Li₄B₁₀H₁₀B₁₂H₁₂ as solid electrolyte for solid-state lithium batteries

Abstract

Hydridoborates are a promising class of solid electrolytes for solid-state batteries, combining liquid-like room temperature ionic conductivity, high (electro-)chemical stability, low gravimetric density, easy processability, and low toxicity. We show that cation exchange is a feasible method to prepare Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ from Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂, respectively, with high yields. Ball milling of an equimolar mixture of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ yields a single phase, isomorphic to the low temperature Li₂B₁₂H₁₂ phase (space group Pa3̄) with ordered [B₁₂H₁₂]²⁻ and disordered [B₁₀H₁₀]²⁻ anions sharing the same position in the structure. The ionic conductivity of the equimolar mixture Li₄B₁₀H₁₀B₁₂H₁₂ exhibits 4 × 10⁻⁴ S cm⁻¹ at 25 °C and 4 × 10⁻³ S cm⁻¹ at 60 °C, respectively, exceeding those of the unmixed phases by several orders of magnitude. The electrolyte possesses an oxidative stability >3 V vs. Li⁺/Li and thermal stability beyond 300 °C, and is evaluated in proof-of-concept solid-state batteries with a lithium metal anode and with titanium disulfide (TiS₂) or lithium iron phosphate (LiFePO₄) as a cathode active material. Discharge capacities of 83% and 73% of the theoretical capacity were achieved for TiS₂ and LiFePO₄, respectively, at the end of the first dis-/charge cycle. For LiFePO₄, the de-/lithiation potential lies outside the electrochemical stability window of the electrolyte, requiring additional measures to protect the electrolyte from decomposition. Our study demonstrates the feasibility of using closo-hydridoborates as ionic conductors in solid-state lithium batteries.

---

Introduction

Solid-state batteries (SSBs) promise to overcome the limitations and challenges of today's state-of-the-art lithium-ion batteries with higher power and energy density while improving operational safety.^1,2 Therefore, SSBs are considered a viable future alternative to lithium-ion batteries and they are expected to reach the market in larger volumes within the next years.³ To fulfill this promise, the solid electrolyte, at the heart of the SSB, must combine high (electro-)chemical stability to enable the use of metal anodes and high voltage cathodes, high ionic conductivity to enable fast charging, and low gravimetric density to maximize the energy density. Other important requirements are ease of processing, low toxicity, and low cost. To date, none of the solid-state conductors investigated meet all these requirements simultaneously.

Alkali metal hydridoborates, formerly known as hydroborates,⁴ and their chemical relatives are an emerging class of solid electrolytes.⁵ In particular, the lithium and sodium hydridoborate and monocarba-hydridoborate salts with closo-caged [B_nH_n]²⁻ or [CB_n−1H_n]⁻ (n = 10 and 12) anions combine oxidative stability beyond 3 V vs. Li⁺/Li and Na⁺/Na, compatibility with lithium and sodium metal anodes and low gravimetric densities (∼1.2 g cm⁻³). They generally exhibit high thermal and chemical stability, soft mechanical properties enabling cold pressing, solution processability, and low toxicity.^6,7 As solid electrolytes, they exhibit a cation transference number of almost unity due to an immobile anion framework.^8,9

Hydridoborates typically reach ionic conductivities of the order of 0.01–0.1 S cm⁻¹ in their high-temperature disordered phases, while exhibiting much lower conductivities (below 10⁻⁷ S cm⁻¹) at room temperature.¹⁰ Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂ undergo phase transitions at 100 °C (ref. 11) and 260 °C,¹² respectively, while their lithium analogues Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ transform to their respective high-temperature phases at 370 °C (ref. 13) and 355 °C.¹⁴ The structural properties, the crystal chemistry, and the relation to the ionic conductivity of the hydridoborates have been summarized in a recent review.¹⁵

Ion conduction in hydridoborates occurs via an anion-assisted hopping mechanism, supported by disorder. As shown by DFT and ab initio molecular dynamics calculations, disorder arises from a mismatch between anion and crystal symmetry and a changing energy landscape due to dynamic reorientations. It is this geometric and dynamic disorder that leads to frustration and to enhanced ionic conductivity.¹⁶ The phase transitions can be shifted to lower temperatures by anionic substitution, e.g., by mechano-chemical mixing of two different hydridoborates, as a means of introducing disorder. The resulting phases typically adopt the crystal structure of a disordered high temperature phase, and increase the room temperature ionic conductivity by several orders of magnitude as compared to their individual compounds. The ionic conductivity of the sodium compounds typically exceeds that of their lithium equivalents due to the more favourable ion size ratio and the lower charge density of the sodium ion compared to the lithium ion.^5,10,16–18

To date, the highest ionic conductivity has been observed in the mixed-anion sodium monocarba-hydridoborate system Na₂(CB₉H₁₀)(CB₁₁H₁₂) (0.07 S cm⁻¹ at 27 °C),¹⁹ while for the analogous lithium system (1 − x)LiCB₉H₁₀–xLiCB₁₁H₁₂ (x = 0.1–0.9), ionic conductivities between 6.9 × 10⁻⁵ and 6.7 × 10⁻³ S cm⁻¹ have been reported at room temperature.^19–21 Monocarba-hydridoborate-based solid electrolytes have been successfully integrated into solid-state battery cells, and stable cycling against alkali metal anodes has been achieved for S |Li(CB₁₁H₁₂)_0.3(CB₉H₁₀)_0.7| Li,Na₃(VOPO₄)₂F|Na₄(CB₁₁H₁₂)₂(B₁₂H₁₂)|Na, and NaCrO₂ |Na₄(CB₁₁H₁₂)₂(B₁₂H₁₂)|Na.^20,22,23 However, the high market price of monocarba-hydridoborates currently still hinders a wider adoption of this material in the battery industry.

In the case of sodium, a mixed-anion hydridoborate Na₄B₁₀H₁₀B₁₂H₁₂ is a carbon-free solid electrolyte that combines low density (1.1 g cm⁻³), high ionic conductivity (0.9 × 10⁻³ S cm⁻¹), stability against sodium metal, and an oxidative stability of ∼3 V vs. Na⁺/Na.^24,25 The electrolyte can be prepared from its precursors Na₂B₁₂H₁₂ and Na₂B₁₀H₁₀ in a 1 : 1 molar ratio either by ball milling or by crystallization from an equimolar solution, enabling solution-based processes. Na₄B₁₀H₁₀B₁₂H₁₂ has been successfully integrated into NaCrO₂|Na₄B₁₀H₁₀B₁₂H₁₂|Na and into NaCrO₂|Na₄B₁₀H₁₀B₁₂H₁₂|Na–Sn cells.^26,27

In the case of lithium, no enhanced ionic conductivity beyond that of the constituent hydridoborates has been reported for the Li₂B₁₀H₁₀–Li₂B₁₂H₁₂ system. Recently, Zhou et al.²⁸ investigated the ionic conductivity and electrochemical properties of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ and their mixtures prepared by ball milling of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ in different stoichiometric ratios, followed by a heat treatment at 380 °C under H₂ atmosphere (450 bar) to prevent sample decomposition by hydrogen loss. Unlike other mixed-anion hydridoborate systems, they found no evidence of solid solution formation within the Li₂B₁₀H₁₀–Li₂B₁₂H₁₂ system. Instead, Zhou et al. described their compounds as physical mixtures of the constituent hydridoborates.²⁸

Here we show that Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ can be prepared by cation exchange from Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂ with high yields. In contrast to the findings of Zhou et al., we show that single-phase equimolar mixtures of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ can be prepared by ball milling. The mixture designated Li₄B₁₀H₁₀B₁₂H₁₂ exhibits an ionic conductivity of 4 × 10⁻⁴ S cm⁻¹ at 25 °C and 4 × 10⁻³ S cm⁻¹ at 60 °C, respectively, and is evaluated in proof-of-concept solid-state battery cells with a lithium metal anode and with titanium disulfide (TiS₂) or lithium iron phosphate (LiFePO₄) as the cathode active material. Our results show that carbon-free hydridoborates are also viable candidates as solid electrolytes for lithium-based solid-state batteries.

Materials and methods

Materials preparation

Hydridoborates were stored and handled in a glovebox under argon (MBraun, H₂O and O₂ content <0.1 ppm) or under vacuum Na₂B₁₂H₁₂, Na₂B₁₀H₁₀, Li₂B₁₀H₁₀, and Li₂B₁₂H₁₂·4H₂O were purchased from Katchem Na₂B₁₂H₁₂ was used and characterized as received and Na₂B₁₀H₁₀ was first dried under vacuum for 6 h at 160 °C. Although no water content is specified for the Li₂B₁₀H₁₀, a residual water content of ∼0.2 weight % was determined by thermogravimetry (TG) (Fig. S1†). After the determination of suitable drying conditions by combined differential scanning calorimetry (DSC) and TG, as-received Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂·4H₂O were vacuum dried (p < 10⁻³ mbar) at 160 °C for 12 h and at 210 °C for 14 h using Schlenk technique. The successful removal of water was confirmed by X-ray diffraction (XRD) as shown in Fig. S2.†

Cation exchange from Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂ to Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ was carried out using AmberLite™ IRC120H (Sigma-Aldrich) as exchange resin. Protonation of the resin was achieved by hydrochloric acid (HCl, Sigma-Aldrich), neutralization by lithium hydroxide (LiOH, Sigma-Aldrich).

Li₂B₁₀H₁₀ : Li₂B₁₂H₁₂ samples with stoichiometric ratios of 1 : 2, 1 : 1, and 2 : 1 were prepared by mixing the appropriate amounts of dried precursors using mechanical ball milling using a shaker mill (Spex 8000M) with 5 mm balls and with a ball-to-sample mass ratio of 10 to 1 for one hour in four 15 minutes intervals with 5 min breaks to avoid overheating. To prevent exposure to air, the milling vial was sealed in a gas-tight bag under argon atmosphere in the glovebox. After completion of the milling process, the milling vial was transferred back to the glovebox.

For the electrochemical impedance spectroscopy measurements, pellets of 30 mg each were prepared by uniaxial cold pressing using a hardened steel die with a diameter of 6.35 mm. To avoid contact with air, the die was sealed in a gas-tight bag in the glovebox and transferred to a hydraulic press (Specac), where a pressure of 3.2 t (990 MPa) was applied for 3 min. The die was then returned to the glovebox.

For the proof-of-concept batteries, cathode composites were hand-mixed from the solid electrolyte (Li₄B₁₂H₁₂B₁₀H₁₀), the cathode active material TiS₂ (Alfa Aeser) or LiFePO₄ (MTI), carbon black (Super C65, Imerys) as a conductive carbon additive, and polyvinylidene fluoride (PVDF, Kynar HSV 900, Arkema) as a binder (in the case of LiFePO₄ cathodes). 100 mg of the solid electrolyte Li₄B₁₂H₁₂B₁₀H₁₀ and typically 0.5 mg or 1 mg of the cathode composite were stacked and pressed into 12 mm diameter pellets under a pressure of 5 t (434 MPa) for 3 min.

Materials characterization

For the XRD measurements, samples of the as-prepared powders were ground and sealed in 0.7 mm diameter borosilicate capillaries under inert conditions in the glovebox. An X-ray diffractometer (MalvernPanalytical Empyrean) was used at an accelerating voltage of 45 kV and a tube current of 40 mA. A Cu K_α radiation selecting focusing mirror was used to focus the incident radiation onto the sample. The diffraction patterns were recorded using an X'Celerator detector in the angular range between 10° and 50° with a step size of 0.004°. Rietveld refinements of the patterns were performed using TOPAS.²⁹ The atomic coordinates of Li⁺ cations and [B₁₂H₁₂]²⁻ and [B₁₀H₁₀]²⁻ anions were fixed to published values,³⁰ the latter modelled as rigid bodies and only cell parameters, peak shapes, scale factors, atomic displacement factors, and anion orientation were refined.

Combined DSC and TG were performed under helium flow at a heating/cooling rate of 5 K min⁻¹ (Netzsch, STA 449 F3 Jupiter). The sample (typically 5–10 mg) was sealed in Al pans in the glovebox under inert atmosphere. The pans were punctured immediately prior to the combined DCS/TG measurement to minimize air contact.

Magic-angle spinning (MAS) solid-state nuclear magnetic resonance (NMR) measurements were performed on a 4 mm cross-polarization MAS probe (Bruker Avance 400). ¹¹B and ²³Na NMR spectra were recorded as single-pulse experiments at 128.4 MHz (typically 512 scans) using MAS rates of 13 kHz and π/12 pulse lengths of 1.5 μs with 50 kHz SPINAL-64 proton decoupling during acquisition. The data were processed with a line broadening of 10 Hz, and the boron background resonance of the probe was subtracted from the spectra with data recorded under the same measurement conditions from a boron-free sample.

Electrochemical characterisation

Electrochemical impedance spectroscopy (EIS) measurements were performed using an Alpha-AT impedance analyzer (Novocontrol). To enhance the electronic contact, the pellets were sandwiched between two 6 mm diameter indium foils (purity 99.995%, 0.1 mm in thickness, Sigma-Aldrich) in an airtight sample holder. The diameter of the indium foil was smaller than the diameter of the pellet to avoid short-circuiting. Temperature control was achieved using a nitrogen stream. Each sample was subjected to two heating/cooling cycles. The first cycle was from −20 °C to 100 °C, and the second was from −25 °C to 120 °C. For each temperature, the EIS measurement was started after the temperature equilibrated at the set temperature ±0.5 °C for at least 1 min. Impedance was measured in the range from 2 MHz to 5 Hz (only data below 1 MHz were used for fitting) with a voltage amplitude of 10 mV. The pellet resistance R was extracted from the last measurement at each temperature and taken from the intercept of the fitted semi-circles and/or linear spikes with the x-axis in the Nyquist plots. Li⁺ conductivity is then calculated from σ = (1/R) × (d/A), where d and A are the thickness and the area of the pellet, respectively. Oxidative stability was determined according to the method described by Asakura et al.³¹

For galvanostatic charge–discharge measurements, an aluminum foil (purity >99.3%, thickness 15 μm, MTI) was attached as a current collector to the cathode side of the pellets, consists of the solid electrolyte covered on one side with the cathode composite. On the anode side, the pellets were covered with lithium metal (purity 99.9%, 0.75 mm thickness, Alfa Aesar) and a copper foil (purity 99.98%, 0.025 mm thickness, Sigma-Aldrich) as a current collector with diameters of 10 mm and 12 mm, respectively. The whole stack was then transferred to a home-made pressure cell and a pressure of 1.71 MPa was applied. Charge–discharge measurements were performed at different C-rates and in a potential range from 1.6 V to 2.5 V vs. Li⁺/Li for cells with TiS₂ and from 2.0 V to 3.6 V vs. Li⁺/Li for cells with LiFePO₄, using a multi-channel potentiostat (BioLogic VMP3). Cell assembly and measurements were performed in the glovebox.

Results and discussion

Preparation of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂via cation exchange from Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂

We have successfully prepared Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ by cation exchange from Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂ by modifying the procedure described by Blake³² for the synthesis of M₂(B_xH_x) (x = 10 and 12) as follows. We use an ion exchange column with the AmberLite™ IRC120H ion exchange resin. The process is shown schematically in Fig. 1. In the first step, the resin is loaded into a vertically mounted column, rinsed with deionized water and protonated with 0.75 M HCl (Fig. 1A). The column is then rinsed with deionized water to remove excess HCl and an aqueous solution of Na₂B₁₀H₁₀ (or Na₂B₁₂H₁₂) is introduced (Fig. 1B). At this stage, sodium ions exchange with protons and bind to the resin, while the anions remain in solution. Subsequent rinsing with deionized water results in a solution of H₂B₁₀H₁₀ (or H₂B₁₂H₁₂), which is neutralized with LiOH (Fig. 1C) to form Li₂B₁₀H₁₀ (or Li₂B₁₂H₁₂).

Fig. 1 Schematic representation of the preparation of Li₂B₁₂H₁₂ from Na₂B₁₂H₁₂ using an ion exchange column. The reaction proceeds in three steps: (A) protonation of the resin, (B) introduction of an aqueous solution of Na₂B₁₀H₁₀ or Na₂B₁₂H₁₂, and (C) rinsing with deionized water and subsequent neutralization with LiOH.

The yield of the process was 96% for Li₂B₁₀H₁₀ and 87% for Li₂B₁₂H₁₂, which was determined from (i) the weights of the isolated products relative to the starting materials (Na₂B₁₀H₁₀ or Na₂B₁₂H₁₂, respectively) and (ii) the amount of LiOH required to neutralize the H₂B₁₀H₁₀ (H₂B₁₂H₁₂) solution. The ion exchange column can be used to prepare Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ individually in separate runs or to prepare mixtures of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ starting from the respective aqueous solution of Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂ in a predefined ratio. The preparation of individual Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ or the direct synthesis of the stoichiometric mixture results in a product of high chemical purity. XRD patterns show almost phase-pure products (see Fig. 2). ²³Na NMR further confirms the successful ion exchange. Only traces of sodium (see Fig. S3†) and no unwanted B–H compounds were detected by ¹¹B NMR. Most importantly, the Li salts prepared by ion exchange show similar ionic conductivity to that of the commercial Li salts, as we will discuss in the next section.

Fig. 2 XRD comparison of commercial and ion-exchanged (a) Li₂B₁₀H₁₀ and (b) Li₂B₁₂H₁₂.

Ball-milled mixtures of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂

Fig. 3a shows the ionic conductivity of dried and ball-milled Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ as well as the ones of ball-milled Li₂B₁₀H₁₀ : Li₂B₁₂H₁₂ in a 2 : 1, 1 : 1, and 1 : 2 molar ratio. The 1 : 1 mixture exhibits the highest conductivity of 4 × 10⁻⁴ S cm⁻¹ at room temperature, exceeding that of the dried, non-milled and non-mixed precursors by several orders of magnitude. In addition, the conductivity of this sample is also higher than that of the other two mixtures (Li₂B₁₀H₁₀ : Li₂B₁₂H₁₂ ratios of 1 : 2 and 2 : 1, respectively). We will restrict our discussion to the 1 : 1 mixture and refer to it as Li₄B₁₀H₁₀B₁₂H₁₂. Note that ball milling increases the conductivity of the pure phases by at least two orders of magnitude (data labelled “bm” in Fig. 3a).

Fig. 3 (a) Temperature-dependent lithium-ion conductivity of the dried, dried and ball-milled precursors Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂, and ball-milled mixtures of the dried precursors in stoichiometric ratios as indicated. (b) XRD pattern of the ball-milled 1 : 1 stoichiometric mixture of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ (blue), the result of the Rietveld refinement (red), and the difference curve (grey). The markers indicate the position of the Bragg reflections of the indicated phases. (c) ¹¹B NMR spectra of Li₂B₁₂H₁₂, Li₂B₁₀H₁₀ and Li₄B₁₀H₁₀B₁₂H₁₂. (d) DSC signals of the ball-milled 1 : 1 stoichiometric mixture of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ over six consecutive cycles.

Nanosizing by mechano-chemical milling of hydridoborides is a well-established strategy to enhance the ionic conductivity of hydrido-borate based solid electrolytes by several orders of magnitude. The effect the milling process on the ionic conductivity is subject of current research. There is common agreement that ball milling results in the increase in structural disorder, stabilizing the high temperature phase.^33–35

In the case of Li₂B₁₂H₁₂, the conductivity at 30 °C increases from 5 × 10⁻⁸ S cm⁻¹ to 6 × 10⁻⁶ S cm⁻¹ by 4 × 15 min of ball milling. Our result for the dried Li₂B₁₂H₁₂ is in agreement with the values measured by Kim et al.,³⁵ Tang et al.,³⁶ and Zhou et al.,²⁸ who recorded ionic conductivities of 2 × 10⁻⁸ S cm⁻¹, 7 × 10⁻⁸ S cm⁻¹ (extrapolated), and 5.4 × 10⁻⁸ S cm⁻¹, respectively. The observed increase in ionic conductivity by several orders of magnitude with ball milling is in agreement with previous work by Kim et al.,³⁵ and Tang et al.,³⁶ which reported conductivity of 2 × 10⁻⁵ S cm⁻¹ (after 20 h of milling) and 1 × 10⁻⁵ S cm⁻¹ (after 5 h of milling) for Li₂B₁₂H₁₂ at 30 °C, respectively. Prolonged ball milling leads to higher conductivity but is also associated with partial decomposition.³⁵

However, the current literature is not consistent on this point. Zhou et al.²⁸ found “only a minor effect from ball milling” and reported a conductivity of 5.4 × 10⁻⁸ and 7.0 × 10⁻⁸ S cm⁻¹ at 30 °C for pristine and ball-milled Li₂B₁₂H₁₂, respectively, while Teprovich et al. reported a room-temperature conductivity of 3.1 × 10⁻⁴ S cm⁻¹ after 5 min of milling.³⁷ Zhou et al. also did not observe an increase in ionic conductivity upon anion mixing nor the formation of a new phase upon ball milling. In contrast to our results, they observed “no indication of any interaction, reaction or formation of a solid solution in the composite samples (1 − x)Li₂B₁₂H₁₂–xLi₂B₁₀H₁₀”.²⁸ We suggest that different sample preparation and treatment, in particular different ball-milling conditions (e.g., planetary ball mill used by Zhou et al. versus high-energy ball mill in the present study) and the heat treatments used by Zhou et al. may lead to chemical modifications. In addition, different levels of impurities, e.g., due to residual solvents and defects in the crystal structure, may be responsible for the conflicting reports.

The crystal structure of the equimolar mixture was investigated by powder XRD. The reflections observed in the XRD pattern are consistent with a single-phase mixture that matches the one of Li₂B₁₂H₁₂ in its room temperature polymorph. Fig. 3b shows the measured powder XRD pattern (blue) together with the result of the Rietveld refinement using published crystallographic data of the room-temperature phase of Li₂B₁₂H₁₂ as a starting model (red),³⁰ and the grey line represents the difference curve. LiCl was identified as a minor impurity (<1 weight %). Thereby the anions remain intact. The ¹¹B MAS NMR spectrum of the equimolar mixture comprises the resonances of the constituents in the precursors (see Fig. 3c). Furthermore, the NMR spectra depicted in Fig. 3c indicate an increase in rotational motion of the ions in the 1 : 1 mixture as compared to the starting materials. This increase in rotational motion leads to a significant narrowing of the resonances of the equimolar mixture as compared to the resonances of the pure precursors. This narrowing clearly shows that the two phases in the 1 : 1 mixture do not simply coexist as a physical mixture, but actually form a new phase with a higher degree of rotational motion. This behavior is typical for highly conductive mixed-anion hydridoborates and has also been observed for Na₄B₁₀H₁₀B₁₂H₁₂.²⁴

There is a peculiarity within the Li₂B₁₀H₁₀–Li₂B₁₂H₁₂ system; the ordered low-temperature phase of Li₂B₁₂H₁₂ (Pa3̄)³⁰ and the disordered high-temperature phase of Li₂B₁₀H₁₀ (Pa3̄ or Fm3̄m)^28,38 crystallize with the same cubic closed packed (CCP) arrangement of the respective anions and display nearly the same lattice parameters of ∼9.58 Å. This facilitates the incorporation of the smaller [B₁₀H₁₀]²⁻ anion into the structure of Li₂B₁₂H₁₂, randomly replacing [B₁₂H₁₂]²⁻ anions. For other cations, the ordered A₂B₁₂H₁₂ (A = Na, K) phase is not cubic and/or the disordered A₂B₁₀H₁₀ does not crystallize in ccp symmetry.¹⁵

We attribute the increase in the conductivity of the mixed phase to the enhanced rotational motion of the anions in the mixed phase. Within the icosahedral [B₁₂H₁₂]²⁻ anion, all boron atoms are equivalent, resulting in a single resonance in the ¹¹B NMR spectrum (Fig. 3c). Rotational jumps, therefore, lead to symmetry equivalent positions. The [B₁₀H₁₀]²⁻ anion has an elongated shape with two inequivalent boron positions, giving rise to two resonances in the ¹¹B NMR spectrum (Fig. 3c). The elongated [B₁₀H₁₀]²⁻ anion has a distinct symmetry axis and therefore a lower symmetry than the cube. Within the cubic symmetry of the crystal, [B₁₀H₁₀]²⁻ anions can align along different principal lattice directions and thus adopt different orientations. Consequently, the [B₁₀H₁₀]²⁻ anions are unlike the [B₁₂H₁₂]²⁻ ones orientationally disordered. X-ray diffraction does not distinguish between different types of rotation and does not yield the axis of rotation. However, similar behavior can be observed in the chemically and structurally related carborane C₂B₁₀H₁₂. In C₂B₁₀H₁₂, the transition between the ordered and disordered phases involves an intermediate, partially disordered phase (Pa3̄), which shows uniaxial rotation of the anion.³⁹ A similar transition with two high-temperature phases has been proposed for Li₂B₁₀H₁₀ in ref. 33,³⁷ and two phases have indeed been observed.²⁷

In addition to the ionic conductivity, a wide electrochemical stability window is a crucial property to enable high-voltage batteries. The oxidative stability of hydridoborates measured by linear sweep voltammetry has previously been overestimated, and an oxidative stability limit above 6 V vs. Li⁺/Li has been reported for Li₂B₁₂H₁₂.³⁷ This overestimation is due to the low electronic conductivity of the hydridoborates, which results in slow decomposition kinetics, making it difficult to observe an anodic current, especially at higher scan rates.³¹ We have determined the oxidative stability of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ by voltammetric methods following the method of Asakura et al.,³¹ which uses the addition of conductive carbon to accelerate electrolyte decomposition kinetics and a low scan rate of 10 μV s⁻¹. The oxidative stability limit at 120 °C was determined to be 3.1 V and 3.6 V vs. Li⁺/Li for Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂, respectively (see Fig. S4†).

Li₄B₁₀H₁₀B₁₂H₁₂ also has a high thermal stability. Fig. 3d shows the DSC signals in six consecutive cycles. The sample can be reversibly heated and cooled between room temperature and 300 °C without any sign of a phase transition or decomposition.

Li₄B₁₀H₁₀B₁₂H₁₂ obtained by the simultaneous ion exchange of Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂ shows a slightly lower ionic conductivity (see Fig. S5†). The following electrochemical cells were assembled using the commercial precursors.

Solution-based synthesis of mixtures of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂

Attempts to prepare the equimolar mixture by crystallization from solution, analogous to the 1 : 1 equimolar mixture of Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂, were not successful. In addition to isopropanol, the solvent of choice for the sodium case, we investigated more than 10 different solvents with different polarity. The results are summarized in Table 1.

Table 1 Solvents used for the attempted synthesis of mixtures of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂

Solvent (decreasing polarity)	Structure	Ultrasonic bath time	Solution formed
a No salt could be recovered from the solution; the solvent could not be completely removed.
Water		—	Yes
Methanol		—	Yes
Ethanol		1 h	Yes
2-Propanol		1 h	Yes
2-Butanol		1 h	Yes
Acetonitrile		1 h	Yes
Dimethyl sulfoxide (DMSO)^a		1 h	Yes
Dimethylformamide (DMF)^a		1 h	Yes
Acetone		1 h	Yes
Dichloromethane (DCM)		1 h	No
2-Butanone		1 h	Yes (reacts)
Tetrahydrofuran (THF)		2 h	No
Toluene		2 h	No
Cyclohexane		2 h	No

---

For each solvent, 100 mg of the equimolar mixture of Li₂B₁₂H₁₂ and Li₂B₁₀H₁₀ was added to 20 mL of the solvent. The solubility of the salts is strongly dependent on the polarity of the solvent. While polar solvents such as water or methanol readily dissolve both salts, ultrasonication is required to dissolve them in less polar solvents such as ethanol, 2-propanol or acetone. Non-polar solvents such as cyclohexane do not dissolve either Li₂B₁₂H₁₂ or Li₂B₁₀H₁₀. 2-Butanone was the only solvent among the ones investigated to react observably with the closo-hydridoborates, forming a yellowish solution. Subsequently, the anion mixture was recovered from the solution and examined by XRD (shown in Fig. S6†). Except for 2-butanone, they all show the coexistence of the initial phases. In the case of 2-butanone, the XRD pattern and the NMR spectrum of the recovered salt resembled the ones of Li₂B₁₂H₁₂ indicating a chemical reaction that consumed Li₂B₁₀H₁₀. We suspect the formation of X-ray amorphous phases; however, we did not further investigate the reaction between Li₂B₁₀H₁₀ and 2-butanone.

Proof-of-concept battery cells

Proof-of-concept batteries using lithium metal anodes and TiS₂ or LiFePO₄ as the cathode active material (CAM) were assembled in home-made pressure cells with a 12 mm diameter active area. Different mass ratios of CAM : Li₄B₁₀H₁₀B₁₂H₁₂ were investigated. In the case of TiS₂ the ratio of Super C65 as a conductive carbon additive was kept constant at 10 weight %, maintaining a mass loading of 0.9 mg_CAM cm⁻². TiS₂ was chosen as CAM because it has been successfully used in combination with other hydridoborate-based solid electrolytes.¹⁸ Its lithium de-/intercalation potential of ≤2.5 V vs. Li⁺/Li lies well below the oxidative stability limit of the Li₄B₁₀H₁₀B₁₂H₁₂ electrolyte.

Fig. 4a shows the charge/discharge voltage profiles of a TiS₂|Li₄B₁₀H₁₀B₁₂H₁₂|Li cell cycled at C/20 (2x), C/10 (5x), C/5 (5x) and C/2 (2x) (1C = 239 mA g_CAM⁻¹), under an applied pressure of 1.71 MPa at 60 °C. The cathode composite consisted of 40 weight % TiS₂, 50 weight % solid electrolyte and 10 weight % C65. galvanostatic cycling of the cells at C-rates between C/20 and 1C under 1.71 MPa at 60 °C shows a strong dependence of the capacity fading on the TiS₂ : Li₄B₁₀H₁₀B₁₂H₁₂ ratio. Initially, regardless of the TiS₂ : Li₄B₁₀H₁₀B₁₂H₁₂ ratio, all cells show a significant decrease in discharge capacity in the first few cycles. At the TiS₂ : Li₄B₁₀H₁₀B₁₂H₁₂ ratios of 40 : 50 and 60 : 30, discharge capacities of up to 200 mA h g⁻¹ (corresponding to a TiS₂ utilization rate of 83% compared to the theoretical capacity of 239 mA h g⁻¹) were achieved at the end of the first cycle, as shown in Fig. 4b. With further cycling, the capacity loss between successive cycles is less pronounced. The discharge capacity decreases from 180 to 170 mA h g⁻¹ in the first five cycles at C/10, and from 160 to 155 mA h g⁻¹ in the subsequent five cycles at C/5. For higher TiS₂ : Li₄B₁₀H₁₀B₁₂H₁₂ ratios of 80 : 10, the discharge capacity drops more rapidly in the first cycles, which we attribute to a loss of percolation of the ion-conducing pathways. A cycling experiment with a TiS₂ : Li₄B₁₀H₁₀B₁₂H₁₂ ratio of 40 : 50, cycled at C/5 after two initial cycles at C/10, showed discharge capacities of 250, 160, and 135 mA h g⁻¹ in the first three cycles (see Fig. S7a†). The cell then stabilizes over the next 35 cycles and reaches a discharge capacity of 105 mA h g⁻¹ until it fails in the cycle 37.

Fig. 4 (a) Charge–discharge voltage profiles for the first 17 cycles at indicated C-rates of a TiS₂|Li₄B₁₀H₁₀B₁₂H₁₂|Li cell under 1.71 MPa at 60 °C. The cathode composite contained 40 weight % TiS₂, 50 weight % solid electrolyte, and 10 weight % carbon. (b) C-rate tests of composite TiS₂|Li₄B₁₀H₁₀B₁₂H₁₂|Li cells with weight ratios of TiS₂ : Li₄B₁₀H₁₀B₁₂H₁₂ : carbon as indicated in the legend.

Analogous cells using LiFePO₄ as the cathode active material, i.e., cathode composites of LiFePO₄ : Li₄B₁₀H₁₀B₁₂H₁₂ : C65 with a mass loading of 1 mg_CAM cm⁻², showed low initial discharge capacities below 10 mA h g⁻¹ when cycled at C/20 at 60 °C. The initial discharge capacity increased after reducing the mass loading to 0.5 mg_CAM cm⁻² and adding 5 weight % polyvinylidene fluoride (PVDF) as a binder. The highest initial discharge capacity of 110 mA h g⁻¹ (corresponding to a LiFePO₄ utilization ratio of 73% compared to the theoretical capacity of 150 mA h g⁻¹) was obtained with a cathode composite having a LiFePO₄ : Li₄B₁₀H₁₀B₁₂H₁₂ : C65 : PVDF weight ratio of 30 : 50 : 15 : 05. The charge–discharge curves with this cathode composite at different C-rates are shown in Fig. 5a. Fig. 5b shows the corresponding discharge capacity at different C-rates, together with those obtained with different cathode compositions. The cells suffer from severe capacity loss, typically more than 50% during the first five cycles at C/20. The discharge capacity also decreases significantly with increasing C-rates, as shown in Fig. 5b. The pronounced capacity degradation can be explained by electrolyte decomposition as the lithium de-/intercalation potential of LiFePO₄ of 3.5 V vs. Li⁺/Li is above the oxidative stability limit of the [B₁₀H₁₀]²⁻ ion.

Fig. 5 (a) Charge–discharge voltage profiles for the first 25 cycles at indicated C-rates of a LiFePO₄| Li₄B₁₀H₁₀B₁₂H₁₂|Li cell under 1.71 MPa at 60 °C. The cathode composite contained 30 weight % LiFePO₄, 50 weight % Li₄B₁₀H₁₀B₁₂H₁₂, 15 weight % carbon, and 5 weight % PVDF. (b) Rate capability of the LiFePO₄|Li₄B₁₀H₁₀B₁₂H₁₂|Li cells at different C-rates with ratios of LiFePO₄ : Li₄B₁₀H₁₀B₁₂H₁₂ : carbon : binder as indicated in the legend.

A cycling experiment of a cell with a cathode composed of 30% LiFePO₄, 50% solid electrolyte, 15% carbon, and 5% PVDF (by mass), cycled at C/5 after two initial cycles at C/10, showed discharge capacities of 87, 69, and 37 mA h g⁻¹ in the first three cycles (see Fig. S7b†). Thereafter, the capacity loss per cycle decreases, and the discharge capacity drops below 10 mA h g⁻¹ within 30 cycles.

Conclusions

We have successfully used cation exchange to prepare Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ from Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂, respectively, demonstrating the stability of the [B₁₀H₁₀]²⁻ and [B₁₀H₁₀]²⁻ anions in aqueous solution.

Furthermore, we have prepared stoichiometric mixtures of Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ by ball milling, which exhibit ionic conductivities exceeding those of the parent compounds by several orders of magnitude. While the conductivity of the 1 : 1 mixture of mixed Na₂B₁₀H₁₀ and Na₂B₁₂H₁₂ was reported to be around 9 × 10⁻⁴ S cm⁻¹ at room temperature, the conductivity of the lithium analogue was found to be 4 × 10⁻⁴ S cm⁻¹. Within the mixture, the constituent anions remain intact. Li₂B₁₀H₁₀ and Li₂B₁₂H₁₂ are oxidatively stable up to 3.1 V and 3.6 V vs. Li⁺/Li, respectively. The resulting compound is thermally stable up to at least 300 °C in an inert atmosphere. This result is consistent with several other hydridoborates, in which (i) superionic conductivity is achieved by anion mixing and (ii) the ionic conductivity of the sodium salts exceeds the ones of the analogous lithium salts. However, the Li₂B₁₀H₁₀–Li₂B₁₂H₁₂ mixture is unique in that it allows the coexistence of one disordered anion with the second one ordered in the same crystal structure.

Li₄B₁₀H₁₀B₁₂H₁₂ has been used as the solid electrolyte in solid-state cells with a lithium metal anode and TiS₂ or LiFePO₄ as the cathode active material.

Despite calculations suggesting that closo-hidridoborates such as Li₂B₁₂H₁₂ may not be electrochemically stable versus lithium,⁷ stable cycling of solid-state hydrioborate-based batteries have been demonstrated in the past,^20,22,23,40 indicating kinetic stabilization, probably due to the formation of protective interfaces. Morphological instabilities and dendrite formation, due to the mechanical properties of hydridoborates, especially from their relatively low shear moduli, may be a greater challenge on the anode side than electrochemical stability.⁷ To mitigate this obstacle, an applied pressure of 1.71 MPa was used during cycling in the current study.

Discharge capacities of 83% and 73% of the theoretical capacity were achieved on first discharge for TiS₂ and LiFePO₄ as cathode active material, respectively. However, both cell types require high amounts of solid electrolyte in the cathode composite and suffer from severe capacity fading. The mass loading and the active material to solid electrolyte ratio need to be further optimized to build a competitive solid-state battery. Cathode active materials such as LiFePO₄, where the lithiation potential lies outside the electrochemical stability window of the electrolyte, require measures to protect Li₄B₁₀H₁₀B₁₂H₁₂ from decomposition to achieve stable cycling. We believe that further cathode engineering, e.g., optimization of the cathode composition and microstructure, and electrolyte engineering either by chemical modification of the electrolyte itself or by the addition of nano-sized oxides, will make hydridoborate-based solid electrolytes suitable for competitive high-performance solid-state batteries.

Author contributions

A. Garcia and G. Müller prepared the samples, determined the thermal stability, the ionic conductivity and performed the battery cycling. A. Garcia performed the ion exchange. R. Černý performed the Rietveld refinement and structural analysis, D. Rentsch performed the NMR measurements, R. Asakura designed the pressure cells and performed the oxidative stability measurements. C. Battaglia and A. Remhof conceived and supervised the project and acquired the funding. A. Remhof drafted the manuscript with input from all authors. All authors participated in the preparation of the manuscript and discussing the results.

Conflicts of interest

The authors declare no conflicts of interest or competing interests.

Acknowledgements

The authors thank Innosuisse – Swiss Innovation Agency for funding under contract number 49729.1 IP-EE. The NMR hardware was partially granted by the Swiss National Science Foundation (SNSF, grant no. 206021_150638/1).

Notes and references

1. S. Randau, D. A. Weber, O. Kötz, R. Koerver, P. Braun, A. Weber, E. Ivers-Tiffée, T. Adermann, J. Kulisch, W. G. Zeier, F. H. Richter and J. Janek, Nat. Energy, 2020, 5, 259.
2. A. Manthiram, X. Yu and S. Wang, Nat. Rev. Mater., 2017, 2, 16103.
3. T. Schmaltz, T. Wicke, L. Weymann, P. Voß, C. Neef and A. Thielmann, Solid-State Battery Roadmap 2035+, Fraunhofer ISI, Karlsruhe, Germany, 2022.
4. M. A. Beckett, B. Brellochs, I. T. Chizhevsky, T. Damhus, K.-H. Hellwich, J. D. Kennedy, R. Laitinen, W. H. Powell, D. Rabinovich, C. Viñas and A. Yerin, Pure Appl. Chem., 2020, 92, 355.
5. M. Brighi, F. Murgia and R. Černý, Cell Rep. Phys. Sci., 2020, 1, 100217.
6. R. Asakura, A. Remhof and C. Battaglia, in Solid State Batteries Volume 1: Emerging Materials and Applications, ed. R. K. Gupta, American Chemical Society, Washington, DC, 2022, ch. 14, vol. 1413, p. 353.
7. Z. Lu and F. Ciucci, Chem. Mater., 2017, 29, 9308.
8. B. Zhang, R. Tan, L. Yang, J. Zheng, K. Zhang, S. Mo, Z. Lin and F. Pan, Energy Stor. Mater., 2018, 10, 139.
9. W. S. Tang, A. Unemoto, W. Zhou, V. Stavila, M. Matsuo, H. Wu, S.-I. Orimo and T. J. Udovic, Energy Environ. Sci., 2015, 8, 3637.
10. W. S. Tang, M. Matsuo, H. Wu, V. Stavila, W. Zhou, A. A. Talin, A. V. Soloninin, R. V. Skoryunov, O. A. Babanova, A. V. Skripov, A. Unemoto, S. I. Orimo and T. J. Udovic, Adv. Energy Mater., 2016, 6, 1502237.
11. T. J. Udovic, M. Matsuo, W. S. Tang, H. Wu, V. Stavila, A. V. Soloninin, R. V. Skoryunov, O. A. Babanova, A. V. Skripov, J. J. Rush, A. Unemoto, H. Takamura and S. Orimo, Adv. Mater., 2014, 26, 7622.
12. T. J. Udovic, M. Matsuo, A. Unemoto, N. Verdal, V. Stavila, A. V. Skripov, J. J. Rush, H. Takamura and S.-I. Orimo, Chem. Commun., 2014, 50, 3750.
13. H. Wu, W. S. Tang, V. Stavila, W. Zhou, J. J. Rush and T. J. Udovic, J. Phys. Chem. C, 2015, 119, 6481.
14. M. Paskevicius, M. P. Pitt, D. H. Brown, D. A. Sheppard, S. Chumphongphan and C. E. Buckley, Phys. Chem. Chem. Phys., 2013, 15, 15825.
15. R. Černý, M. Brighi and F. Murgia, Chemistry, 2020, 2, 805.
16. M. Jørgensen, P. T. Shea, A. W. Tomich, J. B. Varley, M. Bercx, S. Lovera, R. Černý, W. Zhou, T. J. Udovic, V. Lavallo, T. R. Jensen, B. C. Wood and V. Stavila, Chem. Mater., 2020, 32, 1475.
17. S. Payandeh, R. Asakura, P. Avramidou, D. Rentsch, Z. Łodziana, R. Černý, A. Remhof and C. Battaglia, Chem. Mater., 2020, 32, 1101.
18. S. Payandeh, D. Rentsch, Z. Łodziana, R. Asakura, L. Bigler, R. Černý, C. Battaglia and A. Remhof, Adv. Funct. Mater., 2021, 31, 2010046.
19. W. S. Tang, K. Yoshida, A. V. Soloninin, R. V. Skoryunov, O. A. Babanova, A. V. Skripov, M. Dimitrievska, V. Stavila, S.-I. Orimo and T. J. Udovic, ACS Energy Lett., 2016, 1, 659.
20. S. Kim, H. Oguchi, N. Toyama, T. Sato, S. Takagi, T. Otomo, D. Arunkumar, N. Kuwata, J. Kawamura and S.-I. Orimo, Nat. Commun., 2019, 10, 1081.
21. S. Kim, K. Kisu, S. Takagi, H. Oguchi and S.-I. Orimo, ACS Appl. Energy Mater., 2020, 3, 4831.
22. R. Asakura, D. Reber, L. Duchêne, S. Payandeh, A. Remhof, H. Hagemann and C. Battaglia, Energy Environ. Sci., 2020, 13, 5048.
23. F. Murgia, M. Brighi and R. Černý, Electrochem. Commun., 2019, 106, 106534.
24. L. Duchêne, R. S. Kühnel, D. Rentsch, A. Remhof, H. Hagemann and C. Battaglia, Chem. Commun., 2017, 53, 4195.
25. R. Asakura, L. Duchêne, S. Payandeh, D. Rentsch, H. Hagemann, C. Battaglia and A. Remhof, ACS Appl. Mater. Interfaces, 2021, 13, 55319.
26. L. Duchêne, R. S. Kühnel, E. Stilp, E. Cuervo Reyes, A. Remhof, H. Hagemann and C. Battaglia, Energy Environ. Sci., 2017, 10, 2609.
27. L. Duchêne, D. H. Kim, Y. B. Song, S. Jun, R. Moury, A. Remhof, H. Hagemann, Y. S. Jung and C. Battaglia, Energy Stor. Mater., 2020, 26, 543.
28. C. Zhou, J. B. Grinderslev, L. N. Skov, M. Jørgensen, Y. Li, J. Skibsted, Y. Yan and T. R. Jensen, J. Mater. Chem. A, 2022, 10, 16137.
29. A. Coelho, J. Appl. Crystallogr., 2000, 33, 899.
30. J.-H. Her, M. Yousufuddin, W. Zhou, S. S. Jalisatgi, J. G. Kulleck, J. A. Zan, S.-J. Hwang, J. R. C. Bowman and T. J. Udovic, Inorg. Chem., 2008, 47, 9757.
31. R. Asakura, L. Duchêne, R.-S. Kühnel, A. Remhof, H. Hagemann and C. Battaglia, ACS Appl. Energy Mater., 2019, 2, 6924.
32. A. Blake, Electronic Thesis Collection, Master thesis, Pittsburg State University, 2015.
33. L. M. de Kort, V. Gulino, P. E. de Jongh and P. Ngene, J. Alloys Compd., 2022, 901, 163474.
34. V. Gulino, L. Barberis, P. Ngene, M. Baricco and P. E. de Jongh, ACS Appl. Energy Mater., 2020, 3, 4941.
35. S. Kim, N. Toyama, H. Oguchi, T. Sato, S. Takagi, T. Ikeshoji and S.-I. Orimo, Chem. Mater., 2018, 30, 386.
36. W. S. Tang, M. Matsuo, H. Wu, V. Stavila, A. Unemoto, S.-I. Orimo and T. J. Udovic, Energy Stor. Mater., 2016, 4, 79.
37. J. A. Teprovich, H. Colón-Mercado, A. L. Washington Ii, P. A. Ward, S. Greenway, D. M. Missimer, H. Hartman, J. Velten, J. H. Christian and R. Zidan, J. Mater. Chem. A, 2015, 3, 22853.
38. R. Moury, Z. Łodziana, A. Remhof, L. Duchêne, E. Roedern, A. Gigante and H. Hagemann, Inorg. Chem., 2022, 61, 5224.
39. M. Brighi, F. Murgia, Z. Łodziana and R. Černý, Inorg. Chem., 2022, 61, 5813.
40. C. Zhou, H. Sun, Q. Wang, J. B. Grinderslev, D. Liu, Y. Yan and T. R. Jensen, J. Alloys Compd., 2023, 938, 168689.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ta03914e

---