Pure H conduction in oxyhydrides

See allHide authors and affiliations

Science  18 Mar 2016:
Vol. 351, Issue 6279, pp. 1314-1317
DOI: 10.1126/science.aac9185

Transporting the hydrogen anion

Hydrogen cation (H+) transport is common in both biological systems and engineered ones such as fuel cells. In contrast, the transport of hydrogen anions (H) is far less common and is usually coupled with or compromised by the parallel transport of electrons. Kobayashi et al. examined the transport of H in a series of rare-earth lithium oxyhydrides (see the Perspective by Yamaguchi). They prevented electronic conduction by using Li+ as a countercation. In an electrochemical cell, the oxyhydride material acted as a solid-state electrolyte for H, which suggests an alternative avenue for developing energy storage devices.

Science, this issue p. 1314; see also p. 1262


A variety of proton (H+)–conducting oxides are known, including those used in electrochemical devices such as fuel cells. In contrast, pure H conduction, not mixed with electron conduction, has not been demonstrated for oxide-based materials. Considering that hydride ions have an ionic size appropriate for fast transport and also a strong reducing ability suitable for high-energy storage and conversion devices, we prepared a series of K2NiF4-type oxyhydrides, La2-x-ySrx+yLiH1-x+yO3-y, in the hope of observing such H conductors. The performance of an all-solid-state TiH2/o-La2LiHO3 (x = y = 0, o: orthorhombic)/Ti cell provided conclusive evidence of pure H conduction.

Electric conduction is generally classified into two categories, electronic and ionic, which are attributed to the transport of electrons and ions in materials, respectively. Ionic charge carriers include a variety of species, such as Li+, H+, Ag+, Cu+, F, and O2–, and their conductors have found applications in energy devices such as fuel cells and batteries (15), for example. The conduction of hydride ions, H, is also attractive. These are similar in size to oxide and fluoride ions (6) and show strong reducing properties with a standard redox potential of H/H2 (–2.3 V), which is close to that of Mg/Mg2+ (–2.4 V). Hydride ion conductors may therefore be applied in energy storage and conversion devices with high energy densities.

In contrast to proton conduction that takes place widely in oxides (7) and other systems, pure H conduction has been verified only for a few hydrides of alkaline earth metals such as BaH2 (813). Unfortunately, utilization of the hydrides is difficult because of their structural inflexibility, which makes control of the lattice structure to create smooth transport pathways and control of the conducting hydride ion content difficult. We have considered oxyhydrides, where hydride ions and oxide ions share the anion sublattices, as candidate hydride conductors equipped with flexible anion sublattices. Known oxyhydrides include A2BHxO4-x (K2NiF4 structure; A: La, Ce, Nd, Pr, Sr; B: Co, V, Li; 0 < x ≤ 1), Sr3Co2O4.33H0.84 (Ruddlesden-Popper structure), ATiO3-xHx (perovskite structure; A: Ba, Sr, Ca) (1419), and [Ca24Al28O64]4+·4H (mayenite structure) (2022). However, none of these displays pure H conductivity.

Hydride ions have been reported to act as electron donors in oxide-based materials (2024), transferring electrons from hydride ions to the lattice. This causes conduction of electrons accompanied by a characteristic change in the hydrogen charge from H to H+. Indeed, the perovskite and mayenite-type oxyhydrides are dominated by electron conduction caused by the dissociation of hydride ions into electrons and protons (16, 2022, 25). Taking this into consideration, preventing electron donation from hydride ions in the crystal lattice may be important for achieving pure H conduction in the oxide framework structure. In this study, we attempted to prepare a series of K2NiF4-type oxyhydrides, La2-x-ySrx+yLiH1-x+yO3-y (0 ≤ x ≤ 1, 0 ≤ y ≤ 2, 0 ≤ x + y ≤ 2), which are equipped with cation sublattices featuring cations more electron-donating than H and anion sublattices that exhibit flexibility in the storage of H, O2–, and vacancies.

To aid in the understanding of the compositional and structural features of the present system, a few representative examples are shown below: La2LiHO3 (x = y = 0), Sr2LiH3O (x = 0, y = 2), La2-xSrxLiH1-xO3 (0 ≤ x ≤ 1, y = 0), and La1-xSr1+xLiH2-xO2 (0 ≤ x ≤ 1, y = 1). Solid solutions between La2LiHO3 and Sr2LiH3O are represented as La2-ySryLiH1+yO3-y (x = 0, 0 ≤ y ≤ 2). Here, the H:O ratio changes accordingly as La is substituted with Sr, maintaining the simple A2BX4 composition (A: La, Sr; B: Li; X: O, H). In La2-xSrxLiH1-xO3 and La1-xSr1+xLiH2-xO2, by contrast, the anion sublattice contains vacancies. Several starting compositions (see table S1) contained in the colored compositional range in Fig. 1A were synthesized by treating the appropriate starting materials in sealed Au capsules at high pressure and high temperature (26).

Fig. 1 La2-x-ySrx+yLiH1-x+yO3-y (0 ≤ x < 1, 0 ≤ y ≤ 2) oxyhydride system.

(A) Compositional range of the oxyhydride. The compositions for the standard (maintaining A2BX4) and anion-deficient samples are represented by black and blue markers, respectively. Filled markers indicate that ionic conductivities were measured for these compositions. (B) Comparison of the synchrotron XRD profiles for La2-ySryLiH1+yO3-y (x = 0, 0 ≤ y ≤ 2). (C) Magnified view of typical peaks.

The x-ray diffraction (XRD) pattern of La2LiHO3 (x = y = 0) could be assigned to the K2NiF4-type structure, but the lattice symmetry was found to change between tetragonal (I4/mmm, t-La2LiHO3) and orthorhombic (Immm, o-La2LiHO3), depending on certain experimental factors, including the LiH/La2O3 ratio of the starting materials (1/1 and 2/ 1 for t- and o-La2LiHO3, respectively), heating temperature, and pressure (fig. S1). All other samples were prepared under the same conditions as outlined for o-La2LiHO3. Regarding the Sr-substituted series of La2-ySryLiH1+yO3-y, the diffraction peaks continuously shifted to lower angles with increasing y (Fig. 1B) and the lattice symmetry changed from Immm (y < 1) to I4/mmm (y ≥ 1) (Fig. 1C).

The compositions and structures of La2-ySryLiH1+yO3-y (y = 0, 1, 2) were determined by x-ray and neutron Rietveld analyses. Details of the analyses are given in figs. S2 and S3, in tables S2 and S3, and in the supplementary text. In La2LiHO3, the two apical sites of the LiX6 octahedra are occupied only by O2–, as illustrated schematically in Fig. 2, whereas the four in-plane apexes are occupied by O2– and H. These results indicate that the highly charged cations (i.e., La3+ and Sr2+) require highly charged anions around them. LaSrLiH2O2 is composed of tetragonal (LiH2) and (LaSrO2)+ layers alternately stacked along the c axis. The further increase in hydride content up to Sr2LiH3O results in the formation of (Sr2HO)+ layers. In the series of compositions, we note here that there exists a K2NiF4-type, H–free oxide, La2LiO3.5, in which the anion vacancies are randomly distributed in the basal (LiO0.75)0.5- layers (27).

Fig. 2 Crystal structures of t-La2LiHO3 and La2-ySryLiH1+yO3-y (y = 0, 1, 2).

(A) A unit cell of the prepared oxyhydrides. Lanthanum (and strontium) ions occupy the A sites of the K2NiF4-type structure that are 12-fold coordinated with anions. Lithium occupies the B sites that are octahedrally coordinated with anions. The structure is composed of perovskite and rock-salt structure blocks stacked along the c axis. The perovskite-type layers are separated by rock-salt layers.

It is notable that t-La2LiHO3 contains anion vacancies (V(H,O)) with the chemical formula La2Li(H0.53O1.21V(H,O)0.26)O2, exhibiting H, O2–, and V(H,O) disorder at the axial sites of the LiX6 octahedra. In contrast, the orthorhombic phase, o-La2LiHO3, is stoichiometric, and H and O2– are ordered on the axial anion sites. The symmetry change can be attributed to the order-disorder transition of H and O2– in the axial sites, both with and without vacancies. The XRD results for a pair of anion-deficient series, La2-xSrxLiH1-xO3 and La1-xSr1+xLiH2-xO2, are shown in figs. S4 and S5. The occupancy parameters determined by a neutron Rietveld analysis for La0.7Sr1.3LiH1.7O2 (x = 0.3, y = 1) of gH1 = 0.938(2), gH2 = 0.118(3), and gO1 = 0.882(3) lead to a composition of La0.7Sr1.3Li(H1.88V(H)0.12)H0.24O1.76 (fig. S6 and table S4). A vacancy was introduced with x at the LiH4 plane together with H/O2– anion mixing at the apical sites.

The valence state of each constituent atom was estimated for La2LiHO3, LaSrLiH2O2, and Sr2LiH3O by means of valence charge integration over its Voronoi cell (table S5). The valences of hydrogen and oxygen in all materials were estimated as approximately –0.8 to –1.0 and –1.3 to –1.6, respectively, which indicates that these elements exist as H and O2–. Electronic density of states calculations also corroborate the presence of hydride ions, as can be seen in fig. S7, where these ions have localized electrons between approximately 0 and –5 eV below the Fermi level. The Li–H bond was confirmed to be ionic in nature.

The conductivities of the oxyhydrides were estimated from the impedance plots, which are characteristic of typical ionic conductors, as shown in fig. S8. Figure 3, A to C, shows Arrhenius plots of the conductivity of La2-x-ySrx+yLiH1-x+yO3-y. The conductivity and activation energy values (see table S6) demonstrate that the ionic conductivity varies with the compositions of both x and y in La2-x-ySrx+yH1-x+yO3-y. Figure 3A shows the temperature dependence of the conductivity for o-La2LiHO3 (x = 0, y = 0), LaSrLiH2O2 (x = 0, y = 1), and Sr2LiH3O (x = 0, y = 2). The conductivity increases with increasing y, with the highest conductivity of 3.2 × 10−5 S cm–1 at 573 K being observed for Sr2LiH3O (y = 2). The introduction of hydride ions into the anion sites of the K2NiF4 structure improved the ionic conductivity, suggesting that the primary charge carriers seem to be these hydride ions. The conduction is further facilitated by the introduction of vacancies, as can be seen both for La2-xSrxLiH1-xO3 (0 ≤ x ≤ 0.2, y = 0, Fig. 3B) and La1-xSr1+xLiH2-xO2 (0 ≤ x ≤ 0.4, y = 1, Fig. 3C), up to 2.1 × 10−4 S cm–1 for La0.6Sr1.4LiH1.6O2 at 590 K (activation energy ~68.4 kJ mol–1). This increase in conductivity with the introduction of vacancies indicates that the structural defects can affect the ionic diffusion. To further identify the nature of charge carriers, we measured the electrical conductivity of La0.6Sr1.4LiH1.6O2 (x = 0.4, y = 1.0) by the Hebb–Wagner polarization method (28) using an asymmetric (–) Pd/La0.6Sr1.4LiH1.6O2/Mo (+) cell at 480 and 590 K. The total electrical conductivities (electrons + holes) at the irreversible Mo-electrolyte interface of the cell at 480 and 590 K were 2.9 × 10−8 and 4.1 × 10−7 S cm–1, respectively. It is evident that La0.6Sr1.4LiH1.6O2 is a pure ionic conductor (fig. S9 and table S7).

Fig. 3 Temperature dependence of the ionic conductivities of La2-x-ySrx+yLiH1-x+yO3-y.

(A) Compositions for La2-ySryLiH1+yO3-y (x = 0, y = 0, 1, and 2) with a fixed cation/anion ratio of (A2B)/X4, where A, B, and X are La(Sr), Li, and O(H), respectively. Anion-deficient series, (B) La2-xSrxLiH1-xO3 (y = 0, 0 ≤ x ≤ 0.2) and (C) La1-xSr1+xLiH2-xO2 (y = 1, 0 ≤ x ≤ 0.4).

An all-solid-state cell was then constructed with o-La2LiHO3 as the solid electrolyte. The electrode configuration—namely, a powdered mixture of the electrode and electrolyte materials—was similar to those used for the all-solid-state lithium battery (1). Figure 4A shows the discharge curve of the Ti/o-La2LiHO3/TiH2 cell, displaying a constant discharge current of 0.5 μA at 300°C. The cell showed an initial open circuit voltage of 0.28 V, which is consistent with the theoretical value calculated from the standard Gibbs energy of formation of TiH2 (29). During the electrochemical reaction, the cell voltage dropped rapidly from 0.28 to 0.06 V and then decreased gradually to 0.0 V. This steep drop-off in the first reaction step corresponds to an increase in hydride ion content at the anode, according to the constant current discharge reaction: Ti + xH → TiHx + xewhere the reaction at the cathode is as follows:

TiH2 + xe → TiH2-x + xH
Fig. 4 All-solid-state hydride cell.

(A) Discharge curve for a solid-state battery with the Ti/o-La2LiHO3/TiH2 structure. The inset shows an illustration of the cell and the proposed electrochemical reaction. (B) X-ray diffraction patterns for the electrolyte (o-La2LiHO3), cathode (TiH2 + o-La2LiHO3), and anode (Ti + o-La2LiHO3) materials after the reaction. Magnifications are supplied for ranges 13° to 13.8° and 15.1° to 15.8°.

These discharge reactions were confirmed by observation of the phases that appeared following the reaction. Figure 4B shows the synchrotron XRD patterns for the cathode, electrolyte, and anode, both before and after the reaction. The absence of any variation in the diffraction patterns of the electrolyte indicates that the La2LiHO3 electrolyte is stable when in contact with the Ti and TiH2 electrodes during the reaction. Phase changes detected for the cathode and anode materials are consistent with those expected from the Ti-H phase diagram (29), where the δ-TiH2 Embedded Image phase releases hydrogen and is transformed into α-Ti (P63/mmc) through a two-phase (α-TiHb + δ-TiH2-a) coexistence region, which is found below ~573 K (fig. S10). In the case of the cathode, additional diffraction peaks corresponding to P63/mmc symmetry were detected. In addition, the signals corresponded to a shift of TiH2 to a higher angle, thus indicating that lattice shrinkage takes place with the release of hydrogen from TiH2. In the case of the anode, peaks corresponding to Embedded Image symmetry were detected. These results indicate that during the electrochemical reaction, hydride ions are released from the TiH2 cathode and diffuse into the Ti anode through the o-La2LiHO3 (fig. S11).

In conclusion, pure H conduction was realized in the La2-x-ySrx+yLiH1-x+yO3-y system. The present success in the construction of an all-solid-state electrochemical cell exhibiting H diffusion confirms not only the capability of the oxyhydride to act as an H solid electrolyte but also the possibility of developing electrochemical solid devices based on H conduction.

Supplementary Materials

Materials and Methods

Supplemental Text

Figs. S1 to S11

Tables S1 to S7

References (3039)

References and Notes

  1. Details of synthesis condition for the oxyhydrides are described in supplementary materials.
Acknowledgments: This research was supported by JST, PRESTO, and Grant-in-Aid for Young Scientists (A) no. 15H05497 and (B) no. 24750209; Grant-in-Aid for Challenging Exploratory Research nos. 15K13803, 23655191, and 25620180; and Grant-in-Aid for Scientific Research on Innovative Areas nos. 25106005 and 25106009, from the Japan Society for the Promotion of Science. Synchrotron and neutron radiation experiments were carried out as four projects approved by the Japan Synchrotron Radiation Research Institute (JASRI) (Proposal no. 2013A1704), the Japan Proton Accelerator Research Complex (J-PARC) (Proposal no. 2010A0058), the Spallation Neutron Source (SNS) in the Oakridge National Laboratory (Proposal no. IPTS5808), and the Neutron Scattering Program Advisory Committee of IMSS, KEK (Proposal no. 2014S10). A part of neutron experiments (Proposal no. 2014S10) was performed at BL09 Special environment neutron powder diffractometer (SPICA) developed by the Research and Development Initiative for Scientific Innovation of New Generation Batteries (RISING) project of the New Energy and Industrial Technology Development Organization (NEDO). Supercomputing time on the Academic Center for Computing and Media Studies (ACCMS) at Kyoto University is gratefully acknowledged. Further information regarding the materials and methods is included in the supplementary materials. G.K., A.W., M.H., and R.K. have filed for a patent application with the Japan Patent Office under no. JP2015-22868 on the H–- conductive oxyhydride system and its manufacture.

Stay Connected to Science

Navigate This Article