Solid Electrolyte with Oxidation Tolerance Provides a High‐Capacity Li2S‐Based Positive Electrode for All‐Solid‐State Li/S Batteries

The electrochemical window of solid electrolytes (SEs) plays a crucial role in designing active material–SE interfaces in high‐energy‐density all‐solid‐state batteries (ASSBs). However, the suitable electrochemical window for individual active materials is not yet investigated, as the electrochemical window of SEs is overestimated. In this study, the oxidation onset voltages (OOVs) of several SEs, namely those compatible with Li2S as a high‐capacity positive electrode material are determined. Results reveal that SEs with low OOVs decrease the capacity and increase the interfacial resistance of the corresponding ASSBs. The OOVs of SEs must exceed that of Li2S by more than 0.2 V to achieve high capacity, which in turn depends on SE ionic conductivity. Therefore, an Li2S positive electrode is combined with pseudobinary Li‐oxyacid salts as SEs, exhibiting high OOVs and ionic conductivities, to afford a high‐capacity (500 Wh kg−1) ASSB with high Li2S content.


Introduction
All-solid-state batteries (ASSBs) with solid electrolytes (SEs) have attracted considerable attention due to the advantages of their increased safety, high energy, and high power densities. [1][2][3] Although the use of high-energy-density active materials, [4,5] thick electrodes, [6] and thin SE layers [7] has allowed the realization of ASSBs with high gravimetric energy densities, these values are still lower than those of Li-ion batteries (maximum 300 Wh kg −1 ). [8] This behavior can be partly ascribed to the high resistance of the sulfide SE-electrode material interfaces, [9,10] which is due to side reactions [11,12] and the loss of interfacial contacts. [13][14][15] Furthermore, electrochemical material. In Li 2 S-LiI-C composite itself as the positive electrode, the ionic conductive resistance across positive electrode would have the large impact on the capacity of Li 2 S-LiI because the ionic conductivity of Li 2 S-LiI was quite low (<5 × 10 −6 S cm −1 ). [44] For that reason, sulfide SE was necessary as the ionic conduction path across the positive electrode. Thus, we designed the electrode structure dividing ionically nanoscale (Li 2 S-LiI-C) and microscale (sulfide SE) network, which was prepared by hand mixing with Li 2 S-LiI-C and sulfide SEs (Li 2 S-LiI-C + SE) to minimize the capacity contribution of sulfide SE. Surprisingly, the Li 2 S-LiI-C + SE electrode have demonstrated a full utilization of Li 2 S. [43] We have interpreted two points from the results. One is that SEs even with low ionic conductivity functions as the ionically nanoscale network, which is enabled by short ionic diffusion distance and Li chemical diffusion promoted by jobsharing mechanism in composite of SEs and C. [45,46] Another one is that sulfide SEs would be not enough as the ionically nanoscale conduction network due to the decline of ionic conductivity on the charging process; sulfide SE functions as the active material in the interface between sulphide SE and C from the charging process. Actually, the ion conductive resistance of ASSBs with Li 6 PS 5 Cl-C composite electrode have been investigated by impedance measurements. [47] During charging process, the oxidation of Li 6 PS 5 Cl slows down lithium ion transport and increases an overpotential of sulfur-based positive electrodes. [47] Furthermore, the superior property of LiI to sulfide SE is only higher electrochemical oxidation stability. [20] The electrochemical oxidation of sulfide SEs lowering the lithium ion transport indicates that SEs should have the proper electrochemical oxidation stability even though 2 V-class sulfur-based electrodes.
In this study, we probed the relationship between the oxidation stability of SEs and the discharge capacity of ASSBs with Li 2 S positive electrodes. Here, the oxidation stabilities of several Li salts were examined to verify their applicability as the SE, and the contribution of the capacity of sulfide Li 3 PS 4 (LPS) was considered as negligible. In lithium ion batteries, the oxidation stability of liquid based electrolytes has been determined by the onset oxidation voltage (OOV) in linear sweep voltammetry (LSV) to discuss an initial electrolyte electrochemical decomposition reactions and factors controlling them. [48,49] In the same manner, the SE oxidation stability was characterized for SE-C nanocomposites by the OOV in LSV, while the ASSB discharge capacity was measured using a positive electrode comprising Li 2 S and SE-C composites. The corresponding flow diagram is shown in Scheme 1.
We have found that the experimentally determined OOVs, used as an indicator of SE oxidation stability, agree well with those previously predicted by first-principles calculations. [19,20] The capacities of ASSBs with Li 2 S positive electrodes significantly increase up to 1000 mAh g −1 when the employed SEs have OOVs of >2.6 V versus Li, which is 0.2 V higher than that Scheme 1. Experimental procedure used in the present study. Li 2 S capacities were characterized for all-solid-state batteries (ASSBs) with positive electrodes comprising Li 2 S-Li-salt-C composites and Li 3 PS 4 (LPS). Oxidation stabilities were characterized by linear sweep voltammetry (LSV) of allsolid-state cells (ASSCs) with working electrodes comprising Li-salt-C composites and LPS. Li 2 S capacity was correlated with the oxidation stability of the Li salts in the Li 2 S-Li-salt-C composites.
www.afm-journal.de www.advancedsciencenews.com of Li 2 S. Furthermore, the discharge capacity of Li 2 S using SEs with high OOVs depends on the ionic conductivity of the SEs. The developed, high-Li 2 S-content positive electrodes using pseudobinary Li-oxyacid salts with critical OOVs and conductivities exhibit high capacities, which are comparable to the theoretical value of the Li 2 S-based positive electrode (500 Wh kg −1 ).

Results and Discussion
The commonly used Li 2 S positive electrodes comprise Li 2 S, carbon, and an SE prepared by ball-milling to form electronically and ionically conductive nanoscale networks, thus achieving high reversible capacity. [50][51][52] Therefore, the capacity per gram of Li 2 S (mAh g Li2S ) is significantly affected by the electrochemical properties of SEs in the ionically conductive nanoscale networks of the positive electrodes. Most studies have employed sulfide SEs as nanoscale, ionically conductive networks (Figure 1a). [50][51][52] When a common Li 2 S positive electrode comprising Li 2 S, C, and LPS as the sulfide SE was prepared by ball-milling, the corresponding ASSB (Li-In/LPS/ Li 2 S-LPS-C) exhibited a capacity (1600 mAh g −1 ) above the theoretical capacity of Li 2 S (Figure 1b). This indicated that LPS acted as an active material, which is in line with the results detailed by previous reports. [41,42] Therefore, given the difficulty of determining the capacity of an Li 2 S positive electrode in the presence of sulfide SEs, non-sulfide SEs were employed in the current study. As first-principles calculations of SE electrochemical windows suggest that SE oxidation voltage depends on the SE anion type, [19,20] we used Li salts with different anions (LiF, LiCl, LiBr, LiI, Li 3 BO 3 , Li 2 CO 3 , Li 3 PO 4 , Li 2 SO 4 , LiBH 4 , Li 3 N, and Li 3 P) as the nanoscale, ionically conductive networks of the positive electrodes. To prepare Li 2 S-Li-salt-C composites, we mixed Li 2 S with one of the Li salts and C by ball-milling. Since the ionic conductivities of the Li salts were much lower than that of the sulfide SE, the ionic conduction path across the positive electrode layers could significantly impact Li 2 S capacity. [6,53] To mitigate this effect, we employed sulfide SEs to obtain an ion-conduction path across the electrode layer (i.e., the so-called microscale, ionically conductive network). To exclude the capacity contribution of sulfide SEs, their area of contact with C was minimized as much as possible. [54,55] In SEM images and EDX mappings for mixture of sulfide SE and C prepared by hand mixing and ball-milling, sulfide SE was mixed with C in microscale and nanoscale, respectively. [41] Mixing method have an impact on their contact area. Figure 1c shows the charge-discharge curves of ASSBs that were fabricated from a mixture (LPS + C), prepared by the hand-mixing of LPS and C, and a composite (LPS-C), prepared by the ballmilling of LPS and C. In ASSBs (Li-In/LPS/LPS-C) with the LPS-C composite, LPS acted as an active material, whereas ASSBs (Li-In/LPS/LPS + C) with the LPS + C mixture exhibited www.afm-journal.de www.advancedsciencenews.com a capacity of only 5 mAh g −1 . Thus, hand-mixing successfully minimized the capacity contribution of LPS, which allowed LPS to be employed as a microscale, ionically conductive network for the positive electrodes comprising Li 2 S-Li-salt-C composites. Figure 1d illustrates the structure of the positive electrode with nano-and microscale ionically conductive networks, while Figure 1e shows cross-sectional scanning electron microscopy (SEM) images and energy-dispersive X-ray spectroscopy (EDX) mappings of the positive electrode layer comprising an Li 2 S-LiI-C composite and LPS. In the Li 2 S-LiI-C composite, Li 2 S, LiI, and C were uniformly dispersed, but the composite as a whole was clearly separated from LPS. This allowed the capacity contribution of LPS to be neglected.
The presence and dispersion of Li salts in the Li 2 S-Li-salt-C composite were probed by X-ray diffraction (XRD), SEM-EDX, and X-ray photoelectron spectroscopy (XPS). Figure S1, Supporting Information shows the XRD patterns of Li 2 S-Li-salt-C recorded in the presence of Si as an internal standard. The pattern of Li 2 S-LiF-C featured peaks corresponding to the cubic structure of Li 2 S and LiF ( Figure S1a, Supporting Information), while the patterns of all other composites showed no peaks attributable to Li salts and C. No diffraction peaks were observed for Li 2 S-Li 3 N-C, which indicated that all components were present in an amorphous state ( Figure S1c, Supporting Information). In the patterns of all the Li 2 S-Li-salt-C composites, except Li 2 S-LiF-C and Li 2 S-Li 3 N-C, all diffraction peaks could be indexed to the cubic structure of Li 2 S. Figures S2-S4, Supporting Information show SEM-EDX images of Li 2 S-Lisalt-C composites, demonstrating that all elements (and hence, the starting materials) were homogeneously mixed and that nanoscale electronically and ionically conductive networks were formed upon ball-milling. Figure S5, Supporting Information shows the particle sizes and particle size distributions of all composites. Li 2 S-Li-halide-C composites contained particles with sizes on the scale of sub-micrometer to several micrometers, Li 2 S-Li-oxyacid-C composites contained particles several micrometers in size, and Li 2 S-Li 3 N-C composites contained particles several to ten micrometers in size. Finally, Li 2 S-LiBH 4 -C composites contained particles with sizes of 10-30 µm. Li 2 S-Li-salt-C composites were evaluated by XPS to probe side reactions occurring between the starting materials ( Figures S6-S8, Supporting Information). Ball-milling did not significantly affect the F 1s, Cl 2p, Br 3p, and I 3d spectra of the composites containing LiF, LiCl, LiBr, and LiI, respectively ( Figure S6, Supporting Information), nor the B 1s, C 1s, P 2p, and S 2p spectra of the composites comprising Li 3 BO 3 , Li 2 CO 3 , Li 3 PO 4 , and Li 2 SO 4 , respectively ( Figure S7, Supporting Information). This indicates that the corresponding Li-halide-and Li-oxyacid-containing composites were present. Figure S8, Supporting Information shows the XPS spectra of Li 2 S-LiBH 4 -C, Li 2 S-Li 3 N-C, and Li 2 S-Li 3 P-C. The B 1s spectrum of Li 2 S-LiBH 4 -C featured peaks at 68.5 and 70.5 eV attributable to LiBH 4 and Li 3 BO 3 , respectively, indicating that LiBH 4 was partially oxidized during ball-milling. The N 2p spectrum of Li 2 S-Li 3 N-C featured peaks at 396 and 398 eV, which were higher binding energies than those of Li 3 N, indicating the oxidation of this compound during ball-milling. However, these peaks were below 400 eV and were therefore ascribed to anionic N. The peak of P2p 3/2 at 130 and 132 eV attributable to P 0 [56,57] and oxygen-phosphorus moieties [58][59][60] are minor component and the most of peak were located at the lower binding energy compared to 130 eV. In the references, [61][62][63] the binding energy of P2p 3/2 for Li 3 P was located around from 126 to 130 eV. The P 2p spectrum of Li 2 S-Li 3 P-C featured peaks of P 3− and P − , indicating the partial oxidation of Li 3 P. . Notably, the last three values of those listed above were lower than the discharge capacities of ASSBs with Li 2 S-Li-halide-C and Li 2 S-Li-oxyacid-C composites. Furthermore, ASSBs with Li 2 S-LiBH 4 -C, Li 2 S-Li 3 N-C, and Li 2 S-Li 3 P-C had large irreversible capacities. In Li 2 S-Li 3 P-C, Li 3 P acted as an active material ( Figure S9, Supporting Information), and the Li 2 S capacity in this case was calculated by subtracting the capacity of Li 3 P from the overall capacity (for a more detailed discussion, see the Supporting Information).
The OOVs of the SEs were determined by the LSV characterization of all-solid-state cells (ASSCs) containing Li 2 S-free composites, which were prepared by the ball-milling of each Li salt and C. The working electrodes were then prepared by the handmixing of LPS and the Li-salt-C composites. A schematic illustration of these ASSCs (Li-In/LPS/Li-salt-C + LPS) with the working electrodes used to measure the OOVs of the Li salts is shown in Figure 2d, while the LSV curves of the ASSCs (Li-In/ LPS/Li-salt-C + LPS) with the Li-salt-C composites are shown in Figure S10, Supporting Information. The intersection of the base line extrapolation with the curve showing the maximum peak current was defined as the OOV. ASSCs (Li-In/LPS/Lisalt-C + LPS) with Li-salt-C composites, except LiF-C, showed oxidation currents. The OOVs determined from the LSV curves and first-principles calculations are summarized in Table S1, Supporting Information and Figure 2e. In the case of the composites comprising Li halides, LiBH 4 , and Li 3 P, the measured OOVs were almost consistent with the calculated oxidation voltages. However, the measured OOVs for the Li-oxyacid-and Li 3 N-containing composites were more than 1 V, lower and higher, respectively, than the calculated oxidation voltages. [19,20] A more detailed discussion is presented in the Supporting Information Figure 3a shows the relationship between the discharge capacities of the ASSBs with Li 2 S-Li-salt-C composites and the experimentally measured OOVs of the Li salts. The discharge capacity of ASSBs (Li-In/LPS/Li 2 S-LiF-C + LPS) with Li 2 S-LiF-C was ruled out in the relationship because the OOV and ionic conductivity of LiF could not be measured. However, it should be noted the effect of LiF as ionic conductive path on the capacity of Li 2 S. LiF is capable of improving capacity and cycle life as SE interphase (SEI) for lithium metal and high voltage positive electrodes due to wide electrochemical stability. [64,65] On the other hand, LiF accelerates an aggregation of organic SEI component and plays the role of glue in the SEI film. [66] The interface between LiF and Li 2 CO 3 promotes Li + ion defects and works as a high conductive SEI. [67] Therefore, the www.afm-journal.de www.advancedsciencenews.com ionic conduction of LiF only would not be enough in SEI. Actually, the capacity of ASSBs (Li-In/LPS/Li 2 S-LiF-C + LPS) with Li 2 S-LiF-C is the lowest in the Li 2 S-Li-halide-C and LiF only is not effective in the ion conduction path for Li 2 S active material. The discharge capacities of the ASSBs (Li-In/LPS/Li 2 S-Li-salt-C + LPS) with Li 2 S-Li-salt-C composites were under 200 mAh g −1 when the Li salts had OOVs under 2.5 V versus Li, Interestingly, LiBH 4 caused the low discharge capacity even though OOV of LiBH 4 was slightly higher than that of Li 2 S. On the other hand, the discharge capacities of the same ASSBs exceeded 600 mAh g −1 when the Li salts had OOVs of >2.6 V versus Li, indicating that the Li salts acted as nanoscale, ionic conductive networks in these cases.
Although the OOVs of LiBH 4 and LiI only differed by 0.1 V, the discharge capacities of the ASSBs (Li-In/LPS/Li 2 S-LiBH 4 -C + LPS or Li 2 S-LiI-C + LPS) with Li 2 S-LiBH 4 -C and Li 2 S-LiI-C were significantly different. To clarify the origin of this behavior, we compared the Nyquist plots of these two ASSBs (Li-In/LPS/Li 2 S-LiBH 4 -C + LPS or Li 2 S-LiI-C + LPS) after full charging (Figure 3b). In the case of the ASSB (Li-In/LPS/Li 2 S-LiBH 4 -C + LPS) containing Li 2 S-LiBH 4 -C, we observed a large semi-circle representing interfacial resistance. In the charging process, it is predicted that LiBH 4 is oxidized to an Li-poor oxidant Li 2 B 6 H 12 , [20] which would be formed at the interface between Li-salt and C. The formed Li-poor compounds did not act as nanoscale, ionically conductive networks. As the results, the interface area between Li 2 S and LiBH 4 was decreased, which increases the interfacial resistance ( Figure 3e). On the other hand, the ASSB (Li-In/LPS/Li 2 S-LiI-C + LPS) with Li 2 S-LiI-C had a smaller interfacial resistance than that of the ASSB (Li-In/LPS/Li 2 S-LiBH 4 -C + LPS) with Li 2 S-LiBH 4 -C. The cut off voltage of the charging process was 3.6 V versus Li and LiI (Li-salts with OOVs of > 2.6 V vs Li) would be oxidized as well as LiBH 4 . In coating materials for high voltage active materials, even though the oxidation voltage of coating materials (Li 2 CO 3 , [ [20] and LiZr 2 (PO 4 ) 3 [70] ) is lower than the cut off voltage of ASSBs with high voltage active materials, the coating materials have significantly decreased the interfacial resistance. [71][72][73][74][75] Therefore, those coating materials would be also oxidized on the high voltage active materials but they still work as the Li ion conductive material. Based on the study about coating materials for high voltage electrode and the obtained Nyquist plots (Figure 3b), we speculate that the oxidation of LiI was limited and the interface area between Li 2 S and LiI would be maintained ( Figure 3e).Furthermore, when the Li salts exhibited OOVs of >2.6 V versus Li, the discharge capacities of ASSBs with Li 2 S-Li-salt-C composites decreased with increasing OOV. The  [19,20] with the oxidation onset voltages (OOVs) determined from the LSV curves of ASSCs (Li-In/LPS/Li-salt-C + SE) with Li-salt-C composites. www.advancedsciencenews.com discharge capacities of ASSBs (Li-In/LPS/Li 2 S-Li-halide-C + LPS) with Li 2 S-Li-halide-C composites decreased depending on the halide in the order of LiI > LiBr > LiCl. However, the OOVs of these salts decreased in the opposite order. Figure 3c shows the impedance spectra of ASSBs (Li-In/LPS/Li 2 S-Li-halide-C + LPS) with Li 2 S-Li-halide-C after charging to 600 mAh g −1 , that is, to a state-of-charge (SOC) value of 50%. Straight lines with phase angles of 45° were observed in the Nyquist plots, which increased to angles of ≈90° with a decrease in frequency. This is in accordance with the finite-length diffusion model. [76] The impedance (Z) of finite diffusion in an electrode composed of spherical particles is defined in Equation (1), wherein variables C and R d are calculated using Equations (2) and (3), respectively. [77] where R ct is the charge transfer resistance, ω is the angular frequency, E is the open-circuit voltage (OCV), c s is the concentration of the electroactive species in the active electrode material, r is the diffusion length, D is the diffusion coefficient of the electroactive species, and F is the faraday constant. Additionally, R d in Equation (3) is the ratio of applied voltage to the flux of an electroactive species in the active electrode material; thus, we assumed that R d in our study represents the diffusion resistance of the Li flux in the Li 2 S-Li-halide-C composite (R d Li2S-Li-halide-C ). The semi-circle in the plot attributable to R ct was not observed. Therefore, the capacities of the ASSBs with Li 2 S-Li-salt-C depend on the diffusion resistance of the Li flux in the Li 2 S-Li-salt-C composite when the Li salts had OOVs of >2.6 V versus Li. The intersection points of the tangent of the slope with the phase angles of 45° and ≈90° are defined as transitions from semi-infinite diffusion to finite diffusion in active materials. [78] Furthermore, the real axis of the intersection was determined to be represented by R 1 5 d (Figure S11, Supporting Information) in Equation (1). The values of R d Li2S-LiCl-C , R d Li2S-LiBr-C , and R d Li2S-LiI-C were 2250, 1670, and 675 Ω cm 2 , respectively. Figure 3d shows that the discharge capacities of the ASSBs (Li-In/LPS/Li 2 S-Li-halide-C + LPS) with the Li 2 S-Li-halide-C composites increased with decreasing diffusion resistance of the Li flux in the composites. Therefore, the different capacities of these ASSBs likely originate from the varying Li fluxes in the Li 2 S-Li-halide-C composites (Figure 3e).
For the Li fluxes in the Li 2 S-Li-halide-C composites, the Li halides would work as the nanoscale, ionically conductive network rather than Li 2 S during charging. We focused on the ionic conductivities of the Li salts themselves, which were measured from the alternating current (AC) impedance measurements of their compressed powder pellets (Table S1, Figure S12, Supporting Information). Figure 3d shows that the discharge capacities of the ASSBs (Li-In/LPS/Li 2 S-Li-halide-C + LPS) with Li 2 S-Li-halide-C The development of broadly applicable energy storage devices with energy densities of ≈500 Wh kg −1 is an important research target. [9] In particular, calculations assuming a 20-µm-thick Li negative electrode, 20-µm-thick SE layer, and 10-µm-thick current collectors (Equation (S12), Table S2, Figure S13, Supporting Information) suggest that ASSBs with an energy density of 500 Wh kg −1 require an Li 2 S content of >40 wt% and a capacity of >1000 mAh g Li2S ; however, the Li 2 S content reported in most papers is 30-40 wt% (Table S3, Supporting Information). Herein, positive electrodes with an Li 2 S content of 40 wt% were prepared using LiI as a nanoscale, ionically conductive network (Li 2 S:LiI:C:LPS = 40:10:10:40, w/w/w/w). This Li halide was chosen because the Li 2 S-LiI-C positive electrode (Li 2 S:LiI:C:LPS = 30:20:10:40, w/w/w/w) exhibited the highest capacity ( Figure 3a). Figure S14, Supporting Information presents the charge-discharge curves of the ASSBs (Li-In/LPS/Li 2 S-LiI-C + LPS, 30 and 40 wt% Li 2 S) with 30 and 40 wt% Li 2 S. The battery capacity in latter case was much lower than that in the former case. The volume ratio of LiI (LiI/Li 2 S + LiI + C) in Li 2 S-LiI-C was then decreased from 18 vol% (Li 2 S:LiI:C = 30:20:10, w/w/w) to 8 vol% (Li 2 S:LiI:C = 40:10:10, w/w/w) based on the densities of Li 2 S (1.66 g cm −3 ), LiI (4 g cm −3 ), and carbon (2 g cm −3 ). As ionically conductive LiI has a relatively high density, this decrease in its volume within the composite decreased the capacity of the ASSB (Li-In/LPS/Li 2 S-LiI-C + LPS, 40 wt% Li 2 S) with 40 wt% Li 2 S. Therefore, we concluded that SEs must have a low density as well as critical OOV and ionic conductivity values.
Li 3 PO 4 exhibits a suitable OOV (2.6 V vs Li) and a relatively low mass density (2.5 g cm −3 ) but features a suboptimal ionic conductivity (Table S1, Supporting Information). Therefore, if the ionic conductivity of this salt is improved, it can be used as a nanoscale, ionically conductive network to increase the Li 2 S content, while maintaining its high reversible capacity. To improve the ionic conductivity of Li 3 PO 4 , there are several strategy such as solid solution with aliovalent element substitution/doping at both cationic and anionic sites, [79,80] interface engineering due to the addition of insulating fine particles, [81,82] and composite ceramic/polymer electrolytes. [83] Here, we focused on the use of the mixed anion effect to improve the conductivity of Li-ion-conducting glasses in pseudobinary systems, [84] because the XRD pattern of Li 2 S-Li 3 PO 4 -C revealed the amorphous nature of its Li 3 PO 4 component ( Figure S1b, Supporting Information). There are possible combinations and compositions of the pseudobinary systems. For the present study, an Li 3 PO 4 -Li 2 SO 4 pseudobinary system, prepared by the ball-milling of Li 3 PO 4 and Li 2 SO 4 , was investigated because Li 2 SO 4 has higher conductivity and OOV values than those of Li 3 PO 4 . Furthermore, Li 3 PO 4 -Li 2 SO 4 glass (50:50 mol%) could be prepared by rapid quenching. [85] The conductivity of this system was obtained from the AC impedance of the corresponding compressed powder pellet. Figure 4a shows the Nyquist plots of Li 3 PO 4 -Li 2 SO 4 , pure Li 3 PO 4 , and pure Li 2 SO 4 . The semi-circle and spike observed in the Nyquist plot of Li 3 PO 4 -Li 2 SO 4 indicated that this system exhibited an ionic conductivity of 10 −8 S cm −1 , a value greater than those of pure Li 3 PO 4 and Li 2 SO 4 . The improved ionic conductivity has been seen in many mixed glass former glasses although the mechanism has not been fully explained due to a nonlinear and nonadditive change in the ionic conductivity. In Li 3 BO 3 -Li 2 SO 4 glass system, the elastic moduli are decreased with increasing the Li 2 SO 4 content, which enables the densification of the pellet by cold pressing. [86] Thus, one of the reasons in the improved ionic conductivity would be a reduced grain boundary resistance in Li 3 PO 4 -Li 2 SO 4 .
An Li 3 PO 4 -Li 2 SO 4 -C composite was then prepared by ballmilling to measure the OOV of Li 3 PO 4 -Li 2 SO 4 . Figure 4b   , respectively. Notably, ASSBs with Li 2 S-Li 3 PO 4 -Li 2 SO 4 -C exhibited high capacity even at a high Li 2 S content. Li 3 PO 4 -Li 2 SO 4 functions as the nanoscale, ionically conductive network even though the ionic conductivity of Li 3 PO 4 -Li 2 SO 4 was 10 −8 S cm −1 which was much lower than that of LPS. C.-C. Chen et al., have found the ultrafast Ag diffusion in the composite of RbAg 4 I 5 and graphite. [45] At an interface of RbAg 4 I 5 and graphite, Ag + ions and electrons conduct in RbAg 4 I 5 and graphite, which increases the diffusion coefficient by about 10 000 times of RbAg 4 I 5 : Job-sharing chemical diffusion. [45] In our study, the Li chemical diffusion in Li 2 S-Li 3 PO 4 -Li 2 SO 4 -C would be enabled by the job-sharing chemical diffusion, in which Li + ions and electrons conduct in Li 3 PO 4 -Li 2 SO 4 and C.
For an ASSB (Li-In/LPS/Li 2 S-Li-salt-C + LPS) with an Li 2 Sbased positive electrode to achieve an energy density of 500 Wh kg −1 , it should exhibit a reversible capacity of 1167 mAh g Li2S with an Li 2 S content of 40 wt% or a reversible capacity of 1000 mAh g Li2S −1 with an Li 2 S content of 50 wt% ( Figure S13, Supporting Information). These values are based on calculations assuming a certain anode, SE layer, and current collector (Equation (S12), Table S2, Supporting Information). Figure S15 , respectively, that is, they increased with increasing temperature because of the concomitant decrease in internal resistance. Figure S15b, Supporting Information shows the charge-discharge curves of the ASSBs (Li-In/LPS/Li 2 S-Li 3 PO 4 -Li 2 SO 4 -C + LPS, 50 wt% Li 2 S, 2, 5, or 10 mg cm −2 , 60 °C) containing Li 2 S-Li 3 PO 4 -Li 2 SO 4 -C with 50 wt% of Li 2 S at Li 2 S loadings of 2, 5, and 10 mg cm −2 , revealing that the respective discharge capacities equaled 750, 800, and 500 mAh g Li2S . Notably, the discharge capacity at 10 mg cm −2 was lower than those at 2 and 5 mg cm −2 , which indicated that the microscale, ionically conductive network across the positive electrode affected the charge-discharge capacities of this electrode at high loadings. [6,50] To mitigate the effect of the microscale, ionically conductive network on charge-discharge capacity, a fine sulfide SE Li-P-S-I with the diameter of 2-3 µm was used in the positive electrode instead of a bulky sulfide SE. Figure S15c, Supporting  Table S4, Supporting Information and presented in Figure 5f. For ease of comparison, the previously obtained capacities of the Li 2 S-based positive electrode were standardized with respect to positive electrode mass, and the capacities of ASSBs with Li 2 S-based positive electrodes (both areal and per gram of positive electrode) were also included (Table S3, Supporting Information). Figure 5f displays the predicted energy densities of ASSBs based on the capacity per gram of positive electrode and the areal capacity calculated by Equation (4) (assuming an SE layer and current collectors) (Table S2, Supporting Information). The ASSBs with Li 2 S and Li 2 S-LiI nanocomposites exhibited high gravimetric and areal capacities, reaching the predicted energy density of 300-400 Wh kg −1 . However, these gravimetric and areal capacities were still insufficient to reach an energy density of 400-500 Wh kg −1 . On the other hand, the Li 2 S positive electrode containing Li 3 PO 4 -Li 2 SO 4 could potentially be used to fabricate ASSBs with an energy density of 400-500 Wh kg −1 . Thus, SEs with critical OOVs, high ionic conductivities, and low mass densities were www.afm-journal.de www.advancedsciencenews.com found to be suitable for the construction of high-Li 2 S-content positive electrodes with high reversible capacity.

Conclusion
We successfully demonstrated the use of SEs with critical oxidation stability to increase the capacity of Li 2 S as the active material of ASSBs (Li-In/LPS/Li 2 S-Li-salt-C + SE). The capacities of ASSBs with an Li 2 S positive electrode significantly increased when the employed SEs had OOVs of >2.6 V versus Li, in which case the capacities were limited by the ionic conductivities of these SEs. Therefore, it was concluded that proper SEs should exhibit an OOV of >2.6 V versus Li and have the maximum possible ionic conductivity to act as nanoscale, ionically conductive networks in Li 2 S positive electrodes. This obtained insight was used to fabricate a pseudobinary Li-oxyacid SE with the critical OOV (> 2.6 V vs Li) and ionic conductivity of 10 −8 S cm −1 , which allowed us to achieve a high capacity of >1000 mAh g Li2S −1 with a high Li 2 S content (40 to 50 wt%) in the positive electrode. In addition, the corresponding positive electrode could enable the design of ASSBs with an energy density of 400-500 Wh kg −1 . The adopted strategy will be also suitable for the development of both 2-V-class, sulfur-based positive electrode materials and high-voltage positive electrode materials. Furthermore, these results are expected to aid the design of artificial SE interfaces on electrode active materials and deepen our understanding of the role of these interfaces in Li-ion batteries. ASSBs with High-Energy Li 2

S Positive Electrodes: Fabrication and Electrochemical
Characterization: Li 2 S-Li 3 PO 4 -Li 2 SO 4 -C positive electrodes with high Li 2 S contents were prepared as shown above using Li 2 S:Li 3 PO 4 :Li 2 SO 4 :C:SE weight ratios of 40:5:5:10:40 and 50:6.25:6.25:12.5:25. In Li 2 S-Li 3 PO 4 -Li 2 SO 4 -C positive electrodes with 50 wt% Li 2 S, the tested Li 2 S loadings equaled 2, 5, and 10 mg cm −2 (corresponding to electrode loadings of 4, 10, and 20 mg cm −2 , respectively). For an Li 2 S loading of 10 mg cm −2 , a fine sulfide SE powder was used to mitigate the ion conductivity resistance in the positive electrodes. ASSBs with an Li-In/Li 2 S positive electrode were fabricated as described above. Charge-discharge tests of constantly tightened cells were conducted at a current density of 1.17 mA cm −2 (0.1 C, 1 C = 1167 mA g −1 ) and temperatures of 25 and 60 °C under Ar using a charge-discharge device (BTS-2004, Nagano Co.).
XRD Measurements: XRD patterns were recorded using a diffractometer (SmartLab, Rigaku Corp.) equipped with a Cu K α radiation source and a detector (Dtex2000) and operated at 45 kV and 200 mA.
Field Emission-Scanning Electron Microscopy and EDX Measurements: Sample morphology was probed by field emission-scanning electron microscopy (FE-SEM; SU8220; Hitachi High-Technologies Corp.). The distribution of elements in the particles was investigated by EDX (EMAX Evolution; Horiba Ltd.) at an acceleration voltage of 15 kV and room temperature. The dispersion of Li 2 S-Li-salt-C and LPS in the positive electrodes was probed by SEM imaging of the cross sections of the corresponding ASSBs. The ASSB pellets were cut, and the cross sections were polished using an Ar ion milling system (IM4000; Hitachi High-Technologies Corp.) and then observed by FE-SEM.
XPS Measurements: XPS measurements were conducted using a K-Alpha instrument (Thermo Fisher Scientific Inc.) with a monochromatic Al K α source (1486.6 eV) to probe the side reactions within the Li 2 S positive electrode. The observed binding energies were calibrated with respect to the adventitious C 1s peak. Ar + ion-etching was done to remove the surface impurities. After elimination of the surface impurities, etching procedure was continued to obtain bulk information.
To suppress the damages of the samples caused by Ar + ion-etching, Ar clusters were emitted from Ar + ion gun during the etching process. To obtain the spectra, the pass energy and scan number were set as 50 eV and 50 times. [87] Conductivity Measurements: Ionic conductivities for Li salts pressed at 370 MPa were determined using AC impedance measurements (SI-1260; Solartron Analytical) at frequencies in the range from 100 kHz to 0.1 Hz and an applied voltage of 100 mV. The pellet diameter and thickness equaled 10 and ≈0.5 mm, respectively.
The all samples were sensitive with atmosphere and the all experiments were conducted in a dry Ar-filled glove box and transferred to a test chamber using an Ar-filled transfer vessel.
Calculation of Capacity per Gram of Positive Electrode and Areal Capacity for ASSBs with Energy Densities of 300, 400, and 500 Wh kg −1 : The required capacity per gram of positive electrode (C positive electrode /mAh g −1 ) was calculated from the average discharge voltage (V Ave /V), areal capacity (C Area /mAh cm −2 ), total weight of the SE layer, Li metal, current collectors (w/g cm −2 ), and the target energy density per gram of cell (E Cell /Wh kg −1 ), as shown below.
The derivation of this equation and the total weight are shown and listed in Equations (S1)-(S12) and Table S2, Supporting Information. The average discharge voltage was determined as 1.8 V versus Li from the discharge curves of ASSBs with an Li 2 S-based positive electrode, whereas the areal capacity was set as a variable.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.