Degradation mechanism of all‐solid‐state lithium‐ion batteries with argyrodite Li7−xPS6−xClx sulfide through high‐temperature cycling test

Sulfide‐based all‐solid‐state lithium‐ion batteries (LIBs) are promising replacements for conventional liquid electrolyte LIBs. However, their degradation mechanisms and analysis methods are poorly understood. Herein, the degradation mechanism of an argyrodite‐type sulfide‐based all‐solid‐state prototype LIB cell is reported. Furthermore, an analysis method for all‐solid‐state batteries using charge/discharge cycle tests at 100°C followed by the disassembly analysis of cells before and after accelerated degradation tests is reported. Based on the findings of this study, the degradation of the prototype cell is classified as follows: (i) solid electrolyte (SE) oxidation in the positive electrode, which recovers battery capacity and increases resistance; (ii) SE reduction in the negative electrode, which decreases capacity; (iii) lithium deposition on/in the negative electrode, which decreases capacity; and (iv) capacity loss of the positive electrode, which decreases capacity. These degradation reactions appear to occur simultaneously. These findings are expected to aid the development of sulfide‐based solid‐electrolyte LIBs with improved safety and energy densities.


| INTRODUCTION
Globally, research and development related to electric vehicles are increasing because of the ongoing environmental and energy crises. In addition to the energy density and input/output characteristics of storage batteries, their durability and safety are essential factors that affect the application of battery technology in electric vehicles. Lithium-ion batteries (LIBs) have been applied in electric vehicles to take advantage of their excellent performance. Many previous studies have focused on various aspects of the degradation mechanisms of LIBs, such as the degradation of positive and negative electrodes by the deactivation of active materials (AMs) due to changes in the surface structure of particles 1,2 and the electronic disconnection between the AMs and the conductive carbon network or current collector [3][4][5][6][7] ; and solid electrolyte interphase formation due to side reactions such as decomposition of the electrolyte solution 8 or lithium-metal deposition, 9,10 resulting in irreversible lithium-ion formation. Recently, all-solid-state LIBs have received significant attention as replacements for conventional LIBs, which contain liquid electrolytes. Although all-solid-state LIBs can be broadly classified into those with oxide-based solid electrolytes (SEs) and those with sulfide-based SEs, the expectations for sulfide-based SEs are higher because of their good formability and high ionic conductivities. 11 Lithium superionic conductors (thio-LISICON), 12,13 Li 10 -GeP 2 S 12 (LGPS)-type materials, 14 lithium argyrodite, 15,16 and Li 7 P 3 S 11 glass-ceramics 17 are representative sulfidebased SEs and have been reported to exhibit high ionic conductivities. However, the potential windows of sulfide-based SEs are narrow, 18,19 and they are reported to be oxidized or reduced relatively easily. [20][21][22] In particular, in the positive electrode environment (where the AM and sulfide-based SE are directly in contact with each other), the high potential of the AM causes Li + ions to migrate from the SE near the interface, thereby forming a space-charge layer that results in significant interfacial resistance, 23,24 which is considered a significant problem. This phenomenon can be explained by the oxidation of the sulfide-based SE. 21 Several techniques have been proposed as countermeasures, including the introduction of a buffer layer, such as LiNbO 3 , at the interface between the AM and SE, suppressing the oxidation reaction, and reducing the interfacial resistance. 23,25 However, most studies have focused on each electrode reaction, particularly in half-cells, and there are few reports concerning the durability of sulfide-based allsolid-state LIBs (full-cells). A significant difference between a full-cell and a half-cell is whether additional Li + ions are supplied from lithium metal. During degradation of a liquid-type LIB (full-cell), which does not contain lithium metal, the active Li + ions that are originally in the positive electrode are consumed by side reactions, and the state-of-charge (SOC) of the positive electrode irreversibly increases compared with that of the negative electrode (SOC gap). 26,27 However, in the case of a half-cell, the SOC gap does not occur because additional Li + ions are supplied from the lithium metal used as a counter electrode, despite the consumption of the original Li + ions in the working-electrode material by the degradation reactions. Therefore, it is important to evaluate full cells that do not have an additional source of Li + ions. For liquid-type LIBs, analytical methods have been established for studying the degradation behavior, such as Li + ion consumption in the SE interphase and electrode degradation, and many degradation mechanisms have been proposed. [26][27][28] However, to date, there have been few reports on the degradation mechanisms of all-solid-state batteries because the analytical methods applied to conventional LIBs are not always applicable to all-solid-state batteries. For example, it is difficult to divide all-solid-state batteries into electrodes. In particular, there is no established method to evaluate the amount of consumed active Li + ions in all-solid-state batteries and measure the capacities of each electrode.
In this study, we prepared a prototype argyrodite-type sulfide-based all-solid-state LIBs whose electrodes could be divided to analyze each electrode before and after degradation tests. Furthermore, we conducted charge/ discharge cycle tests at a high temperature of 100°C, followed by disassembly analyses using X-ray photoelectron spectroscopy (XPS), scanning electron microscopy (SEM), inductively coupled plasma atomic emission spectrometry (ICP-AES), and electrochemical measurements (half-cell) before and after degradation to study the overall degradation mechanism of the sulfide-based all-solid-state LIB. In particular, the ICP-AES measurements enabled us to determine the consumption of active Li + ions in the negative electrode; for this, we used two types of solvents to extract Li + ions from various side reaction products. We believe that the results of this study will serve as a basis for the qualitative and quantitative evaluation and analysis of the degradation mechanisms of all-solid-state batteries and will contribute to the development of durable all-solid-state batteries.

| Materials and cell preparation
The developed laminate-type single-layer all-solid-state LIB (prototype cell, approximately 8 mAh), as depicted in Supporting Information: Figure S1, was used for the charge/discharge cycle tests. The positive electrode layer was a composite electrode consisting of an AM, SE, conductive carbon, and binder. LiNi 0.5 Co 0.2 Mn 0.3 O 2 (NCM; Sumitomo Metal Mining), which was coated with LiNbO 3 via a previously reported process, 29,30 was used as the positive electrode AM. Argyrodite-structured Li 7−x PS 6−x Cl x (x ≈ 1.2 × 10 −3 S cm −1 ; Mitsui Mining & Smelting) was used as the SE. Carbon fiber (Showa Denko K.K.) was used as the conductive carbon. Rubber polymer was used as the binder. The negative electrode layer was a composite electrode consisting of graphite (Mitsubishi Chemical Holdings), Li 7−x PS 6−x Cl x , and rubber polymer. The separator layer consisted of the SE and binder. A slurry for fabricating the positive electrode was prepared by mixing the AM, SE, carbon, and binder in a ratio of 83:15:1:1 (wt%), respectively, in an organic solvent. A slurry for fabricating the negative electrode was prepared by mixing the AM, SE, and binder in a ratio of 68:30:2 (wt%), respectively, in the organic solvent. A slurry for the separator layer was prepared by mixing the Li 7−x PS 6−x Cl x and styrene-butadiene rubber in a ratio of 97:3 (wt%) in the organic solvent. The positive and negative electrodes and the separator layer were obtained by casting the appropriate slurry onto stainless steel foils and drying under a vacuum at 100°C for 4 h. The positive electrode, negative electrode, and separator were cut into squares with 2, 2.5, and 2.5 cm sides, respectively, and isolated from each stainless steel foil. The prototype cell was assembled using a two-step pressing process as follows: the negative electrode and separator were pressed together at 392 MPa, and the positive electrode, pressed separator/negative electrode, and current collectors (stainless steel foil) were subsequently layered and uniformly pressed at 980 MPa. Then, the pressed positive electrode/separator/negative electrode stack was sealed in a laminated film. All processing was performed in an inert atmosphere (Ar)-filled glove box with a dew point below −76°C and O 2 concentration below 1 ppm. During the subsequent charge/discharge cycling tests, the prototype cell was maintained under a pressure of 200 MPa. The designed capacities of the positive and negative electrodes were approximately 2 and 2.4 mAh cm −2 , respectively, resulting in the prototype cell having a capacity of approximately 8 mAh, depending on the specific capacity of the positive electrode.

| Charge/discharge cycling tests
All prototype cells were activated and analyzed during three charge/discharge cycles. Then, the prototype cells were charged at 0.8 mA and 4.2 V in constant-current (CC) mode and was further charged in a CC-constant voltage mode. Further, they were discharged at a C/3 rate (2.38 mA) to reach 3.0 V in the CC mode at 100°C. Every 7th day, we performed charge/discharge tests under the same current conditions at 25°C and alternating current (AC) impedance measurements at a depth-of-discharge of 50% at 25°C with an AC excitation of ±5 mV over the frequency range of 10 MHz-7 MHz. The charge/discharge cycling tests were continued for 28 days, excluding the period required for the performance measurements at 25°C.

| Disassembly analysis
Disassembly analysis was performed on the prototype cells before and after the high-temperature cycling tests. The cells were discharged at a C/50 rate to the lower voltage limit immediately before disassembly. The following analyses were performed under Ar or vacuum.
For XPS measurements, the positive and negative electrodes were stripped of their current-collector foils, and their surfaces were slightly scraped. Next, XPS measurements were performed using a monochromatized Al-K α radiation source (hν = 1486.6 eV). The binding energies of the spectra were calibrated to the Cl 2p peak at 198.1 eV. The P 2p, S 2p, Nb 3d, and C 1s XPS spectra were analyzed as markers of various battery components.
The deposited phase, the surface of the negative electrode, and cross-sections of all-solid-state batteries prepared using an ion-milling system (IM4000 model; Hitachi High-Technologies) were observed using SEM (FE-SEM, S8020 model; Hitachi High-Technologies). Furthermore, the elemental composition of the surface of the negative electrode was evaluated using energydispersive X-ray spectroscopy (EDS).
The Li and P contents of the SE in the positive and negative electrodes were measured using ICP-AES (SPS 3520; SII Nano Technology) by dissolving each electrode in anhydrous ethanol. The Li and P content in the negative electrode, which included the deposits observed by FE-SEM, as well as the SE, were measured using ICP-AES by dissolving the negative electrode in acid.
The electrochemical properties of the disassembled composite electrodes were investigated using a custommade cell with an internal diameter of 1 cm. To enable the separate analysis of the electrodes of a prototype cell, another cell was prepared without the 980 MPa pressurization step and subjected to the same cycling tests. Halfcells were assembled using the disassembled positive or negative electrode as the working electrode, an SE separator, and an In-Li alloy as the counter electrode, which exhibited a flat potential plateau (0.62 V vs. Li + / Li). 31 Charge/discharge tests on the negative and positive electrode half-cells were performed in the potential ranges of −0.615 to 0.88 V versus Li + /In-Li (0.005-1.5 V vs. Li + /Li) and 3.68-1.88 V versus Li + /In-Li (4.3-2.5 V vs. Li + /Li), respectively. The initial and degraded electrodes were tested at C/50 and C/200, respectively. Figure 1 depicts the capacity retention during charging and discharging at 100°C (Figure 1A), normalized capacity retention measured at 25°C (Figure 1B), and AC impedance spectra obtained at 25°C ( Figure 1C). Supporting Information: Figure S2 depicts the charge/ discharge curves obtained during cycling tests at 100°C. The capacity of the prototype cell during charging/ discharging at 100°C gradually decreased over 28 days to 33%, and the capacity measured at 25°C significantly decreased at the beginning of testing. The difference in capacity retention at various evaluation temperatures is attributed to the capacity at 25°C being affected by an increase in internal resistance. It is evident from Supporting Information: Figure S3 that the resistance increased during cycling. The decrease in capacity at 100°C, at which point the resistance of the cell is low, appears to be because of the active Li + ions; this is similar to the degradation mechanism of conventional LIBs with a liquid electrolyte. 26,32,33 Photographs of the disassembled cells before and after the high-temperature cycling tests are shown in Supporting Information: Figure S4. Figure 2 depicts the XPS spectra of the positive and negative electrodes. Supporting Information: Figures S5 and S6 depict the ratios of the S 2p and P 2p peaks obtained by deconvolution. For the positive electrode, although the peaks attributed to lithium argyrodite represent the main peaks corresponding to sulfur and phosphorus before the cycling tests, the amounts of degradation components such as PO x , P-S-P, -S-S-, and elemental sulfur increased after the high-temperature cycling tests. It was also observed that the position of the Nb 3d peak, which corresponded to the LiNbO 3 coating on the NCM, shifted to a higher binding energy, indicating the extraction of lithium and oxygen from LiNbO 3 to Nb 2 O 5 , thus forming Li 1−x NbO 3−x/2 . 34 For the negative electrode, degradation components such as reduced P (assuming Li 3 P), Li 2 S, and CO 3 2− increased after the high-temperature cycling tests. SEM images are depicted in Figure 3 and Supporting Information: Figure S7. No remarkable change in the positive electrode was observed in the electrode cross-sectional images. For the negative electrode, inplane cracks initiating at the graphite edge were observed on the SE, which was attributed to the expansion and contraction of graphite during charging and discharging. 35 Moreover, deposits were observed in the cracks and on the surface of the current-collector side of the electrode. EDS analysis of the deposits detected only carbon and oxygen, as depicted in Figure 4.

| RESULTS AND DISCUSSION
The observed Li/P ratios, measured using ICP-AES, are listed in Table 1. The Li/P ratio of the SE in the positive electrode decreased from 6.1 to 4.5 after the cycling tests at a high temperature. In contrast, the measured Li/P ratio of the negative electrode dissolved in ethanol increased from 6.4 to 8.4 after cycling and that of the electrode dissolved in acid increased from 6.7 to 9.5 after cycling. The difference in the increase in the Li/P ratio between the ethanolextracted and acid-extracted samples appeared to be linked to the detection of the deposit, which is insoluble in ethanol.
The capacity retention curves of the prototype cells prepared with and without the 980 MPa pressurization step during charging and discharging at 100°C are shown in Supporting Information: Figure S8. The capacity retention curves of both cells are very similar.  even after cycling, almost no capacity loss was observed.
The degradation mechanisms of the sulfide-based allsolid-state LIB based on the obtained analysis results are summarized in Figure 6, which is a modified scheme from our previous report. 36 The degradation reactions of the prototype cell can be divided into four parts: (i) SE oxidation in the positive electrode, (ii) SE reduction in the negative electrode, (iii) lithium deposition on/in the negative electrode, and (iv) capacity loss of the positive electrode.

(i) SE oxidation of the positive electrode
Based on the increase in P-S-P, -S-S-, and sulfur and the decrease in argyrodite indicated by the XPS measurements, as well as the decrease in the Li/P ratio indicated by the ICP-AES measurements, Li 7−x PS 6−x Cl x in the positive electrode was considered to have been oxidatively F I G U R E 4 Energy dispersive X-ray spectroscopy maps of the negative electrode surface showing the deposited phase after the cycling tests. SEM, scanning electron microscopy.
T A B L E 1 Li/P ratios of each electrode material measured by ICP-AES.

Positive electrode (dissolved in EtOH)
Negative electrode (dissolved in EtOH)
Li PS Cl + Li + e , (ii) SE reduction at the negative electrode From the increase in the amounts of reduced P and Li 2 S components, decrease in the amount of argyrodite indicated by XPS, and increase in Li/P ratio indicated by ICP-AES, the Li 7−x PS 6−x Cl x in the negative electrode is considered to have been reduced and decomposed while capturing Li + ions, as shown in Equations (5)-(7). 21 For convenience, the basic composition, Li 6 PS 5 Cl (x = 1), is used as the starting material in Equation (5). Li P S + 7Li + 7e 3Li S + Li P.
(iii) Li deposition on (or in) the negative electrode The deposits on the negative electrode observed using SEM were columnar, similar to those observed in previous works. [37][38][39] Such deposits have been reported to be electrodeposited lithium metal. Moreover, only carbon and oxygen were detected by SEM-EDS, because this method cannot identify elements with a low atomic number, including lithium. These results suggest that the columnar deposits are lithium carbonate generated via the electrodeposition of lithium metal, as shown in Equation (8), followed by carbonation.
The sources of Li + ions and electrons consumed during (ii) SE reduction and (iii) lithium deposition on (or in) the negative electrode were assumed to be inactive Li + ions and electrons released by (i) SE oxidation in the positive electrode and the active Li + ions and electrons released from the active NCM, which originally participated during charging/discharging, as shown in Equation (9).  Table 2 lists the amounts of Li + ions released in (i) (ΔLi + (i) ), captured in (ii) (ΔLi + (ii) ), and captured in (iii) (ΔLi + (iii) ), which were calculated from the increase or decrease in the Li/P ratio and the nominal amount of SE in each electrode. The amount of Li + ions released by the reaction shown in Equation (9) (ΔLi + NCM ) was calculated using Equation (10), and the results are listed in Table 2.
The parameter ΔLi + NCM represents the slippage of the positive electrode/negative electrode reaction areas, as reported in a study of the degradation mechanisms of LIBs with liquid electrolytes. 26 Because ΔLi + NCM is approximately 54% of the initial cell capacity, as indicated by the blue arrow in Figure 6, which is similar to the capacity decrease (67%) of the prototype cell after cycling tests at 100°C, the degradation of the all-solidstate battery is considered to be adequately described by processes (i)-(iii).
Some all-solid-state LIBs containing Li 4 Ti 5 O 12 as the negative electrode AM have shown good durability. 40 This is attributed to the high potential of Li 4 Ti 5 O 12 that avoids degradation mechanisms (ii) and (iii).

(iv) Capacity loss of the positive electrode
The SOCs of the half-cells of the positive and negative electrodes after cycling tests at 100°C immediately after assembly were approximately 0% and 10%, respectively. Therefore, a small retrograde slippage between the positive and negative electrode reaction areas was observed. This was probably because the capacity decreased to 60% of that of the positive electrode, as depicted in Figure 5C, which offset the slippage between the positive and negative electrode reaction areas listed in Table 2, as depicted by the red arrow in Figure 6.   Although we could not find any definitive causes for the degradation of the positive electrode observed in our analysis, a former study attributed such degradation to structural changes, such as cracking and subdivision of the AM, and chemical decomposition at the interface between the AM and SE. 41 We considered that the reactions indicated by the red and blue arrows in Figure 6 occurred independently; it is important to avoid both reactions to prevent the capacity decrease. Finally, the increase in the charge transfer and ionic resistances of the composite (pathway resistance) 22 in the positive electrode can be attributed to the presence of sulfur, which is known to be a Li + ion insulator, 42 and was produced by (i) SE oxidation, as shown in Equations (1)-(4). In contrast, Li 2 S and Li 3 P, which are produced by (ii) SE reduction in the negative electrode, are reported to be ionic conductors 42,43 and the increase in the resistance in the negative electrode is probably restrictive. The AC impedance measurements of the positive and negative electrode symmetric cells after charge/discharge cycling at 100°C are depicted in Supporting Information: Figure S9. Although the impedance of the positive electrode symmetric cell increased, that of the negative electrode symmetric cell remained almost unchanged. The results for the symmetric cell support the hypothesis that positive electrode degradation causes an increase in cell resistance. Further, the PO x detected in the XPS spectrum of the positive electrode can be attributed to the reaction of oxygen released from NCM or LiNbO 3 with the PS 4 3− unit in the SE, thus forming a PS 4−x O x 3− unit. 29,34 It is also considered that the PS 4−x O x 3− unit causes an increase in the charge transfer and pathway resistances in the positive electrode because the ionic conductivity of Li 2 S-P 2 S 5 is higher than that of Li 2 S-P 2 O 5 . 44,45

| CONCLUSION
We performed charge/discharge cycling tests at 100°C and disassembly analysis of argyrodite-type sulfide-based all-solid-state prototype cells before and after accelerated degradation. Based on the findings, we developed an overall degradation mechanism for the sulfide-based allsolid-state LIBs. The degradation reactions of the prototype cell can be divided into four processes: (i) SE oxidation of the positive electrode, which causes capacity recovery by releasing inactive Li + ions and increasing the resistance by forming sulfur; (ii) SE reduction at the negative electrode, which causes a decrease in the capacity by capturing active Li + ions; (iii) Li deposition on/in the negative electrode, which causes a decrease in the capacity by capturing active Li + ions; (iv) capacity loss of the positive electrode and decrease in the capacity, which cause a decrease in the overall capacity. Although (i) is specific to all-solid-state LIBs, the other three processes are similar to the known degradation mechanisms of conventional LIBs. Notably, all these degradation reactions occur simultaneously. For the development of highly durable all-solid-state batteries, all degradation reactions must be identified and resolved. The findings of this study will potentially encourage the development of highly durable all-solid-state batteries and the reported analysis technique could enable the identification of the degradation mechanisms of other battery technologies and electrode materials.