Hydroxyperovskites: An Overlooked Class of Potential Functional Materials

Abstract

While there is enormous interest in studying oxide perovskites with stoichiometries based upon or derived from ABO₃, including oxygen-deficient compositions and organometallics, other closely related topologies have been overlooked. Hydroxyperovskites are such a group. Their structures are perovskite-like octahedral frameworks with vacant cavity A sites, and all oxygen atoms form hydroxyl groups. There are fifteen naturally occurring hydroxyperovskites and numerous synthetic analogues. There are two stoichiometries: BB′(OH)₆ and B(OH)₃. The former consist of alternating divalent and tetravalent cations (B = Mg, Ca, Mn²⁺, Fe²⁺, Co²⁺, Cu²⁺, Zn; B′ = Sn, Ge). B(OH)₃ structures have only trivalent cations (Al, Fe³⁺, Ga). The properties and behavior of solid solutions in hydroxyperovskites are largely unexplored. This article summarizes our current knowledge of the crystallography and crystal chemistry of hydroxyperovskites and suggests productive areas of research in relation to their potential as functional materials. It should be evident that much of the findings remains to be discovered.

1. Introduction

Rather than provide a comprehensive review of the literature, the aim of this article is to outline the key structural features of hydroxyperovskites with a view to their potential as functional materials in the hope that it will stimulate more research on this understudied yet fascinating group. Hydroxyperovskites are ReO₃-type frameworks of corner-linked octahedra in which all oxygen atoms form hydroxyl groups. Unlike perovskites sensu stricto, the cavity A site is vacant. This feature combined with extensive hydrogen-bonded linkages between octahedra leads to highly tilted octahedra.

There are two distinct groups that are defined by the occupancy of cation sites: (i) B(OH)₃ in which a trivalent cation occupies the B site; (ii) in which there is an alternation of B and B′ sites with B = monovalent or divalent cations and B′ = tetravalent or pentavalent cations. Figure 1 shows the structure of MgSi(OH)₆ in which MgO₆ and SiO₆ octahedra alternate, and all oxygen atoms are protonated. Fifteen naturally occurring hydroxyperovskites are known, and several synthetic hydroxyperovskites have been prepared. These are summarized in

Table 1. Compilation of natural and synthetic hydroxyperovskites.

Hydroxyperovskite	Formula	Space Group	Tilt System
bernalite	Fe(OH)3	Pmmn	a+b+c+
söhngeite	Ga(OH)3	P42/nmc	a+a+c−
dzahlindite	In(OH)3	$Im\overline{3}$	a+a+a+
schoenfliesite	MgSn(OH)6	$Pn\overline{3}$	a+a+a+
burtite	CaSn(OH)6	$Pn\overline{3}$	a+a+a+
wickmanite	MnSn(OH)6	$Pn\overline{3}$	a+a+a+
tetrawickmanite	MnSn(OH)6	P42/n	a+a+c−
natanite	FeSn(OH)6	$Pn\overline{3}$	a+a+a+
jeanbandyite	FeSnO(OH)5 *	$Pn\overline{3}$	a+a+a+
mushistonite	CuSn(OH)6	P42/n	a+a+c−
vismirnovite	ZnSn(OH)6	$Pn\overline{3}$	a+a+a+
stottite	FeGe(OH)6	P42/n	a+a+c−
nancyrossite	FeGeO6H5	P42/n	a+a+c−
zincostottite	ZnGe(OH)6	P42/n	a+a+c−
mopungite	NaSb(OH)6	P42/n	a+a+c−
synthetic	FeSnO6H5	$Pn\overline{3}$	a+a+a+
synthetic	MnSn(OH)6	$Pn\overline{3}$	a+a+a+
synthetic	CoSn(OH)6	$Pn\overline{3}$	a+a+a+
synthetic	CuSn(OH)6	Pnnn	a+b+c+
synthetic	MgSi(OH)6	P21	a−a−b+ **
synthetic	Sc(OH)3	$Im\overline{3}$	a+a+a+
synthetic	d-Al(OH)3	P212121	a+b−b−

* formula reported by [1] ** equivalent to a⁺b⁻b⁻ [2] with a change of axes.

.

Synthetic BB′(OH)₆ stannate hydroxyperovskites can been prepared easily using the following reaction, where X is the divalent cation: K₂SnO₃.3H₂O + XCl₂ → XSn(OH)6+ 2 KCl. Potassium chloride is removed by repeated washing (e.g., eight times) with an alkaline aqueous solution of 40 wt.% KOH. An analogous reaction can be written for Na₂SnO₃.3H₂O as a reagent, which has also been used. No synthesis method has yet been found for germanates. Deuterated hydroxyperovskites have been prepared using 98% D₂O as the solvent and KOD (in D₂O) as the washing agent. The original synthesis method is described by [3].

2. Framework Topologies

Examples of hydroxyperovskite topologies are shown in Figure 2. The sense of rotation of an octahedron is defined relative to the crystallographic axes of the aristotype structure. There are 23 distinct combinations of tilts of octahedra. These different combinations are described using a notation [4] in which the sense and degree of tilt are defined with reference to the three axes of the aristotype, which may or may not coincide with the axes of the unit cell of the hydroxyperovskite, e.g., they may be at 45° to aristotype axes. In this notation, the sense of tilt of successive octahedra viewed parallel to an axis (based on directions of the aristotype axes) takes a sign + or −, corresponding to in-phase or out-of-phase (“reverse”) tilts, respectively. In hydroxyperovskites, the aristotype has the space group $Pm\overline{3}$m.

This notation can be applied to hydroxyperovskites. For example, tetragonal space group P4₂/n has symbol a⁺a⁺c⁻, with two equal in-phase tilts along x and y axes and a reverse tilt along z. Cubic hydroxyperovskites (space groups Pn$\overline{3}$ and Im$\overline{3}$) have tilt symbol a⁺a⁺a⁺, i.e., three equal in-phase tilts. Tilt angles are 14–17% in hydroxyperovskites. The proportion of the unit cell occupied by polyhedra is nearly constant at 20–21% for a wide range of compositions.

3. Hydrogen Bonding Topologies

When high-quality crystals are available, it is possible not only to locate and refine H positions but also to identify partially occupied H sites by single crystal X-ray diffraction [5,6,7]. The few hydroxyperovskite structures that have been determined by neutron diffraction [8,9] confirm the existence of H sites with half occupancy.

An important effect to take into account when refining cubic and tetragonal hydroxyperovskites is merohedral twinning, as it is almost always present to a greater or lesser degree in cubic and tetragonal structures, particularly in the former. The presence of merohedral twinning in hydroxyperovskites is easily detected in difference-Fourier maps as “split” oxygen positions.

The realization that some weak peaks of unmodelled electron density in differenceFourier maps, typically with values of 0.4–0.90 electrons, correspond to partially occupied H sites is an important step towards understanding the crystal chemistry of hydroxyperovskites. The framework topologies of hydroxyperovskites have obvious sites for the formation of hydrogen bonded O-H····O bridges, as shown in Figure 2 for In(OH)₃. Hence, it is relatively easy to recognize plausible H sites in difference-Fourier (F_obs–F_calc) maps and attempt to refine their positions. An example of SHELX [10] output from the difference Fourier map, shown as a “Q-peak” list, for a H-free model of FeGe(OH)₆ is shown in

Table 2. SHELX [10] output of difference Fourier peaks for a H-free model of FeGe(OH)₆ refined to convergence after the first eight cycles of least squares. This list contains peaks, highlighted in yellow, corresponding to all five non-equivalent H sites. Note the Q1 peak has a value consistent with a nearly full site; whereas, peaks Q2, Q3, Q4, and Q6 have values nearer to half occupancy. Distances to the nearest atoms (all oxygen) range from 0.77 to 0.88 Å and are consistent with O-H bond lengths determined by X-ray diffraction. These five Q peaks were assigned to H, their occupancies fixed at 1.0 (Q1) and 0.5 (Q2, 3, 4, 6), and then added to the model for refinement.

	x	y	z	sof	Uiso	peak	atom (Å, label)
Q1	0.1930	0.0691	0.7615	1	0.05	0.90	0.88	O3
Q2	0.0267	0.2285	0.1815	1	0.05	0.56	0.77	O1
Q3	0.1698	0.2469	0.0788	1	0.05	0.53	0.87	O1
Q4	0.2685	0.5402	0.1719	1	0.05	0.51	0.84	O2
Q5	-0.0192	0.2367	0.2477	1	0.05	0.47	1.35	O1
Q6	0.2668	0.6803	0.0700	1	0.05	0.46	0.83	O2
Q7	0.2419	0.5811	0.0011	1	0.05	0.44	0.52	O2
Q8	0.0165	0.1272	0.0201	1	0.05	0.33	0.96	O1
Q9	0.3574	0.5227	0.0069	1	0.05	0.32	0.87	O2
Q10	0.0793	0.1338	0.8681	1	0.05	0.28	0.92	O3

, with full and half-occupied H sites highlighted. All five H sites of the P4₂/n structure appear in the first difference-Fourier map after eight cycles of least squares refinement. In Table 2, peak Q1 has a residual of 0.90 electrons, and peaks Q2, Q3, Q4, and Q6 have values ranging from 0.46 to 0.56 electrons. Distances of these peaks from the nearest atoms (all oxygen) are 0.77–0.88 Å and are consistent with O-H bond lengths determined by X-ray diffraction, which are typically shorter than the “true” values found by neutron diffraction. Note that peak Q5 is 1.35 Å from the nearest oxygen and, therefore, is unlikely to be an H site. Other Q peaks in the list are either too distant from or too near to oxygen atoms and/or have low intensities. The provisional assignment of half and full occupancies for the H sites in FeGe(OH)₆ from the refinement of SCXRD data has been confirmed recently by single-crystal neutron diffraction, for which refined occupancies were found to be 1.00 ± 0.02 and 0.50 ± 0.03, in complete agreement with those inferred from SCXRD. For refinements using X-ray data, the occupancies of full- and half-occupied H sites are fixed at 1.0 and 0.5, respectively. These constraints usually allow for the stable refinement of H positions and an overall atom displacement parameter, applying a soft restraint on O-H bond length, e.g., 0.85 ± 0.03 Å.

When considering the hydrogen bonding configurations of hydroxyperovskites, it is instructive to recognize the different connectivities of O-H····O bridges between octahedra. Three topologically distinct components have been found so far that arise from different tilt systems. These are illustrated in Figure 3 and comprise (i) isolated four-membered rings (squares), (ii) “crankshafts”, and (iii) chains. Structures with three in-phase tilts (not necessarily equal) + + + have only four-membered rings; those with one reverse tilt, and two in-phase tilts (+ + −) have isolated rings and crankshafts; those with one in-phase, and two reverse tilts (+ − −) lack isolated rings but have chains and crankshafts.

Hydrogen bonding topology has been shown to correlate with physical properties and structural behavior. For example, anisotropic compressional behavior has been correlated with the density and type of hydrogen-bonded linkages along different crystallographic directions. For example, in the case of FeGe(OH)₆ (tetragonal) had axial moduli K_a = 81 GPa and K_c = 73 GPa, and that the lower compressibility parallel to the a-axis correlated with the orientation of the O-H····O crankshafts parallel to this axis [11]. In contrast, there is no O-H····O connectivity parallel to the c-axis, which may lead to less structural resistance to compression. The arrangement of crankshafts and isolated squares in FeGe(OH)₆ is shown in Figure 4.

4. Ordered Structures, Substructures and Disordered Structures

Problems associated with strong pseudosymmetry, and the assignment of wrong space groups are well-known in the determination of perovskite structures. The same is true of hydroxyperovskites, for which a choice between centrosymmetric and non-centrosymmetric space groups is a recurrent challenge. Here, we distinguish between (i) “H-ordered” phases, for which structure refinement indicates fully occupied H sites; (ii) “substructures”, for which refinements indicate some or all H sites are half-occupied; and (iii) “H-disordered” phases, for which structure refinement indicates have partially occupied H sites but not half-occupied H sites.

It is becoming evident that most published structures of hydroxyperovskites with half-occupied H sites are likely to correspond to substructures. The advent of a new generation of in-house high-flux X-ray sources combined with high-resolution, hybrid photon counting detectors has allowed the detection of weak reflections associated with the ordered structures, rather than substructures, of hydroxyperovskites. These detectors have enormous dynamic ranges (e.g., 10⁹) that avoid saturation and allow longer counting times for the detection of weak reflections from the ordered structure. For example, using a combination of a high-flux microsource and a Dectris Eiger area detector, we discovered that stottite, FeGe(OH)₆, has an ordered structure with space group P4₃ and a doubled c dimension, relative to the previously recognized P4₂/n structure. The reconstructed hk0 and h0l X-ray diffraction patterns of FeGe(OH)₆ are shown in Figure 5. Numerous weak violators of the n-glide of P4₂/n are evident in the hk0 pattern, and the h0l pattern has many reflections corresponding to doubling of the c unit cell axis. The P4₃ structure of FeGe(OH)₆ derived from the X-ray data is shown in Figure 6; all H sites are fully occupied. All previous refinements of its structure were in space group P4₂/n (on a 7.550 Å × 7.550 Å × 7.466 Å cell) with four of the five non-equivalent H sites being half-occupied (the ring site is fully occupied). This result suggests that the origin of the half-occupied H sites in FeGe(OH)₆ is due to the refinement of a substructure (P4₂/n).

To date, there are only three hydroxyperovskites for which the refined structures have fully occupied H sites: FeGe(OH)₆, tetragonal space group P4₃ with a doubled unit cell (see above); synthetic d-Al(OH)₃, orthorhombic space group P2₁2₁2₁ [12]; synthetic MgSi(OH)₆, monoclinic space group P2₁ [7,13]. The structures of d-Al(OH)₃ and MgSi(OH)₆ are shown in Figure 7 and are clearly related. The alternation of Mg and Si octahedra results in the loss of two sets of screw axes relative to d-Al(OH)₃ to give a non-centrosymmetric monoclinic (but metrically orthorhombic) structure for MgSi(OH)₆. The hydrogen bonding topologies of these two structures are, nonetheless, analogous, and both involve the same combination of O-H····O chains and crankshafts. The possible significance of the P2₁ structure topology is discussed below in the section on functionality.

The structure of CuSn(OH)₆ has long been debated [14,15,16]. The original description [14] reported that the mineral is tetragonal with space group P4₂/nnm. This space group implies the significant distortion of Cu(OH)₆ and Sn(OH)₆ octahedra due to the presence of mirror planes, which bisect them.

In the absence of definitive single-crystal studies due to the extreme scarcity of natural material (the mineral mushistonite), it has been necessary to resort to using the synthetic analogue e.g., [15,16]. The synthetic products are invariably very fine-grained (1–2 mm) and, as such, are unsuitable for structure determination using in-house single crystal equipment, other than synchrotron radiation. This hydroxyperovskite had been assigned space group P4₂/n with tilt system a⁺a⁺c⁻, and its structure evaluated using Rietveld refinement of X-ray powder data in this space group [15]. Recently, a revised structure for CuSn(OH)₆ has been reported [16] based upon Rietveld refinement of X-ray and neutron powder diffraction patterns for synthetic CuSn(OD)₆ powder with all H replaced by D. The revised structure is orthorhombic with centrosymmetric space group Pnnn and tilt system a⁺b⁺c⁺. This structure has only isolated four-membered O-H····O rings, as expected for a +++ tilt system (see above). It has nine non-equivalent H sites, all of which were modelled as having partial occupancies of 0.2, 0.3, 0.4, or 0.5 (refined and then fixed at these values for stable refinement); two of the three oxygen atoms are bonded to three partially occupied H sites, summing to 0.7 H atom in both cases. This array of partially occupied H sites was interpreted as evidence of proton disorder [16].

An alternative possibility is that the Pnnn structure with partially occupied H sites may be a substructure. An ordered structure having space group Pnn2, which has the same a⁺b⁺c⁺ tilt system as Pnnn, may be an alternative structure with fully occupied H sites. A structure with space group Pnn2 is non-centrosymmetric and polar (point group mm2) and would, like MgSi(OH)₆ (space group P2₁), be of potential interest as a ferroelectric.

Given the existence of the P4₃ structure of FeGe(OH)₆, we suggest that all structures with partial, and especially half-occupied, H sites be tested against possible ordered structures. If it can be shown that no ordered candidates are consistent with the diffraction data, then there is a good basis for considering proton disorder. As far as we are aware, the Pnnn structure for CuSn(OH)₆ [16] is the only one that has partially occupied sites that are not half-occupied. In this regard, it may be a genuinely H-disordered phase. Determining the correct space group of the ordered structures of hydroxyperovskites is of fundamental importance if, as we suggest, substructures with half-occupied H sites may in reality be non-centrosymmetric and the reported centrosymmetry due to refining a substructure.

5. Novel H-Deficient Hydroxyperovskites: Fe³⁺SnO₆H₅ and Fe³⁺GeO₆H₅

Natural hydroxyperovskites with less than six H per formula unit have been recognized [1,17]. Synthetic Fe³⁺SnO(OH)₅ has also been reported [1]. The H deficiency arises from the oxidation of Fe²⁺ to Fe³⁺ with charge balance being achieved by the loss of a proton.

The original description of the structure of jeanbandyite (cubic, Pn$\overline{3}$) [18] proposed the chemical formula (Fe³⁺_1-x, ⎕_x)(Sn_1-y, ⎕_y)(OH)₆ (⎕ = vacancy). The presence of vacancies in the original study was subsequently shown to be incorrect [1], and a revised formula was proposed Fe³⁺SnO(OH)₅. However, this revised formula implies that one in six oxygen atoms is not bonded to H. This is an unlikely scenario, as such an oxygen would be very under-bonded, e.g., a bond valence sum of ~1.13 valence units (vu). There are four formula units in the unit cell of jeanbandyite and, therefore, 24 oxygen atoms. The chemical formula Fe³⁺SnO(OH)₅ implies that four of these oxygens in the unit cell will not be bonded to H.

Recently, a new natural H-deficient hydroxyperovskite Nancy-rossite has been reported [17] with chemical formula Fe³⁺GeO₆H₅. This formula was proposed, in preference to Fe³⁺GeO(OH)₅, because it avoids the “missing” hydroxyl and recognizes the possibility of novel H behavior. For example, these structures may have dynamic H disorders over all hydroxyl-like H sites. Diffraction methods see only a “static” time-averaged structure that does not allow for the recognition of dynamic behavior.

Figure 8 shows the Raman spectra in the O-H stretching region for Fe³⁺SnO₆H₅, Fe³⁺GeO₆H_5, and FeGe(OH)₆ [17]. The similarity of the spectra of the two H-deficient species, both having two very broad barely resolved bands, contrasts with the well-resolved bands of the FeGe(OH)₆ spectrum. The width of these two bands is far greater than those of the bands of FeGe(OH)₆. Furthermore, the five resolved O-H bands of FeGe(OH)₆ lie within an envelope defined by the Fe³⁺GeO₆H₅ spectrum. The significance of the spectra of Fe³⁺SnO₆H₅ and Fe³⁺GeO₆H₅ remains to be determined. One possibility is that the broad, relatively ill-defined O-H spectra of Fe³⁺SnO₆H₅ and Fe³⁺GeO₆H₅ reflect a dynamically disordered arrangement of H with multiple H sites, possibly; such a scenario contrasts with that of the ordered “static” arrangement in FeGe(OH)₆, with well-defined H sites.

A further intriguing aspect of Fe³⁺SnO₆H₅ and Fe³⁺GeO₆H₅ is that they have anomalous optical properties relative to their symmetry. Fe³⁺SnO₆H₅ is cubic but has uniaxial optics [18]; Fe³⁺GeO₆H₅ has tetragonal symmetry but is clearly optically biaxial, with an optic axial angle of 73° [17]. Anomalous optical properties have also been reported for CaSn(OH)₆ [19] and stottite [20], and so the phenomenon may be characteristic of hydroxyperovskites more generally. The origin of anomalous optical properties in hydroxyperovskites has not been investigated in detail.

6. Thermal Studies of Hydroxyperovskites

Very few studies of the structural effects of temperature on hydroxyperovskites have been made, and with the one exception [6], these only describe decomposition, which occurs at 480–530K and appears to be a stability limit for these structures at room pressure [21]. While this is a much lower thermal limit than in perovskites, it is conceivable that the activation of proton conduction occurs below 500K.

Two examples of polymorphism are known: MnSn(OH)₆ and Ga(OH)₃. Both involve cubic (Pn$\overline{3}$, Im$\overline{3}$) and tetragonal (P4₂/n, P4₂/nmc) structures. Natural samples of MnSn(OH)₆ occur as wickmanite (Pn$\overline{3}$) and tetrawickmanite (P4₂/n); the synthetic analogue of tetrawickmanite has not been reported. In the case of Ga(OH)₃, A thermal study of a natural crystal (the mineral söhngeite) using single-crystal X-ray diffraction [6], observed a transformation on heating from a tetragonal (P4₂/n or P4₂/nmc) structure to cubic (Im$\overline{3}$) at ~350K; the cubic phase is retained on cooling to 293K. This transformation involves a change in the tilt system (a⁺a⁺c⁻ → a⁺a⁺a⁺) and, therefore, a major reconstruction of the hydrogen bonding configuration, with loss of O-H···O crankshafts. The framework octahedra of the two polymorphs are almost identical, and the transformation appears to be solely driven by H behavior.

7. Hydroxyperovskites at High Pressure

Experiments at high pressure can provide key information about the energetics of interactions between coordination polyhedra, as well as transformational behavior. A number of high-pressure studies of hydroxyperovskites have been made. Compared with perovskites with an occupied cavity site and bulk moduli of 200–250 GPa, hydroxyperovskites are relatively soft structures, with bulk moduli of 70–85 GPa; CaSn(OH)₆ is exceptionally compressible (K₀ = 35 GPa) and lies well away from the well-defined linear trend of bulk modulus versus ambient unit cell volume of other hydroxyperovskites. The origin of the anomalously high compressibility of CaSn(OH)₆ is not known.

Hydroxyperovskites are stable to very high pressures (>20 GPa). The synthetic high-pressure hydroyperovskite MgSi(OH)₆ has been studied by Earth scientists, as it is considered to be important in the hosting and transport of crustal water into the Earth’s deep mantle at subduction zones [22]. This phase is quenchable from high pressure (6−14 GPa) and temperature (673−823K). At room temperature, it is stable to at least 40 GPa [23].

The high-P behavior of synthetic CuSn(OH)₆ has been studied [15] (reported as having space group P4₂/n) and (Cu_0.4Zn_0.6)Sn(OH)₆ (Pn$\overline{3}$) to 25 GPa at 293 K using synchrotron powder XRD. They found that CuSn(OH)₆ was much more compressible (K₀ = 59.7 GPa) than (Cu_0.4Zn_0.6)Sn(OH)₆ (K₀ = 75.8 GPa). This difference is due to the high compressibility of CuSn(OH)₆ parallel to the c-axis (K₀ = 38 GPa) compared to being parallel to the a-axis (K₀ = 79 GPa), the latter being in line with values for most other hydroxyperovskites. The high compressibility parallel to the c-axis is attributed to the long Cu-O bond of the octahedron being sub-parallel (22°) to the axis. The lower compressibility of (Cu_0.4Zn_0.6)Sn(OH)₆ can be ascribed to the moderating effect of Zn upon the Jahn–Teller distortion of the Cu(OH)₆ octahedron that prevents a transition to a tetragonal structure, which would allow the distortion to be developed.

8. Solid Solutions

In principle, modifying compositions by forming solid solutions can provide a means of tailoring properties for technological applications, and hydroxyperovskites likely have significant capacity for solid solution. For example, natural samples of schonfliesite, ideally MgSn(OH)₆, have substantial substitution of Mg by Mn²⁺ and Fe²⁺ [24]. However, very little is known about the crystal chemistry and thermochemistry of hydroxyperovskite solid solutions. Almost all published studies have focused on compositional endmembers in order to minimize variables for ease of interpretation. The marked effect of solid solution upon compressibility in (Cu_1-xZn_x)Sn(OH)₆ solid solutions has been described above [15]. The same study found that, at 298K, the transformation from cubic to tetragonal structures in these solid solutions occurs at a composition near (Cu_0.4Zn_0.6)Sn(OH)₆. However, cubic and tetragonal phases coexist in the range (Cu_0.5Zn_0.5)Sn(OH)₆ − (Cu_0.8Zn_0.2)Sn(OH)₆. Whether or not this two-phase range corresponds to immiscibility remains to be determined.

9. Functionality in Hydroxyperovskites

9.1. Photocatalysts and Electrolytic Cells

Several studies have recognized that hydroxyperovskites have potential applications as photocatalysts and in electrolytic cells. Photocatalysis in these materials arises from exposure to ultraviolet radiation, which stimulates the production of free hydroxyl radicals that decompose organic molecules. CoSn(OH)₆ and ZnSn(OH)₆ have been shown to perform well as photocatalysts [25]. Hydroxyperovskites are proving to have considerable potential as oxygen evolution reaction (OER) catalysts as anode materials for hydrogen fuel cells [25,26,27,28]. Discovering suitable OER anode materials is an urgent priority for their commercial development.

9.2. Low-Temperature Proton Conductors?

It is their potential as proton conductors combined with their ease of synthesis, at least for the stannates, that hydroxyperovskites can provide opportunities for exploring novel proton behavior that has functional significance. There is much interest in the development of low-temperature proton-conducting ceramics (LT-PCCs), motivated by the possibility of significantly lower manufacturing and operating costs compared with high-temperature ceramics, such as perovskites. A comprehensive review of the state-of-play of research on LT-PCCs and their potential for use as fuel cell components, sensors, and hydrogen-separation membranes has been published recently [26]. Our contention is that hydroxy-perovskites should be considered part of the LT-PCC landscape. The definition of an LT-PCC [26] is a material that conducts protons at a temperature between 298 K and 673 K.

LT-PCCs have advantageous properties, relative to high-temperature solid oxide materials, [26,27,28] that could lead to a potentially viable and commercially desirable product. Features of the hydroxyperovskite structure that suggest their relevance to LT-PCC research are (a) hydrogen bonding of moderate strength as inferred from O_D-O_A distances of 2.7–3.0 Å [29]; (b) the high probability of H hopping between double potential wells, including tunnelling [16]; (c) those with tilts other than + + + have high hydrogen bonding connectivity (crankshafts, chains), providing pathways that may facilitate proton migration.

Surface proton conductivity can be a major feature of micro-/nanoscale LT-PCCs [26]. In this regard, the fine-grained nature of synthetic analogues of hydroxyperovskites would be advantageous. The ease of synthesis of high-purity stannate hydroxyperovskites is a further attraction.

The effects of temperature on ionic conductivity in a wide range of synthetic hydroxyperovskite compositions, including their decomposition at ~500K have been reported [21]. Although it was stated that, “in hydroxide perovskites [=hydroxyperovskites], the mobility of internal protons is expected to show high proton conductivity on providing thermal energy”, no attempt to measure proton conduction in the samples was made, and so the occurrence and mechanism(s) of the proposed “high proton conductivity” were not elucidated. As far as we are aware, no further studies of proton conduction in hydroxyperovskites have been made. As such, proton conduction in fully protonated hydroxyperovskites remains to be explored.

As noted above for CuSn(OH)₆, was refined with a structure with nine H sites, all of which are partially occupied [16] and it was proposed that this configuration was evidence of proton disorder, i.e., the observed partial occupancy of multiple sites represents a time-averaged (as seen by diffraction) “dynamic” state. This would appear to contrast with the concept of an ordered structure described above, in which the picture obtained by diffraction is of H localized at H sites. However, in the CuSn(OH)₆ structure obtained, most of the partially occupied sites form hydroxyl groups, i.e., if protons are mobile, then they are associated with hydroxyl-type sites. In this sense they are localized rather than being distributed throughout the structure. Such a model is not very different from one in which the H sites are fully occupied, and there is proton hopping between hydroxyl H sites. An obvious scenario is hopping along crankshaft sites. Double potential wells are often associated with hydrogen-bonded O-H, and hopping along crankshafts via successive double wells may be a viable mechanism for proton conduction. Proton conduction may require thermal activation [27,28].

A further point is that the revised CuSn(OH)₆ structure [16] has +++ tilts and so has only isolated O-H····O rings. As such, it is possible that, if proton conduction occurs, the mechanism or pathway is different from structures having crankshafts. Clearly, this is an interesting topic that deserves more attention.

From a practical viewpoint, there are some challenges in the preparation of samples of synthetic hydroxyperovskites that are suitable for conductivity measurements. For fine-grained powders (grain sizes often < 10 mm), as are usually synthesized, this will be a challenging task, as it requires an effective means of preparing suitable void-free pellets to which electrodes can be attached. High-purity natural crystals are available for some hydroxyperovskites that are up to 1 mm in diameter [e.g., FeGe(OH)₆ and are more suitable for electrode attachment. From an applications perspective, a means of growingmillimeter-sized crystals of hydroxyperovskite would seem to be a priority.

Conventionally, protonation (“H-doping”) of perovskites involves a two-step process: (i) the synthesis of an oxygen-deficient structure, followed by (ii) the dissociation of H₂O at vacant oxygen sites to form an OH group and “free” proton, for example BaSc_0.667W_0.333O₃ → BaSc_0.8W_0.2O_2.8 → BaSc_0.8W_0.2O_2.8(OH)_0.2 + 0.2H⁺ [30]. These perovskite-based H-doped protonic conductors have applications in fuel cells and electrolytic cells.

In contrast to the approach used for perovskites, which involves adding protons, proton conduction in hydroxyperovskites, if it occurs, may be achieved by partial deprotonation, with the creation of vacant H sites that facilitate H migration. In this regard, the H-deficient hydroxyperovskites FeSnO₆H₅ and FeGeO₆H₅ could provide insights into H behavior that relate to the potential of such materials as proton conductors. Oxidized FeSn(OH)₆, FeSnO₆H₅, has been synthesized [31], and it seems reasonable to expect other transition element stannates to oxidize to analogous compounds under appropriate conditions. Other candidates are CoSn(OH)₆ and MnSn(OH)₆, both of which are easily synthesized.

9.3. Ferroelectrics?

So far, only one hydroxyperovskite with a polar space group has been found, MgSi(OH)₆, which has space group P2₁. Nonetheless, its occurrence indicates the possibility of ferroelectric behavior in hydroxyperovskites. As noted above, d-Al(OH)₃ has a structure that is closely related to that of MgSi(OH)₆ from which it can be derived by the loss of two screw axes due to the alternation of Mg(OH)₆ and Si(OH)₆ octahedra. As such, the d-Al(OH)₃ structure could form the basis for deriving compositional variants that have cation ordering (e.g., alternating sites) that lowers symmetry to P2₁; the preparation of a d-(Al, Fe³⁺) (OH)₃ series is an obvious possibility to attempt, as the cations have contrasting properties that may induce ordering.

10. Conclusions

This paper has outlined the key crystallographic features of hydroxyperovskites and highlighted those that are relevant to their potential application as functional materials. It will be evident that, despite recent progress in characterizing the structural states and some properties of hydroxyperovskites, there is much to discover. Based upon the current understanding of their crystal chemistry, we have proposed several avenues for exploring their properties and potential as functional materials. In addition to their recently recognized function as photocatalysts, we suggest that they are plausible candidates for being novel proton conductors, both as fully protonated and partially protonated structures. Distinguishing between H-ordered structures, substructures, and genuinely H-disordered structures is a priority for recognizing likely proton conductors as well as for identifying mechanisms and pathways of proton conduction. The significance of non-centrosymmetry of the H-ordered structures, as opposed to centrosymmetric substructures, is ripe for investigation, particularly with a view to identifying polar structures and synthesizing ferroelectrics. A key step will be to prepare synthetics or acquire natural samples that are suitable for measuring physical properties, such as conductivity.