Nido‐Hydroborate‐Based Electrolytes for All‐Solid‐State Lithium Batteries

Hydroborate‐based solid electrolytes have recently been successfully employed in high voltage, room temperature all‐solid‐state sodium batteries. The transfer to analogous lithium systems has failed up to now due to the lower conductivity of the corresponding lithium compounds and their high cost. Here LiB11H14 nido‐hydroborate as a cost‐effective building block and its high‐purity synthesis is introduced. The crystal structures of anhydrous LiB11H14 as well as of LiB11H14‐based mixed‐anion solid electrolytes are solved and high ionic conductivities of 1.1 × 10−4 S cm−1 for Li2(B11H14)(CB11H12) and 1.1 × 10−3 S cm−1 for Li3(B11H14)(CB9H10)2 are obtained, respectively. LiB11H14 exhibits an oxidative stability limit of 2.6 V versus Li+/Li and the proposed decomposition products are discussed based on density functional theory calculations. Strategies are discussed to improve the stability of these compounds by modifying the chemical structure of the nido‐hydroborate cage. Galvanostatic cycling in symmetric cells with two lithium metal electrodes shows a small overpotential increase from 22.5 to 30 mV after 620 h (up to 0.5 mAh cm−2), demonstrating that the electrolyte is compatible with metallic anodes. Finally, the Li2(B11H14)(CB11H12) electrolyte is employed in a proof‐of‐concept half cell with a TiS2 cathode with a capacity retention of 82% after 150 cycles at C/5.


Introduction
Solid-state electrolytes combining liquid-like ionic conductivity, stability against metallic anodes, compatibility with highvoltage cathodes, processability, and low environmental impact and cost are essential for next-generation all-solid-state batteries 11 H 14 LiB 11 H 14 was synthesized based on the reaction of n-butyllithium (LiC 4 H 9 , BuLi) with (CH 3 ) 3  After THF was removed, water was added to filter unreacted, insoluble (CH 3 ) 3 NHB 11 H 14 , resulting in an aqueous solution of LiB 11 H 14 (H 2 O) n (1 < n < 2). The complete removal of the coordinated water without decomposition of the B 11 H 14 − cage is challenging. Our initial attempts using heat treatment led to the formation of LiB 11 H 13 R impurities (R = OH, OB 11 H 13 Li, O-Butyl(O-Bu)) and LiB 11 H 12 O (Samples A-D, Table S1, Supporting Information). Figure 1a shows the open-cage molecular structure of the nido-hydroborate anion. The impurities were identified by 11 B{ 1 H} NMR (see Figure 1b; and Table S1 (Supporting Information); for discussion, see Section 2.2 below).

Experimental Procedure to Synthesize Solvent-Free LiB
The synthesis of pure LiB 11 H 14 was made possible by using an excess of n-butyllithium in the synthesis (Equation (1)). This leads to a partial reduction of the sample, resulting in a mixture of LiB 11 It is important to mention that H 2 is released for the synthesis of (CH 3 ) 3 NHB 11 H 14 as a precursor. [13] [31] Therefore, H 2 release during the synthesis should be considered for the large-scale production of these materials. However, similar compounds such as M 2 B 12 H 12 (M = Li, Na) in which H 2 is released during the synthesis are commercialized and produced on a kg scale. [32] Figure 1c displays the respective 11 B{ 1 H} NMR spectra. The complete removal of water was evidenced by powder X-ray powder diffraction (PXRD), IR spectroscopy, and thermogravimetric analysis (TGA, see Figures S1 and S2, Supporting Information). In the last step, LiB 11 H 14 is separated by adding THF to the mixture to filter THF-insoluble LiOH and by subsequent drying at 100 °C for 16 h to remove THF. This procedure also allows the removal of coordinated water from commercial LiB 11 H 14 (H 2 O) n samples and is described in more detail in the Supporting Information (Samples E, G). Finally, anion mixing was used as a known strategy to improve ionic conductivity. [15,33] The pure LiB 11 H 14 was ball-milled with LiCB 9 H 10 , LiCB 11 H 12, or Li 2 B 12 H 12 in different molar ratios (Table S1, Supporting Information). The compositions of the samples were determined by peak shape analysis of the 1D 11 B NMR spectra. The results are summarized in Table S1 (Supporting Information). We found that LiB 11 H 14 partially decomposed during the ball-milling process. This leads to the formation of nonassignable resonances in the 11 B NMR data (Table S1 and Figure S3, Supporting Information).

Identification of LiB 11 H 13 R By-Products of LiB 11 H 14 Synthesis
LiB 11 H 14 (1): The molecular structure of 1 depicted in Figure 1a has been fully determined previously in ref. [13] It can be described as an oblate spheroid with the minor axis (the "height") perpendicular to the opening of the nest of 4.6 Å and a major axis (the "width" or the "diameter") of 5.3 Å (Figures 2 and 3a).  11 11 B HSQC, and f) 11 B-11 B{ 1 H} COSY NMR spectra with the assignment of resonances to the chemical structure of 2 in Sample F. Please note that the signals of the starting material 1 account to ≈93 mol% of the total signal intensity. The NMR spectra shown in b,c) are recorded in THF and d-f) in CD 3 CN solutions. g) Expanded regions of interest of the (-)-ESI-MS with 11 B/ 10 B isotopic patterns assigned to 2-5 for Samples B, C, and F.  (2): The chemical structures of the abovementioned impurities, which were formed during the drying process (Samples, B, C, and F; see Table S1, Supporting Information), were unveiled using NMR and MS ( Figure S4 and Table S2, Supporting Information). The 1D and 2D 1 H and 11 B NMR data of Sample F leads to the identification of LiB 11 H 13 OH (2) as a first by-product as shown in Figure 1d-f. Besides the strong resonances of the starting material, six distinct signals with an intensity ratio of 1:2:2:2:2:1 are observed in the 1D 11 B NMR spectrum ( Figure 1d) and in the 1 H-11 B HSQC NMR spectrum, each of these resonances results in a distinct cross peak assigned to positions 2-7 of the chemical structure (Figure 1a). Note that the cross signal of position 1 in the spectrum is hidden beyond the strong resonances of LiB 11 (Figure 1e), which therefore can be assigned to positions 5-7 of the chemical structure (Figure 1a, bonds of protons H br are shown with dashed lines). From 11 B NMR chemical shift reasons and the relative signal intensity, the resonance at 16.9 ppm can be assigned to B7. Position B2 was assigned from the cross peak between the boron resonances at 16.9 and −22.6 ppm in the 11 B-11 B correlated NMR spectrum ( Figure 1f). The cross peak at −14.3/−22.6 ppm observed in the 11 B-11 B correlated NMR spectrum leads to the unambiguous identification of B1, which was hidden beyond strong resonances of the starting material in the other 1D and 2D NMR data. Subsequently, all 1 H and 11 B NMR chemical shifts of the molecule were assigned (for enlarged regions of 1 H-1 H and 11 B-11 B correlated NMR spectra see Figure S6, Supporting Information). The presence of an OH group in 2 (in Sample F) was confirmed by the correlation observed at 4.00 ppm to the bridging protons H br at −2.21 ppm in the 1 H-1 H correlated NMR spectrum ( Figure S6e, Supporting Information). All other observable 1 H-1 H correlations originate from 2 J or 3 J couplings and support the signal assignments to the chemical structure of the boron cage shown in Figure 1a. Please note that the boronboron bonds of B2B2, B5B5, and B6B7 could not be verified to fully identify the boron cage. For reasons of symmetry, the boron atoms in the first two cases are homotopic and therefore have identical 11 B NMR chemical shifts, and relaxation reasons may be responsible for the missing B6B7 cross peak (see, e.g., [34][35][36] ).
The presence of LiB 11 H 13 OH (2) in Samples B and F was further confirmed by their corresponding mass spectra ( Figure 1g). [37] The 11 B/ 10 B isotopic patterns with the most abundant signals at m/z 149.215 are assigned to the molecular composition of B 11 H 13 OH − . A calculated MS pattern is presented in Figure S7 (Supporting Information) to confirm the assignment to B 11 H 13 OH − and the summary of all MS data is given in Table S2 (Supporting Information).  Figure S9a, Supporting Information), the signal of 2 is considerably reduced with increasing gradient strengths compared to the resonance of 3. This indicates that 3 diffuses less rapidly and must probably have a higher molecular weight than 2. In the mass spectrum of Sample B, the most abundant signal of the 11  This finding confirms the hypothesis from the NMR data that 3 should have a similar molecular composition but a higher molecular weight than 2. The formation of a dimer with a bridging oxygen seems not to be uncommon for closo-hydroborates as it has been previously observed for other species such as [B 10 H 13 -O-B 10 H 13 ]. [38] Li(B 11 H 13 -O-Bu) (4): For species 4, the 1 H-11 B and 11 B-11 B correlated 2D NMR data ( Figure S12, Supporting Information) suggest a boronboron network similar to the data evaluated for 2 and 3. In the 1 H and 13 C NMR spectra ( Figure S13, Supporting Information) resonances assignable to an O-n-butyl (O-Bu) group were identified. For chemical shift reasons, the signal at 70.7 ppm in the 13 C NMR spectrum can be assigned to an n-butyl ether residue, but NMR data further suggest that 4 consists of a B 11 H 13 core unit with an attached butoxy group at the open apex boron. This was confirmed by mass spectrometry of Sample C with a prominent 11 Figure S9b, Supporting Information) the signal of 4 is significantly reduced with increasing gradient strengths compared to the resonance of 3, which is associated with the lower molecular weight of 4 compared to 3. It is important to mention that the formation of compound 4 was observed only when LiB 11 H 14 (H 2 O) n is heattreated in THF. Heat treatment of LiB 11 H 14 in THF did not form compound 4, suggesting that the coordinating water plays a role in CO bond scission of tetrahydrofuran and its attachment to the boron cage. The formation of 4 was reinvestigated in presence of THF-d 8 as a reactant. From the 1D and 2D 1 H, 2 H, 11 B, and 13 C NMR data (see Figure S15 and discussion in the Supporting Information part), we unequivocally identified O-CD 2 -CD 2 -CD 2 -CD 2 H as the butoxy residue of 4.
LiB 11 H 12 O (5): Besides the impurities 1-4 discussed above, a tiny amount of a further impurity was observed in Sample F. In this sample, in addition to the resonances of 1 and 2, crosspeaks of a minor by-product assignable to LiB 11 H 12 O (5) were observed in the 1 H-11 B HSQC NMR spectrum of Sample F when a very low threshold was applied ( Figure S16, Supporting Information). All 1 H and 11 B NMR data of this species are very similar to the data described in ref. [34,35] This assignment was further supported by the mass spectrum of Sample  An overview of the 11 B NMR spectra with integrated regions of selected samples is presented in Figure S17 (Supporting Information) and global assignments of 11 B NMR chemical shift regions of compounds 1-5 are shown in Table S3 (Supporting  Information) Understanding the chemical composition of the impurities is the basis to establish a high purity synthesis method for LiB 11 H 14 , as discussed earlier. In the next sections, the effect of anion mixing (including the impurities) on the crystal structures and ionic conductivities are discussed. Moreover, the effect of impurities on the electrochemical stabilities of the compounds is investigated using density functional theory (DFT) calculations.
The crystal structure of the low temperature (LT) LiB 11 H 14 was solved in an orthorhombic unit cell with space group Pbca and cubic-close packing (ccp) of the B 11 H 14 − anions and was optimized by DFT (Figures S19, S20a, and S22, Supporting Information). Unlike tetrahedral coordination of Na in NaB 11 H 14 , [13] Li + ions in this structure lie in trigonal planar sites formed by three B 11 H 14 − anions and located between octahedral and tetrahedral sites (Figure 2b). Trigonal planar coordination of Li has been also observed for other Li salts of hydroborates, such as Li 2 B 12 H 12 [39] and Li 2 B 10 H 10 . [40] In general, the preferred coordination environment of the cation is defined based on the cation-to-anion size ratio and by using the first Pauling rule (Table S4, Supporting Information). [41,42] Taking 0.73 Å as the size of Li + (crystal radius in tetrahedral coordination [43] ) and 4.6 Å as the estimated size of the B 11 H 14 − anion (apical axis, Figure 3a), the cation-to-anion size ratio is ≈0.16, suggesting triangular coordination. [41] However, in NaB 11 H 14 with a larger cation (Na + ≈1.13 Å), crystal radius in tetrahedral coordination, [43] ), the cation/anion size ratio increases to ≈0.25, and tetrahedral coordination is preferred for Na + ions. Similarly to NaB 11 H 14 , the B 11 H 14 − anion in LT-LiB 11 H 14 is oriented with the open side of the cage away from the Li + cation. This preferred orientation is explained by the more negative closed side of the cage that interacts with the cation (Figure 3b). A comparable trend has been observed for CB 11 H 12 − and CB 9 H 10 − in which the anions are oriented so that the considerably more positive  Figure 3b). [44,26] Upon heating, LiB 11 H 14 undergoes a phase transition at ≈112 °C to a cubic phase with the Fm3 space group and ccp packing of the anions similar to high temperature (HT) phases of Li 2 B 12 H 12 and LiCB 11 H 12 ( Figure 1c; and Figure S2b, Supporting Information). [44,45] The coordination environment of Li remains trigonal planar and 3D conduction channels appear to form for the Li + cations in ccp anion packing ( Figure S20b, Supporting Information). The Li + jumps between tetrahedral and octahedral sites occur through the shared faces where Li + is localized in a disordered manner ( Figure S21a, Supporting Information). [46] Preliminary DFT calculations of the conduction mechanism show Li + hopping near the vicinity of tetrahedral sites, and large Li + -Li + correlations in electrolytes containing a fraction of CB 9 H 10 − . This points out to lower activation energy of mixed anion compounds ( Figure S21b, Supporting Information) Ion conduction mechanism of related compounds is discussed in detail. [47] The sizes and polarities of the hydroborate anions under discussion are summarized in Figure 3. The charges were calculated in the simplified systems that consider one isolated hydroborate anion and cations compensating the charge of the anion. This model neglects the periodic crystal structure; however, it allows observation of the charge redistribution on the anions related to the motion of Li (Figure 2d,e; and Figure S23, Supporting Information). Both structures have the ccp packing of the anions and triangular coordination for Li. In the case of mixing LiB 11 H 14 with LiCB 9 H 10 as an example for a smaller anion (Samples M, N), the HT phase of LiCB 9 H 10 with a hexagonal unit cell and with the space group P31c is stabilized. This is confirmed by the disappearance of the Bragg reflection at 2θ = 8.6° for the LT phase and appearance of a shoulder at 2θ = 16.6° for the HT phase (red and green arrows in Figure S24, Supporting Information). [15,25] In this structure, Li atoms occupy tetrahedral sites sharing a face with an empty octahedral site (Figure 2f). The higher coordination of lithium is based on the smaller size of anions used in this compound and is in favor of Li-ion conductivity (based on the Pauling rule, Table S4, Supporting Information). [42] In addition, the presence of small amount of impurities  Table 1. The atomic coordinates of all solved structures are presented in Tables S8-S12 (Supporting Information). Figure 4a shows the thermal behavior of LiB 11 H 14 and the mixed hydroborate compounds as observed by differential scanning calorimetry. The reversible phase transition from the orthorhombic to the cubic phase of LiB 11 H 14 appears upon heating as an endothermic peak at ≈112 °C (Figure 4a,b; and Figure S26c, Supporting Information). Upon cooling, LiB 11 H 14 transforms back to the orthorhombic phase with an exothermic event at ≈96 °C. Thermo-gravimetric measurements of LiB 11 H 14 show an ≈3% weight loss at ≈180 °C ( Figure S2b, Supporting Information), indicative of sample decomposition. Moreover, with prolonged heat treatment at 120 °C, the sample color changes from white to yellow and extra resonances in the 11 B NMR spectra appear, indicating thermal instability of LiB 11 H 14 above the phase transition ( Figure S25, Supporting Information). The decreased signal intensity in the 11 B NMR spectra indicates the insolubility of the decomposition products, indicative of the formation of dimers or poly-hydroborates with higher molecular masses. [48,49] The phase-transition temperature of LiB 11 H 14 at 112 °C is lower than the ones of Li 2 B 10 H 10 and Li 2 B 12 H 12 (367 and 355 °C, respectively), [40,50] and is comparable to the phase-transition temperature of LiCB 11 H 12 at ≈122 °C. [44] For the LiB 11 H 14 :Li 2 B 12 H 12 (1:1, Sample L) mixture, an irreversible endothermic event is observed at T ≈120 °C for the first heating cycle. In the subsequent cycle, this event shifts to a lower temperature and fades (Figure 4a; and  Figure S26b,e, Supporting Information) which is similarly observed for other mixed anion systems. [13,[15][16][17]33,51] Figure 5 displays the temperature-dependent lithium-ion conductivity of the synthesized compounds. For comparison, the conductivities of Li 2 B 12 H 12 , LiCB 11 H 12 , and LiCB 9 H 10 are also presented. LiB 11 H 14 shows a conductivity of 1.5 × 10 −6 S cm −1 at 25 °C which is similar to the reported conductivity of LiCB 10 H 13 , but higher than the published data for LiCB 9 H 10 , LiCB 11 H 12 , and Li 2 B 12 H 12 . [14,25,44,52,53] The apparent activation energy for LiB 11 H 14 is 0.81 eV (Sample E) at low temperatures and reduces to 0.16 eV above the phase transition at ≈112 °C ( Figures S28a  and S29, Supporting Information).     [13,15,17] A transition from a low-temperature regime with a correlated cationic motion to a high-temperature regime with noncorrelated cationic diffusion has been suggested to be responsible for this behavior. [54] The conductivity trends indicate that small, single charged anions with high polarity are required to form suitable conduction channels for Li + cations and to obtain high conductivities in Li-hydroborates. Although B 11 H 14 − and CB 9 H 10 − have similar sizes and polarities (Figure 3), the prolate (elongated) shapes of the CB 9 H 10 − anion appear to be more favorable for Li + conduction pathways by stabilizing the disordered hexagonal unit cell. However, the synthesis of CB 9 H 10 − is challenging and the use of expensive/toxic reactants cannot yet be avoided. [30] Therefore, in the following, we assess LiB 11 H 14 :LiCB 11 H 12 (1:1, Sample H) as a solid electrolyte (SE) for further electrochemical measurements and battery assembly.

Conductivity
In inorganic solid electrolytes, the ionic conductivity is predominately attributed to cation mobility and the anion mobility is negligible. [55] The electronic conductivity of Sample E was determined to be ≈5 × 10 −9 S cm −1 at 60 °C, by measuring the current decay in a Li/SE/Cu after applying constant voltages of 0.25 to 1 V ( Figure S30, Supporting Information). The obtained electronic conductivity is one order of magnitude lower than 5 × 10 −8 and 2.5 × 10 −8 S cm −1 obtained at 60 °C for the typical solid electrolytes, Li 7 La 3 Zr 2 O 12, and Li 3 PS 4 , respectively. [56]

Electrochemical Performance
To evaluate the stability of the LiB 11 H 14 :LiCB 11 H 12 (1:1) SE against Li metal, a symmetric Li/SE/Li cell was assembled and the time evolution of the impedance spectra was measured at the open-circuit voltage. Impedance spectra were measured at intervals of 2 to 80 h, starting directly after contacting the Li metal with the SE to evaluate the interphase formation rate and its impact on the cell resistance over time. As depicted in Figure 6a, the evolution of the impedance was extracted using an equivalent circuit consisting of a constant phase element and two resistors representing the bulk and interphase resistances. Figure 6a also shows the Nyquist plots recorded at t = 6, 20, 40, 60, and 80 h. The extracted bulk and interphase resistances are presented in Figure 6b,c. Similar to sulfide-based solid electrolytes such as Li 10 GeP 2 S 12 and Li 3 PS 4 , our SE is not stable against Li metal and slowly decomposes over time. [57] The growth rate of the interphase obeys a parabolic rate law with a rate constant of k = 14.6 Ωcm 2 h −0.5 obtained from the slope in Figure 6c, which is lower than the 45.1 and 1394.3 Ωcm 2 h −0.5 reported for Li 10 GeP 2 S 12 and Li 6 PS 5 I, respectively. [58] However, this value is slightly higher than 2.9 and 3.8 Ωcm 2 h −0.5 published for Li 7 P 3 S 11 and Li 6 PS 5 Cl, respectively. [58] The initial interphase resistance immediately after contacting the SE with Li metal is also calculated as 29.6 Ωcm 2 from the intercept in Figure 6c.
As previous studies have confirmed the stability of the CB 11 H 12 − anion against Li/Na metal, [18,19] we attribute the decomposition reaction on the anode side to the reduction of the B 11 H 14 − anion in the electrolyte. To get a better insight into possible decomposition reactions at the anode and cathode, we performed electrochemical stability calculations for LiB 11 H 14 [59,60] (for details of the present approach, see the DFT section in the Supporting Information and Figure S33 and Table S5, Supporting Information). DFT calculations show that LiB 11 H 14 is unstable against Li metal, reducing to Li 2 B 11 H 13 (dashed rectangle in Figure 6f; and Table S5, Supporting Information). This reduction reaction requires the detachment of one proton and was also observed during the synthesis steps and when (CH 3 ) 3 NHB 11 H 14 was reacted with an excess amount of BuLi as a Li source (Figure 1c).
Although LiB 11 H 14 appears to be unstable against Li metal, the SE/Li interphase shows sufficient ionic conductivity to support reversible electrodeposition as shown by the galvanostatic cycling of a symmetric Li/SE/Li cell. The galvanostatic cycling experiment (Figure 6d) reveals the excellent stability of >620 h at a current density of 25 or 50 µA cm −2 , switching the current direction every 1 or 10 h (0.50 mA cm −2 ). A moderate increase in the overpotential from 22.5 to 30 mV after 620 h might be attributed to continuous and slow electrolyte reduction at the Li/SE interphase due to cracking of the formed Li 2 B 11 H 13 layer during electrodeposition. The critical current density (CCD) at which dendrites start to propagate through the SE was measured by increasing the current density in each cycle in a Li/SE/ Li cell while keeping the transferred capacity per half cycle constant at 0.2 mAh cm −2 ( Figure S31, Supporting Information). Using this method, a CCD of 0.16 mA cm −2 at 60 °C is determined. Similar CCD values of 0.04-0.28 mA cm −2 have been recently reported for single-crystalline Li 6.4 Ga 0.2 La 3 Zr 2 O 12 at ambient temperature. [61] Higher critical currents beyond 1 mA cm −2 typically require interface engineering to lower the interfacial resistance and external pressure in the MPa range applied on the cell to counteract void formation at the interface between Li metal and the electrolyte. [62][63][64] In the next step, we experimentally determined the oxidative stability of the SE following the methodology of Asakura  [65] The voltage of a SE/ carbon composite working electrode was swept against a Li metal reference/counter electrode from 2.0 to 4.0 V versus Li + / Li, as shown in Figure 6e. Similar to NaB 11 H 14 , [13] a clear onset of decomposition reactions is observed at ≈2.6 V versus Li + /Li followed by a second onset at 3.2 V versus Li + /Li. In the subsequent cycles, no significant current up to 4.0 V versus Li + /Li is observed during both forward and backward scans (see Figure S32, Supporting Information), indicating the irreversibility of the decomposition products and the formation of a passivating, blocking layer during the first cycle.
DFT calculations were used to predict the electrochemical stability of the LiB 11 H 14 together with other impurity compounds (2,3,5). It should be noted that the calculated electrochemical stability windows of the considered compounds depend on the selected decomposition products. Upon oxidation, electrolytes are typically delithiated. In this study, therefore, several poly-hydroborates with lower lithium content such as LiB 22 Table S5 and Figure S33 for the complete list, Supporting Information). [66] A B 22 H 22 2− compound was reported as the oxidation product of the B 11 H 14 − anion in solution [67] and trace amounts of B 21 H 18 − were observed in the MS data of Sample B (Table S2, Supporting Information). DFT calculations predict a narrow electrochemical window for LiB 11 H 14 (dashed bar in Figure 6f) when only B 18 H 22 as a completely delithiated and neutral borane is considered as the oxidation products ( Figure S33 and Table S5, Supporting Information). The calculated stability window of LiB 11 H 14 is comparable to Li 3 PS 4 [68] and is still larger than the calculated values for Li 10 GeP 2 S 12 . [59,69] The list of the  Table S5, Supporting Information). The electrochemical stability of the other compounds was calculated with and without consideration of LiB 21 H 18 as a final decomposition product in red and gray bars, respectively. g) Cycling performance and Coulombic efficiency of a Li/SE/TiS 2 cell cycled between 1.6 and 2.5 V versus Li + /Li at C/10 (1st to 5th cycle) and C/5 (from 6th cyle) (1C = 0.239 mA g −1 ) under 3.2 MPa. All measurements were performed at 60 °C. electrochemical stability window of the most common solid electrolytes and their phase equilibria at their stability limits are presented in Table S6 (Supporting Information).
The electrochemical stability windows of the rest of the compounds were calculated with and without considering LiB 21 H 18 as a decomposition product (red and gray bars in Figure 6f). DFT calculations suggest LiB 21 H 18 as the most thermodynamically stable compound among the hydroborates considered here and can be perceived as the thermodynamic sink. It means that this compound tends to form as the oxidation product and is stable up to ≈5 V versus Li + /Li before completely delithiating to B 18 H 22 (red bars in Figure 6f).
A second series of calculations was carried out without considering the LiB 21 H 18 as the decomposition product in order to stimulate that the formation of this compound is kinetically hindered. Under these assumptions, Li (Figure 6f) and points toward a new strategy to synthesize them as a pure phase for the next generation of solid electrolytes.
The electronic conductivity of the decomposition products could be estimated by the following measurement. The Li/SE/ Cu cell assembled for the electronic conductivity measurement shows an open-circuit voltage of ≈2.3 V. To measure the current response versus applied potential, constant voltages of 0.25, 0.5, 0.75, and 1 V were applied to the cell. When a constant voltage of ≥0.5 V is applied, the cell potential increases to >2.6 V versus Li + /Li, which is above the electrochemical stability, and the SE starts to decompose. In these cases, smaller current responses are observed, suggesting the lower electronic conductivities of the decomposition products as a preferred requirement for the formation of a stable interphase ( Figure S30b, Supporting Information).
Finally, a proof-of-concept battery using a Li metal anode and TiS 2 cathode was assembled in a home-built pressure cell. TiS 2 was selected as a cathode active material because of its lithium intercalation potential of ≤2.5 V versus Li + /Li matches the oxidative stability limit of the SE. The chemical compatibility between the SE and the active material was further confirmed with PXRD. Figure S34 (Supporting Information) compares the PXRD patterns of the TiS 2 , SE, and the SE/TiS 2 composite mixture annealed at 60 °C for 12 h (under vacuum, p = 10 −3 mbar). Unlike LiBH 4 that is unstable against TiS 2 , leading to spontaneous lithiation upon heating with a shift of the PXRD patterns to lower 2θ angles, [70] no apparent reaction is observed between the SE and TiS 2 by PXRD. No shift of the PXRD patterns and formation of crystalline decomposition products are detected. Galvanostatic cycling of the cell at C/10 and C/5 (1C = 239 mA g −1 ) under 3.2 MPa in Figure 6g shows a reversible discharge capacity of 190 mAh g −1 in the first cycle, corresponding to a TiS 2 utilization ratio of 80% compared to the theoretical capacity of 239 mAh g −1 . [71] Charge/discharge voltage profiles are presented in Figure S35 (Supporting Information). The Coulombic efficiency is as high as 98% in the first cycle, which stays at >99.5% in the later cycles. Switching C-rates to C/5 in the sixth cycle exhibits a minor decrease in discharge capacity (188 mAh g −1 ). The reversible discharge capacity at C/5 retained 82% after 150 cycles (from 6th cycle), indicating stable battery cycling with a negligible oxidative decomposition of the SE. The slow cell degradation over cycling is attributed to the contact loss between the SE and the cathode active material together with the slow reductive decomposition of the electrolyte by the Li metal anode.
Improvements in battery performance, as well as the use of high voltage cathodes, could be made by using coated cathodes. This is a common approach for using sulfide-based solid electrolytes beyond their stability window and in combination with high-voltage cathodes. Summary lists of the electrochemical stability window of the most common solid electrolytes and their performances in all-solid-state lithium batteries are presented in Tables S6 and S7 (Supporting Information). [68,72]

Conclusion
In this work, we present for the first time, the synthesis of anhydrous LiB 11 H 14 as a cost-effective building block for mixed-anion hydroborate lithium solid electrolytes. The impurity side products such as LiB 11  and liquid-like ionic conductivity of up to 1.1 × 10 −3 S cm −1 at 25 °C is achieved in a mixture with slightly smaller LiCB 9 H 10 in the Li 3 (B 11 H 14 )(CB 9 H 10 ) 2 . Therefore, contrary to the common trend that focuses on increasing the anion size and widening the ion conduction path, we propose focusing on small hydroborate anions which are more compatible with the small size of the Li + cation and enable suitable conduction channels.
LiB 11 H 14 is not stable against Li metal; however, the formation of the Li-rich Li 2 B 11 H 13 allows electrodeposition in a symmetric cell with Li metal electrodes at 50 µA cm −2 (up to 0.5 mAh cm −2 ) with a low overpotential increase which is stable over >620 cycles. The oxidative stability of 2.6 V versus Li + /Li is obtained for this electrolyte is similar to Li 3 PS 4 [68] but is narrower than that for closo-hydroborates due to the delocalized hydrogen atoms. [16,17] In order to achieve high gravimetric energy densities, cathode materials such as Li 2 C 6 O 6 [73] or sulfur [18] with a theoretical capacity of 590 and 1675 mAh g −1 , respectively, are suggested, which lie within the stability window. In this regard, we assembled a proof-of-concept battery using TiS 2 cathode which matches the oxidative stability of the electrolyte. The other common strategy is to use protective coatings together with high-voltage cathodes as a common method for sulfidebased ASSBs. [68,72] Finally, the intrinsic stability of the electrolyte could be improved by replacing hydrogen atoms with an

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.