Review on Interface and Interphase Issues in Sulfide Solid-State Electrolytes for All-Solid-State Li-Metal Batteries

All-solid-state batteries have emerged as promising alternatives to conventional Li-ion batteries owing to their higher energy density and safety, which stem from their use of inorganic solidstate electrolytes instead of flammable organic liquid electrolytes. Among various candidates, sulfide solid-state electrolytes are particularly promising for the development of high-energy all-solid-state Li metal batteries because of their high ionic conductivity and deformability. However, a significant challenge remains as their inherent instability in contact with electrodes forms unstable interfaces and interphases, leading to degradation of the battery performance. In this review article, we provide an overview of the key issues for the interfaces and interphases of sulfide solid-state electrolyte systems as well as recent progress in understanding such interface and interphase formation and potential solutions to stabilize them. In addition, we provide perspectives on future research directions in


Introduction
For decades, rechargeable lithium-ion batteries (LIBs) using organic liquid electrolytes have been commercially used as a power source for portable electronic devices [1][2][3], and their use has recently expanded to mid-large-scale electrochemical energy storage applications, including electric vehicles (EVs) and grids [4,5]. For the success of EVs, battery technology must surpass the current limitation of driving ranges by achieving a larger capacity. According to a global technical goal for 2030, the specific energy of secondary batteries must reach 500 Wh kg −1 , which is much higher than the 246 Wh kg −1 of today's typical EV batteries used in the Tesla Model 3 [6][7][8][9]. To achieve such a high energy density, a Li-metal anode is considered to be an unrivaled component, owing mainly to its extremely high theoretical specific capacity (3860 mAh g −1 ) and lowest reduction potential (−3.040 V vs. the standard hydrogen electrode). Safety issues have become more important than ever as battery sizes have increased. However, current LIB technology using liquid electrolytes cannot meet the requirements for the use of a Li-metal anode or safety [10,11].
In this respect, all-solid-state batteries (ASSBs), which use inorganic solid-state electrolytes (SSEs) instead of flammable organic liquid electrolytes, have emerged as promising alternatives to conventional LIBs. SSEs offer several advantages over conventional liquid electrolytes. First, ASSBs are stable even in extreme environmental conditions such as high pressure, temperature, overcharge, or external shock, and there is much less risk of ignition or explosion [12,13]. Second, it is believed that ASSBs can use a Li-metal anode since the sufficient stiffness of SSEs can mitigate the internal short circuit by suppressing the Li dendrite growth during the electrochemical cycles [14]. Third, the volumetric energy density can be improved by reducing its volume since there is no need to pack each cell and a simpler design through bipolar stacking is available for ASSBs [15,16]. Despite these advantages of ASSBs, important challenges remain to be addressed. Many SSEs cell and a simpler design through bipolar stacking is available for ASSBs [15,16]. Despite these advantages of ASSBs, important challenges remain to be addressed. Many SSEs display lower ionic conductivity at room temperature when compared to liquid electrolytes. Moreover, it is difficult to ensure the ion-transfer channel between the SSE and electrodes because of their unstable interfaces.
Of the various kinds of SSEs, sulfide SSEs have attracted significant attention due to their high ionic conductivity (10 −2 -10 −4 S cm −1 ), which is comparable to that of organic liquid electrolytes [11,17,18]. Figure 1 summarizes the ionic conductivities of sulfide SSEs. Although the high ionic conductivity and deformability make sulfide compounds promising SSEs, several important challenges remain. Sulfide SSEs have an intrinsically narrow electrochemical stability window, which results in undesired (electro)chemical reactions between the sulfide SSE and electrodes during battery cycling [19], forming unstable interphases. Such interphase formation leads to operation failures of ASSBs. Therefore, considerable efforts are now focused on improving the interfacial stabilities between sulfide SSEs and electrode materials.
In this review, we present an overview of the limitations and recent achievements in the development of sulfide SSEs, with a particular focus on the stability issue of sulfide SSEs with lithium-metal anodes and various cathode active materials. We will not discuss the conductivities of sulfide SSEs, as this topic is already well documented in previous review articles [11,[20][21][22][23]. We further discuss recently proposed techniques to achieve a favorable interface between sulfide SSEs and electrodes in ASSBs. This review aims to provide not only a comprehensive description of the developments of sulfide SSEs but also insight into potential directions of future research. The magnified view compares the conductivities at room temperature by their structure [17,. Note that the light blue line is a curve from organic liquid electrolytes of 1 M LiPF6/EC-PC (50:50 vol%) for comparison [46]. The magnified view compares the conductivities at room temperature by their structure [17,. Note that the light blue line is a curve from organic liquid electrolytes of 1 M LiPF 6 /EC-PC (50:50 vol%) for comparison [46].

Electrochemical Stability of Sulfide SSEs
The greatest challenge for sulfide SSEs is their very narrow electrochemical stability window, which leads to poor compatibility with electrodes. Although early works claimed that sulfide SSEs have a good electrochemical stability window of 0-4 V (vs. Li metal) based on cyclic voltammetry (CV) measurements [17,28,47], it is highly likely that these works overestimated their electrochemical stability. In the CV tests, the planar electrode geometry limits the contact area, leading to small currents, which can show kinetically extended electrochemical stability instead of a thermodynamic stability window. Later, Han et al. and Dewald et al. proved that the electrochemical stability windows of sulfide SSEs are indeed much narrower than those determined by previous CV tests [48,49]. To increase the 'active' contact area, they mixed Li 10 GeP 2 S 12 (LGPS) thiophosphate and carbon and measured the CV of the mixture, which presented a much lower oxidation limit (<2.5 V vs. Li). In the mixture of LGPS-carbon, LGPS provides Li-transport pathways and carbon works for electron transfer. Therefore, such a configuration could show a 'practical' electrochemical stability window. This is critical since SSEs will encounter carbon additives in the cathode composite of ASSBs. Computational works have also demonstrated that sulfide SSEs have a narrow electrochemical stability window when compared with oxides and halides, as shown in Figure 2 [19,50,51]. Such a narrow electrochemical stability window is a major practical disadvantage of sulfide SSEs since the electrolyte must be stable over a wide range of lithium potentials between the anode chemical potential (0 eV/atom vs. Li) and the potential set by the cathode, which is near <−4.0 eV/atom for typical cathodes.

Electrochemical Stability of Sulfide SSEs
The greatest challenge for sulfide SSEs is their very narrow electrochemical stability window, which leads to poor compatibility with electrodes. Although early works claimed that sulfide SSEs have a good electrochemical stability window of 0-4 V (vs. Li metal) based on cyclic voltammetry (CV) measurements [17,28,47], it is highly likely that these works overestimated their electrochemical stability. In the CV tests, the planar electrode geometry limits the contact area, leading to small currents, which can show kinetically extended electrochemical stability instead of a thermodynamic stability window. Later, Han et al. and Dewald et al. proved that the electrochemical stability windows of sulfide SSEs are indeed much narrower than those determined by previous CV tests [48,49]. To increase the 'active' contact area, they mixed Li10GeP2S12 (LGPS) thiophosphate and carbon and measured the CV of the mixture, which presented a much lower oxidation limit (<2.5 V vs. Li). In the mixture of LGPS-carbon, LGPS provides Li-transport pathways and carbon works for electron transfer. Therefore, such a configuration could show a 'practical' electrochemical stability window. This is critical since SSEs will encounter carbon additives in the cathode composite of ASSBs. Computational works have also demonstrated that sulfide SSEs have a narrow electrochemical stability window when compared with oxides and halides, as shown in Figure 2 [19,50,51]. Such a narrow electrochemical stability window is a major practical disadvantage of sulfide SSEs since the electrolyte must be stable over a wide range of lithium potentials between the anode chemical potential (0 eV/atom vs. Li) and the potential set by the cathode, which is near <−4.0 eV/atom for typical cathodes.  Theoretically driven electrochemical stability ranges of various electrolyte materials. Note that the yellow-colored sulfide electrolytes show low upper limits of less than 3.0 V. Reproduced with permission from reference [19], Copyright 2016 American Chemical Society.

Interfacial Stability
The narrow electrochemical stability window of sulfide SSEs leads to poor compatibility with both the anode and cathode. At the interface with electrodes, sulfide SSEs suffer from significant material degradation when electron-transport pathways are provided. Figure 3 summarizes the interfaces in sulfide-SSE-based ASSBs. With a Li metal anode, sulfide SSEs are reduced and form a solid-electrolyte interphase (SEI) layer between the Li metal and SSE. Depending on the reduction products, the interface can be passivated or cause the continuous reduction of the SSE, thus leading to battery failure [19]. Another important issue at the anode side is dendritic Li growth during the charging of ASSBs. Although it was believed that rigid SSEs could prevent dendritic growth in the early stages of research, many reports have since shown Li dendritic growth and penetration through SSEs [52][53][54][55]. The situation is much more complicated on the cathode composite side. A cathode active material needs to be homogeneously mixed with an SSE and carbon to provide Li-ion and electron pathways. As a result, the cathode composite has more interfaces, namely cathode/SSE, carbon/SSE, and cathode/carbon. We will discuss recent findings of interfacial degradation mechanisms and approaches to overcome them in the following sections.

Interfacial Stability
The narrow electrochemical stability window of sulfide SSEs leads to poor compatibility with both the anode and cathode. At the interface with electrodes, sulfide SSEs suffer from significant material degradation when electron-transport pathways are provided. Figure 3 summarizes the interfaces in sulfide-SSE-based ASSBs. With a Li metal anode, sulfide SSEs are reduced and form a solid-electrolyte interphase (SEI) layer between the Li metal and SSE. Depending on the reduction products, the interface can be passivated or cause the continuous reduction of the SSE, thus leading to battery failure [19]. Another important issue at the anode side is dendritic Li growth during the charging of ASSBs. Although it was believed that rigid SSEs could prevent dendritic growth in the early stages of research, many reports have since shown Li dendritic growth and penetration through SSEs [52][53][54][55]. The situation is much more complicated on the cathode composite side. A cathode active material needs to be homogeneously mixed with an SSE and carbon to provide Li-ion and electron pathways. As a result, the cathode composite has more interfaces, namely cathode/SSE, carbon/SSE, and cathode/carbon. We will discuss recent findings of interfacial degradation mechanisms and approaches to overcome them in the following sections.

Interface with Li-Metal Anode
Due to the inherently narrow electrochemical stability window, many sulfide SSEs decompose spontaneously simply by contacting a Li-metal anode [19]. Several efforts have been made to evaluate Li-metal/sulfide-SSE interfaces both theoretically and experimentally [18,50,[56][57][58][59][60][61][62][63][64][65][66][67]. The interfaces formed between Li metal and the SSE can be classified into the following three types [68]. Type I interfaces are stable and ideal interfaces formed when the SSE does not react with Li metal electrochemically (Figure 4a). Most binary compounds, including Li2S and Li3P, form this 'type I' interface with Li metal. However, if the SSE is not electrochemically stable with Li metal, as most of the sulfide compounds are, the interphase will form at the interface as products of the SSE decompose. If the interphase material is highly conductive for both ionic and electronic transport, nonpassivating interphases will form, and the reduction of SSE continuously occurs (Type II). Thus, the reaction front rapidly propagates into the SSE, leading to a short circuit and failure of the SSB (Figure 4b). When the interphases exhibit ionic conductivity but not electrical conductivity (Type III), the interface becomes relatively stable (passivating interphase) because the decomposition reaction of the SSE is suppressed significantly (Figure 4c).

Interface with Li-Metal Anode
Due to the inherently narrow electrochemical stability window, many sulfide SSEs decompose spontaneously simply by contacting a Li-metal anode [19]. Several efforts have been made to evaluate Li-metal/sulfide-SSE interfaces both theoretically and experimentally [18,50,[56][57][58][59][60][61][62][63][64][65][66][67]. The interfaces formed between Li metal and the SSE can be classified into the following three types [68]. Type I interfaces are stable and ideal interfaces formed when the SSE does not react with Li metal electrochemically (Figure 4a). Most binary compounds, including Li 2 S and Li 3 P, form this 'type I' interface with Li metal. However, if the SSE is not electrochemically stable with Li metal, as most of the sulfide compounds are, the interphase will form at the interface as products of the SSE decompose. If the interphase material is highly conductive for both ionic and electronic transport, non-passivating interphases will form, and the reduction of SSE continuously occurs (Type II). Thus, the reaction front rapidly propagates into the SSE, leading to a short circuit and failure of the SSB (Figure 4b). When the interphases exhibit ionic conductivity but not electrical conductivity (Type III), the interface becomes relatively stable (passivating interphase) because the decomposition reaction of the SSE is suppressed significantly (Figure 4c). In situ X-ray photoelectron spectroscopy (XPS) measurements during Li vapor deposition of various sulfide SSEs by Wenzel et al. provide experimental evidence of the formation of passivating interphases [69][70][71][72]. For the crystalline Li7P3S11 and Li6PS5X (X = Cl, Br, or I) argyrodites, Li2S, Li3P, and LiX (X = Cl, Br, or I) were formed at the interphase (Figure 5a), consistent with the theoretical predictions [73]. Despite the possible existence of the various reduced P phases in P 2p signals, the authors concluded that this reduced P phase likely originated from the reactions of Li3P with residual oxygen and water in the UHV chamber. Some studies have pointed out that the oxygen (O) contamination in sulfide SSEs can lead to phase segregation in the lithium phosphate (Li3PO4) phase at the initial stage of the interphase formation reaction [74]. It remains debatable whether the oxygen-containing interphase formation should be attributed to the pre-existing oxygen in the sulfide SSEs or contamination from air exposure during the sample transfer. Correlative analysis from time-resolved electrochemical impedance spectroscopy (EIS) and XPS confirms that an increase in the thickness and resistivity of the interphase is gradually slowed down as more Li metal is deposited (Figure 5b) in the Li7P3S11 system, suggesting that their interphase stabilizes the interface themselves (the formation of a passivating interphase). Similar passivating interphase formation has been reported in Li6PS5Cl (LPSCl) and Li6PS5Br [71].
In contrast to the passivating-layer formation of Li7P3S11 and Li-argyrodite (Li6PS5Cl and Li6PS5Br)-related interphases, Li thiophosphates such as LGPS and LSPS (Li10SnP2S12) are known to form a non-passivating interphase upon contact with Li metal due to the presence of components (i.e., Ge 4+ and Sn 4+ ) that provide electron-conducting pathways when reduced [71,75]. Wenzel et al. conducted in situ XPS experiments on LGPS material and reported its interfacial instability with Li metal [72]. Figure 5c shows the XPS spectra changes of LGPS before and after deposition of 31-nm Li metal. In addition to the formation of Li3P and Li2S, the reduction of Ge 4+ to metallic Ge species is clearly detected and denoted as "reduced Ge" and "Ge metal or Li15Ge4" in Figure 5c. The fraction of Li2S continues to increase even after 2 h (Figure 5d), indicating that the interfacial reaction is not yet finished at the end of the last deposition step and that the SEI is still growing ( Figure  5e). Compared with that in germanium-free Li7P3S11, the SEI formed in Li10GeP2S12 grows much faster. It is interesting that the decomposition of LGPS continues given that the amount of Ge reduction measured in this study is much less than the amount of formed Li3P and Li2S. The authors attributed this to the formation of electronically conducting Ge phases that also have high Li diffusivity. Such non-passivating interphase formation leads In situ X-ray photoelectron spectroscopy (XPS) measurements during Li vapor deposition of various sulfide SSEs by Wenzel et al. provide experimental evidence of the formation of passivating interphases [69][70][71][72]. For the crystalline Li 7 P 3 S 11 and Li 6 PS 5 X (X = Cl, Br, or I) argyrodites, Li 2 S, Li 3 P, and LiX (X = Cl, Br, or I) were formed at the interphase (Figure 5a), consistent with the theoretical predictions [73]. Despite the possible existence of the various reduced P phases in P 2p signals, the authors concluded that this reduced P phase likely originated from the reactions of Li 3 P with residual oxygen and water in the UHV chamber. Some studies have pointed out that the oxygen (O) contamination in sulfide SSEs can lead to phase segregation in the lithium phosphate (Li 3 PO 4 ) phase at the initial stage of the interphase formation reaction [74]. It remains debatable whether the oxygen-containing interphase formation should be attributed to the pre-existing oxygen in the sulfide SSEs or contamination from air exposure during the sample transfer. Correlative analysis from time-resolved electrochemical impedance spectroscopy (EIS) and XPS confirms that an increase in the thickness and resistivity of the interphase is gradually slowed down as more Li metal is deposited (Figure 5b) in the Li 7 P 3 S 11 system, suggesting that their interphase stabilizes the interface themselves (the formation of a passivating interphase). Similar passivating interphase formation has been reported in Li 6 PS 5 Cl (LPSCl) and Li 6 PS 5 Br [71].
In contrast to the passivating-layer formation of Li 7 P 3 S 11 and Li-argyrodite (Li 6 PS 5 Cl and Li 6 PS 5 Br)-related interphases, Li thiophosphates such as LGPS and LSPS (Li 10 SnP 2 S 12 ) are known to form a non-passivating interphase upon contact with Li metal due to the presence of components (i.e., Ge 4+ and Sn 4+ ) that provide electron-conducting pathways when reduced [71,75]. Wenzel et al. conducted in situ XPS experiments on LGPS material and reported its interfacial instability with Li metal [72]. Figure 5c shows the XPS spectra changes of LGPS before and after deposition of 31-nm Li metal. In addition to the formation of Li 3 P and Li 2 S, the reduction of Ge 4+ to metallic Ge species is clearly detected and denoted as "reduced Ge" and "Ge metal or Li 15 Ge 4 " in Figure 5c. The fraction of Li 2 S continues to increase even after 2 h (Figure 5d), indicating that the interfacial reaction is not yet finished at the end of the last deposition step and that the SEI is still growing (Figure 5e). Compared with that in germanium-free Li 7 P 3 S 11 , the SEI formed in Li 10 GeP 2 S 12 grows much faster. It is interesting that the decomposition of LGPS continues given that the amount of Ge reduction measured in this study is much less than the amount of formed Li 3 P and Li 2 S. The authors attributed this to the formation of electronically conducting Ge phases that also have high Li diffusivity. Such non-passivating interphase formation leads to the high coulombic loss of LGPS, compared to LPS (specifically, Li 9.6 P 3 S 12 ), as shown in Figure 5f [24].
Electrochem 2021, 2, FOR PEER REVIEW 6 to the high coulombic loss of LGPS, compared to LPS (specifically, Li9.6P3S12), as shown in Figure 5f [24]. Various solutions have been reported to engineer the compatibility between SSEs and Li metals. For example, Yang et al. suggested antiperovskite lithium oxychloride, Li3OCl, as a potential artificial passivating compound to stabilize the interface between SSEs and Li metal [77]. As also well predicted in Figure 2 (located at the top of the oxides) [19], antiperovskites are theoretically stable at 0 V, owing to the absence of reducible non-Li cations in the composition [10]. In their experiments, Li-symmetric cell shows a stable voltage polarization profile >100 h at 0.035 mA cm −2 . However, their approach did not Various solutions have been reported to engineer the compatibility between SSEs and Li metals. For example, Yang et al. suggested antiperovskite lithium oxychloride, Li 3 OCl, as a potential artificial passivating compound to stabilize the interface between SSEs and Li metal [77]. As also well predicted in Figure 2 (located at the top of the oxides) [19], antiperovskites are theoretically stable at 0 V, owing to the absence of reducible non-Li cations in the composition [10]. In their experiments, Li-symmetric cell shows a stable voltage polarization profile >100 h at 0.035 mA cm −2 . However, their approach did not expand on how to incorporate their artificial passivating compound into the full cell systems. Recently, Ye et al. proposed an interesting solution for the poor interfacial stability of sulfide SSEs with Li metal [76]. For example, the Li-LGPS-Li symmetric cell exhibited very poor cyclability at 0.25 mA cm −2 (corresponding to 0.5C), with a voltage spark after 3 cycles. Even for LPSCl, which is well known for being more electrochemically stable, a short-circuit occurred after over 150 h of operation under the same test condition. However, by employing combined strategies of multilayer stacking of the electrolytes, a less-stable electrolyte sandwiched between more-stable SSEs, and covering of the Li-metal anode with a graphite film, the authors demonstrate superior cycling performance for 1800 h under the same condition (0.25 mA −2 ) without any observable voltage drops for the Li/G-LPSCl-LGPS-LPSCl-G/Li symmetric cell. Further, this cell is also stable for up to 15 h of operation at 20 mA cm −2 (corresponding to 40C, Figure 5g). The authors suggested that the multilayer design can prevent lithium dendrite penetration because the decomposition products of SSEs at the interface between LGPS and LPSCl fill and heal cracks. However, this mechanism has not yet been fully proven.
Recently, operando XPS (opXPS) measurements were developed and applied to investigate a Li/Li 2 S-P 2 S 5 interface by Wood et al. This technique mimics the chargingdischarging process of a battery system using an electron gun (for lithium plating, charging) and ultraviolet-light source (for lithium stripping, discharging) [74]. Figure 6a presents schematics of the operando XPS technique employed in this study. By following the general trends in the phase-composition changes during charge/discharge half-cycles (Figure 6b), the opXPS profiles having depth information of the interphase reveal that the interphase has a layered configuration composed of Li 0 /Li 2 O/Li 3 P/Li 2 S/LPS (Figure 5c). The authors claim that Li 2 S and Li 3 P phases form first on the LPS, followed by Li 2 O phase formation, and finally Li 0 deposition. The authors argued that Li 2 O converts into Li 2 O 2 and that Li 3 P transforms into the Li-deficient phase Li 3−x P via delithiation during discharging (lithium-stripping process), as shown in Figure 6d, since Li 2 O and Li 3 P are the redox-active species.  [76]. For example, the Li-LGPS-Li symmetric cell exhibited very poor cyclability at 0.25 mA cm −2 (corresponding to 0.5C), with a voltage spark after 3 cycles. Even for LPSCl, which is well known for being more electrochemically stable, a short-circuit occurred after over 150 h of operation under the same test condition. However, by employing combined strategies of multilayer stacking of the electrolytes, a lessstable electrolyte sandwiched between more-stable SSEs, and covering of the Li-metal anode with a graphite film, the authors demonstrate superior cycling performance for 1800 h under the same condition (0.25 mA −2 ) without any observable voltage drops for the Li/G-LPSCl-LGPS-LPSCl-G/Li symmetric cell. Further, this cell is also stable for up to 15 h of operation at 20 mA cm −2 (corresponding to 40C, Figure 5g). The authors suggested that the multilayer design can prevent lithium dendrite penetration because the decomposition products of SSEs at the interface between LGPS and LPSCl fill and heal cracks. However, this mechanism has not yet been fully proven.
Recently, operando XPS (opXPS) measurements were developed and applied to investigate a Li/Li2S-P2S5 interface by Wood et al. This technique mimics the charging-discharging process of a battery system using an electron gun (for lithium plating, charging) and ultraviolet-light source (for lithium stripping, discharging) [74]. Figure 6a presents schematics of the operando XPS technique employed in this study. By following the general trends in the phase-composition changes during charge/discharge half-cycles (Figure 6b), the opXPS profiles having depth information of the interphase reveal that the interphase has a layered configuration composed of Li 0 /Li2O/Li3P/Li2S/LPS (Figure 5c). The authors claim that Li2S and Li3P phases form first on the LPS, followed by Li2O phase formation, and finally Li 0 deposition. The authors argued that Li2O converts into Li2O2 and that Li3P transforms into the Li-deficient phase Li3−xP via delithiation during discharging (lithium-stripping process), as shown in Figure 6d, since Li2O and Li3P are the redox-active species.

Li-Dendrite Penetration
Dendritic Li growth and its penetration through SSEs is a major issue that must be addressed for the commercialization of ASSBs. In general, Li dendrites grow at relatively

Li-Dendrite Penetration
Dendritic Li growth and its penetration through SSEs is a major issue that must be addressed for the commercialization of ASSBs. In general, Li dendrites grow at relatively high current rates even in SSE systems, and the dendrites lead to a short circuit of the battery cell. In addition, the Li dendrite will form a new interface once it penetrates a SSE, and then, it will decompose the SSE and form a SEI layer, which will lead to deterioration of the battery performance. For example, 0.75Li 2 S-0.25P 2 S 5 glass electrolyte shows a voltage drop in a Li-symmetric cell when the current is larger than 0.4 mA cm −2 ; this voltage drop indicates the short circuit of the cell [78]. Similarly, other sulfide SSEs have relatively low critical currents (<1.5 mA cm −2 ) that result in short circuits [79][80][81], which are much lower critical currents than those for liquid electrolytes (>10 mA cm −2 ) [82] A very recent work by Otoyama et al. confirmed that Li metal can penetrate through the Li 3 PS 4 SSE using in situ X-ray computed tomography (X-ray CT) [83]. Figure 7a presents the galvanostatic cycling results of the Li/Li 3 PS 4 /Li symmetric cell under three different current rates of 0.24, 0.40, and 0.80 mA cm −2 . Tomograms in the XY, YZ, and ZX planes were obtained from the X-ray CT measurements (Figure 7b). The authors claimed that a crack is formed when Li 3 PS 4 is decomposed in contact with Li metal since the volume expanded. Then, Li metal could easily penetrate the cracks and form new interfaces of Li/Li 3 PS 4 . Such a reduction-expansion-cracking reaction is repeated and thus makes large cracks and form Li clusters inside the Li 3 PS 4 electrolyte, finally causing a short circuit (Figure 7c,d). A very recent study by Ning et al. using in situ X-ray CT in Li/Li 6 PS 5 Cl/Li system also demonstrated that the cracking is initiated near the surface with the plated electrode, and that the cracks propagate across the electrolyte from the plated to the stripped electrode [84], which agrees well with previous observations [83]. Similarly, Kazyak et al. observed Li dendrite penetration through Li 3 PS 4 electrolyte using operando optical microscopy [54].
Electrochem 2021, 2, FOR PEER REVIEW 8 high current rates even in SSE systems, and the dendrites lead to a short circuit of the battery cell. In addition, the Li dendrite will form a new interface once it penetrates a SSE, and then, it will decompose the SSE and form a SEI layer, which will lead to deterioration of the battery performance. For example, 0.75Li2S-0.25P2S5 glass electrolyte shows a voltage drop in a Li-symmetric cell when the current is larger than 0.4 mA cm −2 ; this voltage drop indicates the short circuit of the cell [78]. Similarly, other sulfide SSEs have relatively low critical currents (<1.5 mA cm −2 ) that result in short circuits [79][80][81], which are much lower critical currents than those for liquid electrolytes (>10 mA cm −2 ) [82] A very recent work by Otoyama et al. confirmed that Li metal can penetrate through the Li3PS4 SSE using in situ X-ray computed tomography (X-ray CT) [83]. Figure 7a presents the galvanostatic cycling results of the Li/Li3PS4/Li symmetric cell under three different current rates of 0.24, 0.40, and 0.80 mA cm −2 . Tomograms in the XY, YZ, and ZX planes were obtained from the X-ray CT measurements (Figure 7b). The authors claimed that a crack is formed when Li3PS4 is decomposed in contact with Li metal since the volume expanded. Then, Li metal could easily penetrate the cracks and form new interfaces of Li/Li3PS4. Such a reduction-expansion-cracking reaction is repeated and thus makes large cracks and form Li clusters inside the Li3PS4 electrolyte, finally causing a short circuit (Figure 7c,d). A very recent study by Ning et al. using in situ X-ray CT in Li/Li6PS5Cl/Li system also demonstrated that the cracking is initiated near the surface with the plated electrode, and that the cracks propagate across the electrolyte from the plated to the stripped electrode [84], which agrees well with previous observations [83]. Similarly, Kazyak et al. observed Li dendrite penetration through Li3PS4 electrolyte using operando optical microscopy [54]. It is widely known that Li dendrite grows even in SSEs. However, it remains unclear how Li dendrite can penetrate through inorganic SSEs. Some plausible scenarios have been suggested (Figure 8a). First, Han et al. proposed that the relatively high electronic It is widely known that Li dendrite grows even in SSEs. However, it remains unclear how Li dendrite can penetrate through inorganic SSEs. Some plausible scenarios have been suggested (Figure 8a). First, Han et al. proposed that the relatively high electronic conductivity of SSEs is the origin of Li-dendrite formation within the electrolyte after monitoring the dynamic evolution of Li concentration profiles in SSEs during Li-metal plating using time-resolved operando neutron depth profiling (Figure 8b-d) [85]. They quantified the net amount of Li transported and compared it with the cumulative electric charge (Figure 8c,d). In the LiPON system, they did not observe a noticeable difference between the total amount of Li transported and the cumulative charge (Figure 8c). In contrast, a smaller amount of accumulated Li is observed than the cumulative charge in the Li 3 PS 4 system (Figure 8d). Thus, they suggested that the Li dendrites are formed in the deep and undetectable region of the SSEs. Similar behavior is also observed in the LLZO electrolyte system. Given that the electronic conductivities of LLZO (10 −9 S cm −1 ) and Li 3 PS 4 (10 −8 S cm −1 ) are much higher than that of LiPON (10 −15 -10 −12 S cm −1 ) [86][87][88], they concluded that the relatively high electronic conductivity results in the Li dendrite formation. Second, in a recent work by Bruce's group, it was proposed that voids formed during stripping increase the 'effective' current significantly because of the poor contact between Li metal and SSEs, thereby resulting in the dendritic formation upon plating in the Li 6 PS 5 Cl electrolyte system [89]. Figure 8e summarizes the proposed mechanism of Li-dendrite formation. The authors claimed that the high stacking pressure of the ASSBs can overcome the Li-dendrite growth issue: the Li-symmetric cell with 7 MPa did not show a noticeable polarization increase during 100 cycles at 1.0 mA cm −2 , whereas the cell with 3 MPa showed a substantial increase of polarization during stripping and short circuit after 8 cycles at the same current density. A high stacking pressure forms more uniform interface between a SSE and Li metal, thereby increases contact areas. Thus, higher current density is needed to make Li dendrite penetration through the SSE at high stacking pressures. In a very recent computational work, void formation in Li metal during stripping was shown to depend on the lithiophilicity of the interface [90]. For example, Li 2 O and LiPON have lithiophilic interfaces with Li metal and do not show accumulated voids at the interface during the striping process in their density functional theory (DFT) and kinetic Monte Carlo (KMC) computations. In contrast, LiF with lithiophobicity shows accumulated void formation at the LiF/Li interface. This finding suggests that interface and coating design play important roles in suppressing Li dendrite growth. Third, it was proposed that the relatively high stacking pressure in ASSBs accelerates Li penetration through the SSEs [91][92][93]. Although some degree of stack pressure is required to ensure contact between an SSE and Li-metal anode [94], too high of a stack pressure results in Li dendrite penetration through the SSE and thus a short circuit. Recently, Meng's group investigated the effect of stack pressure of ASSBs on the short-circuit behavior using the pressuremonitoring system shown in Figure 8f [91]. They observed the charge/discharge profiles of Li/Li 6 PS 5 Cl/Li symmetric cells at 75 µA cm −2 with different stack pressures from 5 to 75 MPa (Figure 8g). Interestingly, the cell with a very high stack pressure of 75 MPa showed a short circuit even before the Li plating and stripping test was conducted. The authors claimed that the short circuit at such high stack pressure is due to the creeping of Li metal to flow within the pores of Li 6 PS 5 Cl. When the stack pressure was lowered, the cell ran longer before the short circuit occurs. For example, for the cell with a stack pressure of 5 MPa, a short circuit was not observed within 1000 h of plating and stripping. In a full-cell configuration, the cell with a stack pressure of 5 MPa showed no short-circuit behavior and a typical charge-discharge profile, delivering a specific capacity of 150 mAh g −1 (Figure 8h). In contrast, the cell was short-circuited when the pressure was increased to 25 MPa. Large low-density structures were also observed in the SSE using X-ray tomography after the electrochemical Li plating and stripping process at 20 MPa. It was speculated that these low-density structures are Li dendrites and SEI formed between Li metal and Li 6 PS 5 Cl. Based on their observations, it is clear that the stack pressure is an important parameter to control Li plating and stripping behavior in ASSB systems; this idea is also supported by a recent work by Hänsel et al. [93]. In their work, they investigated the effect of the thickness of Li 6 PS 5 Cl at no current condition and observed that the time required for short circuit monotonically increases with the thickness of the Li 6 PS 5 Cl SSE. This finding demonstrates that the short circuit at high stack pressure occurs via Li extrusion through the micropores of SSEs. The same group also discovered that the characteristics of SEI affect Li and Na exclusion, and thus short-circuit behaviors, considerably [92]. In their comparison study, they showed very sharp contrast between Li 3 PS 4 /Li and Na 3 PS 4 /Na systems: Li 3 PS 4 showed short circuit at 45 MPa after 55 h, whereas Na 3 PS 4 did not show such behavior at the same stack pressure even after 250 h. The authors claimed that the root cause of these contrasting observations is the different characteristics of the SEI formed between Li 3 PS 4 and Li metal vs. Na 3 PS 4 and Na metal. Whereas the Li 3 PS 4 /Li system forms a very thin passivating layer, the SEI continues to grow at the Na 3 PS 4 /Na interface because of the relatively high electronic conductivity of Na 3 P and Na 2 S compared to that of their Li counterparts, confirmed by the density of states (DOS) computations. These results indicate that the thickness and mechanical properties of the SEI formed between an SSE and Li-metal anode are crucial parameters to enable the realization of Li-metal ASSBs.
Electrochem 2021, 2, FOR PEER REVIEW 10 for short circuit monotonically increases with the thickness of the Li6PS5Cl SSE. This finding demonstrates that the short circuit at high stack pressure occurs via Li extrusion through the micropores of SSEs. The same group also discovered that the characteristics of SEI affect Li and Na exclusion, and thus short-circuit behaviors, considerably [92]. In their comparison study, they showed very sharp contrast between Li3PS4/Li and Na3PS4/Na systems: Li3PS4 showed short circuit at 45 MPa after 55 h, whereas Na3PS4 did not show such behavior at the same stack pressure even after 250 h. The authors claimed that the root cause of these contrasting observations is the different characteristics of the SEI formed between Li3PS4 and Li metal vs. Na3PS4 and Na metal. Whereas the Li3PS4/Li system forms a very thin passivating layer, the SEI continues to grow at the Na3PS4/Na interface because of the relatively high electronic conductivity of Na3P and Na2S compared to that of their Li counterparts, confirmed by the density of states (DOS) computations. These results indicate that the thickness and mechanical properties of the SEI formed between an SSE and Li-metal anode are crucial parameters to enable the realization of Li-metal ASSBs.

Interface with Cathodes
The cathode side has complicated interfaces since it contains various components to provide both electron and ion-conduction pathways. Typically, the cathode composite consists of a cathode active material, electron-conducting carbon, and an ion-conducting SSE. Therefore, it forms three distinct interfaces, cathode/carbon, cathode/SSE, and carbon/SSE (as shown in Figure 3). At the early stage of sulfide SSE development, sulfide SSEs were thought to be stable upon electrochemical oxidation [17,28,47]. However, recent works have shown that most of the sulfide SSEs are prone to instability during oxidation when electron pathways are provided [48,49,[95][96][97][98]. Specifically, Zhang et al. investigated the interfacial properties of a carbon-containing cathode composite with a LGPS SSE [96]. Figure 9a presents a schematic illustration showing how the electron-conducting carbon additives affect the electrochemical reaction processes at the LiCoO 2 /LGPS interface. They found that the carbon additive contributes to an additional slope during the charging process and degrades the LGPS electrolyte, which increases the overpotential and thus decreases the specific capacity. In their ex situ XPS characterizations, it was confirmed that the LiCoO 2 /LGPS cathode composite with carbon additive shows much more pronounced decomposition of LGPS and produces more insulating sulfide products than the composite without carbon after 100 cycles. In addition, as shown in Figure 9b, the cathode composite without carbon additive exhibits better cycling stability after 100 cycles at 115 µA cm −2 (0.1C). Yoon et al. also demonstrated similar degradation mechanisms at the interface between LGPS and carbon [95]. Both groups observed that the degradation of the LGPS/carbon interface does not depend on the structure and morphology of the carbon additives. It is a great challenge because the sulfide SSE will decompose once electron pathways are provided, which is essential to enable ASSBs. However, we expect that it is possible to slow down such degradations by using a low carbon content or low surface area of carbon.
Chemical and electrochemical degradation of sulfide SSEs at the interface with the cathode material is another important hurdle to overcome. Sakuda et al. observed the interfacial layer between a LiCoO 2 cathode and Li 2 S-P 2 S 5 sulfide-SSE after the first charging using TEM and EDS, as shown in Figure 9c [99]. The authors reported the mutual diffusions of Co, P, and S at the interface between the LiCoO 2 cathode and sulfide SSE. The formation of the interfacial layer increases the resistance of the cell significantly. In contrast, the authors were able to suppress the mutual diffusion and the formation of the interfacial layer by coating a Li 2 SiO 3 thin film on the LiCoO 2 cathode particle, thereby improving the specific capacity. Similarly, the diffusion of S, P, and Cl into LiNi 0.8 Co 0.1 Mn 0.1 O 2 cathode particle was confirmed by the SEM/EDS technique [100]. Later, Walther et al. further studied the degradation mechanisms of the cathode composite of Li 6 PS 5 Cl/LiNi 0.6 Co 0.2 Mn 0.2 O 2 [101]. They characterized the interphase formation using XPS combined with time-of-flight secondaryion mass spectrometry (ToF-SIMS) to determine the composition and microstructure of interphases. Phosphate and sulfate compounds were detected by their ex situ XPS analysis after 100 cycles, which agrees with a report by Visbal et al. [102,103]. ToF-SIMS obtains compositional and morphological information on the interphases beyond the capability of XPS. To investigate the spatial fragment distribution, the depth profiles were reconstructed, as shown in Figure 9d. The results show that the intensities of the phosphates and sulfates are dominated by the interface regions, demonstrating the formation of the SEI with PO x and SO x enriched. In the investigation of the interphase formation between the Li 6 PS 5 Cl SSE and LiMn 2 O 4 cathode by Auvergniot et al., it was observed that the Li 6 PS 5 Cl electrolyte decomposes upon electrochemical cycling by scanning Auger microscopy (SAM) analysis. Figure 9e and f present the SAM mapping results before and after cycling, respectively. Before the electrochemical cycling, the S and Cl signals overlapped. However, the S-rich and Cl-rich regions are clearly separated after 22 cycles. The authors also observed such SSE decomposition and SEI formation in other oxide cathode systems, including LiCoO 2 and LiNi 1/3 Co 1/3 Mn 1/3 O 2 [103].
LiCoO2/LGPS/Li4Ti5O12 and 82% retention after 100 cycles in LiNi0.5Mn1.5O4/LGPS/Li4T full-cell configurations. In ex situ XPS characterizations, no significant decomposition tions of LGPS were observed. The effects of mechanical constriction on the electrochem stability window of sulfide SSEs are well summarized in a recent review article [105]. H ever, it is questionable if such a high stack pressure (>250 MPa) of the cell is practically sible and how much the cost of manufacturing ASSBs would increase.  A recent study by Li's group presented interesting experimental results that the oxidation stability of a sulfide SSE, LGPS, can be improved significantly and that the oxidation stability window can be expanded up to 10 V vs. Li metal by high mechanical stack pressure even if conductive carbon (10 wt%) is added [104]. The authors claimed that the high mechanical constriction can kinetically stabilize sulfide SSEs based on their DFT computations and experimental observations. Note that the authors used graphite-film-covered Li metal to avoid Li metal penetration through the LGPS electrolyte at high stack pressure. They presented good cycling stability: 91% retention after 360 cycles in LiCoO 2 /LGPS/Li 4 Ti 5 O 12 and 82% retention after 100 cycles in LiNi 0.5 Mn 1.5 O 4 /LGPS/Li 4 Ti 5 O 12 full-cell configurations. In ex situ XPS characterizations, no significant decomposition reactions of LGPS were observed. The effects of mechanical constriction on the electrochemical stability window of sulfide SSEs are well summarized in a recent review article [105]. However, it is questionable if such a high stack pressure (>250 MPa) of the cell is practically feasible and how much the cost of manufacturing ASSBs would increase.
A representative strategy to avoid (electro)chemical degradation of the interface between a cathode and a sulfide SSE is the 'cathode coating' approach. Many ternary-metal oxide compounds have been studied for use as cathode coatings, including LiNbO 3 , LiTaO 3 , Li 4 [24,56,99,[106][107][108][109][110][111][112][113][114]. In theory, cathode coatings can separate an SSE from the low Li chemical potentials of the cathode in its highly charged state and prevent electrochemical self-decomposition of the SSE. However, such oxide coating compounds could not solve the interfacial issue between the cathode and sulfide SSE completely. In many cases, S/O exchange between coating materials and sulfide SSEs occurs, driven by the strong binding energy of PO 4 groups, as confirmed by computational studies [51,115]. Recently, Zhang et al. observed a sharp contrast between two coating materials, Li 2 ZrO 3 and Li 3 B 11 O 18 [116]. They show that Li extraction occurs at Li 2 ZrO 3 during the first charge process and that it loses its protection capability for the cathode. It was also observed that metal sulfides form within the Li 2 ZrO 3 coating after 50 cycles. Therefore, it presents~60% capacity retention after 100 cycles (Figure 10a). In contrast, Li 3 B 11 O 18 coating material shows much improved cycling stability,~80% retention after 100 cycles. The Li 3 B 11 O 18 coating is also confirmed to be well maintained even after 10 cycles using STEM, as shown in Figure 10b,c.

Perspectives and Outlooks
The promise of ASSBs arises from the high safety and use of energy-dense Li-metal anodes. At a very early stage, it was expected that 'hard' and 'non-flammable' SSEs could prevent the dendritic growth of 'soft' Li metals. However, many studies have since demonstrated that Li metals can still penetrate SSEs at relatively high current rates [52,55,118,119]. One simple way to suppress Li dendrite growth is to ensure good contact between the SSE and electrode in order to increase the effective contact area, thereby reducing the effective current. Relatively high stacking pressure can ensure contact between the Li metal and SSE and suppress Li dendrite growth, as demonstrated by Bruce's group [89]. However, too high of a stacking pressure can lead to Li extrusion and thereby Li dendrite penetration through the micropores of the SSE [91,93]. Another strategy is to form a lithi-  Figure (a-c) are adopted from reference [116] with permission. (d) Electrochemical stability windows of common Li-ion conductors. (e) Reaction energies ∆E rxt at various interfaces of fully lithiated cathode/SSE, fully lithiated cathode/coating, and coating/SSE in meV/atom. Figure (d,e) are reprinted from reference [115]. Copyright 2019, with permission from Elsevier.
Yihan et al. conducted a computational investigation to identify Li-containing cathode coating materials [115]. In their work, the phase stability, electrochemical and chemical stability against both cathodes and SSEs, and ionic conductivity were considered. The computational screening identifies polyanionic oxide compounds as promising cathode coating materials, including phosphates and borates. Figure 10d shows the electrochemical stability windows of six polyanionic oxides selected by Yihan et al. For better comparison, they plotted the electrochemical stability windows of common sulfide SSEs: Li 6 PS 5 Cl (LPSCl), LGPS, the glass-ceramic LPS, garnet-type LLZO, and ternary non-polyanionic coating materials such as Li 2 ZrO 3 , LiNbO 3 , and LiTaO 3 . Polyanionic oxides have much higher oxidation stability (>4.5 V) than non-polyanionic oxides (3-4 V). They explained that the higher oxidation limit of polyanionic oxides is attributable to the highly covalent bonds between M (here, P and B) and oxygen in MO x polyhedral units, which lowers the energy of the oxygen orbitals. This effect decreases the chemical reactivity of the polyanionic compounds with sulfide SSEs and increases the oxidation stability limit significantly. They also computed the reaction energy between each component, including the sulfide SSEs, oxide cathodes, and coating materials, as presented in Figure 10e. As expected, sulfide SSEs show very high reactivity with most of the oxide cathode materials. Interestingly, the reactivity between an oxide coating and a sulfide SSE is similar with that between an oxide cathode and a sulfide SSE (>100 meV/atom). In contrast, the polyanionic coating materials show less reactivity against sulfide SSEs. Similarly, Liu et al. explored the fluorides for cathode coatings using high-throughput computational screening [117]. By screening the phase stability, electrochemical stability, chemical stability, and Li-ion conductivity, they identified Li 2 MF 6 (M = Si, Ge, Zr, Ti) compounds as good coating materials. However, their work did not aim for ASSB applications; thus, they did not include the chemical reactivity between SSEs and coating materials, which merit investigation.

Perspectives and Outlooks
The promise of ASSBs arises from the high safety and use of energy-dense Li-metal anodes. At a very early stage, it was expected that 'hard' and 'non-flammable' SSEs could prevent the dendritic growth of 'soft' Li metals. However, many studies have since demonstrated that Li metals can still penetrate SSEs at relatively high current rates [52,55,118,119]. One simple way to suppress Li dendrite growth is to ensure good contact between the SSE and electrode in order to increase the effective contact area, thereby reducing the effective current. Relatively high stacking pressure can ensure contact between the Li metal and SSE and suppress Li dendrite growth, as demonstrated by Bruce's group [89]. However, too high of a stacking pressure can lead to Li extrusion and thereby Li dendrite penetration through the micropores of the SSE [91,93]. Another strategy is to form a lithiophilic interphase that ensures contact between the Li metal and SSE [90]. This strategy has been widely studied with some successes shown not only in sulfide but also in oxide SSEs [120][121][122][123]. However, it is vital to better understand Li-dendrite growth mechanisms. Although there are some plausible explanations, as discussed in Section 3.2, the mechanism is not yet fully understood. We do not expect that a single theory can explain everything but suspect that instead the origin of Li-dendrite formation could depend on the properties of SSEs.
There have been significant efforts to understand SEI formation on both cathodes and anodes since the chemical evolution of SEI plays a critical role in determining the electrochemical properties of ASSBs [71,72,74,99,101,103]. However, our current understanding is still limited, including with respect to the thickness of the interphases and the nature of their microstructures. This lack of knowledge can be mainly attributed to the buried nature inside the solid-state bulk cell, which is thus not readily accessible using most conventional analysis techniques. Further, the extreme reactivity of the Li metal, sulfide SSE, and interphases to moisture, oxygen, and even the electron beam are other important challenges for their characterization. Such high reactivity requires air-tight sample preparation and transfer capabilities for studying buried interfaces since typical sample handling can damage these highly reactive materials. In addition, due to vulnerability to the e-beam, the sample can be damaged during the electron microscopy analyses. Thus, most previous works have been limited to low magnification to minimize the electron-beam dose rate to the sample. However, owing to recent developments of electron microscopy technology, such as a multi-capable TEM sample holders (e.g., offering both cryo and air-tight transfer capabilities in a single holder), high-sensitive electron detectors, etc., it is expected to play a major role soon in investigations of the morphological/structural/chemical properties of the interphase.
Although several works have proposed coating materials that can stabilize the interface between the SSE and cathode active materials [115][116][117], another critical issue remains, namely the oxidation of the SSE at the interfaces with the conductive additive (e.g., carbon) and current collector (since an SSE is subject to the full cathode potential at such interfaces). Using a low carbon additive content would reduce such oxidation of the SSE. However, a better strategy is required. One alternative could be a dual coating method that uses carbon nanowire or nano-mesh with uniform cathode active material coating and then SSE coating. This structure will be able to separate electron and ion pathways. Another strategy is to form a passivating interface between a sulfide SSE and carbon.

Summary
Sulfide SSEs have attracted significant interest as a key component to realize ASSBs due to their high ionic conductivity and deformability. However, there are remaining challenges. Herein, we provided an overview of the major challenges in interfaces and interphases formed in sulfide SSE systems and the recent progress that has been made in understanding such interface and interphase formation. Then, efforts to address interface and interphase issues in sulfide SSEs were introduced, including several strategies to stabilize the interface or form passivating interphase. Finally, we provided our perspectives on future research directions, with the hope that this review will help guide potential efforts for developing long-lasting and fast-charging solid-state batteries.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.