Introduction

Various high-performance lithium-ion rechargeable batteries, such as all-solid-state batteries, have been developed to address the demand for technological development under the challenges of climate change [1] and realize a sustainable carbon-neutral society [2,3,4]. The performance of all-solid-state batteries mostly depends on the electrochemical properties and lithium-ion conductivity of the electrolyte [5]. Although conventional all-solid-state lithium-ion batteries have low-rate capabilities and energy densities, recent studies have demonstrated that lithium–phosphorus–sulfide solid electrolytes (SE) show improved ionic conductivity [6,7,8] and may be easily integrated into battery production because of their mechanical softness [9] and facile processing [10].

Among the phosphorus sulfides, we have discovered various Li7-αPS6-αXα (X = Cl, Br, I) argyrodites that exhibit high ionic conductivities [11,12,13,14,15,16,17], denoted as Li7-α(PS4)(S2-αXα) [18]. In a Li7-x–y(PS4)(S2-x-yClxBry) system [18], Li5.4(PS4)(S0.4Cl1.0Br0.6) showed excellent electrochemical stability in Li | Li5.4(PS4)(S0.4Cl1.0Br0.6) | Li(Ni0.8Co0.1Mn0.1)O2–Li5.4(PS4)(S0.4Cl1.0Br0.6) batteries. Nevertheless, the discharge capacity of this battery (140 mAh/g) was still lower than those of other all-solid-state batteries (> 140 mAh/g) [19].

To explain the low discharge capacity, we measured the electronic and ionic conductivities of cathode mixtures. Changing the amount of cathode active material from 70 to 90 wt% increased the electronic conductivity of the cathode mixture, but reduced its ionic conductivity. Unfortunately, both high electronic and lithium-ion conductivities are necessary for generating an effective composite cathode. In addition, much predomination of electronic conductivity contributes to the discharge capacity [19].

Several studies have attempted to enhance the capacity of various sulfide solid electrolytes using a carbon additive (CA) to increase the electronic conductivity of the cathode [20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42]. The discharge capacity of the Li | β-Li3PS4 | Li(Ni0.6Co0.2Mn0.2)O2–β-Li3PS4 battery increased with a CA-modified cathode. However, the capacity retention rate was ~ 85% after 50 cycles [23]. Similarly, the retention rate of the In-Li | Li6.0(PS4)(S1.0Cl1.0) | Li(Ni0.6Co0.2Mn0.2)O2–Li6.0(PS4)(S1.0Cl1.0) battery was 79% after 50 cycles [42]. With a CA, the capacity decreased with repeated cycles, although the initial capacity was high [23]. The decreased capacity after cycling was likely related to a decomposition reaction of SE, although the mechanism remains unclear [43, 44]. However, it is unclear whether an all-solid-state battery with a CA-modified cathode can maintain a consistent capacity after cycling.

We hypothesized that a CA-modified cathode would enhance the capacity of an all-solid-state battery using a high ion-conductive Li5.4(PS4)(S0.4Cl1.0Br0.6) solid electrolyte. We previously showed that batteries with a cathode mixture consisting of 70 wt% active material and 30 wt% Li5.4(PS4)(S0.4Cl1.0Br0.6) had the lowest capacity [19]; however, the capacity of the battery may be rescued by adding CA to increase the electronic conductivity of the cathode mixture [19]. Unfortunately, the effects of CA cathode modifications in all-solid-state Li5.4(PS4)(S0.4Cl1.0Br0.6) batteries are unclear. To increase the battery capacity, the cathode mixture should preferably have a high active material ratio. However, we selected 70 wt% active material and 30 wt% Li5.4(PS4)(S0.4Cl1.0Br0.6) to investigate the effect of electronic conductivity on capacity during cycling. We measured the changes in the discharge capacity and capacity retention rates of the Li | Li5.4(PS4)(S0.4Cl1.0Br0.6) | Li(Ni0.8Co0.1Mn0.1)O2–Li5.4(PS4)(S0.4Cl1.0Br0.6)–CA battery using cycling tests, followed by impedance analysis to evaluate the battery resistance during cycling.

Methods

Synthesis of solid electrolyte and cathode mixture

Li5.4(PS4)(S0.4Cl0.1Br0.6) powder and LiNbO3-coated Li(Ni0.8Co0.1Mn0.1)O2 were synthesized and the thickness of the coated layer (4.2 nm) was calculated, as we have previously described [18, 19]. Mixed cathode powders were prepared using LiNbO3-coated Li(Ni0.8Co0.1Mn0.1)O2, Li5.4(PS4)(S0.4Cl1.0Br0.6), and CA (Li100, Denka) in weight ratios of 70:30:0 (0 vol%), 70:30:3.4 (5 vol%), 70:30:7.2 (10 vol%), and 70:30:21.5 (25 vol%), respectively. CA made from acetylene black was selected in this study because it has previously been reported to maintain a high-capacity retention rate even after cycling [20]. The densities for Li(Ni0.8Co0.1Mn0.1)O2 and Li5.4(PS4)(S0.4Cl1.0Br0.6) were selected as previously described [19] and the CA load was 2.16 g/cm3 [45]. The volume ratios were calculated from these densities and weight ratios. Volume ratio is used to describe the percolation of ions and electrons within a three-dimensional composite cathode [46]. The cathode mixtures were placed in a ZrO2 pot (45 ml) containing a ZrO2 ball (2.0 mm diameter; 34 g) in an argon-filled glovebox for dry ball-milling. The mixing condition was the same as in our previous study [19].

All-solid-state battery fabrication and electrochemical measurements

We fabricated a battery with a Li | Li5.4(PS4)(S0.4Cl1.0Br0.6) | Li(Ni0.8Co0.1Mn0.1)O2–Li5.4(PS4)(S0.4Cl1.0Br0.6)–CA structure. The total amount of cathode active material was 18 mg. Lithium foil (10 mm diameter, 0.2 mm thickness; Honjo Metal, Osaka, Japan) was used as the anode. The solid electrolyte (100 mg) was pressed into 10-mm-diameter pellets at 300 MPa. The cathode mixtures were pressed into 10-mm-diameter pellets at 600 MPa to form a cathode electrode layer. Finally, lithium metal was attached to the opposite side of the cathode and pressed at 100 MPa.

Electrochemical measurements were performed while the battery pellets were loaded at 20 MPa using a screw and torque wrench. The battery was charged and discharged between 2.5 and 4.3 V at 298 K using a potentio-galvanostat (VMP-3, Biologic, Seyssinet-Pariset, France). The atmosphere contained less than 1 ppm moisture and oxygen. The current density was fixed at 0.24 or 1.2 mA/cm2, corresponding to 0.1 C and 0.5 C, respectively. Impedance spectra were collected using the potentio-galvanostat. The charge and discharge capacity values at the 1st, 2nd, 10th, 20th, 30th, 40th, and 50th cycles were measured at 0.1 C. The values at all other cycles were measured at 0.5 C to accelerate the capacity degradation. Impedance spectra were collected at the 1st, 10th, 20th, 30th, 40th, and 50th cycles under a state of charge (SOC) of 0%, 50%, and 100%. Before conducting the impedance measurements, the charge and discharge operations were stopped for 5 min. Impedance spectra were measured for the open-cell state with a voltage amplitude of 10 mV over a frequency range of 106 to 0.01 Hz at 298 K. All measurements were conducted under 1 ppm of moisture and oxygen. For an all-solid-state battery comprising a sulfide solid electrolyte, 300 to 600 MPa can be applied to manufacture the pellet and 10 to 70 MPa during cycling [18, 19, 47,48,49]. Impedance spectra were fitted using Zmeam software (Zmeam_v109002) [50].

Ionic and electronic conductivity measurements of cathode mixtures

The ionic and electronic conductivities of the cathode mixtures were measured as previously described [19]. The ionic conductivities of the cathode mixtures were measured using an electron-blocking cell with a Li | Li5.4(PS4)(S0.4Cl1.0Br0.6) | Li(Ni0.8Co0.1Mn0.1)O2–Li5.4(PS4)(S0.4Cl1.0Br0.6)–CA | Li5.4(PS4)(S0.4Cl1.0Br0.6) | Li structure. Cathode mixtures (total weight: 200 mg) sandwiched with 50 mg of Li5.4(PS4)(S0.4Cl1.0Br0.6) from both sides were pressed into 10-mm-diameter pellets at 600 MPa. Subsequently, Li foil (10 mm φ, thickness: 0.2 mm; Honjo Metal) was applied on both ends and pressed at 100 MPa. Constant voltages (Eapp_i) of 10, 20, 30, 40, and 50 mV were applied for 2 h. The resistance of the electron-blocking cell was calculated from the slope between Eapp_i and current using the current and voltage values obtained after 2 h. The electron-blocking cell resistance includes the solid electrolyte and mixture resistances. The resistance of the solid electrolyte and cell length of Li5.4(PS4)(S0.4Cl1.0Br0.6) (equivalent to 16.2 Ω and 7.4 × 10−2 cm at 100 mg, respectively) were calculated [18] and subtracted from the electron-blocking cell resistance. The ionic conductivity was calculated using the surface area (0.785 cm2), subtracted resistance, and cell length. The measurements were performed while the electron-blocking cell was compressed at 20 MPa using a screw and torque wrench in an atmosphere with 1 ppm moisture and oxygen.

The electronic conductivity of the cathode composites was measured using an ion-blocking electrode of stainless steel (SUS) | Li(Ni0.8Co0.1Mn0.1)O2–Li5.4(PS4)(S0.4Cl1.0Br0.6)–CA | SUS. The ion-blocking electrode (total weight: 200 mg) was pressed into 10-mm-diameter pellets under 600 MPa. Thereafter, constant voltages (Eapp_e) of 10, 20, 30, 40, and 50 mV were applied to the electrode pellets for 2 h, and the resistance of the composite was calculated from the slope between Eapp_e and current. This calculation does not require correction of the solid electrolyte resistance. The measurements were performed while the electrode pellets were compressed at 20 MPa using a screw and torque wrench in an atmosphere with 1 ppm moisture and oxygen.

Results and discussion

Charge and discharge capacities of Li | Li5.4(PS4)(S0.4Cl1.0Br0.6) | Li(Ni0.8Co0.1Mn0.1)O2–Li5.4(PS4)(S0.4Cl1.0Br0.6)-CA batteries

In our previous study, the battery using LiNbO3-coated Li(Ni0.8Co0.1Mn0.1)O2 and Li5.4(PS4)(S0.4Cl1.0Br0.6) in weight ratios of 70:30 without CA exhibited the lowest discharge capacity among the batteries comprising different ratios of Li(Ni0.8Co0.1Mn0.1)O2 and Li5.4(PS4)(S0.4Cl1.0Br0.6) [19]. The discharge capacity and capacity retention rate of the battery using 70:30 LiNbO3-coated Li(Ni0.8Co0.1Mn0.1)O2 and Li5.4(PS4)(S0.4Cl1.0Br0.6) after 50 cycles were 3.1 mAh/g and 99.1%, respectively. Compared with that when using 0 vol% CA in the cathode mixture [LiNbO3-coated Li(Ni0.8Co0.1Mn0.1)O2 and Li5.4(PS4)(S0.4Cl1.0Br0.6) in weight ratios of 70:30], the discharge capacity increased from 3.1 to 167 mAh/g after 50 cycles when using 5 vol% (Fig. 1a). The capacity retention rate and coulombic efficiency after 50 charge/discharge cycles were 95.4% and 99.9% when using 0.1 C and 0.5 C, respectively. This represents a much higher capacity retention rate compared with that of the In-Li | Li6.0(PS4)(S1.0Cl1.0) | Li(Ni0.6Co0.2Mn0.2)O2–Li6.0(PS4)(S1.0Cl1.0) battery [42] after 50 cycles (79%). However, the discharge capacity decreased when the amount of CA exceeded 5 vol%. The discharge capacity of the battery using 25 vol% was 25 mAh/g at most in the initial cycle and the battery suddenly short-circuited after 6 cycles. The first explanation for this phenomenon is that decomposed products formed at the SE/CA interface, such as the decomposition of solid electrolyte observed by cyclic voltammogram in the Li | β-Li3PS4 | β-Li3PS4-CA configuration [43]. The bonding of terminal sulfur such as PS4 tetrahedra in β-Li3PS4 was experimentally confirmed to be decomposed into bridged sulfur compounds (-S-) above 3.5 V [43, 44]. This indicated that PS4 tetrahedra containing solid electrolyte decomposed at high voltage. In our previous study, Li5.4(PS4)(S0.4Cl1.0Br0.6) was stable above 10 V without CA. This suggests that an increased probability of reaction between the CA and Li5.4(PS4)(S0.4Cl1.0Br0.6) exists in batteries using the 25 vol% cathode. Notably, 10 vol% corresponded to a high discharge capacity, almost equivalent to that at 5 vol%. This suggests that Li5.4(PS4)(S0.4Cl1.0Br0.6) is stable and its decomposition is negligible in the 10 vol% cathode. The second explanation is that decomposed products were formed at the Li(Ni0.8Co0.1Mn0.1)O2/Li5.4(PS4)(S0.4Cl1.0Br0.6) interface. We previously reported amorphous impurities generated at the Li(Ni0.8Co0.1Mn0.1)O2/Li5.4(PS4)(S0.4Cl1.0Br0.6) interface during cycling, which increased impedance. Because we used a similar composition of cathode active material and solid electrolyte, we presumed that amorphous impurities formed in systems with high CA contents [19]. A third explanation is that impurities were generated at the Li(Ni0.8Co0.1Mn0.1)O2/CA interface, though this requires validation [23]. Consequently, we hypothesized that the short circuit in 25 vol% CA batteries was due to the decomposition reactions at the Li5.4(PS4)(S0.4Cl1.0Br0.6)/CA and Li(Ni0.8Co0.1Mn0.1)O2/Li5.4(PS4)(S0.4Cl1.0Br0.6) interfaces.

Fig. 1
figure 1

a Cycling durability (filled circles) and coulombic efficiency (asterisks) of the all-solid-state battery using (red) 0 vol%, (black) 5 vol%, (blue) 10 vol%, and (orange) 25 vol% cathode mixtures measured at 0.1 C. b Charge/discharge curve of 2nd cycle at 0.1 C. The red, black, blue, and orange lines represent 0 vol%, 5 vol%, 10 vol%, and 25 vol%, respectively. c Capacity curve of all-solid-state battery using 5 vol% cathode at 3rd and 49th cycle charged/discharged at 0.5 C

Full size image

We previously measured the electronic and ionic conductivities of cathode mixtures without CA and found that changing the cathode active material content from 70 to 90 wt% promoted electronic conductivity while reducing ionic conductivity. Because electronic conductivity changed more drastically than ionic conductivity, it is likely that electronic conductivity was the key factor affecting discharge capacity [19]. We adjusted the amount of CA (from 0 to 25 vol%) in the battery comprising 70:30 LiNbO3-coated Li(Ni0.8Co0.1Mn0.1)O2 and Li5.4(PS4)(S0.4Cl1.0Br0.6) and found that the electronic conductivities of the cathode mixtures increased greatly from 2.6 × 10−8 to 1.1 S/cm with a CA modification (Fig. 2).

Fig. 2
figure 2

Ionic and electronic conductivities of the cathode mixture with increasing amounts of CA

Full size image

The lithium-ion conductivity between 0 and 25 vol% was about 2.9 × 108 times smaller than the electronic conductivity. The small degree of change in ionic conductivity was likely due to the weak influence on Li5.4(PS4)(S0.4Cl1.0Br0.6) conduction path structure [18, 19].

The all-solid-state battery using the Li(Ni0.8Co0.1Mn1.0)O2 and β-Li3PS4 cathode (ionic conductivity: ~ 0.1 mS/cm) mixture showed a high capacity (120 mAh/g) without CA [51,52,53,54]. In contrast, we measured 3.1 mAh/g after 50 cycles in our all-solid-state battery using a high-ion-conductive solid electrolyte, Li5.4(PS4)(S0.4Cl1.0Br0.6) (12 mS/cm), without a CA-modified cathode. Battery capacity may decrease even if the ionic or electronic conductivity of the cathode is high [55, 56], likely because the charge/discharge reaction caused by charge transfer involves both lithium-ion and electron conduction in the electrode [55, 56]. These results emphasize the importance of the electronic conductivity of the cathode mixture when optimizing battery capacity in a system incorporating a high-ion-conductive Li5.4(PS4)(S0.4Cl1.0Br0.6) solid electrolyte. This high ionic conductivity is also attributed to the high discharge capacity at 0.5 C [55, 56]. Compared with the discharge capacity (< 10 mAh/g) of the Li | β-Li3PS4 | Li(Ni0.8Co0.1Mn1.0)O2–β-Li3PS4 battery [51], the all-solid-state battery with 5 vol% CA recorded 137 mAh/g and 134 mAh/g at the 3rd and 49th cycles, respectively (Fig. 1c).

Impedance spectra of Li | Li5.4(PS4)(S0.4Cl1.0Br0.6) | Li(Ni0.8Co0.1Mn0.1)O2–Li5.4(PS4)(S0.4Cl1.0Br0.6)-CA batteries

Impedance spectra are shown in Fig. 3a–d. We previously reported that Li5.4(PS4)(S0.4Cl1.0Br0.6) showed constant resistance at a high frequency due to its high chemical stability. However, we could not separate the bulk and grain boundary due to the measurement frequency region and temperature [18.19], as previously reported [57]. In the present study, we found that Li5.4(PS4)(S0.4Cl1.0Br0.6) was stable even at up to 10 vol% CA in the all-solid-state battery comprising Li | Li5.4(PS4)(S0.4Cl1.0Br0.6) | Li(Ni0.8Co0.1Mn0.1)O2–Li5.4(PS4)(S0.4Cl1.0Br0.6)-CA. Tables S1–S4 show the impedance estimates calculated using an equivalent circuit model (Fig. 4a–d).

Fig. 3
figure 3

Impedance spectra of battery using a 0 vol%, b 5 vol%, c 10 vol%, and d 25 vol% CA, respectively, after 50 cycles

Full size image
Fig. 4
figure 4

Impedance spectra at the 10th cycle using a 0 vol%, b 5 vol%, c 10 vol%, and d 25 vol% CA cathode mixture fitted using the inset equivalent circuit models, as previously reported [19]. The spectra were fitted using Zmeam software [50]

Full size image

In the battery using the highest capacity 5-vol% CA cathode mixture, resistance increased from 53 to 56 Ω during 50 cycles. We previously reported that the impedance of a high-capacity battery increased from 85 to 135 Ω during 50 cycles [19], which was presumed to be due to the formation of amorphous impurities. We also found that Li5.4(PS4)(S0.4Cl1.0Br0.6) was stable against Li metal [18]. Li5.4(PS4)(S0.4Cl1.0Br0.6) is electrochemically stable during cycling because the calculated resistances at high frequency using an equivalent circuit model remained unchanged during 50 cycles (Fig. 4a–d), Tables S1–S4). Previous results indicated that the battery using the 5 vol% CA cathode mixture was highly stable after 50 cycles.

Assuming a homogeneous electrode, Warburg impedance can be measured at low frequencies as a finite length of material diffusion [58, 59]. In all-solid-state batteries, the Warburg coefficient reflects the ease with which lithium-ion diffuses within the electrode [60]. Warburg impedance did not appear when measuring impedance in a Li | SE | Li cell configuration after flowing a current equivalent to 50% of SOC (Fig. S1). When the Warburg coefficients were compared, it was observed that lithium-ion tended to diffuse more easily into positive electrodes incorporating 5 and 10 vol% CA-modified cathodes. In contrast, Li diffusion was limited at 0 and 25 vol% (Tables S1–S4), reflecting the change in capacity. Although Warburg impedance in this study is assumed to be homogeneous in cathodes, actual cathodes are composites of cathode active materials, solid electrolytes, and CA. In the case of a composite structure, macroscopic Li diffusion would be strongly influenced by its morphology [58, 59]. However, we did not analyze morphology-dependent changes in Li diffusion during cycling because the pellets containing CA were too brittle to be observed via scanning electron microscopy (Fig. S2). Therefore, clarifying the relationship between cathode performance, impedance spectra, and cathode morphology will be the next step in manufacturing high-performance composite cathodes.

CA modification was previously reported to increase capacity by increasing cathode active material [Li(Ni0.6Co0.6Mn0.6)O2] utilization rates. However, as the utilization rate increased, deterioration reactions at the β-Li3PS4/CA interface progressed, causing a decrease in capacity during cycling [23]. In this study, the utilization rate of Li(Ni0.8Co0.1Mn0.1)O2 increased by adding CA, capacity decrease during cycling was small, and there was little increase in impedance, indicating limited decomposition reactions at the Li5.4(PS4)(S0.4Cl1.0Br0.6)/CA interface.

Conclusion

This study evaluated the discharge capacity of an all-solid-state battery with a Li | Li5.4(PS4)(S0.4Cl1.0Br0.6) | Li(Ni0.8Co0.1Mn0.1)O2–Li5.4(PS4)(S0.4Cl1.0Br0.6)-CA structure. In conventional all-solid-state batteries, the initial capacity was high with a CA but decreased with cycling. The CA improved the electronic conductivity of the cathode mixture and increased the discharge capacity from 3.1 to 167 mAh/g after 50 cycles. The resistance of the battery increased from 53 to 56 Ω during 50 charge/discharge cycles. These findings demonstrate that the battery capacity of an all-solid-state battery employing a high-ionic conductive Li5.4(PS4)(S0.4Cl1.0Br0.6) can be increased with a CA modification that enhances the electronic conductivity of the cathode. This study presents a proven framework for developing an all-solid-state battery comprising halogen-rich argyrodite (Li7-α(PS4)(S2-αXα); α > 1) with enhanced ionic conductivities by controlling the electronic conductivity of the cathode. We plan to further improve battery performance by elucidating the relationship between the battery performance, actual cathode structure, and impedance during cycling in a future study.