Room-Temperature Solid-State Lithium-Ion Battery Using a LiBH4-MgO Composite Electrolyte.

LiBH4 has been widely studied as a solid-state electrolyte in Li-ion batteries working at 120 °C due to the low ionic conductivity at room temperature. In this work, by mixing with MgO, the Li-ion conductivity of LiBH4 has been improved. The optimum composition of the mixture is 53 v/v % of MgO, showing a Li-ion conductivity of 2.86 × 10-4 S cm-1 at 20 °C. The formation of the composite does not affect the electrochemical stability window, which is similar to that of pure LiBH4 (about 2.2 V vs Li+/Li). The mixture has been incorporated as the electrolyte in a TiS2/Li all-solid-state Li-ion battery. A test at room temperature showed that only five cycles already resulted in cell failure. On the other hand, it was possible to form a stable solid electrolyte interphase by applying several charge/discharge cycles at 60 °C. Afterward, the battery worked at room temperature for up to 30 cycles with a capacity retention of about 80%.

Introduction

Li-ion batteries (LIBs) are widely used in portable devices and play a major role in the fast-growing electro-mobility market. Solid-state electrolytes (SSEs) are promising candidates for resolving the intrinsic limitations of the organic liquid electrolyte currently employed in LIBs, such as the low cation transference number, the incompatibility (due to the uneven Li plating resulting in shortcuts) and reactivity with lithium metal anodes, and flammability.[1−3] Such drawbacks limit the cell energy density and require major safety precautions. SSEs can overcome these hindrances or bottleneck limitations, thanks to their intrinsic stiffness, which makes them less prone to dendrite penetration.[4] Moreover, superior chemical stability allows the use of metallic lithium as a negative electrode.[5,6] The improved safety naturally comes from the solid nature of the electrolyte. An SSE must fulfill several requirements to be employed in an all-solid-state battery (SSB), such as a Li-ion conductivity higher than 10–3 S cm–1 at room temperature (RT), a negligible electronic conductivity, and a wide electrochemical stability window.[3]

Different classes of materials have been proposed as SSEs, among which are complex hydrides.[1,7] Lithium borohydride (LiBH4) has been extensively studied as an SSE, thanks to a remarkable ionic conductivity (σ) above 120 °C, combined with a low density (0.666 g/cm3). In fact, the LiBH4 RT-polymorph has an orthorhombic unit cell, space group (s.g.) Pnma, showing low Li-ion conductivity (10–8 S cm–1), while around 110 °C,[8] it shows a polymorphic orthorhombic-to-hexagonal (s.g. P63mc) transition, rising the ionic conductivity of several orders of magnitude (∼10–3 S cm–1 at 120 °C).[9] Recently, LiBH4 has been reported to be electrochemically stable up to about 2 V versus Li+/Li,[10,11] reducing the value of 5 V versus Li+/Li previously overestimated.[12]

Halide substitution was adopted to enhance the RT Li+ conductivity in LiBH4. A solid solution is formed by substituting BH4– with I–, Br–, and Cl–, stabilizing the hexagonal polymorph at RT.[10,13,14] Mixing other complex anions with LiBH4 (e.g., NH2– and NH2–) leads to the formation of compounds with different structures, with enhanced ionic conductivity at RT.[15,16] Recently, it was reported that partial dehydrogenation of LiBH4 leads to different Li–B–H complexes with significantly higher Li-ion conductivity (∼2.7 × 10–4 S cm–1 at 35 °C).[17]

An alternative method to improve the Li-ion conductivity of the orthorhombic LiBH4 (o-LiBH4) is by nanoconfinement in suitable scaffolds or mixing it with oxides, forming oxide-based composites.[18−22] In this case, the improved Li-ion conductivity relies on the formation of a conductive interface, described by a core–shell model.[20,23] The fraction of LiBH4 (the core) in direct contact with the oxide (the shell) forms an interfacial layer, featuring a Li-ion conductivity enhancement. The presence of different dynamics, due to the occurrence of slow and fast diffusivity of Li ions, has been detected for binary composites by 7Li solid-state NMR spectroscopy.[22,24,25] The effective thickness of such an interfacial layer has been calculated by Suwarno et al.,[21] amounting to 1.94 ± 0.13 nm for a SiO2 nanoscaffold. In the LiBH4–SiO2 system,[26] Li conductivity is optimized when LiBH4 completely fills the silica pores with a thickness of the interfacial layer of about 2 nm, which is in good agreement with the results obtained by Suwarno.[21] Recently, Gulino et al.(26) reported two composite systems (LiBH4–ZrO2 and LiBH4–MgO) that have shown improved RT conductivity (∼10–4 S cm–1) compared to o-LiBH4. Liu et al.(27) reported that two-dimensional MoS2 and LiBH4 composites yielded an RT ionic conductivity of (∼10–4 S cm–1).

LiBH4 was studied as an SSE in several SSBs[28] using TiS2[29] or sulfur[30] as cathode materials. These SSBs operate at a temperature of about 120 °C, allowing a sufficiently high Li-ion conductivity due to the presence of the hexagonal polymorph of LiBH4. The working potential of both TiS2 and S is about 2 V versus Li+/Li,[29,30] close to the electrochemical stability window of LiBH4, which is one of the factors explaining the progressive capacity fading. In order to decrease the SSB-operating temperature, Unemoto et al.(31) used a LiBH4–P2S5 mixture as the SSE, while Das et al.(32) reported an SSB working at 55 °C using LiBH4 nanoconfined in silica (MCM-41) as the SSE.

Optimization of the composition of LiBH4-based composites as an SSE can reduce the working temperature of the LiBH4-based SSBs. Therefore, the aim of this work is to design a LiBH4-based system as an improved SSE for RT SSBs. The LiBH4–MgO system was selected, and the effect of the composition on the Li-ion conductivity was first established. The electrochemical stability window, measured by cyclic voltammetry (CV), was then determined in order to investigate the effect of the oxide matrix on the electrochemical stability of LiBH4. A solid-state cell configuration, that is, TiS2|SSE|Li, was selected as the electrochemical system, focusing on the characterization of the composite LiBH4–MgO as the SSE for SSBs. These composites allowed to decrease the operating temperature of the LiBH4-based SSB down to 60 °C and even down to RT, compared to the 120 °C of pure LiBH4. Several cycles at 60 °C, and probably the concomitant formation of a stable solid electrolyte interphase, allowed us to successfully operate the SSB at RT for more than 30 cycles, with a discharge capacity retention of 80%.

Experimental Section

Synthesis

The Li-ion conductivity in the LiBH4–MgO system has been evaluated for different compositions following the procedure suggested in ref (26). Three samples were synthetized, with a v/v % of MgO corresponding to a fraction of pore filling equal to 1/3, 1, and 3 (CE26, CE53, and CE74, respectively) assuming that LiBH4 fills the pores of MgO after ball milling. LiBH4 (purity > 95%, Alfa Aesar) was mixed with MgO (Steam Chemicals) in different ratios (Table ).

Table 1

Composition, Fraction of the Pore Filled, and Thickness of LiBH4 of Investigated Samples

sample name	oxide fraction (wt %)	oxide fraction (v/v %)	fraction of pore filleda (%)	thickness of LiBH4b (nm)
26CE	65.0	26	323	3.8
53CE	85.7	53	100	1.2
74CE	94.0	74	38	0.4

Data obtained from the ratio between the LiBH4 volume per gram of MgO and the pore volume (Vp).

Thickness of the LiBH4 layer covering the surface of the oxide (BET), assuming a uniform layer of LiBH4 on the oxide surface.

In order to remove the physisorbed/chemisorbed water, MgO pellets were dried in a furnace for 6 h, under dynamic vacuum (by rotary pump), at 300 °C. Before the mechanochemical treatment for the preparation of the different compositions, the as-received LiBH4 was ball-milled for 2 h at 500 rpm in a Fritsch Pulverisette 7 planetary mill and was used as the starting material for the LiBH4–oxide composite. All samples were ball-milled for three periods of 10 min at 300 rpm, separated by 1 min breaks, in 80 mL stainless-steel vials, with stainless-steel spheres (10 mm diameter). The ball-to-sample mass ratio used was equal to 30:1. The mechanochemical treatment was performed under an argon atmosphere for all samples. Due to the air sensitivity of the samples, they have been manipulated in a glovebox (MBraun Lab Star Glove Box) filled with argon, with residual impurities (<1 ppm O2 and <1 ppm H2O).

Characterization

Structural Characterization

The density of MgO has been taken as 3.58 g/cm3 from the literature.[33] The surface properties of MgO were analyzed by N2 adsorption at 77 K in a TriStar Plus II gas-volumetric apparatus (Micromeritics, Norcross, GA, USA). The specific surface area (SBET) was calculated by fitting the experimental data points with a Brunauer–Emmett–Teller isotherm[34] and was 215 m2/g. The pore volume (Vp) was derived from the volume of absorbed nitrogen at p/p0 = 0.95 and was 0.25 cm3/g. Powder X-ray diffraction (PXRD) analysis has been performed on the as-prepared composites, see the Supporting Information.

Electrochemical Impedance Spectroscopy

Li-ion conductivity data for the samples were obtained by collecting the electrochemical impedance spectroscopy (EIS) spectra following the procedure reported in ref (10). The EIS measurements were performed using an HP4192A LF impedance analyzer and a Novocontrol (BDS 1200) sample cell in the temperature range of 20 < T < 130 °C (every 10 °C). By using an axial hydraulic press (60 MPa), the mixtures were pelletized, with a diameter of 10 mm and a thickness of about 0.2–0.6 mm. Impedance data were analyzed via the EqC software[35] following the data validation described in ref (36). All fits performed resulted in a χ-squared test (χ2 test) < 10–3.

Cyclic Voltammetry

CV was used to analyze the oxidative limit of the electrochemical stability window. The desired LiBH4–MgO composite was mixed with carbon black (CB, Ketjenblack EC600JD, Akzo Nobel Chemicals) in a weight ratio of 95:5 using an agate mortar.[11] A two-layered pellet was obtained by pressing the 8 mg CE-composite mixture and about 25 mg of the composite at 240 MPa using a uniaxial hydraulic press (diameter 6 mm). The pellet thus prepared was tested in a two-electrode 3/4″ PTFE Swagelok-type cell, with a lithium disk (99.9%, Sigma-Aldrich) as the counter and reference electrode (SSE side) and a stainless-steel disk as the working electrode (SSE + CB side). The cells were tested with a potentiostat/galvanostat Biologic MPG-2 after a 4 h rest at 60 °C. CV measurements have been performed into a 1.3 < V < 5 V versus Li+/Li voltage region at a scanning rate of 20 μV s–1.

Battery Assembly and the Electrochemical Test

TiS2 (99.9%, Sigma-Aldrich) was selected as the active material for the cathode. The TiS2 and LiBH4 powders were mixed in an agate mortar using a 1:1 weight ratio. The resulting mixture was used as the positive electrode. A two-layered pellet was prepared with 2 mg of the positive electrode mixture and 25 mg of the SSE by cold pressing at 240 MPa using a uniaxial hydraulic press (diameter 6 mm). A pure lithium disk was used as the negative electrode. The assembled bulk-type TiS2/SSE/Li SSB was placed in a 3/4″ PTFE Swagelok-type cell. The galvanostatic cycling was performed in the voltage range of 1.7–2.5 V at 60 °C and at RT. EIS measurements were also performed in the symmetric configuration Li/SSE/Li and on the TiS2/SSE/Li SSB during the rest time, before the battery cycling, and after each discharge and charge.

Results and Discussion

Tailoring of the LiBH4–MgO Composition

The AC conductivity of the composites as a function of the inverse temperature is shown in Figure a. The 20 °C impedance spectra (composed by a single arc and a low-frequency linear dispersion) are plotted together in the Nyquist plot in Figure S1, and the EIS-fitted values are reported in Table S1.

Figure 1

(a) Li-ion conductivity, obtained from a temperature-dependent EIS cycle during the second heating, for LiBH4–MgO composites with different oxide fractions. The gray line corresponds to the Li-ion conductivity of pure LiBH4.[10] (b) Li-ion conductivity at 20, 60, and 120 °C as a function of the pore filling. Dashed lines are a guide for the eyes.

Figure a shows that at RT, sample CE53 shows the highest Li-ion conductivity (2.86 × 10–4 S cm–1 at 20 °C), about 4 orders of magnitude higher than that of pure LiBH4. The 20 °C Li-ion conductivities of CE26 and CE74 were 1.07 × 10–4 and 5.94 × 10–6 S cm–1, respectively. The Li-ion conductivity of sample CE26 at 40 °C (2.57 × 10–4 S cm–1) is in agreement with the data already reported for the same composition and at the same temperature (1.80 × 10–4 S cm–1).[26] The improved Li-ion conductivity likely relies on the formation of a conductive interface, described by a core–shell model.[20] It was demonstrated, by solid-state NMR, that the interface layer between LiBH4 and silica is characterized by a high ion dynamics, for both BH4– and Li+, and cannot be defined with a clear crystal structure.[37] The exact relation between the interface layer structure and the dynamics of ions needs further investigation.

The dependence on the pore filling of the Li conductivity at various temperatures is shown in Figure b, assuming that LiBH4 completely fills the pore of the oxide during the mechanochemical treatment. The maximum σ value is observed for a 100% pore volume filling (Figure b), confirming the trend previously reported by Gulino et al.(26)

In order to explore the phase composition of the different composite, after synthesis, the PXRD patterns were collected (Figure S2). Orthorhombic LiBH4 was detected only for sample CE26, that is, for a pore filling higher than 100%, indicating that the excess of hydride contained in this composite is partially present as the RT polymorph.

From the trend of data reported in Figure a, it is worth noting that different conductive regimes are likely present, indicating a complex temperature-dependent Li-ion conduction mechanism in the investigated temperature range. This type of behavior has been previously reported for complex hydrides,[38] as well as for different classes of materials studied as SSEs,[39] and can be assigned to different ion–ion interaction regimes. Further investigation, that is, combining solid-state NMR and large frequency and temperature-range EIS measurements,[38,40] is needed in order to clarify this aspect.

The activation energy for the CE53 sample was obtained by fitting linearly (R2 > 0.999) the ln(σT) versus 1/T data shown in Figure a below 60 °C, where data suggest an Arrhenius-type temperature dependence. The so-obtained Ea is equal to 0.29 ± 0.03 eV below 60 °C. The obtained Ea is considerably lower than the average value reported in the literature for pure LiBH4 (0.75 ± 0.07 eV),[41] but it is similar to values observed for other SSEs.[42]

For sample CE74, the pore-filling fraction is lower than 100%; therefore, the highly conductive phase does not percolate throughout the sample and the conductive pathway is interrupted by the oxide. A similar effect occurs in sample CE26, where the excess of the low conductive orthorhombic LiBH4 interrupts the conductive pathway.

For the CE53 sample, the Li-ion conductivity is about 1 order of magnitude higher than the values previously reported for LiBH4–SiO2-based composite SSEs: at 40 °C Blanchard et al. obtained 1.0 × 10–5 S cm–1 for 28 v/v % of MCM-41,[20] Choi et al. obtained 1.5 × 10–5 S cm–1 for 55 v/v % of fumed silica,[19] and Gulino et al. obtained 4.1 × 10–5 S cm–1 for 20 v/v % of SiO2.[26]

The Li-ion conductivity of CE74 and CE53 composites does not show, at 110 °C, the typical step due to the phase transition of LiBH4, which is slightly visible for the CE26 sample (see also Figure S3). This suggests that the contribution of the hexagonal phase of LiBH4 is negligible in the CE74 and CE53 samples at high temperatures, whereas it is relevant for the CE26 sample.[26] This behavior can be explained by assuming a homogeneous distribution of the LiBH4 layer on MgO, as reported in our previous work,[26] providing estimated values for the interface layer thickness (see Table ), which are similar to that obtained by Suwarno et al.(21)

The highly conductive layer of LiBH4 in direct contact with the oxide, whose thickness has been estimated to be about 2 nm, does not undergo a structural phase transition.[21,26] Indeed, for samples CE74 and CE53, the calculated thicknesses of LiBH4 are 0.4 and 1.2 nm, respectively (see Table ), which are close to the values estimated in the literature.[21,26] On the other hand, the calculated thickness of LiBH4 for composite CE26 amounts to 3.8 nm, explaining the slight increase of the Li-ion conductivity above 110 °C (Figure a) and confirming that the bulk LiBH4 contributes to the Li-ion conductivity after the phase transition.

Electrochemical Stability

Next to a high Li-ion conductivity, a suitable SSE should have a wide electrochemical stability window and a good chemical compatibility with electrodes. The electrochemical stability of the CE53 sample has been evaluated by CV at 60 °C (Figure a). The interface between a solid electrolyte pellet and a flat metallic Au working electrode can result in a low contact surface area and thus a high interface resistance, which can cause difficulty in the signal detection, for example, overestimating the electrochemical stability window. Therefore, in order to increase the probed surface, a carbon material was added, as described elsewhere.[43,44]

Figure 2

(a) Linear sweep voltammograms of Li|CE53|CE53-C|stainless-steel cells at a scan rate of 20 μV s–1 from 1.3 to 4.0 V vs Li+/Li at 60 °C. (b) Eonset estimation from two linear regression lines of the nonfaradaic background current and faradaic anodic current.

The oxidative limit (Eonset) was determined from the intersection of two linear regression lines (R2 > 0.99) of the background current and the faradaic oxidative current at a positive potential versus Li+/Li (Figure b) following the approach suggested by Asakura et al.(11) It falls at about 2.3 V versus Li+/Li, which is in agreement with the values reported for LiBH4 by Asakura et al.(11) and Gulino et al.(10) This result suggests that the addition of the MgO matrix only affects the ionic conductivity, leaving the electrochemical stability of LiBH4 unchanged.

The chemical compatibility toward metallic lithium was evaluated with galvanostatic cycling in a Li|CE53|Li symmetrical cell at 60 °C, allowing to also determine the extent of the reversible lithium plating/stripping at the SSE surface and, at the same time, its reductive stability. The results, shown in Figure , demonstrate that lithium is plated and stripped reversibly for over 90 h.

Figure 3

Galvanostatic cycling profiles of the symmetrical Li|CE53|Li cell at 60 °C with a current density of 25 μA cm–2 for 30 min sweeps.

The cell polarization is rather steady at 20 mV for the whole period, indicating a long-term stability and that no parasitic reactions between the SSE and lithium occur in this low potential region. Indeed, a well-performing electrochemical material should show a stable polarization, reflecting the electrolyte resistivity and interfacial effects, in case they are present. LiBH4 has been already reported to be able to plat/strip lithium for a longer time and at higher current densities.[17,45]

The contact resistance of the cell was calculated by multiplying the cell resistance (after having subtracted the SSE contribution), divided by a factor of 2 (since the two interfaces are considered equivalent), by the contact surface. It turns out to be about 565 Ω cm2, a much higher value than that reported by Kim et al.(46) on carborane SSEs (<1 Ω cm2). In the absence of a proper cell stack pressure, which would guarantee an intimate contact between lithium and the SSE surfaces, a high contact resistance, increasing during cycling, is expected,[47,48] explaining the observed value.

Battery Test

The electrochemical properties of the CE53 composite SSE were tested in an SSB, selecting TiS2 and Li as positive and negative electrodes, respectively. Lithium increases the energy density with respect to the commercial graphitic anodes,[4] while TiS2 is widely used for LiBH4-based SSBs.[29] In the current study, TiS2 has also been selected to obtain a valid comparison with the system reported by Unemoto et al.(29)

The Li-ion conductivity of CE53 at 20 °C amounts to 2.86 × 10–4 S cm–1 and it is sufficient to operate the battery at RT at low current regimes.[3] Therefore, a freshly prepared cell was built and cycled at RT, without any conditioning. Figure depicts the galvanostatic cycling with a potential limitation (GCPL) profile of the Li|CE53|TiS2 cell, operating with a current density of 24 mA g–1 (C/10).

Figure 4

Voltage profiles of the Li|CE53|TiS cell for a rate of C/10 (i.e., 24 mA/g) at RT.

Figure shows that it was possible to collect data for only five cycles before the cell failure and several spikes are visible in the charge profiles, but they are completely missing during discharges. These results suggest an inhomogeneous Li plating and it may be induced by the imposition of a current density exceeding the so-called critical current density,[49,50] which is also a function of the cell stack pressure.[48]

A significant difference between the discharge capacity at the first and the second cycle (52 and 173 mA h g–1, respectively) has been observed. Unemoto et al.(29) observed a similar behavior for the Li|LiBH4|TiS2 battery, operating at 120 °C. By probing the evolution of LiBH4 into Li2B12H12, it has been related to a partial instability of the TiS2/LiBH4 interface, forming H2 and additional Li, that self-diffuses into TiS2, self-discharging the battery.[29] The recently reported value of 2.2/2.3 V versus Li+/Li for the electrochemical window of LiBH4 clarifies that the TiS2/LiBH4 interfacial instability arises when the LiBH4 oxidation potential is exceeded.[10,11] In the present case, the self-discharging reaction to form a solid electrolyte interface (SEI), as evidenced by the capacity difference between the first and the second cycle, is much less extended than that reported by Unemoto et al.(29) (i.e., discharge capacities of 80 and 205 mA h g–1 for the first and second cycles, respectively). This is probably due to a kinetic limitation in the reaction of LiBH4 to form likely Li2B12H12 as a consequence of the lower temperature.

In order to gain further insights into the effect of the temperature on the formation of the SEI, a freshly prepared cell was built and cycled at 60 °C. Figure a shows the GCPL profile of the Li|CE53|TiS2 cell operating at 60 °C and with a current density of 11.8 mA g–1 (corresponding to C/20).

Figure 5

(a) Voltage profiles of the Li|CE53|TiS2 cell for a rate of C/20 (11.8 mA g–1) at 60 °C. (b) Discharge/charge specific capacity, Coulombic efficiency (discharge capacity over charge capacity), and discharge capacity retention ratio as a function of cycle number for the same cell. The capacity of the battery is expressed per gram of TiS2.

For the sake of clarity, only selected galvanostatic profiles are shown. A high capacity retention was observed over 65 cycles when the battery operated at 60 °C. A rather low discharge capacity has been observed at the first cycle (101 mA h g–1). On the other hand, its value at the second cycle amounts to 175 mA h g–1 that corresponds to about 73% of the theoretical capacity of TiS2 (239 mA h g–1).[51,52]

The self-discharge due to the formation of the Li2B12H12 SEI is also observed in this case but with a higher extent with respect to the test performed at RT (Figure ). The higher self-discharge observed is consistent with a faster kinetics of the reaction of LiBH4 to form Li2B12H12, which is favored by the higher temperature.

Figure b shows the discharge/charge capacity, Coulombic efficiency, and discharge capacity retention ratio to the second discharge as a function of cycle number. The capacity retention has been calculated with respect to the second discharge capacity since in the first run the capacity is overestimated due to the self-discharge reaction occurring at the TiS2/LiBH4 interface. The discharge capacity retention is more than 80% after 65 cycles, which is promising for a stable battery operation at this temperature.

It is worth noting that after the 30th cycle (Figure a), a pronounced decrease of the capacity is observed (i.e., discharge capacities of 174, 147, and 139 mA h g–1 for the 2nd, the 30th, and the 65th cycle, respectively). On the other hand, the Coulombic efficiency is always close to 99.9% up to the 10th cycle (Figure b), indicating that the capacity fading occurs during the charge, namely, at a voltage higher than the oxidative stability of LiBH4.

To clarify the cause of this effect, the cell impedance was monitored after each charge/discharge cycle and results are reported in Figure S4 in the Supporting Information. The contact resistance (R1), at the very first discharge, is <1 Ω (similar to that obtained by Kim et al.),[46] and afterward, it quickly increases in the first 10 cycles and then reaches a steady exponential increase. This behavior can be understood considering the sum of two contributions. The first one is the formation of the Li2B12H12-based SEI layer, that is, poor conductivity at 60 °C (lower than 10–6 S cm–1).[53] The decrease of the capacity fading at the 30th cycle could correspond to the achievement of the maximal SEI thickness. The second contribution, being the cell not supported by an appropriate stack pressure, is likely due to a continuous loss of contact at the electrode interfaces. In fact, during the delithiation of TiS2, a volume contraction of ΔV/V = −9.7% is experienced, which also explains the rather high contact resistance after each charge cycle (Figure S4 in the Supporting Information), that is, after Li plating at the negative electrode. The LiTiS2 ΔV/V value was estimated assuming a complete deintercalation; so, considering the unit cell volume at the two intercalation extremes, x = 0 and at x = 1. In addition, when considering the negative side of the cell, a similar trend of the contact resistance has been reported by Krauskopf et al.,[54] when the current density is higher than a critical value (i.e., 200 μA cm–1 at RT) without a suitable external applied pressure. Therefore, we assume that the contact degradation happens at both the cathode and anode and that it is the main reason of the capacity fading after the 30th cycle.

In order to investigate a possible dependence of SEI formation from the amount of LiBH4 in the SSE, that is, from the LiBH4/MgO volume ratio, a similar cell was built using the CE26 composite as the electrolyte. Figure S5 in the Supporting Information shows the GCPL profile of Li|CE26|TiS2 at 60 °C, with a current density of 4.8 mA g–1 (corresponding to C/50). Also, in this case, the capacity obtained on the second cycle (176 mA h g–1) is higher compared to that observed for the first one (82 mA h g–1), confirming the occurrence of a self-discharge reaction, as described above. Interestingly, when comparing the properties of cells with CE26 and CE53 as SSEs, the same behavior is observed, that is, a stabilization of the capacity fading after a certain number of cycles. In this case, the change is observed at the 58th cycle, a time-shift that can likely be due to the slower kinetics imposed by lower current rates (C/50 instead of C/20).

The high Li-ion conductivity of the CE53 composite and the formation of an SEI at a high temperature allow the battery to operate at RT. Therefore, after 65 cycles performed at 60 °C, the temperature of the test for the Li|CE53|TiS2 cell decreased to RT. A 4 h rest has been applied in order to equilibrate the temperature. Figure a shows the galvanostatic profiles using the same current density of 11.8 mA g–1 (C/20). For the sake of clarity, only selected galvanostatic profiles are shown.

Figure 6

(a) Voltage profiles of the Li|CE53|TiS2 cell for a rate of C/20 at RT after 65 cycles at 60 °C at the C/20 rate. (b) Discharge/charge specific capacity, Coulombic efficiency (discharge capacity over charge capacity), and discharge capacity retention ratio (to the second discharge capacity) as a function of cycle number for the same cell.

Figure b shows the discharge/charge capacity, Coulombic efficiency, and capacity retention ratio to the second discharge as a function of cycle number. It is worth noting that the battery successfully operates at RT for more than 30 cycles, with a discharge capacity retention of 80%.

The capacity of the Li|CE26|TiS2 cell at RT for the first cycle is 51 mA h g–1 (Figure ), compared with the value of 139 mA h g–1 obtained for the 65th cycle at 60 °C. In order to clarify this difference, Figure S6 in the Supporting Information shows the voltage profiles of the 65th charge and discharge cycles shown in Figure compared with those corresponding to the 1st cycle at RT (Figure ). It is clear that the decrease of the working temperature causes an increase of the cell polarization, indicating that the capacity drop (i.e., between the last cycle at 60 °C and the first cycle at RT) can be assigned to kinetic limitations due to the low temperature. During the 4 h rest, the evolution of the cell resistance was monitored by means of in situ EIS (Figure S7a in the Supporting Information). The increase of the cell resistance during the 4 h rest at RT is mainly related to the decrease of the ionic conductivity of the SSE. Figure S7b in the Supporting Information shows the contact resistance (R1) after each charge discharge cycle at RT. Assuming that the formation of the SEI was completed during the first cycles at 60 °C, the steady increase of R1 can be assigned to just a continuous contact loss, as also observed after the 30th cycle at 60 °C (Figure S4b in the Supporting Information). Despite this, CE53 allowed to operate the Li|CE53|TiS2 system at RT, once the SEI was formed.

Comparing the behavior of the battery previously conditioned at 60 °C with respect to that operating directly at RT (Figure ), it is evident that the high-temperature treatment stabilizes the interface by forming a stable interface compound such as the Li2B12H12-based SEI, as suggested by Unemoto et al.,[29] allowing the cell to operate at RT. The longer life cycle suggests the prevention of electrolyte decomposition.

As mentioned above, Unemoto et al.(29) reported results on an SSB very similar to that investigated in the present study but using pure LiBH4 instead of a composite as the SSE and operating at 120 °C since h-LiBH4 was necessary to achieve the high Li-ion conductivity. The obtained capacity retention was 88% after 300 cycles at a C/5 rate. It has been reported that the higher working temperature increases the Li diffusion,[54,55] strongly limiting the void formation at the electrode interface during the stripping process. This effect limits the contact resistance evolution, significantly reducing the capacity fading, as observed by Unemoto et al.(29) even in the absence of an appropriate stack pressure, unlike the reported case at 60 °C. In contrast, the results reported here show that a drastical decrease of the operating temperature (i.e., from 120 °C to RT) is possible, thanks to a LiBH4–MgO nanocomposite as the solid electrolyte and the formation, at 60 °C, of a stable SEI. It is the first time that the formation of a stable SEI at a temperature (60 °C) higher than the operating one (RT) is applied to a hydride solid electrolyte and it is probably also relevant for other SSEs working in full solid-state LIBs, as already reported by Rodrigues et al.(56)

Clearly, optimization of the battery, to realize a long cycle-life SSB, would lead to a different electrode choice. For instance, elemental sulfur, which has a high theoretical capacity (1672 mA h g–1)[57] and a redox potential of ∼2.2 V, similar to the oxidative limit of the LiBH4 electrochemical window (i.e., 2.2 V vs Li+/Li), would be a suitable electrode, possibly adopting an infiltration procedure for the electrode preparation, that is, dissolving the SSE in an opportune solvent and crystallizing it directly on the cathode material.[58,59] In addition, the optimization of the external cell stack pressure, recently suggested to be rather small (i.e., 5 MPa),[60] would reduce the contact resistance, leading to a lower capacity fading.[47,54]

Finally, the cost of solid electrolyte materials is important, as well as effective large-scale production. The U.S. Department of Energy’s Advanced Research Projects Agency-Energy (ARPA-E) has adopted an ambitious target of 10 $/m2 for the cost area of solid electrolyte materials,[61,62] considering a 10 μm thickness. The SSE synthetized in this work is composed by easily available raw materials making the cost less than 2 $/g, corresponding to about 80 $/m2. Considering that the up-scaling might decrease the cost, LiBH4-based composites could be considered as competitive candidates to be used in Li-ion SSBs.

Conclusions

In this work, fast ionic conductors in the solid state, based on the LiBH4–MgO system, were investigated. The samples were mechanochemically synthetized. The Li-ion conductivity of LiBH4 was improved in all cases, and the samples containing 53 v/v % of MgO showed the best enhancement (2.86 × 10–4 S cm–1 at 20 °C), since the volume fraction of LiBH4 allowed to completely fill the pore volume of MgO. The formation of a highly conductive layer does not affect the electrochemical stability window, which is similar to that of pure LiBH4 (i.e., about 2.2 V vs Li+/Li).

A test at RT in a TiS2/Li SSB allowed only five cycles before the cell failure. From a battery test at 60 °C, the incorporation of the solid electrolyte in the battery showed that a stable SEI is formed during the first charge/discharge cycles, causing an initial increase in contact resistance but limiting a further decomposition of the composite electrolyte. Afterward, the battery worked at RT for up to 30 cycles, with a specific capacity of about 50 mA h g–1.

In conclusion, it has been demonstrated that the SEI, formed at 60 °C, allowed to reduce the operating temperature of the SSB down to RT. Therefore, a possible novel strategy to obtain an SSB working at RT, using complex hydrides as electrolytes, can be established by the formation of a stable SEI at higher temperatures. Despite this proof-of-concept, further optimization is mandatory to obtain an efficient battery (e.g., electrode choice and casting and cell stack pressure).