Rational design of chloride ion transport channels in an open borate framework

Abstract

Chloride-ion solid electrolytes have been widely investigated as the core components of membranes, gas sensors, and all-solid-state batteries with high energy densities. Several fast chloride-ion solid electrolytes operating at ambient-to-intermediate temperatures have been reported, ranging from inorganic metal chlorides and metal–organic compounds to complexes. However, these chlorides have the major drawback of being thermally and chemically unstable, and design strategies for chloride-ion conductors have not yet been developed compared with those for cation conducting solids, which restricts their practical applications. Here, we report that water-insoluble and thermally stable borate chloride (Ca_1−xLa_x)₂B₅O₉Cl_1+2x exhibits one-dimensional chloride-ion transport defined by the dimensionality of an open borate framework. Experimental and first-principles molecular dynamics simulations indicate that the predominant chloride-ion conduction is attributed to cooperative diffusion through interstitial chloride sites. This conduction is also strongly influenced by the association with La species. These results show that the rational design of chloride-ion channels and conduction paths based on the dimensionality of the borate framework would provide a new direction for the development of the variety and conducting properties of chemically and thermally stable chloride-ion solid electrolytes.

---

Introduction

Solid electrolytes, in which specific ions can migrate as charge carriers, have been extensively investigated because of a wide range of attractive applications, including sensors, separators, thermoelectric convertors, and all-solid-state batteries.^1–4 The types of solid electrolytes range from inorganic salts, polymers, and complexes to metal–organic frameworks, among which inorganic solid electrolytes have the advantages of low-to-high temperature operation owing to their high thermal and chemical stability, as exemplified by yttria-stabilized zirconia for solid oxide fuel cells,⁵ Na-β alumina for sodium-sulfur batteries,⁶ and Li₆PS₅Cl for Li-ion batteries.⁷ Recently, increasing attention has been paid to anion-conducting solid electrolytes, particularly for new types of rechargeable batteries that transfer anion charges, such as H⁻, F⁻, and Cl⁻.⁸ These anions are not only abundant elemental resources available worldwide but also offer the potential to create a new range of applications that considerably differ from conventional ones.

At present, the variety of chloride-ion conducting solid electrolytes is poor in comparison with the rich variety of cation-conducting solid electrolytes.^9–11 However, inorganic metal chlorides, typically simple binary and ternary chlorides, have been continuously investigated over the last century. In particular, PbCl₂ (cotunnite-type structure) and Cs(Sn/Pb)Cl₃ (perovskite-type structure) achieve high ionic conductivities via cation substitution even near room temperature, for example, 2.3 × 10⁻⁴ S cm⁻¹ at 160 °C for Pb_0.98K_0.02Cl_1.98 ¹² and 1.3–3.45 × 10⁻⁴ S cm⁻¹ at 25 °C for CsSn_0.9Bi_0.1Cl_3.1,¹³ CsSn_0.9In_0.067Cl₃,¹⁴ and CsSn_0.95Mn_0.05Cl₃.¹⁵ These chlorides have also been reported to exhibit promising electrochemical properties for all-solid-state chloride-ion batteries (CIBs). However, the major drawback of these chlorides, as well as other metal chlorides, is their low thermal and chemical stability, including low melting points (e.g. 501 °C for PbCl₂¹⁶ and 368 °C for CsSnCl₃ ¹⁷) and high water solubility.¹⁸ For perovskite-type chlorides, a cubic-to-monoclinic phase transition occurs under humid conditions, resulting in a significant reduction in chloride ion conductivity. In contrast, a layered mixed-anion compound La_1−xCa_xOCl_1−x (0 ≤ x ≤ 0.2) with a tetragonal matlockite-type structure possesses great structural stability against humidity and high temperature.¹⁹ Moreover, La_0.8Ca_0.2OCl_0.8 (x = 0.2) prepared via ball milling exhibited chloride-ion conductivity as high as 7 × 10⁻⁴ and 1 × 10⁻² S cm⁻¹ at 400 and 700 °C, respectively, which are comparable to those of PbCl₂ and CsPbCl₃. Although the lowest operating temperature is higher than those of doped PbCl₂ and CsSnCl₃, anion lattice engineering using both oxide and chloride ions is an attractive approach for the design of new chloride-ion solid electrolytes with high structural stability and high ionic conductivity even at moderate temperatures. On the other hand, making Cl⁻ transport dominant in mixed-anion sublattices, where the migration barrier for Cl⁻ ions tends to be higher than that for O²⁻ ions, remains a major challenge.^20,21 Indeed, even layered oxychlorides such as Ba₃M₂O₅Cl₂ (M = Y, Sc),²² Sr₂ScO₃Cl,²² Bi_4−xSr_xNbO_8−dCl,²³ and MBi_2−xTe_xO_4+x/2Cl (M = La, Lu),^24,25 which adopt two-dimensional (2D) chloride layers similar to those in LaOCl, favor oxide-ion conduction.

Crystalline borates are well known to exhibit a rich variety of open frameworks with 0D–3D dimensionality, which are mainly composed of triangular BO₃ and/or tetrahedral BO₄ units.²⁶ The negatively charged borate framework typically forms open channels at the molecular scale, which serve as conducting pathways for mobile cations, including Li⁺ and Na⁺.^27–30 Importantly, borate frameworks rarely produce oxygen deficiencies due to the strong covalent interactions between B and O atoms,³¹ and therefore no oxide-ion conduction can be expected. To examine possible chloride ion conduction in open framework borates, we chose a hilgardite-type oxychloride, Ca₂B₅O₉Cl,³² which is water-insoluble and thermally stable even at 700 °C (ESI, Fig. S1†). This borate chloride has been studied for applications of deep UV non-linear optics and photoluminescence,^33–35 but not as an ionic conductor. Each Cl⁻ ion forms a flattened tetrahedron with Ca²⁺ ions in the ab plane, confining the chloride ions to one-dimensional channels. In this paper, we report that La-doped Ca₂B₅O₉Cl exhibits one-dimensional chloride-ion conduction within the (B₅O₉)³⁻ borate framework, which reaches 1.32 × 10⁻⁴ S cm⁻¹ at 700 °C with 5% La doping. Single-crystal structure analysis revealed an increased concentration of Cl⁻ ions by La³⁺ doping in chloride-ion transport channels, involving significant chlorine deficiencies at the original site and site disorder along the channel directions. First-principles molecular dynamics simulation indicated one-dimensional diffusion of chloride ions through the interstitialcy mechanism influenced by the association between La and Cl defects. The open borate framework would provide a useful tool for rational design of chloride-ion channels and conduction paths, further developing chloride-ion conductors.

Results and discussion

Colorless transparent single crystals of Ca₂B₅O₉Cl were grown by the self-flux method using a CaCl₂ molten salt and extracted by washing the products with water (Fig. S2†). The crystal structure of Ca₂B₅O₉Cl, determined by single-crystal structure analysis at room temperature, is shown in Fig. 1. The borate chloride crystallizes in an orthorhombic cell (space group Pnn2) with cell parameters a = 11.3572(5) Å, b = 11.2309(4) Å, c = 6.3500(3) Å, and V = 809.95(6) ų. The crystallographic data including the atomic coordinates (Tables S1–S3†) are consistent with previous reports.³² We notice that the Cl1 atoms placed at 2b Wyckoff positions are significantly underbonded against the surrounding Ca atoms compared with the Cl2 atoms at 2a positions: the bond-valence-sum (BVS) value for the Cl1 atom is 0.69, considerably smaller than 0.98 for the Cl2 atom, reflecting the difference in the ClCa₄ tetrahedral volume. The underbonding state of the Cl1 atom would be favorable for Cl⁻ ion migration. However, the nearest neighbor (NN) distance between Cl ions in each 1D open channel is 6.3500(3) Å, much longer than those of fast chloride ion conductors (for example, 3.413(6) Å for PbCl₂ and 3.416(3) Å for LaOCl).^36,37 Therefore, the introduction of interstitial chloride ions or site disorder into chloride ion channels is essential for facilitating Cl⁻ ion migration in Ca₂B₅O₉Cl. To examine the possibility of additional chloride ion insertion via the aliovalent substitution of Ca²⁺ by La³⁺, a flux method using a CaCl₂ molten salt was employed again under conditions similar to those of undoped Ca₂B₅O₉Cl, but La₂O₃ was used as a La source with an atomic ratio of Ca/La = 31/1. La-doped Ca₂B₅O₉Cl single crystals with similar morphologies and dimensions were obtained (Fig. S2†). SEM-EDX elemental analysis estimated the atomic ratio of Ca : La to be 1.80 : 0.21, that is, approximately 10% substitution of La for Ca (Fig. S3†). The cell volume was also found to increase by 1.2% (a = 11.3807(5) Å, b = 11.3109(6) Å, c = 6.3692(3) Å, and V = 819.88(7) ų), thereby maintaining structural symmetry. The cell expansion can be explained by considering the introduction of La³⁺, which has a larger ionic radius (1.30 Å) than Ca²⁺ (1.26 Å),³⁸ and the insertion of additional chloride ions, as discussed below. Single-crystal structure analysis at the initial stages could readily identify the borate framework composed of (B₅O₉)³⁻ units and the random occupations of two nonequivalent 4c sites by Ca and La atoms in ratios of 0.89/0.11 (Ca1/La1) and 0.96/0.04 (Ca2/La2), respectively. The total atomic ratio of Ca to La determined by SCXRD was consistent with the results of the EDX measurements. Further structural analysis revealed not only a substantial reduction in the occupancy (g = 0.65) of the Cl1A atom corresponding to the Cl1 atom in the undoped phase but also the presence of interstitial chloride ions within the Cl1 channels at Cl1B: 1.10(4) Å, Cl1C: 2.12(2) Å, Cl1D: 1.10(2) Å, and Cl1E: 2.10(4) Å from the Cl1A site, which are even shorter than the NN distances between the chloride ions in the undoped phase. The site occupancies of Cl1B–Cl1E were refined to 0.102, 0.216, 0.214, and 0.123, respectively, based on the overall charge valence. The chloride site disorder in the Cl1 channels contrasts with the absence of interstitial chloride ions and almost no Cl2 site vacancies (g = 0.954) within the Cl2 channels. The final refined chemical composition was (Ca_0.92La_0.08)₂B₅O₉Cl_1.16. The final refined crystal structure and crystallographic data are shown in Fig. 1 and Tables S1, S4, and S5.†

Fig. 1 Crystal structures of Ca₂B₅O₉Cl (left panels) and (Ca_0.92La_0.08)₂B₅O₉Cl_1.16 (right panels) and the local coordination environment around Ca²⁺ ions, highlighting the Cl⁻ ion distribution. Site occupancy factors for the Ca/La and Cl atoms are presented in parentheses.

To understand the defect chemistry of Ca₂B₅O₉Cl, the formation energies of point defects were investigated using first-principles calculations. Fig. 2a shows the plots of the calculated formation energies against the variation of the Fermi level under the assumed equilibrium condition of oxidation and chloridation limits. For each defect type, the most stable sites are depicted. The gradient of the lines in the plot indicates the charge states of defects. A positive (negative) slope of the lines of defect formation energy indicates a donor (acceptor)-type defect. The Fermi level position in the band gap balances the concentrations of positive and negative point defects and electronic carriers (holes and electrons) to meet the condition of charge neutrality. In the Ca₂B₅O₉Cl system, the concentrations of positive and negative point defects determine the Fermi level because of its high insulation (Fig. S4†). As seen in Fig. 2a, and have the lowest formation energies for positive and negative point defects, respectively, and thus, these defects should be dominant over the other defects. and stand for a La atom on a Ca site with single positive charge and a Cl atom on an interstitial site with single negative charge, respectively, in Kröger–Vink notation.

Fig. 2 (a) Formation energies of point defects as a function of the Fermi level on oxidation and chloridation limits. For each defect type, only the most stable sites are depicted. (b) Obtained structure with the most stable . (c and d) Obtained structure seen from the a-axis of perfect crystal and point defect models with . Only Cl and Ca are drawn in the range of the internal coordinates of the a-axis from 0.6 to 1.0. The structure models were drawn using the VESTA program.³⁹

The association energy between and was also estimated to be 0.70 eV from all possible configurations of in the cell containing the most stable . The formation energies of the complex defects of and are lower than those of isolated and under the condition of charge neutrality, suggesting the presence of these associated defects in La-doped Ca₂B₅O₉Cl. Fig. 2b shows the calculated structure containing the most stable , which is located in the Cl1 channel. Fig. 2c and d show the structures of the perfect crystal and defect models, respectively, viewed along the a-axis. For clarity, only Ca and Cl atoms are drawn in the range of internal coordinates of the a-axis from 0.6 to 1.0. It should be noted that the Cl atom at the original Cl1 site closest to the interstitial Cl atom is significantly displaced along the c-axis. This result is consistent with the single-crystal structure analysis, revealing complex Cl-site disorder in the Cl1 channel.

From the chloride site disorder observed by the single-crystal structure analysis and first-principles calculations, we can expect Cl⁻ ion conduction, especially along the Cl1 channels. To investigate the possible chloride ion conductivity and carrier concentration dependence, polycrystalline samples of (Ca_1−xLa_x)₂B₅O₉Cl_1+2x (x = 0–0.15) solid solutions were synthesized using a conventional solid-state reaction. Fig. 3 shows the room-temperature powder X-ray diffraction (PXRD) patterns of the La-doped Ca₂B₅O₉Cl along with the data for the undoped phase. All solid solutions could be obtained as the main phase, but several tiny peaks assigned to LaBO₃ and/or La(BO₂)₂Cl were detected, suggesting deviation from the nominal chemical compositions. The cell volume increased monotonically with an increase in x ≤ 0.10; however, it remained almost unchanged at 0.125 ≤ x. It is likely that the solid solubility limit was between 0.10 and 0.125.

Fig. 3 (a) Powder X-ray diffraction patterns of (Ca_1−xLa_x)₂B₅O₉Cl_1+2x (x = 0–0.15) solid solutions recorded at room temperature. (b) The cell volume as a function of x in (Ca_1−xLa_x)₂B₅O₉Cl_1+2x. The dashed line is a guide to the eye.

Fig. 4a shows the typical Nyquist plots of (Ca_1−xLa_x)₂B₅O₉Cl_1+2x (x = 0, 0.05, 0.10) at 600 °C, recorded by impedance methods in an Ar gas atmosphere. The data obtained at 400, 500, and 700 °C are presented in Fig. S5.† For each sample, the plot was fitted using a total resistance (R_total), its constant phase element (CPE_total), an electrode–electrolyte interface resistance (R_i), and its constant phase element (CPE_i), where the total components include the contributions from bulk and grain-boundary components. Fig. 4b shows the temperature dependence of the ac conductivity (σ_ac) of (Ca_1−xLa_x)₂B₅O₉Cl_1+2x (x = 0–0.10) solid solutions, estimated from the total resistance. The σ_ac of the undoped phase is comparable to the ionic conductivity of LaOCl at 600–700 °C, and on the order of 10⁻⁵ S cm⁻¹ at 700 °C. Interestingly, 5% La doping into Ca₂B₅O₉Cl improved the conductivity of x = 0 by more than one order of magnitude. x = 0.05 exhibits the highest conductivity of 1.32 × 10⁻⁴ S cm⁻¹ at 700 °C. The conductivities of (Ca_1−xLa_x)₂B₅O₉Cl_1+2x are lower than those of ball-milled La_0.8Ca_0.2OCl_0.8 but become comparable to those of non-ball-milled La_0.8Ca_0.2OCl_0.8 with increasing temperature (Fig. 4b and S6†). The enhanced conductivity of (Ca_1−xLa_x)₂B₅O₉Cl_1+2x can be attributed to the increased carrier (or chloride ion) concentration and the presence of interstitial chloride ions. However, a further increase in the amount of La doping tends to reduce the conductivity, which is probably because of the overdoping effects associated with clustering among the dopants (i.e., La³⁺ ions) and interstitial chloride ions. In fact, no sign of sample degradation for x = 0.10 was observed in the PXRD and thermogravimetric measurements (Fig. S7†). The activation energy (E_a) of (Ca_1−xLa_x)₂B₅O₉Cl_1+2x estimated using the Arrhenius equation is 1.67, 1.1, and 1.21 eV for x = 0, 0.05, and 0.10, respectively. 5–10% La doping into Ca₂B₅O₉Cl reduces the activation energy in comparison with that of the undoped phase. To gain insight into the conducting species in (Ca_1−xLa_x)₂B₅O₉Cl_1+2x, polarization measurements for x = 0.05 were carried out at 700 °C in an Ar gas atmosphere (Fig. S8†). The time evolution of σ_dc/σ_ac, where σ_dc stands for the dc conductivity, exhibited a considerably high degree of polarization (σ_dc/σ_ac < 0.1 over 30 min), suggesting that ions are the dominant migrating species, not electrons, as expected from its high insulating nature.

Fig. 4 (a) Nyquist plots of (Ca_1−xLa_x)₂B₅O₉Cl_1+2x (x = 0–0.10) solid solutions recorded under an Ar gas atmosphere at 600 °C with the inset showing the equivalent circuit model used to fit the data. R_total and CPE_total stand for a total resistance and its constant phase element, respectively, which are attributed to bulk and grain-boundary components, and R_i and CPE_i for an electrode–electrolyte interface resistance and its constant phase element, respectively. The inset plot shows the magnified image of the data for x = 0.05. Fitting curves are indicated by black lines. (b) Arrhenius plots of the ac conductivity (σ_ac) of (Ca_1−xLa_x)₂B₅O₉Cl_1+2x (x = 0–0.10) as a function of temperature. The data for LaOCl (dash line) and non-ball-milled La_0.8Ca_0.2OCl_0.8 (dotted line) are also shown together.⁴⁰

Based on the results of the single-crystal structure analysis and point defect calculations, should be a dominant conduction species. To identify chloride ion diffusion and its conduction mechanism, first-principles molecular dynamics calculations were performed for the association model involving and (Fig. 5a). The O_i model, which has a higher formation energy than , was also used to determine whether oxide ion conduction occurred. The Cl-ion distribution density revealed 1D Cl⁻ ion diffusion along the c-axis in the Cl1 channel during the simulation of the model, while the Cl ions in other channels remained stationary (Fig. 5b). The root mean square displacement (RMSD) result clearly showed only Cl jumps associated with the diffusion of Cl atoms, in contrast to the absence of jumps of the other elements, including O atoms (Fig. 5c). Furthermore, the root square displacement (RSD) calculations for Cl atoms revealed that three Cl atoms in the Cl1 channel cooperatively jump (Fig. S9†), suggesting Cl⁻ ion diffusion via the interstitialcy mechanism. In contrast, no jumps of the O atoms were observed in the O_i model (Fig. S10†). Therefore, we can conclude that is the dominant point defect for ionic conduction.

Fig. 5 (a and b) The most stable structure containing complex defects of and and the corresponding atomic distribution of Cl during MD simulation. (c) Root mean square displacements during MD simulations of the model containing complex defects of and . Migration energy of in models with (d) the point defect and (e) complex defects of and .

Chloride ion migration in Ca₂B₅O₉Cl was also investigated by nudged elastic band (NEB) calculations^41,42 for two models of the point defect and associated defects with and via the interstitialcy mechanism. The estimated energy barrier for the Cl ion migration in the former model was 0.52 eV (Fig. 5d). This value is much lower than the experimental activation energies of (Ca_1−xLa_x)₂B₅O₉Cl_1+2x. The discrepancy between the experimental and calculated values can be attributed to the formation of and the association between and in the undoped and La-doped phases, respectively. Assuming that conducting is perfectly released from the association with La in the dilute state of the La-doped system, the activation energy can be expressed as the sum of the migration energy of and the association energy of . This value was 1.22 eV, which is higher than that of x = 0.05, implying the possibility that conducting Cl is not fully released from the association with the dopant like other ionic conductors.⁴³ Thus, the migration energy of in the association model of a 1 × 1 × 3 supercell with Ca₂₃La₁Cl₁₃B₆₀O₁₀₈ (x = 0.0417) was calculated (Fig. 5e), yielding 1.05 eV. This value agrees well with the experimental E_a value for x = 0.05, which suggests that the association with significantly affects the activation energy of conduction in La-doped Ca₂B₅O₉Cl. Since the migration energy of free from association with La in Ca₂B₅O₉Cl is comparable to those of related metal (oxy)chlorides (e.g. 0.54 eV for La_0.8Ca_0.2OCl_0.8 and 0.15–0.69 eV for CsSnCl₃), optimizing dopant species and compositions could improve the ionic conductivity of Ca₂B₅O₉Cl.

Conclusions

In summary, we demonstrated a unique strategy to design chloride-ion conducting channels using an open borate framework compound, (Ca_1−xLa_x)₂B₅O₉Cl_1+2x, which is the first to exhibit one-dimensional chloride-ion transport defined by the dimensionality of the borate framework. The aliovalent substitution of La³⁺ for Ca²⁺ causes an increase in the concentration of chloride ions in one of the two non-equivalent chloride-ion channels surrounded by covalent (B₅O₉)³⁻ units, involving significant chlorine deficiencies at the original site and site disorder along the channels. No oxygen deficiencies or disorder occurred, in contrast to oxide-ion conducting oxyhalides. First-principles molecular dynamics simulations show one-dimensional cooperative diffusion of interstitial chloride ions along the chloride-ion channels. The calculations of the formation energies of point defects suggest the presence of non-negligible association between and , which enhances the energy barrier for chloride ion migration, compared with that without defect association. As presented in this study, an open borate framework built with rigid covalent B–O bonds can be a useful tool for rationally designing 0D–3D chloride-ion conduction paths based on its structural dimensionality, which will contribute not only to understanding of the chloride-ion transport phenomena but also to improving the poor variety of chloride-ion conduction. In addition, it has significant advantages in terms of low environmental impact and abundant resources. The chloride-ion conductivity of the present compound is lower than those of previously reported perovskite chlorides and doped PbCl₂; however, further exploration using a rich variety of open borate frameworks^26,44 could contribute to the realization of safe, low-temperature-driven fast chloride-ion conducting solid electrolytes with thermal and chemical stability for all-solid-state chloride ion batteries.

Data availability

Data supporting this article have been included as part of the ESI.† Crystallographic data for Ca₂B₅O₉Cl and (Ca_0.92La_0.08)₂B₅O₉Cl_1.16 have been deposited at the CCDC with the deposition numbers 2353931 and 2353932, respectively.

Author contributions

Y. T. conceived and designed the experiments. Y. T., Y. M., N. N., Y. M., N. I., and K. Y. performed all experimental work. K. S. performed all first-principles calculations. Y. T., Y. M., N. N., and K. S. wrote the draft paper. All authors discussed the results and commented on the manuscript.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was partly supported by the JSPS KAKENHI (22K14465, 22H05146, and 23K23046), Core-to-Core Program (JPJSCCA20200004), JSPS Bilateral Program (No. JPJSBP120237714). MANA is supported by the World Premier International Research Center Initiative (WPI), MEXT, Japan. Y.T. acknowledges the grant from the Sumitomo Foundation.

Notes and references

1. B. Lei, S. Bai, S. Ju, C. Yin, C. Chen and J. Zhang, Mater. Res. Express, 2021, 8, 062001.
2. N. Yamazoe and N. Miura, MRS Bull., 1999, 24, 37–43.
3. T. Famprikis, P. Canepa, J. A. Dawson, M. S. Islam and C. Masquelier, Nat. Mater., 2019, 18, 1278–1291.
4. T. Nestler, E. Roedern, N. F. Uvarov, J. Hanzig, G. Antonio Elia and M. de Vivanco, Phys. Sci. Rev., 2019, 4, 20170115.
5. S. Singhal, Solid State Ionics, 2000, 135, 305–313.
6. M. P. Fertig, K. Skadell, M. Schulz, C. Dirksen, P. Adelhelm and M. Stelter, Batteries Supercaps, 2022, 5, e20210013.
7. H. J. Deiseroth, S. T. Kong, H. Eckert, J. Vannahme, C. Reiner, T. Zaiß and M. Schlosser, Angew. Chem., Int. Ed., 2008, 47, 755–758.
8. Q. Liu, Y. Wang, X. Yang, D. Zhou, X. Wang, P. Jaumaux, F. Kang, B. Li, X. Ji and G. Wang, Chem, 2021, 7, 1993–2021.
9. I. V. Murin, O. V. Glumov and N. A. Mel'Nikova, Russ. J. Electrochem., 2009, 45, 411–416.
10. B. Wu, J. Luxa, J. Šturala, S. Wei, L. Děkanovský, A. K. Parameswaran, M. Li and Z. Sofer, Energy Environ. Mater., 2024, 7, e12530.
11. D. Newman, D. Hazlett and K. Mucker, Solid State Ionics, 1981, 3–4, 389–392.
12. R. Sakamoto, N. Shirai, A. Inoishi and S. Okada, ChemElectroChem, 2021, 8, 4441–4444.
13. T. Xia, Q. Li, X. Zhao and X. Shen, Adv. Mater., 2024, 36, 2310565.
14. G. Karkera, M. Soans, A. Akbaş, R. Witter, H. Euchner, T. Diemant, M. A. Cambaz, Z. Meng, B. Dasari, S. G. Chandrappa, P. W. Menezes and M. Fichtner, Adv. Energy Mater., 2023, 13, 2300982.
15. R. Sakamoto, N. Shirai, L. Zhao, A. Inoishi, H. Sakaebe and S. Okada, Electrochemistry, 2023, 91, 23–00041.
16. J. G. Speight, Lange’s Handbook of Chemistry, McGraw-Hill, New York, 16th edn, 2005.
17. L. Peedikakkandy and P. Bhargava, RSC Adv., 2016, 6, 19857–19860.
18. IUPAC-NIST Solubility Database, Version 1.1 NIST Standard Reference Database 106.
19. N. Imanaka, K. Okamoto and G. Y. Adachi, Angew. Chem., Int. Ed., 2002, 41, 3890–3892.
20. H. Matsumoto, T. Miyake and H. Iwahara, Mater. Res. Bull., 2001, 36, 1177–1184.
21. K. Shitara, A. Kuwabara, K. Hibino, K. Fujii, M. Yashima, J. R. Hester, M. Umeda, N. Nunotani and N. Imanaka, Dalton Trans., 2021, 50, 151–156.
22. H. Yaguchi, K. Fujii, Y. Tsuchiya, H. Ogino, Y. Tsujimoto and M. Yashima, ACS Appl. Energy Mater., 2022, 5, 295–304.
23. M. Kluczny, J. T. Song, T. Akbay, E. Niwa, A. Takagaki and T. Ishihara, J. Mater. Chem. A, 2022, 10, 2550–2558.
24. N. Ueno, H. Yaguchi, K. Fujii and M. Yashima, J. Am. Chem. Soc., 2024, 11235–11244.
25. H. Yaguchi, D. Morikawa, T. Saito, K. Tsuda and M. Yashima, Adv. Funct. Mater., 2023, 33, 2214082.
26. F. C. Hawthorne, Mineral. Mag., 2014, 78, 957–1027.
27. Z. Zhang, Y. Wang, B. Zhang, Z. Yang and S. Pan, Angew. Chem., Int. Ed., 2018, 57, 6577–6581.
28. F. Strauss, G. Rousse, D. Alves Dalla Corte, C. Giacobbe, R. Dominko and J. M. Tarascon, Inorg. Chem., 2018, 57, 11646–11654.
29. B. Calès, A. Levasseur, C. Fouassier, J. M. Réau and P. Hagenmuller, Solid State Commun., 1977, 24, 323–325.
30. L. Duchêne, S. Lunghammer, T. Burankova, W. C. Liao, J. P. Embs, C. Copéret, H. M. R. Wilkening, A. Remhof, H. Hagemann and C. Battaglia, Chem. Mater., 2019, 31, 3449–3460.
31. X. Li, L. Yang, Z. Zhu, X. Wang, P. Chen, S. Huang, X. Wei, G. Cai, P. Manuel, S. Yang, J. Lin, X. Kuang and J. Sun, Sci. China Mater., 2022, 65, 2737–2745.
32. Z. Žak and F. Hanic, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem., 1976, 32, 1784–1787.
33. Z. Wang, J. He, B. Hu, P. Shan, Z. Xiong, R. Su, C. He, X. Yang and X. Long, ACS Appl. Mater. Interfaces, 2020, 12, 4632–4637.
34. V. P. Dotsenko, I. V. Berezovskaya, N. P. Efryushina, E. V. Shabanov and A. S. Voloshinovskii, Phys. Status Solidi A, 2006, 203, 892–897.
35. J. Liang, X. Yang and S. Xiao, Opt. Mater., 2023, 135, 113269.
36. M. Lumbreras, J. Protas, S. Jebbari, G. J. Dirksen and J. Schoonman, Solid State Ionics, 1986, 20, 295–304.
37. L. H. Brixner and E. P. Moore, Acta Crystallogr., Sect. C: Cryst. Struct. Commun., 1983, 39, 1316.
38. R. D. Shannon, Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr., 1976, 32, 751–767.
39. K. Momma and F. Izumi, J. Appl. Crystallogr., 2011, 44, 1272–1276.
40. N. Nunotani, M. R. I. Bin Misran, M. Inada, T. Uchiyama, Y. Uchimoto and N. Imanaka, J. Am. Ceram. Soc., 2020, 103, 297–303.
41. G. Henkelman, B. P. Uberuaga and H. Jónsson, J. Chem. Phys., 2000, 113, 9901–9904.
42. G. Henkelman and H. Jónsson, J. Chem. Phys., 2000, 113, 9978–9985.
43. J. Koettgen, S. Grieshammer, P. Hein, B. O. H. Grope, M. Nakayama and M. Martin, Phys. Chem. Chem. Phys., 2018, 20, 14291–14321.
44. M. Mutailipu, K. R. Poeppelmeier and S. Pan, Chem. Rev., 2021, 121, 1130–1202.

---

Footnote

† Electronic supplementary information (ESI) available. CCDC 2353931 and 2353932. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4ta04624b

---