mean molecular weight to something comparable to its surroundings. In ad- dition, the lower the mass of the primary star from which material is accreted, the less efficient the mixing will be. This is because less massive stars take longer to evolve, so the accreting star has more time to burn its own material, raising the mean molecular weight of its innermost regions and preventing mixing reaching these depths. Figure 4 displays the evolution of [C/H] with luminos- ity when accreting 0.001, 0.01 and 0.1 M⊙ of material from a 1.5 M⊙ companion. Note that the three models without ther- mohaline mixing have different [C/H] values because of the existence of a thin convective envelope, which mixes some of the material during accretion. Figure 5 displays the evo- lution of [C/H] with luminosity when accreting 0.02 M⊙ of material from companions of different mass. These plots are representative of the full data set. For the models without thermohaline mixing, the only episode of mixing occurs between luminosities of around log L/L⊙ = 0.7 − 1.4. This is due to the occurrence of first dredge-up as the star ascends the giant branch and a deep convective envelope develops. In some of the models there is a small amount of mixing into a surface convection zone during the accretion phase. In the models with thermohaline mixing there is a sharp drop in the [C/H] values at low lu- minosities while the star is on the main sequence. If a large quantity of material is accreted then it becomes mixed to a depth lower than the convective envelope reaches during first dredge-up. In this case, we do not see a decrease in [C/H] at first dredge-up. If the accreted material does not mix as deeply, when first dredge-up occurs the material be- Figure 4. The evolution of the surface [C/H] ratio as a function of luminosity when accreting 0.1 M⊙ (solid line), 0.01 M⊙ (dot- ted line) and 0.001 M⊙ (dashed line) of material from a 1.5 M⊙ companion. The mass of all the models after accretion has fin- ished is 0.8 M⊙. The top panel displays models without thermo- haline mixing while the lower one is for models with thermoha- line mixing. The tracks begin from the point at which accretion finishes. Crosses denoted the CEMPs of Lucatello et al. (2006) while squares denote the Ba-rich CEMPs of Aoki et al. (2007) and Aoki et al. (2008). comes further diluted and the [C/H] value falls further. This can be seen in lower panel of Figure 4. The model which ac- cretes 0.001 M⊙ (the dashed line in the lower panel of this figure) only mixes down to a mass co-ordinate of 0.63 M⊙ whereas the convective envelope reaches a mass co-ordinate of around 0.35 M⊙ at its maximum depth. As first dredge- up occurs, we see a marked decrease in the [C/H] value as expected. Both sets of models have problems explaining the ob- served patterns of [C/H] as a function of luminosity. Mod- els without thermohaline mixing struggle to populate those turn-off stars with low [C/H] values and unless a large quan- tity of material is accreted, they suffer too much dilution at first dredge-up. Models with thermohaline mixing do not produce the highest [C/H] values as the accreted matter is mixed with the pristine stellar material too quickly to be ob- served. In addition, they suggest that we should be able to observe turn-off stars with comparatively low [C/H] values. No such objects have so far been detected and this is not a selection effect (see the discussion in Aoki et al. 2007). Both model sets favour accretion from 1.5-2.5 M⊙ compan- Mixing processes in CEMP stars 7 Figure 5. The evolution of the surface [C/H] ratio as a func- tion of luminosity when accreting 0.02 M⊙ of material for a 1 M⊙ (solid line), 1.5 M⊙ (dotted line), 2 M⊙ (short-dashed line), 3 M⊙ (long-dashed line) and a 3.5 M⊙ (dot-dashed line) companion. The mass of all the models after accretion has finished is 0.8 M⊙. The top panel displays models without thermohaline mixing while the lower one is for models with thermohaline mixing. The tracks begin from the point at which accretion finishes. Crosses denoted the CEMPs of Lucatello et al. (2006) while squares denote the Ba-rich CEMPs of Aoki et al. (2007) and Aoki et al. (2008). Figure 6. The evolution of the surface [N/H] ratio as a function of luminosity when accreting 0.1 M⊙ (solid line), 0.01 M⊙ (dot- ted line) and 0.001 M⊙ (dashed line) of material from a 3.5 M⊙ companion. The mass of all the models after accretion has fin- ished is 0.8 M⊙. The top panel displays models without thermo- haline mixing while the lower one is for models with thermoha- line mixing. The tracks begin from the point at which accretion finishes. Crosses denoted the CEMPs of Lucatello et al. (2006) while squares denote the Ba-rich CEMPs of Aoki et al. (2007) and Aoki et al. (2008). ions as accreting different masses from these stars can repro- duce the spread in the [C/H] values at luminosities above log L/L⊙ = 1.4. The evolution of [N/H] as a function of luminosity when accreting a varying amount of mass from a companion of 3.5 M⊙ is shown in Figure 6. Figure 7 shows the evolution of [N/H] as a function of luminosity when accreting 0.02 M⊙ of material from companions of different mass. The mod- els without thermohaline mixing show exactly the same be- haviour as was noted for [C/H], with first dredge-up causing a major dilution of the accreted material. In the case that the accreted material is nitrogen-poor (as it is for accre- tion from companions below 3 M⊙), there is a slight rise in [N/H] at the end of first-dredge-up, following the sharp drop caused by the dilution. This occurs because the convective envelope reaches down to those regions of the star where carbon originally present in the star has been processed to nitrogen via the CN cycle. This rise in [N/H] is exactly what would be expected in a 'normal' metal-poor star. The picture is very different for models including ther- mohaline mixing. As with the behaviour of [C/H], there is initially a sharp drop in [N/H] on the main sequence as the accreted material mixes with the stellar interior. However, as the star evolves up the giant branch (i.e. to higher lu- minosity), a sharp increase in [N/H] can occur during first dredge-up. This occurs because when carbon is mixed deep into the stellar interior it can be processed to nitrogen and subsequently brought to the surface during first dredge-up (Stancliffe et al. 2007). This effect can be seen in the lower panel of Figure 7, most notably for the case of accretion from a 1.5 M⊙ companion (the dashed line). The extent of this effect depends on how much carbon is present in the accreted material and to what depth it is mixed. Nitrogen is a real problem for these models. Only the 3 and 3.5 M⊙ models produce nitrogen abundances that are close to the observed values, with the 3.5 M⊙ model being able to reproduce the spread in [N/H] values. Without ther- mohaline mixing, it is difficult to reproduce the low [N/H] values in turn-off stars, whereas the thermohaline mixing models can populate this region provided that only a small amount of material is accreted. It should be noted that such massive companions do not reproduce the [C/H] ob- 8 R. J. Stancliffe & E. Glebbeek Figure 7. The evolution of the surface [N/H] ratio as a func- tion of luminosity when accreting 0.02 M⊙of material for a 1 M⊙ (solid line), 1.5 M⊙ (dotted line), 2 M⊙ (short-dashed line), 3 M⊙ (long-dashed line) and a 3.5 M⊙ (dot-dashed line) companion. The mass of all the models after accretion has finished is 0.8 M⊙. The top panel displays models without thermohaline mixing while the lower one is for models with thermohaline mixing. The tracks begin from the point at which accretion finishes. Crosses denoted the CEMPs of Lucatello et al. (2006) while squares denote the Ba-rich CEMPs of Aoki et al. (2007) and Aoki et al. (2008). servations very well. It has been noted that it is hard to match both the carbon and nitrogen enrichments from AGB sources (Johnson et al. 2007). The difficulty in matching the models to the observa- tions is almost certainly a problem on the theoretical side. It is well known that extra-mixing mechanisms must be ac- tive during the TP-AGB (and indeed at other phases in the lives of stars). Progress at determining the physical cause of this unknown extra mixing, dubbed 'cool bottom pro- cessing' (CBP) has not been forthcoming. We know that some extra-mixing processes must happen in order to get a 13C pocket to provide neutrons for the s-process. However, we note that our yields are comparable to those of Herwig (2004) who included convective overshooting at the base of both the intershell convection zone and the convective enve- lope. It is also interesting to note that there is a degree of N-richness associated with nearly all CEMP stars. Whatever the extra-mixing mechanism is, it may have to produce just the right amount of mixing over a range of stellar masses which should seriously constrain the physical process. One possible alternative explanation for the simulta- neous enhancement of both carbon and nitrogen is the oc- currence of a proton ingestion episode or flash-driven deep- mixing (FDDM). These two labels refer to the situation that occurs when protons are ingested into the intershell convection zone and burn at very high temperatures. Such episodes should produce enhancements of both C and N (Fujimoto et al. 2000). In order for the intershell convec- tion zone to be able to penetrate into H-rich regions the hydrogen burning shell cannot present an effective entropy barrier. This only happens at very low metallicity when the efficiency of CNO burning is substantially reduced due to the absence of CNO nuclei. This does not seem to happen at [Fe/H]≈ −2 but may account for some of the CEMPs with lower metallicity. 4.1 The effect of gravitational settling There are certainly problems with models including thermo- haline mixing from an observational perspective. They do not seem to reproduce those objects with high [C/H] ratios and they imply there should be a population of low [C/H] objects that are not observed. However from the theoreti- cal perspective there are problems with models that do not include thermohaline mixing -- one cannot ignore physics! If thermohaline mixing does not seem to be effective, there must be some mechanism to suppress it. Thompson et al. (2008) have suggested that the effects of gravitational settling may inhibit the action of thermo- haline mixing. Gravitational settling would lead to helium and other heavy isotopes diffusing away from the stellar sur- face during the main sequence. This would raise the mean molecular weight, µ, just below the stellar surface, providing a positive µ-gradient that would inhibit the action of ther- mohaline mixing (see Figure 12 of Thompson et al. 2008). However, we note that this µ-barrier is not very extensive in their model. At its peak µ is just below 0.6 which is some- what less than the µ of the accreted material. For example, the mean molecular weight of material ejected by the 1.5 M⊙ model is 0.621. This barrier is also located close to the stel- lar surface (in mass). One would therefore expect that the accretion of a sufficiently large amount of material will easily overcome the barrier. To test this, we have implemented the physics of gravi- tational settling in the code. The rate of change of the abun- dance, Xi, of an isotope i in the absence of nuclear reactions and convective mixing is given by ∂Xi ∂t = − 1 r2 ∂ ∂r (r2ρXivD) (1) where ρ is the density, r is the radius and vD is the diffusion velocity of the species (Pelletier et al. 1986). The diffusion velocity is given by vD = −D12 (cid:16) ∂ ln ci ∂r + [mi − µ] g kT − α12 ∂ ln T ∂r (cid:17) (2) where µ is the mean molecular weight of the material, g is the local gravity, ci = n(i)/n(H) is the concentration of the isotope i relative to hydrogen and mi is the mass of the isotope i (Michaud & Vauclair 1991). D12 and α12 are the atomic and thermal diffusion coefficients and are taken from Paquette et al. (1986). The right-hand side of equation 1 is added to the standard equation for the composition change, which is then solved as usual. We re-ran the thermohaline mixing models with grav- itational settling included in the calculation. Again, it is instructive to consider the structural changes of an individ- ual model to understand the processes involved. Figure 8 shows the mean molecular weight of the interior of a 0.7 M⊙ star accreting 0.1 M⊙ of material from a 1.5 M⊙ compan- ion when both thermohaline mixing and gravitational set- tling are taken into account. At the zero-age main sequence, the composition of the material is uniform as no significant H-burning has occurred (the solid line in Figure 8). Grav- itational settling, which acts over long timescales, leads to helium (and other heavy elements) settling from the surface and hydrogen being enhanced there. This leads to the mean molecular weight decreasing in the surface regions (down to a fractional mass of around dM/M ≈ 0.01) and a mean molecular weight gradient building up (see the long-dashed line of the figure). It is this which may potentially inhibit any thermohaline mixing. Note that the mean molecular weight remains constant down to around dM/M ≈ 0.01 because of the presence of a shallow convection zone. Material of higher mean molecular weight is then accreted from the companion (short-dashed line). In this example, a large quantity of ma- terial is accreted and the mean molecular weight gradient is quickly overwhelmed. The accreted material then mixes via thermohaline mixing. Once the equilibrium configuration is reached, the effects of gravitational settling begin to assert themselves again, with helium and heavy elements becom- ing depleted in the surface layers by the end of the main sequence (dotted line). For this reason, the mean molecular weight in the outermost layers is reduced for a second time. Finally, the star evolves up the giant branch and first dredge- up occurs, homogenising the composition of the envelope (dot-dashed line) and producing a similar mean molecular weight profile to that of the model without gravitational settling. Table 3 shows the maximum extent of mixing of ac- creted material. We find that gravitational settling is most significant when accreting from a low mass companion. This is to be expected as settling in low-mass stars acts over long (i.e. gigayear) timescales and so it takes time to build up an effective mean molecular weight barrier. If accretion hap- pens before the barrier is fully established, as it does in the case of accretion from all but the two lowest mass compan- ions, then thermohaline mixing proceeds in almost the same way as it does in the absence of gravitational settling. As the amount of material accreted is decreased, the effects of the µ-barrier become more pronounced. The ac- creted material suffers more dilution with the H-enriched, He-depleted surface of the star which reduces its mean molecular weight, slowing the rate of thermohaline mixing. In the case that the amount of material accreted is very small, the barrier can be completely effective and mixing does not proceed to great depths. Again, this behaviour is expected. The µ-barrier is formed only in the outermost parts of the star and if a large quantity of material is ac- creted it quickly overwhelms the barrier. When only a small amount of material is accreted the barrier is effective at in- hibiting thermohaline mixing. If 0.001 M⊙ of material from a 1 M⊙ companion is accreted, it will only mix with less than 1% of the star, in contrast to the 12% seen to occur with thermohaline mixing when gravitational settling is not considered. Mixing processes in CEMP stars 9 t h g i e w r a l u c e l o m n a e M 0.7 0.68 0.66 0.64 0.62 0.6 0.58 0.56 0.54 0.52 0.0001 ZAMS pre-accretion post-accretion end of MS post 1st dredge-up 0.001 0.01 dM/M 0.1 1 Figure 8. Mean molecular weight of the stellar material as a function of the fractional mass (the mass above the considered layer) for the model accreting 0.1 M⊙ of material from a 1.5 M⊙ companion with thermohaline mixing and gravitational settling included. The displayed mean molecular weight profiles are: at the zero-age main sequence (ZAMS, solid line), just prior to the on- set of accretion (long-dashed line), just after accretion has ended (short-dashed line), at the end of the main sequence (dotted line) and just after the end of first dredge-up (dot-dashed line). The total stellar mass has increased from 0.7 M⊙ to 0.8 M⊙ by the end of the accretion phase. The steps in the mean molecular weight profile near the surface of the end of main sequence model are an artefact caused by the finite precision with which the mass co-ordinate is written in the output file. M1 ( M⊙) 0.1 0.05 Macc 0.02 ( M⊙) 0.01 1 1.5 2 2.5 3 3.5 0.43 0.24 0.16 0.15 0.17 0.07 0.55 0.34 0.25 0.25 0.26 0.17 0.69 0.46 0.39 0.39 0.38 0.30 0.76 0.56 0.48 0.47 0.47 0.38 0.005 0.002 0.001 0.78 0.71 0.56 0.58 0.58 0.47 0.795 0.795 0.73 0.74 0.72 0.64 0.797 0.797 0.794 0.794 0.791 0.74 Table 3. Mass coordinate of the greatest extent of thermohaline mixing (in solar masses), as a function of companion mass, M1, and the amount of material accreted, Macc for the models in- cluding gravitational settling. At first dredge-up, the base of the convective envelope reaches a mass co-ordinate of around 0.35 M⊙ at its maximum depth. This effect can also be seen in Figure 9, which displays the evolution of [C/H] as function of luminosity when ma- terial is accreted from a 1.5 M⊙ companion. The decline in [C/H], followed by the sudden rise around log L/L⊙ = 0.7 is caused by gravitational settling on the main sequence both reducing the surface carbon abundance and enhancing that of hydrogen. The sudden rise occurs as first dredge-up sets in, with the deepening convective envelope mixing the stel- lar interior. When accreting 0.1 M⊙ the material is quickly mixed in to the same degree in both models. If less mate- rial is accreted, we begin to notice differences in the degree of mixing. In the case of accreting 0.001 M⊙ the µ-barrier keeps the [C/H] ratio over 1 dex higher than in the case with thermohaline mixing alone. In fact, the µ-barrier keeps the [C/H] value higher on the main sequence (log L/L⊙ < 0.7) than it is in the case when 0.01 M⊙ is accreted. This is in 10 R. J. Stancliffe & E. Glebbeek Figure 9. The evolution of the surface [C/H] ratio as a function of luminosity when accreting 0.1 M⊙ (solid line), 0.01 M⊙ (dot- ted line) and 0.001 M⊙ (dashed line) of material from a 1.5 M⊙ companion. The mass of all the models after accretion has fin- ished is 0.8 M⊙. The top panel displays models with thermoha- line mixing and gravitational settling while the lower one is for models with thermohaline mixing only. The tracks begin from the point at which accretion finishes. Crosses denoted the CEMPs of Lucatello et al. (2006) while squares denote the Ba-rich CEMPs of Aoki et al. (2007) and Aoki et al. (2008). contrast to the situation with thermohaline mixing alone, where the model which accretes 0.001 M⊙ (the dashed line in Figure 9) has a [C/H] value about 0.6 dex lower than that of the model which accretes 0.01 M⊙ throughout its whole evolution. With the inclusion of gravitational settling, we no longer obtain models with low [C/H] values at low luminosities (provided the initial [C/H] value is high enough). How- ever, gravitational settling does present an added prob- lem. As carbon settles from the stellar surface during the star's main-sequence lifetime, the [C/H] ratio drops. This makes it difficult for us to reproduce the more luminous turn-off objects with [C/H] values in the range -1 -- 0. This is a serious problem for any model that includes gravita- tional settling. If more material is accreted, then a higher [C/H] value results throughout the main sequence. For ex- ample, if 0.2 M⊙ of material were to be accreted from a 1.5 M⊙ companion, the resulting object would have a [C/H] value with a minimum of just -0.5. We note that we have not included the effects of radiative levitation. How- ever, based on the simulations of Richard et al. (2002) and Richard, Michaud & Richer (2002), it seems that radiative levitation should be ineffective for carbon. There is also still the problem of the high [C/H] ob- jects. The highest [C/H] values seem to correspond to those models which have accreted the most mass from their com- panions. In these cases, the µ-barrier is ineffective and ther- mohaline mixing proceeds unhindered. The highest [C/H] value obtained in our models is about [C/H] ≈ −0.2. If more material is accreted, we can reach a higher [C/H]. For example, accreting 0.2 M⊙ of material from a 1.5 M⊙ com- panion produces [C/H] ≈ 0. This is in line with the upper [C/H] values of the Ba-rich stars of Aoki et al. (2007) and Aoki et al. (2008). However, the sample of C-rich stars of Lucatello et al. (2006) has three objects with [C/H] > 0.6. It is difficult to explain these objects with models involv- ing thermohaline mixing as our AGB models only just pro- duce [C/H] values of around this. Any amount of mixing will easily reduce this value. However, we note that these three high [C/H] CEMPs are all somewhat more metal-rich (with [Fe/H] greater than around -1.8) than the models of this work so these models may not be directly comparable. 5 DISCUSSION As suggested by Thompson et al. (2008) a means of sup- pressing the action of thermohaline mixing is necessary to prevent the destruction of lithium in some CEMPs. Lithium is a fragile element that is easily destroyed at temperatures of around 2 × 106 K which are found at around 0.01 M⊙ be- low the stellar surface of the low-mass secondary while on the main sequence. As we have seen, gravitational settling will only allow the accreted matter to remain at the sur- face if the mass of accreted material is small. In the case of CS22964-161, the CEMP binary of Thompson et al. (2008), the measured [C/H] value of -1.2 fits well with accretion from a 1-1.5 M⊙ companion when about 0.001 M⊙ is transferred to the recipient star. We would expect lithium to survive in this model because mixing does not reach down to re- gions where the temperature is high enough for lithium to burn. However, we have not tried to model this. We leave the treatment of the behaviour of the light elements to future work. Denissenkov & Pinsonneault (2008b) suggested that the shift in the [C/H] and [N/H] values at first dredge-up requires that the average CEMP has a depth of 0.2 M⊙ of ac- creted material. They point out that this can either be from accreting 0.2 M⊙ of material that remains unmixed, or from accreted material mixing to a depth of 0.2 M⊙. This would imply that, on average, no more than about 0.01 M⊙ could be accreted from a 1 M⊙ companion (this value rises slightly to around 0.05 M⊙ when gravitational settling is taken into account), with this value decreasing rapidly as the compan- ion mass increases. This assumes that thermohaline mixing is as efficient as is presented in this work. If thermohaline mixing is less efficient, then a greater amount of material could be accreted. Aoki et al. (2008) have suggested that the [C/H] val- ues measured for turn-off stars represent the composition of the donor AGB star. This cannot be reconciled with models including thermohaline mixing, as mixing leads to a large decrease in the [C/H] value. For the star to retain the com- Mixing processes in CEMP stars 11 the models presented herein, one would need to do a pop- ulation synthesis calculation for each model set. Attempts at modelling the CEMP population are currently under way (Izzard et al., in prep). If thermohaline mixing cannot be inhibited by gravi- tational settling in the case that a large quantity of ma- terial is accreted, we must find another way to reduce its efficiency. A possible mechanism to do this has been put forward by Denissenkov & Pinsonneault (2008a). These au- thors point out that rotationally-induced horizontal turbu- lent diffusion can suppress thermohaline convection and ex- amine the specific case of mixing on the giant branch (see e.g. Eggleton et al. 2006; Charbonnel & Zahn 2007). The molec- ular weight inversion found there is at least two orders of magnitude smaller than in the case of accretion in CEMPs, so one presumes the turbulent diffusion would need to be much stronger to suppress thermohaline mixing in this situ- ation. The interaction between thermohaline mixing and ro- tation is an intriguing possibility and merits further study. We are unable to investigate this at present as our code does not include rotational physics. 6 CONCLUSIONS We have modelled the accretion of AGB material on to low- mass stars in order to determine what chemical signatures may be observed in CEMP stars. We have examined three specific cases: canonical evolution including only convective mixing, the inclusion of thermohaline mixing and the inclu- sion of both thermohaline mixing and gravitational settling. We find that thermohaline mixing can lead to accreted ma- terial being mixed with between 16 and 88% of the accreting star, depending on the mass and composition of the accreted material. When gravitational settling is included, thermoha- line mixing is severely inhibited when only a small amount of material (around a few 10−3 M⊙) is accreted because of the presence of a µ-barrier formed by the settling of helium. This barrier is less and less effective as the amount of ac- creted material is increased. It is also less effective for more massive companions as the barrier has had less time to be established before accretion on to the secondary occurs. Models without thermohaline mixing produce turn-off objects with [C/H] values that are too high to match obser- vations. Unless a substantial quantity of material has been accreted, these models also suffer from too much dilution at first dredge-up. Models with thermohaline mixing cannot re- produce the highest [C/H] values observed and predict the existence of low [C/H] objects which are not observed. In order to reproduce the highest [C/H] values, which seem to require the accretion of a large quantity of material from a companion, some alternative mechanism to suppress ther- mohaline convection is required. The inclusion of gravita- tional settling can solve the problem of the low [C/H] objects at low luminosity as the µ-barrier prevents mixing when the amount of accreted material is small. However, the inclu- sion of this physics presents another serious problem namely that carbon will settle from the surface during the main se- quence, making it extremely difficult to form turn-off stars with [C/H] values of -1 -- 0. There also appears to be a problem with the abundance predictions of the AGB models. They do not predict sub- Figure 10. The distribution of [C/H] for turn-off stars and gi- ant stars for the data of Lucatello et al. (2006) and the combined data of Aoki et al. (2007) and Aoki et al. (2008). Stars are iden- tified as turn-off objects if log L/L⊙ < 0.7 (i.e. they have not yet undergone first dredge-up) and are displayed in the upper panels. Stars are defined as post-first dredge-up giants if log L/L⊙ > 1.5 and are displayed in the lower panels. The total number of stars, Ntot, in each plot is also displayed. position of the AGB donor a substantial amount of material would have to have been accreted from the primary, such that the accreted material dominates the original stellar ma- terial. However, the accretion and subsequent thermohaline mixing of a large quantity of material does not agree with the observed differences in distribution of [C/H] from turn- off to giant stars (see Figure 10, right-hand column). Effi- cient thermoahline mixing of a large quantity of accreted material leads to very little change in [C/H] at first dredge- up. This cannot be the case in the majority of CEMP stars. In addition, if [C/H] ≈ 0 does represent the composition of the donor then there is a major problem with the AGB models which predict substantially larger values. This is not improbable given how little is known about the mass-loss rate at low metallicity and the efficiency of third dredge-up. The existence of high [C/H] CEMPs suggests that ther- mohaline mixing is not efficient in all cases. The difference in the distribution of [C/H] for turn-off and giant stars in the sample of Aoki et al. (2008) suggests that some, but not all, turn-off stars with [C/H] ≈ 0 do not become mixed after first dredge-up while some do (see Figure 10). After first dredge- up there is still a large proportion of stars with [C/H] ≈ 0 which suggests that little dilution of material has taken place in these objects during first dredge-up. These objects must have experienced efficient thermohaline mixing, while the others remained unmixed to some degree until first dredge- up. There is less evidence for this phenomenon in the data set of Lucatello et al. (2006), which displays a distinct shift in the peak [C/H] value between turn-off and giant stars (see the left-hand column of Figure 10). In order to predict the distributions of [C/H] for turn-off stars and giants based on 12 R. J. Stancliffe & E. Glebbeek stantial enhancements of both carbon and nitrogen at the same time (except in a very narrow mass range). Further work needs to be done to find a mechanism capable of pro- ducing the observed abundance patterns. Lucatello S., Beers T. C., Christlieb N., Barklem P. S., Rossi S., Marsteller B., Sivarani T., Lee Y. S., 2006, ApJ.Lett., 652, L37 Lucatello S., Gratton R. G., Beers T. C., Carretta E., 2005a, ApJ, 625, 833 Lucatello S., Tsangarides S., Beers T. C., Carretta E., Gratton R. G., Ryan S. G., 2005b, Ap.J., 625, 825 Marigo P., 2002, A&A, 387, 507 Michaud G., Vauclair S., 1991, in Cox A. N., Livingston W. C., Matthews M. S., eds, Solar Interior and Atmo- sphere Element separation by atomic diffusion.. Univer- sity of Arizona Press, Tuscon, pp 304 -- 325 Paquette C., Pelletier C., Fontaine G., Michaud G., 1986, ApJS, 61, 177 Pelletier C., Fontaine G., Wesemael F., Michaud G., Weg- ner G., 1986, ApJ, 307, 242 Pols O. R., Tout C. A., Eggleton P. P., Han Z., 1995, MN- RAS, 274, 964 Reimers D., 1975, Memoires of the Societe Royale des Sci- ences de Liege, 8, 369 Richard O., Michaud G., Richer J., 2002, ApJ, 580, 1100 Richard O., Michaud G., Richer J., Turcotte S., Turck- Chi`eze S., VandenBerg D. A., 2002, ApJ, 568, 979 Scalo J. M., Ulrich R. K., 1975, ApJ, 200, 682 Stancliffe R. J., 2005, PhD thesis, University of Cambridge Stancliffe R. J., Glebbeek E., Izzard R. G., Pols O. R., 2007, A&A, 464, L57 Stancliffe R. J., Jeffery C. S., 2007, MNRAS, 375, 1280 Stancliffe R. J., Lugaro M. A., Ugalde C., Tout C. A., Gorres J., Wiescher M., 2005, MNRAS, 360, 375 Stancliffe R. J., Tout C. A., Pols O. R., 2004, MNRAS, 352, 984 Straniero O., Dom´ınguez I., Cristallo S., Gallino R., 2003, Pub. Astron. Soc. Aust., 20, 389 Thompson I. B., Ivans I. I., Bisterzo S., Sneden C., Gallino R., Vauclair S., Burley G. S., Shectman S. A., Preston G. W., 2008, ApJ, 677, 556 Vassiliadis E., Wood P. R., 1993, ApJ, 413, 641 Wood P. R., 1981, in Iben Jr. I., Renzini A., eds, Physical Processes in Red Giants Vol. 88 of Astrophysics and Space Science Library, The conditions for dredge-up of carbon during the helium shell flash and the production of carbon stars. pp 135 -- 139 APPENDIX A: STELLAR YIELDS Here we present the final gross yields (i.e. the total mass of each isotope ejected) from each of the AGB models for all of the isotopes in the nucleosynthesis network. 7 ACKNOWLEDGEMENTS The authors thank Sara Lucatello for making her data avail- able and for helpful discussions. They also thank the ref- eree, Zhanwen Han, for useful comments which have im- proved the manuscript. RJS is now funded by the Australian Research Council's Discovery Projects scheme under grant DP0879472. He is grateful to Churchill College for the Ju- nior Research Fellowship which has supported him for the last three years. EG thanks the NWO for funding. REFERENCES Anders E., Grevesse N., 1989, Geo.Cosmo.Acta, 53, 197 Aoki W., Beers T. C., Christlieb N., Norris J. E., Ryan S. G., Tsangarides S., 2007, ApJ., 655, 492 Aoki W., Beers T. C., Sivarani T., Marsteller B., Lee Y. S., Honda S., Norris J. E., Ryan S. G., Carollo D., 2008, ApJ, 678, 1351 Aoki W., Ryan S. G., Tsangarides S., Norris J. E., Beers T. C., Ando H., 2003, Elemental Abundances in Old Stars and Damped Lyman-α Systems, 25th meeting of the IAU, Joint Discussion 15, 22 July 2003, Sydney, Australia, 15 Beers T. C., Christlieb N., 2005, ARA&A, 43, 531 Bitzaraki O. M., Tout C. A., Rovithis-Livaniou H., 2003, New Astronomy, 8, 23 Blocker T., 1995, A&A, 297, 727 Bondi H., Hoyle F., 1944, MNRAS, 104, 273 Charbonnel C., Zahn J.-P., 2007, A&A, 467, L15 Chen X., Han Z., 2004, MNRAS, 355, 1182 Cristallo S., Straniero O., Lederer M. T., Aringer B., 2007, ApJ, 667, 489 Denissenkov P. A., Pinsonneault M., 2008a, ApJ, in press, arXiv:0708.3864 Denissenkov P. A., Pinsonneault M., 2008b, ApJ, in press, arXiv:0709.4240 Eggleton P. P., 1971, MNRAS, 151, 351 Eggleton P. P., Dearborn D. S. P., Lattanzio J. C., 2006, Science, 314, 1580 Eldridge J. J., Tout C. A., 2004, MNRAS, 348, 201 Fujimoto M. Y., Ikeda Y., Iben I. J., 2000, ApJ.Lett., 529, L25 Herwig F., 2004, ApJS., 155, 651 Iglesias C. A., Rogers F. J., 1996, ApJ, 464, 943 Izzard R. G., Glebbeek E., Stancliffe R. J., Pols O. R., in prep. Johnson J. A., Herwig F., Beers T. C., Christlieb N., 2007, ApJ, 658, 1203 Karakas A., Lattanzio J. C., 2007, Publications of the As- tronomical Society of Australia, 24, 103 Karakas A. I., Lattanzio J. C., Pols O. R., 2002, Pub. As- tron. Soc. Aust., 19, 515 Keeley D. A., 1970, ApJ, 161, 643 Kippenhahn R., Ruschenplatt G., Thomas H.-C., 1980, A&A, 91, 175 Mixing processes in CEMP stars 13 Initial mass ( M⊙ ) 1H 4He 12C 14N 16O 20Ne 1 1.5 2 2.5 3 3.5 2.76×10−1 6.02×10−1 8.96×10−1 1.19×100 1.53×100 1.80×100 1.06×10−1 2.58×10−1 4.04×10−1 4.91×10−1 6.11×10−1 8.25×10−1 1.06×10−3 1.06×10−3 2.14×10−2 2.11×10−2 1.24×10−2 5.53×10−3 1.28×10−5 2.43×10−5 3.44×10−5 3.48×10−5 7.94×10−4 1.74×10−2 7.29×10−5 4.77×10−4 6.32×10−4 6.53×10−4 5.33×10−4 6.85×10−4 3.90×10−6 9.72×10−6 2.11×10−5 2.72×10−5 2.61×10−5 3.82×10−5 Mass lost Final age ( M⊙ ) (yrs) 3.84×10−1 8.72×10−1 1.32×100 1.71×100 2.15×100 2.65×100 5.83×109 1.63×109 7.48×108 4.22×108 2.70×108 1.88×108 Initial mass mass ( M⊙ ) 2H 3He 7Li 7Be 11B 13C 14C 15N 17O 18O 1 1.5 2 2.5 3 3.5 8.49×10−16 1.15×10−16 2.06×10−14 8.37×10−15 6.60×10−11 9.52×10−17 1.67×10−4 2.35×10−4 2.22×10−4 2.45×10−4 9.25×10−5 3.25×10−6 1.04×10−11 1.56×10−11 1.56×10−11 3.73×10−12 7.39×10−11 1.71×10−12 3.23×10−11 3.32×10−11 5.22×10−11 1.51×10−8 3.54×10−7 8.49×10−10 1.65×10−12 4.59×10−12 7.03×10−11 7.08×10−11 1.91×10−12 3.20×10−11 8.64×10−7 8.65×10−7 9.60×10−7 2.70×10−6 1.18×10−3 3.38×10−04 5.61×10−8 3.81×10−7 8.80×10−7 5.06×10−7 8.19×10−8 4.85×10−9 8.87×10−9 2.31×10−8 1.11×10−8 1.68×10−8 2.86×10−8 5.63×10−7 1.79×10−7 7.73×10−7 1.99×10−6 1.62×10−6 2.63×10−6 4.16×10−5 3.07×10−8 8.33×10−8 1.03×10−7 1.44×10−7 3.96×10−8 2.59×10−9 Initial mass mass ( M⊙ ) 19F 21Ne 22Ne 22Na 23Na 24Mg 25Mg 26Mg 26Alm 26Alg 1 1.5 2 2.5 3 3.5 7.87×10−8 7.70×10−7 1.05×10−6 3.18×10−7 7.31×10−8 5.68×10−9 9.86×10−9 8.26×10−8 7.13×10−7 9.63×10−7 3.60×10−7 3.70×10−7 1.79×10−5 3.33×10−4 1.09×10−3 6.97×10−4 1.94×10−4 1.57×10−4 2.24×10−13 5.01×10−13 4.78×10−12 7.71×10−12 9.34×10−10 3.59×10−7 3.03×10−7 4.44×10−6 2.19×10−5 1.57×10−5 6.28×10−6 1.56×10−4 1.05×10−6 3.32×10−6 1.13×10−5 1.26×10−5 8.21×10−6 1.39×10−5 2.05×10−7 3.49×10−6 3.24×10−5 5.21×10−5 2.75×10−5 7.20×10−5 1.89×10−7 1.95×10−6 2.87×10−5 9.36×10−5 7.07×10−5 2.00×10−4 Initial mass mass ( M⊙ ) 27Al 28Si 29Si 30Si 31P 32S 33S 34S 0.00 0.00 0.00 0.00 0.00 0.00 56Fe 1 1.5 2 2.5 3 3.5 1.12×10−7 2.61×10−7 6.07×10−7 1.65×10−6 1.74×10−6 4.18×10−6 1.26×10−6 2.85×10−6 4.42×10−6 6.41×10−6 7.86×10−6 1.12×10−5 6.64×10−8 1.54×10−7 2.59×10−7 4.64×10−7 5.60×10−7 1.04×10−6 4.57×10−8 1.08×10−7 1.85×10−7 3.47×10−7 4.58×10−7 9.34×10−7 1.62×10−7 3.93×10−7 6.32×10−7 1.01×10−6 1.29×10−6 2.61×10−6 7.64×10−7 1.72×10−6 2.59×10−6 3.38×10−6 4.34×10−6 5.61×10−6 6.29×10−9 1.49×10−8 2.30×10−8 2.98×10−8 3.75×10−8 5.32×10−8 3.59×10−8 8.01×10−8 1.19×10−7 1.55×10−7 2.00×10−7 2.47×10−7 2.24×10−6 5.00×10−6 7.40×10−6 9.62×10−6 1.24×10−5 1.51×10−5 Initial mass mass ( M⊙ ) 58Fe 59Fe 60Fe 59Co 58Ni 59Ni 60Ni 61Ni g 1 1.5 2 2.5 3 3.5 8.86×10−9 4.64×10−8 1.25×10−7 8.85×10−8 6.92×10−8 7.15×10−8 6.59×10−16 1.36×10−14 3.16×10−11 4.90×10−10 5.83×10−10 1.87×10−10 2.53×10−12 1.72×10−10 1.11×10−8 5.23×10−8 3.30×10−8 5.35×10−8 6.85×10−9 2.13×10−8 5.18×10−8 4.85×10−8 4.70×10−8 5.12×10−8 9.49×10−8 2.09×10−7 3.08×10−7 4.05×10−7 5.24×10−7 6.40×10−7 3.55×10−11 3.16×10−10 2.09×10−10 1.52×10−10 1.08×10−10 9.91×10−11 3.76×10−8 8.35×10−8 1.22×10−7 1.60×10−7 2.07×10−7 2.53×10−7 1.71×10−9 4.52×10−9 6.24×10−9 7.43×10−9 9.39×10−9 1.13×10−8 5.36×10−10 6.22×10−9 2.25×10−8 4.47×10−8 2.86×10−8 1.27×10−7 3.08×10−10 1.03×10−9 2.81×10−9 2.47×10−9 2.39×10−9 3.28×10−7 57Fe 5.56×10−8 1.34×10−7 2.04×10−7 2.47×10−7 3.09×10−7 3.75×10−7 Table A1. Final gross yields in stellar masses for all the isotopes in the nucleosynthesis network.
astro-ph/9911180
2
9911
1999-11-27T12:16:13
The redshift-space two-point correlation function of galaxy groups in the CfA2 and SSRS2 surveys
[ "astro-ph" ]
We measure the two-point redshift-space correlation function of loose groups of galaxies, xi_GG, for the combined CfA2 and SSRS2 surveys. Our combined group catalog constitutes the largest homogeneous sample available (885 groups). We compare xi_GG with the correlation function of galaxies, xi_gg, in the same volume. We find that groups are significantly more clustered than galaxies: <xi_GG/xi_gg>=1.64+-0.16. A similar result holds when we analyze a volume-limited sample (distance limit 78 h^-1 Mpc) of 139 groups. For these groups, with median velocity dispersion sigma_v sim 200 km s^-1 and mean group separation d sim 16 h^-1 Mpc, we find that the correlation length is s_0=8+-1 h^-1 Mpc, which is significantly smaller than that found for rich clusters. We conclude that clustering properties of loose groups of galaxies are intermediate between galaxies and rich clusters. Moreover, we find evidence that group clustering depends on physical properties of groups: correlation strengthens for increasing sigma_v.
astro-ph
astro-ph
A&A manuscript no. (will be inserted by hand later) Your thesaurus codes are: 02(12.12.1; 11.03.1; 11.19.7) ASTRONOMY AND ASTROPHYSICS September 3, 2018 9 9 9 1 v o N 7 2 2 v 0 8 1 1 1 9 9 / h p - o r t s a : v i X r a The redshift -- space two -- point correlation function of galaxy groups in the CfA2 and SSRS2 surveys Marisa Girardi1, Walter Boschin2, and Luiz N. da Costa3,4 1 Dipartimento di Astronomia, Universit`a degli Studi di Trieste, Via Tiepolo 11, I-34100 - Trieste, Italy ([email protected]) 2 Osservatorio Astronomico di Trieste, Via Tiepolo 11, I-34100 - Trieste, Italy ([email protected]) 3 European Southern Observatory, Karl-Schwarzschild-Str.2, D-85748 Garching bei Munchen, Germany ([email protected]) 4 Departamento de Astronomia CNPq/Observatorio Nacional, Rua General Jose Cristino, 77, Rio de Janeiro, Brazil 20.921 Received 9 August 1999 / accepted 5 November 1999 Abstract. We measure the two -- point redshift -- space cor- relation function of loose groups of galaxies, ξGG(s) , for the combined CfA2 and SSRS2 surveys. Our combined group catalog constitutes the largest homogeneous sam- ple available (885 groups). We compare ξGG(s) with the correlation function of galaxies, ξgg(s) , in the same vol- ume. We find that groups are significantly more clustered than galaxies: < ξGG/ξgg > =1.64 ± 0.16. A similar result holds when we analyze a volume -- limited sample (distance limit 78 h−1 Mpc) of 139 groups. For these groups, with median velocity dispersion σv ∼ 200 km s−1 and mean group separation d ∼ 16 h−1 Mpc, we find that the corre- lation length is s0 = 8 ± 1 h−1 Mpc, which is significantly smaller than that found for rich clusters. We conclude that clustering properties of loose groups of galaxies are inter- mediate between galaxies and rich clusters. Moreover, we find evidence that group clustering depends on physical properties of groups: correlation strengthens for increas- ing σv. Key words: cosmology: large-scale structure of Universe -- galaxies: clusters: general -- galaxies: statistics 1. Introduction Loose groups of galaxies, the low -- mass tail of the mass dis- tribution of galaxy systems, fill an important gap in the mass range from galaxies to rich clusters. Until now, clus- tering properties of loose groups have been studied on the basis of rather small samples. The results are consequently uncertain and even contradictory. Nevertheless, cluster- ing properties of groups are shown to be robust against the choice of the identification algorithm, provided sys- tems are identified with comparable number overdensity thresholds (Frederic 1995b). The two -- point correlation function, CF, of galaxies and of galaxy systems constitutes an important measure Send offprint requests to: M. Girardi ([email protected]) of the large -- scale distribution of galaxies (e.g., Davis & Peebles 1983; Bahcall & Soneira 1983; de Lapparent et al. 1988; Tucker et al. 1997; Croft et al. 1997). From galaxies to clusters the two -- point correlation function in redshift -- space, ξ(s), (e.g., Peebles 1980), is consistent, within errors, with a power -- law form ξ(s) = (s/s0)−γ with γ ∼ 1.5 − 2 for a variety of systems. The correla- tion length, s0, ranges from about 5 -- 7.5 h−1 Mpc for galaxies (e.g., Davis & Peebles 1983; Loveday et al. 1996; Tucker et al. 1997; Willmer et al. 1998; Guzzo et al. 1999) to s0 ∼> 15 h−1 Mpc for galaxy clusters (e.g., Bahcall & Soneira 1983; Postman et al. 1991; Peacock & West 1992; Croft et al. 1997; Abadi et al. 1998; Borgani et al. 1999; Miller et al. 1999; Moscardini et al. 1999). As far as loose groups are concerned, previous deter- minations of the CF are very uncertain. From the study of 137 groups (within CfA1) and 87 groups (within SSRS1), Jing & Zhang (1988) and Maia & da Costa (1990) re- spectively find that the group -- group CF, ξGG(s) , has a lower amplitude than the galaxy -- galaxy CF, ξgg(s) . An- alyzing 128 groups in a sub -- volume of CfA2N, Ramella et al. (1990; hereafter RGH90) find that the amplitudes of ξGG(s) and ξgg(s) are consistent (see also Kalinkov & Kuneva 1990). Finally, Trasarti-Battistoni et al. (1997) study 192 groups in the Perseus -- Pisces region and find that the amplitude of ξGG(s) exceeds that of ξgg(s) . The theoretical expectations for the relative strength of the group and galaxy clustering are also contradictory. Frederic (1995b) determines the correlation function for galaxy and group halos in CDM numerical simulations by Gelb (1992) and finds that groups are more strongly cor- related than galaxies. In contrast, Kashlinsky (1987), on the basis of an analytical approach to the clustering prop- erties of collapsed systems of different masses, concludes that groups and individual galaxies should be correlated with the same amplitude. Here we compute the two -- point correlation function (in redshift space) for 885 groups of galaxies identified in the combined CfA2 and SSRS2 redshift surveys. This sample is characterized by its large extent (more than five 2 M. Girardi et al.: The correlation function of galaxy groups Table 1. Data Samples Sample NG CfA2N 395 CfA2S 139 SSRS2N 123 228 SSRS2S TOTAL 885 Ng 5426 2104 1697 3083 12290 times the volumes previously studied) and by the homo- geneity of the identification process (the friends -- of -- friends algorithm FOFA; Ramella et al. 1997 -- hereafter RPG97). Moreover, we compare the group -- group CF to that com- puted for galaxies in order to determine the relative clus- tering properties of groups and galaxies. Because we use the same galaxy sample where groups are identified, we avoid possible effects of fluctuations due to the volume sampled. In Sect. 2 we briefly describe the data; in Sect. 3 we describe the estimation of the two -- point correlation func- tion; in Sect. 4 we compute the correlation function of groups and compare it to that for galaxies; in Sect. 5 we summarize our results and draw our conclusions. Throughout the paper, errors are at the 68% confi- dence level, and the Hubble constant is H0 = 100 h Mpc−1 km s−1 . 2. Galaxy and groups catalogs We extract the sample of galaxies from the CfA2 North (CfA2N) and South (CfA2S) (Geller & Huchra 1989; Huchra et al. 1995; Falco et al. 1999), and the SSRS2 North (SSRS2N) and South (SSRS2S) (da Costa et al. 1998) redshift surveys. These surveys are complete to mB(0) ≃ 15.5 and cover more than one -- third of the sky, i.e. most of the extragalactic sky. The original papers con- tain detailed descriptions of the observations and of the data reduction. The velocities we use are heliocentric; they include corrections for solar motions with respect to the Local Group and for infall toward the center of the Virgo cluster (see RPG97 for details). As in previous analyses of the CfA2 surveys (e.g., Park et al. 1994; Marzke et al. 1995), we discard regions of large galactic extinction. The total sample includes 13435 galaxies with radial velocity V < 15000 km s−1 . We use the catalogs of groups identified within CfA2N by RPG97 and within SSRS2 by Ramella et al. (in prepa- ration). The identification method is a friends -- of -- friends algorithm (FOFA) which selects systems of at least three members above a given number density threshold in red- shift space. In particular, RPG97 and Ramella et al. (in preparation) use the number density threshold δρN/ρN = 80 and a line -- of -- sight link V0 = 350 km s−1 at the fidu- cial velocity Vf = 1000 km s−1 . We run FOFA with these parameters on CfA2S and produce a group catalog for this survey, too. The combined catalog contains a total of Fig. 1. We show the projection on the sky of the group and galaxy samples (upper and lower panel, respectively). 885 groups that constitute a homogeneous set of systems objectively identified in redshift space. The group catalogs are limited to radial velocities V ≤ 12000 km s−1 , but members are allowed out to V ≤ 15000 km s−1 . We confine the galaxy sample to V ≤ 12000 km s−1 and are left with a total of 12290 galaxies. Table 1 lists the numbers of groups, NG, and the num- bers of galaxies, Ng, for each sample. Fig. 1 shows the distribution of the galaxy and group samples on the sky. 3. Estimation of the correlation function We compute the two -- point correlation functions in red- shift space for groups and galaxies (hereafter ξGG(s) , and ξgg(s) , respectively). The formalism in the two cases is the same. We define the separation in the redshift space, s, as : qV 2 i + V 2 j − 2ViVjcosθij H0 s = , (1) where Vi and Vj are the velocities of two groups (or galax- ies) separated by an angle θij on the sky. Following Hamil- ton (1993) we estimate ξ(s) with: ξ(s) = DD(s)RR(s) [DR(s)]2 − 1, (2) where DD(s), RR(s), and DR(s) are the number of data -- data, random -- random, and data -- random pairs, with sep- arations in the interval (s, s + ds). We build the control M. Girardi et al.: The correlation function of galaxy groups 3 sample by filling the survey volume with a uniform ran- dom distribution of the same number of points as in the data. The points are distributed in depth according to the selection function of the surveys, Φ(V ). In order to decrease the statistical fluctuations in the determination of ξ(s), we average the results obtained us- ing several different realizations of the control sample. We compute 50 realizations in the case of groups and 5 in the case of galaxies. Unless otherwise specified, we compute the "weighted" correlation function by substituting the counts of pairs with P wiwj, the weighted sum of pairs, which takes into account the selection effects of the sample used. In the case of a sample characterized by the same selection function, Φ(V ), volumes are equally weighted and wi = 1/Φ(Vi) is the weight of a group (or galaxy) with velocity Vi. The appropriate selection function for a magnitude -- limited sample (e.g., de Lapparent et al. 1988; Park et al. 1994; Willmer et al. 1998) is: Φ(V ) = R M(V ) −∞ φ(M )dM R Mmax −∞ φ(M )dM , (3) where φ(M ) is the Schechter (1976) form of the lumi- nosity function. Mmax is a low luminosity cut -- off. We chose Mmax = −14.5, the absolute magnitude correspond- ing to the limiting apparent magnitude of the survey at the fiducial velocity Vf = 1000 km s−1 . This value of Vf is the same as in RPG97. The Schechter parameters of the galaxy luminosity function, before the Malmquist bias correction, are: M ∗ = −19.1, α = −1.1 for CfA2, and M ∗ = −19.7, α = −1.2 for SSRS2 (Marzke et al. 1994; 1998). We assume that the group selection function is the same as for galaxies. In fact, the velocity distributions of groups, NG(V ), and of galaxies, Ng(V ), are not sig- nificantly different according to the Kolmogorov -- Smirnov test (cf. Fig. 2). RGH90 and Trasarti-Battistoni et al. (1997), and Frederic (1995b) make the same assumption for observed and simulated catalogs, respectively. The different luminosity functions of CfA2 and SSRS2 correspond to different selection functions. For this reason we assign to a group (galaxy) i, belonging to the subsam- ple k, the weight given by: wi = 1 Φk(Vi)nk , (4) Fig. 2. The velocity distributions of groups (solid line) and of galaxies (dotted line) are compared. The error bands represent 1 − σ Poissonian errors. Fig. 3. The group-group (filled circles) , and galaxy- galaxy (open circles) CFs, as computed for the combined sample. The error bands represent 1 − σ bootstrap errors. pairs of groups (galaxies) linking two different subsam- ples. In this way we also avoid crossing large unsurveyed regions of the sky. We compute the errors on ξ(s) from 100 bootstrap re -- samplings of the data (e.g., Mo et al. 1992). Note that the bootstrap -- resampling technique, which overestimates the error in individual bins, represents a conservative choice in this work. where nk is the mean number density of groups (galaxies) of that subsample (e.g. Hermit et al. 1996). We compute the density as nk = 1/Vk · Σi[1/Φ(Vi)], where the sum is over all the groups (galaxies) of the subsample volume, Vk (Yahil et al. 1991). In our analysis, Vk is the effective volume of the subsample. Because the different selection functions, we also build a control sample for each sub- sample separately, and, conservatively, we do not consider 4. The group -- group correlation function We plot the group -- group CF, ξGG(s) , in Fig. 3. In the same figure we also plot the galaxy -- galaxy CF, ξgg(s) . On small scales (s ∼< 3.5 Mpc ) ξGG(s) starts dropping because of the anti -- correlation due to the typical size of groups. On large scales (s ∼> 15 h−1 Mpc) the signal -- to -- noise ratio of ξGG(s) drops drastically. We thus limit our analysis to the separation range 3.5 ∼< s ∼< 15 h−1 Mpc. 4 M. Girardi et al.: The correlation function of galaxy groups of the triples and quadruples in group catalogs could be spurious. We consider the 321 rich groups with Nmem ≥ 5, and find again that groups are more correlated than galax- ies. Moreover, we find evidence that the CF computed for richer groups is higher than the CF computed for poorer groups, i.e. < ξGG,rich/ξgg >= 1.48 ± 0.16 and < ξGG,poor/ξgg >= 1.10 ± 0.14 (cf. Fig. 5). The component of spurious groups among poor groups could be responsible for the lower amplitude of the ξGG(s) of poor groups compared to the ξGG(s) of rich groups. In fact, spurious groups should be distributed like non -- member galaxies. However, at least part of the ob- served higher clustering amplitude of rich groups could be due to the existence of a clustering amplitude vs rich- ness relationship. The relationship has been discussed for a variety of systems by several authors (e.g. Bahcall & West 1992; Croft et al. 1997; Miller et al. 1999). Richness is usually taken as a measure of the mass of the system, mass being the real, interesting physical quantity directly related to the predictions of cosmological models. 4.2. The CF in the volume -- limited sample For our groups, richness is not a good physical parame- ter. A better parameter is the group (line -- of -- sight) veloc- ity dispersion, σv (e.g. RPG97). In a magnitude -- limited sample any group selection based on velocity dispersion will affect the selection function in an "a priori" un- known way. To avoid this problem, we analyze the volume -- limited group sample built by Ramella et al. (in prepa- ration) who run an appropriately modified version of FOFA within volume -- limited sub -- samples of the CfA2 and SSRS2 galaxy surveys. In particular, we consider the 139 distance limited groups within V ≤ 7800 km s−1 , roughly corresponding to the effective depth of CfA2. We cut the volume -- limited galaxy catalogs in the same way. Within this sample we compute the "unweighted" CF es- timator, i.e. we set w = 1 for all groups/galaxies. For this sample the useful s -- range is 3.5 ∼< s ∼< 12 h−1 Mpc (see Fig. 6). We find that the ratio < ξGG/ξgg > of the total volume -- limited sample is < ξGG/ξgg > = 1.58 ± 0.10, sim- ilar to that computed for the magnitude -- limited sample (< ξGG/ξgg > ∼ 1.6). This result reassures us about the reliability of the selection function we assume for groups. In order to check a possible dependence of ξGG(s) on σv, we divide the group volume -- limited sample into two subsamples of equal size, one subsample containing groups with σv ≥ 214 km s−1 , the other including the remain- ing low velocity dispersion groups. We find that high -- σv systems are more correlated than those with low σv (< ξGG/ξgg > =2.14±0.37 and < ξGG/ξgg > =1.29±0.17, respectively). This evidence is in agreement with that found for clusters, and suggests a continuum of cluster- ing properties for all galaxy systems. Fig. 4. The ratio ξGG/ξgg , where the CFs are those com- puted for the combined sample in Fig. 3. Fig. 5. The ratio ξGG/ξgg for the combined sample (thin line), but, now, ξGG(s) is computed also for groups with at least five member galaxies (thick line) and for groups with less than five member galaxies (dashed line). The main physical result in Fig. 3 is that ξGG(s) has a larger amplitude than ξgg(s) . This property of the CFs is also evident in Fig. 4, where we plot the ratio ξGG(s) / ξgg(s) on a linear scale. Over the s -- range of interest, the values of the ratio are roughly constant within the errors. In order to give an estimate of the relative behavior of groups and galaxies we compute the mean of the values of the ratio. We obtain < ξGG/ξgg > =1.64 ± 0.16. 4.1. The CF of rich groups Groups with a number of members Nmem ≥ 5 are gener- ally reliable, as shown both by optical and X -- ray analyses (Ramella et al. 1995; Mahdavi et al. 1997). On the other hand, the reliability of groups with fewer members is of- ten questionable. In particular, the analysis of the CfA2N survey performed by RPG97, and the analysis of a CDM model by Frederic (1995a) show that a significant fraction M. Girardi et al.: The correlation function of galaxy groups 5 4.3. The unweighted CF In order to verify the stability of our results against varia- tions of the weighting scheme, we compute the unweighted CF, ξUW, for the magnitude -- limited sample. We find that, as in the weighted case, the amplitude of ξUW GG (s) is still significantly higher than the amplitude of ξUW gg (s) , GG /ξUW < ξUW gg > =1.18 ± 0.05. We also find that the ampli- GG (s) and ξUW tudes of ξUW gg (s) are both significantly lower than the weighted estimates. The differences between the results of the two weight- ing schemes rise from the fact that the weighted CF weights each volume of space equally and therefore bet- ter traces the clustering of more distant objects. In fact, when we divide the group/galaxy catalogs in two subsam- ples of equal size according to group/galaxy distances, the distant samples (V > 6680 km s−1 ) give CFs GG,distant/ξUW with higher amplitude, i.e. < ξUW GG,nearby >= 2.15 ± 0.31 and < ξUW gg,distant/ξUW gg,nearby >= 1.43 ± 0.08. Moreover, the distant samples (V > 6680 km s−1 ) give < ξUW gg,distant >= 1.43 ± 0.12 in closer agree- ment with the result of the weighted analysis. The ratio ξUW GG /ξUW for the whole sample, as well as for its nearby gg and distant parts, is shown in Fig. 7. GG,distant/ξUW explanation, the fact As for a physical that ξUW gg (s) <ξgg(s) could be the consequence of a depen- dency of clustering on luminosity, since the unweighted CF estimator is more sensitive to the clustering of nearer, fainter groups/galaxies (e.g., Park et al. 1994). In fact, the dependency of clustering on luminosity has been pointed out for the galaxy-galaxy CF (e.g., Benoist et al. 1996; Cappi et al. 1998; Willmer et al. 1998). In addition, the greater strength of ξgg(s) could be explained by different clustering properties in different volumes of the Universe: e.g., Ramella et al. (1992) find that the strength of the galaxy CF is very high in the Great Wall. In the volume we examine the two biggest structures, the Great Wall and the Southern Wall, both lie in distant regions (e.g. da Costa et al. 1994) and therefore their weight is larger in the weighted CF scheme. It is reasonable to expect also that distant groups, which are brighter (and presumably more massive) and which preferably lie in the two big structures, are more strongly correlated than nearby groups leading to the observed ξUW GG (s) <ξGG(s) . 5. Summary and conclusions We measure the two -- point redshift -- space correlation func- tion of loose groups, ξGG(s) , for the combined CfA2 and SSRS2 surveys. Our combined group catalog con- stitutes the largest homogeneous sample available (885 groups). We compare ξGG(s) with the correlation func- tions of galaxies, ξgg(s) , in the same volumes. Our main results are the following: 1. Using the whole sample we find that groups are sig- nificantly more clustered than galaxies, < ξGG/ξgg > Fig. 6. The group-group (filled circles) and galaxy-galaxy (open circles) CFs, as computed for the volume -- limited sample. Fig. 7. The ratio ξUW for the combined sample (thin line), and its nearby (dashed line) and distant (thick line) parts, as computed with the unweighted CF. GG /ξUW gg In this context, it is appropriate to compare the groups of the volume -- limited sample, characterized by the me- dian velocity dispersion σv = 214 km s−1 and by the mean group separation d ∼ 16 h−1 Mpc, to rich clus- ters (σv ∼ 700 km s−1 ; d ∼ 50 h−1 Mpc; e.g. Zabludoff et al. 1993; Peacock & West 1992). We fit ξGG(s) to the form ξ(s) = (s/s0)−γ with a non -- linear weighted least squares method and find γ = 1.9 ± 0.7 and s0 = 8 ± 1 h−1 Mpc. Note that groups show similar slope but signif- icantly smaller correlation length than optically or X -- ray selected clusters, for which s0 ∼> 15 h−1 Mpc (e.g. Bahcall & West 1992; Croft et al. 1997; Abadi et al. 1998; Borgani et al. 1999; Miller et al. 1999). Our results agree with the predictions of those N -- body cosmological simulations that also correctly predict the observed cluster -- cluster CF (e.g. cf. our (s0, d) with Fig. 8 of Governato et al. 1999). 6 M. Girardi et al.: The correlation function of galaxy groups =1.64 ± 0.16, thus consistent with the result by Trasarti-Battistoni et al. (1997), based on a much smaller sample. This ratio can be considered a lower limit considering the possible presence of unphysical groups. 2. Groups are significantly less clustered than clusters. In particular, we find γ = 1.9 ± 0.7 and s0 = 8 ± 1 for 139 groups identified in a volume -- limited sample (V ≤ 7800 km s−1 , median velocity dispersion σv ∼ 200 km s−1 , and mean group separation d ∼ 16 h−1 Mpc). This result can be compared with that of galaxy clusters (s0 ∼ 15 -- 20 h−1 Mpc for systems with σv ∼ 700 km s−1 and d ∼ 50 h−1 Mpc; e.g., Bahcall & West 1992; Croft et al. 1997; Abadi et al. 1998; Borgani et al. 1999; Miller et al. 1999). 3. There is a tendency of clustering amplitude to increase with group velocity dispersion σv, which is the better indicator of group mass at our disposal. We conclude that there is a continuum of clustering prop- erties of galaxy systems, from poor groups to very rich clusters, with correlation length increasing with increas- ing mass of the system. Acknowledgements. We thank Stefano Borgani, Antonaldo Di- aferio, and Margaret Geller for useful discussions. Special thanks to Massimo Ramella for enlightening suggestions. M.G. wishes to acknowledge Osservatorio Astronomico di Trieste for a grant received during the preparation of this work. This work has been partially supported by the Italian Ministry of Univer- sity, Scientific Technological Research (MURST), by the Ital- ian Space Agency (ASI), and by the Italian Research Council (CNR-GNA). References Abadi M. G., Lambas D. G., Muriel H., 1998, ApJ 507, 526 Bahcall N. A., Soneira R. M., 1983, ApJ 270, 20 Bahcall N. A., West M. J., 1992, ApJ 392, 419 Benoist C., Maurogordato S., da Costa L. N., Cappi A., Schaf- Geller M. J., Huchra J. P., 1989, Science 246, 897 Governato F., Babul A., Quinn T., et al., 1999, MNRAS 307, 949 Guzzo L., Bartlett J. G., Cappi A., et al., 1999, A&A submitted Hamilton A. J. S., 1993, ApJ 417, 19 Hermit S., Santiago B. X., Lahav O., et al., 1996, MNRAS 283, 709 Huchra J. P., Geller M. J., Corwin H. G. Jr., 1995, ApJS 99, 391 Jing Y., Zhang J., 1988, A&A 190, L21 Kalinkov M., Kuneva I., 1990, IAU Colloq. 124, 149 Kashlinsky A., 1987, ApJ 317, 19 Loveday J., Efstathiou G., Maddox S. J., Peterson B. A., 1996, ApJ 468, 1 Mahdavi A., Boringher H., Geller M. J., Ramella M., 1997, ApJ 483, 68 Maia M. A. G., da Costa L. N., 1990, ApJ 349, 477 Marzke R. O., Huchra J. P, Geller M. J., 1994, ApJ 428, 43 Marzke R. O., Geller M. J., da Costa L. N., Huchra J. P., 1995, AJ 110, 477 Marzke R. O., da Costa L. N., Pellegrini P. S., Willmer C. N. A., Geller M. J., 1998, ApJ 503, 517 Miller C. J., Ledlow M. J., Batuski D. J., 1999, MNRAS sub- mitted Mo H. J., Jing Y. P., Borner G., 1992, ApJ 392, 452 Moscardini L., Matarrese S., De Grandi S., Lucchin F., 1999, MNRAS submitted Park C., Vogeley M. S., Geller M. J., Huchra J. P., 1994, ApJ 431, 569 Peacock J. A., West M. J., 1992, MNRAS 259, 494 Peebles P. J. E., 1980, in The Large-Scale Structure of the Universe, Princeton Univ. Press, Princeton Postman M., Huchra J. P., Geller M., 1991, ApJ 384, 404 Ramella M., Geller M. J., Huchra J. P., 1990, ApJ 353, 51 (RGH90) Ramella M., Geller M. J., Huchra J. P., 1992, ApJ 384, 396 Ramella M., Geller M. J., Huchra J. P., Thorstensen J. R., 1995, AJ 109, 1469 Ramella M., Pisani A., Geller M. J., 1997, AJ 113, 483 (RPG97) Schechter P. L., 1976, AJ 203, 297 Trasarti-Battistoni R., Invernizzi G., Bonometto S. A., 1997, fer R., 1996, ApJ 472, 452 ApJ 475, 1 Borgani S., Plionis M., Kolokotronis V., 1999, MNRAS 305, Tucker D. L., Oemler A. JR., Kirshner R. P., et al., 1997, 866 MNRAS 285, L5 Cappi A., da Costa L. N., Benoist C., Maurogordato, S. Pelle- Willmer C. N. A., da Costa L. N., Pellegrini P. S., 1998, AJ grini P. S., 1998, AJ 115, 2250 115, 869 Croft R. A. C., Dalton G. B., Efstathiou G., Sutherland W. J., Yahil A., Strauss M. A., Davis M., Huchra J. P., 1991, ApJ Maddox S. J., 1997, MNRAS 291, 305 372, 380 da Costa L. N., Geller M. J., Pellegrini P. S., et al., 1994, ApJ Zabludoff A., Geller M. J., Huchra J. P., Ramella M., 1993, AJ 424, L1 106, 1301 da Costa L. N., Willmer C. N. A., Pellegrini P.S., et al., 1998, AJ 116, 1 Davis M., Peebles P. J. E., 1983, ApJ 267, 465 de Lapparent V., Geller M. J., Huchra J. P., 1988, ApJ 332, 44 Falco E. E., Kurtz M. J., Geller M. J., et al., 1999, PASP 111, 438 Frederic J. J., 1995a, ApJS 97, 259 Frederic J. J., 1995b, ApJS 97, 275 Gelb J. M., 1992, PhD Thesis, M.I.T.
astro-ph/0105207
1
0105
2001-05-11T23:00:05
Diffuse Gamma-Rays from Local Group Galaxies
[ "astro-ph" ]
Diffuse gamma-ray radiation in galaxies is produced by cosmic ray interactions with the interstellar medium. With the completion of EGRET observations, the only extragalactic object from which there has been a positive detection of diffuse gamma-ray emission is the Large Magellanic Cloud. We systematically estimate the expected diffuse gamma-ray flux from Local Group galaxies, and determine their detectability by new generation gamma-ray observatories such as GLAST. For each galaxy, the expected gamma-ray flux depends only on its total gas content and its cosmic ray flux. We present a method for calculating cosmic ray flux in these galaxies in terms of the observed rate of supernova explosions, where cosmic ray acceleration is believed to take place. The difficulty in deriving accurate supernova rates from observational data is a dominant uncertainty in our calculations. We estimate the gamma-ray flux for Local Group galaxies and find that our predictions are consistent with the observations for the LMC and with the observational upper limits for the Small Magellanic Cloud and M31. Both the Andromeda galaxy, with a flux of $\sim 1.0 \times 10^{-8}$ photons sec$^{-1}$ cm$^{-2}$ above 100 MeV, and the SMC, with a flux of $\sim 1.7 \times 10^{-8}$ photons sec$^{-1}$ cm$^{-2}$ above 100 MeV, are expected to be observable by GLAST. M33 is at the limit of detectability with a flux of $\sim 0.11 \times 10^{-8}$ sec$^{-1}$ cm$^{-2}$. Other Local Group galaxies are at least two orders of magnitude below GLAST sensitivity.
astro-ph
astro-ph
DIFFUSE GAMMA-RAYS FROM LOCAL GROUP GALAXIES Vasiliki Pavlidou and Brian D. Fields Department of Astronomy and Center for Theoretical Astrophysics University of Illinois Urbana, IL 61801, USA ABSTRACT Diffuse γ-ray radiation in galaxies is produced by cosmic ray interactions with the interstellar medium. With the completion of EGRET observations, the only extragalac- tic object from which there has been a positive detection of diffuse γ-ray emission is the Large Magellanic Cloud. We systematically estimate the expected diffuse γ-ray flux from Local Group galaxies, and determine their detectability by new generation γ-ray observatories such as GLAST. For each galaxy, the expected γ-ray flux depends only on its total gas content and its cosmic ray flux. We present a method for cal- culating cosmic ray flux in these galaxies in terms of the observed rate of supernova explosions, where cosmic ray acceleration is believed to take place. The difficulty in deriving accurate supernova rates from observational data is a dominant uncertainty in our calculations. We estimate the γ-ray flux for Local Group galaxies and find that our predictions are consistent with the observations for the LMC and with the observational upper limits for the Small Magellanic Cloud and M31. Both the Andromeda galaxy, with a flux of ∼ 1.0 × 10−8 photons sec−1 cm−2 above 100 MeV, and the SMC, with a flux of ∼ 1.7 × 10−8 photons sec−1 cm−2 above 100 MeV, are expected to be observable by GLAST. M33 is at the limit of detectability with a flux of ∼ 0.11 × 10−8 sec−1 cm−2. Other Local Group galaxies are at least two orders of magnitude below GLAST sensitivity. Subject headings: gamma rays: theory -- cosmic rays -- Local Group 1. Introduction All-sky EGRET images (Hunter et al. 1997) dramatically show that the γ-ray flux above 100 MeV is dominated by emission from the Galactic disk. This emission can be well understood (Strong 1996, Sreekumar et al. 1998, Hunter et al. 1997) in terms of cosmic ray interactions with the interstellar medium. At energies & 100 MeV, the generation of diffuse γ−ray emission is dominated -- 2 -- by the decay of π0 produced in collisions between cosmic ray nuclei and interstellar medium nuclei. At lower energies, the dominant emission mechanism is bremsstrahlung from energetic cosmic ray electrons. Given the prominence of the Milky Way diffuse γ-ray emission, it is natural to ask whether we can see the corresponding emission originating from other galaxies. To explore the possibility of such a detection requires an understanding of the γ-ray emissivity due to interactions of cosmic rays with matter. The problem of calculating the γ-ray emissivity per hydrogen atom given a cosmic-ray spectrum has been treated in detail by Stecker (1970; 1973; 1988) and Dermer (1986), who have made use of both experimental data and theoretical models on the relevant collisional cross sections. The results of these studies are also consistent with more recent calculations by Mori (1997) who used Monte Carlo event simulators and recent accelerator data. These studies reproduce the Milky Way (MW) γ-ray flux quite well as a function of energy and angle on the sky (at least in the 30 -- 500 MeV range; at higher energies, the situation is less clear, and inverse Compton scattering of a hard electron component may be important; see Strong, Moskalenko, & Reimer (2000)). By combining this information with an estimate of the gas content and the cosmic ray flux level and spectrum for a given galaxy, one can arrive at a prediction for its total γ-ray flux. Extended studies of this kind have been made for the Magellanic Clouds. Fichtel et al. (1991) calculated the flux expected from the Large Magellanic Cloud (LMC). Their estimate of the cosmic ray flux was based on the assumption that the expansive pressures of the gas, magnetic field and cosmic rays are in dynamic balance with the gravitational attraction of matter. Subsequent EGRET observations (Sreekumar et al 1992) resulted in the detection of the LMC with a γ-ray flux consistent with the prediction by Fichtel et al. (1991). Sreekumar & Fichtel (1991) arrived at three different predictions for the Small Magellanic Cloud (SMC) diffuse γ-ray flux, each one based at a different assumption for the level of the SMC cosmic ray flux: a dynamic balance among thermal, magnetic, and cosmic ray pressures (as in the case of the LMC); a constant ratio of cosmic ray electrons and protons (where the electron flux was deduced from observational synchrotron data); and a "universal" cosmic ray flux (at the same level as in the solar neighborhood). EGRET observations have not led to a positive detection of the SMC in high-energy γ-rays, placing an upper limit to the γ-ray flux which excludes the predictions based on a dynamic balance or a "universal" cosmic ray flux (Sreekumar et al. 1993, Lin et al. 1996). In this way, the debate over the origin of cosmic rays (Galactic versus extragalactic) was settled observationally and the cosmic rays were shown to be originating from within the Galaxy. Apart from the Magellanic Clouds, the only other Local Group galaxy for which there have been theoretical γ-ray flux predictions is M31.1 Ozel & Berkhuijsen (1987) and Ozel & Fichtel 1Theoretical studies have also been made for starburst galaxies such as M82 (Akyuz et al. 1991) and NGC253 (Paglione et al. 1996) since these objects were good candidates for detection by EGRET due to their high supernova rate and presumably high cosmic ray flux. However, no positive detection of these objects has been achieved (Blom et al. 1999). -- 3 -- (1988), based on the observational data available at the time for the distance and gas content of M31, concluded that if the cosmic ray flux in the Andromeda galaxy is comparable to that in the Milky Way, then the galaxy should be detectable by EGRET. However, Blom, Paglione, & Carraminana (1999) showed that EGRET has not detected M31, and have instead placed an upper limit for its γ-ray flux lower than the theoretically predicted value of Ozel & Berkhuijsen (1987). In this work we systematically study the γ-ray emission from Local Group galaxies and its detectability; with the completion of EGRET observations and the prospect of the construction of new, more sensitive γ-ray observatories, such an investigation is timely. We survey the latest avail- able data for the gas contents of various Local Group galaxies, with attention to the uncertainties in the observational inputs, and their impact on the predicted emission. We also present a new method for computing the global mean cosmic ray flux and, thus the γ-ray flux, from the observed properties of the extragalactic sources. To do this we use the ratio of the supernova (SN) rate in each galaxy to the supernova rate of the Milky Way to calibrate the magnitude of the cosmic ray flux. This association is justified by the fact that cosmic ray acceleration is believed to take place in supernova remnants, an idea supported both by theoretical arguments (e.g., Ellison, Drury, & Meyer (1997)) as well as observational evidence (e.g., Koyama et al. 1995, Combi et al. 1998). Furthermore, we find this method leads to an estimate of the LMC γ-ray flux which is in excellent agreement with the level observed by EGRET. Our predictions for the Local Group will be testable by the forthcoming Gamma-ray Large Area Space Telescope (GLAST), which is predicted to be launched in 2005. The proposed design of GLAST, described in De Angelis (2000) and in the GLAST webpage (http://glast.gsfc.nasa.gov), is that of a 1 m2 effective area detector, sensitive to energies from 20 MeV to 300 GeV, with a field of view of 2.4 sr. This large field of view will allow any individual source to be observed for 20% of the duty cycle when GLAST is operating in the normal sky-scanning mode. The single-photon angular resolution will be 3◦ at 100 MeV and about 0◦.2 at 10GeV. The nominal predicted lifetime of the GLAST mission is 5 years, with a goal of 10 years of operation. These specifications imply a sensitivity for GLAST of ∼ 2 × 10−9 photons cm−2 s−1 (5-σ detection for a point source at high Galactic latitudes after a 2-year all sky survey), better by more than an order of magnitude than that of EGRET. Recent work by Digel, Moskalenko, Ormes, Sreekumar, & Williamson (2000) has drawn atten- tion to GLAST's potential for observing the three brightest Local Group sources. These authors presented simulated maps of the diffuse emission from the Magellanic clouds using models of Sreeku- mar, and discussed the observability of M31 on the basis of current observational upper limits. Our work gives theoretical support to this effort, as we make a robust prediction that M31 will be de- tected by GLAST. By considering the entire Local Group, we also find that M33 could be detectable by GLAST. In Section 2 we present the theoretical background of our calculations. We discuss the γ−ray emissivity per H atom for a given cosmic ray spectrum, by now established both theoretically and -- 4 -- observationally. The "leaky box" model for cosmic ray propagation leads to the prediction that the CR flux should be proportional to a mean SN rate in the galaxy under consideration. Using this fact and an estimate of the gas content of each galaxy of the Local Group, we arrive at a prediction for its γ-ray flux. In Section 3 we present the data on the SN rates of the MW and the galaxies under study as well as their gas contents. In Section 4 we describe our results, compare our predictions for the Magellanic Clouds and M31 with the EGRET observations, and compare these and predictions for other Local Group galaxies with the anticipated sensitivity of GLAST. Discussion of our findings and conclusions follows in Section 5. 2. Gamma-Ray Production Inelastic collisions of high-energy cosmic ray protons with the interstellar medium protons results most frequently to the production of neutral pions (π0) which then decay, with a probability ∼ 99%, to two gamma ray photons: p + p → p + p + π0 ֒→ γ + γ If φp E(Tp) is the differential energy spectrum of the cosmic ray protons (protons/(cm2-s-GeV-sr)) and hζ σπ(Tp)i is the inclusive cross-section for the production of π0 from p-p collisions, then the total π0 production rate per target H atom (Γπ0) is given by: Γpp→π0 = 4πZ ∞ 0 dT hζ σπ(T )i φp E (T ) (1) In the case of the Milky Way, various authors (Stecker 1970, 1973, 1988; Cavallo & Gould 1971; Stephens & Badhwar 1981; Dermer 1986; Mori 1997) have calculated Γπ0. The results are consistent with each other within errors. Here we adopt Γpp→π0 ≈ 7.0 × 10−26 π0 (s - H atom )−1 . (2) There is an additional contribution to π0 production by interactions involving nuclei with A > 1 either in cosmic rays or in the interstellar medium (predominantly p + α and α + α). This contribution increases the estimated pion production rate by a multiplicative factor (Stecker 1970), which, when including α as well as nuclei heavier than helium both in the cosmic rays and in the interstellar medium, is ∼ 1.5 (e.g., Mori (1997) and refs. therein). Taking into account this correction and the fact that each pion produces two γ-rays, the total γ-ray emissivity per hydrogen atom from π0 decay only is qπ0 γ = 2.0 × 10−25 photons (s - H atom)−1 , (3) in good agreement with the theoretical calculations cited above. Equation (3) includes γ-rays of all energies. Now if we confine our interest in the energy range > 100MeV, this emissivity decreases -- 5 -- by a factor of 0.76. In the same energy range, Fichtel and Kniffen (1984) have estimated the contribution in the γ-ray emissivity from cosmic ray electron bremsstrahlung to be 55% of the flux produced by neutral pion decay. 2 In this way, we can calculate a total Galactic γ-ray emissivity per hydrogen atom, for photon energies > 100 MeV , originating by all π0-producing CR-ISM collisions as well as bremsstrahlung radiation from cosmic ray electrons: qγ(> 100MeV) = 2.4 × 10−25 photons / (s - H atom). (4) which is consistent with observational data (e.g., Digel et al. 1995) In order to extend this calculation to galaxies other than our own, we must relate the γ- ray production and hence cosmic ray flux, to observable properties of the galaxies. We therefore must account for both the acceleration and propagation of cosmic rays, and their dependence on the galactic environment. The assumption that supernova explosions are the engines of CR acceleration is encoded in simple and direct way. Specifically, we will impose a scaling of the CR source (injection) rate Qp with RG , the mean SN rate in a specific galaxy G: QG p ∝ RG . To describe the propagation of cosmic rays requires more detailed treatment. Cosmic rays propagate diffusively as they spiral along galactic magnetic field lines. The particles sources are acceleration sites (i.e., supernova remnants) which are distributed inhomogeneously; the sinks are energy losses due to a variety of processes, as well as escape from the galaxy. Cosmic ray propagation is thus properly described by a diffusion equation which accounts for all of these effects as a function of position in the galaxy. However, for our purposes of describing a galactic-averaged cosmic ray flux, the treatment of propagation can be stripped to its essential features. To do this, we make the following simplifying assumptions: • The shape of the cosmic ray energy spectrum is the same throughout each galaxy and identical to that in the Milky Way (although the normalization may be different). • The ratios of p to α as well as the ratio of CR electrons to CR protons is constant and same to that in the Milky Way. • The propagation of cosmic rays in the galaxy can be approximated in terms of the "leaky box" model. The first two assumptions appeal to a universality in the underlying physics of cosmic ray acceler- ation by individual supernova remnants. The last assumption treats the galaxy as a single zone to be modeled, as we now see. 2The situation above 500 MeV is perhaps less clear. Strong, Moskalenko, & Reimer (2000) have recently suggested that the dominant source in this regime may be inverse Compton scattering of interstellar radiation off a hard electron component. -- 6 -- According to the "leaky box" model, the galaxy is treated as a homogeneous containment environment where CRs propagate freely with a certain probability per unit time to escape. Within the frame of this model the propagation equation becomes ∂Ni(T, t) ∂t = Qi(T, t) + ∂ ∂T [bi(T )Ni(T, t)] − 1 τesc Ni(T, t). (5) Here Ni(T, t) is the density of particles of species i with kinetic energy between T and T + dT . Qi(T, t) is the source term (particles per volume per time per energy interval dT ) including all sources of species i. As we are concerned with p and α particles, the source term Qi simply represents the supernova acceleration as a source these "primary" species (as opposed to "secondary" particles created in flight, such as Li, Be, and B). Additional spallation contribution and losses are negligible at our level of accuracy. The second term in the right hand side of eq. (5) represents energy losses due to ionization of the ISM, with a rate bi(T ) = − (∂T /∂t)i; in the energy range of interest for π0 production (> 279 MeV), this can be considered unimportant to a good approximation. The loss of CRs that escape from the "leaky box" is included in the last term, with τesc being the mean time spent by the CRs in the containment volume. Although strictly speaking inelastic collisional losses should have also been included in the propagation equation, we have omitted them since the mean free path against collisions is much larger than the one against escape. If in addition we assume a steady state, the left hand side of eq. (5) vanishes and the propa- gation equation for protons assumes the simple form 0 = Qp(T ) − 1 τesc Np(T ) . (6) Physically, this corresponds to an equilibrium between sources (SN acceleration) and sinks (escape). Since the CR proton flux and the corresponding number density are related via φp(T ) = vpNp(T ), we find that the (propagated) flux is given by φp(T ) = ℓescQp(T ), (7) where ℓesc = τescv is the mean free path against escape.3 Thus, to make further progress in estimating the CR flux φp(T ) in the galaxy G, we need to have some understanding of the CR confinement in that galaxy, which enters in eq. (7) through ℓesc. This depends on the details of the magnetic field strength and configuration in these galaxies. The detailed physical origin of the confinement scale ℓesc (or equivalently, the diffusion tensor) is as yet uncertain, but is almost certainly related to the structure of and fluctuations in the galaxy's magnetic field (e.g., Berezinskiı et al. citebbdgp). Lacking any better knowledge, we will assume confinement conditions similar to those in the Milky Way. Specifically, we will assume ℓesc is the 3It is often conventional to define qp = Qp/ρISM and thus equation (7) implies φp(T ) = Λqp(T ) Λ = vpτesc ρISM is the escape pathlength in g cm−2, the "grammage." -- 7 -- same as in the Milky Way. This amounts to an Ansatz that the physical properties that determine ℓesc are dominated by local rather than global properties of the host galaxy. This assumption becomes more plausible the more similar G is to the MW, so we expect our approach to yield better results in the cases of M31 and M33 rather than in the cases of the Magellanic Clouds and other irregular galaxies. (Alternatively, one could turn the problem around, and with γ-ray observations of these objects, one can measure or limit the cosmic ray confinement in these objects.) Under this assumption, the CR flux is proportional to the SN rate in G: So from equations (1) and (8) we get φG p φM W p = RG RMW = fG ΓG π0 = fGΓM W π0 which, following the same procedure that has lead to eq. (4), finally gives γ (> 100MeV) = 2.36 × 10−25fG photons (s - H atom)−1 . qG (8) (9) (10) Assuming that the CR flux level and spectrum remains the same over the whole galaxy, this emissivity is space-independent. Thus, the γ−ray flux from galaxy G which lies at a distance d and has a gas content of Mgas will simply be or, using eq. (10) for qG γ , F G γ = 1 4πd2 Mgas mp qG γ , F G γ (> 100MeV) = 2.34 × 10−8fG(cid:18) Mgas 108M⊙(cid:19)(cid:18) d 100 kpc(cid:19)−2 photons cm−2 s −1 . (11) The largest fraction of the gas present in galaxies is in the form of neutral hydrogen which is detected via 21cm H I observations. The integrated H I mass in a galaxy is related to the integrated H I flux R Svdv by the scaling MH I ∝ d2. This leads to the fortunate circumstance that, although the calculated gas mass for a given galaxy depends on the assumed distance to that galaxy, the ratio Σ = Mgas d2 is associated only with quantities which are directly observable and is independent of any assump- tion on the distance. Thus, eq. (11) can finally be re-written to express the γ-ray flux of photons > 100 MeV from galaxy G F G γ = 2.34 × 10−8fG(cid:18) Σ 104M⊙kpc−2(cid:19) photons cm−2 s −1 . (12) in terms of the ratio fG of the supernova rate in G to that of the Milky Way, and the gas mass-to- distance squared ratio Σ. -- 8 -- Table 1: Observed Properties of Selected Local Group Galaxies Galaxy LMC SMC M31 M33 NGC6822 IC10 SN rate (century−1) 0.1(2), 0.23 (3),0.49(4) 0.065(3), 0.12(4) 0.9 (9), 1.21(4), 1.25 (7) 0.28(8), 0.35(9), 0.68(4) 0.04(10) 0.082-0.11(14) Adopted Σ (104 M⊙ kpc −2) f 0.14 0.04 0.45 0.17 0.02 0.04 H I 22 ± 6(1),(11),(12),(13) 17 ± 4(1),(5) 0.9 ± 0.2(1),(6) 0.26 ± 0.05(1) 0.05 ± 0.02(1) 0.016 ± 0.003(15) H2 4.63(13) 0.76(13) 0.06(16) 0.004(17) 0.006(18) & 10−5 (19) Total 26.6 17.8 0.92 0.264 0.056 0.016 References: (1) Huchtmeier & Richter (1989), references therein; (2) Chu & Kennicutt (1988); (3) Kennicutt & Hodge(1986); (4) Tammann et al. (1994); (5) Stanimirovi´c et al. (1999); (6) Braun & Walterbos (1992); (7) Braun & Walterbos (1993); (8) Gordon et al. (1998); (9) Berkhuijsen (1984); (10) Timmes & Woosley (1997); (11) Luks & Rohlfs (1992); (12) Kim et al. (1998); (13) Westerlund (1997); (14) Thronson et al. (1990); (15) Shostak & Skillman (1989), also references therein (16) Dame et al. (1993); (17) Wilson & Scoville (1989); (18) Israel (1997); (19) Wilson & Reid (1991) 3. Data Equation (12) shows that the information we need to calculate the expected γ-ray flux from each galaxy is the SN rate of the galaxy, and the ratio Σ. A summary of these data and the relevant references are presented in Table 1. The indicated ranges are the span in the published observational data and are not representative of the error of each measurement as estimated by the corresponding authors. This should give a sense of the systematic uncertainties, but one should bear in mind that the overall error could be larger, particularly for the supernova rates. The value quoted in the Σ column is the mean value of all the measurements found in the indicated references while the error is just the square root of the sample variance. In order to calculate fG we also need the SN rate of the MW. Different authors have pro- duced results which cover a range of roughly an order of magnitude, depending on the method of calculation. The three main methods used are extragalactic SN discoveries, SN-related Galactic data relating to massive star formation, chemical evolution, and nuclear γ-ray lines, and analysis of the historical record of Galactic SN explosions. Dragicevich et al. (1999) critically surveyed RMW determinations by different methods. The different results quoted in that work are given in Figure 1. As noted by Dragicevich et al. , the Galactic supernova rates estimated using data from historical supernovae tend to be higher than those based on extragalactic or nuclear/γ-ray line constraints. This discrepancy might arise if our location in the Galaxy is "special" -- e.g., near a spiral arm where star formation is enhanced. In addition, Hatano, Fisher, & Branch (1997) suggest that a large fraction (∼ 50%) supernovae are "ultradim" (MV ∼ −13), possibly due to localized -- 9 -- shrouding effects. Such a population would be missed in extragalactic surveys, but would appear in the historical record, and would still contribute to nucleosynthesis and accelerate cosmic rays. In our study, we will use extragalactic supernova data for Local Group objects, and thus the lower estimates of the Galactic supernova rate are the appropriate ones to use. As a "best bet" we will adopt Dragicevich et al. 's (1999) recommended value of 2.5 SN per century. 4. Results We now combine eq. (12) with the data presented in Table 1 to predict γ−ray flux levels for photons with energies above 100 MeV originating in the interaction between cosmic rays and interstellar medium in galaxies of the Local Group. We will use the Magellanic Clouds to verify the applicability of our method, since for these systems we can compare our predictions with both EGRET observational results (detection for the LMC, upper limit for the SMC) and other predictions using models based on entirely different assumptions: dynamic equilibrium for the LMC (Fichtel et al. 1991), constant cosmic ray proton-to-electron ratio for the SMC (Sreekumar & Fichtel 1991). We will then use our model to proceed to predictions of γ−ray fluxes in galaxies of the Local Group for which there have been either no previous studies or, in the case of M31, only partial treatments. For the latter, Ozel & Berkhuijsen (1987) based on the gas content alone arrived at a result in which the cosmic ray flux scaling between the Milky Way and M31 was treated as a free multiplicative parameter, left to be decided observationally. Our predictions, and their implications for GLAST, are summarized in Table 2, with detailed discussion of each galaxy appearing below. In Table 2, all values refer to γ-rays > 100 MeV. The "GLAST Significance" column refers to the formal significance expected to be achieved after a 2-year (nominal GLAST duty cycle) and 10-year (GLAST lifetime goal) all-sky survey. The "On-Target 5 σ Exposure Time" column refers to the total exposure of the object needed to achieve a 5 σ detection. When GLAST is operating in the normal sky-scanning mode, each individual source is in the field of view for only ∼ 20% of the time for each duty cycle, so the GLAST operation time required to achieve a detection of the same significance is typically 5-6 times the on-target exposure time quoted (assuming a field of view for GLAST between 2 and 2.4 sr). All significances and exposure times were calculated according to the GLAST specifications as described in De Angelis (2000) and in the GLAST Science Requirements Document (http://glast.gsfc.nasa.gov/science/aosrd). We have used an effective collector area of 8000 cm2 (GLAST requirement, 0.8× GLAST goal), and taken into account the (Galactic and extragalactic) γ−ray background noise levels as measured by EGRET for the Galactic coordinates of each object. Whenever the angular extent of an object in the sky (as derived from 21 cm maps) exceeded the 1-photon angular resolution of GLAST, the actual size of the object was used as the relevant solid angle for the collection of background noise. The same calculation, when applied to a point source of flux 2 × 10−9photons cm−2 s−1 located at high Galactic latitude, predicted a 5-σ detection after a 2-year all-sky survey, in accordance to the -- 10 -- Table 2: Predicted Gamma-Ray Flux and GLAST Requirements for Selected Local Group Galaxies Galaxy LMC SMC M31 M33 NGC6822 IC10 EGRET Value/Limit (14.4 ± 4.7) × 10−8 Flux > 100 MeV (photons cm−2 s−1) GLAST Significance GLAST On-Target Prediction 5σ Exposure Time 11 × 10−8 1.7 × 10−8 1.0 × 10−8 0.11 × 10−8 2.6 × 10−11 2.1 × 10−11 4.6 × 10−3 yr 2.1 × 10−2 yr 4.1 × 10−2 yr 93 σ 43 σ 31 σ 4.1 σ 0.09 σ 0.05 σ 2.31 yr ≫ 10 yr ≫ 10 yr 2 years 10 years < 4 × 10−8 < 1.6 × 10−8 N/A N/A N/A 42 σ 19 σ 13 σ 1.9 σ 0.04 σ 0.02 σ sensitivity derived in the GLAST Science Requirements Document. 4.1. Magellanic Clouds 4.1.1. Large Magellanic Cloud The LMC is the only galaxy other than the Milky Way for which there has been a positive detection of its diffuse γ−ray emission, and is therefore the only one of the systems of interest for which any prediction can be directly tested against observations. For the LMC, Table 1 suggests a mean Σ equal to 26.6 × 104M⊙ kpc−2 and a mean SN rate of 0.27 century−1 which, combined with a Galactic SN rate of 2.5 century−1, gives fLMC = 0.11. Inserting these data in eq. (12) we derive a γ-ray flux for photons with energies > 100 MeV of 6.8 × 10−8 photons cm−2 s−1. However, of the 3 references quoted in Table 1 for the SN rate of the LMC, the lowest one (2) in the table, equal to 0.1 SN century−1), which is based on a count of observed SN (ref. remnants, is derived only as a lower limit to the LMC SN rate. The other two estimates are based on extragalactic SN discoveries in morphologically similar galaxies (ref. 4 in Table 1) and on the massive star formation rate (ref. 3 in Table 1). If we use the mean value of the latter two as our best estimate for RLMC we get fLMC = 0.14. As far as the gas mass is concerned, although the more recent 21cm surveys tend to give rather low values for Σ (Luks & Rohlfs 1992, Kim et al. 1998), the gas mass estimates in those cases are assuming an optically thin medium. However, recent studies of the cool gas in the LMC by Marx-Zimmer et al. (2000) are not in favor of this assumption, which indicates that the gas masses might in fact be significantly underestimated. On this basis, we will adopt for our calculation the higher estimate from Westerlund (1997) which predicts a ΣH I = 28 × 104M⊙ kpc−2, and thus Σtot = 32.6 × 104M⊙ kpc−2. -- 11 -- These values of f and Σ, if used in eq. (12), yield a total γ-ray flux of F LMC γ = 11 × 10−8 photons cm−2 s −1 . (13) This value is in excellent agreement with the observed value of (14.4±4.7)×10−8 photons cm−2 s −1 (Hartman et al. 1999). This consistency gives us confidence in our method of computing galactic cosmic ray fluxes. Indeed, one could even turn the argument around, and tentatively interpret this agreement as an indication that the cosmic ray confinement in the LMC is comparable to that of the Milky Way, Λesc ∼ 10 g cm−2. Given the large differences in the size and structure of these two galaxies, this result would suggest that cosmic ray confinement is not strongly dependent on a galaxy's global properties, but more closely connected to local physics, such as the magnetic field strength configuration and fluctuations. If confirmed, therefore, this result would give new and unique information about the nature of cosmic ray confinement and propagation. As noted below (§4.2), this result can be tested by combining the flux measurements for multiple Local Group sources, particularly by comparing the LMC to the Small Magellanic Cloud. Of course, the uncertainties in the observational data, especially the Galactic SN rate, can have an important impact on our calculation. For example, if we adopt a Galactic SN rate of 1 century−1 and keeping the same values for all other data, the LMC flux could be as large as 27 ×10−8photons cm−2 s−1 On the other hand, if the Galactic SN rate is as high as 5 century−1, this estimate would drop to 5.3 ×10−8 photons cm−2 s−2. With these caveats in mind, our best estimate for the flux gives a very strong detection (for- mally, at the 42σ level) in the first 2 years of sky-scanning GLAST operation, with the 5-σ detection feasible after an on-target observation time of less than 2 days. 4.1.2. Small Magellanic Cloud The γ−ray flux level of the Small Magellanic Cloud has been the object of both theoretical and observational studies in the past, which have been used to resolve the debate of the origin of the cosmic rays: if the origin of the cosmic rays were cosmological and the level of the CR flux were universal the γ−ray flux should have been easily detectable by EGRET (Sreekumar & Fichtel, 1991). However, the SMC was not detected by EGRET, and an observational upper limit was placed instead on its γ−ray flux. (Sreekumar et al. 1993, Lin et al. 1996). This observational upper limit was also lower than the value Sreekumar & Fichtel (1991) predicted if the cosmic rays, gas and magnetic fields in the SMC were in a state of dynamic equilibrium. Thus, Sreekumar et al. (1993) attributed the non-detection of the SMC by EGRET to poor confinement of cosmic rays in the SMC. In the framework of our model, we will investigate whether CR flux levels of the SMC low -- 12 -- enough so as to prohibit its detection by EGRET can also be explained in terms of a low supernova rate and consequently a low cosmic ray acceleration rate, even if we assume similar confinement conditions with those of the Milky Way. Although the differences between the morphologies of the MW and the Magellanic Clouds might raise questions concerning the validity of the latter assumption, the good agreement of our results with observations in the case of the LMC encourage the application of our method in the case of the SMC as well. For the SMC, the mean value of Σ is equal to 17.8 × 104M⊙ kpc−2 and the average of the quoted SN rates is 0.09 SN per century (Table 1). Thus, eq. 12 (using again a RMW equal to 2.5 century−1) predicts a γ-ray flux of F SMC γ = 1.7 × 10−8 photons cm−2 s−1 . (14) This value is consistent with the current observational upper limit of 4 × 10−8 photons cm−2 s−1 of Lin et al. (1996). Using this "best bet" value for the SMC γ−ray flux, we find that GLAST will detect the SMC with a 19 σ significance after a 2-year all-sky survey. The total exposure time needed to achieve a 5 σ detection is only 8 days. We thus see that the observed SN rate for the SMC is, by itself, sufficient to explain the observational upper limit placed by EGRET, even under the assumption of confinement conditions that do not differ from those in the Milky Way. Whether confinement in the SMC plays an additional role in lowering the global cosmic ray and subsequently γ−ray flux is a very interesting question, left to be answered observationally by GLAST. If now we use the upper limit values for the ΣSMC and fSMC of 21.8 × 104M⊙ kpc−2 and 0.05 correspondingly, the resulting γ−ray flux reaches 2.6 × 10−8 photons cm−2 s−1, still below the current observational upper limit. In the other end of the range, using the lowest estimates for the Σ and R of the SMC, the result we get is equal to 0.85 × 10−8 photons cm−2 s−1 , well above the anticipated sensitivity of GLAST. In fact, the SMC γ−ray flux would be still detectable by GLAST even if, in addition to using the lowest available values for Σ and RSMC, we adopted a Galactic SN rate higher than our "best bet" by a factor of 3. Although our assumption for a "leaky box" CR confinement is more questionable in the case of the Magellanic Clouds, our prediction is consistent the upper limit set by EGRET. Furthermore, it coincides with the detailed calculations of Sreekumar & Fichtel (1991). These authors used synchrotron radiation measurements to deduce the cosmic ray intensity and distribution in the SMC (assuming a CR proton-to-electron ratio same as that of the MW) and predicted a total flux of γ-rays above 100 MeV equal to 1.7 × 10−8 photons cm−2 s−1 . The perfect agreement between this prediction and ours is fortuitous, but nevertheless increases our confidence in our basic model. In the cases of galaxies such as M31 and M33, we expect the CR confinement conditions to be much more similar to those in the MW (observations indicating the opposite would not only be a surprising but a very interesting result in itself). Thus, we expect our predictions for these galaxies to be, within our uncertainty limitations, even more reliable. -- 13 -- 4.2. M31 The mean observed value of Σ for M31 is 0.92 × 104M⊙ kpc−2 while the average of RM31 as measured with all 3 different methods (observations of SN remnants, star formation rates and mor- phology arguments) is 1.12 per century (Table 1). The average RM31 corresponds to an M31/MW supernova rate ratio fM31 = 0.45. It is worth pointing out that M31 has an unusually low Hα and far infrared emission (e.g., Pagini et al. citepagnini), which imply a low star formation rate, and hence a low Type II supernova rate for such a large galaxy. As we will see, GLAST should detect M31, an thus provide an important new measure of the M31 supernova rate. Using our adopted gas content and supernova rate for M31, eq. (12) then predicts a total γ-ray flux for energies above 100 MeV F M31 γ = 1.0 × 10−8 photons cm−2 s −1 . (15) This value is consistent with the observational upper limit of 1.6 × 10−8 photons cm−2 s−1 set by Blom et al. (1999), but only slightly lower than the EGRET sensitivity. As we see in Table 2, our predicted γ−ray flux for M31 would be detected by GLAST in its first 2-year all-sky survey with a 14 σ significance. After a projected lifetime of 10 years, and assuming continuous sky-scanning operation, this significance would rise to 31 σ. Our calculation gives theoretical support to the case made by Digel et al. (2000), who noted that if the flux from Andromeda lies just below the Blom et al. limit, then GLAST will readily be able to detect it. Here we find that indeed, the flux should be just at the level of 1×10−8 photons cm−2 s −1 suggested by Digel et al., though the uncertainty range is considerable, as we now see. To estimate the effect of the various uncertainties in our calculation, we observe that if we had used the highest available observed values for Σ and RM31 as given in Table 1 but kept the Galactic SN rate fixed at 2.5 per century, the resulting flux of 1.3 × 10−8 photons cm−2 s−1 would still be lower than the observational upper limit. With Σ and RM31 values both at the upper end of their respective ranges, F M31 would remain below the EGRET limit for RMW as low as 2 per century. γ At the other end of the range of the available M31 data, using a Σ as low as 0.7× 104M⊙ kpc−2 and RM31 of 0.9 per century, we derive a predicted flux of 0.6 × 10−8 photons cm−2 s−1, still easily above the expected GLAST sensitivity. If in addition we take into account the uncertainty in the RMW , we can see that even with the most pessimistic estimates for the M31 gas content and SN rate, the γ-ray flux remains above the GLAST detectability limit for a Galactic SN rate up to 7 per century. In sum, our analysis leads to the robust prediction that a γ-ray observatory with sensitivity similar to that expected for GLAST will detect the diffuse γ−ray signature of M31. Once there is a positive detection, depending on the strength of the signal and the available spatial resolution, this opens several avenues of further analysis: -- 14 -- • The spatial distribution of neutral hydrogen in M31 has been observed to exhibit an asym- metry, with the west part of the galaxy having a significantly lower column density than the northeast part. For the relevant γ-ray fluxes of the two areas (a few times 10−9 up to 10−8 photons cm−2 s−1), the angular resolution of GLAST will be 10′ − 20′ (Digel et al. 2000). Given the ∼ 2◦ size of M31 in 21 cm (which is the relevant angular size for γ−ray emission), this asymmetry is easily within the angular resolution capabilities of GLAST and is thus expected to be observed in the γ−ray signal as well. If the M31 flux is high enough, one might even hope to observe effects of the magnetic torus (e.g., Beck, Brandenburg, Moss, Shukurov, & Sokoloff (1996)) and star forming ring (e.g., Pagani et al. (1999)) at radius 10 kpc. A similar morphological feature in the Milky Way, the H2 ring extending in radius from 4 to 8 kpc (e.g., Bronfman et al. 1988) was first detected in early γ−ray surveys and then shown to be consistent with subsequent CO molecular surveys (Stecker, Solomon, Scoville, & Ryter 1975). • An observational measurement of the diffuse γ−ray flux from M31 could be used to reverse the arguments presented in section 2 and make inferences regarding the cosmic ray flux in M31: From the γ−ray flux as measured from Earth, F M31 , and using the distance and gas content of M31, we can calculate the γ-ray production rate per target H-atom. This result, combined with an assumption for the cosmic ray energy spectrum, would provide information on the cosmic ray flux level in M31 which could then be compared with the MW cosmic ray flux to determine whether they are comparable and whether the MW has a typical cosmic ray activity for galaxies with similar morphology. (Note here that we assume that the contribution from point sources in M31 would be negligible.) γ • A measured value of the γ−ray flux from M31 could also be used to determine the ratio of the supernova rates between M31 and the Milky Way, assuming that the gas content of M31 is known to a good accuracy: fM31 = RM31 RMW = 0.45 F M31 10−8 photons cm−2 s−1! γ (16) using eq. 12. This ratio could then be used e.g. to determine the MW supernova rate or to to infer the ratio of formation rates for high-mass stars, and, assuming a constant initial mass function, the ratio of total star formation rates between the two galaxies. (Here we rely on our assumption that all cosmic rays are produced by supernovae.) These last two points can be self-consistently checked once GLAST has information on more than one extragalactic source. For example, the ratios of the flux measurements among the ex- tragalactic sources is a direct measure of the ratios of the cosmic ray densities. In our model, these ratios gives the ratio of the product of Rℓesc. Thus, with information about the supernova rates, one can directly measure the ratios of cosmic ray confinements; this allows one to compare these values between the LMC and SMC, and to contrast the clouds with M31 (and M33). One -- 15 -- might then assume, in the case of the clouds, that their confinements are equal, and compare the inferred supernova rates with values estimated by other means. Alternatively, one could assume (as we have) that the confinements are these same as in the MW, and then use the ensemble of measurements of Local Group sources to reduce the uncertainty in the inferred Milky Way SN rate. Another potential cross-check would come from the comparison of Local Group > 100 MeV emission and γ-ray line emission from radioactive decays (Timmes & Woosley (1997)). The line flux is sensitive to the supernova rate only, so that the ratio of the > 100 MeV continuum to the lines gives direct information about the cosmic ray confinement. Unfortunately, such a comparison may have to wait, as the Timmes & Woosley calculations predict that the line fluxes lie below the expected sensitivity of the forthcoming International Gamma-Ray Astrophysics Laboratory (INTEGRAL) mission. Even a non-detection of M31 in high energy γ-rays would be, apart from very surprising, a result of high theoretical interest, since the only parameter entering our calculations other that f and Σ is ℓesc (the escape mean free path) -- which in turn depends on the confinement conditions of the cosmic ray nuclei in M31. A non-detection would thus lead us to reconsider our assumption of similar confinement conditions with M31. Such a result would provide observational clues for the confinement in an extragalactic environment (morphologically similar to the Milky Way) and for the nucleonic component of cosmic rays. 4.3. Other Local Group Galaxies Using the mean values for the gas content-related Σ and the supernova rate shown in Table 1 for the case of M33, eq. (12) gives a γ−ray flux of F M33 γ = 0.11 × 10−8 photons cm−2 s−1 , (17) which is slightly below the sensitivity limit of GLAST. As seen in Table 2, after a 10 year sky survey, the detection is at the 4.2σ level (comparable to the current LMC significance). If we use the highest available estimates for the M33 gas content and supernova rate, the flux would rise to 0.2 × 10−8 photons cm−2 s−1, which would allow for a 5σ detection after a sky survey of 4.39 years. Alternatively, even if eq. (17) is accurate, M33 will be detectable at the 5σ level within 10 years if the GLAST effective area and field of view achieve their "goal" levels (as opposed to the "required" levels we have used). Thus, there is grounds for optimism that a detection might be possible, if either the gas content and supernova rate have been slightly underestimated, or the actual sensitivity of the GLAST observatory is slightly improved as compared to the current prediction. We therefore urge observers to be aware of the possibility of detecting (or placing a limit on) flux from M33. The next best candidates for detection (highest combination of Σ and f ) in the local group are NGC6822 and IC10. However, using the data shown in Table 1 for these galaxies, eq. (12) -- 16 -- predicts fluxes which are comparable within our uncertainty limits and equal to about 0.002 × 10−8 photons cm−2 s−1. Such tiny fluxes would require exposure GLAST times of decades, and thus appear to lie beyond reach for the foreseeable future. Thus, apart from the Magellanic Clouds, M31 and M33, no other Local Group galaxy seems to be a good candidate for the detection of its diffuse γ-ray flux by γ-ray observatories in the near future. There might, however, be other promising candidates, which lie outside the Local Group. Starburst galaxies, despite the fact that they lie at a greater distance that Local Group galaxies, have a significantly higher supernova rate, as well as high gas contents, which should result to high γ-ray production rate and a flux that might be detectable by highly sensitive observatories (Paglione et al. 1996). 5. Discussion Diffuse, high-energy (& 100 MeV) γ-rays provide the most direct evidence for the extension of the cosmic ray ion component throughout our Galaxy. The observation of such radiation from extragalactic systems would provide unique information, as the mere detection of high-energy γ-rays confirms the presence of cosmic ray ions, and the photon flux can be used to infer the cosmic ray flux, and thus can constrain extragalactic cosmic ray properties. Unfortunately, the only extragalactic object detected thus far is the LMC (Sreekumar et al (1992)). In anticipation of future high-energy γ-ray observatories such as GLAST, we have estimated the γ-ray flux due to diffuse emission for Local Group galaxies. To do this, we have used a simple "leaky box" model of cosmic ray propagation, and taken supernova blasts to be the engines of cosmic ray acceleration. Our model makes different assumptions than Fichtel et al. 's (1991) more detailed treatment of the LMC, but both give similar results, and are in reasonable agreement with LMC γ-ray observations. Applying our model to other Local Group galaxies, we predict that M31 has a γ-ray flux above 100 MeV of about 1.0 × 10−8 photons cm−2 s−1, with an uncertainty of about a factor of 3. Fortunately, despite this large error budget, we can conclude that M31 should be observable by GLAST, and we therefore strongly urge that M31 be looked for in GLAST maps. A detection will provide important and unique information about cosmic rays in a galaxy similar to our own. In addition, we find that the SMC should have a flux of about 1.4 × 10−8 photons cm−2 s−1, readily detectable by GLAST. The comparison among the LMC, SMC, and M31 γ-ray luminosities will provide new information about cosmic ray densities and confinement, and supernova rates, in these systems and in the Milky Way. The high-energy γ-ray flux from other Local Group galaxies is much smaller. Other than M31 and the Magellanic clouds, the only system that is potentially observable is M33, with a flux of about 0.1×10−8 photons cm−2 s−1. If GLAST can stretch to reach its sensitivity goals, this too will be observable. All other Local Group galaxies have emission that is at least 2 orders of magnitude smaller. -- 17 -- We thank David Branch, Jim Buckley, You-Hua Chu, Robert Gruendl and Kostas Tassis for enlightening discussions. We thank the referee for constructive comments which have improved this paper. The work of V.P. was partially supported by a scholarship from the Greek State Scholarship Foundation. Akyuz, A., Brouillet, N. & Ozel, M.E. 1991, A&A, 248, 419 REFERENCES Beck, R., Brandenburg, A., Moss, D. Sukurov, A., & Sokoloff, D. 1996, ARAA, 34, 155 Berezinskiı, V.S., Bulanov, S.V., Dogiel, V.A., Ginzburg, V.L., & Ptuskin, V.S. 1990, Astrophysics of Cosmic Rays, Amsterdam: North-Holland Berkhuijsen, E.M. 1984, A &A, 140, 431 Blom, J. J., Paglione, T. A. D. and Carraminana, A. 1999, ApJ, 516, 744 Braun, R. & Walterbos, R.A.M. 1992, ApJ, 386, 120 Braun, R. & Walterbos, R.A.M. 1993, A&A, 98, 327 Bronfman, L., Cohen, R. S., Alvarez, H., May, J., & Thaddeus, P. 1988, ApJ, 324, 248 Cavallo, G. & Gould, R.J. 1971, Nuovo Cimento, 2B, 77 Chu, Y.-H. & Kennicutt, R.C. Jr. 1988, ApJ, 96, 6 Combi, J.A., Romero, G.E. & Benaglia, P. 1998, A&A, 333, L91 Cram, T.R., Roberts, M.S. & Whitehurst, R.N. 1980, A&AS, 40, 215 Dame, T. M., Koper, E., Israel, F. P. & Thaddeus, P. 1993, ApJ 418, 730 Dermer, C. D. 1986, A&A, 157, 223 De Angelis, A. 2000, to be published in the Proceedings of the 3rd International Workshop "New Worlds in Astroparticle Physics," astro-ph/0009271 Digel, S.W., Hunter, S.D. & Mukherjee, R. 1995, ApJ, 441, 270 Digel., S.W., Moskalenko, I.V., Ormes, J.F., Sreekumar, P., & Williamson, P.R. 2000, AIP Conf.Proc., 528, 449 Dragicevich, P.M., Blair, D.G. & Burman, R.R. 1999, MNRAS, 302, 693 -- 18 -- Ellison, D.C., Drury, L.O., & Meyer, J.P. 1997, ApJ, 487, 197 Fichtel, C.E. & Kniffen, D.A., A&A, 134, 13 Fichtel, C.E., Ozel, M.E., Stone, R.G. & Sreekumar, P. 1991, ApJ, 374, 134 Gordon, S.M., Kirshner, R.P., Long, K. S., Blair, W. P., Duric, N. & Smith, R.C. 1998, ApJS, 117, 89 Hartman R.C. et al. 1999, ApJS, 123, 79 Hatano, K., Fisher, A., & Branch, D. 1997, MNRAS, 290, 360 Huchtmeier, W. K. 1973, A&A, 22, 91 Huchtmeier, W. K., & Richter, O. 1989, A general catalog of H I observations of galaxies: the reference catalog, (Springer-Verlag: New York) Huchtmeier, W. K, Karachentsev, I.D., Karachentseva, V.E. & Ehle, M. 2000, A&A 141, 469 Hunter, S.D. et al 1997, ApJ 481, 205 Israel, F. P., 1997, A&A 317, 65 Kennicutt, R. C., Jr. & Hodge, P. W. 1986, ApJ, 306,130 Kim, S., Staveley-Smith, L., Dopita, M.A., Freeman, K.C., Sault, R.J., Kesteven, M.J. & Mc- Connell, D. 1998, ApJ, 503, 674 Koyama, K. et al. 1995, Nature, 378, 255 Lin, Y.C. et al. 1996, ApJS, 105, 331 Long, K.S., Blair, W.P., Kirshner, R.P. & Winkler, P.F. 1990, ApJS, 72, 61 Luks, Th. & Rohlfs, K. 1992, A & A, 263, 41 Marx-Zimmer, M, Herbstmeier, U. Dickey, J.M., Zimmer, E., Staveley-Smith, L. & Mebold, U. 2000, A & A, 354, 787 Mori, M. 1997, ApJ, 478, 225 Ozel, M.E. & Berkhuijsen, E.M. 1987, A&A, 172, 378 Ozel, M.E. & Fichtel, C.E. 1988, ApJ 335, 135 Pagani, L., et al. 1999, A&A, 351, 447 Paglione, T.A.D., Marscher, A.P., Jackson, J.M. & Bertsch, D.L. 1996, ApJ, 460, 295 -- 19 -- Shostak, G.S. & Skillman, E.D. 1989, A & A, 214, 33 Sreekumar, P. & Fichtel, C.E. 1991, A&A, 251, 447 Sreekumar, P. et al 1992, ApJ, 400, L67 Sreekumar, P. et al 1993, Phys.Rev.L., 70, 127 Sreekumar, P. et al 1998, ApJ, 494, 523 Stanimirovi´c, S.,Staveley-Smith, L., Dickey, J.M., Sault, R.J. & Snowden, S.L. 1999, MNRAS, 302, 417 Stecker, F.W. 1970, Ap. and Space Sci., 6, 377 Stecker, F.W. 1973, ApJ, 185, 499 Stecker, F.W., Solomon, P.M., Scoville, N.Z. & Ryter, C.E. 1975, ApJ, 201, 90 Stecker, F.W. 1988, in Cosmic Gamma Rays, Neutrinos and Related Astrophysics, ed. M.M. Shapiro & J.P. Wefel (Dordrecht:Reidel), 85 Stephens, S.A. & Badhwar, G.D. 1981, Ap&SS, 76, 213 Strong, A.W. 1996, SSRv, 76, 205 Strong, A.W., Moskalenko, I.V., & Reimer, O. 2000, ApJ, 537, 763 Tammann, G.A., Loffler, W. & Schroder, A. 1994, ApJS, 92, 487 Timmes, F. X. & Woosley, S. 1997, ApJ, 489, 160 Thronson, H. A. Jr., Hunter, D. A., Casey, S. & Harper, D. A. 1990. ApJ, 355, 94 Westerlund, B.E. 1997, The Magellanic Clouds, (Cambridge: Cambridge Univ. Press), 28 Wilson, C.D. & Scoville, N. 1989, ApJ 347, 743 Wilson, C.D. & Wilson, Reid, I. N. 1991, ApJL 366, 11 This preprint was prepared with the AAS LATEX macros v5.0. -- 20 -- Fig. 1. -- Milky Way supernova rates as calculated by different authors, from the tabulation of Dragicevich et al. (1999). The estimates from extragalactic SN discoveries have been standardized to h = 0.75 and a Milky Way blue luminosity of (2.3 ± 0.6) × 1010L⊙,B.
0809.4945
1
0809
2008-09-29T11:43:48
Near-Infrared Polarimetry toward the Galactic Center
[ "astro-ph" ]
Near-infrared polarimetry of point sources reveals the presence of a toroidal magnetic field in the central 20' x 20' region of our Galaxy. Comparing the Stokes parameters between high extinction stars and relatively low extinction ones, we have obtained a polarization originating from magnetically aligned dust grains at the central region of our Galaxy of at most 1-2 kpc. The derived direction of the magnetic field is in good agreement with that obtained from far-infrared/submillimeter observations, which detect polarized thermal emission from dust in the molecular clouds at the Galactic center. Our results show that by subtracting foreground components, near-infrared polarimetry allows investigation of the magnetic field structure at the Galactic center. The distribution of the position angles shows a peak at around 20deg, nearly parallel to the direction of the Galactic plane, suggesting a toroidal magnetic configuration.
astro-ph
astro-ph
Astronomical Polarimetry 2008: Science from Small to Large Telescopes ASP Conference Series, Vol. 4**, 2009 Bastien and Manset Near-Infrared Polarimetry toward the Galactic Center Shogo Nishiyama1, Motohide Tamura2, Hirofumi Hatano3, Saori Kanai3, Mikio Kurita3, Shuji Sato3, Tetsuya Nagata1 Near-infrared polarimetry of point sources reveals the presence of Abstract. a toroidal magnetic field in the central 20' x 20' region of our Galaxy. Comparing the Stokes parameters between high extinction stars and relatively low extinc- tion ones, we have obtained a polarization originating from magnetically aligned dust grains at the central region of our Galaxy of at most 1-2 kpc. The derived direction of the magnetic field is in good agreement with that obtained from far- infrared/submillimeter observations, which detect polarized thermal emission from dust in the molecular clouds at the Galactic center. Our results show that by subtracting foreground components, near-infrared polarimetry allows investi- gation of the magnetic field structure at the Galactic center. The distribution of the position angles shows a peak at around 20◦, nearly parallel to the direction of the Galactic plane, suggesting a toroidal magnetic configuration. 1. Introduction The magnetic field configuration at the Galactic center (GC) has been inves- tigated with a wide variety of methods. Recent far-infrared (FIR) and sub- millimeter (sub-mm) polarimetric observations point out that the magnetic field is generally parallel to the Galactic plane (Novak et al. 2000, 2003; Chuss et al. 2003). This is contrast to the poloidal field traced by non-thermal radio fila- ments, most of which align nearly perpendicular to the Galactic plane. Previous near-infrared (NIR) polarization measurements of the GC were discussed in terms of selective absorption by the intervening interstellar dust grains in the Galactic disk. To our knowledge, no one has studied the magnetic field configuration at the GC with NIR polarimetry. In this paper, we present results of NIR polarimetric observations toward the GC. We demonstrate that NIR polarization of point sources can provide information on the magnetic field structure at the central region of our Galaxy. 2. Observations and Data Analysis We observed a 20' × 20' area centered at the position of Sgr A* with the NIR polarimetric camera SIRPOL on the IRSF telescope. SIRPOL consists of a single-beam polarimeter and NIR imaging camera SIRIUS. SIRIUS provides 1Department of Astronomy, Kyoto University, Kyoto 606-8502, Japan 2National Astronomical Observatory of Japan, Mitaka, Tokyo 181-8588, Japan 3Department of Astrophysics, Nagoya University, Nagoya 464-8602, Japan 1 2 images of a 7.′7 × 7.′7 area of sky in three NIR wavebands, J (1.25µm), H (1.63µm), and KS (2.14µm) simultaneously with a scale of 0.′′45 pixel−1. The Stokes parameters I, Q, and U for point sources were determined from aperture polarimetry of combined images. Based on intensities for each wave plate angle, we calculated the Stokes parameters I, Q, and U as I = (I0.◦ 0 + 5. The degree of I22.◦ 0, and U = I22.◦ polarization P and the position angle θ were derived by 5)/2, Q = I0.◦ 5 − I67.◦ 0 + I67.◦ 5 + I45.◦ 0 − I45.◦ P = q(Q2 + U 2)/I, θ = 1 2 arctan(U/Q). We show a KS-band vector map binned by 0.′5 × 0.′5 in Fig. 1. The vectors are superposed on the three color (J, H, KS) composite image of the same region. At a first glance, most of the vectors are in order, and are nearly parallel to the Galactic plane. Moving north-eastward across the image, the position angles slightly rotate clockwise. At a few positions where the number density of stars is small, and hence strong foreground extinction exists, the vectors have irregular directions particularly at the northwestern corner. These irregularity might be explained by the inherent magnetic field configuration in foreground dark clouds. 3. Magnetic Field Configuration at the Galactic Center Polarimetric measurements of stars of different distances reveal the three dimen- sional distribution of magnetic field orientations. From the stars at the close side in the Galactic bulge (referred to hereafter as "blue stars" due to their relatively small reddening), we can obtain P and θ, which are affected mainly by interstel- lar dust grains in the Galactic disk. The light from stars at the far side in the bulge (hereafter "red stars") is transmitted through the dust in the disk and the bulge. Therefore, using both blue and red stars, we can obtain the polarization originating in the bulge. The procedure is as follows. At first, we divided the field into 10 × 10 sub-fields, and made H − KS histograms for each sub-field. Using the histograms, we evaluated a peak value of the histogram (H − KS)peak. For H − KS color, we divided the stars into two groups: "blue" and "red" stars in the bulge. The "blue" stars are redder than H − KS = 1.0 and bluer than (H − KS)peak. The stars with H − KS > (H − KS)peak are selected as "red" stars. Next, Q/I and U/I histograms in the KS band were constructed for the blue and red stars in each sub-field. We calculated their means as < Q/I >B, < Q/I >R, < U/I >B, and < U/I >R. We then obtained P and θ for "red minus blue" components using the following equations (Goodrich 1986): PR−B = s(cid:18)(cid:28) Q θR−B = 1 2 I (cid:29)R −(cid:28) Q I (cid:29)B(cid:19)2 I (cid:29)R −(cid:28) U +(cid:18)(cid:28) U I (cid:29)B(cid:17)(cid:30)(cid:16)(cid:28) Q I (cid:29)B(cid:19)2 I (cid:29)R −(cid:28) U I (cid:29)R −(cid:28) Q I (cid:29)B(cid:17)(cid:21) . , arctan(cid:20)(cid:16)(cid:28) U The errors of < Q/I > and < U/I > were calculated from the standard error on the mean σ/√N of the Q/I and U/I histograms, where σ is the standard 3 Figure 1. KS-band polarization vector map superposed on the three color (J, H, KS) composite image of the Galactic center. The Galactic center is the bright yellow blob in the center. The mean PKS and θKS are calculated for each 0.′5 × 0.′5 grid. deviation and N is the number of stars (see Nishiyama et al. 2009, for more detail). We show a vector map for PR−B and θR−B in Fig. 2. The average of PR−B and θR−B are obtained as 0.85 % and 16.◦0, only for grids where the polarization is detected with PR−B/δPR−B ≥ 2. The histogram of θR−B has a peak at ∼ 20◦, which roughly coincides with the angle of the Galactic plane. A similar result is also obtained for the H-band polarization. This coincidence shows the basically toroidal geometry of the magnetic field. 4. Discussion The direction of the magnetic field at the GC has been investigated from po- larized dust emission in the FIR and sub-mm wavelengths. As seen in Fig. 2, 4 the magnetic field configuration we obtained at the GC shows a good agree- ment globally with those obtained by Dotson et al. (2000), Novak et al. (2000), and Chuss et al. (2003), which are the highest angular resolution polarimetry data sets in the FIR/sub-mm wavelengths. The polarized FIR/sub-mm emis- sion comes from molecular clouds, which are known to be located in the GC. Therefore we conclude that the position angles derived from our NIR polarimetry represent the direction of the magnetic field in the GC. i ] n m c r a [ δ ∆ 10 5 0 -5 -10 5.0 % 2.5 % l = 0o b = 0o 10 5 0 ∆α [arcmin] -5 -10 Figure 2. KS-band vector map derived from the Galactic center component (PR−B & θR−B, thick bars). The length of the bars is proportional to the measured degree of polarization, and their orientation is drawn parallel to the inferred magnetic field direction. The polarization map derived from FIR/sub- mm observations (thin bars) is also shown. The data sets of FIR/sub-mm wavelengths are from 60 µm & 100 µm polarimetry by Dotson et al. (2000), and 350 µm polarimetry by Novak et al. (2000) and Chuss et al. (2003). 5 From H −KS color differences between peaks in the H −KS histograms and mean colors of the blue and red stars, we tried to estimate the depth of the region where we have mapped the magnetic field configuration. Using AKS /EH−KS = 1.44 (Nishiyama et al. 2006) and the model of dust distribution by Davies et al. (1997), we obtained the average distances of 0.5 kpc from the GC for the blue stars, and 1.0 kpc for the red stars. This suggests that the polarization shown in Fig. 2 occurs between the average distances of (R0 − 0.5) kpc and (R0 + 1.0) kpc from the Sun, arising probably from the central 1−2 kpc region of our Galaxy (where R0 is the distance between the GC and the Sun). We have shown that the polarization of starlight can be a probe of the magnetic field near the GC. Morris (1998) enumerated five different ways in which the magnetic field near the GC has been studied: morphology, polarization angle, Faraday rotation of the radio continuum, Zeeman effect, and polarized dust emission in FIR/sub-mm wavelengths. The wide field-of-view of the NIR polarimeter SIRPOL, and the statistical treatment of tens of thousands of stars enable us to study the magnetic field near the GC; that is, the NIR polarimetry of starlight is a new way to investigate the magnetic field in the GC. NIR polarimetry has the advantage of providing information about the mag- netic field at locations where FIR/sub-mm emission is weak. NIR polarization of starlight is attributed to extinction along the line of sight by aligned dust grains, while FIR/sub-mm polarization is due to emission from the aligned dust. Hence, NIR polarimetry can investigate the magnetic field in regions where FIR/sub- mm polarimetry is absent, if background stars exist. This advantage is clearly shown in Fig. 2. We have detected polarization at positions where thin bars are not shown. The distribution of the position angles in most of the observed regions including such low emission regions shows a globally toroidal magnetic configuration at the GC. Acknowledgments. SN is financially supported by the Japan Society for the Promotion of Science (JSPS) through the JSPS Research Fellowship for Young Scientists. This work was also supported by KAKENHI, Grant-in-Aid for Young Scientists (B) 19740111, and Grant-in-Aid for Scientific Research (A) 19204018. References Chuss, D. T., et al. 2003, ApJ, 599, 1116 Davies, J. I., Trewhella, M., Jones, H., Lisk, C., Madden, A., Moss, J. 1997, MNRAS, 288, 679 Dotson, J. D., Davidson, J., Dowell, C. D., Schleuning, D. A., & Hildebrand, R. H. 2000, ApJS, 128, 335 Goodrich, R. W. 1986, ApJ, 311, 882 Kandori, R., et al. 2006, Proc. SPIE, 6269, 159 Morris, M. 1998, in IAU Symp. 184, The Central Regions of the Galaxy and Galaxies, ed. Y. Sofue (Dordrecht: Kluwer), 331 Nishiyama, S., et al. 2006, ApJ, 638, 839 Nishiyama, S., et al. 2009, ApJ, accepted, arXiv:0809.3089 [astro-ph] Novak, G., Dotson, J. L., Dowell, C. D., Hildebrand, R. H., Renbarger, T., & Schleuning, D. A. 2000, ApJ, 529, 241 Novak, G., Chuss, D. T., Renbarger, T., Griffin, G. S., Newcomb, M. G., Peterson, J. B., Loewenstein, R. F., Pernic, D., & Dotson, J. L. 2003, ApJ, 583, L83
0811.3756
1
0811
2008-11-24T08:53:31
Analytical Collapsing Solutions to Pressureless Navier-Stokes-Poisson Equations with Density-dependent Viscosity $\theta=1/2$ in $R^{2}$
[ "astro-ph", "math-ph", "math-ph" ]
We study the 2-dimensional Navier-Stokes-Poisson equations with density-dependent viscosity $\theta=1/2$ without pressure of gaseous stars in astrophysics. The analytical solutions with collapsing in radial symmetry, are constructed in this paper.
astro-ph
astro-ph
Analytical Collapsing Solutions to Pressureless Navier-Stokes-Poisson Equations with Density-dependent Viscosity θ = 1/2 in R2 Yuen Manwai∗ Department of Applied Mathematics, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong Revised 24-Nov-2008v2 Abstract We study the 2-dimensional Navier -- Stokes-Poisson equations with density-dependent vis- cosity θ = 1/2 without pressure of gaseous stars in astrophysics. The analytical solutions with collapsing in radial symmetry, are constructed in this paper. 1 Introduction The evolution of a self-gravitating fluid can be formulated by the Navier-Stokes-Poisson equations of the following form: ρt+∇ • (ρu) =0, (ρu)t+∇ • (ρu ⊗ u) + ∇P =−ρ∇Φ + vis(ρ, u), ∆Φ(t, x) = α(N )ρ,   where α(N ) is a constant related to the unit ball in RN : α(1) = 2; α(2) = 2π and For N ≥ 3, α(N ) = N (N − 2)V (N ) = N (N − 2) πN/2 Γ(N/2 + 1) , (1) (2) where V (N ) is the volume of the unit ball in RN and Γ is a Gamma function. And as usual, ρ = ρ(t, x) and u = u(t, x) ∈ RN are the density, the velocity respectively. P = P (ρ) is the pressure. ∗E-mail address: [email protected] 1 2 M.W.Yuen In the above system, the self-gravitational potential field Φ = Φ(t, x) is determined by the density ρ through the Poisson equation. And vis(ρ, u) is the viscosity function: vis(ρ, u) = ▽(µ(ρ) ▽ •u). Here we under a common assumption for: µ(ρ) . = κρθ (3) (4) and κ and θ ≥ 0 are the constants. In particular, when θ = 0, it returns the expression for the u dependent only viscosity function: And the vector Laplacian in u(t, r) can be expressed: vis(ρ, u) = κ∆u. ∆u = urr + N − 1 r ur − N − 1 r2 u. (5) (6) The equations (1)1 and (1)2 (vis(ρ, u) 6= 0) are the compressible Navier-Stokes equations with forcing term. The equation (1)3 is the Poisson equation through which the gravitational potential is determined by the density distribution of the density itself. Thus, we call the system (1) the Navier-Stokes-Poisson equations. Here, if the vis(ρ, u) = 0, the system is called the Euler-Poisson equations. In this case, the equations can be viewed as a prefect gas model. For N = 3, (1) is a classical (nonrelativistic) description of a galaxy, in astrophysics. See [2], [6] for a detail about the system. P = P (ρ) is the pressure. The γ-law can be applied on the pressure P (ρ), i.e. P (ρ) = Kργ . = ργ γ , (7) which is a commonly the hypothesis. The constant γ = cP /cv ≥ 1, where cP , cv are the specific heats per unit mass under constant pressure and constant volume respectively, is the ratio of the specific heats, that is, the adiabatic exponent in (7). In particular, the fluid is called isothermal if γ = 1. With K = 0, we call the system is pressureless. For the 3-dimensional case, we are interested in the hydrostatic equilibrium specified by u = 0. According to [2], the ratio between the core density ρ(0) and the mean density ρ for 6/5 < γ < 2 is given by ρ ρ(0) =(cid:18)−3 z y (z)(cid:19)z=z0 where y is the solution of the Lane-Emden equation with n = 1/(γ − 1), y(z) + 2 z y(z) + y(z)n = 0, y(0) = α > 0, y(0) = 0, n = 1 γ − 1 , (8) (9) Analytical Collasping Solutions and z0 is the first zero of y(z0) = 0. We can solve the Lane-Emden equation analytically for 3 (10) yanal(z) . = n = 0; n = 1; , n = 5,   1 − 1 6 z2, sin z , z 1 p1 + z2/3 and for the other values, only numerical values can be obtained. It can be shown that for n < 5, the radius of polytropic models is finite; for n ≥ 5, the radius is infinite. Gambin [4] and Bezard [1] obtained the existence results about the explicitly stationary solution (u = 0) for γ = 6/5 in Euler-Poisson equations: where A is constant. ρ =(cid:18) 3KA2 2π (cid:19)5/4 (cid:0)1 + A2r2(cid:1)−5/2 , The Poisson equation (1)3 can be solved as Φ(t, x) =ZRN G(x − y)ρ(t, y)dy, (11) (12) where G is the Green's function for the Poisson equation in the N -dimensional spaces defined by G(x) . =  x, N = 1; log x, N = 2; −1 xN −2 , N ≥ 3. (13) In the following, we always seek solutions in radial symmetry. Thus, the Poisson equation (1)3 is transformed to rN −1Φrr (t, x) + (N − 1) rN −2Φr=α (N ) ρrN −1, (14) Φr = α (N ) rN −1 Z r 0 ρ(t, s)sN −1ds In this paper, we concern the analytical solutions with core collapsing for the 2-dimensional pressureless Navier-Stokes-Poisson equations with the density-dependent viscosity. And our aim is to construct a family of such core collapsing solutions. Historically in astrophysics, Goldreich and Weber [5] constructed the analytical collapsing (blowup) solution of the 3-dimensional Euler-Poisson equations for γ = 4/3 for the non-rotating gas spheres. After that, Makino [7] obtained the rigorously mathematical proof of the existence of such kind of collapsing solutions. And in [3], the extension of the above collapsing solutions to the higher dimensional cases (N ≥ 4). After that, for the 2-dimensional case, Yuen constructed the analytical collapsing solutions for γ = 1, in [8]. For the construction of the analytical solutions to the Navier-Stokes equations in RN , the Navier-Stokes-Poisson equations in R3 without pressure with θ = 1 and in R4 without pressure with θ = 5/4, readers may refer Yuen's recent results in [9], [10], [11] respectively. 4 M.W.Yuen In this article, the analytical collapsing solutions are constructed in the pressureless Navier -- Stokes-Poisson equations with density-dependent viscosity in R2, with θ = 1/2, in radial symmetry: ρt + uρr + ρur + 1 r ρu = 0, ρ (ut + uur) = − 2πρ r Z r 0 ρ(t, s)sds + [κρ1/2]rur + (κρ1/2)(urr + 1 r ur − 1 r2 u), (15) in the form of the following theorem.   Theorem 1 For the 2-dimensional pressureless Navier -- Stokes-Poisson equations with θ = 1/2, in radial symmetry, (15), there exists a family of solutions, ρ(t, r) = 1 (T − Ct)2y( 2π√κ y(z)+ 1 z y(z) − C   T −Ct )2 , u(t, r) = −C T − Ct r r; (16) 1 y(z)2 = 0, y(0) = α > 0, y(0) = 0, where T > 0, κ > 0, C 6= 0 and α are constants. In particular, for C > 0, the solutions collapse in the finite time T /C. 2 Separable Blowup Solutions Before presenting the proof of Theorem 1, we prepare some lemmas. First, we obtain the solutions for the continuity equation of mass in radial symmetry (15)1. Lemma 2 For the equation of conservation of mass in radial symmetry: ρt + uρr + ρur + 1 r ρu = 0, there exist solutions, ρ(t, r) = 1 (T − Ct)2 y( r T −Ct )2 , u(t, r) = −C T − Ct r, with the form y 6= 0 and y ∈ C1, C and T > 0 are constants. Proof. We just plug (18) into (17). Then (17) (18) ρt + uρr + ρur + ρu 1 r (−2)(−C) = + + r T −Ct )2 (−2) (T − Ct)2y( (T − Ct)3 y( (−C)r T − Ct 1 r 1 (T − Ct)2y( r T −Ct )2 + · (−2)(−1)(−C)r (T − Ct)4 y( y( T −Ct ) + T − Ct r r r T −Ct )3 (−C) T − Ct = 0. · r T −Ct ) y( r T −Ct )3 1 (T − Ct)2y( r T −Ct )2 (−C) T − Ct Analytical Collasping Solutions 5 The proof is completed. Besides, we need the lemma for stating the property of the function y(z). The similar lemma was already given in Lemmas 9 and 10, [8], by the fixed point theorem. For the completeness, the proof is also presented here. Lemma 3 For the ordinary differential equation, y(z) − σ y(z)2 = 0 y(0) = α > 0, y(0) = 0, 1 z y(z)+   y(z) = ∞. where σ is a positive constant, has a solution y(z) ∈ C2 and lim Proof. By integrating (19), we have, z→+∞ · y(z) = σ z Z z 0 1 y(s)2 sds ≥ 0. (19) (20) Thus, for 0 < z < z0, y(x) has a uniform lower upper bound As we obtained the local existence in Lemma 3, there are two possibilities: y(z) ≥ y(0) = α > 0. (1)y(z) only exists in some finite interval [0, z0]: (1a) lim z→z0− upper bound, i.e. y(z) ≤ α0 for some constant α0. (2)y(z) exists in [0, +∞): (2a) lim y(z) ≤ β for some positive constant β. We claim that possibility (1) does not exist. We need to reject (1b) first: If the statement (1b) is y(z) = ∞; (2b)y(z) has an uniformly upper bound, i.e. y(z) = ∞; (1b)y(z) has an uniformly z→+∞ true, (20) becomes σz 2α2 = σ z Z z 0 s α2 ds ≥ · y(z). (21) · Thus, y(z) is bounded in [0, z0]. Therefore, we can use the fixed point theorem again to obtain a large domain of existence, such that [0, z0 +δ] for some positive number δ. There is a contradiction. Therefore, (1b) is rejected. Next, we do not accept (1a) because of the following reason: It is impossible that as from (21), · y(z) has an upper bound in [0, z0]: Thus, (22) becomes, σz0 2α2 ≥ · y(z). lim z→z0− y(z) = ∞, (22) · y(s)ds y(z0) = y(0) +Z z0 ≤ α +Z z0 0 σz0 2α2 ds = α + 0 σ(z0)2 2α2 6 Since y(z) is bounded above in [0, z0], it contracts the statement (1a), such that So, we can exclude the possibility (1). We claim that the possibility (2b) doesn't exist. It is because M.W.Yuen lim z→z0− y(z) = ∞. · y(z) = Then, we have, σ z Z z 0 s y(s)2 ds ≥ σ z Z z 0 s β2 ds = σz 2β2 . y(z) ≥ α + σ 4β2 z2. (23) By letting z → ∞, (23) turns out to be, y(z) = ∞. Since a contradiction is established, we exclude the possibility (2b). Thus, the equation (19) exists in [0, +∞) and lim y(z) = ∞. This completes the proof. z→+∞ Here we are already to give the proof of Theorem 1. Proof of Theorem 2. From Lemma 2, it is clear for that (16) satisfy (15)1. For the momentum equation (15)2, we get, ρ(t, s)sds − [µ(ρ)]rur − µ(ρ)(urr + 1 r ur − 1 r2 u) r r ∂ = ρ(ut + uur) + 2πρ r + Z0 = ρ(cid:20) (−C)(−1)(−C) (−C) (T − Ct)2 T − Ct T − Ct − µ(ρ)(cid:18)0 + (−C) 1 − (κρ1/2)r r C−ct )2#1/2 ∂r " T −Ct )2#−1/2 2" T −Ct )2 sds Z0 κ (T − Ct)2y( 1 Z0 (T − Ct)2y( ρ y( T −Ct ) (T − Ct) κ (T − Ct)2y( = −C √κ 2πρ 2πρ + = + 1 r r s r r r r r r · T − Ct(cid:21) + (−C) (−1) T − Ct − 1 r2 C T − Ct + 2πρ r r r 2πρ Z0 (−1) T − Ct Z0 r (−2) (T − Ct)2y( r T −Ct )3 T −Ct )2 sds r 1 (T − Ct)2y( r(cid:19) 1 (T − Ct)2y( T −Ct ) y( T − Ct r T −Ct )2 sds r C T − Ct (24) (25) (26) (27) (28) (29) (30) 1 (T − Ct)2y( s T −Ct )2 sds Z0 y( r r/(T −Ct) = = −ρ (T − Ct)  (T − Ct)  −ρ C √κ y( r T − Ct C √κ y( r T − Ct ) − ) − 2π r(T − Ct) Z0 2π r T −Ct 1 s T −Ct )2 sds  y(τ )2 τ dτ  1 . Analytical Collasping Solutions By letting τ = r/(T − Ct), it follows: = = r y( −ρ (T − Ct) C √κ  T − Ct T − Ct(cid:19) . Q(cid:18) r (T − Ct) −ρ r/(T −Ct) ) − 2π r T −Ct Z0 1 y(τ )2 τ dτ  7 (31) And denote z = r/(T − ct), Q( r T − Ct ) = Q(z) = C √κ y(z)− 2π z z Z0 1 y(τ )2 τ dτ. Differentiate Q(z) with respect to z, Q(z) = C √κ y(z) − 2π y(z)2 + 2π z2 z Z0 1 y(τ )2 τ dτ C √κ = − y(z) z + 2π z2 z Z0 1 y(τ )2 τ dτ 1 z = − Q(z), where the above result is due to the fact that we choose the following ordinary differential equation: y(z)+ 1 z y(z) − 2π√κ C 1 y(z)2 = 0 y(0) = α > 0, y(0)= 0.   With Q(0) = 0, this implies that Q(z) = 0. Thus, the momentum equation (15)2 is satisfied. With Lemma 3 about y(z), we are able to show that the family of the solutions collapse in finite time T /C. This completes the proof. The statement about the blowup rate will be immediately followed: Corollary 4 The collapsing rate of the solution (16) is lim t→T /C− ρ(t, 0)(T − Ct)2 ≥ O(1). (32) Remark 5 Besides, if we consider the 2-dimensional Navier-Stokes equations with the repulsive force in radial symmetry with θ = 1/2, ρt + uρr + ρur + 1 r ρu = 0, ρ (ut + uur) = +   the special solutions are: 0 ρ(t, s)sds + [κρ1/2]ur + (κρ1/2)(urr + 2πρ r R r 1 r ur − 1 r2 u), (33) ρ(t, r) = y(z)+ 1 z y(z) +   1 (T − Ct)2y(cid:16) r 2π√κ 1 T −Ct(cid:17)2 , u(t, r) = −C T − Ct y(z)2 = 0, y(0) = α 6= 0, C y(0) = 0, r (34) 8 References M.W.Yuen [1] M. Bezard, Existence locale de solutions pour les equations d'Euler-Poisson. (French) [Local Existence of Solutions for Euler-Poisson Equations] Japan J. Indust. Appl. Math. 10 (1993), no. 3, 431 -- 450. [2] S. Chandrasekhar, An Introduction to the Study of Stellar Structure, Univ. of Chicago Press, 1939. [3] Y.B. Deng, J.L. Xiang and T. Yang, Blowup Phenomena of Solutions to Euler-Poisson Equa- tions, J. Math. Anal. Appl. 286 (1)(2003), 295-306. [4] P. Gamblin, Solution reguliere a temps petit pour l'equation d'Euler-Poisson. (French) [Small- time Regular Solution for the Euler-Poisson Equation] Comm. Partial Differential Equations 18 (1993), no. 5-6, 731 -- 745. [5] P.Goldreich, S. Weber, Homologously Collapsing Stellar Cores, Astrophys, J. 238, 991 (1980). [6] R. Kippenhahn, A,Weigert, Stellar Sturture and Evolution, Springer-Verlag, 1990. [7] T. Makino, Blowing up Solutions of the Euler-Poission Equation for the Evolution of the Gaseous Stars, Transport Theory and Statistical Physics 21 (1992), 615-624. [8] M.W. Yuen, Analytical Blowup Solutions to the 2-dimensional Isothermal Euler-Poisson Equa- tions of Gaseous Stars, J. Math. Anal. Appl. 341 (1)(2008), 445-456. [9] M.W. Yuen, Analyitcal Solutions to the Navier-Stokes Equations, Journal of Mathematical Physics, 49 (2008) No. 11, 113102, 10pp. [10] M. W. Yuen, Analytical Solutions to the 3-dimensional Pressuless Navier-Stokes-Poisson Equations with Density-dependent Viscosity, Submitted, arXiv:0811.0379v1. [11] M. W. Yuen, Analytical Solutions to the 4-dimensional Pressuless Navier-Stokes-Poisson Equations with Density-dependent Viscosity, Submitted, arXiv:0811.1323v1.