Datasets:

SouthernCrossAI
/

Arxiv

Dataset card Data Studio Files Files and versions Community

Split (1)

train · 132k rows

Filename stringlengths 22 64	Paragraph stringlengths 8 5.57k
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	These dark state gates were implemented and the qubit superposition states that were created as a result of the rotations, were characterized using quantum state ) tomography. In Figure 5a the readout of a particular superposition state, i is shown as an example. The three curves correspond to projection measurements on the x-, y- and z-axis respectively. As expected, we see that both the y and the z measurement contain equal contributions from either state, but the x measurement gives a projection onto one of the states. States corresponding to the six diﬀerent (positive and negative) axes on the Bloch sphere, were prepared, and characterized, and the resulting state vectors inscribed in a Bloch sphere are displayed in Figure 5b.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	i.e. the overlap between the theoretical and the experimental representations. For the six states displayed in Figure 5b, the total ﬁdelities were all between 0.84 and 0.92. As mentioned above the total procedure consists of two rotations, ﬁrst one to create the state, then another one to perform the measurement in the right basis. Thus, the ﬁdelity for a single qubit gate can be expressed as Fgate = √Ftot, and we see that the qubit gate ﬁdelities are in the range of 0.91 < Fgate < 0.96. More information and discussion regarding these results can be found in [8, 18].
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	Even though the dark state qubit gates were realized with good ﬁdelity there are still reasons to search for further improvements. The total duration of the dark state pulses that performs one arbitrary qubit rotation is close to 18 µs. The employed laser power on the other hand, typically provides Rabi frequencies in the order of 1 MHz, which indicates that the qubit rotations could probably be performed an order of magnitude faster, with an optimal scheme. To reduce the operation time, pulses obtained through optimal control theory calculations are currently investigated.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	Optimal control theory provides a framework for ﬁnding the best set of controls to steer a system so that a desired target state or unitary gate is implemented. In some special cases these controls can be determined analytically, see for example [20, 21, 22]. In addition, powerful numerical methods [23] are available that make it possible to explore the physical limits of time-optimal control experiments in cases where no analytical solutions are known. These methods have so far been successfully applied to spin systems [24, 25, 26] and superconducting qubits [27], and trapped ions [28]. Optimized pulses can be designed to account for experimental errors, and, if a realistic model is available, can include multiple system levels and transitions.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	Figure 6: (color online) Two pulse shapes with diﬀerent bandwidths, both designed by optimal control algorithms to perform eﬃcient state-to-state transfers. The left side shows the detected beating and the amplitude envelope, while the right side shows the phase evolution. In both cases the red line is the theoretically desired shape, while the blue line is the experimentally detected shape, and in both cases they are matching very well. Note that in the phase plot, the phase does not have any meaning outside the pulse envelope (since the amplitude there is zero).
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	solution is not necessarily unique, or even at the global maximum, and the shapes that comes out of the calculations are then often strange-looking, without obvious intuitive explanations. In order to characterize how well the AOM system could reproduce these strange shapes, an interference experiment was set up, where the AOM pulse-shaping system is put in one arm and the light passing through this part then interferes with the light from an undisturbed part of the laser. The result is a beating, from which both the amplitude envelope and the phase evolution of the pulses can be obtained via a Fourier transform. Figure 6 shows two examples of such pulses. In panel a) of this ﬁgure, the green, oscillating curve is the detected beating, the blue line is the amplitude envelope of the beating, which is almost perfectly overlapping with the red line, which is the theoretically desired shape (may be hard to see if not in color).
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	for all bandwidths except the highest (16 MHz). The reason for this is not yet fully known, but is believed to be related to complications that arise when multiple levels are involved. From the level diagram (Figure 1) we see that any pulse that centers 4.6 = 9.2 MHz will transition and has a bandwidth less than 2 around the only involve those two levels and nothing else, but any pulse with a higher bandwidth will automatically involve at least the two other excited states as well, which creates a much more complicated situation. For two-level schemes, other groups have also obtained very good results, for example in ion traps [11]. But clearly, there should also be suitable optimal pulses for three-level conﬁgurations. Thus, the good results for the simpler pulses together with the potential of an order of magnitude faster pulses make the optimal control technique seem promising for the future.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	Scaling to large number of qubits is a key issue for a quantum computer. In the present system, the qubit-qubit operation is achieved by the dipole-dipole interaction between two neighboring ions. As stated in Section 2 they must physically sit close enough to from the ground state each other so that when the state of one ion is changed, e.g. to the excited state and the electric ﬁeld from the static dipole moment changes, this will shift the neighboring ion out of resonance with its original transition frequency. In this way, closely lying ions can control each other. For an ion absorbing at a selected frequency, νi, there will be several ions in the vicinity that can control this ion. In the ensemble approach, for two qubits, i and j, at transition frequencies νi and νj, there is a probability, p, that for an arbitrary ion, ai, in qubit i, there will be an ion in qubit j that can control ai. For n qubits, if N denotes the number of ions in a qubit that can 1. For the materials be controlled by ions in the other n studied so far p is in the order of 1%, which means that for a 5-qubit quantum register there will only be 1 out of 108 ions that are useful in each qubit.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	e.g. choosing ions with large static dipole moment, increasing the dopant concentration [29], constructing the qubit-qubit interaction by using a mediator bus ion [12] or adding a specially selected readout ion to detect the state of single ions, thereby enabling the use of single ions instead of an ensemble of ions for each qubit. Considering both advantages and disadvantages in each scheme the single ion readout approach is believed to be the easiest way to achieve the scalability.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	Inside any small volume within a rare-earth-ion doped crystal there is a high probability to ﬁnd several strongly interacting ions which can control each other and each of them can represent one qubit. A readout scheme has been suggested for detecting the qubit state in such a single ion qubit system through the dipole-dipole interaction [12]. The dopant concentration of the readout ion should be very low to make sure there is on average only one readout ion interacting with the laser ﬁeld. To fulﬁll its role, the readout ion is supposed to meet several requirements: (i) The absorption spectrum should be well separated from that of the qubit ion so that the readout procedure does not aﬀect the qubit state. (ii) The excited state lifetime should be short and there should not be any trapping state. Thus, the excitation may be cycled many times producing many photons and giving a strong detection signal. (iii) The homogeneous linewidth should be narrow so that when a nearby qubit ion is transferred between the ground and excited states, the dipole-dipole interaction is able to shift the readout ion in or out of resonance with the readout beam frequency, which consequently turns on or oﬀ the ﬂuorescence. (iv) The inhomogeneous linewidth should be large which provides a larger number of frequency channels.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	Figure 8: (color online) The dipole-dipole interaction between a qubit ion and a near- lying readout ion. If the qubit ion q1 is excited from the state at frequency ν1, the electric ﬁeld generated by the dipole will shift the transition frequency of the readout ion by an amount δν, after which excitation at frequency νr will have no eﬀect and the readout ion ﬂuorescence will be turned oﬀ.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	So far Ce3+ is believed to be the most promising candidate for the readout ion. The 5d zero-phonon-line (ZPL) of Ce3+ ion randomly excitation wavelength for the 4f doped in the YSO-crystal is around 371 nm, which is well separated from the qubit ion transition frequency 606 nm (Pr) or 580 nm (Eu). The Ce3+ homogeneous linewidth is determined by detecting the sum frequency signal from a saturation spectroscopy experiment with diﬀerent amplitude modulation frequencies on the pump and probe beams. The measured homogeneous linewidth was as narrow as 3 MHz. This is a very encouraging result, because based on a reasonable assumption about the Ce dipole moment, the Pr ion should be able to shift the Ce transition line by 30 MHz when they are separated by 7 nm [30]. With a 50 ns lifetime [31, 32] this also means that the homogeneous linewidth is lifetime-limited which is the ideal case for this readout ion concept.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	are stronger for 4f 4f transitions. Clearly this will aﬀect the readout ﬁdelity. Assume the detector quantum eﬃciency is 10% and the ﬂuorescence collection eﬃciency is 30% [33], and a π-pulse is applied on the qubit transition. As the readout laser is turned on, there will then be about 100 ﬂuorescence photons detected during the optical lifetime (150 µs) of the Pr3+ ion, if the qubit was in its state, the ZPL background 1 \| ﬂuorescence might still give around 50 photons due to oﬀ-resonant excitations (the background signal). However, the diﬀerence in the number of photons is certainly still suﬃcient for separating the two cases. One way to reduce the background signal, would be to decrease the number of Ce ions within the laser focus.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	When the single-ion readout system is scaled up by adding new qubits to the (possibly branched) qubit ion chain there will, although there are many frequency channels available within the inhomogeneously broadened qubit transition line, eventually be a new qubit ion (Qnew) that have the same frequency as one of the qubit ions (Qold) already in the chain. In this situation, the two ions (Qold and Qnew) cannot any longer be individually addressed in frequency space. It is conceivable that there are algorithmic solutions to this problem, where single qubit operations could be replaced by two-qubit operations, so that operations are applied to ”the qubit with frequency νi, sitting next to a qubit with frequency νj”. As an alternative, the problem of coinciding transition frequencies can be removed by mounting (closely spaced) electrodes onto the crystal. The part of the qubit chain on which the operations are carried out, is selected by applying a voltage on the appropriate electrode. Eu in Y2SiO5 has a Stark coeﬃcient of 35 kHz/(V/cm) [34], and ions close to the electrode can be shifted into resonance with the excitation pulse. A 1 MV/cm ﬁeld would shift the transition frequency 35 GHz, a detuning much larger than the inhomogeneous transition line width. Simulations show that 20 nm long electrodes separated by 40 nm on a surface may give shifts of the order of 30 GHz within a 60 nm region while ions 20 nm further away would shift less than 10 GHz [12]. However, still smaller electrodes and electrode separations would be favorable and while this is in line with the current technological development we will also now analyze the possibility of entangling spatially remote few-qubit systems.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	Thus, we consider spatially separated qubit chains, each containing a not too large number of qubits. There are several schemes for entangling spatially remote few-qubit systems. To optically entangle a qubit in one system with a qubit in another system the key point is to conﬁgure the photon detection (or absorption) process such that when a photon is detected (absorbed) it cannot in principle be determined which of the two systems that emitted (absorbed) the photon [35, 36]. If the set-up is appropriately conﬁgured, the photon detection/absorption entangles the two qubits.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	We consider a rare earth crystal with a Ce readout ion, close to the surface facing a toroidal microcavity. Around this readout ion there will be a number of qubit ions that interact and can be read out, as described in Section 6. One of these qubit ions is selected as the entangler ion. The purpose is now to entangle entangler ions in diﬀerent rare-earth crystals. Such a scheme requires that also rare-earth ions close to the crystal surface retain their favorable coherence properties, and preliminary data indicate that this indeed is the case [42].
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	In a more complete version of the approach in Figure 9, there would be a large number of microresonators (with their corresponding crystals) along the tapered ﬁbre. Which two crystals that are selected for any speciﬁc entanglement process can be controlled by e.g. adjusting the distance between the tapered ﬁbre and the resonator (cavity) using e.g. micro-actuators. Alternatively, the cavity can be tuned oﬀ resonance with the transition and photon frequency. Tuning the cavity can be carried out using expansion due to temperature. For example, electrodes can be connected to the toroid and the substrate. The temperature and thereby also the frequency change will then be proportional to the dissipated power [46]. In general the system could be quite ﬂexible and versatile, as an example, ions can be tuned in and out of resonance with the cavity using an electric ﬁeld across the crystal.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	As earlier stated, the qubit coherence times can be 30 seconds for Pr ions [7] and for Eu ions coherence times of the order of an hour have been predicted [47]. A strength with the scheme outlined above is therefore that there is ample time both to create entanglement and to carry out operations once the systems have been entangled.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	Figure 9: (color online) Schematic view of an approach to entangle spatially separate few qubit processors.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	Pr3+:Y2SiO5 which has an oscillator strength almost six orders of magnitude lower than free Cs atoms [34], would then yield a Rabi frequency of about 150 kHz. This value is too low for a single photon to act as a π-pulse and we are therefore looking at techniques for enhancing the interaction between an ion just below the surface and an evanescent ﬁeld.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	To summarize, single-qubit operation ﬁdelities above 90% are obtained for ensemble qubits and > 99% single qubit operation ﬁdelity should be readily obtained with single- ion qubits. However, the single ion qubit scheme is contingent upon the ability to read out the hyperﬁne state of single rare-earth ions and work to develop this capacity is still ongoing. Although, based on current data in the investigation of using Ce3+ as a readout ion for such a scheme, there is reason to be quite optimistic about the possibilities to succeed. Finally, it may be pointed out that while quantum state storage in rare-earth- ion-doped materials is pursued vigorously by several groups and now demonstrates excellent progress [49] including storage eﬃciencies above 30% [50], only two groups [51, 8] have published experimental eﬀorts on quantum computing in rare earth crystals and considering the limited eﬀorts, substantial progress has been made.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	This work was supported by the Swedish Research Council, the Knut and Alice Wallen- berg Foundation, the European Commission through the integrated project QAP and the Crafoord Foundation.
Processed_Extracting_high_fidelity_quantum_computer_hardware.txt	Ideality in a ﬁber-taper- coupled microresonator system for application to cavity quantum electrodynamics. Phys. Rev. Lett., 91(4):043902, Jul 2003.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	A possibility for fundamental constants to vary in time is suggested by theories unifying gravity with other interactions. In this article we examine proposals to use optical transitions of Sr, Dy, YbII and YbIII for the search of the time variation of the ﬁne structure constant α. Frequencies of atomic transitions are calculated using relativistic Hartree-Fock method and conﬁgu- ration interaction technique. The eﬀect of variation of α on the frequencies is studied by varying α in computer codes. Accuracy of measurements needed to improve current best limit on the time variation of α is discussed.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	Theories unifying gravity with other interactions suggest that fundamental constants could vary in space-time (see, e.g. [1]). Recent evidence of variation of the ﬁne structure constant α in quasar absorption spectra [2] elevated interest to the search of variation of α in laboratory experiments. Comparing frequencies of diﬀerent atomic transitions over long period of time is a good way to do such search due to extremely high accuracy of mea- surements achieved for certain types of transitions. The best limit on local present-time variation of the ﬁne structure constant published so far was obtained by comparing Hg+ microwave atomic clock vs hydrogen maser [3]. Recently this limit was further improved by more than an order of magnitude in comparing cesium and rubidium atomic clocks [4]. There are also many proposals for the search of variation of α in atomic optical transitions, some of which were analyzed in our previous works (see [5] and references therein). In the present paper we analyze three new proposals involving strontium/calcium, dual beam [6], dysprosium atom [5,7] and ytterbium positive ions Y b+ [8] and Y b2+ [9]. We perform rela- tivistic many-body calculations to link variation of α with the variation of the frequencies of atomic transitions. Then we use this connection to ﬁnd out what accuracy of measurements is needed to improve current best limit on time variation of the ﬁne structure constant.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	In the proposal suggested by S. Bergeson strontium-calcium dual beam is to be used to compare the frequencies of the 1S0 −3 P1 clock transitions in these atoms over a long period of time. Ca and Sr have similar electron structure. However, due to higher nuclear charge, relativistic eﬀects are larger for strontium. If α is changing, corresponding change in frequency of the clock transition for Sr would go considerably faster than for Ca. Precise measurements might be able to indicate this or, at least, put strong constrain on possible variation of α. Calculations of the relativistic eﬀects for Ca were done in our previous work [5]. In present paper we do similar calculations for Sr.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	Experiments with ytterbium positive ion have advantages of greater relativistic eﬀects due to larger nuclear charge and the convenience of working with two diﬀerent transitions of the same element. There are two transitions in Yb+ involving metastable states for which comparison of frequencies is considered. One is quadrupole transition 4f 146s 2S1/2 − 4f 145d 2D5/2 and another is octupole transition 4f 146s 2S1/2−4f 136s2 2F7/2. The quadrupole transition is basically a s−d transition while the octupole one is a f −s transition. According to simple analytical formula presented in Ref. [5] relativistic energy shifts for s electrons, and electrons with high total momentum j (like d and f electrons) are large but have opposite sign. This means that we should expect that two metastable states of Yb+ move in opposite directions if α is changing. This brings extra enhancement to the sensitivity of the measurements for Yb+ to the variation of α. Our accurate calculations presented below support these considerations.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	between the states to variation of α. The enhancement (about eight orders of magnitude) seems to be strong enough to overcome the disadvantage of dealing with states which are not very narrow.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	In the present paper we calculate the values of relativistic energy shifts for Sr, Yb+ and Dy and discuss what accuracy of measurements is needed to improve current best constrain on local time variation of the ﬁne structure constant.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	here κ = (−1)l+j+1/2(j + 1/2), n is the principal quantum number and ˆVHF is the Hartree- Fock potential. The value of relativistic eﬀects is studied by varying the value of α in (2). In particular, non-relativistic limit corresponds to α = 0.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	than those with ˆΣ1 but much more time consuming in calculations. We either neglect them or simulate their eﬀect by introducing screening factors.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	We are now going to discuss the speciﬁcs of the calculations for each atom/ion. Apart from the states of interest we also calculate energies of the other states of the same con- ﬁgurations to ensure that the accuracy is systematically good. We also calculate magnetic g-factors to ensure correct identiﬁcation of states. This is particularly important for dys- prosium.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	It has two 5s-electrons on its outermost shell and we need to consider energy intervals between 1S0 ground state and states of the 5s5p conﬁguration where the 3P1 metastable state is of most interest. The RHF calculations for Sr were done in V N approximation, for a closed-shell atom in its ground state. For the CI calculations we considered Sr as an atom with two valence electrons and followed the similar calculations for Ba [13]. Basis states for the CI+MBPT method were calculated using the B-spline technique [14] with 40 B-splines in a cavity of radius R = 40aB. The same basis functions were used to calculate ˆΣ1 and for the CI calculations. Thirteen lowest states above core in each of the s1/2, p1/2, p3/2, d3/2 and d5/2 waves were used to construct two-electron wave function for both 5s2 and 5s5p conﬁgurations. Large number of basis functions is needed mostly for adequate description of the 5s5p conﬁguration. This is because the V N approximation doesn’t provide us with a good 5p single-electron state. Also, the 5s single-electron state in the 5s5p conﬁguration is diﬀerent from the 5s state in the 5s2 conﬁguration for which Hartree-Fock calculations were done. However, with thirteen states in each wave the saturation of the basis was clearly achieved and adding more states to the basis didn’t change the energy. Two-electron basis states for the CI calculations were obtained by distributing valence electrons over 65 basis states (13 × 5) in all possible ways with a restriction of ﬁxed parity and total momentum.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	The ground state of ytterbium positive ion is 4f 146s 2S1/2 and we need to consider transitions into the 4f 145d 2D5/2 and 4f 136s2 2F7/2 states. Therefore it is convenient to do the RHF calculations in the V N −1 approximation, for the Yb2+ ion with the 4f 14 closed-shell conﬁguration. The 6s, 5d and other basis states for the CI method are calculated then in the ﬁeld of frozen closed-shell core of Yb2+. Then, in the CI calculations, we need to consider all 4f electrons as valence ones since one of the transitions of the interest involves excitation from the 4f subshell. So, the total number of valence electrons in present CI calculations is ﬁfteen. This is very diﬀerent from our previous calculations for Yb+ [5] in which the 4f 136s2 2F7/2 state was not considered and we were able to treat ytterbium ion as a system with one external electron above closed shells.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	Our ﬁnal set of single-electron states for the CI calculations consisted of 4f5/2, 4f7/2, 6s1/2, 5d3/2, 5d5/2 and few more s and f states above 4f and 6s. Note that in contrast with Sr we don’t need many basis functions here because all our single-electron wave functions correspond to the Yb+. This makes initial approximation to be very good and leads to fast convergence of the CI calculations with respect to the basis set used.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	We also don’t include ˆΣ1 in calculations for Yb+. In a case of many valence electrons (ﬁfteen for Yb+) correlations are dominated by correlations between them which are taken into account accurately via the CI technique. Correlations between valence electrons and core electrons mostly manifest themself via screening of the Coulomb interaction between valence electrons. We take this eﬀect into account semiempirically, by introducing screening factors fk. Namely, we multiply every Coulomb integral of the multipolarity k by a numerical factor fk which is chosen to ﬁt the energies. It turns out that good ﬁt for Yb+ is achieved with f2 = 0.8 and fk = 1 for all other k.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	Many-electron basis states for the CI calculations were obtained by allowing all possible single and double excitations from the base conﬁguration with the restriction of ﬁxed parity and total momentum.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	Results for energies of Yb+ are presented in Table I. The theoretical accuracy for energies as compared to the experiment is 2- 3% for the states of interest and is not worse than 5% for other states.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	Dysprosium atom is the most diﬃcult for calculations because of its complicated electron structure. Ground state conﬁguration of Dy is 4f 106s2 which means that there is no realistic RHF approximation which corresponds to a closed-shell system. We do the RHF calculations for Dy in the V N approximation with an open-shell version of the RHF method. Contribution of the 4f electrons into the RHF potential is calculated as for a closed shell and then multiplied by a numerical factor to take into account its fractional occupancy. This factor is 10/14 when interaction of the 4f electrons with other core electrons is considered and 9/13 when interaction of a 4f electron with other 4f electrons is considered. When convergence is achieved we have the 4f and 6s basis states for the CI calculations. To calculate other states of valence electrons we remove one 6s electron, freeze all RHF orbitals, including 4f and 6s and calculate the 6p1/2, 6p3/2, 5d3/2, 5d5/2 and few more d-states above 5d in the ﬁeld of frozen RHF core.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	In the CI calculations states below 4f are considered as core states and all other as valence states. Total number of valence electrons is therefore twelve. As for the case of Yb+ we neglect ˆΣ1 and use screening factors as ﬁtting parameters to improve agreement with experiment. It turns out that best ﬁt for the 4f 106s6p conﬁguration is achieved with f1 = 0.7 and fk = 1 for all other k. No ﬁtting was used for other conﬁgurations.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	computers. On the other hand test calculations with pairs of conﬁgurations showed that mixing of our state of interest with other conﬁgurations is small and can be neglected. We do need however to include mixing with the 4f 95d6d6s, 4f 95d7d6s and 4f 96d26s conﬁgura- tions. This is because our basis 5d state corresponds rather to the 4f 105d6s conﬁguration and extra d-states are needed to correct it.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	The result are presented in Table I. Note that they are considerably better than in our previous calculations [15]. This is because of better basis and more complete CI treatment.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	Note that the width of 3P1 state in Sr may be a problem in this case.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	In the case of Yb+ frequencies of the 2S1/2 −2 D5/2 and 2S1/2 −2 F7/2 are to be compared.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	Note two orders of magnitude improvement in the magnitude in comparison with the Sr-Ca dual beam experiment.
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	This means that to improve current best limit on local time variation of α the frequency of this transition in Dy should be measured to the accuracy of about 10−7 over about a year time interval. This seems to be feasible [7].
Processed_Relativistic_effects_in_Sr,_Dy,_YbII_and_YbIII_and.txt	We are grateful S. Bergeson, P. Blythe, D. Budker,S. Lea, F. Pereira and J.R. Torgerson for useful discussions. Part of the work was done on the computers of the Australian Center for Advanced Computing and Communications. This work is supported by the Australian Research Council.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Abstract In wind- and solar-dominated energy systems it has been assumed that there are synergies between producing electricity and electrolytic hydrogen since electrolysis can use excess electricity that would otherwise be curtailed. However, it remains unclear whether these synergies hold true at higher levels of hydrogen demand and how they compare with benefits of off-grid, islanded hydrogen production, such as better renewable resources and cost savings on electronics due to relaxed power quality standards. Using a mathematical model across two geographical locations for Germany, Spain, Australia, and Great Britain, we explore trade-offs and synergies between integrated and islanded electrolysers. Below a certain threshold, between 5% and 40% hydrogen share depending on the country, integrated electrolysers offer synergies in flexibility and reduced curtailment. Above these thresholds, islanded electrolysers become more favourable. Without cost advantages, systems including islanded electrolysers in Germany achieve up to 21% lower hydrogen costs than systems with only integrated electrolysers. With 25% island cost advantage, this benefit rises to 40% lower hydrogen costs. Our study identifies three investment regimes with country-specific transition points that vary based on island cost advantages and each country’s renewable resources. Based on our results we provide guidelines for countries considering how to deploy electrolysers.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	countries position themselves in the emerging global market. Economic considerations, such as energy dependency, economic feasibility, and potential revenues from domestic markets, must be carefully weighed against each other to formulate optimal individual strategies. The European Union, for example, aims to produce 10 million tonnes of renewable hydrogen domestically by 2030, along with an equal amount of imports [3, 4]. Other countries like Chile or Australia aim to export large amounts of green hydrogen, which will likely exceed the total domestic energy demand by a large amount [5, 6]. Furthermore, the ramp-up of the green hydrogen economy will likely unfold at a very high pace [7]. As countries develop strategies to meet the quickly growing demand for green hydrogen, the question of optimal production methods thus becomes increasingly important in the near future. System plan- ners and policy makers must carefully decide where to locate electrolysers [8] and how to integrate them into the existing energy system.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The implementation of green hydrogen production integrated in power systems and sec- tor-coupled energy systems has been extensively discussed in the scientific literature [9, 10, 11]. It is commonly acknowledged that integrated co-production of electricity and electrolytic hydrogen can benefit renewable energy systems by reducing curtailment of renewable power generation and providing a long-term storage solution [12, 13]. To determine the optimal placement of integrated electrolysers, various consideration have to be taken into account. Emerging electrolysers are either strategically located near hydrogen demand centers and industrial clusters where demand exists [14], or in resource-rich locations where electricity can be provided at low cost. This issue is not least related to the feasibility of expanding the existing power grid and/or creating a new hydrogen transport systems. In this regard, the study by Neumann et al. [10] looks at the impact of a hydrogen transport system in a fully renewable European energy system. The study shows that flexibility in transporting large quantities of hydrogen across the continent could reduce costs by up to 30%, especially at industrial sites.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	In contrast to integrated hydrogen production, the concept of dedicated hydrogen produc- tion sites, where all the electricity from dedicated wind and solar plants is used in electrolysis and which we refer to as “hydrogen islands”, is increasingly being discussed in the literature. Various studies indicate the feasibility of hydrogen production in specialised systems like solar and wind hybrid setups [15, 16, 17, 18, 19]. Specifically, the study by Garcia G. et al. demonstrated the economic feasibility of green hydrogen production in Chile [17] with prices ranging from US$2.09/kg to US$3.28/kg. Against this backdrop, ambitious large-scale green electrolysis projects are being planned in resource-rich countries such as Australia, Namibia, Morocco, and Chile, often as stand-alone systems without grid connections. The concept of “energy islands” is also gaining traction in the North Sea, where hydrogen is planned to be produced in remote areas and transported via pipelines to the mainland [20]. Studies like the one by Semeraro et al. [21], have shown that hydrogen pipelines connecting hydrogen production centers are a cost-competitive method for the transmission of variable renewable energy compared to a high voltage direct current transmission lines.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Thus far no research has explored the contrasting costs and benefits of islanded versus integrated hydrogen systems. Our study aims to address this gap in the literature. We employ optimisation techniques to evaluate the deployment of integrated and islanded elec- trolysers in fully renewable energy systems in individual countries. By contrasting optimal configurations with baseline scenarios—in which either integrated or island electrolysers are built exclusively—we describe characteristic investment regimes tailored to different hydro- gen requirements. In addition, we consider the potential capital cost reductions in islanded systems due to simplified BoP configurations. For clarity and focus, our model has been streamlined to (1) consider only wind and solar energy sources, (2) limit the spatial loca- tions of renewable resources to two per country, and (3) use stylised demand profiles for electricity and hydrogen.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The paper is structured as follows: Section 2 describes the model setup and the model runs. Section 3 presents the results of our model runs. Section 4 discusses the results and their implications, followed by section 5 pointing out limitations. Finally, section 6 concludes the research.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	We employ a streamlined model using the open-source toolbox PyPSA [25] to analyse joint electricity and hydrogen production in individual countries. The model performs greenfield capacity expansion and hourly dispatch in systems based on wind and solar power, aiming to minimise annualised system costs. Each country is represented as two geographical locations with a fixed electricity and hydrogen demand (see Figure 1). The model incorporates wind and solar power, electrolysers, green fuel-powered dispatchable plants, short-term batteries, and large-scale underground hydrogen storage. Country-specific weather data for wind and solar generation are prepared using the open-source package “atlite” [26]. Technology costs projected for the year 2050 are retrieved from the Danish Energy Agency [27] (see Table A3 for a complete list of assumptions). The model is available online as open-source software [28].
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 1: System setup for each scenario used throughout this study: Integration (grey and blue elements), Island (grey and orange elements) and Optimisation (all elements). The two loads are located on the mainland. The hydrogen demand is met by the mainland and/or island hydrogen production, the electricity demand is met by the mainland grid only.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	and offshore island locations, and to include countries with island locations at longer and shorter distances from the mainland electricity grid. An overview of the selected countries is available in Table 1.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	We model each country’s energy system with three interconnected subsystems: one mainland electricity grid, one mainland hydrogen grid, and one hydrogen island (see Figure 1).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The mainland electricity grid serves electricity demand and can also power integrated electrolysers. Energy storage options on the mainland (batteries and hydrogen storage) and green fuel-powered combined cycle turbines help to align variable renewable production with demand profiles. Hourly capacity factors for wind and solar represent spatially aggregated values with weightings prioritising sites with good renewable resources, i.e. assuming renew- able resources are distributed within a country proportionally to the square of the average capacity factor.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The hydrogen island houses “islanded” electrolysers, which draw electricity only from dedicated renewable resources. Hydrogen produced on the island can be transported to the mainland via pipelines, with costs calculated based on capacity and distance (see Table A4). Slightly differing model designs were chosen to reflect country-specific circumstances: For Germany and Great Britain, offshore locations are chosen in the North Sea, where only offshore wind power is available. In Australia and Spain, pipelines are assumed to be above ground, whereas in Germany and Great Britain, they are modeled to be submarine with slightly higher costs, because of the offshore island locations and onshore portions running through densely populated areas.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The hydrogen pipeline connects hydrogen production with storage and consumption. It is designed to meet the hourly hydrogen demands either directly or by drawing from stockpiled hydrogen. All hydrogen must originally come from the mainland electricity grid or the hydrogen island.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 2: Geographical locations for the mainland (yellow) and island (blue) locations for each country considered in this study. The islands are located at sites with particularly good potential for renewable energy. Time-series for wind and solar power generation were obtained for each location using the atlite software [26].
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	We run our model for each country individually, and in three scenarios: (1) Integration: electrolysers are only built in the mainland electricity grid. (2) Island: electrolysers are only built on the island. (3) Optimisation: electrolysers can both be built in an integrated or in an islanded setting. The Optimisation scenario can utilise the advantages of both other scenarios, and its cost results thus represent a lower bound for the other two scenarios.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	demand, going from 100% electricity to 100% hydrogen. We assume the demand for each energy carrier is constant throughout the year, and we use a single hourly wind and solar time series for each location, aggregated over its area.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	As discussed above, producing hydrogen in an islanded setting can potentially provide cost benefits through simplified BoP equipment. We capture this aspect in the model by applying a uniform capital cost advantage to all power electronics elements in the islanded sub-system: capital costs for wind and solar power, electrolysers as well as the inverters used for battery storage. We then run each country, scenario, and hydrogen demand with six levels of these BoP-related capital cost advantages, ranging from 0% (no advantage for the islanded system) to 25% (see Table 2).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	As discussed in the following sections, the optimal deployment of mainland and island elec- trolysers can be classified into characteristic patterns for each country. Figure 3 sketches an idealised example of hydrogen production cost curves to showcase the classification of three optimal electrolyser investment strategies to support the analysis of our model results. We also identify two benefits from islanding electrolysers that determine the switch between these investment strategies, and we quantify the effect of these benefits on reducing hydrogen costs.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The idealised cost curves show the average hydrogen production costs as a function of the hydrogen-to-electricity share of final energy demand for the three different scenarios: Integration, Island, and Optimisation. The costs of hydrogen on the island are unaffected by the electricity production on the mainland and thus remain constant. The costs of hydrogen in the Integration Scenario rise, as the availability of otherwise-curtailed electricity decreases. Due to its higher degrees of freedom, the Optimisation scenario’s costs are always equal to or below the costs of the Integration and Island scenarios.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 3: Idealised hydrogen cost curves for all three scenarios as a function of the hydrogen share of final energy demand. The curves show three investment regimes for the Optimisation scenario: The Integration Regime (left), where electrolysers are optimally deployed on the mainland; the Hybrid Regime (middle), where additional electrolysers are deployed both on the mainland and on the island; and the Island Regime (right), where all additional electrolysers are deployed on the island.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	electrolysers are the cheapest source of additional hydrogen, and thus the Island Regime is characterised by the deployment of all additional electrolysers on the island.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	In this section, we explore the economics of hydrogen production for the selected set of coun- tries. We delve into the key drivers that influence the cost-effectiveness of both integrated and islanded electrolysers. We identify the progression of investment regimes that dictate each country’s electrolyser deployment behaviour. Lastly, we investigate how reductions in capital costs for islanded systems can alter these optimal electrolyser investment strategies.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 4: Average hydrogen costs for all countries in all three scenarios as a function of the hydrogen share of final energy demand. The hydrogen costs in the Optimisation scenario (green) are always equal to or lower than the hydrogen costs of the Integration (blue) and Island (orange) scenarios. In contrast to Spain and Germany, where the system reaches the Island and Hybrid Regime, respec- tively, for Great Britain and Australia the deployment of only integrated electrolysers (Integration Regime) is cost-optimal for all shares of hydrogen demand.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 4 displays the average hydrogen costs for each country and scenario as a function of the hydrogen-to-electricity ratio in final energy demand. By model design, the hydrogen costs for the Island scenario (orange) remain constant across different hydrogen demand shares and are determined by the quality of the renewable resources at the island and the pipeline costs. Between the countries, hydrogen costs for the Island scenario vary from 58€/MWh in Australia to 71€/MWh in Great Britain. In contrast, the hydrogen costs in the Integration scenario (blue) increase with hydrogen share, starting at their lowest values with 0% hydrogen share, initially rising steeply before levelling off at round 30 to 50% hydrogen share, asymptoting towards the cost of hydrogen at 100% hydrogen share.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	We now compare the Island and Integration scenarios. At 100% hydrogen share, the In Island scenario has lower costs than the Integration scenario for Spain and Germany. both countries, the two scenarios reach cost equality at 40% hydrogen share. For Australia, despite superior island resources (see Table 1), pipeline costs render hydrogen production on the island more expensive even for 100% hydrogen share. In Great Britain, the island scenario has better wind resources but lacks solar power, making the average hydrogen costs on the island more expensive than on the mainland. Pipeline costs also contribute to a small degree but are not solely responsible for the cost difference.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Finally, the hydrogen costs in the Optimisation scenario (green) are by design always equal to or below the hydrogen costs of the Integration and Island scenarios. Consequently, for Great Britain and Australia, the Optimisation scenario always deploys the same system as the Integration scenario. For these two countries, the system stays in the Integration Regime and it is always cost-optimal to invest into integrated electrolysers. In contrast, the optimal deployment pattern for electrolysers in Spain and Germany reaches a tipping point: For Spain, the system moves at 40% hydrogen share from the Integration Regime directly into the Island Regime, where it is cost-optimal to deploy all additional electrolysers on the island. For Germany, the system enters the Hybrid Regime at around 5% hydrogen share, where both mainland and island electrolysers are deployed, and remains in this regime until 100% hydrogen share.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	In the following, we delve into the unique characteristics of each regime, we identify two benefits that make islanded electrolysers more cost-effective than integrated electrolysers, and we quantify the impact of these benefits on the average hydrogen costs for each country.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	First, we analyse the Integration Regime which is characterised by the fact that only in- tegrated electrolysers are deployed to accommodate the rising demand for hydrogen. The Integration Regime is prevalent in Spain from 0 to 40%, in Germany from around 0 to 5% hydrogen share, and in Great Britain and Australia for the entire range of hydrogen demand shares.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	that reduce both hydrogen and electricity production costs. These co-production synergies mostly accrue early for smaller shares of hydrogen, with average hydrogen costs quickly con- verging to the cost of electrolysis with no co-production of electricity (see hydrogen costs for the Integration scenario in Figure 4).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The root of this synergy is the utilisation of previously curtailed energy for hydrogen production. Evidence for this is the decrease in curtailment quantities in the Integration Regime for the Integration and Optimisation scenarios, as highlighted in Figure 5. However, it is essential to note that electrolysers are not capable of utilising all the available curtailed energy. Each country exhibits one cost-optimal level of curtailment for electricity production (see 0% hydrogen share in Figure 5) and a different cost-optimal level of curtailment for hydrogen production (see 100% hydrogen share in Figure 5). Both levels of curtailment are determined by each country’s renewable resources. The difference between these optimal curtailment levels delineates the quantity of curtailed energy that can feasibly be directed toward electrolysers as hydrogen demand increases. For example, whilst Spain experiences high levels of curtailment—up to 30% of all generated electricity—only about 8 percentage points of this curtailment are mitigated through the deployment of integrated electrolysers.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 5: Average Curtailment Rate for countries in all three scenarios as a function of the hydrogen share of final energy demand. The Curtailment Rate is defined as the ratio of curtailed energy to total energy demand. Across all countries and scenarios, the electrolyser deployment in the Optimisation scenario (green) reduces the curtailment rate. The extent of reduction is determined by synergies between electricity and hydrogen production on the mainland, and the quality of renewable resources at the mainland and the island.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The Hybrid Regime is characterised by the deployment of electrolysers both on the mainland and on the island to meet additional hydrogen demand. This regime is prevalent in Germany from 10 to 100% hydrogen share (see Figure 6). The Integration Regime gives way at the “Competitiveness Point” (CP), where islanded electrolysers become cost-competitive with integrated electrolysers for the first time. It can also be identified as the hydrogen share where the first islanded electrolyser is deployed in each country (see Figure 6).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 6: Installed electrolyser capacities for all countries in the Optimisation scenario as a function of the hydrogen share of final energy demand. In the first regime (Integration Regime), the system only deploys integrated electrolysers (purple). Once the system moves to the Hybrid Regime, as is the case in Germany at 10%, integrated and islanded electrolysers (yellow) are used in parallel. The Island Regime, which starts in Spain at 40%, is characterised by additional electrolysers being deployed only on the island.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	synergies between the different renewable resource profiles on the mainland and on the island to produce hydrogen in both locations. We call this the Portfolio Benefit.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The synergies responsible for the Portfolio Benefit stem from two sources: The first is a reduced requirement for hydrogen storage, as corroborated by a comparison of the specific hydrogen storage cost components between the Integration and Optimisation scenarios for Germany, depicted in Figure 7. The time series of renewable energy production at both locations align in such a way that hydrogen production schedules throughout the year com- plement each other, thereby minimising the need for hydrogen storage.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	In the Optimisation scenario, Germany deploys electrolysers on the island which run 100% of the time that wind power is available. This leads to a capacity factor of 65% for islanded elec- trolysers compared to about 45% for integrated electrolysers in the Integration scenario (see Figure 8). Islanded electrolysers operate at full capacity (or at least whenever wind energy is available), whilst integrated electrolysers selectively operate during periods of zero-cost (i.e., using curtailed energy) or low-cost electricity and/or when their operation can lead to hydrogen storage reduction. The capacity factor for integrated electrolysers remains also around 45% such that the average capacity factor for the entire electrolyser fleet in the Op- timisation scenario is larger than in the Integration scenario (at least for the range from 10 to 60% hydrogen share).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The value of the Portfolio Benefit is measured as the difference between the hydrogen costs of the Integration and Optimisation scenarios while in the Hybrid Regime (see Figure 4). The only country where the Portfolio Benefit is active is Germany, and at the maximum it reduces hydrogen costs by 21% (15€/MWh) compared to the Integration scenario.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 7: Hydrogen cost breakdown by cost components for the Integration (top row) and Op- timisation (bottom row) scenarios for Spain and Germany. Within the Integration Regime, the two scenarios result in the same setup and cost breakdown. While the cost breakdown does not differ much in the Island regime (left side, Spain), a strong difference can be seen in the Hybrid regime (right side, Germany). Here, the Optimisation scenario benefits from synergies between the mainland and the island, which reduce the need for hydrogen storage and decrease electrolyser investment costs.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	electrolysers start low, but quickly rise as hydrogen demand increases (see Figure 4). And thirdly, when the resource portfolio from the mainland and island locations fits well together to produce hydrogen, the Portfolio Benefit reduces hydrogen costs for the Optimisation scenario compared to the Integration scenario (see Germany starting at 5% hydrogen share in Figure 4).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The Island Regime is defined as the regime under which additional hydrogen demand is solely met by deploying islanded electrolysers. Here, we see that the islanded electrolyser outcompetes the integrated electrolyser due to its better resources. We call this the Supe- riority Benefit for islanded electrolysers. The Superiority Benefit is also quantified as the difference between the hydrogen costs of the Integration and Optimisation scenarios, but only applies under the Island Regime (see Figure 4). The transition to the Island Regime occurs at the “Winout Point” (WP), and this is where the Superiority Benefit outweighs the Portfolio Benefit. Any integrated electrolysers deployed prior to reaching this regime remain in the system at a constant rate relative to the declining electricity demand as hydrogen demand increases.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 8: Electrolysis capacity factors for Spain and Germany for integrated and islanded electrol- ysers in all scenarios. Whereas the capacity factors in Spain stay relatively constant, the capacity factors for integrated electrolysers in the Optimisation scenario (purple) in Germany decrease with the hydrogen share, in favour of maximum islanded hydrogen production (yellow).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	exists only for Spain, starting at 40% hydrogen share, and decreases hydrogen costs by up to 7% (5€/MWh) compared to the Integration scenario.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	In addition to the dynamics of competition between integrated and islanded electrolysers, our study also includes the influence of lower capacity costs, from hereon denoted as cost advantage, for relevant system components on the hydrogen island. On an island, power electronic costs could be saved because grid-level power quality no longer needs to be main- tained, or the conversions to alternating current could be avoided altogether [23]. In this section, we exemplarily discuss the results for Great Britain (see Figure 9).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	We still observe the same dynamics, trade-offs and synergies between integrated and islanded electrolysers, but the locations of the regimes shift. In addition, incorporating cost advantage leads to lower hydrogen costs from islanded electrolysers (see decreasing level of orange curves in Figure 9, as cost advantage increases). As a result for Great Britain, the CP and WP are gradually introduced (CP at 5%, WP at 20% cost advantage), and then shift left as cost advantage increases further. At 5% cost advantage, the decrease in hydrogen costs from islanded electrolysers is not yet sufficient to outright compete with the synergies between integrated electrolysers and the power system. However, some small amount of Portfolio Benefit exists and barely makes the hybrid deployment of integrated and islanded electrolysers beneficial for the overall system.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 9: Hydrogen costs for Great Britain at 0% to 25% capital cost advantage for all three scenarios. The capacity cost advantages pull the costs of islanded hydrogen production down, and thus shift the start of the Hybrid Regime, and the start of the Island Regime to the left.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	at 80% hydrogen share. From hereon, the system deploys all additional electrolysers on the island.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The two benefits identified in our study also change with increasing cost advantage. For example, for Great Britain without cost advantage there is no Superiority Benefit to islanded electrolysers, because renewables on the mainland are actually the superior resource. With 25% cost advantage, the island becomes the superior resource, and applies a benefit of up to 19% lower hydrogen costs (12€/MWh) to islanded electrolysers. Net shifts in favour of islanded electrolysers are similar for Australia with 20% (11€/MWh) lower hydrogen costs, but even larger for Spain and Germany (33% and 37% lower costs, respectively, see Figure A11).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	The Portfolio Benefit also increases significantly as cost advantage increases. For Great Britain, the maximum Portfolio Benefit at 5% cost advantage amounts to 3% (2€/MWh, at 65% hydrogen share), and at 25% cost advantage to 21% (13€/MWh, at 50% hydrogen share).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Finally, it is worth noting that the engineering-based cost advantages contribute to the same effect as the resource advantage of the island over the mainland. Essentially, an island location with strong and favourable renewable resources would exhibit the same boost to the competitiveness of islanded electrolysers as a less favourable location with reduced capital costs.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	renewable electricity system on the mainland and allow otherwise-curtailed electricity to be used. As the share of hydrogen in final energy demand and thus also the amount of inte- grated electrolysers on the mainland increase, Portfolio Benefits can make islanded electrol- ysers competitive. For countries with a high domestic hydrogen demand, and those looking to export hydrogen, the optimal system deployment can include the bulk of electrolysers operating in an islanded setting, especially if islanded hydrogen production yields some of the capital cost advantages studied in this paper. In the following, we briefly discuss these findings and focus on their implications for system planners as well as additional regulatory and technology issues.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	From a system planning perspective, two additional quantities have to be estimated in order to formulate an optimal investment strategy for integrated versus islanded electrolysers: the domestic hydrogen demand and the expected world market price for hydrogen. Assuming these as given, we show, on the basis of Figure 10, how to determine (1) the optimal amount of domestically produced hydrogen (and how much to import or export accordingly to meet domestic hydrogen use), (2) whether to deploy integrated or islanded electrolysers, and/or at which respective shares, and (3) how to unfold the electrolyser deployment over time as hydrogen demand ramps up.
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	To determine the optimal amount of domestically produced hydrogen (1), system plan- ners need to construct their domestic hydrogen production price curve based on the optimal investment regimes for integrated and islanded electrolysers, as we deduce with the Optimi- sation scenario in our paper. The intersection with the world market price then determines the optimal domestic hydrogen production level (see Figure 10).
Processed_Benefits_from_Islanding_Green_Hydrogen_Production.txt	Figure 10: Idealised cost curves to determine domestic hydrogen production and import/export volumes, as well as deployment patterns for electrolysers. The intersection of the domestic hydrogen production costs from the Optimisation scenario (green) and the world market price (grey) determine the domestic hydrogen production level. The intersection of the domestic hydrogen demand and the lower of both price curves determine the level of domestic hydrogen use. The difference between domestic hydrogen production and hydrogen use is balanced by imports from or exports to the world market.

Subsets and Splits