All-Solid-State Lithium Batteries with Sulfide Electrolytes and Oxide Cathodes

Abstract

All-solid-state lithium batteries (ASSLBs) have attracted increasing attention due to their high safety and energy density. Among all corresponding solid electrolytes, sulfide electrolytes are considered to be the most promising ion conductors due to high ionic conductivities. Despite this, many challenges remain in the application of ASSLBs, including the stability of sulfide electrolytes, complex interfacial issues between sulfide electrolytes and oxide electrodes as well as unstable anodic interfaces. Although oxide cathodes remain the most viable electrode materials due to high stability and industrialization degrees, the matching of sulfide electrolytes with oxide cathodes is challenging for commercial use in ASSLBs. Based on this, this review will present an overview of emerging ASSLBs based on sulfide electrolytes and oxide cathodes and highlight critical properties such as compatible electrolyte/electrode interfaces. And by considering the current challenges and opportunities of sulfide electrolyte-based ASSLBs, possible research directions and perspectives are discussed.

Graphic Abstract

1 Introduction

Driven by the emergence of electric vehicles, great efforts have recently been devoted to the development of better rechargeable lithium ion batteries in terms of higher specific energy density, longer cycle lifes and better safety [1]. Although significant efforts have been devoted to lithium ion batteries based on organic liquid electrolytes in the past 30 years, the limited electrochemical and thermal stability, low ion selectivity and poor safety of organic liquid electrolyte-based lithium ion batteries hinder application. In particular, the combustible characteristic of organic electrolytes is a significant security risk for vehicle or grid applications that require large battery sizes. Here, the replacement of liquid electrolytes with solid electrolytes is regarded to be the ultimate solution to these problems in which the nonflammable nature of solid electrolytes can avoid fire and explosion even in a large scale [2,3,4,5,6,7]. Based on this, a variety of solid ionic conductors have been investigated as solid electrolytes to replace liquid ones [8]. However, their application is limited by low ionic conductivity. Sulfide-based electrolytes have been explored as alternatives to oxide-based electrolytes due to the more polarizable nature of sulfur anions, which can allow for more facile lithium ion conduction and higher ionic conductivity [9, 10]. And in recent years, the realization of sulfide electrolytes with unprecedented ionic conductivities equal to or greater than liquid electrolytes has been achieved in which Li₁₀GeP₂S₁₂ and Li_9.54Si_1.74P_1.44S_11.7Cl_0.3 have been demonstrated to possess ionic conductivities of 1.2 × 10⁻² S cm⁻¹ and 2.5 × 10⁻² S cm⁻¹, respectively [11, 12]. Various systems from glass to glass–ceramics and crystalline conductors have also been studied, and an exciting new area in solid electrolytes has been presented. Due to these prominent performances, sulfide electrolytes can provide opportunities in the development of ASSLBs for electric vehicle applications.

Despite these promising findings, many challenges still exist in ASSLBs based on sulfide electrolytes. First, many sulfide electrolytes are unstable at low potentials, making their matching with lithium metal anodes difficult. Second, numerous studies have reported that the complex reactions between electrodes and electrolytes can result in low interfacial charge–transfer kinetics. Third, damage caused by Li dendrite formation in sulfide electrolytes is far worse than predicted and will shorten battery lifespans. Finally, the comprehensive understanding of sulfide electrolytes remains lacking. As for cathodes, lithium transition metal oxides have been extensively used in traditional lithium ion batteries because of their high electrochemical potential and moderate capacity [13, 14] and are commercially successful. However, poor rate and cyclic performances are generally observed in ASSLBs that combine sulfide electrolytes with oxide cathodes due to interfacial instability between the two, which is a major issue limiting the application of oxide cathodes in ASSLBs. In terms of anodes, although the use of lithium metal is necessary to obtain high-energy density, recent research has revealed that Li dendrites can easily form in the bulk or grain boundaries of inorganic sulfide electrolytes [15, 16]. In addition, sulfide electrolytes can show severe thermodynamic and dynamic instability against metallic lithium and the huge volume change of lithium anodes during lithium ion deposition/dissolution can lead to virtually infinite volume expansion to further deteriorate interfacial stability. Overall, the insufficient fundamental understanding of interfacial evolution processes during charge/discharge severally hinders the realization of ASSLBs based on sulfide electrolytes, oxide cathodes and lithium metal anodes.

Herein, we present an overview of the recent progress of ASSLBs using sulfide electrolytes and oxide cathodes. This review starts with a brief description of sulfide electrolytes. The corresponding electrolyte structures, preparation methods and ionic conductivity improvement strategies are summarized in detail. It is then followed by detailed discussions on the mechanisms of interfacial issues between sulfide electrolytes and electrodes with oxide cathodes and lithium metal anodes being the main focus. In addition, various design strategies including electrode coating, electrolyte component tuning and interfacial construction are proposed. At the end, a brief perspective concerning future development is presented.

2 Sulfide Electrolytes

The study of sulfide-type solid electrolytes began in 1981 with a Li₂S–P₂S₅ system [17] and although conductivities were improved through the doping of Li_xMO_y (M = Si, P and Ge), ionic conductivities remained lacking [18]. In 2001, a series of crystalline sulfide electrolytes based on the Li₂S–P₂S₅ system was reported by Kanno et al. [19] that possessed high lithium ion conductivities. As compared with Li₂S–SiS₂, Li₂S–P₂S₅ systems possess better physicochemical properties and have therefore been widely employed in all-solid-state batteries. In particular, Li₂S–P₂S₅ systems possess high room-temperature ionic conductivities of 0.1 × 10⁻³–1 × 10⁻³ S cm⁻¹ and wide electrochemical windows in which the materials in this family are referred to as “thio-LISICON” because their structures are similar to lithium superionic conductor (LISICON) materials [20]. Different from LISICON materials that possess poor room-temperature ionic conductivities, thio-LISICON materials possess much higher ionic conductivities due to the replacement of larger and more polarizable S²⁻ for O²⁻, which can enlarge lithium ion migration tunnels and weaken interactions to result in higher ionic conductivities as compared with oxide analogs [21].

Based on the crystalline state, sulfide electrolytes can be divided into three categories, including glass (amorphous), crystal and glass–ceramic phases in which glassy electrolytes are attractive due to their isotropic ion conduction, zero grain boundary resistance and low costs with conductivities reaching ~ 10⁻⁴ S cm⁻¹ at room temperature. Alternatively, glass–ceramic electrolytes obtained through the crystallization of glassy electrolytes can possess higher conductivities of 10⁻³ S cm⁻¹ (even up to 10⁻² S cm⁻¹). Tatsumisago et al. [22,23,24,25] systematically investigated the conductivity of (100 − x)Li₂S–xP₂S₅ systems by controlling composition and heat treatment temperature and found that for all these electrolytes, glass–ceramic type electrolytes exhibited higher ionic conductivity as compared with glass or crystalline materials. Here, these researchers attributed this ionic conductivity enhancement mainly to the appearance of new metastable thio-LISICON analogs as confirmed by XRD analysis.

Sulfide electrolytes can also be divided into three categories based on structure, including binary systems of Li₂S–MS_x (M = P, Si and Ge), ternary systems of Li₂S–P₂S₅–MeS_x (Me = Si, Ge and Sn) and the argyrodite type. In this section, the crystal structure, conductivity and stability of these sulfide electrolyte systems will be discussed in detail.

2.1 Structure of Different Sulfide Electrolytes

2.1.1 Structure of Binary Sulfide Electrolytes

The chemical composition and conductivities of representative binary sulfide electrolytes include Li₂S–P₂S₅, Li₂S–SiS₂, Li₂S–GeS₂ and Li₂S–SnS₂ are summarized in Table 1. Among these, Li₂S–P₂S₅ has been the most studied system due to superior ionic conductivities. Taking into consideration good compatibility with metallic lithium and relatively high conductivity, Li₃PS₄ with a stoichiometry of 75%Li₂S–25%P₂S₅ has also been extensively studied. In addition, Li₃PS₄ is usually regarded as the stablest composition in Li₂S–P₂S₅ systems and thus far, three types of Li₃PS₄ systems have been reported, including α-Li₃PS₄, β-Li₃PS₄ and γ-Li₃PS₄ (Fig. 1a–c) [26]. Among these crystal phases, the γ-Li₃PS₄ phase possesses the lowest ionic conductivity of 3 × 10⁻⁷ S cm⁻¹, whereas the β-Li₃PS₄ phase can deliver the highest ionic conductivity of ~ 10⁻⁴ S cm⁻¹ in which the different arrangements of PS₄ tetrahedra in Li₃PS₄ are the main reason for these differences in conductivity. In terms of structure, α-Li₃PS₄ is composed of LiS₄ tetrahedra and isolated PS₄ tetrahedra in which one LiS₄ tetrahedron is connected to four LiS₄ tetrahedra through corner sharing, one PS₄ tetrahedron through edge sharing and two PS₄ tetrahedra through corner sharing [27]. By contrast, γ-Li₃PS₄ is stable at low temperatures [27, 28] and possesses an ordered arrangement with the apex of the PS₄ tetrahedron orderly arranged in the same direction and the apex of the LiS₄ tetrahedron showing the same ordering as that of the PS₄ tetrahedron [27]. In terms of the structure of β-Li₃PS₄, this is particularly interesting due to its high lithium ion conductivity in which corresponding PS₄ tetrahedra are isolated from each other and connected with LiS₆ octahedra through edge sharing [29]. Here, lithium ions located in the tetrahedral (Wyckoff 4c) and octahedral (Wyckoff 4b) positions can form face-sharing octahedral–tetrahedral chains across the b-axis [26]. And unlike the ordered arrangement of PS₄ tetrahedra in γ-Li₃PS₄, β-Li₃PS₄ possesses a zig-zagged arrangement of PS₄³⁻ tetrahedra that can provide lithium ion positions in both octahedral and tetrahedral sites, allowing for more mobile lithium ions [27, 29].

Table 1 Summary of the conductivity of representative binary sulfide electrolytes

Fig. 1

Structure of a α-Li₃PS₄. Reprinted with permission from Ref. [27]. Copyright 2010, The Physical Society of Japan. b Structure of β-Li₃PS₄. c Structure of γ-Li₃PS₄. Reprinted with permission from Ref. [29]. Copyright 2007, Elsevier. d Li₇P₃S₁₁ viewed along the [010] direction. Reprinted with permission from Ref. [30]. Copyright 2007, Elsevier

As an example of a binary sulfide electrolyte, a newly emerged Li₇P₃S₁₁ phase developed by Yamane et al. [30] has attracted increasing attention due to its superior conductivity of up to 1.7 × 10⁻² S cm⁻¹ as reported by Seino et al. [31]. Here, the stoichiometry of Li₇P₃S₁₁ corresponds to a 70%Li₂S–30%P₂S₅ mixture with the crystalline structure of Li₇P₃S₁₁ (Fig. 1d) being composed of corner-sharing PS₄³⁻ and P₂S₇⁴⁻ anions in a 1:1 ratio in which nearly all lithium sites are tetrahedrally coordinated (LiS₄) and interconnected by further empty tetrahedral sites (□S₄) to provide three-dimensional diffusion tunnels [26]. And unlike other ionic conductors involving Li₂S–P₂S₅ systems, Chu et al. [32] proposed that the ionic conductivity of Li₇P₃S₁₁ does not proceed through the slow diffusion of isolated defects (i.e., lithium ion vacancies), but rather through the collective motion of multiple defects. Despite the favorable ionic conductivity of Li₇P₃S₁₁, it is metastable and can easily transform into other low-conductivity crystalline structures due to its narrow thermal stability window in which if heated to over 420 °C, it will decompose into Li₃PS₄ and Li₄P₂S₇, the latter of which tends to form Li₄P₂S₆ with extremely low conductivity (10⁻⁸ S cm⁻¹).

2.1.2 Structure of Ternary Sulfide Electrolytes

The electrochemical performance of Li₂S–P₂S₅ binary systems, especially ionic conductivity at room temperature remains unsatisfactory. To further optimize ionic conductivity and electrochemical performance, Li₂S–P₂S₅ ternary systems can be obtained through the introduction of tertiary components such as GeS₂, SiS₂, SnS₂ or Al₂S₃. The conductivities of representative ternary sulfide electrolytes are summarized in Table 2. Of these ternary systems, Li₂S–GeS₂–P₂S₅ is the most studied due to high ionic conductivities in which X-ray diffraction analysis has demonstrated that the introduction of GeS₂ can enlarge lattice structures, which is beneficial to lithium ion transport and enhanced ionic conductivity. This structure was first discovered by Kanno et al. [19] based on a Li₄GeS₄–Li₃PS₄ solid solution system with a structural formula of Li_4−xGe_1−xP_xS₄ in which corresponding Li_4−xGe_1−xP_xS₄ crystals possess a similar structure to γ-Li₃PO₄ and can be divided into three compositional regions depending on monoclinic superstructures with different types of cation ordering, including region I (0 < x $\leqslant$ 0.6), which possesses a monoclinic superlattice cell of a × 3b × 2c that is related to orthorhombic Li₄GeS₄; regions II (0.6 < x < 0.8) and III (0.8 $\leqslant$ x < 1.0) that possess different monoclinic cells of a × 3b × 3c and a × 3b × 2c, respectively, and region II has been reported to possess the highest conductivity of 2.2 × 10⁻³ S cm⁻¹ (Li_3.25Ge_0.25P_0.75S₄) at room temperature among the three regions. This Li_3.25Ge_0.25P_0.75S₄ electrolyte also possesses advantages including negligible electronic conductivity, high electrochemical stability and good physicochemical stability.

Table 2 Summary of the conductivity of representative ternary sulfide electrolytes

In 2011, Kanno et al. [11] discovered a new lithium superionic conductor Li₁₀GeP₂S₁₂ with an extremely high room-temperature ionic conductivity of 12 mS cm⁻¹ that was even higher than that of organic liquid electrolytes used in commercial lithium ion systems. X-ray diffraction patterns indicated that the three-dimensional network structure of Li₁₀GeP₂S₁₂ differed from previously reported thio-LISICONs (Fig. 2a) in which the framework was composed of (Ge_0.5P_0.5)S₄ tetrahedra, PS₄ tetrahedra, LiS₄ tetrahedra and LiS₆ octahedra. Among these (Ge_0.5P_0.5)S₄ tetrahedra and LiS₆ octahedra shared the same edges and can form one-dimensional long chains along the c-axis to act as one-dimensional lithium conduction pathways. These one-dimensional chains were further connected to one another through PS₄ tetrahedra that were connected to LiS₆ octahedra through a common corner (Fig. 2b). As a result, high ionic conductivities can be achieved through the zigzag conduction pathways formed by LiS₄ tetrahedra in 16h and 8f sites along the c-axis direction (Fig. 2c).

Fig. 2

Crystal structure of Li₁₀GeP₂S₁₂. a The framework structure and lithium ions that participate in ionic conduction. b Framework structure of Li₁₀GeP₂S₁₂. One-dimensional (1D) chains formed by LiS₆ octahedra and (Ge_0.5P_0.5)S₄ tetrahedra, which are connected by a common edge. These chains are connected by a common corner with PS₄ tetrahedra. c Conduction pathways of lithium ions. Zigzag conduction pathways along the c axis are indicated. Lithium ions in the LiS4 tetrahedra (16h site) and LiS4 tetrahedra (8f site) participate in ionic conduction. Thermal ellipsoids are drawn with a 30% probability. The anisotropic character of the thermal vibration of lithium ions in three tetrahedral sites gives rise to 1D conduction pathways. Reprinted with permission from Ref. [11]. Copyright 2011, Nature Publishing Group

Despite the high ionic conductivity of Li₁₀GeP₂S₁₂, the high cost of elemental Ge and its instability with lithium metal limit large-scale application in ASSLBs. Because of this, many cheaper elements have been employed to replace Ge. For example, Ong et al. [33] investigated the ionic conductivity of a Li_10±1MP₂X₁₂ family of materials (M = Ge, Si, Sn, Al or P and X = O, S or Se) using first principle calculations and found that conductivity will not be affected by substituting Ge with relatively cheaper Sn or Si. Bron et al. [34] also reported the successful synthesis of Li₁₀SnP₂S₁₂ with a high grain conductivity of 7 mS cm⁻¹ and a total conductivity of 4 mS cm⁻¹ in which structural characterizations indicated that Li₁₀SnP₂S₁₂ was isostructural to the Ge analog with regard to the P4₂/mc space group, thus ensuring high ionic conductivity. This high conductivity as well as the reduced cost of Li₁₀SnP₂S₁₂ makes it an attractive and affordable candidate for application in ASSLBs. White et al. [35] further prepared Li₁₀SiP₂S₁₂ as another Li₁₀GeP₂S₁₂ analog through the replacement of Ge by Si and obtained a conductivity of 2.3 mS cm⁻¹, which was greater than that of Li₁₀GeP₂S₁₂ prepared in the same conditions. Elemental Al has also been used to replace Ge. For example, Zhou et al. [36] prepared Li₁₁AlP₂S₁₂ with a Li₁₀GeP₂S₁₂ analogous structure through sintering and reported that although their ternary electrolyte provided an ionic conductivity (0.82 mS cm⁻¹) that was lower than that of Li₁₀GeP₂S₁₂, it provided a wider electrochemical window and excellent electrochemical performances.

2.1.3 Argyrodite-Type Solid Electrolytes

Argyrodite is another important class of solid electrolytes derived from the mineral Ag₈GeS₆ and possesses high Ag⁺ conductivities [37]. In addition, the replacement of Ag⁺ ions by other cations does not damage the cubic structure of argyrodite. And based on efforts to enhance lithium ion mobility, lithium argyrodites have been explored as solid electrolytes. A summary of the conductivity of representative argyrodite-type solid electrolytes is presented in Table 3. Li₇PS₆ as a lithium argyrodite was reported to have a cubic phase at high temperature or an orthorhombic phase at low temperature, and the cubic high-temperature phase possesses higher ionic conductivity and can be stabilized by the replacement of sulfur by halogen anions [38]. Deiseroth et al. [39] explored a series of lithium argyrodites with the general formula Li₆PS₅X (X=Cl, Br, I) as characterized by the partial replacement of S²⁻ by halogen anions. Theoretical calculations demonstrated that the incorporation of halogen ions can stabilize the corresponding cubic structure at room temperature and improve ionic conductivity [40]. For example, Li₆PS₅Cl and Li₆PS₅Br prepared through solid phase methods can reach ionic conductivities of 1.9 × 10⁻³ S cm⁻¹ and 6.8 × 10⁻³ S cm⁻¹, respectively [41], whereas the ionic conductivity of Li₆PS₅I is low. Here, high-resolution neutron and X-ray diffraction analyses showed that aside from lithium distribution disorder, disorder caused by Cl⁻ and Br⁻ can promote lithium ion mobility. However, I⁻ cannot exchange with S²⁻ due to its large size, resulting in more ordered structure and low conductivity, which cannot compete with those of Li₆PS₅Cl and Li₆PS₅Br [42]. In addition, Li₆PS₅X electrolytes can also possess wide electrochemical windows and cheaper precursors of this system allow for great potential as practical candidates for further application.

Table 3 Summary of the conductivity of representative argyrodite-type solid electrolytes

Li₆PS₅X can crystallize into a structure based on the tetrahedral close packing of anions (a cubic unit cell with the space group F4̅3m and a space group number 216, see Fig. 3). The halide anions X⁻ can form a face-centered cubic lattice (4a) with isolated PS₄ tetrahedra on octahedral sites (P on 4b) and free S in half of the tetrahedral sites (4d), while Li⁺ ions are randomly distributed over the tetrahedral interstices (48h and 24g sites) [7, 43]. In the case of Li₆PS₅I, S and I atoms at 4d and 4a sites are fully ordered due to the small size mismatch between S and I, while it is fully disordered for Li₆PS₅Cl. Li₆PS₅Br consists of both ordered and disordered structures with 84% S and 16% Br at 4d sites and 60% Br and 40% S at 4a sites [7]. Here, the absence of disorder in Li₆PS₅I will lead to higher activation barriers for ion migration and lower ionic conductivities as compared with other argyrodite systems.

Fig. 3

a Crystal structure of Li₆PS₅X in the case of X = I, b rate-determining step of the inter-cage 48h–48h jump in lithium ion migration as shown by LiS₃I polyhedra. Reprinted with permission from Ref. [43]. Copyright 2018, American Chemical Society

2.2 Synthesis Methods

In general, sulfide electrolyte preparation methods can be divided into three categories, including solid-state reaction, mechanical ball-milling and wet chemistry methods. Of these methods, the solid-state reaction method involves the heating of a precursor material mixture in stoichiometric ratios at the melting point followed by the cooling of the molten sample to room temperature. The preparation possess of the solid-state method is harsh and direct solid-state methods can produce impurities due to lithium and sulfur volatilization under high temperatures. Now, these issues can be alleviated through sealing techniques in which precursor powders can be pressed into pellets and filled into quartz ampoules that are sealed under vacuum conditions. By providing a sealed sintering environment, electrolyte purity can be ensured and high conductivities up to 10⁻³ S cm⁻¹ can be achieved [43,44,45]. For example, Seino et al. [31] synthesized a Li₂S–P₂S₅ glass ceramic electrolyte that possessed a high conductivity of 1.7 × 10⁻² S cm⁻¹ through the solid-state reaction followed by hot pressing in which a mixture of Li₂S and P₂S₅ was sealed in a quartz tube and heated at 700 °C for 2 h followed by rapid quenching in ice water and hot-press sintering under 94 MPa at 280 °C or 300 °C to obtain the densified glass ceramic sample. These researchers also reported that grain boundary resistances can be reduced through optimized heat treatment conditions and allow for higher lithium ion mobility than that of organic liquid electrolytes. Ternary sulfide electrolytes with high conductivities can also be obtained through the solid-state method. For example, Kanno et al. obtained Li₁₀GeP₂S₁₂ by reacting stoichiometric quantities of Li₂S, GeS₂ and P₂S₅ at 550 °C in an evacuated quartz tube followed by a slow cooling process, whereas for Li_9.54Si_1.74P_1.44S_11.7Cl_0.3, the synthesis process was almost the same but with different initial precursor materials and reaction temperatures [11, 12]. The solid-state reaction method can further be used in the synthesis of argyrodite-type solid electrolytes. As examples, Deiseroth et al. [39] and Kraft et al. [46] both used the simple direct heating of precursor mixtures to prepare Li₆PS₅X despite long heating times, even up to 2 weeks. However, Wang et al. [47] also investigated the influence of sintering temperature and time on the structure and conductivity of Li₆PS₅Cl and found that 10 min of heating at 550 °C was sufficient to obtain Li₆PS₅Cl with a high conductivity of 3.15 × 10⁻³ S cm⁻¹. Similarly, Yu et al. [48] achieved a higher conductivity of 5 × 10⁻³ S cm⁻¹ for Li₆PS₅Cl after long-term annealing with pellets pressed under high pressure.

Mechanical ball-milling, especially high-energy ball-milling, has been widely used to prepare sulfide electrolytes in which precursor particles can collide, diffuse and react if induced energy through high-speed impact is sufficient [49]. And unlike solid-state reactions that require intricate processes involving mixing, pulverization and high-temperature treatment, mechanical ball-milling processes are simple, with the entire process being able to be completed at room temperature. More importantly, the preparation of amorphous electrolytes can be easily achieved through mechanical ball-milling due to low reaction temperatures. Using this technology, a series of binary and ternary electrolytes has been synthesized, including 70Li₂S–30P₂S₅ (8.6 mS cm⁻¹) [50], 77.5Li₂S–22.5P₂S₅ (1 mS cm⁻¹) [51], 75Li₂S–25P₂S₅ (0.5 mS cm⁻¹) [52] and ternary Li₂S–M_xS_y–P₂S₅ electrolytes [53]. High-energy ball-milling can also widen the region of amorphous-sample formation as compared with the solid-state method. For example, Adams et al. [41] recently synthesized Li₆PS₅X (X = Cl, Br, I) argyrodite through mechanical ball-milling followed by annealing and reported that the use of their process allowed for faster synthesis than earlier reported methods with longer annealing times. Subsequently, a series of studies was conducted in their group to optimize the synthesis route, allowing for crystalline phases that can exhibit high ionic conductivities reaching 7 × 10⁻⁴ S cm⁻¹ for X = Cl or Br [42, 54]. Boulineau et al. [55] were also able to obtain Li₆PS₅Cl with an ionic conductivity of 1.33 × 10⁻³ S cm⁻¹ and a wide electrochemical window by optimizing ball-milling times to 10 h. The attractive electrochemical properties were attributed to the spontaneous formation of crystallized argyrodite during ball-milling. Similarly, Yu et al. [56] investigated the effects of mechanical milling times and subsequent heat treatment processes on electrolyte structure and were able to improve the ionic conductivity of Li₆PS₅Cl to 1.1 × 10⁻³ S cm⁻¹, revealing that further annealing can increase crystallinity and contact between grains to result in higher ionic conductivity. A major disadvantage of the ball-milling process is the long processing times. In addition, contamination as induced by the milling ball can compromise the purity of final electrolytes. In fact, it is difficult to completely distinguish ball-milling from solid-state processes because the latter requires ball-milling to mix materials in advance. Likewise, annealing after ball-milling is usually required to further improve the conductivity or crystallinity of electrolytes and the combination of ball-milling with solid-state methods can effectively reduce the reaction time of solid electrolyte synthesis.

In recent years, wet chemical methods by using liquid solvents as media have received increasing attention in the synthesis of sulfide electrolytes due to simple procedures, time-saving processes and the uniformity of resulting electrolytes. The use of the wet chemical method for the synthesis of Li₂S–P₂S₅ binary systems was first developed by Liu et al. [57] in 2013 in which the reaction between Li₂S and P₂S₅ was mediated by tetrahydrofuran to form β-Li₃PS₄. Subsequently, wet chemical methods have become more popular in the synthesis of sulfide electrolytes in which by using stable organic liquid solvents with low boiling points as reaction media, perfect homogeneity can be achieved and solvents can be removed through evaporation. Unfortunately, commonly adopted organic solvents such as anhydrous acetonitrile [58], tetrahydrofuran [57], ethyl acetate [59], N-methyl formamide [60] and 1,2-dimethoxyethane [61] are not environmentally friendly and the conductivity of electrolytes synthesized through the liquid-phase method is usually lower than those obtained through high-energy mechanical ball-milling methods. It was presumed that impurities from solvents and remaining amorphous phases precipitated at the interface between crystal particles were responsible for the low conductivity [61, 62]. Similar to the case of binary electrolytes, the purpose of the liquid-phase method in synthesizing argyrodite-type solid electrolytes is to improve homogeneity through dispersion in liquid media. Tatsumisago et al. [63] were the first to prepare argyrodite-type solid electrolytes through the liquid-phase method in which preparation conditions such as solvent, dissolution time and drying temperature were examined, resulting in a Li₆PS₅Br electrolyte with a relatively high ionic conductivity of 1.9 × 10⁻⁴ S cm⁻¹. And with the further development of their synthetic procedure, these researchers further improved the conductivity to 10⁻³ S cm⁻¹ [64]. Choi et al. [65] also prepared argyrodite-type Li₆PS₅Cl solid electrolytes using a liquid-phase process with ethyl acetate as the solvent along with subsequent heat treatment at 550 °C to obtain a Li₆PS₅Cl solid electrolyte with a relatively high conductivity of 1.1 × 10⁻³ S cm⁻¹ at room temperature. In addition, Zhou et al. [66] synthesized Li₆PS₅X (X = Cl, Br) through a liquid-phase method using tetrahydrofuran/ethanol mixtures as solvate complexes that enabled shorter reaction times as compared with the use of dimethoxyethane and acetonitrile to achieve a high conductivity of 3.9 × 10⁻³ S cm⁻¹ for a resulting Li₆PS₅Cl electrolyte. Furthermore, Yubuchi et al. [64] employed a similar method using tetrahydrofuran as a solvent and were able to obtain a Li₆PS₅Br electrolyte with a conductivity of 3.1 × 10⁻³ S cm⁻¹. The preparation of ternary electrolytes by using wet chemical methods has seldomly been reported, possibly due to the unstable characteristic of high valent cations in organic solvents. Overall, the liquid-phase method is more applicable in scalable manufacturing processes as compared with solid-state and mechanical ball-milling methods and can be applied to synthesize various compounds through appropriate solvent selection. Another unique advantage of the liquid-phase method is that resulting morphology and particles sizes can be controllably adjusted by changing reaction conditions [65]. Despite all of this, the involvement of toxic organic solutions restricts application.

2.3 Strategies for Improving Electrochemical Properties

2.3.1 Cation Substitution

Lithium ion diffusion in solid materials mainly depends on migration tunnels in which a lithium ion can diffuse from one lattice site to adjacent vacant sites as individual lithium ion hops and is governed by corresponding crystalline structures, specifically the size of migration tunnels. Higher lithium ion concentrations can also enhance conductivity because more lithium ions can participate in diffusion. Various strategies have been proposed to enhance the room-temperature conductivity of sulfide electrolytes. In general, aliovalent cation substitution has proven to be effective in which the incorporation of different valence cations can induce new vacancies and the migration of ion concentrations can increase in order to compensate for the valence imbalance of main skeletons. As examples, Ge et al. [44] reported that a 2 mol% Ni₂P-doped Li₇P₃S₁₁ electrolyte can exhibit an enhanced conductivity of 2.22  × 10⁻³ S cm⁻¹ that is 1.6 times higher than that of the pristine electrolyte. Xu et al. [67] prepared Li₇P_2.9S_10.85Mo_0.01 through high-energy ball-milling using Li₂S, P₂S₅ and high-quality MoS₂ as starting materials with a stoichiometric ratio of 7:2.9:0.1. The resultant Li₇P_2.9S_10.85Mo_0.01 electrolyte possessed a high room-temperature conductivity of 4.8 × 10⁻³ S cm⁻¹. Wu et al. [68] prepared a series of 70Li₂S·(30 − x)P₂S₅·xSeS₂ (x = 0, 0.3, 0.5, 1, 3 and 5) glass–ceramic electrolytes through ball-milling to obtain an optimal conductivity of 5.28 × 10⁻³ S cm⁻¹ at 20 °C in 70Li₂S·29P₂S₅·1SeS₂ and ascribed this conductivity to the partial replacement of P₂S₅ by SeS₂. Yamauchi et al. [69] also reported that the addition of LiBH₄ to (100 − x)(0.75Li₂S·0.25P₂S₅)·xLiBH₄ glass electrolytes can not only improve conductivity but also enlarge electrochemical windows by up to 5 V versus Li/Li⁺. Furthermore, Kraft et al. [43] systematically explored the influence of aliovalent substitution in Li_6+xP_1−xGe_xS₅I and found that with an increased Ge content, anion site disorder is induced and activation barriers for ionic migration are significantly reduced, leading to a high ionic conductivity of (18.4 ± 2.7) mS cm⁻¹ upon sintering.

2.3.2 Anion Substitution

Doping with appropriate amounts of oxides can simultaneously enhance the ionic conductivity and stability of Li metal anodes. For example, Tao et al. [52] reported that a 75Li₂S–25P₂S₅ electrolyte with 1 mol% P₂O₅ doping presented an enhanced conductivity of 8 × 10⁻⁴ S cm⁻¹ that was 56% higher than that of the undoped electrolyte. In addition, 1P₂O₅–75Li₂S–24P₂S₅ exhibited good electrochemical stability with lithium metal. Similar conclusions were also drawn by Huang et al. [70], who reported that a Li₃PO₄-doped Li₇P₃S₁₁ electrolyte exhibited higher conductivities than the pristine material (1.87 × 10⁻³ S cm⁻¹ as compared with 1.07 × 10⁻³ S cm⁻¹). In terms of ternary sulfide electrolytes, Kim et al. [71] investigated new Li₁₀SiP₂S_12−xO_x solid-state electrolytes through the substitution of O for S and reported a maximum ionic conductivity of 3.1 × 10⁻³ S cm⁻¹ at x = 0.7.

The addition of halides can also significantly improve the ionic conductivity of Li₂S–P₂S₅ sulfide electrolytes. For example, Ujiie et al. [72] compared the doping effects of different halides and found that the introduction of LiBr can effectively increase the conductivity of a Li₇P₃S₁₁ glass–ceramic electrolyte to 6.5 × 10⁻³ S cm⁻¹. (100 − x)(0.7Li₂S·0.3P₂S₅)·xLiI glass and glass–ceramic electrolytes were also prepared through mechanical milling in the composition range of 0 $\leqslant$ x (mol%) $\leqslant$ 20 by the same group in which the conductivity of the glass electrolytes increased with an increasing LiI content and resulted in an optimal conductivity of 5.6 × 10⁻⁴ S cm⁻¹ at x = 20 [73]. More importantly, they also found that the 80(0.7Li₂S·0.3P₂S₅)·20LiI glass electrolyte exhibited a wide electrochemical window up to 10 V (vs. Li/Li⁺) according to cyclic voltammetry results. However, the conductivity will decrease sharply if the LiI content increases continually.

As for Li₆PS₅X argyrodite electrolytes, the selection of different halide anions is crucial to the determination of conductive properties in which structural changes can strongly affect activation barriers and further optimize electrolytes to obtain higher ionic conductivities. After systematic investigations, the optimized electrolyte system is determined as Li₆PS₅Cl [55, 66]. Adeli et al. found that in the halide-rich Li_6−xPS_5−xCl_1+x system the Cl⁻/S²⁻ ratio has remarkable impacts on lithium ion diffusivity in the lattice in which Li_5.5PS_4.5Cl_1.5 can reach a high room-temperature conductivity of (12.0 ± 0.2) mS cm⁻¹, which was almost four-fold greater than Li₆PS₅Cl under identical processing conditions. Here, these researchers suggested that the weakened interactions between mobile lithium ions and surrounding framework anions as incurred by the substitution of Cl⁻ for S⁻ played a major role in the enhancement of lithium ion diffusivity [74].

2.3.3 Multi-element Substitution

Aside from single element doping, multi-element substitution, especially dual-cation doping, has been demonstrated to be an effective method to enhance the electrochemical performance of electrolytes due to the synergistic effect between different cations [75, 76]. For example, Yang et al. [77] studied the influence of multi-element doping on the ion channel width and the activation energy in which a Sn–Se co-doped Li₁₀GeP₂S₁₂ electrolyte was synthesized through a solid-state reaction method. Here, these researchers reported that in contrast to the limited benefits of single element doping, their Sn–Se dual-doped 5Li₂S·P₂S₅·0.6GeS_2·0.4SnSe₂ (Li₁₀Ge_0.6Sn_0.4P₂S_11.2Se_0.8) electrolyte demonstrated a high ionic conductivity of 2.75 × 10⁻³ S cm⁻¹ and extremely low activation energy of 16 kJ mol⁻¹ at room temperature, which were some of the lowest reported values for lithium ion conductors, demonstrating the potential of the Sn–Se dual-doped Li₁₀GeP₂S₁₂ as a promising electrolyte for advanced all-solid-state batteries. Kanno et al. [78] also prepared a new Li_10+δ(Sn_ySi_1−y)_1+δP_2−δS₁₂ sulfide electrolyte through dual element substitution and found that changing Sn/Si ratios and (Sn⁴⁺ and Si⁴⁺)/P⁵⁺ ratios can adjust lithium conduction channel sizes and optimize conductivity, resulting in an optimized ionic conductivity of 1.1 × 10⁻² S cm⁻¹. Although this obtained conductivity was close to the value for the original Li₁₀GeP₂S₁₂ compound, the Ge-free electrolyte is more suitable for practical application.

2.3.4 Densification

The formation of micro-cracks and pores is inevitable in the processes of mass transport and grain growth in solid electrolytes. However, low-conductivity impurities tend to concentrate in these areas and block continuous lithium ion migration tunnels. In addition, electric charge accumulating in these defects can lead to the formation of lithium dendrites, which may further elevate lithium ion migration energy barriers. Because of this, the elimination of these cracks and pores through densification is a reasonable solution to improve ionic conductivity. For example, Chu et al. [32] reported that Li₇P₃S₁₁ prepared through spark–plasma sintering showed a high ionic conductivity of 1.16 × 10⁻² S cm⁻¹ at 27 °C that was higher than that prepared through general methods. Amorphous electrolyte materials can also be used for glass–ceramic electrolytes to fill cracks and pores and lower lithium ion migration energy barriers. Minami et al. [79] further reported that by optimizing heat treatment parameters, the conductivity of a Li₇P₃S₁₁ glass–ceramic electrolyte can reach 5.2 × 10⁻³ S cm⁻¹ that was significantly higher than that of a Li₇P₃S₁₁ crystal electrolyte.

3 Sulfide Electrolyte/Oxide Cathode Interfaces

Oxide cathodes including lithium transition metal oxides and phosphates have been extensively used in commercial lithium ion batteries because of their high electrochemical potential and moderate capacity [80,81,82]. The combination of sulfide electrolytes with oxide cathodes can allow for the development of new ASSLBs with high safety and energy density. However, interfacial instability limits the application of oxide cathodes in ASSLBs. Because of this, extensive research has been conducted in this field and an overview of common interfacial behaviors between sulfide electrolytes and oxide cathodes is summarized in this section along with recent progress on strategies to improve interfacial stability.

3.1 Interface Behavior

In general, interfacial behaviors include the following aspects: (1) space charge layer effects, (2) interfacial reactions and (3) contact loss.

3.1.1 Space Charge Layer

Space charge layers are formed at the interface between two ionic conductors with significantly different lithium chemical potentials and can induce high interfacial resistances that severely affect the high-rate charge/discharge ability of ASSLBs. Because oxide cathodes are usually mixed conductors with high ion and electron conductivities, whereas sulfide electrolytes are single lithium ion conductors, contact between oxide cathodes with sulfide electrolytes will cause lithium ions to move from the sulfide electrolyte to the oxide cathode due to large chemical potential differences, resulting in the formation of space charge layers. Furthermore, because of the high electron conductivity of oxide cathodes, electrons can eliminate lithium ions on cathode sides to cause lithium ions to continue diffuse from the sulfide side to reach equilibrium. As a result, lithium-deficient space charge layers can form on the sulfide electrolyte side at equilibrium and grow after initial charging [83]. Takada et al. [84] investigated the effects of space charge layers and found that although the thickness of the space charge layer is only about 10 nm for a Li_1−xMn₂O₄/Li_3.25Ge_0.25P_0.75S₄ interface, it can lead to huge electrode resistances as high as 1 × 10⁴ Ω. In addition, Takada et al. [85] also found that space charge layers were detrimental for high-rate charge/discharge abilities in In-Li/LiCoO₂ cells in which only 4% of initial capacity was retained at 5 mA cm⁻² in an ASSLB.

3.1.2 Interfacial Reactions

The interfaces between electrodes and electrolytes in ASSLBs are different from that in conventional organic liquid electrolyte-based lithium ion batteries in which electrochemical reactions are carried out through the solid–solid interface. Based on this, many studies have been conducted to understand the possible causes of these interfacial reactions. For example, Tateyama et al. [86] investigated cation mutual diffusion properties at sulfide electrolyte/cathode interfaces using DFT + U treatment. Their results for a representative LiCoO₂/β-Li₃PS₄ solid electrolyte interface revealed that Co and P exchange energy was negative, suggesting that elemental diffusion is energetically preferable and that interfaces kinetically is stabilized after diffusion. In addition, these researchers found that repetitions of these elemental diffusions can induce detrimental transformations at interfaces, such as phase transitions or resistive-layer formations. Calculated thermodynamic data also revealed that sulfide electrolytes possess higher reaction energy with high-voltage oxide cathodes because of high cathode potentials and strong reactions between PS₄ groups and oxide cathodes to form PO₄ groups [87,88,89]. Moreover, nearly all types of sulfide electrolytes have been found to be able to react with oxide cathodes, especially high-voltage cathodes [81, 90,91,92]. Kwak et al. [91] observed the interface between LiCoO₂ and Li₂S–P₂S₅ using transmission electron microscopy in which a 10-nm interfacial layer was found (Fig. 4a) and the coexistence of Co, S and P elements was observed (Fig. 4b), indicating that elements of LiCoO₂ and Li₂S–P₂S₅ can mutually diffuse. A similar interfacial layer was also found between a high Ni cathode and an O-doped binary sulfide electrolyte 75Li₂S–22P₂S₅–3Li₂SO₄ in which large amounts of S and P penetrated into the oxide cathode and resulted in poor rate capability and capacity retention. Ternary sulfide electrolytes and argyrodite solid electrolytes have further been found to be unstable after contact with oxide cathodes. According to calculations by Mo et al. [92], Li₁₀GeP₂S₁₂ can decompose into Li₂S, Li₂SO₄, Li₃PO₄ and Li₄GeO₄ and argyrodite Li₆PS₅Cl can decompose into elemental S, lithium polysulfides, P₂S_x(x $\geqslant$ 5), phosphates and LiCl at the interface with different oxide cathodes as confirmed by XPS (Fig. 4c). Many components of interfacial layers are ionically insulating, which will further induce high interfacial resistances and even worse, some components are even electronically conductive (e.g., cobalt sulfides) and will lead to the continuous growth of ionically insulating interfacial layers [32]. As a result, interfacial reactions can explain experimentally observed high internal resistances that cause poor rate and cycle performances.

Fig. 4

a Cross-sectional high-angle annular dark-field transmission electron microscope image of the interface between a LiCoO₂ electrode and a Li₂S–P₂S₅ solid electrolyte. b Energy dispersive spectroscopy line profiles for Co, P and S elements near a LiCoO₂ electrode/Li₂S–P₂S₅ solid electrolyte interface after initial charging. Reprinted with permission from Ref. [90]. Copyright 2009, American Chemical Society. c S 2p, P 2p and Li 1s XPS spectra of a composite LiCoO₂ electrode in a LiCoO₂/Li₆PS₅Cl/Li-In half-battery: before cycling (pristine), after 25 cycles and after 25 cycles with increasing etching depths of the electrode from 5 to 20 μm. Reprinted with permission from Ref. [81]. Copyright 2017, American Chemical Society

3.1.3 Contact Loss

Although sulfide electrolytes are soft and deformable enough to form better interfacial contact than oxide electrolytes, the repeated volume change of cathodes will inevitably cause contact loss between electrolytes and active material particles, resulting in increased interfacial resistance and capacity loss [93,94,95,96]. For example, Koerver et al. [93] investigated LiNi_0.8Co_0.1Mn_0.1O₂ composite morphology using scanning electron microscopy. The as-prepared sample showed close and intimate contact between the active material and the electrolyte before electrochemical cycling (Fig. 5a). But after 50 cycles, an obvious spherical gap appeared at the interface (Fig. 5b). Here, this mechanical deflation caused contact loss between the cathode and the electrolyte during cycling and had severe negative effects on battery performance because no additional solid electrolytes can fill the emerging voids. As a result, the corresponding ASSLB showed poor rate performances and capacity retention capabilities at low current rates with no reversible capacity at 1 C and a continuous capacity loss of 1%–2% per cycle at 0.1 C.

Fig. 5

a Scanning electron microscopy (SEM) image of a composite cathode of LiNi_0.8Co_0.1Mn_0.1O₂ (NCM811) in an ASSLB before electrochemical cycling. b SEM image of the composite cathode after 50 cycles. Reprinted with permission from Ref. [93]. Copyright 2017, American Chemical Society

3.2 Interfacial Engineering

Due to poor interfacial behaviors, many approaches have been proposed to improve interfacial stability and promote the practical application of ASSLBs. These methods mainly include cathode coating, electrolyte compositional tuning and other methods.

3.2.1 Cathode Coating

The introduction of protective coatings on cathodes is the most efficient method to achieve the practical application of ASSLBs [97]. As shown in Fig. 6, an electrochemically stable interfacial coating layer can serve as a bridge to mitigate lithium chemical potential differences between electrolytes and cathodes and significantly extend electrochemical windows and improve interfacial stability [98]. Based on the many types of materials used as effective coating layers (Table 4), ideal interfacial coating layers should possess several characteristics, including: (1) coating materials should be electronically insulating and possess similar lithium chemical potentials to oxide cathodes to efficiently shield sulfide electrolytes from high electrode potentials and reduce space charge layer effects; (2) ideal interfacial coating layers should possess wider electrochemical windows than sulfide electrolytes to prevent interfacial decomposition on cathodes [87]; (3) ideal interfacial coating layers need to possess excellent lithium ion conductivity because lithium ions have to migrate through coating layers during battery cycling but also be selective toward other ions to suppress elemental diffusion; and finally, (4) because volume change (i.e., expansion and contraction) continuously occurs during battery cycling, coating materials should match with active material lattices and be mechanically plastic in response to active material deformation to maintain intimate contact between interfaces [99]. And according to these prerequisites for coating materials, various coating methods have been studied to improve battery performance in which ideal coating methods should be simple, nondestructive and cost-effective and corresponding ASSLB performances should be appreciably improved.

Fig. 6

Schematic diagram of an electrochemical window profile in an ASSLB. Reprinted with permission from Ref. [98]. Copyright 2015, American Chemical Society

Table 4 Summary of coating materials on oxide cathode surfaces in ASSLBs

3.2.1.1 Liquid-Phase Method

The liquid-phase method is the most common coating method due to associated low costs and simple procedures [91, 100, 101] in which two routes are most commonly adopted. In the first approach, a cathode is added into a precursor solution and a coating layer is obtained through co-precipitation or solution–evaporation. Although this approach is simple, economical and does not depend on expensive equipment, the thickness, homogeneity and morphology of resulting coating layers are difficult to control [90]. As for the second approach, precursor solution is sprayed onto a cathode material using a fluidized bed or through spray drying to obtain homogeneous coating layers with varying thicknesses [102]. This approach is more suitable for large-scale production processes. Based on these approaches, researchers have achieved a variety of coating layers, of which oxide-based LiNbO₃ is currently the most prevalent due to its excellent ionic conductivity (in the order of 10⁻⁵–10⁻⁶ S cm⁻¹), low electronic conductivity (below 10⁻¹¹ S cm⁻¹) [101, 103, 104], good lattice matching with oxide cathodes and wide electrochemical windows [87]. Zhang et al. [100] designed a novel double buffer layer strategy using the liquid-phase method (Fig. 7a) in which LiNi_0.8Co_0.15Al_0.05O₂ (NCA) was first designed as a core–shell structure through self-coating involving the coating of Ni-rich NCA with Al-rich NCA. Secondly, a thin LiNbO₃ layer was coated onto the core–shelled NCA (CS-NCA) through the solution–evaporation method by using metal lithium and Nb(C₂H₅O)₅ as precursors and ethanol as the solvent. Here, these researchers reported that the complete coating of the resulting CS-NCA with a thinner inactive LiNbO₃ layer was easier to achieve as compared with pristine NCA cathodes (Fig. 7b) and that after coating with the thin inactive LiNbO₃ buffer layer, space charge layers and interfacial reactions can be significantly suppressed in which CS-NCA@LiNbO₃ showed the lowest interfacial resistance (Fig. 7c). Spray coating can also be for LiNbO₃ coatings. For example, Takada et al. [84] investigated the effects of LiNbO₃ coating layers on the electrode properties of LiMn₂O₄ and LiCoO₂ cathodes through spray coating. Here, the thickness of the coating layers was controlled within 0–20 nm by altering the precursor solution amount as calculated from the specific gravity of LiNbO₃ and the BET surface area of the cathodes. As a result, these researchers found that LiNbO₃ coating layers can reduce electrode resistances by two orders of magnitude. In addition, these researchers found that the change trend in electrode resistance was similar for LiMn₂O₄ and LiCoO₂, suggesting that the space charge layer was the main cause for high interfacial resistance between high-voltage oxide cathodes and sulfide electrolytes because interfacial reactivity to sulfide electrolytes cannot be the same for LiMn₂O₄ and LiCoO₂ and that LiNbO₃ coating layers can effectively suppress space charge layers.

Fig. 7

a Schematic of the preparation process of CS-NCA@LiNbO₃. b HRTEM image of a CS-NCA@LiNbO₃ material. c Nyquist plots of charged ASSLBs by using NCA, CS-NCA and CS-NCA@LiNbO₃ cathodes after 2 cycles at 60 °C. EIS tests were conducted at room temperature. Reprinted with permission from Ref. [100]. Copyright 2018, Elsevier

In addition to LiNbO₃, other coating materials have also been successfully applied through liquid-phase methods. For example, Kwak et al. [91] employed LiInO₂–LiI as a novel coating layer in which LiI was added to improve the ionic conductivity of the coating layer and found the optimal coating amount to be 0.5 wt.% and that increases of coating amounts resulted in decreased capacity, implying that coating layers can introduce extra resistance if interlayers are too thick. Hayashi et al. [105] also investigated the electrochemical performance of LiCoO₂ coated with Li₂SiO₃ and SiO₂ and reported that batteries with Li₂SO₃-coated and SiO₂-coated LiCoO₂ showed higher discharge capacities as compared with batteries with uncoated LiCoO₂. Here, these researchers reported that the lithium ion conductive Li₂SiO₃ coating layer was more effective than inert SiO₂ in the improvement of battery performance and demonstrated that good ionic conduction was important in effective coating layers. These researchers further investigated interfacial behaviors using transmission electron microscopy and found that Co elemental diffusion from LiCoO₂ cathodes to Li₂S–P₂S₅ electrolytes can effectively be suppressed by means of Li₂SiO₃ coatings [90].

3.2.1.2 Atomic Layer Deposition (ALD)

ALD is an attractive thin-film decomposition technique that can achieve atomic-scale coatings with excellent uniformity [106,107,108,109]. Different from traditional chemical vapor deposition, ALD is a self-limiting chemical reaction process that can achieve the thickness of about 0.1–0.2 nm for each ALD layer [110, 111] to allow for the precise control of coating layer thicknesses. To investigate the effects of ALD on ASSLB cycling stability, Lee et al. [112] conducted the ALD coating of ultrathin Al₂O₃ onto LiCoO₂ particle surfaces using trimethylaluminum and H₂O as precursors at a deposition temperature of 180 °C and reported that corresponding batteries using LiCoO₂ with 2–4 ALD Al₂O₃ layers exhibited improved capacity retention as compared with those using uncoated ones. Microstructural and elemental analyses also showed that the thickness of unfavorable interfacial layers can be reduced in coated LiCoO₂ as compared with uncoated LiCoO₂ (Fig. 8a, b), suggesting greatly suppressed interfacial reactions in coated LiCoO₂. As a result, the ALD Al₂O₃-coated LiCoO₂ in this study delivered superior cycling performances with 90% capacity retention within 25 cycles, whereas only 70% capacity retention was achieved for the uncoated LiCoO₂, indicating that the ALD Al₂O₃ film was stable enough to inhibit Co dissolution and interfacial reactions. Despite these performances, the nonionic conductivity of metal oxides (e.g., Al₂O₃) can increase resistances and therefore, lithium-containing thin films should be more promising candidate materials due to their good ionic conductivity and excellent electrochemical stability. For example, Sun et al. [111] successfully achieved LiNbO_x thin-film deposition with well-controlled thicknesses and composition through ALD at a deposition temperature of 235 °C using lithium tert-butoxide and niobium ethoxide as Li and Nb sources in which different thicknesses and composition were obtained by adjusting ALD cycle numbers and subcycle ratios of Li and Nb, respectively. A uniform and continuous LiNbO_x thin film with an optimal lithium ion conductivity of 6 × 10⁻⁸ S cm⁻¹ at room temperature can be obtained at a Li/Nb subcycle ratio of 1:4. In another example, Li et al. [113] developed an ALD LiNbO_x-coated LiNi_0.8Mn_0.1Co_0.1O₂ cathode to improve interfacial stability between LiNi_0.8Mn_0.1Co_0.1O₂ and Li₁₀GeP₂S₁₂ (Fig. 8c). These researchers also investigated the thickness effect of LiNbO_x coating layers in which 2-, 5- and 10-nm-thick LiNbO_x layers were deposited onto LiNi_0.8Mn_0.1Co_0.1O₂ surfaces. As a result, they found that 5-nm-thick LiNbO_x-coated LiNi_0.8Mn_0.1Co_0.1O₂ presented optimal electrochemical performances in terms of significantly improved cycling capacity and stability (Fig. 8d) because of a stabler interface and lower interfacial resistances. With the development of ALD techniques, many lithium-containing coating materials have also been successfully achieved, such as Li₄Ti₅O₁₂ [114], LiSiO_x [115], LiPO_x [116], LiTaO_x [117], LiPON [118] and Li₃BO₃–Li₂CO₃ [119], all of which show promise as interfacial coating layers between cathodes and sulfide electrolytes because of their relatively high ionic conductivity and excellent electrochemical stability. Overall, the ALD technique can play an important role in addressing interfacial issues between oxide cathodes and sulfide electrolytes through well-controlled coatings.

Fig. 8

a HAADF TEM image of an uncoated LiCoO₂/Li_3.15Ge_0.15P_0.85S₄ interface after the 33rd charge. b HAADF TEM image of an ALD Al₂O₃-coated LiCoO₂/Li_3.15Ge_0.15P_0.85S₄ interface after the 33rd charge. Reprinted with permission from Ref. [112]. Copyright 2012, The Electrochemical Society. c HRTEM image of an ALD LiNbO_x-coated LiNi_0.8Co_0.1Mn_0.1O₂. d Cycling performances of bare and LiNbO_x–LiNi_0.8Co_0.1Mn_0.1O₂ in an ASSLB. Reprinted with permission from Ref. [113]. Copyright 2019, American Chemical Society

3.2.1.3 Physical Vapor Deposition (PVD) and Chemical Vapor Deposition (CVD)

Common PVD processes involve sputtering and evaporation in which pulsed laser deposition and magnetron sputtering are typical PVD techniques. Here, Tatsumisago et al. [120] extensively used the pulsed laser deposition method to mitigate interfacial impedance between cathodes and electrolytes. For example, they successfully deposited a 100-nm Li₃PO₄ thin film onto a 5 V class LiNi_0.5Mn_1.5O₄ cathode using pulsed laser deposition at room temperature (Fig. 9a) and reported that with an 80Li₂S–20P₂S₅ solid electrolyte, the coating of Li₃PO₄ onto the LiNi_0.5Mn_1.5O₄ cathode allowed for operation in an assembled ASSLB, whereas uncoated LiNi_0.5Mn_1.5O₄ could not achieve reversible cycling, indicating that the coating of Li₃PO₄ can significantly reduce interfacial impedance between cathodes and sulfide electrolytes. In addition, these researchers deposited the sulfide electrolyte onto an oxide cathode through the pulsed laser deposition method [121, 122] and achieved closer interfacial contact (Fig. 9b). The results indicated that a corresponding ASSLB using the sulfide electrolyte-coated LiCoO₂ delivered lower interfacial resistances and higher discharge capacities as compared with that using bare LiCoO₂.

Fig. 9

a SEM image of pulsed laser deposition Li₃PO₄-coated LiNi_0.5Mn_1.5O₄ particles. Reprinted with permission from Ref. [120]. Copyright 2015, Elsevier. b Schematic of a typical ASSLB and an ASSLB based on sulfide electrolyte-coated LiCoO₂ particles. Reprinted with permission from Ref. [122]. Copyright 2010, Elsevier. c TEM image of CVD diamond-like carbon-coated LiNi_0.8Co_0.15Al_0.05O₂. Reprinted with permission from Ref. [126]. Copyright 2016, Elsevier

Aside from pulsed laser deposition, magnetron sputtering is also an attractive physical coating method that can achieve uniform thin-film depositions on cathode surfaces through the sputtering of elemental (metallic) targets in inert gas (e.g., Ar) or inert/reactive mixture gas (e.g., Ar/O₂). This coating method has been applied successfully in organic liquid electrolyte-based lithium ion battery systems. For example, Zhou et al. [123] successfully coated Li₄Ti₅O₁₂ onto LiCoO₂ surfaces and reported significantly improved interfacial kinetics, cycling stability and rate performances for Li₄Ti₅O₁₂-coated LiCoO₂ as compared with bare LiCoO₂. Many other oxides can also be used as targets to coat cathodes to suppress interfacial reactions and improve cycling performances [124, 125]. Although the magnetron sputtering technique has yet to be widely used in ASSLB applications, it possesses significant application potential in the coating of proper interfacial materials onto oxide cathodes to improve interfacial compatibility between oxide cathodes and sulfide electrolytes.

CVD is another effective coating method that uses thermally induced chemical reactions on heated substrate surfaces. The difference between CVD and PVD is that PVD uses physical forces, whereas CVD uses chemical processes. Recently, CVD has also been applied in the coating of cathode materials to improve interfacial stability. Aihara et al. [126] used acetylene gas as a precursor and obtained a diamond-like carbon-coated (DLC) LiNi_0.8Co_0.15Al_0.05O₂ through CVD at a deposition temperature of 250 °C. Here, these researchers reported that the resulting diamond-like carbon layer possessed a thickness of around 4 nm (Fig. 9c) as well as low electronic conductivity (unlike traditional graphitic carbon) and moderate ionic conductivity, which effectively hindered side reactions and provided an alternative method to improve the performance of ASSLBs by using nonmetal oxide materials.

3.2.2 Electrolyte Composition Tuning

Although cathode coatings can effectively improve interfacial stability, intrinsic tuning through electrolyte composition tuning to improve the inherent stability of solid electrolytes is also attractive. In general, electrolytes should possess high ionic conductivity, low electronic conductivity as well as close electrochemical potentials and low lattice mismatches with cathodes; however, perfect electrolytes remain unavailable. Here, the most commonly adopted strategy to improve electrolyte performance involves oxygen anion substitution due to inherently lower lattice mismatches with cathodes and the higher electrochemical stability of oxides. In addition, substitution of oxygen atoms for sulfur atoms in sulfide electrolytes can inhibit the diffusion of oxygen from oxide cathodes to sulfide electrolytes. For example, Kawamoto et al. [127] prepared a 7Li₂O–68Li₂S–25P₂S₅ electrolyte through ball-milling and found that the resulting 7Li₂O–68Li₂S–25P₂S₅ electrolyte can effectively inhibit elemental diffusion and side reactions to reduce interfacial resistance growth and capacity fading as compared with ASSLBs using a 75Li₂S–25P₂S₅ electrolyte.

In addition to binary and ternary sulfide electrolytes, argyrodite Li₆PS₅X electrolytes have also been widely studied to improve interfacial stability with oxide cathodes. For example, Shao et al. [128] investigated the electrochemical stability of hali-chalcogenide Li₆PA₅Cl argyrodites involving a system of materials including pristine Li₆PS₅Cl, partial oxygen-substituted Li₆PO₄SCl and full oxygen-substituted Li₆PO₅Cl as formulated through extensive theoretical simulations. Here, results based on these theoretical calculations indicated that the Li₆PO₄SCl electrolyte possessed a wide electrochemical window from 0 to 2.62 V and the Li₆PO₅Cl electrolyte possessed a wider electrochemical window from 0 to 3.49 V, and that the electrochemical window of the Li₆PS₅Cl electrolyte was narrow, only being 0.49 V (1.7–2.19 V), indicating that the replacement of S by O can significantly improve the electrochemical stability window of argyrodite sulfides and suppress interfacial reactions under high cathode potentials. Zhang et al. [129] also studied an oxygen-substituted Li₆PS₅Br solid electrolyte Li₆PS_4.7O_0.3Br and reported that an ASSLB using the Li₆PS_4.7O_0.3Br as an electrolyte and NCM811 as a cathode achieved discharge capacities of 108.7 mAh g⁻¹ at 0.1 C and 47.4 mAh g⁻¹ at 0.8 C, which presented higher electrochemical performances as compared with an ASSLB using Li₆PS₅Br as the electrolyte. Here, the enhanced cycle and rate performances of the Li₆PS_4.7O_0.3Br-based ASSLB were attributed to the inhibition of space charge layer effects and interfacial reactions.

Clearly, compositional tuning is essential in the improvement of the intrinsic interfacial stability of sulfide electrolytes. Oxygen substitution, as the most prevalent modification method, can significantly improve the compatibility of oxide cathodes/sulfide-based solid electrolytes. However, mechanisms concerning how these substitutions can improve interfacial stability remain unclear and significant research is required to elucidate impact mechanisms and promote the development of novel electrolytes.

3.2.3 Others

Overall, cathode coating and electrolyte compositional tuning are two of the most promising approaches to improve interfacial stability. In addition to these two major methods, other methods have also been extensively employed to improve interfacial performance. As one method, the nano-crystallization of cathodes or electrolytes can increase contact areas between cathodes and electrolyte particles to reduce interfacial impedance. For example, Peng et al. [130] systematically investigated the fundamental lithium storage behaviors of LiNi_0.8Co_0.15Al_0.05O₂ in ASSLBs. They found that particle sizes can be decreased and surface impurities (e.g., Li₂CO₃) can be removed through ball-milling and that surface defects as caused by ball-milling such as cracks and pores can be eliminated through post-annealing. The decrease in particle size caused by ball-milling can increase the contact area between cathodes and sulfide electrolytes, and the interfacial impedances can be further reduced through post-annealing. Here, these researchers reported that their resulting LiNi_0.8Co_0.15Al_0.05O₂ in a LiNi_0.8Co_0.15Al_0.05O₂/Li₁₀GeP₂S₁₂/Li-In all-solid-state battery delivered an enhanced discharge capacity of 146 mAh g⁻¹ at room temperature. To obtain uniform and continuous electrode–electrolyte interfaces, composite electrodes can also generally be prepared through the mixing of electrode active materials, sulfide electrolytes and conduction additives. For example, Navarro et al. [131] studied the electrochemical performance of a LiNi_1/3Co_1/3Mn_1/3O₂ composite cathode using a Li₇P₃S₁₁ solid electrolyte prepared through the liquid-phase method and ball-milling in which particle sizes of about 500 nm can be obtained through the liquid-phase process, whereas particle sizes larger than 10 μm can be obtained through ball-milling. As a result, the composite cathode composed of LiNi_1/3Co_1/3Mn_1/3O₂ and small-sized Li₇P₃S₁₁ prepared through the liquid-phase method exhibited excellent electrochemical performances (Fig. 10a). It is difficult to achieve uniform and continuous interfaces in traditional composite cathodes prepared through mechanical dry-mixing with solid electrolyte particles. To address this issue (Fig. 10b), Jung et al. [132] infiltrated a Li₆PS₅Cl solution in ethanol into a cathode using a dip-coating method followed by the subsequent removal of the solvent and heat treatment at 180 °C under vacuum to obtain a densified composite cathode with very low porosity through cold pressing. Similarly, Tatsumisago et al. [133] and Navarro et al. [134] prepared composite cathodes through infiltration with sulfide electrolyte solutions to construct favorable electrode–electrolyte interfaces.

Fig. 10

a Initial discharge curves of an ASSLB using LiNi_1/3Co_1/3Mn_1/3O₂ as the active material and different particle sizes of Li₇P₃S₁₁ as ionic conductors in the composite cathode. Reprinted with permission from Ref. [131]. Copyright 2018, Elsevier. b Schematic illustrating the filtration of cathodes with solution-processable Li₆PS₅Cl. Reprinted with permission from Ref. [132]. Copyright 2017, American Chemical Society

4 Sulfide Electrolyte/Anode Interfaces

4.1 Lithium Metal Anodes

Lithium metal is considered to be the most promising candidate for anodes in lithium batteries to meet the increasing energy demands of the modern world because of its extremely high theoretical specific capacity (3860 mAh g⁻¹) and low redox potential (−3.040 V vs. SHE). However, undesirable lithium dendrite growth and low Coulombic efficiency during repeated lithium plating and stripping hamper widespread application [135, 136].

Conventionally, solid electrolytes with high lithium ion transference numbers, mechanical strength and high chemical–electrochemical stability are expected to suppress Li dendrite growth and side reactions at Li/electrolyte interfaces. However, recent research has revealed that lithium dendrites can easily form in the grain boundaries of inorganic electrolytes [137]. In addition, ternary sulfide electrolytes such as Li₁₀GeP₂S₁₂ can show severe thermodynamic and dynamic instability against metallic lithium. Moreover, the huge volume change of lithium anodes during lithium ion deposition/dissolution can lead to virtually infinite volume expansion, which further deteriorates interfacial stability. As a consequence, the large-scale commercialization of high-energy density and safe ASSLBs by using sulfide electrolytes remains challenging.

4.1.1 Interfacial Reactions

Theoretical simulations have revealed that only a few solid electrolytes are thermodynamically stable in contact with lithium due to its strong reducing ability [92]. In terms of solid electrolyte/electrode interfaces, a simplified view of the molecular orbital theory structure is schematically shown in Fig. 11a, b. In the case of the stable interface (Fig. 11a), the lowest unoccupied molar orbital (LUMO) energy of a solid electrolyte should be higher than that of the highest occupied molar orbital (HOMO) of Li (or the Fermi level of Li) to prevent the transfer of electrons from the Fermi level of Li to the conduction band of the solid electrolyte. In the case of the instable interface (Fig. 11b), the Fermi level (HOMO) of Li is higher than that of the conduction band (LUMO) of a solid electrolyte, and electrons can transfer from lithium metal to the solid electrolyte to result in inevitable interfacial reactions. Likewise, the LUMO energy of cathodes should be higher than that HOMO of sulfide electrolytes so that electron transfer from the valence band of solid electrolytes to the Fermi level of cathodes can be prevented, allowing sulfide electrolytes to be thermodynamically stable with cathodes in which cathodes reacting with solid-state electrolytes can lead to altered conduction properties and the consumption of active materials (i.e., capacity loss).

Fig. 11

Band structures for a stable and b unstable sulfide solid electrolyte/electrode interfaces. c Types of interphases between Li metal and the solid electrolyte. Reprinted with permission from Ref. [138]. Copyright 2015, Elsevier

Considering this interfacial behavior and interphase formation on solid electrolytes, Wenzel et al. [138] defined three types of interfaces based on thermodynamics (Fig. 11c), including: (1) interfaces that are thermodynamically stable against lithium in which no reactions occur at the interface; (2) interfaces that are thermodynamically unstable against lithium at which mixed conducting interphases with reasonably high ionic and electronic conductivities can form, resulting in high electronic conductivity interphases that cannot bridge the high chemical potential of lithium and leading to the continued reduction of inner fresh solid electrolytes [92]; and (3) interfaces that are thermodynamically unstable against lithium but with interphases that are kinetically stable against Li.

As for binary sulfide electrolytes with glass or glass–ceramic structures, P₂S₅, B₂S₃ or SiS₂ anions can form matrixes and Li₂S can provide lithium ion conduction. Here, increasing the Li₂S content can increase ionic conductivity due to increased mobile-ion concentrations; however, crystal structure stability also decreases. And due to the smaller electronic conductivity of interphases, binary sulfide electrolytes generally show thermodynamic instability but dynamic stability against lithium metal anodes. For example, the reaction between Li₇P₃S₁₁ and Li⁰ is Li₇P₃S₁₁ + 24 Li → Li₂S + 3Li₃P [139], which is similar to other reactants such as Li₃PS₄ [140].

For ternary systems, Wenzel et al. [139, 141] found that strong increases in the overall impedance of a Li/Li₁₀GeP₂S₁₂/Li cell as well as the solid electrolyte interphase layer can be observed as compared with a Li₂S–P₂S₅ system in which operando XPS data showed that the content of Li₃P, Li₂S and Ge⁰ interphases increased with lithium deposition time. For binary sulfide electrolytes such as Li₇P₃S₁₁ however, increases in Li₃P and Li₂S interphases resulted in an obvious decelerating trend, meaning that the interfacial decomposition of Li₁₀GeP₂S₁₂ can result in the deterioration of charge–transfer kinetics and the rapid increase of cell resistances in which the possible decomposition reaction can be expressed as: Li₁₀GeP₂S₁₂ + 20Li → 12Li₂S + 2Li₃P + Ge. Camacho-Forero et al. [142] also summarized the anion decomposition process of sulfide electrolytes through density functional theory optimizations and ab initio molecular dynamics (AIMD) simulations. As shown in Fig. 12, the PS₄³⁻ of Li₁₀GeP₂S₁₂, Li₃PS₄ and Li₇P₃S₁₁ decomposes sequentially through the breaking of each P–S bond. In most cases, final decomposition species involve S, P and some PS_x species remaining stable for long periods of time. These researchers also reported that the GeS₄ group from Li₁₀GeP₂S₁₂ decomposed similarly to PS₄ anions and that P₂S₇ anions (from Li₇P₃S₁₁) reduced to PS₃ and PS₄, which can further break down to S and P. Overall, this study provided new insights into the time evolution and transient phenomena of interfacial structures, allowing for the identification of major interfacial products through theoretical calculation. Ong et al. [33] further reported that the replacement of Ge with cheaper Si or Sn still possessed limited influences on phase stability and electrochemical stability in Li_10±1MP₂X₁₂ (M = Ge, Si, Sn, Al or P, and X = O, S or Se). Bron et al. [143] tested the time-dependent parallel resistance R_par of Li₁₀GeP₂S₁₂, Li₁₀SiP₂S₁₂, Li₁₀SnP₂S₁₂ and 95(0.8Li₂S·0.2P₂S₅)·5LiI as well as the corresponding time-dependent ionic resistance R_ion to find that the parallel resistance and ionic resistance of Li₁₀GeP₂S₁₂, Li₁₀SiP₂S₁₂ and Li₁₀SnP₂S₁₂ samples increased with contact time, reflecting continuous chemical reactions between lithium metal and sulfide electrolytes, whereas a stable solid electrolyte interphase formed at the interface between 95(0.8Li₂S·0.2P₂S₅)·5LiI and lithium metal. As for argyrodite-type solid electrolytes, Wenzel et al. [144] reported that all three compounds including Li₆PS₅Cl, Li₆PS₅Br and Li₆PS₅I decomposed in contact with lithium and that growing interphases can lead to increased interfacial resistances, but that Li₆PS₅Cl was the stablest solid electrolyte against lithium because of its slower increasing resistance trend. In situ XPS data further showed that the main interphases of Li₆PS₅Cl/Li were Li₃P, Li₂S and Li, but that LiCl can inevitably be generated based on stoichiometry in which a possible decomposition reaction is: Li₆PS₅X + 8Li → 5Li₂S + Li₃P + LiX.

Fig. 12

Anion decomposition mechanism based on density functional theory optimization and AIMD simulation. Reprinted with permission from Ref. [142]. Copyright 2018, Elsevier

4.1.2 Strategies to Inhibit Interfacial Reactions

Various approaches have been proposed to alleviate interfacial reaction issues and mainly include the design of electrolyte components and the construction of artificial protective layers.

4.1.2.1 Electrolyte Component Optimizations

Although Li₃PS₄ can exhibit better stability against lithium reduction than other sulfide electrolytes with higher valence ions, interfacial reactions still exist, which can lead to large interfacial impedances in corresponding ASSLBs during charge–discharge processes. Here, theoretical calculations and experimental results have shown that oxygen doping can improve interfacial stability in which the stabilization of crystal phases and the improvement of ionic conductivities have been widely demonstrated. In addition, simulations of sulfide electrolyte/lithium interfaces indicated that oxygen doping can prevent interfacial reactions and avoid the formation of Li₂S-like buffer layers [145]. Furthermore, the replacement of P⁵⁺ with large radius ions can improve chemical stability in addition to enhancing ionic conductivity. For example, Xie et al. [146] recently prepared a Li₃P_0.98Sb_0.02S_3.95O_0.05 electrolyte through the simultaneous introduction of Sb⁵⁺ and O²⁺ into a Li₃PS₄ structure in which structural characterizations showed that Sb and O can partially occupy P and S sites with no new phases observed, resulting in a higher ionic conductivity of 1.08 mS cm⁻¹ at room temperature and excellent stability against lithium through this dual doping method. Similarly, Liu et al. [147] successfully synthesized Li_3.06P_0.98Zn_0.02S_3.98O_0.02 through the aliovalent substitution of 2 mol% ZnO in which P⁵⁺ and S²⁻ were partially substituted by Zn²⁺ and O²⁻ separately and reported a wider electrochemical window and better stability against lithium metal.

Theoretical calculations also indicate that the Li₁₀GeP₂S₁₂ ternary electrolyte is in a metastable phase that is not stable against lithium reduction at low voltages or lithium extraction with self-decomposition at high voltages [148]. Here, Sun et al. [149] suggested that oxygen doping can effectively suppress lithium anode reduction in ternary systems and Hu et al. [150] reported that both Coulombic interactions and Van der Waals forces can contribute to the structural stability of Li₁₀GeP₂S₁₂ in which if lithium ions were partially substituted by other divalent or trivalent cations, interactions between S²⁻ and di/trivalent cations can be enhanced, allowing for the reduction of Li₁₀GeP₂S₁₂ total energy and enhanced structural stability. Sun et al. [151] also partially substituted Li⁺ by divalent Ba²⁺ cations to improve the structural stability of Li₁₀GeP₂S₁₂ and reported that due to stable interactions between Ba²⁺ and S²⁻, an optimized composition of Li_9.4Ba_0.3GeP₂S₁₂ exhibited lower polarization and better stability against lithium reduction.

Oxygen doping is also applicable for argyrodite electrolytes. Different from other sulfide electrolytes, oxygen atoms in argyrodite electrolytes prefer to substitute S atoms at free S²⁻ sites rather than those at PS₄ tetrahedral sites. For example, Zhang et al. [129] systematically investigated the electrochemical properties of Li₆PS_5−xO_xBr (0 $\leqslant$ x $\leqslant$ 1) solid electrolytes to study the effects of oxygen doping and found that without deteriorating ionic conductivities, an O-doped electrolyte can exhibit much better stability against lithium as compared with its undoped counterparts, the corresponding ASSLB using this electrolyte exhibited higher capacity and better cycling performance than that with oxygen free electrolyte. What′s more, using lithium metal as the anode, a LiNi_0.8Co_0.1Mn_0.1O₂ cathode achieved stable cycling for 92 cycles, whereas that with undoped electrolyte failed after only two cycles.

4.1.2.2 Construction of Artificial Protective Layers

The addition of artificial protective layers has been extensively studied, and due to the severe instability of ternary sulfide electrolytes/Li interfaces, recent studies have mainly focused on the modification of Li_10±1MP₂S₁₂ (M = Ge, Si, Sn). For example, Zhang et al. [152] prepared a manipulated LiH₂PO₄ protective layer on the surface of lithium foil to circumvent the intrinsic chemical instability issues of Li₁₀GeP₂S₁₂ to lithium metal through the reaction of H₃PO₄ with lithium metal (Fig. 13a). As a result, their Li/Li symmetric cell showed that the LiH₂PO₄ protective layer can play a positive role in the stabilization of the Li₁₀GeP₂S₁₂/Li interface and enhance the stability of Li₁₀GeP₂S₁₂ to lithium metal. Lithium deficiency at sulfide electrolytes/Li interfaces as caused by interfacial reactions has also been proposed as a major cause of interfacial resistance; however, it is difficult to noninvasively observe lithium distribution in solid electrolytes by using traditional probing methods. To address this, Chien et al. [153] employed ⁷Li magnetic resonance imaging to observe lithium distribution during electrochemical cycling and found that significant lithium loss occurred at the electrode/electrolyte interface upon electrochemical cycling. To address this issue, these researchers used a PEO-LiTFSI polymer electrolyte film to improve Li₁₀GeP₂S₁₂/Li interfacial stability in which results showed significant improvements in lithium distribution homogeneity as well as enhanced cycling stability in a corresponding ASSLB (Fig. 13b). Inspired by high ionic conductivity and good thermal stability in lithium ion batteries, succinonitrile-based plastic crystal electrolytes have also been adopted as buffer layers to address the instability of sulfide electrolytes to lithium metal. For example, Wang et al. [154] used a solid-state plastic crystal electrolyte to address interfacial issues between sulfide electrolytes and lithium metal and achieved significant progress toward high-energy density ASSLBs (Fig. 13c). Here, these researchers reported that if lithium metal was in direct contact with sulfide electrolytes, the sulfide electrolyte can easily reduce to form a high-resistance interphase that hindered lithium ion migration at the interface, whereas the coating of a layer of the plastic crystal electrolyte as an intermediate layer at the interface between lithium metal and the sulfide electrolyte can greatly suppress interfacial reactions between the sulfide electrolyte and lithium metal. As a result, a corresponding ASSLB with a LiFePO₄ cathode delivered an enhanced initial capacity of 148 mAh g⁻¹ at 0.1 C and 131 mAh g⁻¹ at 0.5 C. These researchers also reported that the chemical compatibility between the sulfide electrolyte and the plastic crystal electrolyte ensured the long-term cycling stability of the ASSLB. Gao et al. [155] further dripped a highly concentrated liquid electrolyte into a Li₁₀GeP₂S₁₂/Li interface and found that nanocomposites derived from commercial organic or inorganic lithium salts can act as interphases and that these composite interphases not only possessed high ionic conductivity but also cleanly separated lithium and Li₁₀GeP₂S₁₂ to result in stable interfaces. Overall, these promising results indicate that the rational design of buffer layer composition and structure to enhance interfacial compatibility between lithium and sulfide electrolytes is important.

Fig. 13

a Schematic of the preparation of an in situ LiH₂PO₄ protective layer and a LiCoO₂/Li₁₀GeP₂S₁₂/LiH₂PO₄-Li ASSLB with an optimized structure. Reprinted with permission from Ref. [152]. Copyright 2018, American Chemical Society. b Lithium density profiles at different depths of electrochemically cycled Li₁₀GeP₂S₁₂ pellets. Reprinted with permission from Ref. [153]. Copyright 2018, American Chemical Society. c Schematic of ASSLBs with plastic crystal electrolyte interlayers. Reprinted with permission from Ref. [154]. Copyright 2019, Wiley–VCH

To take advantage of individual components, inorganic–organic composite electrolytes have also been frequently adopted to improve rigid interfacial issues and compatibility with lithium metal. For example, Ju et al. [156] fabricated a poly(vinyl carbonate) and Li₁₀SnP₂S₁₂ composite electrolyte (PVCA–LSnPS) through in situ polymerization and reported that the resulting PVCA–LSnPS composite possessed an indispensable combination of high ionic conductivity, wide electrochemical windows and large lithium ion transference numbers. More importantly, this composite electrolyte possessed good compatibility with lithium metal as engineered through in situ polymerization, leading to significant interfacial impedance reductions in corresponding solid-state Li–Li symmetric cells.

4.1.3 Lithium Dendrites

According to Li dendrite growth mechanisms in organic liquid electrolyte/polymer electrolyte batteries, researchers have generally proposed that the growth of dendrites can be physically limited by high shear modulus solid electrolytes (G_{solid electrolyte} > 1.8G_Li) [157]. However, recent research has shown that lithium dendrites are also found in ASSLBs using sulfide electrolytes, presenting a challenge in the application of lithium anodes in which lithium dendrites in ASSLBs are believed to be associated with the inherent physical defects of sulfide electrolytes or corresponding deterioration during cell cycling.

In general, lithium dendrite formation tends to occur under several situations, including: (1) along the grain boundary and voids of solid electrolytes in which low lithium ion diffusivity at grain boundaries was reported to be the intrinsic reason for dendrite formation along grain boundaries [158]; (2) pre-existing defects in the surface of or inside solid electrolytes, such as cracks, in which generated stress can further extend cracks and further promote lithium dendrite propagation [159]; (3) inhomogeneous lithium plating due to insufficient interfacial contact between lithium and sulfide electrolytes [160]; and (4) high electronic conductivity solid electrolytes that will accelerate dendrite formation and growth.

In one example, Nagao et al. [158] investigated lithium deposition/dissolution in an ASSLB with an 80Li₂S·20P₂S₅ electrolyte through in situ SEM using a stainless steel current collector with chamfered corners in which to observe the interface between the electrolyte and the electrode, SEM observations were carried out with a stage tilt at an angle of 30° (Fig. 14a). As a result, these researchers found that current density played a crucial role in deposition behavior in which at currents above 1 mA cm⁻², lithium deposited locally and triggered large cracks in the solid electrolyte that led to the short circuiting of the cell. In addition, these researchers reported that deposited lithium can push the solid electrolyte out to form pillared deposits 6.6 μm in length at the interface between the electrolyte and the electrode after short circuiting (Fig. 14b), whereas at current densities lower than 0.05 mA cm⁻², lithium metal deposited uniformly on the solid electrolyte surface and no cracks or pillared deposits formed, suggesting that the homogeneous deposition of lithium and the suppression of lithium growth along grain boundaries were important to achieve highly reversible lithium deposition and dissolution. These researchers also reported that the softening of amorphous electrolytes followed by pressurization can allow for solid electrolytes with fewer grain boundaries and pores, which is beneficial to the inhibition of dendrite growth.

Fig. 14

a Cell schematic for in situ SEM observation and b lithium deposition at 5 mA cm⁻² for 10 min after short circuiting. Reprinted with permission from Ref. [158]. Copyright 2013, The Royal Society of Chemistry. c Transmission optical microscopy and d fracture surface SEM images of a lithium metal network in a solid-state electrolyte. Reprinted with permission from Ref. [161]. Copyright 2017, Wiley–VCH. e Schematic of a Li/Li₆PS₅Cl interface cycled at an overall current density above critical current density. Reprinted with permission from Ref. [163]. Copyright 2019, Nature Publishing Group

Chiang et al. [161] further developed a new method to monitor lithium penetration in different types of solid electrolytes during lithium electrodeposition and found that the onset of lithium dendrite formation depended on the roughness of the solid electrolyte surface, particularly the size and density of defects. More importantly, these researchers found based on typical transmission optical microscopy and fracture surface SEM imaging that above a critical current density, lithium plating in surface cracks can produce crack-tip stress that can further drive crack propagation and extend to complex networks (Fig. 14c, d), meaning that the failure mechanism for brittle electrolytes is Griffith-like and that the suppression of lithium dendrite formation in solid-state electrolytes requires scrupulous attention to the minimization of interfacial defects.

Recently, Han  et al. [162] also investigated the origins of dendrite formation in solid electrolytes by monitoring the dynamic evolution of lithium concentration profiles in different solid electrolytes during lithium electrodeposition. They found that lithium metal can directly deposit inside Li₇La₃Zr₂O₁₂ and Li₃PS₄ solid-state electrolytes, whereas no dendrites were found in LiPON, suggesting that high electronic conductivity was most likely responsible for dendrite formation in solid electrolytes. Because of this, electronic conductivity is considered to be another critical criterion in the evaluation of solid-state electrolytes. In another study, Kasemchainan et al. [163] investigated the process of lithium plating/stripping at a Li/Li₆PS₅Cl interface and suggested that critical current density for stripping was the crucial factor affecting dendrite growth in which if stripping current removed lithium faster than what can be supplied, voids would form at the interface, leading to preferential lithium deposition in subsequent lithium plating and the formation of lithium dendrites near these voids (Fig. 14e). These researchers also found that pressure-dependent creep rather than lithium diffusion-dominated lithium transport at the interface and that considerable pressure was crucial for homogeneous lithium deposition and can effectively increase critical current density to achieve high power density or large rate capability.

4.1.4 Modifications to Alleviate Lithium Dendrite Formation

Due to high reactivity, the suppression of lithium dendrite growth in sulfide electrolytes is challenging because the mechanisms for “unexpected” dendrite formation are unclear and currently, methods to restrain lithium dendrite growth are mainly focused on the optimization of electrolyte composition and the application of artificial protective layers. In terms of the optimization of electrolyte composition, LiI has been demonstrated to be an effective additive to suppress lithium dendrite growth. For example, Suyama et al. [164] investigated the lithium dissolution/deposition behaviors of an all-solid-state lithium symmetric cell using a Li₃PS₄–LiI electrolyte system and their electrochemical tests showed that the introduction of LiI enhanced lithium dissolution/deposition performances in which the optimized electrolyte (46 mol% LiI) delivered ultralong cycling for 3400 h at 1.25 mA cm⁻² without short circuiting and with a high areal capacity of 7.5 mAh cm⁻². Here, structural analysis revealed that electrolyte reduction by lithium metal can be inhibited with the addition of LiI in which good interfacial contact was maintained even after prolonged cycling, clearly demonstrating that the introduction of LiI improved the tolerance of sulfide electrolytes to lithium metal reduction. Similarly, Han et al. [160] demonstrated that lithium dendrites in a Li₂S–P₂S₅ glass electrolyte can be effectively suppressed by incorporating LiI into the electrolyte in which the ionically conductive but electronically insulating characteristic of LiI can improve lithium ion migration at the interface to suppress dendrite growth. As a result, these researchers reported significantly improved critical current densities, reaching 3.90 mA cm⁻² at 100 °C as well as stable cycling for 200 h at 1.50 mA cm⁻² in the corresponding lithium symmetric battery.

Aside from electrolyte composition optimization, the construction of buffer layers between lithium metal and electrolytes has also been demonstrated to be an effective method to suppress lithium dendrite formation. For example, Xu et al. [165] demonstrated that the coating of a uniform LiF (or LiI) interfacial layer at a Li/Li₇P₃S₁₁ interface can effectively inhibit lithium dendrite growth in which LiF or LiI can form on lithium surfaces after exposing lithium to methoxyperfluorobutane and I₂ gas at 150 °C (Fig. 15a). These researchers also reported that LiF interlayers exhibited much higher capabilities than LiI in the suppression of lithium dendrite formation due to the higher interfacial energy of LiF. More importantly, these researchers reported that even if the interfacial layer was broken by lithium dendrites, fresh lithium dendrites will be consumed by penetrative methoxyperfluorobutane (HFE) to form new solid electrolyte interphases (Fig. 15b). And as a result of this modification, a corresponding lithium symmetric battery showed enhanced cycling performances for over 200 cycles at 0.5 mA cm⁻² and 0.1 mAh cm⁻² and an assembled LiNbO₃@LiCoO₂/Li₇P₃S₁₁/LiF@Li ASSLB exhibited a reversible discharge capacity of 118.9 mAh g⁻¹ at 0.1 mA cm⁻² and retained 96.8 mAh g⁻¹ after 100 cycles, demonstrating greatly enhanced performances as compared with ASSLBs using pure lithium metal anodes. Wang et al. [166] further developed an organic–inorganic hybrid interlayer (alucone) between Li₁₀SnP₂S₁₂ and lithium metal through molecular layer deposition (Fig. 15c) and found that the artificial interfacial layer can serve as an SEI to intrinsically block electron transfer at the Li/Li₁₀SnP₂S₁₂ interface to completely suppress interfacial reactions and lithium dendrite growth. As a result, an ASSLB employing the modified lithium metal anode exhibited smaller polarization, higher capacity and longer cycle lifes than that using bare lithium metal. More importantly, these researchers reported that their organic–inorganic hybrid molecular layer deposition method resulted in better flexibility than inorganic coatings and can better accommodate stress/strain as caused by volume change.

Fig. 15

a Schematic of LiF and LiI coating processes on lithium metal surfaces. b Schematic of lithium stripping/plating behaviors in bare lithium and LiF or LiI coated lithium with HFE or iodine infiltrated electrolytes. Reprinted with permission from Ref. [165]. Copyright 2018, Wiley–VCH. c Structure of a solid-state battery and chemical structure of the alucone layer. Reprinted with permission from Ref. [166]. Copyright 2018, Wiley–VCH

4.2 Other Anodes

Challenges such as uncontrollable lithium dendrite growth, interfacial reactions as well as volume effects facing ASSLBs are difficult to resolve in the short term. However, many anode materials used in traditional lithium ion batteries have been explored for application in ASSLBs, such as graphite, silicon and Li₅Ti₇O₁₂. In addition, the high environmental stability of these materials can lower the manufacturing costs of ASSLBs.

4.2.1 Graphite and Silicon

In general, the main issue between graphite or silicon anode powder materials and sulfide electrolytes is inferior contact. For example, Takeuchi et al. [167] prepared a graphite–sulfide electrolyte composite anode through spark–plasma–sintering (SPS) in which electrochemical tests showed that a corresponding ASSLB with a Li₂S cathode delivered a discharge capacity of 750 mAh g⁻¹ (normalized by Li₂S mass). And although this capacity was lower than that of a battery employing an In/Li₂S anode (ca. 920 mAh g⁻¹), the estimate energy density was higher due to the low redox potential of graphite (0.1 V vs. Li/Li⁺). More importantly, the graphite–sulfide composite anode exhibited enhanced rate capabilities as compared with the simply blended graphite–sulfide anode due to the intimate contact as a result of the SPS method. The selection of suitable electrolytes can also maintain electrochemical stability. For example, Takada et al. [168] designed a unique bilayer electrolyte structure in which a LiI–Li₂S–P₂S₅ glass electrolyte was used at the anode to suppress electrochemical reduction and Li₃PO₄–Li₂S–SiS₂ or Li₂S–GeS₂–P₂S₅ glass electrolytes were used at the cathode to suppress electrochemical oxidation. As a result, electrochemical testing showed no significant reactions between the LiI–Li₂S–P₂S₅ glass electrolyte and the graphite anode during lithium intercalation/deintercalation. This construction also allowed for the use of graphite as an anode and LiCoO₂ as a cathode in an ASSLB to achieve comparable performances to commercial lithium ion batteries. Similarly, Yamamoto et al. [169] found that Li₇P₂S₈I, a type of LiI–Li₂S–P₂S₅ electrolytes, exhibited wide electrochemical windows and was suitable to be paired with graphite anodes in which a half-cell with a graphite anode delivered an initial discharge of 372 mAh g⁻¹ and highly reversible capacities.

Despite the high theoretical capacity of silicon (4200 mAh g⁻¹ for Li_4.4Si), large volume change (~ 400%) during lithiation and de-lithiation as well as poor electronic conductivity must be overcome to allow for application. Based on this, great efforts have been devoted to the use of Si as anodes in traditional lithium ion batteries with strategies including the nano-crystallization of silicon particles, the construction of stress–relief buffer matrixes and the use of physical compartments to accommodate volume expansion [170]. Overall, the underlying mechanism of these strategies is to enhance adhesion between active materials and conducting matrixes to maintain better electrical contact during repeated volume change. In recent years, researchers have also applied Si in solid-state batteries. However, most studies have been based on dry-mixing with extremely low mass loadings and resulted in poor electrical contact and low energy density. Kim et al. [171] prepared a sheet-like composite anode by infiltrating a traditional Si anode with a solution-processable Li₆PS₅Cl electrolyte (Fig. 16a) and reported that after the evaporation of the solvent, the electrolyte can solidify on the surface of the silicon particles to allow for compact ionic contact and favorable ionic transport. EDX mapping of the anode further showed that the penetration of the Li₆PS₅Cl solution into the Si electrode allowed for intimate contact between Si and Li₆PS₅Cl and electrochemical testing showed that a corresponding half-cell delivered a high capacity of 3246 mAh g⁻¹ at 0.25 mA cm⁻² that was much better than that of a traditional dry-mixed Si anode. In addition, a corresponding all-solid-state full cell using a LiCO₂ cathode was able to provide a high-energy density of 338 Wh kg⁻¹. Many results have also shown that encapsulating Si particles in robust matrixes can further suppress the volumetric expansion and pulverization of Si. Based on commercial considerations, Lee et al. [172] prepared a silicon–carbon composite derived from the industrial waste product coal–tar–pitch as an anode in an ASSLB in which the pyrolysis of coal–tar–pitch produced a mixed conducting amorphous carbon to encapsulate Si particles. Here, structural characterizations showed that the pitch-derived carbon was robust enough to suppress the volume change of silicon and resulted in impressive electrochemical properties in which an optimized sample displayed stable specific capacities of 653.5 mAh g⁻¹ (a mass electrode) and 1089.2 mAh g⁻¹ (a mass Si–C) with Coulombic efficiency > 99%.

Fig. 16

a Schematic and photographs illustrating the infiltration process of conventional Si composite electrodes with solution-processable solid electrolytes. Reprinted with permission from Ref. [171]. Copyright 2019, Elsevier. b Schematic of the synthesis of a Li₄Ti₅O₁₂@LPS + VGCFs composite electrode and the assembly of a solid-state battery. Reprinted with permission from Ref. [173]. Copyright 2018, Elsevier. c Schematic of the infiltration of Li₆PS₅Br into a Li₅Ti₇O₁₂@CNT electrode by using the liquid-phase technique. Reprinted with permission from Ref. [133]. Copyright 2019, Elsevier

4.2.2 Li₄Ti₅O₁₂

Considering the good reversibility and negligible volume change of Li₄Ti₅O₁₂ in charge–discharge processes, the combination of Li₄Ti₅O₁₂ anodes with sulfide electrolytes should enable ASSLBs with high rate and cycle performances. And to enhance electrochemical performances, many strategies have been conducted to increase the contact area and enhance the compactness of contact interfaces. Similar to graphite and silicon anodes, these strategies mainly focus on liquid-phase or infiltration methods. For example, Cao et al. [173] developed a liquid-phase approach to in situ coat 70Li₂S–30P₂S₅ onto Li₄Ti₅O₁₂ surfaces using mineral spirit as a solvent (Fig. 16b) and reported that a cell employing the Li₄Ti₅O₁₂@70Li₂S–30P₂S₅ and vapor-grown carbon fibers as the working electrode exhibited excellent rate capacities and cycling stability. In addition, these researchers successfully prepared a Li₆PS₅Cl-coated Li₄Ti₅O₁₂ using a kind of paint as the solvent [174] in which SEM and energy dispersive spectroscopy mapping showed the homogenous coating of nano-sized Li₆PS₅Cl onto the Li₄Ti₅O₁₂ surface, allowing for a stable interface between the active material and the solid electrolyte and resulting in low interfacial resistances and excellent electrochemical activities. Similar to that of silicon anodes, the infiltration method can also be used to enhance compactness and facilitate electric/ionic conduction. For example, Yubuchi et al. [133] prepared a homogenous Li₄Ti₅O₁₂-carbon nanotube-Li₆PS₅Br composite electrode using infiltration followed by cold pressing (Fig. 16c) to achieve an electrochemically active interface with a large contact area and favorable electric/ionic conduction pathways, resulting in a corresponding ASSLB demonstrating an improved capacity of 100 mAh g⁻¹ at 4 °C and 100 °C.

5 Summary and Outlook

ASSLBs based on sulfide electrolytes are attracting significant interest due to their potential to address safety concerns and improve energy density. The metal oxide cathodes have been extensively used in commercial lithium ion batteries. Based on this, this review has comprehensively presented the progress in ASSLBs using sulfide electrolytes and oxide cathodes. In addition, related interfacial issues at sulfide electrolyte/oxide cathode interfaces as well as unstable anodic interfaces have been systematically discussed. Moreover, major challenges as well as corresponding improvement strategies for ASSLBs using lithium anodes have been proposed. Different sulfide electrolytes have also been summarized and ionic conductivities have been compared based on the type and the preparation method. In terms of practical commercialization, the stability of sulfide electrolytes was also considered. As for oxide-based cathodes, various types of interfacial behaviors including space charge layer effects, interfacial reactions and contact losses were classified and discussed in detail.

Overall, to improve interfacial stability and promote practical application, various strategies have been adopted, mainly including cathode coating and electrolyte compositional tuning. New strategies such as the reduction of particle size and the mixing of electrolytes with electrodes through ball-milling have also been proposed to enhance the performance of ASSLBs. As for anodes, main methods including artificial solid electrolyte interphase construction and electrolyte component optimization have routinely been used to minimize interfacial resistance. And despite Li₃PS₄ and derivatives exhibiting relative stability with lithium metal, the capacity utilization and working current density of lithium metal anodes are far from meeting practical requirements. Therefore, the resolution of interfacial contact and dendrite formation issues for lithium metal anodes used with sulfide electrolytes remains challenging.

Despite these challenges, ASSLBs with oxide-based cathodes and sulfide electrolytes are still considered to be promising candidates for next-generation energy storage systems in which enhancements in electrolyte conductivity and interfacial stability are still needed. Here, the adoption of straightforward strategies alone is insufficient for the practical application of ASSLBs and combined solutions such as suitable electrolyte components and artificial interfaces between electrolytes and electrodes need to be considered to achieve joint modifications for high-performance ASSLBs. Several design principles and solutions regarding ASSLBs with oxide-based cathodes and sulfide electrolytes are listed as follow:

• (1) New sulfide electrolytes. Although sulfide electrolytes possess unparalleled ionic conductivities, poor electrochemical and chemical stability greatly limits practical application. Therefore, the exploration for new electrolytes is needed to further improve the performance of corresponding ASSLBs. Here, both ionic and electronic conductivities need to be taken into consideration and recently, theoretical calculations have been reported to be an important tool in the design of new electrolytes by allowing for the prediction or selection of suitable components and physicochemical properties in advance. Overall, a combination of experimental results with theoretical calculations is necessary to explore new electrolytes.

• (2) Optimized electrolyte/cathode interfaces. The different chemical potentials of lithium ions between sulfide electrolytes and oxide cathodes can induce high interfacial resistance that can severely affect the high-rate capability of ASSLBs. Therefore, optimized interfaces are necessary to act as bridges to mitigate lithium chemical potential differences between electrolytes and cathodes. These interfaces should possess high ionic conductivity to ensure smooth lithium ion migration as well as good compatibility with electrodes and stability to shield sulfide electrolytes from high electrode potentials.

• (3) Dendrite control in lithium anodes. Low relative density, grain boundaries, defects and high electronic conductivity in sulfide electrolytes are believed to be the reasons for lithium dendrite formation. Here, the mechanisms of lithium dendrite formation in solid electrolytes are being actively studied and a single method to suppress all aspects of lithium dendrite formation remains lacking. Therefore, combined modifications are better choices to achieve dendrite-free anodes in ASSLBs, such as the construction of artificial interfaces, the design of 3D lithium matrixes and the adoption of lithium alloys. In addition, the development of new sulfide electrolytes with high ionic conductivity and low electronic conductivity can also inhibit dendrite formation.

• (4) Optimized battery assembly. Well-designed battery assembly technologies are also important. For example, suitable pressures are crucial to achieve homogeneous lithium deposition and can effectively increase critical current densities to achieve high power density ASSLBs.