Communication pubs.acs.org/JACS
Cubic Fluorite-Type CaH2 with a Small Bandgap Hiroshi Mizoguchi,*,† SangWon Park,†,‡ Takashi Honda,§,¶ Kazutaka Ikeda,§,¶ Toshiya Otomo,§,¶ and Hideo Hosono*,†,‡,⊥ †
Materials Research Center for Element Strategy, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama 226-8503, Japan ‡ Laboratory for Materials Research, Institute of Innovative Research, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama 226-8503, Japan § Institute of Materials Structure Science, High Energy Accelerator Research Organization (KEK), Tsukuba, Ibaraki 305-0801, Japan ¶ J-PARC Center, KEK, Tokai, 319-1106, Japan ⊥ ACCEL, Japan Science and Technology Agency, 4-1-8 Honcho, Kawaguchi, Saitama 332-0012, Japan S Supporting Information *
we encounter much more unusual structural trends in salt-like hydrides.6 As shown in Table S1, MH adopts rocksalt-type crystal structure for M = Li−Cs, as could be expected from the large ionicities of these compounds. However, the crystal structures appearing for MH2 phases do not follow the tendency seen for ionic compounds, e.g., fluorides. Like MgF2, MgH2 takes a rutile-type structure, but MH2 (M = Ca, Sr, or Ba) adopts the PbCl2-type (cotunnite-type) rather than the fluorite-type preferred by the corresponding fluorides. Meanwhile, ScH2, YH2, or LnH2 (Ln: lanthanide ion) crystallizes in fluorite-type structures, whereas low oxidation state including Ln2+ does not appear in fluorides generally. Among these curious hydrides, CaH2 plays an important role as it lies in boundary region between rutile-type, fluorite-type, and PbCl2-type structure. The PbCl2-type CaH2 structure observed under ambient conditions, which we will denote as o-CaH2 (αCaH2), seems to capture a brief intermediate point as the relative size of H− ion shrinks drastically across the series from MgH2 to CaH2 due to the size flexibility of the H− ion. Its role as a boundary phase is further highlighted by the discovery that it undergoes a pressure-induced structural phase transition from PbCl2-type to InNi2-type at ∼15 GPa,7 as well as the existence high temperature phase (β-CaH2) over the 1053−1256 K whose crystal structure has not been confirmed experimentally.8 Curiously, the fluorite-type structure that is anticipated to be a low pressure phase relative to the PbCl2-type has not yet been observed for CaH2. Finding conditions that fluorite-type CaH2 could be stabilized would thus be interesting from the point of view of understanding the structural trends within this system. Furthermore, new polymorphs of CaH2 could be of interest for catalytic applications, as Ru-loaded o-CaH2 has attracted attention recently as an efficient catalyst for ammonia synthesis.9 Here we report Ca1−xMxH2+x (M = Y or La, x = ∼0.15), with fluorite-type structures. These phases are the first alkaline earth metal hydrides adopting fluorite-type structure. This cubic hydride exhibits a bandgap (2.5 eV) much smaller than that of the original o-CaH2 (4.4 eV), and is, in fact, the smallest among alkali or alkaline earth hydrides reported to
ABSTRACT: A cubic variant of CaH2 adopting a fluoritetype crystal structure was synthesized by cationic substitution with La or Y, yielding the first alkaline earth hydride-based with fluorite-type framework. The material has a bandgap of ∼2.5 eV (greenish yellow in color), which is much smaller than that of orthorhombic PbCl2type CaH2 (4.4 eV) and is, in fact, the smallest among alkaline or alkaline earth metal hydrides reported to date. Analysis of the density functional theory band structure of cubic-CaH2 indicates that its conduction band minimum is formed mainly by the interaction between the Ca 3d eg orbitals around the crystallographic cavity defined by cubes of H− ions. The use of such cavities in the creation of lowlying conduction band minima by semiconductors is extremely rare, and has similarities to inorganic electrides. n inorganic solids, the hydride ion, H− might be expected to behave like the F− ion, due to their of equal charge and similar ionic sizes.1,2 Similar to F−, H− commonly forms saltlike ionic compounds with positive cations such as alkali-metal ions in spite of its relatively small electron affinity. However, they differ in the extent to which simple radius ratio considerations can rationalize their structural chemistries, as is summarized for MX and MX2 (X = F or H) compounds in Table S1 in the Supporting Information.3 Generally, the crystal structures of MF2 compounds, as well as the corresponding oxides, or oxyfluorides with large ionicities, shift systematically from the rutile-type, to the fluorite- and PbCl2-types, depending on the cation/anion radius ratio.3 The anionic coordination number of cation increases from 6 to 8 or 9. Fluorite-type structure appears for MF2 with M = Ca, Sr, or Ba. The highpressure phases of MF2 compounds can be similarly understood when one factors in the larger compressibility of the anions over the cations. Under pressure, the structures tend to shift as though the cation has a larger relative size.4 For example, CaF2 transforms from the fluorite-type to the PbCl2type at 9.5 GPa.5 Even more striking is the formation of the InNi2-type crystal structure with 11 coordination number of indium-type positions for X at extreme pressures. In contrast,
I
© 2017 American Chemical Society
Received: June 3, 2017 Published: August 14, 2017 11317
DOI: 10.1021/jacs.7b05746 J. Am. Chem. Soc. 2017, 139, 11317−11320
Communication
Journal of the American Chemical Society
dense packing than c-CaH2.4 Thus, we can conclude that the stability of c-CaH2 relative to o-CaH2 will be greatest at low pressures. Figure 2a−c shows calculated band structure of hypothetical c-CaH2 along with isostructural ScH2 and TiH2 (using the
date. A chemical bonding analysis through density functional theory (DFT) band structure calculations will trace this unique bandgap to the dominate role that tiny crystallographic cavity spaces defined by H-cubes play in their conduction band minima. This bonding situation arises from Ca−Ca covalent interactions around the cavity centers, that is, Ca multicenter bonding states stemming from the Ca 3d orbitals with relatively deep binding energy. We began our investigation by exploring the phase stability of two polymorphisms of CaH2 by theoretical methods: oCaH2 and the hypothetical cubic phase with fluorite-type structure we denote as c-CaH2 (Figure 1). The two structures
Figure 2. Electronic band structures of MH2 hydrides with fluoritetype structures. (a) Hypothetical c-CaH2, (b) ScH2, and (c) TiH2.11 The Brillouin zone for the cubic F-lattice, along with labels for the special k-points, is shown in Figure S1 of the Supporting Information. In the band diagrams, the contribution from M 3d eg orbitals are highlighted with a fat-band representation in red. Schematic orbital interactions at the points labeled 1−4 in the band diagram of CaH2 are shown in panel d. The gray, green, or blue orbitals belong to the ions at z = 0, z = 1/2, or z = 1/4 (or 3/4), respectively.
Figure 1. Crystal structures of CaH2. (a) Orthorhombic phase (PbCl2type) and (b) a hypothetical cubic fluorite-type phase, overlaid with a 3D electron density isosurface map of the DFT band structure in region from ECBM to ECBM + 0.1 eV at 1.2 × 10−4 Å−3. (c) 2D cross section along the (110) plane of the partial electron density map displayed as an isosurface in panel b.
experimentally determined crystal structures11) for comparison. In these panels, fatbands are used to highlight the orbital contributions from the Metal 3d eg, the most likely contributor to the conduction band. The band structure of these three hydrides is very similar to that we presented for the isostructural LnH2 in a previous report,12 except for the position of Fermi energy (EF). To understand the specific chemical bonding appearing in c-CaH2, we focus on four key points labeled in Figure 2a, the orbital character for each of which is shown in Figure 2d. The occupied states are based on the H 1s, ranging from the H 1s−Ca 4s σ bonding state indicated as point 1 (Γ, −6.5 eV) to the H 1s−H 1s σ* antibonding state comprising the valence band maximum (VBM) at point 2 (Γ, 0 eV), respectively. The full bandwidth related to the occupied H 1s states, ∼6.5 eV, derives mainly on the H 1s−H 1s interaction, as well as stabilization by interactions with the Ca 4s. Though both the maximum and minimum of the valence band occur at the Γ point, the conduction band minimum (CBM) is located at the X point (point 4), indicating that c-CaH2 is an indirect type semiconductor with the calculated bandgap 0.2 eV. At the X point, contributions from the Ca 3d are evident near EF. Ca 3d t2g−H 1s σ interaction happens in the valence band at point 3 (X, −2.9 eV). The remaining Ca 3d orbitals, the eg set, form unique multicenter bond through the cavity surrounded by eight H− anions (point 4, +0.2 eV, CBM). At this point, Ca−Ca interactions can give rise to a low-lying bonding state without interference with H 1s, due to the restrictions of orbital symmetry. To better see the character of the band near the CBM, electron density maps in the energy region from ECBM to ECBM +0.1 eV are shown in Figure 1b,c. Notably, the
bear a close relationship suggesting that their energy difference could be small. In these two structures, the Ca ion sublattices adopt the two simplest variants of close-packing the hexagonal closest packed (hcp) and cubic closest packed arrangements, respectively, in fact correspond to the β- and α-polymorphs of elemental Ca. These two arrangements offer different interstitial spaces for the H ions to occupy. In o-CaH2, there are two crystallographic sites for hydrogen, H1 and H2, which are coordinated with four and five Ca ions, respectively. Each Ca2+ ion is then bound to nine H− ions. In c-CaH2, on the other hand, the H atoms occupy the tetrahedral holes of the Ca sublattice (sites we label as Ht due to their Td site symmetry) This arrangement creates a H sublattice simple cubic structure, half of whose cubic holes are empty, leaving a tiny cavity site, V, with Oh site symmetry at the center of the fcc unit cell. To compare relative energies of these polymorphs, DFT periodic calculations including structural relaxation were performed. (see Table S2 in the Supporting Information). The total energy (Etot) values indicate that o-CaH2 is slightly stable rather than cCaH2 at 0 K. As MH2 compounds of alkaline-earth are generally considered to be ionic compounds, the lattice energy values were also calculated using MADEL10 for comparison. The results suggest that electrostatic interactions favor o-CaH2. This idea is supported by the larger bond valence sum of the Ca ions in o-CaH2, indicating the electrostatic effect of ionic packing, that is, atomic density, is a crucial factor for competition between these crystal structures. These findings agree well with the tendency of pressure-induced phase transitions in ionic systems, which point to o-CaH2 being a 11318
DOI: 10.1021/jacs.7b05746 J. Am. Chem. Soc. 2017, 139, 11317−11320
Communication
Journal of the American Chemical Society accumulation of electron density from the Ca 3dy2 orbital at the X point, together with other Ca 3d eg orbitals at the equivalent k-points, results in the appearance of the electron density peak at the cavity site. This can be regarded as being based on multicenter bonding states involving six Ca ions. This electronic structure makes a sharp contrast with most compound semiconductors such as those of the III-V and IIVI types. In these more typical cases, the CBM states are mainly focused on the constituent ions (especially on the cations). Cavity spaces rarely play roles in the CBM or other electronically active states. Instead, the shape of the CBM electron density is very similar to that of CBM of the antifluorite-type Mg2Si semiconductor,13 or the states near the EF of the LaH2 electride.12 Here, we emphasize the similarity with electride compounds. In inorganic electrides, the crystallographic interspaces are occupied by anionic electrons, which have high chemical reactivity owing to their small work functions.14 We can find systematic variation of the band structure from going from c-CaH2 to MH2 fluorite-type phases with M = Sc, Ti, or V (see Table S1). Following the general periodic trends, we expect that M moves to right in the periodic table, its 3d orbital energy level will deepen, while its band filling will increase. At the same time, the cationic size of the M site will decrease. All of these effects are evident in Figure 2, where the 3d orbital band energies drops from Ca to Sc, and to Ti. Simultaneously, the decrease of lattice sizes due to the variation of cationic size enhances the interaction between H ions, resulting in the increase of the VB width of H 1s. As combined result of these effects, the VBM, that is, H 1s−H 1s σ* antibonding state starts to overlap energetically with the CBM, leading to the indirect bandgap closing. Simultaneously, the EF climbs to higher bands as the electron count rises. On the basis of these calculations, we attempted the synthesis of CaH2 by ionic substitution via solid state reactions, and the details are given in the Supporting Information. In our first experimental approach, we attempted anionic substitution of CaF2 by H− ion. The powder X-ray diffraction (XRD) patterns in Figure S2a in the Supporting Information confirm the formation of the cubic phases, Ca(F1−xHx)2, as reported in the literature.15 The systematic shift of lattice constants evident in Figure S2b supports the formation of solid solution in the range, 0 < x < 0.6. These lattice constants almost agree with the value obtained from the DFT geometrical optimization of cCaH2 (5.433 Å). All of the samples appeared white in color and were insulating. Figure 3a shows the diffuse reflectance spectra of the solid solutions along with spectra of some related phases. As shown in Figure 3b, the absorption edge of the solid solution is shifted drastically to lower energy by H−substitution, with the bandgap reached getting as low as 3.5 eV for Ca(F0.4H0.6)2. This decrease in bandgap originates mainly from the shallower VB energy position of the filled H 1s level relative to that of the F 2p. Notably, the estimated bandgap of Ca(F0.4H0.6)2 (3.5 eV) is not only smaller than that of CaF2 but also that of o-CaH2 (4.4 eV), which agrees well with the tendency revealed by DFT calculations in Table S2. This reflects the differences in electronic structure achievable through the polymorphism of CaH2, especially, in the CB electronic structure derived from the Ca 4s/3d orbitals. In our efforts to synthesize c-CaH2, we also tried cationic substitution of o-CaH2 using MH3 (M = Y or La). As shown in Figure S3 of the Supporting Information, the powder XRD patterns for the Ca0.87Y0.13Hx and Ca0.84La0.16Hx compounds obtained can be indexed with cubic F-centered lattices. The
Figure 3. (a) Diffuse reflectance spectra of o-CaH2, c-Ca(F1−xHx)2 (0 < x < 0.6), and c-(Ca0.84La0.16)H2.12. (b) Estimated bandgaps. Those of Ca0.84La0.16H2.12 and o-CaH2 are shown by a green dashed line and blue square, respectively. (c) Photo of o-CaH2 and Ca0.84La0.16H2.12 pellets.
cationic compositions measured by EPMA agreed well with nominal ones. Figure S4 in the Supporting Information shows the thermal desorption spectrum (TDS) of the Ca0.84La0.16Hx powder, which yielded the chemical composition Ca0.84La0.16H2.12, which is consistent with an assignment of an approximately +3 the valence state to the La. The XRD pattern of the cubic phases are similar to that of α-Ca with fcc-type structure. Thus, the positions of H ions were determined via neutron powder diffraction (NPD) analysis, in which the total H-content is fixed to the value from the TDS measurement. Results of the Rietveld structure refinements for Ca0.84La0.16H2.12 are given in Figure 4, Tables S3 and S4, and
Figure 4. Rietveld refinement results for the structure Ca0.84La0.16H2.12 at 300 K from TOF-NPD data. Observed (gray dots), calculated (black line), and difference profiles are presented. The vertical bars at the bottom show the calculated positions of the Bragg reflections of Ca0.84La0.16H2.12.
CIF file in the Supporting Information. The best fitting was obtained using the model based on fluorite-type structure, in which H partially at the V site with 16% occupancy. In this way, the excess H atoms needed to balance the charges of the La3+ cations are accommodated by the V sites. The formation of these c-Ca1−xMxH2+x compounds by cationic substitution reminds us yttria stabilized zirconia (YSZ). In contrast with white color of o-CaH2, the obtained M-substituted cubic phases appear yellow in color (Figure 3c) and are insulating. The diffuse reflectance spectrum is presented in Figure 3a,b alongside those measured for c-Ca(F1−xHx)2 Notably, the estimated bandgap of 2.5 eV is much smaller than that of Ca(F0.4H0.6)2 (3.5 eV), and agrees well with the apparent color. The calculated bandgap of c-CaH2 from DFT, 0.2 eV (Table S2) is far smaller than this experimental value, which is 11319
DOI: 10.1021/jacs.7b05746 J. Am. Chem. Soc. 2017, 139, 11317−11320
Communication
Journal of the American Chemical Society attributed to the well-known bandgap problem of DFT.16 As far as we know, c-Ca1−xMxH2+x has the smallest bandgap among the hydrides of alkali and alkaline-earth metals. As shown in Figure 3a, the optical spectra of all the cubic phases stand up slowly around bandgaps. This feature originates from the indirect type transition, which is shown in the band structure of Figure 2a. In conclusion, CaH2-based having fluorite-type structures were synthesized by partial substitution of Ca2+ with La3+. The obtained yellow insulating sample has bandgap of ∼2.5 eV, which is much smaller than that of o-CaH2 (4.4 eV). Analysis of band structure by DFT calculations indicates that Ca 3d orbitals play an important role near EF in c-CaH2. The CBM state is formed mainly by the interaction between Ca 3d eg around crystallographic cavity spaces.
■
(11) (a) Venturini, E. L.; Morosin, B. Phys. Lett. A 1977, 61, 326− 328. (b) Irving, P. E.; Beevers, C. J. Metall. Trans. 1971, 2, 613−615. (12) Mizoguchi, H.; Okunaka, M.; Kitano, M.; Matsuishi, S.; Yokoyama, T.; Hosono, H. Inorg. Chem. 2016, 55, 8833−8838. (13) Mizoguchi, H.; Muraba, Y.; Fredrickson, D. C.; Matsuishi, S.; Kamiya, T.; Hosono, H. Angew. Chem., Int. Ed. 2017, 56, 10135− 10139. (14) (a) Dye, J. L. Acc. Chem. Res. 2009, 42, 1564−1572. (b) Matsuishi, S.; Toda, Y.; Miyakawa, M.; Hayashi, K.; Kamiya, T.; Hirano, M.; Tanaka, I.; Hosono, H. Science 2003, 301, 626−629. (c) Miao, M.; Hoffmann, R. J. Am. Chem. Soc. 2015, 137, 3631−3637. (15) (a) Brice, J. F.; Courtois, A.; Aubry, J. J. Solid State Chem. 1978, 24, 381−387. (b) Leveque, R.; Zanne, M.; Vergnat-Grandjean, D.; Brice, J. F. J. Solid State Chem. 1980, 33, 233−243. (16) Jones, R. O.; Gunnarsson, O. Rev. Mod. Phys. 1989, 61, 689− 746.
ASSOCIATED CONTENT
* Supporting Information S
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b05746. Geometry optimization for orthorhombic and cubic CaH2, Brillouin zone for cubic F-lattice, powder XRD patterns for Ca(F1−xHx)2 and Ca1−xMxH2+y (M = Y or La), and TDS spectrum for Ca0.84La0.16H2+y (PDF) Data for Ca0.84La0.16H2.12 (CIF)
■
AUTHOR INFORMATION
Corresponding Authors
*
[email protected] *
[email protected] ORCID
Hiroshi Mizoguchi: 0000-0002-0992-7449 Takashi Honda: 0000-0003-4121-8957 Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This study was supported by the JSPS KAKENHI Grant (No. 17H06153); by the JST ACCEL project; by the Neutron Scattering Program Advisory Committee of IMSS, KEK (Proposal No. 2014S06). We are grateful to Drs. D. C. Fredrickson, S. Matsuishi, and M. Kitano (Tokyo Institute of Technology) for useful discussion.
■
REFERENCES
(1) Messer, C. E. J. Solid State Chem. 1970, 2, 144−155. (2) Shannon, R. D. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976, 32, 751−767. (3) Wells, A. F. Structural Inorganic Chemistry, 5th ed.; Oxford University Press: Oxford, 1984. (4) Leger, J. M.; Haines, J. Eur. J. Solid State Inorg. Chem. 1997, 34, 785−796. (5) Speziale, S.; Duffy, T. S. Phys. Chem. Miner. 2002, 29, 465−472. (6) Ye, X.; Hoffmann, R.; Ashcroft, N. W. J. Phys. Chem. C 2015, 119, 5614−5625. (7) Tse, J. S.; Klug, D. D.; Desgreniers, S.; Smith, J. S.; Flacau, R.; Liu, Z.; Hu, J.; Chen, N.; Jiang, D. T. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 75, 134108. (8) Okamoto, H. J. Phase Equilib. Diffus. 2015, 36, 292−303. (9) Kitano, M.; Inoue, Y.; Ishikawa, H.; Yamagata, K.; Nakao, T.; Tada, T.; Matsuishi, S.; Yokoyama, T.; Hara, M.; Hosono, H. Chem. Sci. 2016, 7, 4036−4043. (10) Momma, K.; Izumi, F. J. Appl. Crystallogr. 2011, 44, 1272−1276. 11320
DOI: 10.1021/jacs.7b05746 J. Am. Chem. Soc. 2017, 139, 11317−11320
