A Partial Anion Disorder in SrVO₂H Induced by Biaxial Tensile Strain

Abstract

SrVO₂H, obtained by a topochemical reaction of SrVO₃ perovskite using CaH₂, is an anion-ordered phase with hydride anions exclusively at the apical site. In this study, we conducted a CaH₂ reduction of SrVO₃ thin films epitaxially grown on KTaO₃ (KTO) substrates. When reacted at 530 °C for 12 h, we observed an intermediate phase characterized by a smaller tetragonality of c/a = 0.96 (vs. c/a = 0.93 for SrVO₂H), while a longer reaction of 24 h resulted in the known phase of SrVO₂H. This fact suggests that the intermediate phase is a metastable state stabilized by applying tensile strain from the KTO substrate (1.4%). In addition, secondary ion mass spectrometry (SIMS) revealed that the intermediate phase has a hydrogen content close to that of SrVO₂H, suggesting a partially disordered anion arrangement. Such kinetic trapping of an intermediate state by biaxial epitaxial strain not only helps to acquire a new state of matter but also advances our understanding of topochemical reaction processes in extended solids.

1. Introduction

Mixed-anion compounds have been attracting a great deal of attention due to their potential applications such as photocatalysts, anionic conductors, and thermoelectric materials [1]. In terms of structural chemistry, what distinguishes mixed-anion compounds from simple oxides is a heteroleptic coordination environment around a metal centre, which induces new degrees of freedom including cis and trans geometries for the MO₄X₂ octahedron. For example, the perovskite oxynitrides AMO₂N (e.g., SrNbO₂N, SrTaO₂N [2], and BaTaO₂N [3]) are composed of only cis-MO₄N₂ octahedra, which occurs in order to maximize more covalent M d_π–N p_π bonds [2]. This local constraint gives rise to correlated disorder with various types of zigzag –M–N–M–N– chains assembled within the perovskite framework [4], as originally stated by Pauling for the structure of ice [5]. Other mixed-anion oxides such as BaScO₂H, prepared under high pressure, and SrFeO₂F, prepared by the fluorination of SrFeO₂, exhibit a cis preference, but only partially [6,7].

Local coordination geometry can also be directed by the d-electron count. The vanadium oxyhydride SrVO₂H (V³⁺; d²) (Figure 1b), prepared by topochemical hydride reduction of perovskite SrVO₃ (Figure 1a) contains only trans-VO₄H₂ octahedra, as triply degenerate t_2g states are lifted by the trans-structure, leading to a stable (d_xz/yz)²(d_xy)⁰ configuration [8] as shown in Figure 1d. The resulting fully anion-ordered structure of [SrH] and [VO₂] layers can be regarded as a half-filled (d_xz/yz)² system. Hydride anions at the apical sites function as “π-blocker” ligands [9], making this oxyhydride a quasi-two-dimensional Mott insulator [10]. A partial trans-preference has been proposed for the 6H-type perovskite BaVO_3−xH_x (0.5 < x < 0.9) where the d-electron count, n, is 1.5 < n < 1.9 [11].

Transformation from cis to trans or vice versa is an important and challenging issue that may allow for tuning chemical and physical properties. As mentioned above, the TaO₄N₂ octahedra in ATaO₂N are exclusively of cis configuration, but the short-range-ordered state with zigzag –M–N–M–N– chains is not polar (or ferroelectric), which is unfortunate given the more polarizable nature of nitride anions compared with oxide anions. However, the recent thin film study of SrTaO₂N and its Ca-substituted solid solution has shown that the application of compressive biaxial strain from a SrTiO₃ (STO) substrate induces a partial transformation to trans [12,13], leading to ferroelectric behavior [12].

The present study addresses a case where trans-VO₄H₂ octahedra in SrVO₂H are converted partially to cis coordination. It has been previously shown that SrVO₂H thin films, under small compressive strain (−0.7%) from SrTiO₃ (001) (a = 3.905 Å) substrates have a fully anion-ordered phase (with 100% trans-coordination) [14], as in the bulk case [8]. Here, we applied tensile strain (1.4%) by using a KTaO₃ (001) substrate (KTO, a = 3.989 Å). While the reaction of the SrVO₃/KTO thin film with CaH₂ for 24 h gave the known SrVO₂H phase, reacting it for 12 h gave rise to an intermediate state, which is potentially a partially anion-disordered SrVO₂H, with [SrH_1−xO_x] and [VO_2−xH_x] layers. We discuss the possible origin of the formation of the intermediate phase in view of the kinetics.

2. Results and Discussion

2.1. Low-Temperature Reduction of SrVO₃ Films

SrVO₃ films with a thickness of about 100 nm were grown on KTO substrates and were reacted with CaH₂ at 530 °C for 12 and 24 h, respectively. The lower reaction temperature compared to that of the powder samples (610 °C) [8] is due to a short anion diffusion path for thin films, as in the case of SrVO₂H/STO [14]. The reduced films were investigated using high-resolution X-ray diffraction (XRD). The out-of-plane θ–2θ XRD scans are shown in Figure 2a, along with the as-deposited film (i.e., before reduction). The 2θ peak at 47.4°, corresponding to c = 3.83 Å, can be assigned to the 002 reflection of SrVO₃ (a = 3.841 Å for bulk [15]). Both reduced samples exhibited only 00l peaks, indicating a good orientation of the films along the c axis. For the 24h-reacted SrVO₃ film, the 002 peak is significantly shifted to a higher angle of 49.4° corresponding to c = 3.68 Å, which is close to that of bulk SrVO₂H (c = 3.6671(3) Å [8]). Most remarkably, the 002 peak for the 12h-reacted film appears between these two peaks. This peak is absent in bulk [8] and thin film (on the STO substrate) [14] samples of reduced SrVO₃, suggesting the emergence of an intermediate phase. This intermediate phase could not be assigned simply as SrVO_3−δ with random anion vacancies, because the 00l peaks should appear at lower angles than for SrVO₃ [8,14]. The FWHM of the θ–2θ peaks for the film reacted for 12 h is broader than the film reacted for 24 h (Figure 2a), implying lower crystallinity and/or a certain compositional distribution. Note that epitaxial SrVO₂H films cannot be obtained on an (La_0.3Sr_0.7)(Al_0.65Ta_0.35)O₃ (LSAT) substrate, which provides a compressive strain of −1.7% [14], and this was reproduced in our study (not shown). It is also reported that even small tensile strain from a DyScO₃ substrate (0.3%) destabilizes the SrVO₂H phase [14].

2.2. Structural CharacterizatIon

To further investigate the crystal structures of the three films on the KTO substrates, XRD reciprocal space mapping (RSM) measurements were performed around the 103 reflection, and the results are shown in Figure 2b–d. The in-plane positions of the 103 reflections of the 12h- and 24h-reacted films are found to be near the q_x positions of the KTO substrates (shown by the dashed lines), suggesting that both films are grown almost coherently, unlike the as-deposited SrVO₃ film that is relatively relaxed. There was no domain formation or orientation change, which was often observed in strained films [16] or films after metal hydride treatments [17,18]. The estimated lattice parameters for the as-deposited SrVO₃ film are a = 3.85 Å and c = 3.83 Å, indicating a small tetragonal lattice distortion (c/a = 0.995) compared to the bulk (cubic; a = 3.841 Å [15]), although RSM showed a partial relaxation (Figure 2b). The cell constants (a = 3.95 Å and c = 3.68 Å) for the 24h-reacted film indicate that the reduced film is slightly under tension, compared with bulk SrVO₂H (a = 3.9331(4) Å, c = 3.6671(3) Å) [8]. One can therefore presume that tensile strain has been continuously applied to the films via the KTO substrate throughout the whole reaction process from SrVO₃ to SrVO₂H. Note that even if the KTO substrate is reduced, the corresponding cell expansion and resulting effect on strain should be very small, as XRD data did not show any anomalies.

The hydride exchange was examined with secondary ion mass spectrometry (SIMS). We measured two areas (100 × 100 µm²) of each film in order to investigate the hydrogen content and its distribution. Figure 3 shows the SIMS profiles of the ratio of secondary ion intensities ¹H/¹⁸O for the films after 12 h and 24 h reduction. For both films, a high hydrogen signal was observed, with clear contrast at the boundary between the substrate and the film. The hydrogen signal in the KTO substrate is much smaller than the target films, though it is 10–100 times higher than the case of LSAT [19,20], which might reflect some reactivity with CaH₂. For the film reacted for 24 h (i.e., SrVO₂H), homogeneous hydrogen distribution is confirmed, with ¹H/¹⁸O values at the two areas of 7.2 (2SD 0.5) (Figure 3, blue) and 7.5 (2SD 1.0). Most importantly, hydrogen is almost homogeneously distributed in the 12h-reacted film, with ¹H/¹⁸O values of 7.4 (2SD 1.4) (Figure 3, red) and 8.2 (2SD 0.8), similarly to those of the 24h-reacted film. This result strongly suggests that the intermediate phase has roughly the same hydride concentration as SrVO₂H, though it might have some level of anion nonstoichiometry. This interpretation is supported by the close cell volumes (57.5 ų and 57.4 ų for 12h- and 24h-reacted films, respectively). The intermediate tetragonal phase is less anisotropic with a c/a = 0.96, compared to the 24h-reacted film and the bulk sample of SrVO₂H with a c/a = 0.93.

Since the anisotropic lattice parameter of bulk SrVO₂H obviously arises from the full anion (H⁻/O²⁻) order, the most natural origin of the suppressed anisotropy for the intermediate phase would be a partial anti-site disorder, where the structure consists of alternating stacks of [SrH_1−xO_x] and [VO_2−xH_x] layers (where 0 < x < 1). One can roughly estimate the degree of order (1 − 3x/2) to be 0.57 for the 12h-reacted film (c/a = 0.96, x = 0.29) by simply interpolating the c/a value between the ‘hypothetical’ fully disordered cubic phase with [SrH_1/3O_2/3] and [VO_4/3H_2/3] layers (c/a = 1, x = 2/3), and the fully ordered one (c/a = 0.93, x = 0). The partial anti-site disorder in the intermediate phase indicates the presence of cis-VO₄H₂ octahedra (Figure 4b). The broad FWHM observed in the XRD profiles for the 12h-reacted film (Figure 2a) implies either a certain local variation in anion ordering or buckled [VO_2−xH_x] layers as observed in Ln-doped SrFeO_2+δ [21]. Further studies are necessary to clarify the anion arrangement of this intermediate phase, using, e.g., linearly polarized X-ray absorption spectroscopy as utilized to determine cis/trans-TaO₄N₂ configuration in Ca-doped SrTaO₂N [13].

2.3. The Role of Tensile Strain

The previous bulk and thin film studies showed that SrVO₂H with a fully anion-ordered (thus all trans) structure is a stoichiometric phase (or a line phase) [8,14]. A recent study showed that a small amount of Ti substitution (5%) of V is able to change anionic composition and crystal symmetry, leading to the cubic perovskite SrV_0.95Ti_0.05O_1.5H_1.5 (a = 3.888(1) Å) [22]. The present study demonstrates that an intermediate tetragonal phase appears when the SrVO₃ film on the KTO substrate reacts with CaH₂ in a short period of time (12 h), and its composition is roughly given by SrVO₂H. The structure is most likely a partially anion-disordered SrVO₂H (PAD-SrVO₂H), thus having partial cis-geometry. Given the absence of such an intermediate phase in the case of the STO substrate with compressive strain [14], the application of tensile strain by the KTO substrate potentially plays a key role in stabilizing PAD-SrVO₂H (Figure 4a).

Recall that a longer time treatment with CaH₂ (e.g., 24 h) resulted in the anion-ordered SrVO₂H (Figure 1b). This fact strongly suggests that the partially disordered phase is a kinetically trapped transient state, shown in the reaction coordinate diagram in Figure 4b. In the bulk form, the energy of PAD-SrVO₂H’s state is rather high and the potential well of this state is shallow (meaning a low activation energy to the final phase of SrVO₂H) so that hydride reduction of SrVO₃ leads rapidly to the anion-ordered SrVO₂H (Figure 4b, left). On the other hand, when tensile strain is applied, the PAD-SrVO₂H state may be kinetically trapped if the strain can stabilize this state (Figure 4b, right), but the increase in the activation energy is not high enough to stop the further reaction from occurring at 530 °C. The origin is not clear, but the stabilization of PAD-SrVO₂H by tensile strain might be rationalized from examples with other perovskites, where tensile strain tends to induce oxygen deficiencies (or hydride anions, in our case) at equatorial sites [23].

It is worth considering the reaction route from SrVO₃ to SrVO₂H, which involves H⁻/O²⁻ exchange at the surface and hydride/oxide diffusion in the film [24]. In perovskites, oxide anion diffusion occurs along the octahedral edge (the nearest-neighbor pathway). The involvement of the equatorial oxide anion can be easily understood by the low-temperature conversion from SrFeO₃ to SrFeO₂ via a SrFeO_2.5 intermediate [25]. Hopping of oxide anions (or oxygen vacancies) involving the equatorial site is also suggested in the layered perovskite oxide Sr₃Fe₂O_7−x (0 ≤ x ≤ 1) [26]. The importance of nearest-neighbor anion hopping via octahedral edges has been pointed out for BaTi(O,H)₃ [27]. Accordingly, in our case H⁻/O²⁻ anion migration predominantly takes place (via anion vacancies) along the V(O,H)₆ octahedral edge, rather than the next-nearest-neighbor hopping between the apical sites. If this is the case, when the H⁻/O²⁻ exchange proceeds and the hydride content reaches (close to) 1, a partially anion-disordered phase with a composition close to SrVO₂H will be formed, which is subsequently transformed into the fully anion-ordered SrVO₂H.

3. Materials and Methods

Epitaxial films of the SrVO₃ precursor with a thickness of ~100 nm were grown by using pulsed laser deposition on (001)-oriented KTO substrates. A target of the Sr₂V₂O₇ pellet, prepared at 800 °C in air, was ablated by a KrF excimer laser (λ = 248 nm) with an energy density of 0.6 J/cm² and 1 Hz repetition rate. The films were deposited at a substrate temperature of 700–750 °C under vacuum (~1 × 10⁻⁶ Pa). The SrVO₃ film obtained was embedded in 0.2 g of CaH₂ powder in a quartz tube, sealed under vacuum (<4 × 10⁻² Pa), and then heated at 530 °C for 12 or 24 h. After these reactions, the residual CaH₂ and the resulting CaO byproducts on the surface of the films were removed by washing with anhydrous acetone. XRD measurements were performed at room temperature with a RigakuSmartLab diffractometer and a Cu Kα radiation (Rigaku, Akishima, Japan). SIMS measurements were performed for detecting hydrogen in the films using a Cameca ims-4f-E7 SIMS (Cameca, Gennevilliers France) at Kyoto University. The primary ion was ¹³³Cs⁺ ion, accelerated at 14.5 keV to obtain depth profiles at 100 × 100 μm² areas for each film.

4. Conclusions

We found the intermediate tetragonal phase between SrVO₃ perovskite and anion-ordered SrVO₂H when the SrVO₃ thin film (epitaxially grown on the KTO substrate) was treated with CaH₂ for 12 h at 530 °C. XRD and SIMS results show that this intermediate phase is less anisotropic with c/a = 0.96 (vs. 0.93 in SrVO₂H) and similar hydrogen content, implying the formation of a partially anion-disordered phase of roughly anion stoichiometric SrVO₂H. Since a longer reaction time resulted in the known anion-ordered phase of SrVO₂H, this phase is considered an intermediate state, possibly stabilized by applying tensile strain from the KTO substrate. This phase, with partial cis-structure, is also rationalized when considering the reaction process from SrVO₃ to SrVO₂H, which should involve anion migration through the octahedral edge. Stabilization of such an intermediate state has been demonstrated in many systems tuned by various parameters. For example, high pressure and temperature can kinetically trap a metastable state of dichalcogenides, MnS₂, CuS₂, and ZnS₂ [28,29,30]. The metastable ordered state of a ternary alloy is discussed in terms of the kinetics of atomic species in crystallization [31]. In the present study, we suggest that strain engineering offers a useful strategy to kinetically trap metastable states. Despite numerous examples, topochemical reactions in extended solids have remained poorly understood. The present study sheds light on the underlying reaction mechanism.