Reactivity of Xenon with Ice at Planetary Conditions
Chryste `le Sanloup*
UPMC Univ Paris 06, UMR CNRS 7193, ISTEP, 75005 Paris, France
Stanimir A. Bonev
Lawrence Livermore National Laboratory, Livermore, California 94550, USA and Department of Physics,
Dalhousie University, Halifax, Nova Scotia B3H 3J5, Canada
Universite ´ Paris-Est, Laboratoire Mode ´lisation et Simulation MultiEchelle, MSME UMR 8208 CNRS,
77454 Marne-La-Valle ´e, France
Helen E. Maynard-Casely†
School of Chemistry, University of Edinburgh, United Kingdom
(Received 23 December 2012; published 24 June 2013)
We report results from high pressure and temperature experiments that provide evidence for the
reactivity of xenon with water ice at pressures above 50 GPa and a temperature of 1500 K—conditions
that are found in the interiors of Uranus and Neptune. The x-ray data are sufficient to determine a
hexagonal lattice with four Xe atoms per unit cell and several possible distributions of O atoms. The
measurements are supplemented with ab initio calculations, on the basis of which a crystallographic
structure with a Xe4O12H12primitive cell is proposed. The newly discovered compound is formed in the
stability fields of superionic ice and ?-O2, and has the same oxygen subnetwork as the latter. Furthermore,
it has a weakly metallic character and likely undergoes sublattice melting of the H subsystem. Our
findings indicate that Xe is expected to be depleted in the atmospheres of the giant planets as a result of
sequestration at depth.
DOI: 10.1103/PhysRevLett.110.265501PACS numbers: 62.50.?p, 61.05.cp, 96.15.Bc, 96.30.Pj
Knowledge of the chemistry of xenon (Xe) with plane-
tary materials under high pressure P and temperature T
conditions is a prerequisite for understanding the abun-
dance of Xe and its isotopic ratios in the atmospheres of
giant planets. The noble gases Ne, Ar, Kr, and Xe are
indeed among the most critical heavy elements , as their
abundancesconstrain the modelsforgiant planetformation
and the origin of their atmospheres. The field of Xe chem-
istry at ambient P was initiated with the synthesis of
fluoride salts  and later extended to H, C, O, N, S, other
halogens, and even metals (Au, Hg), so that close to a
hundred Xe compounds are currently known [3,4].
However, the use of pressure as a means to enforce Xe to
bond to other elements has seldom been investigated .
Geophysicists first started to explore this track as a poten-
tial explanation for the observed Xe deficiency in the
terrestrial and Martian atmospheres . Xe-Fe compounds
were shown to be unstable up to 70 GPa , although up to
0.8 mol% Xe could be alloyed to Fe at 300 GPa . It was
later proposed that Xe depletion from Earth’s atmosphere
could occur by Xe substitution for Si in the silicates
network under the P-T conditions of the deep crust and
mantle, therefore forming Xe-doped oxides [9–11],
a synthesis shortly followed by that of pure XeO2at
ambient P .
Water ice is one of the most prevalent substances in the
Solar System, with the majority of it existing at high P and
T conditions in the interiors of giant planets. Xe is among
the gases that stabilizes clathrate hydrate structures
through van der Waals interactions. However, Xe hydrates
are stable only up to 2.5 GPa, before dissociating into Xe
plus ice VII . The chemistry of Xe with water has
nonetheless been successfully explored by UV photolysis
of water in solid Xe at ambient pressure. Among the
compounds hitherto obtained are HXeOH  and, the
latest one to date, HXeOXeH , which results from
the insertion of Xe atoms in the water molecule. These
results reopened the perspective to synthesize covalent
compounds in the Xe-H2O system by applying extreme
In this Letter we report the first experimental evidence
for the reactivity of Xe with water ice at conditions found
in the interior of giant planets. The resulting crystalline
structure is resolved from x-ray diffraction data with addi-
tional input from ab initio calculations. Our analysis indi-
cates the participation of H in the structure formation, and
brackets for the H content are given with an emphasis on
High P-high T conditions were produced in laser-heated
diamond-anvil cells and x-ray diffraction experiments
were carried out in situ on the ID27 beam line of the
PRL 110, 265501 (2013)
28 JUNE 2013
? 2013 American Physical Society
ESRF. A ring was laser cut in a 2:5 ?m thick platinum (Pt)
foil and inserted in a 80 ?m hole drilled in a rhenium
gasket on top of two ruby spheres, with Pt acting as a
coupler with the IR laser. Milli-Q deionised water was then
loaded in the sample chamber, and Xewas added cryogeni-
cally in a N2atmosphere. The sample did not contain N2
after closing the cell as attested by the absence of the N2
vibron in Raman spectra. The pressure was measured from
the ruby fluorescence signal , and the temperature was
calculated from the thermal equation of state of Pt .
Temperatures were also measured from the sample thermal
emission spectra. However, volumes measured from the
x-ray diffraction patterns show that the sample T was lower
than the surface T as previously shown when a metallic
laser coupler is used . Data were collected on a MAR-
CCD using a monochromatic x-ray beam (? ¼ 0:3738?A)
focused to a 3 ? 4 ?m area on the sample. We carried out
experiments at P ranging from 17 to 80 GPa and T below
the melting point of water ice.
During laser heating at 50 GPa, we observed two phases
above 1500 K in addition to Xe, Pt, and bcc ice and will
refer to them as phase-1 and phase-2 (see Fig. 1). X-ray
diffraction peaks from phase-1 can be assigned to a face-
centered cubic structure. Phase-1 is not quenchable back to
room T and is equally observed in blank experiments on
Pt-water mixtures. No blank experiment was conducted on
the Xe-Pt system as it has been experimentally explored at
P-T conditions covering those of the present work and no
reaction was observed . Phase-2 is stable back to
and was systematically synthesized upon
heating up to 80 GPa, the maximum P investigated here.
The x-ray diffraction signal from phase-2 is indexed by a
hexagonal cell with a ¼ b ¼ 5:0539 ? 0:0003?A and
c ¼ 8:210 ? 0:001?A at 58 GPa and 1500 K. Its com-
pressibility as obtained from a second-order Birch-
Murnaghan equation of state is 77 ? 5 GPa at room T
(fitted with zero pressure volume parameter V0¼
255:8?A3), and 67 ? 5 GPa at 1500 K (V0¼ 269:8?A3).
The thermal expansion coefficient at high P is ð1:8 ?
0:1Þ ? 10?5K?1(see Fig. 2).
The x-ray diffraction pattern at 58 GPa and 1500 K has a
sufficiently high powder-quality pattern for the atomic
positions of the Xe and oxygen (O) to be refined;
P63=mmc is the highest-symmetry space group allowed.
From density considerations, four atoms of Xe can be
accommodated in the unit cell. The solution is however
not unique for the O atoms, with three possibilities equally
well satisfying the x-ray pattern intensities. The three
possible structures are with either 12, 14, or 16 O atoms
per unit cell. In order to determine the correct solution for
O atoms and to investigate the presence of hydrogen in the
structure, we have carried out a theoretical analysis.
First principles density functional theory calculations
were carried out with VASP  using projector augmented
wave pseudopotentials with 8 and 6 valence electrons for
the Xe and O atoms, respectively, and a 500 eV planewave
cutoff. Full structural optimizations were performed with a
uniform 6 ? 6 ? 6 k-point grid sampling of the Brillouin
zone, which ensures the desired degree of convergence
(meV=atom). The choice of exchange-correlation func-
tional can lead to significant variations in the specific
volume of the Xe-O-H systems being studied. In order to
provide an estimate for the theoretical uncertainties
originating from the exchange-correlation approximation,
we have carried out calculations within the local
density approximation, the Perdew-Burke-Ernzerhof 
2-θ angle (degrees)
Intensity (arb. units)
H2O-Pt run 62 GPa
Xe-H2O-Pt run 58 GPa
300 K quench
300 K quench
after heating a Xe þ H2O þ Pt sample (black lines) and upon
and after heating a H2O þ Pt sample (gray lines). Circles in-
dicate peaks from phase-1 and asterisks indicate peaks from
X-ray diffraction patterns. Spectra obtained upon and
4050 6070 80
Experimental data, room T
id., 1,500 K
Cell parameters (Å)
FIG. 2 (color online).
volume (left) and cell parameters (right). Solid curves are a
second-order Birch-Murnaghan fit to the data. Theoretical cal-
culations: squares are results obtained with the PBEsol func-
tional, the error bars extend from results obtained using the local
density approximation (lower bar) to results within the PBE
Pressure dependence of the unit cell
PRL 110, 265501 (2013)
28 JUNE 2013
generalized gradient approximation (PBE), and the PBE
revised for solids  (PBEsol). Additionally, for selected
structures van der Waals interactions were included using
the vdW-D2 method  as implemented in
However, this did not change the results.
Given the fact that the newly synthesized structure is
thermodynamically stable only at high T, as well as its
complexity, we have not attempted to perform a complete
phase space crystalline search. Instead, we have limited the
theoretical analysis to within the constraints imposed by
the experimental data. Starting from the experimental lat-
tice parameters and refined atomic positions, structural
relaxations were initially performed on systems with Xe
and O atoms only. The optimizations with 16, 14, and 12 O
atoms resulted in theoretically stable structures that differ
significantly from the experimental fits—respectively
þ1%, ?7%, and ?5% in the a lattice parameter, and
þ6%, þ8%, and þ9% in the c lattice parameter (within
the PBE). The internal O coordinates also change signifi-
cantly. These structural differences, which are accompa-
nied by large energy differences on the order of tens of
eV=atom, are beyond the uncertainties of both the theo-
retical and experimental methods. The fact that no H-free
structure could be identified is in agreement with a recent
theoretical investigation  where no thermodynamically
stable xenon oxides were found below 83 GPa.
Analysis for the effect of hydrogen was carried out by
initially placing H atoms into the system both at random
positions and at selected high-symmetry points. This was
followed by full structural optimization at 58 GPa. To
ensure that the entire phase space for H positions had
been sampled, we also performed selective molecular dy-
namics where the H subsystem was heated while Xe and O
atoms were held fixed. This procedurewas repeated with 4,
12, and 24 H atoms. For the structures with 16 and 14
O atoms per unit cell, adding H atoms increases even
further the discrepancy with the experimental fits.
Therefore, the structures with 16 and 14 O atoms can be
ruled out conclusively. On the other hand, introducing H in
calculations with 12 O atoms per unit cell yields structures
that match the x-ray data. The best fitting result is for 12
H atoms, for which the finite temperature lattice parame-
ters, optimized to give an isotropic stress tensor at each
P–300 K point, are compared with the measured ones
(see Fig. 2). Using 4 H atoms, the calculated structure
has an excellent agreement with the measured volume
(see Fig. 2); however, the H coordinates cannot be repro-
duced within any hexagonal space group. The structures
with 24 H atoms do not give satisfactory agreement with
the experimental cell parameters. It is nonetheless difficult
to determine theoretically the exact H content—between 4
and 12 atoms per unit cell. In fact, it may be varied due the
diffusive character of H at high temperature as described
thermodynamic stability analysis at 1500 K, which goes
beyond the scope of this Letter.
The optimized Xe and O atomic positions further refined
for the Xe4O12H12structure by a Rietveld fit to the data
(see Fig. 3) are a 12(k) site (x, y, z) for O with x ¼ 0:157 ?
0:002, y ¼ 0:314 ? 0:003, z ¼ 0:622 ? 0:002, and a 4(f)
site (1=3, 2=3, z) for Xe with z ¼ 0:071 ? 0:001. The H
atoms,according tothetheoreticalcalculations,are located
on a 12(j) site (x, y, 1=4) with x ¼ 0:9427 ? 0:002, y ¼
0:2886 ? 0:004. The structure consists of two Xe2O6H6
units per unit cell (see the inset of Fig. 3), and the Xe-O
distance is 2:21 ? 0:01?A. This distance is similar to that
observed in the HXeOH and HXeOXeH molecules, and
equal to 2.208 A˚and 2.15 A˚, respectively [14,15].
In order to examine the finite temperature properties of
the Xe4O12H12structure, we performed density functional
theory molecular dynamics simulations on supercells con-
taining 224 and 504 atoms and ?-point sampling of the
Brillouin zone. While the simulations with the two cell
sizes give very close results, we determined that a denser
k-point grid is required for well converged results.
Nevertheless, it is worth noting the observed tendency of
the H subsystem to become diffusive at finite T. Its self-
diffusion coefficient is estimated to be 5?A2ps?1at 60 GPa
and 1500 K. The diffusivity of H is similar to what is
observed in H2O at these conditions, except that while
water is superionic Xe4O12H12is metallic. The electronic
properties of the latter, computed within the PBEsol,
are shown in Fig. 4. Notice that the electronic states near
the Fermi level are due to the Xe and O atoms, so that the
metallic character would be preserved even if H undergoes
subatomic melting at high T.
4.0 6.0 8.0 10.0 12.0 14.0
FIG. 3 (color online).
at 58 GPa and 1500 K. Vertical ticks indicate diffraction peaks
from phase-2 (blue) and from Xe (black). The pattern has been
cut between 9.4 and 12.3 degrees, as the signal in this region is
overloaded by Pt and water ice diffraction peaks which would
prevent an accurate refinement of the less intense Xe4O12H12
peaks. Insets: structure of the Xe2O6H6unit (two per cell), and
inner half of the corresponding 2D diffraction pattern. [The ring
at 10 degrees is the (111) Pt peak; all rings inwards are from
either phase-2 or Xe].
Rietveld fit of the Xe4O12H12structure
PRL 110, 265501 (2013)
28 JUNE 2013
No oxygen phase was detected in the x-ray diffraction
patterns. However, the Xe4O12H12structure we propose
may remarkably be obtained by the interpenetration of the
high-P Xe and ?-O2hcp structures . The O atoms in
the unit cell of the newly discovered phase can be matched
by summing four ?-O2unit cells expanded by 25% along
the c axis, i.e., along the intramolecular bond, resulting
within 1% in an identical atomic volume for O between
both phases. ?-O2has been described to be stable below
the O2melting line between 15 and 20 GPa , but recent
experiments carried out at much higher P evidenced an-
other domain of stability just below the melting line above
44 GPa . The stability field of the new Xe compound as
mapped in this study thus coincides with that of high P
?-O2. In contrast to other O2solid phases, ?-O2is char-
acterized by its orientational disorder and a high degree
of charge transfers that provides an explanation for the
relatively short Xe-O distance in our present phase. It is
also interesting to note that the Xe4O12H12structure bears
similarities to the hypothetical clathrate V structure
proposed  for 4X-8Y-68H2O, where X ¼ CCl4and
Y ¼ Xe, with partial occupancy for the latter. Both are
hexagonal with the P63=mmc space group, an ...ABAB...
stacking sequence, and a c=a ratio of 1.61.
Although theoretical calculations reproduced the experi-
mental results without considering Pt, its role in the
energetics of the reaction has to be considered as its
hydrogenation is an exothermic reaction . A PtH
x-ray diffraction signal  was observed locally upon
heating at 40 GPa, a P value lower than that of Xe reac-
tivity. It is therefore not possible to decipher between
newly formed and inherited PtH from previous heating
cycles at lower P. PtH was never observed simultaneously
with the Xe compound (see Fig. 1), but that could be due to
the very small x-ray beam (3 ? 4 ?m) and high diffusion
rate of hydrogen. In addition, Raman spectra collected
after heating showed no sign of a H2vibron. We therefore
propose the following global reaction:
2Xe þ 6H2O þ ð12 ? x=2ÞPt
! Xe2O6Hx=2þ ð12 ? x=2ÞPtH:
The reaction must have a negative ?V=V value to be
pressure induced, which further brackets the H content
between 8 (?V=V ¼ ?0:2% per mole) and 12 (?V=V ¼
?2:2% per mole) atoms per cell.
Water ice and Xe are consequently expected to react at
P-T conditions higher than 50 GPa and 1500 K. These
conditions are gathered inside giant planets, and particu-
larly within Uranus and Neptune that contain a larger
proportion of planetary ices. These conditions are also
gathered within subducted slabs in the deep Earth, pro-
vided that sea-water carries Xe within the deep mantle
. The observed enrichment of the Jovian atmosphere
in noble gases by a factor of 2 compared to the solar
abundances has been explained by different scenarios
including their clathration and/or absorption on water ice
in preplanetary objects and their further release into the
planet’s atmosphere . However, Xe abundance is over-
estimated in these scenarios. Any reactivity with Xe at
depth would explain this mismatch and future probe mis-
sions to Saturn , Uranus, and Neptune could provide a
test of the present results. These planetary implications
also fit into the wider context of noble gases sequestration
at depth inside giant planets as demonstrated for Ne in the
interiors of Jupiter .
We acknowledge the ESRF for provision of beam time
on ID27 and LLNL for computational resources. We thank
M. Mezouar and E. Gregoryanz for their help with collect-
ing in situ x-ray diffraction data, and Y. Noel and M.
Marques for useful discussions. C.S. is funded by the
Community’s Seventh Framework Programme (Grants
No. FP7/2007-2013 and No. 259649). S.A.B. performed
work at LLNL under the auspices of the U.S. Department
of Energy under Grant No. DE-AC52-07NA27344.
*Present address: School of Physics and Astronomy and
Center for Science at Extreme Conditions, University of
Edinburgh, Edinburgh EH9 3JZ, United Kingdom.
†Present address: Bragg Institute, Australian Nuclear
Science and Technology Organisation, Menai, New
South Wales 2234, Australia.
 T. Guillot, Annu. Rev. Earth Planet Sci. 33, 493 (2005).
 N. Bartlett, Proc. Chem. Soc. 218 (1962).
 W. Grochala, Chem. Soc. Rev. 36, 1632 (2007).
 R.B. Gerber, Annu. Rev. Phys. Chem. 55, 55 (2004).
EDOS (states per atom per eV)
FIG. 4 (color online).
Xe4O12H12computed at 60 GPa within the PBEsol. The total
EDOS are shown with a solid line. The dashed lines indicate
partial EDOS projected on different atomic species (as indicated
in the figure legend). The energy scale is shifted so that the Fermi
level is at 0.
Electronic density of states (EDOS) of
PRL 110, 265501 (2013)
28 JUNE 2013
 M. Somayazulu, P. Dera, A.F. Goncharov, S.A. Gramsch, Download full-text
P. Liermann, W. Yang, Z. Liu, H.k. Mao, and R.J.
Hemley, Nat. Chem. 2, 50 (2009).
 E. Anders and T. Owen, Science 198, 453 (1977).
 W.A. Caldwell, J.H. Nguyen, B.G. Pfrommer, F. Mauri,
S.G. Louie, and R. Jeanloz, Science 277, 930 (1997).
 K.M. Lee and G. Steinle-Neumann, J. Geophys. Res. 111,
 C. Sanloup, B.C. Schmidt, E.C. Perez, A. Jambon, E.
Gregoryanz, and M. Mezouar, Science 310, 1174 (2005).
 M.I.J. Probert, J. Phys. Condens. Matter 22, 025501
 C. Sanloup, B.C. Schmidt, G. Gudfinnsson, A. Dewaele,
and M. Mezouar, Geochim. Cosmochim. Acta 75, 6271
 D.S. Brock and G.J. Schrobilgen, J. Am. Chem. Soc. 133,
 C. Sanloup, H.-K. Mao, and R.J. Hemley, Proc. Natl.
Acad. Sci. U.S.A. 99, 25 (2002).
 M. Pettersson, J. Lundell, and M. Ra ¨sa ¨nen, Eur. J. Inorg.
Chem. 1999, 729 (1999).
 L. Khriachtchev, K. Isokoski, A. Cohen, M. Ra ¨sa ¨nen, and
R.B. Gerber, J. Am. Chem. Soc. 130, 6114 (2008).
 H.K. Mao, J. Xu, and P.M. Bell, J. Geophys. Res. 91,
 C.-S. Zha, K. Mibe, W.A. Bassett, O. Tschauner, H.-K.
Mao, and R.J. Hemley, J. Appl. Phys. 103, 054908
 J.F. Lin, M. Santoro, V.V. Struzhkin, H.K. Mao, and R.J.
Hemley, Rev. Sci. Instrum. 75, 3302 (2004).
 G. Kresse and J. Hafner, Phys. Rev. B 47, 558 (1993); G.
Kresse and X. Furthmuller, Comput. Mater. Sci. 6, 15
 J.P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett.
77, 3865 (1996).
 J.P. Perdew, A. Ruzsinszky, G.I. Csonka, O.A. Vydrov,
G.E. Scuseria, L.A. Constantin, X. Zhou, and K. Burke,
Phys. Rev. Lett. 100, 136406 (2008).
 S. Grimme, J. Comput. Chem. 27, 1787 (2006).
 Q. Zhu, D.Y. Jung, A.R. Oganov, C.W. Glass, C. Gatti,
and A.O. Lyakhov, Nat. Chem. 5, 61 (2013).
 L.F. Lundegaard, C. Guillaume, M.I. McMahon, E.
Gregoryanz, and M. Merlini, J. Chem. Phys. 130,
 M. Santoro, E. Gregoryanz, H.-K. Mao, and R.J. Hemley,
Phys. Rev. Lett. 93, 265701 (2004).
 A.F. Goncharov,T.R.R.N.
J. Chem. Phys. 135, 084512 (2011).
 E.A. Smelik and H.E. King, Z. Kristallogr. 211, 84
 L. Visscher, T. Saue, W.C. Nieuwpoort, K. Faegri, and O.
Gropen, J. Chem. Phys. 99, 6704 (1993).
 T. Scheler, O. Degtyareva, M. Marques, C.L. Guillaume,
J.E. Proctor, S. Evans, and E. Gregoryanz, Phys. Rev. B
83, 214106 (2011).
 G. Holland and C.J. Ballentine, Nature (London) 441, 186
 F. Hersant, D. Gautier, and J.I. Lunine, Planet. Space Sci.
52, 623 (2004).
 B. Marty, T. Guillot, A. Coustenis, N. Achilleos, Y.
Alibert, S. Asmar, D. Atkinson, S. Atreya, G. Babasides,
K. Baines, T. Balint, D. Banfield, S. Barber, B. Bezard,
G.L. Bjoraker, M. Blanc, S. Bolton, N. Chanover, S.
Charnoz, E. Chassefiere, J.E. Colwell, E. Deangelis, M.
Dougherty, P. Drossart, F.M. Flasar, T. Fouchet, R.
Frampton, I. Franchi, D. Gautier, L. Gurvits, R. Hueso,
B. Kazeminejad, T. Krimigis, A. Jambon, G. Jones, Y.
Langevin, M. Leese, E. Lellouch, J. Lunine, A. Milillo, P.
Mahaffy, B. Mauk, A. Morse, M. Moreira, X. Moussas, C.
Murray, I. Mueller-Wodarg, T.C. Owen, S. Pogrebenko,
R. Prange, P. Read, A. Sanchez-Lavega, P. Sarda, D. Stam,
G. Tinetti, P. Zarka, J. Zarnecki, and Kronos Consortium,
Exp. Astron. 23, 947 (2009).
 H.F. Wilson and B. Militzer, Phys. Rev. Lett. 104, 121101
PRL 110, 265501 (2013)
28 JUNE 2013