Mechanochemically Prepared Li2S–P2S5–LiBH4 Solid Electrolytes with an Argyrodite Structure

Solid electrolytes with compositions of (100 – x)(0.75Li2S·0.25P2S5)·xLiBH4 (mol %, 0 ≤ x ≤ 100) were mechanochemically prepared from the 75Li2S·25P2S5 (mol %) glass and LiBH4 crystal. The samples with x ≥ 43 have crystalline phases and those with x ≤ 33 formed a glassy phase. The crystalline phase was identified as argyrodite Li6PS5(BH4). The x = 50 sample formed a crystalline phase and demonstrated a high lithium-ion conductivity of 1.8 × 10–3 S cm–1 at 25 °C with an activation energy of 16 kJ·mol–1. The argyrodite-type crystal with a BH4– anion that occupies the halide site is a novel and promising solid electrolyte.


INTRODUCTION
All-solid-state lithium-ion batteries have attracted significant attention as next-generation batteries. 1 The performance of the all-solid-state batteries strongly depends on the conductivities of the solid electrolytes. 2 Recently, sulfide-based solid electrolytes have become more widespread because they possess high conductivities (>10 −3 S cm −1 at 25°C) 2−8 and good mechanical properties for the formation of effective interfaces between solids. 9,10 Adding lithium halides to glass electrolytes is well-known to be effective in improving conductivity. For example, LiI in addition to sulfide-based glasses resulted in a high conductivity of 10 −3 S cm −1 at 25°C. 3,11,12 Argyrodite Li 6 PS 5 X (X = Cl, Br, and I) also displayed high conductivities of over 10 −3 S cm −1 at 25°C. 13−16 Lithium borohydride (LiBH 4 ) has been reported to exhibit high lithium-ion conductivity and high stability to reduction at high temperatures. 17,18 LiBH 4 is also attractive as an additive salt in the solid electrolyte. Previous investigations 19−21 have revealed that the addition of LiBH 4 to sulfide-based glassy solid electrolytes by a mechanochemical process increases their conductivities. 20 The glasses were fabricated with compositions ranging from x = 0 to 33 in (100 − x)(0.75Li 2 S·0.25P 2 S 5 )· xLiBH 4 (mol %). It was discovered that crystals precipitated in samples with x ≥ 43. 19,20 A similar crystalline phase was also observed in the LiBH 4 −P 2 S 5 system. 21 In this study, we mechanochemically prepared solid electrolytes with crystalline phases with a composition of (100 − x)(0.75Li 2 S·0.25P 2 S 5 )·xLiBH 4 . We determined that the prepared crystal had an argyrodite phase and that the BH 4 − anion occupied the halide site of the argyrodite structure. The solid electrolytes with the argyrodite-type phase exhibited a high lithium-ion conductivity of 1.8 × 10 −3 S cm −1 at 25°C.
X-ray diffraction (XRD) measurements with Cu Kα radiation (λ = 0.1542 nm) were performed using a diffractometer (Rigaku, Ultima IV). An air-sealing holder was used to prevent exposure of the samples to air. Rietveld refinements were performed using RIETAN-FP. 22 The XRD data were collected in the 2θ range between 10°and 80°at a scan rate of 0.1°· min −1 and a step size of 0.02°. The structure was refined starting from the published XRD data by Rayavarapu et al. 15 Differential scanning calorimetry (DSC) was performed on the obtained samples sealed in an Al pan in a glovebox using a thermal analyzer (Seiko Instruments Inc., DSC6200) at a scanning rate of 10°C min −1 . Raman spectra were obtained using a Raman spectrometer (HORIBA, LabRAM HR-800) with a 325 nm He−Cd laser. To measure electrical conductivities, the samples were first placed under 360 MPa pressure at room temperature to form pellets 10 mm in diameter and 1−1.5 mm in thickness. ac impedance measurements were performed using an impedance analyzer (Solartron, 1260) in the frequency range of 0.1 Hz to 8 MHz. The all-solidstate cells were constructed in the same way as previously reported elsewhere except that newly developed solid electrolytes were used. 9 3. RESULTS AND DISCUSSION Figure 1 shows the XRD patterns of the mechanochemically prepared (100 − x)(0.75Li 2 S·0.25P 2 S 5 )·xLiBH 4 samples. The XRD patterns for x = 0 and 33 display characteristic halo patterns; it has been previously reported that the samples with x ≤ 33 have a glassy state. 20 For the samples with 43 ≤ x ≤ 82, unknown diffraction peaks, which were not attributed to the starting materials of Li 2 S, P 2 S 5 , or LiBH 4 , were observed in the XRD patterns. In the XRD pattern of the x = 82 sample, the peaks attributed to LiBH 4 23 were also observed, suggesting that unreacted LiBH 4 was present in a small quantity in the x = 82 sample.
Gas generations were detected in Li 2 S−P 2 S 5 −LiI samples under 220°C. For example, the weight loss due to the gas generation was detected at 180°C in the x = 50 sample by thermogravimetric analysis conducted before the DSC measurements. Thus, the DSC measurements were conducted below the gas generation temperature. The gas generation would be caused by thermal decomposition of the samples. This suggests that this series of materials can be prepared only by the low-temperature process. An endothermic peak at 113°C in the LiBH 4 curve can be attributable to a phase transition from orthorhombic to hexagonal, 24 which is observed at 113°C . A similar endothermic peak is not observed for the (100 − x)(0.75Li 2 S·0.25P 2 S 5 )·xLiBH 4 samples except for the x = 82 sample, which partially includes the LiBH 4 crystal. The DSC curves for the x = 43, 50, and 54 samples exhibit broad exothermic peaks at approximately 60−160°C. The x = 0 sample has an exothermic peak at 213°C, which has been reported to arise from the crystallization of the glass. 25 Thus, it is considered that the broad exothermic profiles observed in the x = 43, 50, and 54 samples are attributable to the crystallization and/or increase in the crystallinity. Figure 3 shows the Raman spectra of the (100 − x)(0.75Li 2 S·0.25P 2 S 5 )·xLiBH 4 samples and the LiBH 4 crystal between 200 and 3000 cm −1 . The band at 420 cm −1 can be attributed to the PS 4 3− ions 26 and is observed for samples containing LiBH 4 . Samples containing LiBH 4 also display a broad band at approximately 2330 cm −1 . This band can be attributed to BH 4 − ion vibrations. Vibration of the internal BH 4 − ions results in strong bands at 2275 and 2300 cm −1 for the orthorhombic low-temperature phase and a broad band at approximately 2300 cm −1 for the hexagonal high-temperature phase. 27 The bands observed in the spectra of the samples with x ≤ 54 are similar to those observed in the high-temperature phase. The band observed in the spectrum of the samples with x = 82 is similar to that observed in the low-temperature phase. 27 The rotation of the BH 4 − ions probably accelerates in the high-temperature phase, 28,29 where there is delocalization of

ACS Omega
Article the negative charge on the BH 4 − ions. The BH 4 − and Li + ions have weakened electrostatic interactions in the high-temperature LiBH 4 phase and the x ≤ 54 samples prepared in this study.
The temperature dependence of the conductivities of the 50(0.75Li 2 S·0.25P 2 S 5 )·50LiBH 4 solid electrolyte (x = 50) and 75Li 2 S·25P 2 S 5 glass (x = 0) is shown in Figure 4. The LiBH 4 crystal data reported by Matsuo et al. 17 are also shown for comparison. In the temperature range from 20 to 70°C, the conductivities of the samples with x = 50 obey the Arrhenius law and the conduction activation energy is determined to be 16 kJ mol −1 . Although higher conductivities are obtained above 70°C, accurate values are difficult to measure because of the extremely low resistance. The crystalline sample with x = 50 exhibits a higher conductivity (1.8 × 10 −3 S cm −1 ) at 25°C than those of the 75Li 2 S·25P 2 S 5 glass samples (2.7 × 10 −4 S cm −1 ) and the LiBH 4 crystal high-temperature phase (1.0 × 10 −3 S cm −1 at 115°C 17 ).
The influence of composition on the conductivity at room temperature and activation energies for the (100 − x)(0.75Li 2 S· 0.25P 2 S 5 )·xLiBH 4 solid electrolytes is shown in Figure 5. The experimental results of the glassy electrolytes with x = 0, 11, and 33 20 are also shown. In the glass-forming region, the conductivity of the glasses increases and the activation energies decrease with increasing LiBH 4 content. The x = 50 crystalline sample displays higher conductivity and lower activation energy than those of the other compositions, suggesting that the crystalline phase exhibits high lithium-ion conductivity. The conductivity dramatically decreased in the x = 54 and x = 82 samples. The conductivity of the x = 82 sample was less than 10 −6 S cm −1 at 25°C. The reason of the decrease of the conductivity has not been clarified yet. One possibility is that the highly resistive amorphous materials as LiBH 4 were formed at grain boundary of the highly conducting argyrodite crystallites in the sample. The possibility of decreasing the conductivities of crystalline phase with argyrodite phase was also not denied at the present stage.
To characterize the crystal structure, samples with higher crystallinity were prepared by heat treatment at 130°C. Figure  6a shows the XRD patterns of the as-obtained and heat-treated samples with an x = 50 composition. The intensities of the diffraction peaks observed in the spectra of the as-obtained sample increase with the heat treatment, indicating that the exothermic peaks observed in Figure 2 can be attributed to crystallization. The DSC results indicate that the x > 33

ACS Omega
Article samples are primarily composed of crystal and glass components. The pattern of the newly formed crystal phase is similar to that of the argyrodite-type crystal Li 6 PS 5 X (X = Cl, Br, and I), 14,15 suggesting that the obtained crystalline phase is argyrodite-type. The ionic radii of Cl − (0.168 nm) and Br − (0.195 nm) ions are similar to that of BH 4 − ions (0.205 nm). 30 Thus, we presume that the Li 7−x PS 6−x (BH 4 ) x crystal precipitates, where BH 4 − ions occupy the sites of Cl − or Br − ions in the Li 7−x PS 6−x X x crystal. The structural model of the Li 6 PS 5 Cl crystal reported by Rayavarapue et al. 15 was used as a starting point for the Rietveld refinement. Figure 6b shows the Rietveld refinement pattern of the heated x = 50 sample. The refined structural parameters for Li 6 PS 5 (BH 4 ) obtained from the Rietveld refinements are summarized in Table 2. Peak indexing of the XRD data suggests that the new phase exhibits argyrodite structure in the space group F4̅ 3m with a lattice parameter of 1.001(8) nm. The BH 4 − and S 2− ions are disordered over the 4a site (0 0 0) (67% BH 4 , 33% S) and 4d site (3/4 3/4 3/4) (33% BH 4 , 67% S). It was previously reported that disorder in the S 2− /X − (X = Cl, Br, and I) distribution promotes lithium-ion mobility in argyrodite-type crystals. 15 The large I − ions cannot be exchanged for S 2− ions, and the resulting Li 6 PS 5 I is more ordered and exhibits only moderate conductivity. The disorder in the S 2− /BH 4 − distribution indicates that the Li 6 PS 5 (BH 4 ) crystal exhibits high lithium-ion conductivity. The increase in the lattice parameter should also affect the conductivity of the crystal. The argyrodite-type Li 6 PS 5 (BH 4 ) phase refined here is a

ACS Omega
Article novel lithium-ion conductor. In this study, the Rietveld refinements were performed using the XRD data of the low crystallinity sample. The Rietveld refinements for the neutron diffraction data of high crystallinity samples are necessary to determine more accurate atomic coordinates for hydrogen and lithium.
The conductivity of the heat-treated x = 50 sample was 1.9 × 10 −3 S cm −1 at 25°C, which is almost similar to that of asobtained samples. It is clear that the Li 6 PS 5 (BH 4 ) phase is more Li-rich compared with the (100 − x)(0.75Li 2 S·0.25P 2 S 5 )· xLiBH 4 (x = 50) solid electrolytes. This suggests that the glass matrix of the x = 50 sample has a lower Li + ion concentration than its nominal composition. However, the x = 50 solid electrolytes that included the Li 6 PS 5 (BH 4 ) phase and glass matrix exhibited a high lithium-ion conductivity of more than 10 −3 S cm −1 . This suggests that both the glass matrix and Li 6 PS 5 (BH 4 ) phase display high lithium-ion conductivities. It is deemed necessary to synthesize single Li 6 PS 5 (BH 4 ) phase crystals and examine their conductivity in the future.
All-solid-state cells using mechanochemically prepared Li 2 S− P 2 S 5 −LiBH 4 solid electrolytes with an argyrodite structure were assembled in order to show their potentials for practical applications. Figure 7 shows the charge−discharge curves of the all-solid-state cells using (a) Li/TiS 2 , (b) In/LiCoO 2 , and (c) In/LiFePO 4 and the x = 50 sample. All of the all-solid-state cells using Li 2 S−P 2 S 5 −LiBH 4 was successfully charged and discharged with the large capacities comparable to the previously reported all-solid-state cells. 2,31 There are few reports of the all-solid-state cells with a LiFePO 4 positive electrode material. The all-solid-state cell using Li metal was also charged and discharged. The Li 2 S−P 2 S 5 −LiBH 4 solid electrolytes have high potential for practical use.

CONCLUSIONS
In this study, (100 − x)(0.75Li 2 S·0.25P 2 S 5 )·xLiBH 4 solid electrolytes were prepared by a mechanochemical process. Glassy samples were obtained in the composition range for x (mol %) from 0 to 33, and samples with argyrodite crystalline phase were obtained at x > 33. The DSC curves for the samples with LiBH 4 did not exhibit the peak attributed to the phase transition of LiBH 4 . The Raman spectra of the (100 − x)(0.75Li 2 S·0.25P 2 S 5 )·xLiBH 4 samples displayed bands attributed to both BH 4 − and PS 4 3− ions. In the glass-forming region, the conductivity of the glass increased with increasing LiBH 4 content. The x = 50 samples that included newly formed crystal phases also exhibited a high conductivity of 1.8 × 10 −3 S cm −1 at 25°C and a low conduction activation energy of 16 kJ mol −1 . Through Rietveld refinements of the x = 50 sample, the novel crystal phase was identified as an argyrodite-type crystal, Li 6 PS 5 (BH 4 ). The disorder of the S 2− /BH 4 − distribution indicates that the argyrodite-type crystal Li 6 PS 5 (BH 4 ) exhibits high lithium-ion conductivity. wrote the paper. All authors analyzed the data and commented on the manuscript.

Notes
The authors declare no competing financial interest.