The thio-LISICON family [45–50] has the framework of a γ-Li₃PO₄ structure (Fig. 1a) with compositional formula of Li_xM_1-yM′_yS₄ (M = Si, Ge, and M′= P, Al, Zn, Ga, and Sb) of which ionic conductivity ranges from 10⁻⁷ to 10⁻³ S cm⁻¹ at room temperature. The ionic transport properties of thio-LISICON are largely dependent on the size and polarizability of the ions forming the structure as well as the interstitial and/or vacancy defect properties produced by substitution [49]. In order to enhance ionic conductivity, it is desirable to replace oxygen ions with larger sulfur ions that can offer more polarizable framework. One of the important characteristics of this class of materials is that they can contain interstitial lithium ions or form a wide variety of solid solutions by aliovalent substitution, which introduces structural vacancies, enhancing the ionic conductivity. For example, Li_4–xGe_1–xP_xS₄ is a representative composition among various thio-LISICON-based electrolytes developed by aliovalent substitution, e.g., Ge⁴⁺ + Li⁺ ↔ P⁵⁺, in the Li₂S-GeS₂-P₂S₅ system. Li_4–xGe_1–xP_xS₄ is reportedly crystalized into three different types of monoclinic superstructures depending on the concentration of P (phase I: x ≤ ~0.6; phase II: ~0.6 ≤ x ≤ ~0.8; phase-III: x ≤ ~0.8) and shows the highest ionic conductivity of 2.2×10⁻³ S cm⁻¹ at room temperature from phase-III compound [47].

Thereafter, Homma et al. reported a temperature-dependent structural change in a binary Li₂S-P₂S₅ system that did not contain the metal elements such as Ge or Si [45]. This system was highlighted for their greatly improved electrochemical stability window and the high chemical stability upon exposure to the atmosphere compared to other sulfide solid electrolytes. One of the representative glass sulfide, Li₃PS₄ (75Li₂S·25P₂S₅ (mol%)), possesses a γ-Li_3-PO₄-type structure with a hexagonal close-packed sulfide ion array, phosphorus ions distributed in tetrahedral sites, and PS₄ tetrahedra separated from one another (Fig. 1b) [52]. The structure of γ-Li₃PO₄ synthesized at ~470 K exhibits a transition to β-phase at 573 K and to α-phase at 746 K [45]. The structure of the β-phase that evolved from the γ-phase is shown in the Fig. 1b, where the PS₄ tetrahedra of the β-phase structure are separated; has a zigzag arrangement; and are edge-shared with the LiS₆ octahedron. The LiS₆ octahedra are connected through edge-sharing to form one-dimensional (LiS₆) chains [45,53]. Among these three phases, β-phase shows the highest ionic conductivity of 3.0×10⁻² S cm⁻¹ (at 227°C) due to its unique crystal structure.

In glass ceramic, superionic crystalline phases with conductivities greater than 10⁻³ S cm⁻¹ was produced by heating binary Li₂S-P₂S₅ above the glass transition temperature (T_g) [54–58]. Above T_g, the high-temperature phase or metastable phase tends to be crystallized as a primary phase from the super-cooled liquid [59]. The Li₂S-P₂S₅, Li₇P₃S₁₁ or Li_3.25P_0.95S₄ systems tend to be synthesized with a composition of 70 to 80 mol% Li₂S, and exhibit high ionic conductivity of >10⁻³ S cm⁻¹ [54,56]. Li₇P₃S₁₁ prepared at a low annealing temperature of 280°C exhibited high conductivity with a crystal structure as shown in Fig. 2. The ionic conductivities of the cold-pressed and hot-pressed samples were 3.2×10⁻³ and 1.7×10⁻² S cm⁻¹, respectively, and the activation energy was estimated to be 12 kJ mol⁻¹ [26,54].

Numerous studies have employed Raman spectroscopy to understand the structural feature of glass-ceramic. Particularly, the local structure of highly conductive Li₂S-P₂S₅ glass-ceramic was investigated [56], where P₂S₇⁴⁻ and PS₄³⁻ polyanions play a crucial role in enhancing the conductivity of materials based on Li₂S-P₂S₅ [56,57,60]. In the (Li₂S)_x(P₂S₅)_(100–x) tie line, the 75Li₂S-25P₂S₅ glass-ceramic composition and PS₄³⁻ unit exhibited the lowest activation energy and corresponding highest ionic conductivity [61]. The study on the structure of the crystalline phase in glass-ceramic showed that the Li₇P₃S₁₁ glass-ceramics (70Li₂S-30P₂S₅) had a triclinic centrosymmetric space group P-1 structure, as in Fig. 2.

Li₁₀GeP₂S₁₂ was first reported in 2011 as a superionic conductor having three-dimensional framework structure, and thus far, the compounds Li_9.54Si_1.74 P_1.44S_11.7Cl_0.3, Li₁₀Si_0.3Sn_0.7P₂S₁₂, Li₁₀SnP₂S₁₂, Li_10.35Sn_0.27Si_1.08P_1.65S₁₂, Li_9.42Si_1.02P_2.1S_9.96O_2.04, and Li₁₀Ge(P_0.925Sb_0.075)₂S₁₂ have been reported [5,37,62–67]. Li₁₀GeP₂S₁₂ was prepared by reacting stoichiometric amounts of Li₂S, GeS₂, and P₂S₅ in vacuum quartz tubes at 550°C [5,10,47,54,68]. The bulk ionic conductivity was measured to be ~1×10⁻² S⁻¹ cm at 27°C (room temperature) or higher, implying that the material can substitute liquid electrolytes [5,10,69,70]. As shown in Fig. 3a, the LiS₆ octahedron and (Ge_0.5P_0.5)S₄ tetrahedron form a one-dimensional chain along the c-axis and the chains are horizontally connected by the PS₄ tetrahedron through a common edge. The zigzag conduction pathways enable the lithium ion hopping between Li 16h site (Li1) and Li 8f site (Li3) in LiS₄ tetrahedra as shown in Fig. 3a,b. An additional two-dimensional conduction path was confirmed by neutron diffraction Rietveld refinement and computational analysis based on first principles as shown in Fig. 3c [70–72]. A continuous lithium pathway connection between the Li1 and Li4 sites in addition to the one between Li1 and Li3 exhibited three-dimensional ion conduction pathways explaining high ionic conductivity of the LGPS-type solid electrolytes. Based on the aforementioned structural and conduction properties, Li_9.54Si_1.74P_1.44S_11.7Cl_0.3 which has a framework with a mixture of halides and S elements, exhibited the highest conductivity at 25 mS cm⁻¹ [5,64,73,74]. On the other hand, Li_10.35Sn_0.27Si_1.08P_1.65S₁₂ and Li_9.81Sn_0.81P_2.19S₁₂ have a large unit cell volume but Sn substitution results in the lowest conductivity because the lattice softness of the Sn-based structure forms a bottleneck in the z-axis direction and causes strong interaction between Li⁺ and S²⁻ [75].

Argyrodite is a class of chalcogenide compounds structurally related to the mineral Ag₈GeS₆ that contains various fast Ag⁺ or Cu⁺ ion conductors, including A₇PS₅X (A = Ag⁺, Cu⁺) [76,77]. Based on the topology of the cubic Laves phase (i.e., MgCu₂), crystalline structure of argyrodite is determined to consist of tetrahedral of non-metallic atoms (chalcogen/ halogen) with close-packed lattice. In the unit cell of Ag₉AlSe₆ (Z=4) [12], 24 chalcogen atoms create 136 tetrahedral holes, in which four Al³⁺ (ordered) and 36 Ag⁺ ions are located. These Ag⁺ ions are dynamically and/or statically disordered, allowing numerous argyrodite cations to be highly mobile [25].

Recently, Deiseroth et al. reported similar cubic Li⁺ argyrodites with the formula Li₆PS₅X (X = Cl, Br, and I) and Li₇PS₆ [25]. Li₆PS₅X shows a crystal structure analogous to A₇PS₅X (A = Ag⁺, Cu⁺) argyrodite. The P atoms of Li₆PS₅X are predicted to fill four of the 136 tetrahedral sites formed by the S atoms; however, the PS₄ tetrahedra do not share S atoms and is dispersed in the unit cell. Li⁺ ions occupy the remaining 132 tetrahedral sites (formed by S and X) in a partially disordered manner, resulting in the formula (Li⁺)₆(PS₄³⁻)S²⁻X⁻ (Fig. 4a,b). Changes in anions from Cl to Br or I for Li₆PS₅X result in observable alterations in the unit cell volume and site disorder. Increasing the disorder can reduce the activation barrier associated with the ion mobility [78]. On the other hand, new families of argyrodite-based solid electrolytes were developed by cation substitution for P, such as Li–M–Sb–S–I (M = Si, Ge, and Sn) [79–81], Li–P–M–S–I (M = Sn, Si) [29,82], Li–P–Bi–S–O–Cl [83], Li–M–S–O (M=Al, Si) [84,85], and Li–Sn–Si–P–S [86]. The composition containing Li–Si–Sb–I exhibited the highest conductivity of 14.8 mS cm⁻¹ under cold-pressed conditions at room temperature and up to 24 mS cm⁻¹ as a sintered pellet (Fig. 4c) [79]. By replacing P⁵⁺ with larger Ge⁴⁺, Li_6+xP_1–xGe_xS₅ exhibits a conductivity of 5.4 ± 0.8 mS cm⁻¹ in a cold-pressed condition and 18.4 ± 2.7 mS cm⁻¹ in the case of a high-temperature sintered pellet due to the significantly reduced activation barrier with increased unit cell volume [80,87].

In spite of high ionic conductivity of glass and glass ceramic type sulfide solid electrolytes, the correlation between their crystallinity and reactivity to moisture had not been thoroughly investigated. In this regard, Muramatsu et al. investigated the relationship between the structural composition and moisture stability of solid electrolytes by comparison of Li₂S-P₂S₅ glass and Li₂S-P₂S₅ glass ceramic prepared by post heat-treatment of the glass [96]. As an indication of moisture reactivity of the sulfide solid electrolytes, the amount of H₂S gas produced by the reaction to water were measured from a glass containing 67, 70, 75, and 80 mol% of Li₂S and glass-ceramic sulfide solid electrolytes as shown in Fig. 7a. Notably, the amount of H₂S produced in the Li₂SP₂S₅ solid electrolyte varies with the material composition and significant decrease in H₂S gas evolution was observed when Li₂S content was 75 mol% from both solid electrolytes. Even after extended storage time in air, the amount of H₂S produced from the 75Li₂S-25P₂S₅ composition remained relatively unchanged, whereas the 80Li₂S-20P₂S₅ composition exhibited sort of unstable behavior. By analyzing their Raman spectra, the process associated with structural change in the 67Li₂S·33P₂S₅ glass after the air exposure was proposed as described in Fig. 7b. After the exposure, the P₂S₇⁴⁻ ion which is the main component of the 67Li₂S·33P₂S₅ glass, first decomposed to form OH and SH groups. After that, the SH group was further hydrolyzed to form the OH group and H₂S gas. Fig. 7c–e display the Raman spectra of 75Li₂S-25P₂S₅ before and after exposure to the atmosphere for 1 day. The three invariant, apparent peaks within the 200–700 cm⁻¹ range (Fig. 7d) with no additional peak evolution within 1000–4000 cm⁻¹ range (Fig. 7e) indicate the high moisture resistivity of the PS₄³⁻ unit, while the formation of OH or SH groups is much suppressed [97]. T. A. Yersak et al. suggested that a significant amount of H₂S gas generation from the xLi₂S-(1–x)P₂S₅ glass electrolytes after the air exposure is closely related to the population of P₂S₇⁴⁻ and P₂S₆⁴⁻ units in the sulfide glass electrolytes because their labile bridging sulfur and P–P bonds are more susceptible to hydrolysis compared to PS₄³⁻ units [98]. Therefore, introducing PS₄³⁻ unit is highly desirable tactics in designing the moisture resistive glass and glass-ceramic sulfide solid electrolytes.

Combined doping of Li₂O [34,35,91,99] and P₂O₅ [16,23,31] were often utilized to stabilize the structure of sulfide solid electrolytes and thus improve the moisture resistance. Several studies demonstrated that replacing Li₂S with Li₂O, and P₂S₅ with P₂O₅ from argyrodite- and oxysulfide-based solid electrolytes enhanced the structural stability and water-resistivity of the sulfide solid electrolytes by reducing the reactivity of S to H₂O through O substitution for S in the sulfides. Ohtomo et al. showed that adding Li₂O to Li₂S-P₂S₅ glass can reduce the H₂S gas emission from the Li₂S-P₂S₅ glass in direct contact with H₂O [34,35]. It was shown that the Li₂O-doped Li₂S-P₂S₅ glass solid electrolyte primarily consisted of PS₄ and POS₃ units, with only trace amounts of P₂S₆ units. As a result, the moisture reactivity of the Li₂O-doped Li₂S-P₂S₅ was greatly reduced. Specifically, the reactivity (i.e., H₂S generation) of the xLi₂O·(100–x)(0.7Li₂S·0.3P₂S₅) glass (x = 17, 20, and 25) to moisture were evaluated, showing that only negligible amount of H₂S gas was detected from the 70Li₂S-30P₂S₅ glass upon Li₂O addition compared to the unsubstituted 75Li₂S-25P₂S₅ glass at a relative humidity of 80% (Fig. 8a). The doping of Li₂O into Li₂S-P₂S₅ glass further improved the electrochemical properties of the solid electrolyte significantly, despite the decrease in ionic conductivity. [34,35,100]. The decrease in the ionic conductivity by O substitution can be recovered or even be improved (from 1.9×10⁻⁴ to 4.0×10⁻⁴ S cm⁻¹) by post-annealing of the (75–x)Li₂S·25P₂S₅·xLi₂O glasses at 450°C as a glass ceramic of monoclinic thio-LISICON III phase was induced into the glass solid electrolytes (Fig. 8b,c). The heated (75–x)Li₂S·25P₂S₅·xLi₂O glasses further showed an improved thermal and chemical stability, which was substantiated by the shift of the exothermic peak to higher temperature region in the DSC curves (Fig. 8d) [101].

The synergistic effect of Li₂O and P₂O₅ dual doping on argyrodite-based solid electrolyte composition was reported by Xu et al. [23]. In chalcogen-based materials, the difference in the electronegativity between Li and anions must be decreased in order to enhance the ionic conductivity. From their studies, argyrodite-based solid electrolytes with different sulfur-to-oxygen ratios, (Li₆PS_4.25O_0.75Cl, Li₆PS₄OCl and Li₆PSO₄Cl) were synthesized and the doping effect on the structural evolution and the ionic conductivity of the materials were evaluated (Fig. 9a) [102,103]. Both theoretical and experimental demonstrations showed an increase in the ionic conductivity through an excess of Li in the crystal structure, along with chalcogen substitution for halogen anion. Among the off-stoichiometric compounds considered, e.g., Li_6.25PS_4.2O_1.05Cl_0.75, Li_6.25PS₄O_1.25Cl_0.75, and Li_6.25PS_3.5O_1.75Cl_0.75, the last composition exhibited the highest ionic conductivity of 2.80 mS cm⁻¹ at room temperature with improved chemical stability to Li metal anode. Importantly, the improved moisture stability by chalcogen substitution was verified by XRD analysis as shown in Fig. 9b. The comparison of XRD patterns for Li₆PS₅Cl and Li_6.25PS₄O_1.25Cl_0.75, before and after the exposure to humid air for 0.5 hour followed by a post-annealing process, demonstrated that the moisture resistivity of Li_6.25PS₄O_1.25Cl_0.75 was significantly improved, as evidenced by its well-preserved XRD patterns. In contrast, numerous peaks from unidentifiable phases were observed in the XRD pattern after the exposure for Li₆PS₅Cl. Additionally, a much smaller amount of H₂S gas was detected from Li_6.25PS₄O_1.25Cl_0.75 compared to the case with Li₆PS₅Cl, indicating the high moisture resistance of the Li_6.25PS₄O_1.25Cl_0.75 (Fig. 9c). These studies have confirmed oxygen-doping as an effective strategy for simultaneously improving the ionic conductivity and moisture stability of argyrodite-type sulfide solid electrolytes.

The moisture stability of sulfide solid electrolytes can also be improved by aliovalent cation substitution along with oxygen substitution, as shown by the studies of Tatsumisago’s and Xu’s groups regarding ZnO-doped Li₃PS₄. Chen et al. further reported the improved moisture stability and electrochemical properties of argyrodite-based solid electrolytes after ZnO doping [33,36,104]. Notably, ZnO was selected based on theoretical calculations that a certain amount of P⁵⁺ and S²⁻ in Li₃PS₄ could be substituted with Zn²⁺ and O²⁻, respectively. According to the report from Xu’s group, the excellent moisture stability of Li_3.06P_0.98Zn_0.02S_3.98O_0.02 was demonstrated by a significantly reduced amount of H₂S gas evolution (0.0175 cm³ g⁻¹) after exposure to humid air for 180 min (Fig. 10), compared to the amount of H₂S gas (0.07 cm³ g⁻¹) from the undoped Li₃PS₄ after only 20 min of exposure [36]. Their theoretical study on reaction entropy further strengthened their assertion that the higher reactivity of Li₃PS₄ to H₂O, compared to Li_3.06P_0.98Zn_0.02S_3.98O_0.02, was due to a larger entropy loss in the reaction between Li₃PS₄ and H₂O. These studies proposed the entropy of reaction as a useful indicator for designing moisture-stable sulfide solid electrolytes.

In addition to utilizing aliovalent metal oxide dopants to alleviate the reactivity of sulfide solid electrolytes to moisture, Hayashi et al. used metal oxides with high chemical reactivity to H₂S to achieve high moisture stability [16]. In their study, nanoscale ZnO particles were added to the P₂O₅-doped oxysulfide-based glasses (75Li₂S-(25–x)P₂S₅-xP₂O₅, where 0 ≤ x ≤ 10) by simple physical mixing. Here, ZnO was intended to act as a chemical H₂S absorbent. Despite of the decrease in ionic conductivity (Fig. 11a), the P₂O₅ (x = 10) doping slowed down the reaction rate between the sulfide solid electrolyte and water. Moreover, the addition of 10 wt.% ZnO further reduced the amount of H₂S gas through the reaction, ZnO + H₂S → ZnS + H₂O (Fig. 11b) [33,105]. This is predictable from the large negative values of the Gibbs free energy change (DG) for the reaction, M_xO_y + H₂S → M_xS_y + H₂O, as shown in Table 1 [16,105]. However, despite this positive synergic effect of P₂O₅ and ZnO, further optimization of dopants and additives is required to overcome the lowered ionic conductivity (~10⁻⁵ S cm⁻¹ due to an excess amount of P₂O₅ doping) and the production of H₂O (from the reaction between metal oxides and H₂S) for the actual application of air-stable sulfide solid electrolytes to ASSBs (Fig. 11c).

Diverse metal oxides, including ZnO, Fe₂O₃, and Bi₂O₃, were evaluated as H₂S scavengers, and their effects on the electrochemical properties of ASSBs were further evaluated by various groups [16,33–35,55,100,106]. Especially from Hayashi’s report, unlike other metal oxides, Bi₂O₃ reacts with Li at around 1 V and forms a Li–Bi alloy before lithium plating occurs, as shown in Fig. 12. This study emphasizes that the chemical and electrochemical compatibility of metal elements with the Li plating process should be considered when designing the most appropriate metal oxide additives.

Along with doping PS₄³⁻ or metal oxides on sulfide-based solid electrolytes, capturing H₂O and H₂S is also an efficient strategy for suppressing the water-reactivity of the sulfides. This approach involves the addition, substitution, and doping of metal oxides to sulfide-based solid electrolytes. While various materials including M_xO_y (M = Zn [16,33], Fe [33], Bi [33,83], Nb [20], Sb [107], and Zr [108], zeolite [44], and etc.) have been investigated as a structure stabilizer of sulfide solid electrolytes for the reaction to H₂O, some of them have been proven to be effective physical scavenger of not only H₂S but also H₂O. Lee et al. demonstrated that porous zeolite is effective in absorbing H₂S and H₂O molecules which may inhibit the formation of additional H₂S [44]. As described in Fig. 13a, when zeolite and Li₆PS₅Cl mixture (ZLi₆PS₅Cl) were exposed to humid air, some amount of H₂S was to be produced by a chemical reaction between Li₆PS₅Cl and moisture at the beginning. But, the zeolites could effectively adsorb H₂S and H₂O molecules to prevent further intimate contact between moisture and Li₆PS₅Cl, reducing additional H₂S production in the composite electrolyte compared to the pristine sample (P-Li₆PS₅Cl). The ionic conductivity of original Z-Li₆PS₅Cl (1.27×10⁻³ S cm⁻¹) was almost comparable to P-Li₆PS₅Cl (1.31×10⁻³ S cm⁻¹). But after 1 h of exposure to humid air (RH (room humidity) of 50%), slightly higher conductivity of ZLi₆PS₅Cl (0.39×10⁻³ S cm⁻¹) than P-Li₆PS₅Cl (0.23×10⁻³ S cm⁻¹) was observed with much mitigated H₂S gas evolution over the testing period (Fig. 13b). which could be ascribed to the efficient H₂O and H₂S capturing by the zeolite additive in ZLi₆PS₅Cl (Fig. 13c). In addition, Fig. 13d and e show the comparison of ASSB cycling performance that demonstrates the enhanced discharge capacity and cycling retention of Z-Li₆PS₅Cl than P-Li₆PS₅Cl because the zeolite nanoparticles in Li₆PS₅Cl effectively mitigated continuous H₂S generation.

Recently, J. Y. Jung et al. reported zeolite imidazolate framework-8 as a moisture absorbent in LPSCl SE without modifying the structure of the SE. They demonstrated that the use of those adsorbents enabled a dry processing for ASSB manufacturing in a regular laboratory or processing area at room temperature, removing the need for a dry room facility [109]. In this light, the development of effective moisture absorbents could be a practical solution for manufacturing sulfide SEs in the near future.

The hard and soft, acid and base (HSAB) theory was first introduced by Sahu et al. and it is widely utilized for explaining the chemical stability of sulfide based solid electrolytes toward H₂O or O₂ [39]. According to the HSAB theory, hard acids are small, compactly arranged atoms with high charge density, and weakly polarizable, whereas soft acids are large, loosely arranged atoms with low charge density, and strongly polarizable. Importantly, hard acids prefer to combine with hard bases, whereas soft acids readily react with soft bases. On the basis of the HSAB theory, numerous studies have employed the relatively soft Sn [18,21,24,38,41–43,110], Sb [24,111], As [39,40], and Cu [19] elements to improve the water stability of Li₃PS₄ solid electrolytes [87]. For example, the thiophosphate-based superionic conductors are chemically unstable in the presence of oxygen or H₂O because the hard acid P readily reacts with the hard bases, O or H₂O [38,39]. In this regard, Sahu et al. synthesized Li_3.833Sn_0.833As_0.166S₄ in 2014 by introducing Sn and As in accordance with the β-Li₃PS₄ structure and aimed to increase the ionic conductivity after air exposure based on superior moisture resistance based on HSAB theory [39]. Later, Kaib et al. have proposed a Li₄SnS₄-based ionic conductor that is water-stable and exhibits high ionic conductivity based on the HSAB theory at the same time [38]. Introducing soft acids like Sn and As cations enables greater chemical stability towards oxygen, which is a hard base, while maintaining high ionic conductivity (1.39×10⁻³ S cm⁻¹ at 25°C (Fig. 14a) for the composition Li_3.833Sn_0.833As_0.166S₄. Fig. 14b,c display the XRD patterns of Li_3.833Sn_0.833As_0.166S₄ and β-Li₃PS₄ before and after 48 h of exposure to laboratory atmosphere (18°C, 80% RH), respectively [112]. The comparison demonstrated the high chemical stability of weak acids Sn and As doped Li_3.833Sn_0.833As_0.166S₄ to moisture, which preserved initial crystal structure distinctly.

In 2017, Choi et al. reported Li₄SnS₄ which is soft acid Sn substituted for P in Li₃PS₄, and its LiI composite glass, 0.4LiI-0.6Li₄SnS₄, via a water-based synthetic route to improve the moisture stability of Sn-based solid electrolytes [18]. A solid phase of Li₄SnS₄ was first synthesized, completely dissolved in deionized water (Fig. 15a), which was then precipitated to obtain Li₄SnS₄. It was confirmed that the original structure had been restored through reheating, and the ionic conductivity was measured at 0.14 mS cm⁻¹, indicating a relatively high conductivity level. A comparison of the amount of H₂S released from the aqueous solutions of Li₄SnS₄ and Li₁₀GeP₂S₁₂ is shown in Fig. 15b. In contrast to Li₁₀GeP₂S₁₂, which produces a substantial amount of H₂S, the Li₄SnS₄ solution produces negligible amounts of H₂S. Later in 2020, Kwak et al. revisited Li₄SnS₄-based ionic conductors and further explored an orthorhombic Sb-substituted Li₄SnS₄; Li_4–xSn_1–xSb_xS₄ [21,24]. A high ionic conductivity of ~10⁻⁴ S cm⁻¹ and excellent atmospheric air stability of Li_4–xSn_1–xSb_xS₄ was demonstrated by showing the reduced amount of H₂S gas evolution compared to Li₄SnS₄ while exposing the samples to atmospheric air for 15 mins as shown in Fig. 15c [18,38]. Therefore, the combination of soft acid substitution to sulfides and developing a new glass composition at one time would be a desirable for improving moisture stability of sulfide solid electrolytes.

On the other hand, Liang et al. and Wang et al. reported the incorporation of soft acid Sb into Li₁₀GeP₂S₁₂ (LGPS) and Li₁₀SnP₂S₁₂ (LSPS), respectively [65,66]. The HSAB theory can theoretically explain the instability of LGPS and LSPS to water by ascribing the fact to the weak P–S bond [17,111]. According to the HSAB theory [113], rather than maintaining the P–S bond structure, the hard acid P⁵⁺ in Li₁₀GeP₂S₁₂ readily forms a P–O bond by combining with oxygen in air/water, while preferential binding between S and the H that emits H₂S [114]. The Ge–S bond in Li₁₀GeP₂S₁₂ is relatively more stable because of the strong bond between the soft acid Ge⁴⁺ and soft base S²⁻ [115,116]. Therefore, complete or partial substitution of P in Li₁₀GeP₂S₁₂ with soft acid is necessary to enhance water stability. Sb has many advantageous properties among the many candidates for soft bases to increase the stability of solid electrolyte towards water [81,90,117]. The strong covalent bonding between Sb and S has a low affinity for oxygen, as confirmed by the formation of Na₃SbS₄·xH₂O rather than the breakage of the Sb–S bond [114,118]. In addition, the larger ionic radius of Sb (r_Sb5+ = 60 pm vs. r_P5+ = 38 pm) results in a larger lattice and enlarge Li diffusion pathway [87]. The same principles can be applied to LSPS. For these reasons, the Li₁₀Ge(P_1–xSb_x)₂S₁₂ exhibited an ionic conductivity of 17.3 mS cm⁻¹ which was well maintained even after moisture exposure as shown in Fig. 16a [65]. Liang et al further examined the performance of ASSBs assembled with air exposed Li₁₀Ge(P_0.95Sb_0.05)₂S₁₂ SE as shown in Fig. 17. In the LiCoO₂+SE composite/SE/In configuration, LiCoO₂ was uniformly coated with LiNbO₃ on its surface. From the test, a high initial Coulombic efficiency of 91% and the stable charge/discharge reaction over 100 cycles with initial discharge capacity of ~130 mAh g⁻¹ demonstrated the water stability of air exposed Li₁₀Ge(P_0.95Sb_0.05)₂S₁₂ SE. For the moisture stability test of the LSPS and Li₁₀SnP_1.8Sb_0.2S₁₂ samples, both samples were exposed to an environment with a room humidity (RH) of 18% for 1 h to determine their water stability along with ionic conductivity change. The testing result in Fig. 16b indicates that the moisture stability is improved by Sb doping evidenced by the minimal decrease in ionic conductivity between before and after the air exposure. The improved electrochemical properties of Li₁₀SnP_1.8Sb_0.2S₁₂ compared to LSPS confirmed by the much suppressed overpotential evolution from the Li plating/stripping cycling conducted at a current density of 0.1 mA cm⁻² as shown in Fig. 16c. Therefore, Sb doping on LGPS and LSPS can significantly improve not only the moisture stability of the materials but also the ionic conductivity and electrochemical stability.

In contrast to HSAB strategy that substitutes strong acid P with soft acid cation for the purpose of mitigating the reactivity of sulfide SEs to the moisture, replacing S by more electron pulling Se in the Li₁₀GeP₂S₁₂ (LGPS) to suppress the H₂S gas evolution was first suggested by a theoretical work from J. C. Jiang’s group as shown in Fig. 18 [95]. Based on their combination of electron localization function, electron density difference, and partial density of states analysis, the dissolution energy of H₂O (energy for H₂O ® 2H⁺ + O²⁻) adsorbed at the surface of Se-doped LGPS can be increased due to the localized electron structure from each element at the surface of the LGPS SE. The localized electron structure of each atoms in Li–O and S–H bonds after Se doping suppresses the electron transfer between the atoms in the bond and consequently increases the activation energy barriers of H₂S formation and stabilizes LGPS against reaction with H₂O. Although the electrochemistry of the suggested material has not been reported yet, the strategy of modifying the electron structure to suppress the reactivity of sulfide SE with H₂O seems worth trying for many sulfide SE families.

Z. Yu et al. suggested a new strategy differentiated from the chemical doping or H₂O scavenger approaches described in the previous sections. They introduced removable surface protection layer on the sulfide SEs using the organic amphipathic molecules to suppress the reactivity of sulfide SEs to moisture [119]. In their study, a glass-ceramic Li₇P₂S₈Br_0.5I_0.5 (σ_ion = 5.3 mS cm⁻¹) was coated with a hydrophobic organic layer (1-bromopentane (1BR), C₅H₁₁Br) with the thickness of 0.59 nm by wet physical mixing and drying as depicted in Fig. 19a. The coated SE exhibited much enhanced moisture stability under the ambient condition (relative humidity = 20 %) compared to the uncoated by virtue of suppressed H₂S gas evolution. A first-principle calculation based on density functional theory showed that the hydrophobic long-chain alkyl tail of 1-bromopentane repels water molecules and 1-bromopentane has more negative adsorption energy with the sulfide SE than H₂O, thereby enhancing the moisture stability of the SE as shown in Fig. 19b. This simple coating approach could be considered for mass production process of sulfide solid electrolytes as upkeep expenses for maintaining the extremely dry environment to produce the materials could be very expensive in industry.