Binder-free sheet-type all-solid-state batteries with enhanced rate capabilities and high energy densities

All-solid-state batteries using inorganic solid electrolytes are considered promising energy storage systems because of their safety and long life. Stackable and compact sheet-type all-solid-state batteries are urgently needed for industrial applications such as smart grids and electric vehicles. A binder is usually indispensable to the construction of sheet-type batteries; however, it can decrease the power and cycle performance of the battery. Here we report the first fabrication of a binder-free sheet-type battery. The key to this development is the use of volatile poly(propylene carbonate)-based binders; used to fabricate electrodes, solid electrolyte sheets, and a stacked three-layered sheet, these binders can also be removed by heat treatment. Binder removal leads to enhanced rate capability, excellent cycle stability, and a 2.6-fold increase in the cell-based-energy-density over previously reported sheet-type batteries. This achievement is the first step towards realizing sheet-type batteries with high energy and power density.

Exploration of solvent and aliphatic polycarbonates. The requirements for a good solvent are that it should not affect the ionic conductivity of LPS, should dissolve the binder, and have a moderate vapor pressure. The ionic conductivities of LPS after exposure to various solvents were investigated to evaluate the stability of LPS (Fig. 2a). As the solvent donor number 22 is increased to values greater than 14, solvent-exposed LPS shows a significantly decreased ionic conductivity and its color changes from pale yellow to brown and green. LPS became sticky after exposure to propylene carbonate (PC), with an ionic conductivity that was too low to measure. In contrast, LPS showed negligible changes in color and ionic conductivity when exposed to solvents with a donor number of 9 or less, such as anisole, toluene, 1,2-dichloroethane, and n-decane. Since solvents with higher donor numbers have higher electron-donating abilities, we conclude that solvents with donor numbers greater than 14 decomposed LPS by nucleophilic attack, causing a decrease in ionic conductivity.
We investigated the solubility of aliphatic polycarbonates in solvents with donor numbers of 9 or less (see Supplementary Table S1 online). Poly(ethylene carbonate) (PEC) dissolved only in 1,2-dichloroethane. Positive sheets prepared with PEC/1,2-dichloroethane had a whitish surface, implying that lightly pulverized fine 75Li 2 S·25P 2 S 5 glass powder (f-LPS) floated while sedimentation occurred for heavy LiNbO 3 -coated LiNi 1/3 Co 1/3 Mn 1/3 O 2 (NCM). Moreover, many bulges were observed on the surface of the positive sheets due to the rapid evaporation of 1,2-dichloroethane, which has a high vapor pressure (8.7 kPa, 20 °C) 23 (see Supplementary  Table S2 online). In contrast, poly(propylene carbonate) (PPC) dissolved in both 1,2-dichloroethane and anisole. Positive sheets prepared using PPC/anisole had smooth surfaces with uniform color, implying the homogeneous dispersion of NCM and f-LPS. The slow evaporation of anisole, the vapor pressure of which is low (0.47 kPa, 25 °C) 23 , prevents the segregation of NCM and f-LPS and the formation of cracks and bulges during the drying process. It is worth noting that positive sheets with a pre-compression thickness of 200-300 μm (60-75 μm after compression) were difficult to break by punching even 5 months after fabrication (see Supplementary Fig. S2). Thus, we selected a combination of PPC and anisole for subsequent experiments.
A PPC-containing SE slurry (Fig. 2b) enables the fabrication of bendable free-standing SE sheets (Fig. 2c), which can be punched to afford SE disks (Fig. 2d), and the fabrication of two-layered SE/electrode sheets by slurry coating (Fig. 2e).
Since PC exposure causes a significant decrease in the ionic conductivity of LPS (Fig. 2a), the PC that arises from the depolymerization of PPC (see Supplementary Fig. S1) must be removed from the as-prepared cells quickly under vacuum to minimize the exposure time. We studied the electrochemical properties and the structure of PPC-containing LPS before and after removal of PPC by vacuum heating. Figure 2f shows the ionic conductivity of both pristine and 6 wt.% PPC-containing LPS as a function of the heat treatment temperature. The ionic conductivity of pristine LPS, 4.1 × 10 −4 S cm −1 increased slightly to 5.2 × 10 −4 S cm −1 after heat treatment at temperatures between 160-290 °C 5 . Compared to pristine LPS glass, the conductivity of PPC-containing LPS (PPC; 6 wt.%) decreased significantly to 1.1 × 10 −4 S cm −1 , which suggests that non-conductive PPC interferes with ionic conduction at the interface between LPS particles. The ionic conductivity of PPC-containing LPS gradually increased with heat-treatment temperature, returning to that of pristine LPS between 225-290 °C. Even though the sheet size increased from 10 mm to 40 mm in diameter and PPC content increased from 6 wt.% to 10 wt.%, their conductivities returned to that of pristine LPS after heat treatment at 225 °C. Thus, we concluded that heat treatment at 225 °C is sufficient to decompose PPC and remove PC from PPC-containing LPS. Since both pristine and PPC-containing LPS show a similar color change from pale yellow to dark brown with increasing temperatures (see Supplementary Fig. S3), this change was attributed to the heat treatment rather than the removal of PPC.
We confirmed that PPC-containing LPS before and after heat treatment maintains both a framework structure containing PS 4 3− moieties 24 (see Supplementary Fig. S4a) and a wide electrochemical window of 5 V (see Supplementary Fig. S4b). DC conductivities were evaluated for both pristine and PPC-containing LPS before and after heat treatment to investigate whether the electronic conductivity of LPS is affected by the carbon arising from the decomposition of PPC under an oxygen-free atmosphere (see Supplementary Fig. S4c). While the total conductivity (including ionic and electronic contributions) of all samples were close to those obtained by AC impedance measurements, the electronic conductivities were ~4 orders of magnitude lower than the corresponding total conductivities, suggesting that residual carbon in the LPS layer had a negligible effect. These results demonstrate that the original ionic and electronic conductivity of LPS are retained after the PPC removal under vacuum heat treatment. Effect of binder removal on cell performance. Figure 3a compares the initial charge-discharge curves of binder-containing and binder-free positive half-cells constructed using as-prepared 3 wt.% PPC-containing positive sheets, which had average discharge voltages of 3.04 and 3.15 V and discharge capacities of 123 and 151 mAh g −1 , respectively. The binder-free discharge capacity corresponded to the substantial theoretical capacity of NCM 25 . Impedance measurements were performed to clarify the differences in cell performance (Fig. 3b,c). The resistance at the intersection with the real axis around 10 kHz represents the bulk and grain boundary resistance of the SE layer (R SE ) 26 . The resistance at the semicircles with peak-top frequencies of ~500 and 1 Hz correspond to the charge transfer resistance of NCM/SE (R I ) and In/SE interface (R In/SE ), respectively 26 . While PPC was not used in the SE layer, the R SE value was reduced by ~40 Ω by heat treatment, implying a decrease in the grain boundary and void volume by heating to temperatures near the glass transition temperature of LPS 19,20 . The R I value was drastically reduced from 638 Ω to 14 Ω by the removal of PPC, suggesting an increased contact area between NCM and the SE.
The effect of PPC content on the positive sheets was investigated for binder-containing and binder-free positive half-cells using positive sheets containing 3 wt.% and 6 wt.% PPC (Fig. 4a). The R I of binder-containing half-cell A (PPC: 3 wt.%) was half that of binder-containing half-cell C (PPC: 6 wt.%), increasing the capacity. Binder removal induced drastic decreases in R I , resulting in improved capacity for both cells. Compared to binder-free half-cell D (PPC: 6 wt.%), the binder-free half-cell C (PPC: 3 wt.%) exhibited a higher discharge capacity and one-fifth as much R I . The small R I indicates a large contact area between NCM and the SE. When the volume ratio of PPC/SE in the as-prepared positive sheet (PPC/SE = 0.22 (PPC: 3 wt.%), PPC/SE = 0.45 (PPC: 6 wt.%)) is small enough, voids and cracks between NCM and the SE originating from the removal of PPC can be filled completely during the plastic deformation of the SE by compression. The R I value for the binder-free sheet-type half-cell (PPC; 3 wt.%) was consistent with that of a binder-free pellet-type half-cell, suggesting a similar NCM and SE contact area was achieved by binder removal (Fig. 4b). Conversely, when the PPC content was too low (around 1 wt.%), the positive sheets crumbled easily. Although the PPC content should be adjusted by the specific surface area of the components and composition ratio of active materials, SE, and conductive additive, in this case, a PPC content of around 3 wt.% is desirable for manufacturing positive sheets.
In addition to the improved positive half-cell performance, the binder removal process also improved the charge-discharge capacity of negative half-cells using as-prepared 3 wt.% PPC-containing negative sheets (see Supplementary Fig. S5). The initial charge capacity increased from 257 to 308 mAh g −1 , which was close to the theoretical value for graphite of 330 mAh g −1 estimated for conventional LIBs 27 . The binder removal process was further applied to sheet-type full-cells constructed using an as-prepared SE sheet containing 3 wt.% PPC (~180 μm in thickness) and positive/negative sheets (Fig. 5). Binder removal remarkably improved the average discharge plateau from 3.54 V to 3.62 V, and the discharge capacity from 103 mAh g −1 to 122 mAh g −1 (Fig. 5a). The corresponding impedance profiles show a decrease in R SE (Fig. 5b,c), suggesting that the removal of PPC from the SE layer induces a decrease in the grain boundary and void volume 19,20 . As for the positive half-cells (Fig. 3b,c), R I for the full-cell was also substantially reduced by the removal of PPC from 836 Ω to 60 Ω. Improved rate capability and stable cycling for binder-free cells. Figure 6a compares the rate capability of binder-free and binder-containing positive half-cells. Under a low current density of 0.15 mA cm −2 , both cells show slight differences in their measured capacities (153 mAh g −1 for the former, 155 mAh g −1 for the latter), comparable with the theoretical value of NCM 25 . The use of milder measurement conditions, such as CCCV-charge mode and an In-Li counter electrode, caused the increased initial capacity in the latter cell, compared to Fig. 3a. Notably, with increasing current densities, the binder-containing cells display rapid decays of capacities and negligible capacities, 15 mAh g −1 , starting at 3.0 mA cm −2 . In contrast, the binder-free cell exhibited significantly higher capacity, delivering ~50 mAh g −1 even at 3.0 mA cm −2 . Further cycling at a low current  Thus, we concluded that the rate capability is remarkably improved by reducing the R I value.
Stable charge/discharge cycling was exhibited by binder-free positive half-cells after 25 cycles at the above testing rate (Fig. 6b). The cell shows an initial discharge capacity of 149 mAh g −1 with 100% coulombic efficiency and retains a high capacity of greater than 125 mAh g −1 for at least 175 cycles. The capacity retention rate of binder-free and binder-containing half-cells that were operated at several current densities are shown in Supplementary Fig. S6a, which obtained from the discharge capacity of N th cycle compared to that of (N-4) th cycle in Fig. 6a. The binder-free cell operated at 0.15 mA cm −2 had a capacity retention rate of 99.4% per 5 cycles (30 th /26 th ). Thus, the capacity after the 175 th cycle was estimated to be 121 mAh g −1 (see Supplementary Fig. S6b), which is comparable with the experimental value (125 mAh g −1 ) in Fig. 6b. This agreement suggests the capacity retention rate per 5 cycles may be a reliable measure for estimating the long-term stability under various current densities. The capacity retention rates of the binder-free cell decreased with increasing current densities (see Supplementary Fig. S6a). This decrease can be improved by reducing the dominant cell internal resistance for the half-cell, R SE , by using a thinner SE layer, a SE with a higher ionic conductivity (~10 −2 S cm −1 ) 10 , and higher-temperature operation (~100 °C) 10 . In comparison with a binder-containing cell, the binder-free cell maintains higher capacity retention rates at every current density (see Supplementary Fig. S6a). Therefore, the reduction of R I by the removal of the binder is effective to enhance long-term stability especially for the operation under higher current densities. Stable cycling was also exhibited for binder-free negative half-cells (Fig. 6c). The cell shows an initial charge capacity of 319 mAh g −1 with 88% coulombic efficiency. Notably, extended testing for more than 250 cycles indicated excellent reversible capacity above 240 mAh g −1 with coulombic efficiencies in excess of 99% after a few initial cycles.

Fabrication of binder-free sheet-type full-cells with high cell-based-energy-density.
We fabricated SE sheets with half their original thickness and positive/negative sheets with twice the original loading of active materials to improve the cell-based energy density. Moreover, the capacity ratio of negative to positive sheets, N/P, was tuned to be ~1.2-1.5 to prevent the occurrence of fine short-circuiting associated with the formation of lithium dendrites on the negative layer, although N/P is generally tuned to 1.1 in the industrial production of LIBs. As a result, the cell-based-energy-density of sheet-type binder-free full-cells reached 115 Wh kg −1 , which was calculated from the initial discharge capacity (116 mAh g −1 ), the average discharge voltage (3.51 V), and the weight of the cell without current collectors (PE 16.38 mg, SE 10.93 mg, NE 16.89 mg) (see Supplementary  Fig. S7). The above binder-free sheet-type full-cell after 350 cycles could be taken out from the mold without breaking, indicating its free-standing feature, and its microstructure was observed by SEM (Fig. 7). The thickness of the positive layer, the SE layer, and the negative layer were estimated to be 74, 59, and 134 μm, respectively (Fig. 7a). Despite a relatively thin SE layer, the positive and negative sheets were successfully stacked on both sides of the SE sheet without short-circuiting. NCM particles were dispersed in the SE matrix homogeneously and made intimate contact with the SE (Fig. 7b). In addition, no pulverization of NCM by compression between NCM particles 14 was observed in the cross-sectional view of the positive layer. The SE layer appears as an almost completely dense region without voids, cracks, or grain boundaries (Fig. 7c). Similarly, smooth cross-sections were also observed in hot-pressed SE pellets 19,20 . Even without added pressure, sintering of SE particles may be promoted by heat treatment near the glass transition temperature. The originally spherical graphite powder was crushed with cleavage in the direction perpendicular to the external pressure used in the fabrication of the cell to form a closely packed composite with the SE particles (Fig. 7d). Scarcely any evidence of PPC removal, such as visible voids and cracks in the electrode and electrolyte layers, was found by SEM. Surprisingly, the active materials and SE make intimate contact and are densely packed even after 350 cycles and despite graphite's ~10 vol% volume change 28 .

Discussion
Our work demonstrates the novel fabrication process of sheet-type ASSLBs in which the binder can be removed. Most of the conventional binders for LIBs, such as PVdF, and for sheet-type sulfide-based ASSLBs, such as silicone 12 , SEBS 14,15 , SBS 15 , PEG 16 , and BR 17 decompose around 400 °C under an oxygen-free atmosphere causing carbon deposits, which may affect the electronic conductivity of the separator layer. Therefore, we focused on volatile aliphatic polycarbonates, which depolymerize by unzipping at lower temperatures. The aliphatic polycarbonates tend to be hard to dissolve in solvents with low donor number, while SE tends to be decomposed by solvents with a high donor number. A combination of PPC and anisole fulfilled these conflicting demands.
PPC acts as a suitable dispersant and binder to make homogeneously dispersed slurries that afford bendable free-standing SE sheets, splinter-free electrodes during punching, and two-layered SE/electrode sheets by coating. The removal of PPC even from large-area SE sheets with high PPC content induces no adverse effects on the conductivity of the SE, implying the possibility of preparing large-area sheet-type cells which are important for applications. The binder-free sheet-type full-cells were also free-standing, similar to the previously reported binder 12,15 , polymer mesh 12 and nonwoven 13 -containing sheet-type cells. This property was accomplished by the intimate contact between the SE and the active materials through the plastic deformation of SE 19,20 by compression after heat-treatment. Therefore, the SE can act as a binder after the removal of PPC. The binder removal strategy is a simple, effective, and fruitful approach to producing sheet-type cells. To the best of our knowledge, this report is the first case that realizes binder-free sheet-type sulfide-based ASSLBs by coating.
Long-term stability of the binder-free half-cells was achieved despite operation under much harsher conditions relative to previously reported pellet-type half-cells [7][8][9] , including twice the ratio of positive active materials in the electrode and a two-fold increase in current density. Although previously reported binder-containing half-cells were investigated for only 20 cycles or less 12,14,15 , and some battery exhibits significant capacity fading 12 , we validated the reliability of the binder-free half-cells over 175 cycles. While gradual capacity fading for the binder-free full-cell was observed over 350 cycles, this may be improved by using a SE with high ionic conductivity and high operating temperatures 10 . Work to enhance the reversible cycle stability of these full-cells is currently in progress.
Binder removal drastically decreases R I , which is lower than that reported for SEBS 14,15 and SBS-containing cells 15 , and reaches values close to those seen for binder-free pellet-type cells (Fig. 4c). As a result, the rate capability (Fig. 6a) and capacity retention rates (see Supplementary Fig. S6a) were improved by binder removal, especially under higher current densities. As the thickness of the SE layer decreases and the conductivity of SE (10 −2 S cm −1 ) 10 increases, R I is expected to become the dominant factor in the total cell internal resistance, comprising R I and R SE 29 . Therefore, minimizing R I is critical to improving the rate capability and long-term stability, and is effectively accomplished by removing the binder from sheet-type cells.
Stackable binder-free sheet-type batteries may potentially satisfy the demands of next-generation electronic vehicles in terms of energy density. These batteries exhibited an energy density of 115 Wh kg −1 , which is a ~2.6-5-fold increase compared with conventional pellet-type cells (LiCoO 2 /graphite) 10,11 and a 2.6-fold increase relative to nonwoven-containing sheet-type cells (LiCoO 2 /Li 4 Ti 5 O 12 ) 13 . This feat was accomplished through the fabrication of thinner SE sheets and homogeneously dispersed electrodes with higher active material weight fractions by using PPC.
The results of this study provide a novel and unique process that can enhance rate capability, long-term stability and cell-energy-density, and offer opportunities to build high-performance sheet-type ASSLBs that are not limited to NCM/graphite ASSLBs.

Methods
Evaluation of solvent-exposed LPS and PPC-containing LPS before and after heat treatment.
All processes were performed in a dry Ar atmosphere. Three types of 75Li 2 S · 25P 2 S 5 (Li 3 PS 4 ) were prepared according to previous reports: Li 3 PS 4 glass (LPS) prepared by mechanical milling (~10 μm in diameter, 4 × 10 −4 S cm −1 ) 5 ; pulverized fine Li 3 PS 4 glass (f-LPS) (~3 μm in diameter, 2 × 10 −4 S cm −1 ) 30 ; thin rod-like Li 3 PS 4 (LS-LPS) prepared by liquid-phase shaking (LS) (ca. 3 μm × 500 nm × 100-200 nm, 2 × 10 −4 S cm −1 ) 31,32 . The stability of LPS toward solvents was evaluated by soaking LPS powder in a solvent for 24 h followed by drying for a few days at room temperature. PPC-containing LPS was prepared by mixing LPS and PPC (M w = 357,000, PO units = 3 wt.%, Sumitomo Seika Chemicals) in a weight ratio of 100:6 in anisole and drying the resultant slurry at room temperature for a few days in a dry Ar atmosphere. Heat treatment of the resultant powder was carried out in a stainless-steel container on a hot plate under vacuum for 30 min.
The ionic conductivities of pelletized samples were measured by AC impedance spectroscopy at frequencies ranging from 0.1 Hz to 1 MHz under an applied voltage of 10 mV (FRA1455, Solartron) in a symmetric cell (stainless-steel/sample/stainless-steel). The structure was studied by Raman spectroscopy with a green laser at 532.05 nm (NRS-3100, JASCO). Cyclic voltammetry measurements were conducted to quantify the voltage stability window for the pelletized sample using Li foil as the counter/reference electrode, a stainless-steel disk as the working electrode, and a sweep rate of 5 mV s −1 between −0.1 to 5 V. DC conductivity was measured using a potentiostat (Multistat 1470E, Solartron) at room temperature under a constant DC voltage of 50 mV in a symmetric cell configured as either Li/sample/Li or stainless-steel/sample/stainless-steel, and calculated from the thickness of the pellets and obtained current values.
Fabrication of binder-free sheet-type ASSLBs. Typical electrode and SE sheets were prepared by mixing the active materials LiNbO 3 -coated LiNi 1/3 Co 1/3 Mn 1/3 O 2 (NCM) 33 (18 μm), and drying at room temperature (see Supplementary  Fig. S8). Two-layered SE/electrode sheets were also prepared by coating an SE slurry on the electrode sheet.
Typical half-cells were fabricated using the electrode sheet as the working electrode, SE glass powder as the separator, and In foil or In-Li foil as the counter and reference electrodes. The electrode sheet and SE powder (80 mg) were placed in a 10 mm-diameter polyimide mold and pressed together by applying a pressure of 110 MPa 14 . Then, the bilayer sheet was placed in a stainless-steel container and heated at 225 °C for 30 min under vacuum. An In foil or In-Li foil were placed on the surface of the SE side of the bilayer sheet and compressed using two stainless-steel rods as current collectors under ~100 MPa. The sheet-type full-cells were constructed as described above except for simultaneous stacking of the PPC-containing positive, the SE, and the negative sheet.
Evaluation of the binder-free sheet-type ASSLBs. Charge-discharge cycling was conducted using a charge-discharge measurement device (BTS-2004, Nagano) between cut-off voltages of 2.0-3.7 V vs. Li-In for the positive half-cells, −0.57-0.88 V vs. Li-In for the negative half-cells, and 3.0-4.2 V for the full-cells at room temperature. The capacity calculations are based on the mass of NCM for positive half-cells and full-cells, and graphite for negative half-cells. AC impedance measurements were performed after initial charging at a current density of 0.064 mA cm −2 under frequencies ranging from 0.01 Hz to 1 MHz at an applied voltage of 10 mV. The morphology and microstructure of the full-cell were characterized by scanning electron microscopy (KEYENCE VE-9800) at 10 kV. Cross sections of the specimens were formed by Ar milling using a cooling cross section polisher (CP, JEOL 1B-19520CCP). The samples were transferred in an Ar atmosphere from the glove box to the equipment for CP and SEM.