Synthesis of FeH5: A layered structure with atomic hydrogen slabs

See allHide authors and affiliations

Science  28 Jul 2017:
Vol. 357, Issue 6349, pp. 382-385
DOI: 10.1126/science.aan0961

Atomic hydrogen with an iron assist

Metal polyhydrides can be used at lower pressures to make material that might have atomic hydrogen bonding. Pépin et al. manage to synthesize an incredibly hydrogen-rich FeH5 compound at 130 GPa pressure. The material consists of slabs of four thin planes of atomic hydrogen intercalated with layers of quasicubic FeH3 units. These metal polyhydrides were stable at far more accessible pressures than pure hydrogen. This achievement provides an opportunity to investigate special electrical properties expected from atomic hydrogen bonding, such as superconductivity.

Science, this issue p. 382


High pressure promotes the formation of polyhydrides with unusually high hydrogen-to-metal ratios. These polyhydrides have complex hydrogenic sublattices. We synthesized iron pentahydride (FeH5) by a direct reaction between iron and H2 above 130 gigapascals in a laser-heated diamond anvil cell. FeH5 exhibits a structure built of atomic hydrogen only. It consists of intercalated layers of quasicubic FeH3 units and four-plane slabs of thin atomic hydrogen. The distribution of the valence electron density indicates a bonding between hydrogen and iron atoms but none between hydrogen atoms, presenting a two-dimensional metallic character. The discovery of FeH5 suggests a low-pressure path to make materials that approach bulk dense atomic hydrogen.

Dense metal hydrogen is expected to possess intriguing properties associated with the light mass and quantum behavior of the proton, such as a room-temperature superconductivity or even a superfluid/superconductor fluid state (1, 2). We should observe atomic metal hydrogen around 450 GPa; however, achieving such a pressure with static compression is a major challenge. An alternate approach to high-temperature superconductivity of dense metal hydrogen is through hydrogen-rich alloys at much lower pressures (3). Many density functional theory calculations show, for numerous elements, pressure favoring the stability of polyhydrides with high stoichiometry. These polyhydrides might be superconductors with a critical temperature (Tc) approaching ambient temperature (4, 5). The richness of this line of research has been confirmed by the discovery of superconductivity in hydrogen sulfide at a Tc of 203 K at 150 GPa, explained by the formation of H3S (6, 7). Furthermore, polyhydrides with unusually high H/M (H, hydrogen; M, metal) ratios were shown to exist with the observations of LiH6 and NaH7, which contain H2 units (8, 9). The progression in the discovery of these polyhydrides suggests a compound with a very high H content, which would be a real analog for atomic metal hydrogen.

The existence of alkali and alkaline earth polyhydrides is a topic investigated with a variety of methods and seemed the most promising route to reach very high H stoichiometries (10). However, the polyhydrides of these two groups of elements contain complex hydrogenic sublattices composed of H, H2, H3, or cage units. For other group elements, the richest stoichiometry is generally associated with the presence of quasimolecular H2 or H3 units (1113). On the other hand, trends disclosed by calculations on transition metal polyhydrides seem encouraging. Many H atoms can be bound to a metal atom forming molecules that are stabilized in matrices, potentially forming extended hydride structures. The hydrogen atoms should also remain in the H hydridic form even for the rich stoichiometry. With increasing hydrogen content, calculations predict that the dimensionality of the metal network decreases along with the emergence of layers of H atoms (1416). The use of high hydrogen pressure has been very efficient in producing hydrides of transition metals, most being interstitial hydrides with an H/M ratio close to 1. Few larger H contents are obtained for the interstitial stoichiometric compounds RhH2, IrH3, FeH2, and FeH3 (1719). In contrast, we synthesized a FeH5 polyhydride with a prominent layered structure.

Epsilon iron hydride (ε′-FeHx, x ≤ 1) is an extensively studied archetypical interstitial hydride (20, 21). The compression of Fe in excess hydrogen allows synthesis of FeH2 and FeH3 (19). The high-pressure sequence FeH, FeH2, and FeH3 was the clear experimental confirmation of the trend for hydrogen solubility to increase in metals determined theoretically by formation of unusual polyhydrides.

We compressed a 2-μm-thick Fe polycrystalline sample surrounded by excess hydrogen in a diamond anvil cell, similar to previous work (19). The diamond culets were 100 μm, and we performed on-line yttrium-aluminum-garnet laser heating of the sample up to 150 GPa on the ID27 x-ray diffraction (XRD) beamline of the European Synchrotron Radiation Facility. The temperature was kept below 1500 K to avoid parasitic chemical reaction of hydrogen with the sample chamber wall and prevent diffusion of hydrogen in the diamond anvil. We collected angular XRD patterns before and about 1 min after laser heating. We confirmed the formation of simple cubic FeH3 by laser heating of our samples between 100 and 125 GPa (19). New XRD peaks (Fig. 1A) appeared when the sample was laser-heated at 130 to 140 GPa. We found the same results on a second sample loading. The Rietveld refinement, using the FULLPROF software, of the diffraction pattern recorded at 130 GPa yields a tetragonal unit cell with the symmetry I4/mmm with a = b = 2.404 Å and c = 11.621 Å, with iron atoms in Wyckoff position 4e (0, 0, 0.896) (Fig. 1A). The low scattering power of hydrogen atoms prevents the determination of their positions and consequently the stoichiometry using XRD. We measured the equation of state of this phase by decreasing pressure down to 66 GPa at ambient temperature (Fig. 1B). It decomposes back to FeH3 below this pressure. The volume as a function of pressure [V(P)] data points can be fitted with a Vinet-type equation of state (22) with ambient pressure volume V0 = 16.447 (±0.19) cm3/mol, bulk modulus K0 = 85.9 (±4.5) GPa, and its pressure derivative K0 = 5 (fixed).

Fig. 1 Equation of state and structure determination of FeH5.

(A) Powder x-ray diffraction pattern and Rietveld refinement of the structure (iron atoms only) at 130 GPa for λ = 0.3738 Å. Vertical ticks correspond to the Bragg peaks for I4/mmm-FeH5 and hexagonal close-packed–rhenium (gasket material). The conventional reliability factors for the Rietveld refinement are (in %) RBragg(FeH5) = 8.72, Rp = 19.3, Rwp = 17.3. (Inset) Image plate. (B) Molar volume per formula unit as a function of pressure for iron hydrides (19). The Vinet fits (22) of the data are full lines. Ab initio calculations are the black dashed line. Dash/dotted gray lines depict the molar volumes of ideal mixtures of iron and molecular hydrogen with H:Fe ratio of 4 and 5 using the previously determined equations of state (33, 34). (C) Model representation of the FeH5 structure. Large blue and small red spheres represent the iron and hydrogen atoms, respectively.

We compared the compression curve of this FeHx compound to the equations of state of FeH, FeH2, FeH3, and ideal Fe–H2 solutions (Fig. 1B). We found a larger volume per Fe in our samples than in FeH3, thus suggesting a substantial increase of the H/Fe ratio concentration. The V(P) curve of the FeHx compound lies above the ideal mixing line of Fe and 4/2 × H2 solids and below the ideal mixing line of Fe and 5/2 × H2. This indicates that a stoichiometry of 4 would make FeH4 energetically less favorable with respect to the decomposition into the elements. The volume increase per H atom going from FeH to FeH2 and from FeH2 to FeH3 is very similar—~1.9 Å3 around 100 GPa (19)—and going from FeH3 to FeHx compound is twice this value. This suggests that FeH5 may be an appropriate stoichiometry to explain our data.

We used the ABINIT code (23) and the projector-augmented wave method (24) for first-principles density functional calculations to determine the structural positional parameters of the hydrogen atoms in FeH5 (25). We employed the generalized gradient approximation in the Perdew-Burke-Ernzerhof form (GGA-PBE) (26) that enabled us to satisfactorily calculate the properties of FeH, FeH2, and FeH3 (19). When we assumed the experimental determination of the Fe sublattice, we retained, among all structures compatible with the possible Wyckoff positions of a tetragonal I4/mmm unit cell (with Fe atoms in a 4e position), 53 structures (25) and structurally optimized all of them. The minimum enthalpy at 147 GPa among these various structures is obtained with H atoms in Wyckoff positions 8g (0, 0.5, 0.185), 4e (0, 0, 0.4095), 4e (0, 0, 0.230), and 4c (0, 0.5, 0). This lowest enthalpy structure for FeH5 (Fig. 1C and fig. S1) is a structure formed by alternating layers of FeH3 and hydrogen atoms. Phonons calculation shows that this structure is dynamically stable (25). Furthermore, two molecular dynamic runs in the isobaric-isothermal (N, P, T) ensemble confirm the stability: This structure is unchanged after a molecular dynamic run and is obtained after starting from a higher energy structure (25). Finally, the compression curve calculated for this FeH5 structure is in excellent agreement with our experimental data (Fig. 1B). We found good agreement comparing the experimentally and computationally determined lattice parameters (fig. S2). The c axis is more compressible than the a axis, which is in line with our preferred structure (Fig. 1C), where the strongest bonds are expected to lie in the (ab) plane. The FeH5 structure has the same space group I4/mmm as FeH2, with Wyckoff positions for the Fe atoms that are almost identical [4e (0, 0, 0.853) for FeH2 and 4e (0, 0, 0.896) for FeH5]. FeH5 also has structural similarities to a simple cubic FeH3. The layers containing Fe and H atoms are composed of quasisimple cubic FeH3 units, similar to the FeH3 structure, sharing faces (the Fe–Fe distances are 2.404 Å along the x and y axis and 2.414 Å along the z axis) arranged in an ABAB stacking. These FeH3 units are slightly bigger than in pure FeH3, which has a cell parameter of 2.38 Å at 130 GPa.

The structures of the Fe polyhydrides stable in excess hydrogen have increasing stoichiometry with increasing pressure and, as previously suggested (1416), decreasing dimensionality of the associated metal network due to the intercalation of the H layers (Fig. 2). Experimentally, we did not observe the predicted P21/m-, P213-, and Imma-structured FeH4 (27, 28). Our calculations also allow us to propose a Cmmm-structured FeH4 not previously considered (27, 28) (fig. S5). However, the Hull plot (fig. S3) shows that FeH4 is never stable against decomposition. FeH5 is stable against decomposition between FeH3 and H2 above 55 GPa. Therefore, the high stability of FeH5 above 55 GPa hinders the formation of FeH4.

Fig. 2 Hydrogen in the host lattice.

(A) Evolution of the stoichiometry and of the structures of iron hydrides synthesized in excess of hydrogen as the pressure increases. (B) Volume expansion by adding one H to FeHn, compared to the equation of state of pure atomic hydrogen. The volume expansion is determined as the slope of the equation V(FeHn) versus stoichiometry at various pressures. (C) Comparison of the nearest H–H distances in AlH3 (29), H3S (30), FeH2, FeH3 (19), metal H (25), and FeH5 as a function of pressure.

We compared the H–H nearest-neighbor distance in the dense atomic hydrogen slab of FeH5 to the nearest-neighbor H distance for polyhydrides in which neither quasimolecular H2 nor H3 units are formed (Fig. 2B). At 130 GPa the nearest H–H distance is 1.32 Å, which is considerably lower than the ~1.70 Å value in FeH3. Aluminum hydride (AlH3) has a nonbonded H–H distance 1.53 Å at 130 GPa (29). A similar value of ~1.5 Å was found in the high-temperature superconducting phase of H3S (5, 30). The shortest H–H distance in FeH5 is thus the closest to the shortest H–H distance in atomic metal hydrogen, artificially extended in the same pressure range (Fig. 2C) (25). The volume expansion due to the addition of one hydrogen in the FeHx formula unit is constant at a given pressure and is close to the volume of atomic metal hydrogen (Fig. 2C). Hence, the formation of Fe polyhydrides can be viewed as an ideal mixing of Fe and atomic metal H and associated, for a large fraction of metal H (25), with the formation of a thicker atomic H slab in the structure of the polyhydride.

The electron-density distribution (Fig. 3) gives a qualitative indication of the bonding in FeH5. The valence electron density is nonuniform in space (Fig. 3A). The higher electron density between Fe and H suggests bonding between them. Thirteen H atoms bond to each Fe atom, 12 of them bridging two Fe atoms, and the remaining one has a single Fe–H bond parallel to c (“top” H). The ABAB stacking of FeH3 units allows a compact arrangement of “top” H atoms. Many hydrogen atoms can be bound to a transition metal atom (31), although only FeH, FeH2, and FeH3 molecules could be isolated in matrices at ambient pressure (32). We visualized the electron density distribution in the plane parallel to the stacking direction, also showing the H–Fe bonding (Fig. 3B). The electron density in the plane of the hydrogen atoms is more uniform (Fig. 3C), with a minimum value 3.6 times lower (0.3 el/Å3) than that along Fe–H. It clearly shows that no H–H bonding exists. The topology of the electron density seems to give a different representation of the FeH5 structure than the geometrical one, more as a stacking of alternate FeH5 layers reminiscent of two-dimensional (2D) van der Waals materials. The reality is probably in between, with the atomic H slab interacting with the Fe atoms to stabilize dense atomic H at a much lower pressure than in pure hydrogen, in the spirit of the chemical “precompression” proposed by Ashcroft (3).

Fig. 3 Electron-density distribution at 147 GPa.

(A) Structure of FeH5 showing the isosurface for the electron density of 0.8 el/Å3. (B) Electron density map in the plane y = 0 as shown on the truncated structure. The large hidden spheres at the bottom around x = –2 and 2 bohrs are iron atoms. (C) Electronic density map in a plane perpendicular to the c axis as shown on the truncated structure. The hydrogen atom in position (0, 0, 0.23) is taken as the reference position (0, 0).

We also calculated the band structure and the electronic densities of states (DOS) for FeH5 (Fig. 4). The DOS has a clear steplike character at the bottom of the occupied band, indicating that electronically FeH5 is also a 2D metal. In this range, the DOS is constituted by H 1s and Fe 4s orbitals. Yet, there is a clear depletion of the DOS near the Fermi level, so FeH5 should be considered as a weak metal. The electronic band structure, as drawn in Fig. 4B, shows that the bandgap closure is only effective for a few points at the Brillouin-zone corners.

Fig. 4 Band structure and electronic densities of state of FeH5 at 147 GPa.

(Left) Electronic density of states of FeH5. (Right) Electronic band structure plotted on a path in the kz = 0 plane. (Inset) View of the Brillouin zone of the conventional unit cell, with the path used to plot the band structure (red arrows). The Fermi level is set at 0 eV in both cases.

The series of Fe polyhydrides, from FeH to FeH5, present an example of how hydrogen can combine with transition metals as pressure increases. Remarkably, the structures of these polyhydrides are built of atomic hydrogen only. By increasing pressure, hence the H content, slabs of atomic hydrogen are stabilized, giving, as in an FeH5 structure, structures resembling bulk atomic hydrogen. The next step would now be to quantify in FeH5 some of the intriguing quantum effects of atomic hydrogen.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S6

Tables S1 to S3

Data S1

References (3540)

References and Notes

  1. Materials and methods are available online as supplementary materials.
  2. Acknowledgments: The authors acknowledge the European Synchrotron Radiation Facility for provision of synchrotron radiation on beamline ID27 (proposal HC-1914). Data are available on request. The authors declare no conflict of interest.
View Abstract

Navigate This Article