Hydroborate Solid-State Lithium Battery with High-Voltage NMC811 Cathode

: Hydroborate solid electrolytes offer high ionic conductivity and are stable in contact with alkali metal anodes but are challenging to integrate into batteries with high-voltage cathodes. Here, we demonstrate stable dis-/charge cycling of solid-state Li batteries combining a Li 3 (CB 11 H 12 ) 2 (CB 9 H 10 ) hydroborate electrolyte with a 4 V-class LiNi 0.8 Mn 0.1 Co 0.1 O 2 (NMC811) cathode, exploiting the enhanced kinetic stability of the LiCB 11 H 12 -rich and LiCB 9 H 10 -poor electrolyte composition. Cells with Li metal and InLi anodes achieve a discharge capacity at C/10 of ∼ 145 mAh g − 1 at room temperature and ∼ 175 mAh g − 1 at 60 ° C. InLi cells retain 98% of their initial discharge capacity after 100 cycles at C/5 and 70% after 1000 cycles at C/2. Capacity retentions of 97% after 100 cycles at C/5 and 75% after 350 cycles at C/2 are also achieved with a graphite anode without any excess Li. The energy density per cathode composite weight of 460 Wh kg − 1 is on par with the best solid-state batteries reported to date.

L i-ion batteries are indispensable for a range of applications, from portable electronics to electric mobility.Solid-state batteries with solid electrolytes are expected to expand the energy density beyond the limits of current Li-ion batteries, 1−5 replace their flammable 6,7 liquid organic electrolytes with safer nonflammable alternatives, and enable faster charging. 8The most heavily investigated solid electrolyte classes are polymers, sulfides, oxides, and halides.Each of them has its own advantages and disadvantages.Polymers are flexible and easy to process, but their Li-ion conductivity, typically <1 mS cm −1 at room temperature, and thermal and electrochemical stability are not yet sufficient to establish a competitive battery technology. 9−13 Sulfides can reach Li-ion conductivities above 10 mS cm −1 and are mechanically soft. 8−23 Halides feature Li-ion conductivities of >1 mS cm −1 and high oxidative stability, but are typically unstable against Li metal. 24 −27 In contrast, hydroborates, a specific type of complex metal hydrides, have only recently emerged as solid electrolytes and offer distinct advantages. 28,29−33 Furthermore, hydroborates typically exhibit a low crystallographic density of only 1.0−1.2g cm −3 , 34−36 are mechanically soft, and can be crystallized from solution, which may facilitate large-scale production. 28,37The Na hydroborates typically exhibit higher ionic conductivities up to 70 mS cm −1 at room temperature 38 than their Li analogues, which reach up to 6.7 mS cm −1 . 31,39ome of us successfully demonstrated stable room temperature cycling of a hydroborate solid-state Na battery with a 4 V-class Na 3 (VOPO 4 ) 2 F cathode and a Na metal anode, delivering ∼100 mAh g −1 upon discharge. 32owever, stable long-term cycling of a high-capacity 4 Vclass cathode such as LiNi 0.8 Mn 0.1 Co 0.1 O 2 (NMC811) delivering up to 200 mAh g −1 has not yet been demonstrated.Most previous studies have focused on optimizing the ionic conductivity of Li hydroborates, reaching so far the highest value of 6.7 mS cm −1 for a 3:7 mixture of LiCB 11 H 12 and LiCB 9 H 10 . 31Systematic efforts to investigate and enhance the oxidative stability of Li hydroborates are lacking, which is why stable dis-/charge cycling of cells based on these Li hydroborates has so far been demonstrated exclusively for cells with low-voltage cathodes like S 31,40 or TiS 2 . 30,39,41,42ttempts to enable the integration of 4 V-class cathodes including a LiNbO 3 protective layer 43 or the use of a kinetically more stable, but less conductive Li 2 B 12 H 12 −Al 2 O 3 electrolyte 41 have so far resulted only in poor capacity retention (<80% after 20 cycles), suggesting that interface stability is still insufficient.
Excellent cycling stability over several hundred dis-/charge cycles is achieved with InLi and graphite anodes, demonstrating the advantages of the LiCB 11 H 12 -rich and LiCB 9 H 10 -poor 2:1 composition.Importantly, our results bring the hydroborates to a performance level comparable to the best solid electrolytes reported so far.Unlike many previous reports, we achieve this by using the very same solid electrolyte in the cathode composite, as the separator layer, and in contact with the anode.

■ TRADE-OFF BETWEEN IONIC CONDUCTIVITY AND OXIDATIVE STABILITY
To explore Li-hydroborate-based solid electrolytes that combine high ionic conductivity with high electrochemical oxidative stability, we first revisit the mixed-anion closocarbaborates of LiCB 11 H 12 and LiCB 9 H 10 .Figure 2a shows the temperature-dependent ionic conductivity of ball-milled LiCB 11 H 12 , LiCB 9 H 10 , and their mixtures with molar ratios of 2:1, 1:1, 1:2, and 1:4.All mixtures exhibit higher Li-ion conductivity than either of the two individual components, resulting from the lowering of the phase transition temperature upon mixing as witnessed by differential scanning calorimetry (DSC) shown in Figure S1, analogous to the case of mixedanion Na hydroborates. 38,45,46For ratios of 1:1, 1:2, and 1:4, the phase transition is no longer apparent in the DSC traces, even if the resulting phases are not single phase at room temperature as evidenced by X-ray diffraction (XRD) patterns in Figure S2, which agrees well with a previous study. 39The XRD patterns shown in Figure S2 indicate that the 2:1 mixture is isostructural with LiCB 11 H 12 , while the other mixtures contain multiple phases dominated by the LiCB 9 H 10 phase.This suggests that the larger [CB 11 H 12 ] − -anion-derived facecentered cubic lattice can accommodate the smaller [CB 9 H 10 ] − anions, while the reverse is not possible.
Notably, while the ionic conductivity of LiCB 11 H 12 and LiCB 9 H 10 at 25 °C is only ∼0.1 mS cm −1 , that of their mixtures at 25 °C exceeds 1.5 mS cm −1 for every mixing ratio, which for some ratios is higher than the previously reported values. 39his can be attributed to the difference in ball-milling conditions because a high-energy ball mill was employed in this study, while a planetary ball mill was employed in the previous study.At 60 °C, the 2:1 LiCB 11 H 12 −LiCB 9 H 10 mixture exhibits the highest conductivity of 29.4 mS cm −1 among the mixing ratios studied, thanks to the lowered phasetransition temperature of LiCB 11 H 12 by mixing with LiCB 9 H 10 .These conductivity data suggest that all mixtures can function as highly conductive Li-ion solid electrolytes over a wide temperature range, from room temperature to elevated temperatures.
Subsequently, the electrochemical oxidative stability is examined by voltammetric methods to identify solid electrolytes that are compatible with high-voltage cathodes.This method has been shown to yield experimental stability values in close agreement with thermodynamic stability values obtained from first-principles calculations. 47Figure 2b shows the cyclic voltammograms of LiCB 11 H 12 , LiCB 9 H 10 , and their mixtures between 2.0 and 6.0 V vs Li + /Li at a scan rate of 10 μV s −1 at 60 °C.All samples show a tiny initial oxidation peak at ∼3.0 V vs Li + /Li, which is commonly observed in other closo-hydroborates and is not related to subsequent oxidation. 32,47When we compare the onset potential of the main electrochemical oxidation of the closo-carbaborate anions, [CB 11 H 12 ] − shows the highest value of ∼4.0 V vs Li + /Li, while [CB 9 H 10 ] − shows the lowest value of 3.55 V vs Li + /Li.The oxidation peaks in the mixtures consist of a linear combination of peaks derived from [CB 11 H 12 ] − and [CB 9 H 10 ] − anions, with peak intensities roughly depending on the mixing ratio.Our results thus suggest that the apparent electrochemical oxidative stability of the hydroborate anion mixtures can be increased kinetically by employing LiCB 11 H 12rich electrolyte compositions, i.e. by enriching the electrolyte with the thermodynamically more stable [CB 11 H 12 ] − anions. 29 particular, the 2:1 mixture contains a non-negligible amount of [CB 9 H 10 ] − but succeeds in suppressing the oxidation peak derived from [CB 9 H 10 ] − in Figure 2b.We attribute this to the fact that the major component in the mixture is the thermodynamically more stable [CB 11 H 12 ] − and the thermodynamically less stable [CB 9 H 10 ] − is incorporated into the face-centered cubic LiCB 11 H 12 structure (see Figure S2).This is significant because the former studies on the LiCB 11 H 12 − LiCB 9 H 10 mixtures focused on a LiCB 9 H 10 -rich phase, which limits the oxidative stability to ∼3.6 V vs Li + /Li for the 1:2 mixture (Figure 2b) and hence limits the selection of cathode active materials (e.g., to S or TiS 2 ). 31,39In contrast, the higher oxidative stability of the 2:1 mixture up to ∼3.9 V vs Li + /Li opens up the possibility to use higher-voltage cathodes.This mixing ratio possesses a good balance between ionic conductivity and electrochemical oxidative stability and will be referred to as Li 3 (CB 11 H 12 ) 2 (CB 9 H 10 ).

■ INTEGRATION INTO ALL-SOLID-STATE BATTERIES
The Li 3 (CB 11 H 12 ) 2 (CB 9 H 10 ) solid electrolyte is combined with an in-house developed, high-voltage NMC811 cathode 44,48 and three different anodes, namely Li metal, InLi, and graphite.The cathode active material is bulk-doped with 0.3 mol % Ti and surface-coated with 0.6 mol % TiO 2 , as described elsewhere. 44A schematic of the NMC811-graphite cell cross-section based on SEM images is displayed in Figure 1.The NMC811 composite cathode with an areal capacity of 0.34 mAh cm −2 delivers a high reversible discharge capacity at C/10 of ∼145 mAh g −1 and ∼175 mAh g −1 at room temperature and 60 °C, respectively, when cycled to an upper cutoff voltage of 4.2 V vs a Li metal anode (Figure S3).However, even at 60 °C, these cells short-circuit within the first 20 cycles due to penetration of Li metal through the solid electrolyte (Figure S3c).
Short circuits are prevented with an InLi alloy anode.NMC811-InLi cell cycling results are shown in Figures 3a and  3b, delivering an initial discharge capacity at C/10 of 145 mAh g −1 and 175 mAh g −1 at room temperature and 60 °C, respectively.At 2 C, the cell retains more than 100 mAh g −1 and it can be charged and discharged at 10 C, equivalent to 10 mA cm −2 and 1700 mA g −1 , without affecting the capacity at lower rates (i.e., when returning to C/10).In Figures 3c and  3d, we demonstrate remarkable cycling stability with 98% and 90% capacity retention after 100 cycles at C/5 at room temperature and 60 °C, respectively.The Coulombic efficiency is 75% and 87% for the first cycle, respectively, and >99% for all subsequent cycles, indicating that the interfaces are stabilized after the first cycle. 49A higher cutoff voltage of 4.3 V instead of 4.2 V vs Li + /Li slightly increases the initial capacity at the cost of faster fading (Figure S4), which is why we choose an upper cutoff voltage of 3.6 V vs In/InLi (4.222 V vs Li + /Li) to maintain long-term stability.Even after 1000 cycles at C/2, 70% of the initial discharge capacity is still available (Figures 3e and 3f), and 54% after 2000 cycles (Figure S13).
While the rate capability and cycling stability with the InLi anode are promising, the resulting cell voltage is relatively low (0.622 V lower than with a Li metal anode).In addition, In is not sufficiently abundant and InLi has only about half the specific capacity of graphite (372 mAh g −1 ), considering the limit of 45 at% Li in the InLi alloy to stay on the 0.622 V plateau vs Li + /Li. 50,51In addition, the InLi anode provides a nearly infinite reservoir of Li that can compensate for the losses of active Li in the cell.Therefore, we investigate full cells with a graphite anode, which is commercially relevant, enables a cell voltage only 0.1 V lower than with Li metal anode, 50 and does not provide any excess Li, imposing an even more stringent condition on the capacity retention.
The corresponding cells with the same amount of NMC811 cathode active material possess a discharge capacity of 125 mAh g −1 and 150 mAh g −1 at room temperature and 60 °C, when cycled to an upper cutoff voltage of 4.1 V vs a graphite anode (∼4.2 V vs Li + /Li) (Figure 4a and Figure S5).Importantly, we utilize almost all of the theoretical capacity of graphite (Figure S6), and no more than 5−10% additional anode capacity is needed to balance the cell (Figure S7).Compared to cells with the InLi anode, the discharge capacity is about 20−25 mAh g −1 lower.The difference is independent of the C rate and temperature (Figure S8).Moreover, the cell impedance (Figure S9), the dis-/charge overpotentials (Figure

S10
) and the extra capacity obtained by a constant-voltage step (Figure S11) are very similar between cells with InLi and graphite anodes.For these reasons, we rule out a larger cell resistance or slower kinetics in graphite cells as the cause of the lower discharge capacity.Rather, we consider a lower amount of reversibly active Li for the graphite cells, which can be attributed to the following two phenomena.First, not all of the Li intercalated into the graphite during the first charge can be subsequently deintercalated (Figure S6), which accounts for up to 15 mAh g −1 .Second, if an interphase forms between the solid electrolyte and the NMC811 particles upon charge, the Li consumed in this process cannot be replaced during discharge in the case of a graphite anode, because there is no excess Li available as in the case of the Li and InLi anode.
The high capacity retention of the cells with the graphite anode (up to 97% after 100 cycles, as shown in Figures 4c and  4d, and 75% after 350 cycles, as shown in Figures 4e and 4f), in which no excess Li is available, is clear evidence that Li is not continuously consumed, but instead stable passivating interfaces are formed.Based on the high capacity retention without excess Li, stable dis-/charge cycling up to 4.2 V vs Li + / Li is possible, which is beyond the stability of the Li 3 (CB 11 H 12 ) 2 (CB 9 H 10 ) electrolyte (∼3.9 V vs Li + /Li, Figure 2b).Factors contributing to this stable dis-/charge cycling could be (i) the formation of hydroborate anion dimers with higher stability at the interface 32 and/or (ii) the TiO 2 coating on the NMC811 surface.More research is needed in the future to clarify the interface passivation.Overall, we can conclude that the increased stability of the cathode−solid electrolyte interface results in excellent capacity retention, even without any excess Li.
Figure 5 compares our solid-state cells to cells reported in the literature, which have demonstrated a capacity retention of at least 80% after 100 cycles.Compared to all the previous hydroborate-based solid-state Li batteries (circles), our cells (marked with arrows) exhibit by far the best combination between high cell voltage, high cathode-specific capacity, and cycling stability.More than 80% capacity retention after 100 cycles for Li hydroborate solid-state batteries has previously only been achieved with low-voltage TiS 2 cathodes. 30,39,42ore importantly, the hydroborate-based battery reported in this work is among the leading solid-state Li cells in terms of specific energy per cathode composite weight.For example, the 460 Wh kg − 1 obtained in this work with the Li 3 (CB 11 H 12 ) 2 (CB 9 H 10 ) electrolyte compares well with the 466 Wh kg −1 achieved with a Li 6 PS 5 Cl electrolyte. 52Finally, if we consider full cells without excess Li (filled symbols), we know of only one other report, using an Ag−C composite anode, surpassing our cell. 53For practical applications, the energy density at the cell level is decisive.The crystallographic density of closo-hydroborates as reported in the ICSD structure database (Li 2 B 12 H 12 , 1.18 g cm −3 ; Li 2 B 10 H 10 , 1.03 g cm −3 ; NaCB 9 H 10 , 1.14 g cm −3 ) 34−36 is significantly lower than the density of the commonly used oxide (Li 7 La 3 Zr 2 O 12 : 5.12 g cm −3 ), 54 halide (Li 3 InCl 6 : 2.71 g cm −3 ), 55 and sulfide electrolytes (Li 6 PS 5 Cl: 1.86 g cm −3 ). 56Therefore, we expect a competitive gravimetric energy density at the cell level for an optimized hydroborate-based cell architecture.
To conclude, we demonstrate stable cycling of hydroboratebased solid-state batteries with a state-of-the-art 4 V-class NMC811 cathode.The interface stability between the hydroborate electrolyte and the NMC811 cathode is kinetically improved by using a LiCB 11 H 12 -rich and LiCB 9 H 10 -poor electrolyte composition, which maintains a high Li-ion conductivity.While Li metal penetration remains an issue to be addressed in future studies, cells with InLi and graphite anodes feature a remarkable capacity retention of 98% and 97% after 100 cycles at room temperature, confirming the excellent properties of the hydroborate electrolyte in combination with the NMC811 during dis-/charge cycling.With an energy density per cathode composite weight of 460 Wh kg −1 , our cells are on par with the best solid-state batteries reported to date.
In the future, the cell should be developed toward an industrial prototype by reducing the separator thickness from 900 μm to 20−25 μm, increasing the areal capacity from 1 mAh cm −2 to 4−5 mAh cm −2 , improving the rate capability at room temperature, defining strategies to suppress Li metal penetration from the anode to the cathode, and reducing the costs associated with hydroborate anion synthesis.Only solid-state cells with a capacity retention of at least 80% after 100 cycles are considered.The shape of the symbol indicates the type of solid electrolyte: hydroborates (circles), 30,39,42 polymer and polymer composites (triangles), 57−61 sulfides (squares), 52,53,62−64 halides and halide-sulfide bilayer (pentagons), 65−71 and oxides (diamonds). 72,73Empty symbols indicate cells containing excess Li (oversized Li metal anodes or InLi alloy anodes).Full symbols indicate cells without excess Li (graphite and Ag−C anodes).For comparison, the approximate performance of a state-of-the-art liquid electrolyte Li-ion battery (LIB) is indicated (star), calculated based on NMC with 170−200 mAh g −1 and accounting for the weight of the liquid electrolyte. 74The symbols highlighted with arrows correspond to the values obtained in this work.The values obtained from the literature are summarized in Tables S1−S5.
Experimental section (material preparation and characterization, cell assembly and electrochemical characterization) as well as supplemental figures and tables showing the thermal and structural characterization of the solid electrolyte, cycling stability of NMC811-Li metal cells, the capacity at higher cutoff voltages, the capacity of the graphite used, a comparison of the discharge capacities of the individual cells with InLi anode and graphite anode, Nyquist plots of potentiostatic electrochemical impedance spectroscopy, time evolution of cell voltage vs Li

Figure 1 .
Figure 1.SEM images of the cell cross-section with schematically added current collectors: aluminum current collector, cathode composite with NMC811 and solid electrolyte, solid electrolyte separator, anode composite with graphite and solid electrolyte, and copper current collector (top to bottom).Enlarged views show SEM images of NMC811 particles and vapor-grown carbon fibers (top) and graphite flakes (bottom) embedded in Li 3 (CB 11 H 12 ) 2 (CB 9 H 10 ) solid electrolyte (false-colored green).Current collectors are not drawn to scale.

Figure 3 .
Figure 3. Solid-state cell performance at room temperature (blue) and at 60 °C (red) with Li 3 (CB 11 H 12 ) 2 (CB 9 H 10 ) solid electrolyte, NMC811 composite cathode, and InLi anode, in terms of (a, b) rate capability, (c, d) cycling stability at C/5, and (e, f) cycling stability at C/ 2. The discharge capacity is normalized to the weight of the cathode active material in the lithiated state.

Figure 5 .
Figure5.Performance of stable state-of-the-art solid-state Li batteries reported in the literature, in terms of average discharge cell voltage, specific discharge capacity, and specific energy, normalized by the cathode composite weight in the lithiated state.Only solid-state cells with a capacity retention of at least 80% after 100 cycles are considered.The shape of the symbol indicates the type of solid electrolyte: hydroborates (circles),30,39,42 polymer and polymer composites (triangles), 57−61 sulfides (squares),52,53,62−64  halides and halide-sulfide bilayer (pentagons),65−71  and oxides (diamonds).72,73Empty symbols indicate cells containing excess Li (oversized Li metal anodes or InLi alloy anodes).Full symbols indicate cells without excess Li (graphite and Ag−C anodes).For comparison, the approximate performance of a state-of-the-art liquid electrolyte Li-ion battery (LIB) is indicated (star), calculated based on NMC with 170−200 mAh g −1 and accounting for the weight of the liquid electrolyte.74The symbols highlighted with arrows correspond to the values obtained in this work.The values obtained from the literature are summarized in Tables S1−S5.