A combined theoretical and experimental study of lithium palladium deuteride (Li2PdD2) subjected to pressures up to 50 GPa reveals one structural phase transition near 10 GPa, detected by synchrotron powder x-ray diffraction, and metadynamics simulations. The ambient-pressure tetragonal phase of Li2PdD2 transforms into a monoclinic C2/m phase that is distinct from all known structures of alkali metal–transition metal hydrides/deuterides. The structure of the high-pressure phase was characterized using ab initio computational techniques and from refinement of the powder x-ray diffraction data. In the high-pressure phase, the PdD2 complexes lose molecular integrity and are fused to extended [PdD2] chains. The discovered phase transition and new structure are relevant to the possible hydrogen storage application of Li2PdD2 and alkali metal–transition metal hydrides in general.

Transition metal hydrides have been actively investigated for their rich physical properties and growing applications in industry. Palladium (Pd), in particular, has an exceptional capability of absorbing hydrogen to form hydrides.1 At ambient conditions Pd can absorb up to a few hundred times its own volume of hydrogen. Some Pd hydrides are highly efficient catalysts for reactions such as oxidation, reduction, or carbonylation.2 Pd hydrides have also been examined in hydrogen storage applications, and interesting candidate materials include binary PdHx and ternary AxPdyHz hydrides, where metal A is usually an alkali or alkaline earth metal.3–5 The insertion of an alkali metal into Pd hydrides was suggested to increase the absorption reversibility. Ternary Pd hydrides with alkali metals have been synthesized in multiple stoichiometries, including A2PdH2, A2PdH4, and A2PdH6.6–8 These hydrides have distinct Pd–H interactions in the PdHx complexes, and as a consequence, they show a variety of electronic properties, ranging from insulating to semiconductor and to metallic and superconducting behaviors.

Among Pd ternary hydrides, Li2PdH2 has attracted particular attention. Li2PdH2 is one of the few metallic compounds in this group. At ambient conditions, Li2PdH2 crystallizes in a tetragonal structure [S.G. I4/mmm (139), Z = 2] with linear [Li–H–Pd–H–Li] complexes.9 As hydrogen binds strongly in the complexes, high temperatures are required to release it from Li2PdH2, which yields decomposition products LiPd, LiH, and H2. One approach to facilitate the hydrogen absorption/desorption is applying extreme pressure on Li2PdH2 to reduce the H⋯H distances in the structure, and weaken the Pd–H bonds, possibly decreasing the activation energy barrier. A study of high-pressure behavior of Li2PdH2 can therefore provide important information about its structural characteristics and help us to understand the desorption dynamics under ambient conditions. The structure modifications induced by the application of high pressure may lead to an improved hydrogen storage performance, which serves to motivate our present work, for which the deuterated version, Li2PdD2, was available. Li2PdD2 has the same crystal structure as Li2PdH2 but with hydrogen replaced by deuterium.10 It is expected that they have similar bonding and physico-chemical properties under pressure.

In the present work, the high-pressure structural behavior of Li2PdD2 was studied using synchrotron x-ray powder diffraction and first-principles computations up to 50 GPa. We identified a first-order phase transition at approximately 10 GPa to a monoclinic structure, which remains stable up to 50 GPa. Remarkably, the high-pressure structure of Li2PdD2 consists of novel fused [PdD2] chains, which was for the first time observed in alkali metal–transition metal hydrides. This study reveals that the external pressure can effectively modify the Pd–H interaction, or transition metal-H interaction in general in transition metal hydrides, which is critical to the absorption/release of molecular H2.

An Li2PdD2 sample was synthesized by grinding LiD (0.18 g, 20 mmol) and Pd powder (1.06 g, 10 mmol) with a mortar and pestle, which was then pressed into a pellet in a 1/2 in. die at 10 000 pounds in an argon filled glove box (<1 ppm of oxygen or moisture). An initial attempt to make Li2PdD2 from a piece of the pellet at 400 °C under D2 resulted in incomplete reaction. Several new pieces of the pellet were loaded into an alumina tube inside a pressure vessel and pressurized to 1000 psi with D2. The pressure vessel was heated to 502 °C for 17 h. The pellet piece at the bottom that was touching the alumina tube was dark, but the pieces on top of it had a golden sheen and were partially melted together. One of those top pieces was used for this experiment.

The polycrystalline sample was loaded, along with fine Au powder (acting as a pressure sensor) in a glove box with a controlled Ar atmosphere (<1 ppm of oxygen). The sample was sealed in a DAC inside the glove box and transferred to a gas-loading apparatus, where a Ne pressure of 200 MPa was created. The DAC was opened under the Ne pressure to let the gas in, resealed, and then taken out for further high-pressure experiments. An image plate CCD detector was used to collect pressure dependent X-ray diffraction (XRD) data at the undulator XRD beamline at GeoSoilEnviroCARS, APS, Chicago. The monochromatic x-ray beam (wavelength λ = 0.3344 Å) was focused to a nominal diameter of 4 μm. Pressure was determined using a known ambient temperature equation of state (EOS) of gold.11 Integration of powder diffraction images to yield scattering intensity versus 2θ patterns and initial analysis were performed using the DIOPTAS program.12 Le Bail refinements were performed using the GSAS13 software.

Pressure-induced phase transitions were simulated using the metadynamics method14,15 combined with the projector augmented plane-wave (PAW) method as implemented in the Vienna Ab Initio Simulation (VASP) program.16 Tight PAW potentials17 with the Perdew–Burke–Ernzerhof (PBE) functional18 and a 900 eV energy cutoff were used for the three elements. The Li, Pd, and D potentials employ 1s22s1, 4p64d10, and 1s1 as valence states, respectively. The ambient-pressure I4/mmm structure was used as the starting structure for the simulation of phase transitions at 300 K, in the pressure range of 10–50 GPa, using various supercells that consisted of 4 to 24 Li2PdD2 formula units (f. u.) along with a k-spacing of 2π × 0.08 Å−1 for the Brillouin zone (BZ) sampling. Each metastep consisted of a first principles molecular dynamics (MD) simulation within the canonical (NVT) ensemble for a simulation time of 0.4 ps. To reduce the energy barrier for phase transitions, the supercells were over-pressurized in both hydrostatic and non-hydrostatic conditions. A recent implementation19 of the metadynamics method was used to load the uniaxial stresses to the simulation cell. Enthalpy calculation for candidate structures was conducted with a fine k-spacing of 2π × 0.03 Å−1, which yielded excellent convergence for total energy (within 1 meV/atom).

X-ray diffraction (XRD) patterns of the Li2PdD2 sample at several selected pressures are shown in Fig. 1. At ambient pressure, the XRD pattern can be unambiguously indexed to the I4/mmm structure, in excellent agreement with the previously published results,9 see Fig. 2(a). A slight difference between observed and calculated intensities (assuming continuous Debye rings of uniform intensity) can be attributed to the use of a polycrystalline sample instead of a powder sample due to preferred orientation of the grains. For this reason, together with the large difference in the Z values between Pd and Li/D, we have performed Le Bail refinements. A low intensity peak at ∼5.39° (marked with an asterisk in Fig. 1) is attributed to an unknown, at this stage, impurity which persists up to the highest pressure of this study. We speculate that this impurity could be a Li2Pd compound with a hexagonal structure (PDF#50-1278, ICSD #104774). Although the 2θ position of the most intense Bragg peak of this compound matches the observed impurity peak, a definite identification is not possible from this peak alone.

The evolution of the XRD data under pressure clearly shows a phase transition of the sample at approximately 10 GPa. This is signified by splitting of the XRD peaks, and in particular, the occurrence of a new Bragg peak at a low 2θ angle (∼3°). Such changes of the XRD pattern are consistent with a monoclinic distortion of the tetragonal unit cell. The new phase is found to be stable to at least 48 GPa (the highest obtained pressure) without any sign of a subsequent phase transition. The high-pressure phase remains stable after the pressure is released to near ambient pressure (Fig. 1). To avoid any possible reaction with the atmosphere, the sample was confined in the gasket hole and therefore the lower pressure limit was 2 GPa. Thus, although we could not determine the metastability of the high-pressure phase at ambient conditions, it is clear that the pressure induced phase transition shows considerable hysteresis. A careful inspection of the XRD pattern shows that this phase does not belong to any known structural type of alkali metal–transition metal hydride.20 

First-principles metadynamics simulations were performed to provide insight into the pressure-induced phase transition and identify the structure. The simulation employing the I4/mmm phase as the starting structure revealed several low-enthalpy structures at pressures above 10 GPa. The best structure obtained at this point was a structure with the Pnma space group (see Table I for structural parameters). The Pnma structure becomes energetically more favorable than the I4/mmm structure at around 15 GPa (Fig. 3), emerging as a candidate for the high-pressure phase. The simulated XRD pattern of the Pnma structure also matches fairly well with the experimental data except the absence of a small angle peak at ∼3°. This evidence indicates that the Pnma structure is a good approximation to the experimental high-pressure phase, but not exact, perhaps an intermediate phase. Thus, we performed a new set of metadynamics simulations starting from the Pnma structure, aiming to find a low-energy pathway to the global minimum structure on the energy surface. This time the search yielded a monoclinic structure with the C2/m space group (Z = 6). The result of the full pattern Le Bail refinement of C2/m structure with the pattern collected at 41 GPa is shown in Fig. 2(b). The C2/m structure is validated by an excellent match to experimental XRD pattern (see Table I for structural parameters).

The C2/m structure is calculated to become more thermodynamically stable than the I4/mmm structure at around 4.6 GPa (Fig. 3). This transition pressure agrees well with the experimental transition pressure near 10 GPa. The C2/m structure is also the lowest-enthalpy structure among all competing structures in the studied pressure range, which establishes it as a high-pressure thermodynamic ground-state phase of Li2PdD2. The pressure-dependent cell volume per f. u. (Vpfu) and lattice parameters for both I4/mmm and C2/m structures were calculated and compared with the experimental results [Figs. 4(a) and 4(b)]. For the sake of comparison, here we used the non-standard I2/m setting for the C2/m structure to yield a β angle close to 90°. The experimental and theoretical Vpfu values and lattice parameters match quite well for both structures. The Vpfu reveals a subtle (∼1.4% as determined by the experimental results) but clear discontinuity at the phase transition, induced by a sudden drop of the lattice parameters.

The volume discontinuity at the transition point indicates that the phase transition is of the first order. This is understood by the change of chemical bonding in the Li2PdD2 structure. The tetragonal I4/mmm structure before the phase transition is shown in Fig. 5(a). This structure consists of Li+ cations and [PdD2]2− complexes. Assuming that Li transfers all valence electrons to the PdD2 group, the latter would be hypervalent and therefore need to be stabilized by resonance, i.e., D–Pd D D Pd–D, which becomes favorable in the linear D–Pd–D geometry.21 The Li+ cations are situated in a head-on fashion to D, forming the minimum repeatable unit in the structure. The stability of this structure has been addressed via the interactions of [PdD2]2− complexes with the Li+ counterions.22 In this geometry, the electrostatic attractions between Li+ and D are maximized along the c-direction. On the a-b plane, Li+ and [PdD2]2− are intercalated to form two dimensional slabs [Figs. 6(a) and 6(b)]. The stacking sequence for the slabs is ABAB…, which results in a tetragonal unit cell containing two Li2PdD2 units. At ambient pressure, the calculated Pd–D distance in the I4/mmm structure is 1.69 Å, close to the value in the neutral PdH2 complex.23 The closest Pd–Pd distance in the structure is 3.1 Å, indicating weak interactions between the [PdD2]2− anions.

The C2/m structure (in the I2/m setting) after the phase transition is shown in Fig. 5(b). In this structure, the [PdD2]2− groups lose their molecular integrity and are connected via extended chains. There are two types of chains, I and II, shown in Figs. 5(c) and 5(d). In chain type I, the Pd atoms are each triply coordinated to one terminal and two bridging D atoms. This chain is bent at the bridging D atoms, indicating that some bonding interactions have been developed between Pd atoms.24 The calculated Pd–Pd distance of 2.8 Å (∼7 GPa) is within the bonding distance of Pd. The bridging and terminal Pd–D distances are 1.80 and 1.84 Å, respectively. The surrounding Li+ cations provide an electrostatic stabilization to the chain, through either bifurcated or head-on interactions [Fig. 5(c)]. In chain type II, the [PdD2]2− anions are connected through the Pd–Pd interactions with all hydrogen atoms as terminals and stabilized by head-on Li+–D interactions [Fig. 5(d)]. The two chain types have the same Pd–Pd distances. They are alternatively populated on the a-c plane and intercalated by the Li+ cations [Figs. 6(c) and 6(d)]. The increased coordination of the structure results in a sudden change of the lattice parameters at the phase transition [Fig. 4(b)]. The c vector of the I4/mmm structure becomes the b vector of the I2/m structure accompanying a 7% reduction (calculated at 10 GPa). The a and b vectors split to a/3 and c of the I2/m structure with minor changes in the average length. In addition, at the transition point, the β-angle is increased from 90° to around 91° which transforms the unit cell to monoclinic. The high-pressure I2/m structure has Z = 6, which is three times greater than the I4/mmm structure.

The band structures and projected density of states (DOS) for the I4/mmm and C2/m structures are shown in Figs. 7(a) and 7(b). Both structures are found to have a weak metallic ground state. In both structures, the Fermi level (EF) is located at the bottom of a pseudogap in the DOS, suggesting that the electronic structure of Li2PdD2 is not free-electron-like. The DOS at EF in the I4/mmm (0 GPa) and C2/m (20 GPa) structures are, 0.21 and 0.11 e/eV per f. u., both substantially lower than the value in a free electron gas (3n/2EF, where n is the density of electrons). The low DOS at EF limits the possibility of Li2PdD2 achieving credible electrical conductivity. In Li2PdD2, the entire valence DOS is dominated by the 4d electrons of the Pd atoms. The D 1s DOS mixes strongly with the Pd DOS in the low-energy valence range, which is consistent with the bonding interactions in [PdD2]2−. The contribution from the Li atoms to the total DOS is hardly seen in the valence regime, indicating that the valence electrons are almost completely depleted from the 2s orbital. This electron transfer scenario is confirmed by the Bader charge analysis,25 which reveals the amount of charge loss (−) or gain (+) for Li, Pd, and D atoms as −0.86 e, +0.64 e, and +0.54 e, respectively, in the I4/mmm structure. In the C2/m structure, the charge loss/gain values are very similar, which are −0.83 e, +0.61 e, and +0.51 e, respectively (20 GPa).

A high-pressure phase transition of Li2PdD2 has been discovered by a combined experimental and computational study up to 50 GPa. This first-order phase transition was initially identified at approximately 10 GPa by synchrotron x-ray diffraction measurements and also confirmed with theoretical calculations. The exact crystal structure of the high-pressure phase is fully characterized as a monoclinic C2/m structure via a metadynamics simulation, and it is distinct from all known structures of alkali metal–transition metal hydrides. The observed phase transition via the application of pressure results in an additional bonding coordination. In the high-pressure structure, the PdD2 complexes lose molecular integrity and are fused to extended [PdD2] chains. The crystal structure is stabilized by electrostatic interactions between the PdD2 groups and surrounding Li+ cations, which results in a weak metallic ground state for Li2PdD2 at ambient and high pressures.

This work was supported by Natural Sciences and Engineering Research Council of Canada (NSERC), the DARPA (Grant Nos. W31P4Q1310005 and W31P4Q1210008), and the Deep Carbon Observatory DCO. A.F.G. was partly supported by Chinese Academy of Sciences visiting professorship for senior international scientists (Grant No. 2011T2J20) and Recruitment Program of Foreign Experts. Part of this work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Security, LLC under Contract No. DE-AC52-07NA27344. GSECARS is supported by the U.S. NSF (Nos. EAR-0622171 and DMR-1231586) and DOE Geosciences (No. DE-FG02-94ER14466). Use of the APS was supported by the DOE-BES under Contract No. DE-AC02-06CH11357. Computing resources were provided by the University of Saskatchewan, WestGrid, and Compute Canada. A.E. and A.P.P. thank the Office of Naval Research for supporting this work.

1.
W.
Grochala
and
P. P.
Edwards
,
Chem. Rev.
104
,
1283
(
2004
).
2.
V. V.
Grushin
,
Chem. Rev.
96
,
2011
(
1996
).
3.
W.
Bronger
and
G.
Auffermann
,
Z. Anorg. Allg. Chem.
621
,
1318
(
1995
).
4.
Y.
Sakamoto
,
R.
Nakamura
, and
M.
Ura
,
J. Alloys Compd.
231
,
533
(
1995
).
5.
W.
Bronger
and
G.
Auffermann
,
Chem. Mater.
10
,
2723
(
1998
).
6.
E.
Rönnebro
,
N.
Kitamura
, and
T.
Sakai
,
J. Alloys Compd.
358
,
216
(
2003
).
7.
W.
Bronger
and
G.
Auffermann
,
J. Alloys Compd.
228
,
119
(
1995
).
8.
K.
Puhakainen
,
E.
Stoyanov
,
M. J.
Evans
,
K.
Leinenweber
, and
U.
Häussermann
,
J. Solid State Chem.
183
,
1785
(
2010
).
9.
K.
Kadir
and
D.
Noréus
,
Z. Phys. Chem.
163
,
231
(
1989
).
10.
K.
Kadir
,
M.
Kritikos
,
D.
Noréus
, and
A. F.
Andresen
,
J. Less-Common Met.
172
,
36
(
1991
).
11.
M.
Matsui
,
J. Phys.: Conf. Ser.
215
,
012197
(
2010
).
12.
C.
Prescher
and
V. B.
Prakapenka
,
High Pressure Res.
35
,
223
(
2015
).
13.
A. C.
Larson
and
R. B. V.
Dreele
, Los Alamos National Laboratory Report LAUR 86-748,
2000
.
14.
R.
Martoňák
,
A.
Laio
, and
M.
Parrinello
,
Phys. Rev. Lett.
90
,
075503
(
2003
).
15.
R.
Martoňák
,
D.
Donadio
,
A. R.
Oganov
, and
M.
Parrinello
,
Nat. Mater.
5
,
623
(
2006
).
16.
G.
Kresse
and
J.
Hafner
,
Phys. Rev. B
47
,
558
(
1993
).
17.
G.
Kresse
and
D.
Joubert
,
Phys. Rev. B
59
,
1758
(
1999
).
18.
J. P.
Perdew
,
K.
Burke
, and
M.
Ernzerhof
,
Phys. Rev. B
77
,
3865
(
1996
).
19.
Y.
Yao
and
D. D.
Klug
,
Phys. Rev. B
85
,
214122
(
2012
).
20.
W.
Bronger
,
Angew. Chem., Int. Ed.
30
,
759
(
1991
).
21.
T. K.
Firman
and
C. R.
Landis
,
J. Am. Chem. Soc.
120
,
12650
(
1998
).
22.
M.
Olofsson-Mårtensson
,
U.
Häussermann
,
J.
Tomkinson
, and
D.
Noréus
,
J. Am. Chem. Soc.
122
,
6960
(
2000
).
23.
K.
Balasubramanian
,
P. Y.
Feng
, and
M. Z.
Liao
,
J. Chem. Phys.
88
,
6955
(
1988
).
24.
S.
Krüger
,
S.
Vent
,
F.
Nörtemann
,
M.
Staufer
, and
N.
Rösch
,
J. Chem. Phys.
115
,
2082
(
2001
).
25.
R.
Bader
,
Atoms in Molecules: A Quantum Theory
, International Series of Monographs on Chemistry (
Clarendon
,
Oxford
,
1990
).