Interfacial Enhancement of an Iodide–Oxide–Sulfide Solid Electrolyte for Long-Life All-Solid-State Lithium Metal Batteries

Abstract

Interfacial instability against lithium metal remains a major challenge for sulfide-based solid electrolytes in all-solid-state lithium metal batteries (ASSLMBs). In this study, the previously reported iodide–oxide–sulfide solid electrolyte 10LPSOI was prepared by mechanical ball milling (BM) and a modified liquid-phase shaking (LS) method to investigate the influence of the synthesis route on structural characteristics, particle morphology, ionic transport, and lithium–metal interfacial behavior. Structural and spectroscopic analyses revealed differences in phase characteristics and local phosphorus environments between 10LPSOI-BM and 10LPSOI-LS, while SEM analysis showed that the LS method produced substantially finer and more homogeneous particles. Although 10LPSOI-LS exhibited a lower room-temperature ionic conductivity of 0.7 mS cm^–1 than 10LPSOI-BM at 3.4 mS cm^–1, it achieved a higher critical current density of 0.925 mA cm^–2 compared with 0.30 mA cm^–2 for 10LPSOI-BM. The Li | 10LPSOI-LS | Li symmetric cell maintained relatively stable polarization for about 3092 h at 0.30 mA cm^–2 before a sustained decrease in overpotential was observed. In addition, 10LPSOI-LS showed improved cycling stability in NMC₁₁₁-based cells compared with 10LPSOI-BM under the tested conditions. These results indicate that the modified LS route improves lithium–metal interfacial performance despite the lower ionic conductivity. This improved behavior is correlated with synthesis-dependent differences in particle morphology, phase characteristics, local structure, and surface chemistry. The modified liquid-phase process may also offer potential advantages for scalable preparation of sulfide-based solid electrolytes.

Introduction

Global energy demand and continued dependence on fossil fuels have contributed substantially to greenhouse-gas emissions and associated environmental concerns. − The transportation sector remains strongly dependent on petroleum-based fuels and is therefore an important target for decarbonization. The increasing adoption of electric vehicles (EVs) offers a promising pathway to reduce direct fossil-fuel consumption in transportation. , However, broader implementation of EVs requires energy-storage systems with improved safety, energy density, and long-term cycling performance.

Despite their advantages, EVs face several challenges, primarily arising from the Li-ion battery that serves as their energy storage system. There are two main problems associated with commercial Li-ion batteries: first, the use of liquid electrolytes, and second, the heavy graphite anode. The liquid electrolytes used in commercial Li-ion batteries consist of lithium salts dissolved in organic solvents, which are volatile and can ignite or explode. This poses a serious risk to user safety. At the same time, graphite is commonly used as the anode in Li-ion batteries. However, the energy density of graphite is relatively low. , Considering that approximately 20–30% of an EV’s mass comes from the battery module, it becomes one of the heaviest components of the vehicle. Lower anode energy density requires a larger battery mass to store more energy. As EVs become heavier, their driving range decreases and overall energy efficiency is reduced. The first solution to overcome the issues associated with liquid electrolytes is to replace them with solid electrolytes, which are nonflammable and exhibit high energy density and high-temperature resistance. The second solution to address the heavy weight of battery cells is to replace the low energy density of graphite carbon (372 mAh g^–1) with the high energy density of lithium metal (3,860 mAh g^–1), which is approximately ten times higher, thereby enabling a higher overall energy density with reduced weight. , By utilizing solid electrolytes in combination with lithium metal anodes to form all-solid-state lithium metal batteries (ASSLMBs), this system is believed to be a promising future energy-storage solution for EV applications.

Although ASSLMBs exhibit promising advantages, their practical application still faces critical challenges related to interfacial stability between the solid electrolyte and the lithium metal anode. Nonuniform lithium deposition can promote the formation of lithium dendrites, which may penetrate the solid electrolyte and cause internal short circuits or even complete battery failure when the dendrites reach the cathode side. On the other hand, poor interfacial contact and electrochemical or chemical instability between the solid electrolyte and lithium metal, sometimes referred to as the solid electrolyte interphase (SEI), can increase interfacial resistance and degrade performance during repeated cycling. Therefore, it is necessary to develop solid electrolytes that form a stable SEI, which can withstand degradation and effectively suppress lithium dendrite formation. ,−

Solid electrolytes can generally be categorized into three main classes: oxides, chlorides, and sulfides. In terms of chemical stability, oxide-based solid electrolytes have a clear advantage, as they exhibit excellent stability even under ambient atmospheric conditions. Representative oxide-based solid electrolytes include NASICON-type materials such as Li_1+xAl _x Ti_2‑x(PO₄)₃ (LATP) and garnet-type Li₇La₃Zr₂O₁₂ (LLZO), both of which demonstrate relatively high ionic conductivities on the order of 10^–4 to 10^–3 S cm^–1. Although garnet-type solid electrolytes show good compatibility with lithium metal, their intrinsic brittleness often leads to interfacial issues between the electrodes and the solid electrolyte, such as cracking and poor contact area. Furthermore, chloride- or halide-based solid electrolytes exhibit more promising mechanical properties and improved interfacial characteristics, making them more preferable ionic conductors for cathode composites compared to oxide-based solid electrolytes. Li₃MCl₆ (M = Y, In, Zr) exhibits room-temperature ionic conductivity in the range of 10^–4 to 10^–3 S cm^–1 along with superior oxidative stability. In spite of this, chloride-based solid electrolytes are unstable against lithium metal because their SEI is unstable and continues to degrade during the reduction process, leading to the formation of a non–ionically conductive interphase layer. , This makes chloride-based solid electrolytes unsuitable for ASSLMBs. Compared with the other two classes, sulfide-based solid electrolytes are particularly favorable, as they not only exhibit the highest ionic conductivity but also possess advantageous mechanical properties, such as softness, which enable cold pressing to form battery cells. Basically, sulfide-based solid electrolytes are derived from the 3Li₂S–P₂S₅ system, which forms lithium thiophosphate (Li₃PS₄), regarded as the parent compound of sulfide-based solid electrolytes. Li₃PS₄ exhibits an ionic conductivity of 2.0 × 10^–4 S cm^–1 at room temperature when synthesized by mechanical ball milling. Furthermore, Li₃PS₄ was further developed by the incorporation of GeS₂, leading to the formation of Li₁₀GeP₂S₁₂ (LGPS) with a new structural type known as thio-LISICON (sulfide lithium superionic conductor), which exhibits a remarkable ionic conductivity of 1.2 × 10^–2 S cm^–1 at room temperature. Moreover, an LGPS-type composition, Li_9.54Si_1.74P_1.44S_11.7Cl_0.3, demonstrates an even higher ionic conductivity of 2.5 × 10^–2 S cm^–1 at room temperature while simultaneously mitigating the high cost associated with germanium. Despite their high ionic conductivity, solid electrolytes containing Group IV elements are reported to be unstable against lithium metal. As a result, solid electrolytes incorporating Si, Ge, or Sn are not suitable for application in ASSLMBs. Moving forward, another type of lithium thiophosphate family is formed through halide doping to produce argyrodite solid electrolytes. Argyrodite solid electrolytes, such as Li_6‑xPS_5‑xCl_1+x, exhibit superior ionic conductivity of up to 17 mS cm^–1. However, argyrodite solid electrolytes tend to show low stability against lithium metal anodes.

Besides argyrodite, another halide-doped lithium thiophosphate family that shows promising potential consists of glass–ceramic–type materials such as Li₁₀P₃S₁₂I and Li₇P₂S₈I. , These materials are derivatives of the thio-LISICON structure and retain a glass ceramic nature, which is referred to as thio-LISICON II. In comparison, Li₁₀P₃S₁₂I exhibits a higher ionic conductivity of 9.1 × 10^–3 S cm^–1, whereas Li₇P₂S₈I shows an ionic conductivity in the range of 2 × 10^–3 S cm^–1 at room temperature. , Although Li₇P₂S₈I has a lower ionic conductivity than Li₁₀P₃S₁₂I, it demonstrates better compatibility with the lithium metal anode, which may originate from the higher iodine content in Li₇P₂S₈I that contributes to the formation of a more stable SEI. In order to identify an optimal electrolyte for ASSLMB applications, our group previously found that the addition of 10 mol % Li₃PO₄ to the Li₇P₂S₈I solid electrolyte further enhanced its stability against lithium metal compared to pristine Li₇P₂S₈I. As a result, this composition, referred to as 10LPSOI, has been reported as one of the most stable solid electrolytes against lithium metal. The incorporation of a small amount of oxide compound into a sulfide-based solid electrolyte has been shown to form a stable oxysulfide phase, which can inhibit lithium dendrite growth. ,,

In this study, 10LPSOI was newly prepared by mechanical ball milling (BM) and compared with 10LPSOI synthesized by a modified liquid-phase shaking (LS) method to clarify the influence of the synthesis route on its structural, morphological, and electrochemical properties. Although the beneficial effect of Li₃PO₄ addition to Li₇P₂S₈I has been reported previously using liquid-phase synthesis, the influence of different preparation routes on the resulting material characteristics and lithium–metal interfacial behavior has not been systematically examined. Therefore, the present study focuses on a direct comparison between 10LPSOI-BM and 10LPSOI-LS in terms of phase characteristics, local phosphorus environments, particle morphology, ionic conductivity, and electrochemical stability against lithium metal. We hypothesized that synthesis-dependent differences in structure and morphology could significantly influence lithium–metal interfacial behavior, even when the resulting solid electrolytes exhibit different bulk ionic conductivities. To examine these relationships, X-ray diffraction (XRD), Raman spectroscopy, ³¹P magic-angle-spinning nuclear magnetic resonance (³¹P MAS NMR), scanning electron microscopy (SEM), electrochemical impedance spectroscopy (EIS), X-ray photoelectron spectroscopy (XPS), critical current density measurements, long-term lithium symmetric-cell cycling, and battery performance measurements were conducted. In addition, the modified LS method is discussed in terms of its potential scalability, while the primary focus of this study is the relationship between synthesis-dependent material characteristics and electrochemical behavior. More broadly, improving the safety, durability, and manufacturability of solid-state lithium–metal batteries could contribute to the development of safer and more efficient energy-storage technologies for electric transportation and other applications, supporting the broader transition toward lower-carbon energy systems.

Experimental Method

Synthesis

For clarity and ease of differentiation, 10LPSOI synthesized by mechanical ball milling is hereafter referred to as 10LPSOI-BM, while 10LPSOI prepared by the liquid-phase shaking method is referred to as 10LPSOI-LS. Both 10LPSOI-BM and 10LPSOI-LS were synthesized using synthesis-grade Li₂S (99.9%, Mitsuwa), P₂S₅ (99%, Merck), LiI (99.99%, Sigma-Aldrich), and Li₃PO₄ (Sigma-Aldrich). LiI and Li₃PO₄ were vacuum-dried at 150 °C overnight prior to use. Li₂S, P₂S₅, and LiI were mixed in a molar ratio of 3:1:1, with 10 mol % Li₃PO₄ relative to the 3Li₂S:P₂S₅ ratio, using an agate mortar for approximately 10 min. For 10LPSOI-BM, 2.0 g of the mixed precursor powder was prepared and loaded into a 45 mL zirconia pot inside an Ar-filled glovebox with H₂O and O₂ concentrations maintained below 1 ppm. The pot contained 15 zirconia balls with a diameter of 10 mm and a total mass of approximately 45.6 g, corresponding to a ball-to-powder mass ratio of approximately 22.8:1. The zirconia pot was hermetically sealed inside the glovebox and subsequently removed for mechanical milling outside the glovebox. The sample was milled at 500 rpm using a planetary ball mill (Pulverisette 7, Fritsch Co., Ltd.). The milling process consisted of 15 milling periods of 1 h each, separated by 10 min rest intervals, corresponding to a total effective milling time of 15 h. After milling, the sealed pot was returned to the glovebox before opening, and the powder was reground in an agate mortar for approximately 10 min. The resulting powder was then placed on a titanium plate positioned above a hot plate and covered with a stainless steel cover. Heat treatment was carried out at 180 °C for 2 h inside the Ar-filled glovebox. The heat-treatment temperature of 180 °C was selected based on previous reports for mechanically milled Li₇P₂S₈I, in which heat treatment near this temperature produced high ionic conductivity. A systematic optimization of the heat-treatment temperature for Li₃PO₄-modified 10LPSOI-BM was not performed in the present study. Elemental analysis specifically targeting possible wear-derived contamination from the zirconia milling media was not performed; therefore, trace zirconium contamination cannot be excluded. For 10LPSOI-LS, the synthesis procedure was based on our previous liquid-phase shaking method. The mixed reagents were placed in a 45 mL polypropylene centrifuge tube (Labcon) together with 10 mL of ethyl propionate and 35 g of zirconia balls with a diameter of 4 mm. The suspension was shaken at 1500 rpm for 3 h. After shaking, the solvent was removed using a controlled rotary vacuum pump at room temperature, followed by stepwise heating at 50, 70, 90, and 110 °C, with a total evaporation time of 10 h. The recovered precursor was then placed on a titanium plate positioned above a hot plate and covered with a stainless steel cover. Heat treatment was carried out at 130 °C for 2 h inside an Ar-filled glovebox with H₂O and O₂ concentrations maintained below 1 ppm. Li_5.5PS_4.5Cl_1.5 was synthesized from Li₂S, P₂S₅, and LiCl (99.9%, Sigma-Aldrich) in a molar ratio of 4:1:3 by mechanical ball milling. A total of 2.0 g of precursor powder was placed in a zirconia pot together with 15 zirconia balls with a diameter of 10 mm. The sample was milled at 600 rpm for a total effective milling time of 20 h, consisting of 1 h milling periods separated by 10 min rest intervals. After milling, the powder was placed on a titanium plate positioned on a ceramic hot plate and heat-treated at 440 °C for 2 h inside an Ar-filled glovebox.

Material Characterization

The crystal structures of the retrieved samples were analyzed using X-ray diffraction (XRD; SmartLab, Rigaku) with an airtight holder to prevent exposure to air and humidity. Raman spectra were recorded using a laser Raman spectrometer (Jasco, NRS-4500) to investigate the local structure inside a sealed container, with all measurements conducted in an Ar-filled glovebox. Scanning electron microscopy (SEM) images were obtained using a Hitachi S-4800 microscope. Solid-state ³¹P magic-angle-spinning nuclear magnetic resonance (³¹P MAS NMR; Avance III 400, Bruker) measurements were performed using a standard single-pulse sequence at a spinning rate of 6 kHz. X-ray photoelectron spectroscopy (XPS) was used to analyze the surface states of the solid electrolytes. The pristine 10LPSOI-BM and 10LPSOI-LS samples were measured using an ESCA-3400 system at the Consortium for Lithium Ion Battery Technology and Evaluation Research Center (LIBTEC), whereas the postcycling samples recovered from the Li | 10LPSOI-M | Li symmetric cells were measured at the BL7U beamline of the Aichi Synchrotron Radiation Center. The spectra were processed and analyzed using CasaXPS software. High-energy X-ray total scattering measurements using synchrotron radiation were performed at the BL04B2 beamline of SPring-8.

Electrochemical Performance Measurement

The temperature dependence of the ionic conductivity of the retrieved samples was investigated using alternating current (AC) impedance spectroscopy (Novocontrol Electrochemical Impedance System, NEISYS) over a frequency range from 100 MHz to 10 mHz. The measurement temperature was controlled using a cryogenic temperature controller (Lake Shore Cryotronics, Model 336) in the range of 143 to 363 K, with temperature increments of 20 K. Samples for impedance measurements were prepared by uniaxially pressing approximately 80 mg of powder into pellets with a diameter of about 10 mm. The pressing was carried out at a pressure of 255 MPa using a uniaxial press for 10 min at room temperature. The thickness of the 10LPSOI solid electrolyte pellets was approximately 0.55 mm. Relative density, porosity, pellet-to-pellet thickness variation, and pellet microstructural uniformity were not independently quantified in the present study. The prepared pellets were placed in a polyether ether ketone (PEEK) holder with two stainless steel (SUS) blocking electrodes. Equivalent-circuit fitting of the impedance spectra was attempted; however, because the bulk and grain-boundary responses were not sufficiently resolved to obtain reliable fitting parameters, the fitted values were not used for quantitative interpretation. Accordingly, the resistance values discussed in this work were estimated from the characteristic features of the Nyquist plots. Critical current density (CCD) measurements were conducted using direct current (DC) polarization tests. Lithium metal foil with an initial thickness of approximately 0.1 mm was used as the nonblocking electrode on both sides of the pelletized solid electrolyte, and the cell was sandwiched between stainless steel plungers. The thickness of the lithium metal after cell pressing was not measured. During both the CCD and long-term lithium–metal stability measurements, the symmetric-cell fixture was tightened to a torque of 2 N·m. The prepared symmetric cells were cycled with a current-density increment of 0.025 mA cm^–2 per cycle. Each cycle consisted of 1 h of charging and 1 h of discharging using a charge–discharge system (BTS 2004H, Nagano) under a dry Ar atmosphere at room temperature. The cell preparation for lithium–metal stability measurements was identical to that used for the CCD measurements. For the long-term stability test, a constant current density of 0.30 mA cm^–2 was applied for 1 h during each plating and stripping step, corresponding to an areal capacity of 0.3 mAh cm^–2 per half-cycle. The battery performance was evaluated using both Li–In half-cells and lithium–metal full cells. The positive electrode composites were prepared by thoroughly homogenizing LiNi_0.33Mn_0.33Co_0.33O₂ (NMC₁₁₁), solid electrolyte, and acetylene black (AB) in an agate mortar for approximately 15 min at a weight ratio of 70:30:3. Prior to use, NMC₁₁₁ was coated with 1 wt % LiNbO3 and subsequently heat-treated at 300 °C for 30 min. For the Li–In half-cell, 10LPSOI-M (M = BM or LS) was used as the solid electrolyte in the cathode composite, while Li_5.5PS_4.5Cl_1.5 was used as the separator electrolyte. Approximately 10 mg of cathode composite and 80 mg of separator electrolyte were used. The cells were cycled at 30 °C and 0.1C within a voltage window of 2.0–3.7 V versus Li–In. For the lithium–metal full cell, Li_5.5PS_4.5Cl_1.5 was used as the solid electrolyte in the cathode composite, while 10LPSOI-M (M = BM or LS) was used as the separator electrolyte. Approximately 5 mg of cathode composite was deposited onto one side of the pelletized 10LPSOI-M inside a PEEK container with an inner diameter of 10 mm and then uniaxially pressed at 255 MPa at room temperature for 10 min. This corresponds to an NMC₁₁₁ mass of approximately 3.4 mg and an active-material loading of approximately 4.3 mg cm^–2 based on the 10 mm electrode diameter. Based on an NMC₁₁₁ specific capacity of 170 mAh g^–1 and the initial lithium–metal dimensions (approximately 5 mm in diameter and 0.1 mm in thickness), the estimated negative-to-positive capacity ratio (N/P ratio) was approximately 7.0. The separator-electrolyte mass and thickness were approximately 400 mg and 2.83 mm, respectively. After attaching a lithium metal sheet to the opposite side of the pellet, the cell assembly was subjected to an additional uniaxial pressing step at 127 MPa at room temperature for 5 min. During electrochemical cycling, the cell fixture was maintained at a tightening torque of 10 N·m. The assembled cells were cycled using a charge–discharge system (BST-2004H, Nagano) within a voltage window of 2.6–4.2 V versus Li/Li⁺ at 0.1C and 30 °C. For the electrochemical measurements reported in this study, one independently assembled cell was evaluated for each condition (n = 1); therefore, statistical reproducibility across replicate cells was not assessed.

Results and Discussion

The first parameter examined is the structural difference between the glass ceramic thio-LISICON II phase (tetragonal, P4₂, nmc) and the crystalline Li₄PS₄I phase derived from the Li₇P₂S₈I system. Figure S1 shows the XRD patterns of Li₇P₂S₈I prepared by mechanical ball milling before heat treatment and after heat treatment at 180 and 250 °C. It is evident that the Li₇P₂S₈I precursor becomes largely amorphous after milling. Upon heat treatment at 180 °C, partial crystallization of the thio-LISICON II phase begins to appear, which originally comes from the structure of Li₁₀P₃S₁₂I (ICSD#54663), accompanied by a significant amorphous background. This indicates that the amorphous phase is transforming into a crystalline phase, resulting in a glass ceramic structure in thio-LISICON II. , Furthermore, after Li₇P₂S₈I was heat-treated at 250 °C, the structure fully transformed into crystalline Li₄PS₄I (ICSD#123878), with no residual thio-LISICON phase. This finding is noteworthy, as it indicates that the crystal structure evolves with increasing temperature. It also suggests that obtaining a glass ceramic structure is important, since thio-LISICON II has been reported to exhibit higher ionic conductivity than crystalline Li₄PS₄I. , For clarity, the molar ratios of Li₃PS₄, Li₁₀P₃S₁₂I, Li₇P₂S₈I, and Li₄PS₄I are expressed in eqs –.

$$2\mathrm{L}{\mathrm{i}}_3\mathrm{P}{\mathrm{S}}_4\to 3\mathrm{L}{\mathrm{i}}_2\mathrm{S}+{\mathrm{P}}_2{\mathrm{S}}_5$$ 1

$$\frac{2}{3}\mathrm{L}{\mathrm{i}}_{10}{\mathrm{P}}_3{\mathrm{S}}_{12}\mathrm{I}\to 3\mathrm{L}{\mathrm{i}}_2\mathrm{S}+{\mathrm{P}}_2{\mathrm{S}}_5+\frac{2}{3}\mathrm{L}\mathrm{i}\mathrm{I}$$ 2

$$\mathrm{L}{\mathrm{i}}_7{\mathrm{P}}_2{\mathrm{S}}_8\mathrm{I}\to 3\mathrm{L}{\mathrm{i}}_2\mathrm{S}+{\mathrm{P}}_2{\mathrm{S}}_5+\mathrm{L}\mathrm{i}\mathrm{I}$$ 3

$$2\mathrm{L}{\mathrm{i}}_4\mathrm{P}{\mathrm{S}}_4\mathrm{I}\to 3\mathrm{L}{\mathrm{i}}_2\mathrm{S}+{\mathrm{P}}_2{\mathrm{S}}_5+2\mathrm{L}\mathrm{i}\mathrm{I}$$ 4

It is clear that the compounds described in eqs – are all based on Li₃PS₄ with different amounts of LiI addition. The molar ratio of LiI added in Li₁₀P₃S₁₂I (eq ) is the smallest, followed by Li₇P₂S₈I (eq ), while the largest LiI content is observed for Li₄PS₄I (eq ). In addition to temperature, the amount of LiI also contributes to the tendency for the formation of the stable crystalline Li₄PS₄I phase. Although Li₄PS₄I is reported to be more chemically stable, its ionic conductivity is relatively low. Therefore, a smaller amount of LiI combined with a lower heat-treatment temperature is crucial for achieving high ionic conductivity in the thio-LISICON II glass ceramic phase. Moreover, Li₁₀P₃S₁₂I has been identified as the parent structure of the thio-LISICON II glass ceramic phase and exhibits the highest ionic conductivity within the Li₂S–P₂S₅–LiI system. , However, it has also been reported that its chemical stability does not reach an optimal level. In contrast, Li₇P₂S₈I, which contains a higher LiI content than Li₁₀P₃S₁₂I but a lower LiI content than Li₄PS₄I, achieves an optimal balance between chemical stability and ionic conductivity. Interestingly, the chemical structure of Li₇P₂S₈I can be considered as a combination of Li₁₀P₃S₁₂I and Li₄PS₄I, as expressed in eq , which explains its enhanced stability while maintaining high ionic conductivity.

$$\mathrm{L}{\mathrm{i}}_7{\mathrm{P}}_2{\mathrm{S}}_8\mathrm{I}\to \frac{1}{2}\mathrm{L}{\mathrm{i}}_{10}{\mathrm{P}}_3{\mathrm{S}}_{12}\mathrm{I}+\frac{1}{2}\mathrm{L}{\mathrm{i}}_4\mathrm{P}{\mathrm{S}}_4\mathrm{I}\to 3\mathrm{L}{\mathrm{i}}_2\mathrm{S}+{\mathrm{P}}_2{\mathrm{S}}_5+\mathrm{L}\mathrm{i}\mathrm{I}$$ 5

The next study aims to reproduce the synthesis conditions previously established by our group for 10LPSOI-LS. In our earlier work, 10LPSOI-LS heat-treated at 130 °C for 2 h was found to exhibit the highest ionic conductivity. Figure S2 shows the XRD patterns of 10LPSOI-LS, indicating significant variability in the crystalline structures among the four trials. Most of the samples exhibit residual LiI, high-intensity Li₄PS₄I reflections, the presence of Li₃PS₄, and a low intensity of the thio-LISICON II phase. Ideally, the optimal structure of 10LPSOI-LS should exhibit the highest possible intensity of the thio-LISICON II phase while minimizing the formation of Li₄PS₄I with no Li₃PS₄, and residual starting materials. In addition, it is worth noting that in trials one and four, a pair of closely spaced peaks appears at similar positions around 26–27°, which can be attributed to LiI and Li­(IO₃). The key difference between these two phases is that residual LiI typically produces sharp diffraction peaks with high crystallinity, whereas Li­(IO₃) exhibits broader peaks, indicating lower crystallinity. The formation of Li₃PS₄ inhibits the crystallization and growth of the thio-LISICON II phase, as precrystallization of Li₃PS₄ during the reaction process in the ethyl propionate (EP) medium hinders the incorporation of LiI into the Li₂S–P₂S₅ framework, thereby suppressing the formation and the nucleation of the glass ceramic thio-LISICON II crystalline structure.

To overcome the crystallization instability of 10LPSOI, the precrystallization of Li₃PS₄ must be suppressed to facilitate the incorporation of LiI into the Li₂S–P₂S₅ matrix. Therefore, instead of mixing all reagents simultaneously as in previous work, LiI was first dissolved in 10 mL of EP, followed by the addition of the 3Li₂S–P₂S₅ mixture into the LiI-dissolved EP solution. Figure shows the XRD pattern result of 10LPSOI-BM and 10LPSOI-LS with modified liquid phase shaking method. Both 10LPSOI-BM and 10LPSOI-LS exhibit fully amorphous structures prior to heat treatment. After heat treatment at 180 °C, 10LPSOI-BM exhibits the formation of a highly crystalline thio-LISICON II phase with very low crystallinity intensity of Li₄PS₄I, which is in agreement with eq The relatively low heat-treatment temperature suppresses the formation of Li₄PS₄I, resulting in its low peak intensity. Similarly, 10LPSOI-LS successfully forms a high-intensity thio-LISICON II phase, accompanied by a lower crystalline intensity of Li₄PS₄I compared to the conventional method after being heat-treated at 130 °C. Both 10LPSOI-BM and 10LPSOI-LS solid electrolytes show no remaining residual starting materials, such as Li₂S and LiI. The main difference between 10LPSOI-BM and 10LPSOI-LS is the formation of Li­(IO₃) in 10LPSOI-LS, which indicates a reaction between LiI and oxygen originating from the ethyl propionate solvent. Li­(IO₃) exhibits a significantly lower ionic conductivity (∼10^–12 S cm^–1) compared with LiI (∼10^–8 S cm^–1). , Nevertheless, the influence of Li­(IO₃) on the electrochemical stability of the 10LPSOI-LS solid electrolyte system remains unclear. Quantitative phase analysis was not performed because the XRD patterns contain substantial amorphous contributions and overlapping reflections from the glass–ceramic and secondary phases. Therefore, the amounts of Li­(IO₃), Li₄PS₄I, and other secondary phases are discussed qualitatively based on their diffraction features.

1.

XRD pattern result of 10LPSOI-BM and 10LPSOI-LS before and after heat treatment process.

For further structural comparison, Raman spectroscopy measurements were conducted, as shown in Figure . The results indicate that both 10LPSOI-BM and 10LPSOI-LS exhibit very similar Raman peak features, suggesting comparable local structures. In addition, neither sample shows the presence of D or G bands, which are typically associated with carbon species originating from solvent degradation after heat treatment. This observation is expected for 10LPSOI-BM, as no solvent is used during its synthesis. Importantly, the absence of detectable carbon-related Raman bands in 10LPSOI-LS indicates that no substantial Raman-active carbonaceous species were observed after solvent removal; however, trace surface carbon species cannot be excluded by Raman spectroscopy. Moreover, a sharp peak corresponding to the PS₄ ^3– anion exhibits the highest intensity in both 10LPSOI-BM and 10LPSOI-LS at 418 cm^–1 and 419 cm^–1, respectively, followed by a weaker peak around 283 cm^–1 and a P–S stretching mode at 555 cm^–1. Notably, the main PS₄ ^3– Raman band of 10LPSOI-LS exhibits a slight shift toward higher wavenumbers, suggesting subtle differences in the local structural environment compared to 10LPSOI-BM. Accordingly, peak deconvolution was performed, as the 404–420 cm^–1 region in thiophosphate-based sulfide electrolytes is known to contain overlapping contributions from P₂S₇ ^4– and PS₄ ^3– anion. As a result, 10LPSOI-BM exhibits a higher proportion of P₂S₇ ^4– relative to PS₄ ^3–, whereas 10LPSOI-LS shows the opposite trend, with PS₄ ^3– being more dominant than P₂S₇ ^4–. This difference in anion distribution explains the slight shift observed in the main Raman band between the two solid electrolytes. In addition, both samples exhibit a weak Raman feature at ∼ 385 cm^–1, corresponding to the presence of P₂S₆ ^4– anion.

2.

Raman spectroscopy of 10LPSOI-BM and 10LPSOI-LS after the heat treatment process.

³¹P MAS NMR measurements were conducted to achieve a deeper structural analysis of both 10LPSOI-BM and 10LPSOI-LS, as shown in Figure a. The magnified regions focusing on P–S–O chemical environments are presented in Figure b and c for 10LPSOI-BM and 10LPSOI-LS, respectively. As shown in Figure a, a Li₃PO₄ reference was used to identify the PO₄ ^3– resonance at 12.8 ppm. Both 10LPSOI-BM and 10LPSOI-LS exhibit a PO₄ ^3– signal at the same chemical shift as the reference, originating from the addition of 10 mol % Li₃PO₄. Notably, crystalline Li₃PO₄ was not detected in the XRD patterns shown in Figure , indicating that the introduced Li₃PO₄ exists in an amorphous state and is incorporated within the Li₂S–P₂S₅–LiI matrix. , As observed in Figure b, 10LPSOI-BM exhibits the highest intensity corresponding to the P₂S₇ ^4– anion at 91 ppm, followed by the PS₄ ^3– anion at 86 ppm, while the lowest intensity is attributed to the P₂S₆ ^4– anion at 109 ppm. In contrast, Figure c shows that 10LPSOI-LS exhibits a higher ratio of the PS₄ ^3– anion at 89 ppm compared to the P₂S₇ ^4– anion at 91 ppm, while displaying a similar intensity of the P₂S₆ ^4– anion at 109 ppm to that observed for 10LPSOI-BM. This distribution of thiophosphate anions is consistent with the Raman results shown in Figure . In 10LPSOI-LS, additional peaks are observed at 94.3 and 75 ppm, which can be attributed to PS₄ ^3– tetrahedral units similar to those found in the LGPS structure. These signals correspond to distinct phosphorus environments commonly assigned as PS₄ ^3– (P1) and PS₄ ^3– (P2), respectively. This observation is particularly interesting, as the thio-LISICON II structure is derived from the thio-LISICON framework related to the LGPS structure, which can also be formed within the Li₂S–P₂S₅–LiI system, in agreement with previous findings for the 10LPSOI-LS composition. , At the same time, the PS₄ ^3– tetrahedral units (P1 and P2) are also observed at comparable chemical shift positions in 10LPSOI-BM, where the signals appear broader due to the higher amorphous content of the sample. Oxysulfide-related resonances attributed to POS₃ ^3– environments are observed at different chemical shifts for 10LPSOI-BM and 10LPSOI-LS, appearing at approximately 81.1 and 85.2 ppm, respectively. The difference in chemical shift indicates that the two synthesis routes produce distinct local phosphorus environments involving P–S–O coordination. However, assignment of these resonances to specific P = O, P = S, or oxygen-bridged configurations cannot be established conclusively from the present³¹P MAS NMR measurements alone. Previous studies have associated oxygen-containing thiophosphate environments with improved chemical and electrochemical stability; therefore, the observed differences may be relevant to the distinct interfacial behavior of 10LPSOI-BM and 10LPSOI-LS. ,,

3.

(a) Comparison of the ³¹P MAS NMR spectra. Enlarged views of the selected chemical shift range are shown for (b) 10LPSOI-BM and (c) 10LPSOI-LS after the heat-treatment process.

After examining the chemical and crystal structures, the morphologies of 10LPSOI-BM and 10LPSOI-LS were analyzed by SEM, as shown in Figure a and b, respectively. As observed, 10LPSOI-BM (Figure a) exhibits relatively large particles with an uneven spherical morphology. In contrast, 10LPSOI-LS (Figure b) shows a similar particle shape but with a finer and more homogeneous particle size distribution. It is well-known that liquid-phase synthesis proceeds via bottom-up particle formation via nucleation and grain growth, whereas mechanical ball milling follows a top-down approach by crushing large particles into smaller ones. As a result, liquid-phase synthesis is generally more favorable for producing finer solid-electrolyte particles. For deeper analysis, particle size calculations were performed for both 10LPSOI-BM and 10LPSOI-LS using ImageJ software based on SEM images. Figure S3 presents the particle size distribution histogram of 10LPSOI-BM. From the histogram, the smallest particle size is 0.69 μm and the largest particle size is 35 μm, with an average particle size of 5.9 ± 4.2 μm. The D50 value for 10LPSOI-BM is estimated at approximately 5.1 μm. In contrast, the particle size distribution histogram of 10LPSOI-LS shown in Figure S4 reveals a minimum particle size of 75 nm (0.075 μm), which is significantly smaller than that of 10LPSOI-BM. The largest particle size of 10LPSOI-LS is 4877 nm (4.9 μm), with an average particle size of 637 ± 450 nm (0.64 ± 0.45 μm) and a D50 value of approximately 508 nm (0.51 μm). These results clearly indicate that the liquid-phase shaking method produces approximately 10-fold finer and more homogeneous 10LPSOI particles than the mechanical ball-milling synthesis process.

4.

SEM images with the selected magnification of (a)­10LPSOI-BM and (b)­10LPSOI-LS after the heat treatment process.

Figure a shows the temperature dependence of the ionic conductivity of 10LPSOI-BM and 10LPSOI-LS. At room temperature, 10LPSOI-BM exhibits a high ionic conductivity of 3.4 mS cm^–1 with an activation energy of 28.67 kJ mol^–1. In comparison, 10LPSOI-LS shows a lower but still comparable ionic conductivity of 0.7 mS cm^–1 while exhibiting the same activation energy of 28.67 kJ mol^–1. The similarity in activation energies suggests comparable overall energy barriers for lithium-ion migration over the measured temperature range. However, this does not exclude synthesis-dependent differences in crystallinity, local structure, phase characteristics, or grain-boundary contributions between 10LPSOI-BM and 10LPSOI-LS. At the same time, 10LPSOI-LS exhibits a substantially smaller particle size than 10LPSOI-BM, as shown in Figure . The lower room-temperature ionic conductivity of 10LPSOI-LS may therefore arise from a combination of synthesis-dependent structural and microstructural factors, including differences in phase characteristics, local phosphorus environments, particle size, and interparticle resistance. Figure b presents the Nyquist plots of 10LPSOI-BM and 10LPSOI-LS measured at −130 °C (143 K), allowing the separation and evaluation of bulk and grain-boundary resistances. As a result, 10LPSOI-LS does not exhibit a clearly distinguishable bulk resistance, even at high frequencies up to 100 MHz. The semicircle observed around 1 Hz corresponds to the combined contribution of bulk and grain-boundary resistances, indicating a total resistance of 2.47 × 10⁸ Ω. A similar trend is observed for 10LPSOI-BM, as shown in the inset of Figure b, where the separation of bulk resistance remains challenging. Although 10LPSOI-BM displays two barely distinguishable semicircles, one in the high-frequency region (100 MHz to 1 kHz) that may be attributed to bulk resistance and another in the intermediate-frequency region (100 to 1 Hz) corresponding to the total resistance including grain-boundary contributions, the bulk component is still difficult to isolate due to the incomplete formation of well-defined semicircles, making reliable fitting challenging. Therefore, for 10LPSOI-BM, only the total resistance can be identified at approximately 1 Hz, exhibiting a value of 4.7 × 10⁷ Ω. The impedance spectra are consistent with the glass-ceramic character of both materials, in which overlapping bulk and grain-boundary responses may occur. However, because these contributions are not sufficiently resolved for reliable fitting, the present EIS data do not allow a quantitative comparison of crystallite size or a reliable separation of bulk and grain-boundary resistances between 10LPSOI-BM and 10LPSOI-LS. These observations are consistent with the glass ceramic nature of the materials. Glass ceramics represent a transitional structural state between fully amorphous and fully crystalline phases, in which crystallites are present but not fully developed. In fully crystalline materials, grains consist of crystallites aligned in similar orientations, while grain boundaries correspond to the interfaces between differently oriented crystallites. In contrast, glass ceramics contain randomly oriented and partially developed crystallites, which may limit the formation of well-defined grain structures. This structural characteristic likely explains the absence of a clearly resolved bulk resistance contribution in the impedance spectra.

5.

(a) Temperature dependence of the ionic conductivity of 10LPSOI-BM and 10LPSOI-LS. (b) Nyquist plots measured at −130 °C (143 K) for 10LPSOI-BM and 10LPSOI-LS.

DC polarization measurements were performed to further evaluate the critical current density (CCD) of Li | 10LPSOI-M (M = BM or LS) | Li symmetric cells in order to assess the stability of the 10LPSOI solid electrolytes against lithium metal, which is crucial for ASSLMB applications. The measurements were conducted with an initial current density of 0.1 mA cm^–2, followed by incremental increases of 0.025 mA cm^–2 per cycle. Each cycle consisted of 1 h of charging and 1 h of discharging, and the measurements were conducted at 30 °C. Figure a shows the CCD measurement for 10LPSOI-BM. The initial overpotential was 4.9 mV at a current density of 0.1 mA cm^–2. The voltage gradually increased with increasing current density, reaching a maximum of 10 mV at 0.3 mA cm^–2. In the subsequent cycle, the voltage suddenly dropped to 3 mV, indicating the formation of a low-resistance internal short-circuit pathway. This result indicates the CCD of 10LPSOI-BM is 0.3 mA cm^–2 at 30 °C. The CCD measurement of 10LPSOI-LS is shown in Figure b. The 10LPSOI-LS solid electrolyte exhibits an initial overpotential of 17 mV at a current density of 0.1 mA cm^–2. This value is approximately four times that of 10LPSOI-BM, consistent with its lower ionic conductivity. The reduced ionic conductivity of 10LPSOI-LS indicates higher internal resistance, resulting in a larger overpotential during polarization measurements in accordance with Ohm’s law. Over the subsequent cycles, 10LPSOI-LS exhibits a significant increase in overpotential, reaching a maximum of 149 mV at a current density of 0.925 mA cm^–2 before experiencing an internal short circuit in the following cycle, as indicated by a sudden drop in overpotential to 40.5 mV. These results indicate that 10LPSOI-LS achieves a significantly higher CCD of 0.925 mA cm^–2, compared with 0.30 mA cm^–2 for 10LPSOI-BM. Because each plating or stripping step was maintained for 1 h, these CCD values correspond to areal capacities of 0.925 and 0.30 mAh cm^–2 per half-cycle for 10LPSOI-LS and 10LPSOI-BM, respectively. Despite its lower ionic conductivity, 10LPSOI-LS exhibits improved tolerance toward lithium plating and stripping under the tested conditions. This improved behavior is likely associated with synthesis-dependent differences in particle morphology, phase characteristics, local structure, and surface chemistry. The substantially smaller particle size of 10LPSOI-LS may contribute to improved interfacial contact with lithium metal; however, the present results do not isolate particle size as the sole origin of the enhanced CCD. In addition, the differences observed in the³¹P MAS NMR spectra suggest distinct local phosphorus environments in 10LPSOI-LS and 10LPSOI-BM, although the specific contribution of these local structural features to lithium–metal stability cannot be conclusively established from the present measurements. ,,

6.

Critical current density (CCD) results of (a) 10LPSOI-BM and (b) 10LPSOI-LS at an increment rate of 0.025 mA cm^–2 per cycle at 30 °C. Lithium metal stability measurements at a constant current density of 0.3 mA cm^–2 and 30 °C for (c) 10LPSOI-BM and (d) 10LPSOI-LS. One cycle corresponds to 1 h of charging and 1 h of discharging.

Lithium metal stability measurements were subsequently conducted using separately prepared Li | 10LPSOI-M (M = BM or LS) | Li symmetric cells with the same configuration as those used for the CCD measurements. These measurements were performed to evaluate the electrochemical stability of each solid electrolyte against lithium metal under a constant current density of 0.3 mA cm^–2. The lithium stability test can be considered analogous to a fatigue test in mechanical property evaluation, where the CCD corresponds to the ultimate tensile strength (UTS), and the applied current represents the mechanical load during cyclic deformation. Accordingly, a constant current density of 0.3 mA cm^–2 was selected, as this value corresponds to the CCD of 10LPSOI-BM at 30 °C. The lithium metal stability measurement of 10LPSOI-BM is shown in Figure c. The initial overpotential was 17 mV. However, the cell remained stable for only one cycle (1 h charge and 1 h discharge). In the subsequent cycle, the overpotential dropped to 4.3 mV, consistent with the formation of a low-resistance internal short-circuit pathway. In addition, the lithium metal stability of 10LPSOI-BM was evaluated at a lower constant current density of 0.1 mA cm^–2 at 30 °C, as shown in Figure S5, to further verify its stability against lithium metal in case 0.3 mA cm^–2 was near the CCD limit of 10LPSOI-BM. As a result, the initial overpotential of 10LPSOI-BM at 0.1 mA cm^–2 was 11.7 mV and remained stable up to the 24th cycle (48 h). In the 25th cycle, a sudden drop in overpotential to 9 mV was observed, indicating the development of an internal short-circuit pathway. Although 10LPSOI-BM exhibited improved stability at 0.1 mA cm^–2, the overall cycling performance remains limited. In contrast, the lithium metal stability of 10LPSOI-LS at a current density of 0.3 mA cm^–2 is presented in Figure S6 for the complete data set and in Figure d for the selected voltage range, allowing clearer observation of the charge–discharge profiles. Both figures show consistent electrochemical behavior for 10LPSOI-LS. The initial overpotential was approximately 50 mV, which is reasonably attributed to the higher interfacial resistance of 10LPSOI-LS compared to 10LPSOI-BM. The overpotential remained highly stable throughout cycling, with only minor fluctuations observed. These small variations are likely due to temperature changes in the chamber during opening and closing operations. The cell maintained an overpotential of approximately 50 mV up to the 1546th cycle (3092 h). Thereafter, the overpotential decreased to 41 mV at the 1547th cycle and gradually declined, eventually stabilizing at approximately 12 mV by the 1560th cycle (3120 h). The gradual decrease in overpotential after approximately 3092 h suggests the development of a partial or resistive internal electronic pathway, consistent with soft-short behavior. Because the cell voltage did not collapse to nearly 0 V, unlike 10LPSOI-BM, the internal short-circuit pathway in 10LPSOI-LS appears to have remained incomplete or highly resistive during this stage. However, the extent of lithium dendrite penetration cannot be confirmed without direct postcycling cross-sectional characterization. Overall, a comparison of the lithium metal stability results clearly indicates that, although 10LPSOI-LS has lower ionic conductivity, its stability against lithium metal is significantly higher than that of 10LPSOI-BM. Both solid electrolytes exhibit oxysulfide-related local phosphorus environments, which may be associated with their electrochemical behavior against lithium metal. However, the present measurements do not establish a direct causal relationship between a specific POS₃ ^3– configuration and the improved stability of 10LPSOI-LS. In addition, the substantially smaller particle size of 10LPSOI-LS may promote improved interparticle contact and more uniform lithium-ion transport pathways. Nevertheless, because pellet relative density and porosity were not directly measured, the present study cannot quantitatively determine whether 10LPSOI-LS forms a denser electrolyte layer. Previous studies have shown that interfacial voids and poorly contacted regions can promote nonuniform lithium deposition and dendrite growth, which is consistent with the improved lithium–metal stability observed for the finer-particle 10LPSOI-LS sample. ,

For deeper analysis, EIS measurements were conducted for Li | 10LPSOI-M (M = BM or LS) | Li symmetric cells before cycling and after multiple cycles. The measurements were performed over a frequency range from 1 MHz to 0.01 Hz, with intervals of 25 cycles (50 h). The EIS results obtained during DC polarization for 10LPSOI-BM are shown in Figure S7. Prior to cycling (0 cycles), 10LPSOI-BM exhibited two semicircles in the Nyquist plot at high and low frequencies. The high-frequency semicircle corresponds to a resistance of approximately 18 Ω at 30 °C, which can be attributed to the bulk and grain-boundary resistance of the 10LPSOI-BM solid electrolyte. The low-frequency semicircle, observed in the range of 116–153 Ω, is associated with the interfacial resistance at the solid electrolyte-lithium metal interface. This observation is noteworthy, as it indicates that an interfacial layer had already formed prior to the application of DC current, suggesting a relatively resistive interphase at the solid electrolyte–lithium interface. After 25 cycles, the impedance spectrum evolved into a nearly flat response with significantly reduced impedance magnitude, resembling metallic behavior. This response is consistent with the formation of a low-resistance internal short-circuit pathway and agrees with the DC polarization behavior shown in Figure c. However, the EIS data alone do not establish the physical origin of the short circuit, and lithium dendrite penetration cannot be confirmed without direct postcycling cross-sectional characterization. On the other hand, the EIS measurements of 10LPSOI-LS during DC polarization, as shown in Figure S8, exhibit a larger high-frequency semicircle with a resistance of approximately 59 Ω compared to 10LPSOI-BM. This high-frequency response corresponds to the bulk and grain-boundary resistance of 10LPSOI-LS. The higher resistance is consistent with its lower ionic conductivity relative to 10LPSOI-BM. At lower frequencies, 10LPSOI-LS exhibits a broad semicircle in the range of 80–150 Ω, which corresponds to the interfacial resistance associated with the Li | 10LPSOI-LS | Li interface. After 25 cycles of DC polarization, the 10LPSOI-LS symmetric cell maintains a similar high-frequency resistance of approximately 59 Ω. However, the semicircle broadens as Z″ decreases, suggesting improved interfacial uniformity between 10LPSOI-LS and lithium metal. At the same time, the high-frequency semicircle of 10LPSOI-LS decreases in size, with the interfacial resistance falling within the range of 80–118 Ω and exhibiting a lower Z″ magnitude. This behavior may be attributed to the formation of a more stable SEI after cycling. A similar impedance profile was observed after the Li | 10LPSOI-LS | Li cell was cycled for 50 and 75 cycles, indicating no significant change at the Li | 10LPSOI-LS | Li interface. This result is consistent with the DC polarization stability test shown in Figure d, where 10LPSOI-LS maintained a nearly constant overpotential from the initial cycle up to the 75th cycle (150 h) of charge and discharge. However, because EIS measurements were conducted only up to 75 cycles, changes in impedance during the subsequent long-term cycling period cannot be directly evaluated from the present data. Overall, the EIS results indicate that 10LPSOI-BM exhibits lower solid-electrolyte resistance but undergoes rapid impedance collapse after cycling, whereas 10LPSOI-LS maintains a comparatively stable impedance response over the measured 75-cycle period. These results are consistent with the improved short-term interfacial stability of 10LPSOI-LS under the tested conditions.

To comprehensively analyze the structural changes in both 10LPSOI-BM and 10LPSOI-LS solid electrolytes, XPS measurements were conducted before and after DC polarization during the lithium metal stability test. The spectra of selected elements, including Li 1s, P 2p, S 2p, and O 1s, are presented in Figure . For the post-DC polarization analysis, the solid electrolyte samples were collected from the interface between the solid electrolyte and lithium metal after cell disassembly, either upon completion of the measurement or after short circuiting, and are denoted as “after cycle” in Figure . Because the postcycling 10LPSOI-BM and 10LPSOI-LS samples were recovered after different electrochemical cycling histories, the XPS comparison is interpreted qualitatively and does not represent a cycle-matched comparison of interfacial degradation. The term “pristine” in Figure refers to the solid electrolyte prior to cell assembly in the Li | 10LPSOI-M | Li configuration.

7.

XPS spectral decomposition of the Li 1s, P 2p, S 2p, and O 1s regions of pristine and postcycling 10LPSOI-BM and 10LPSOI-LS. The postcycling samples were collected from the solid electrolyte/Li interface after the lithium–metal stability measurements corresponding to Figure c and d.

The Li 1s spectra of pristine 10LPSOI-BM and 10LPSOI-LS show similar contributions attributed to lithium within the thiophosphate framework, assigned as Li–P–S, and to lithium iodide (Li–I), located at approximately 58.1 eV and 51–53 eV, respectively. These results are consistent with the Li₂S–P₂S₅–LiI composition at a molar ratio of 3:1:1. After cycling, both 10LPSOI-BM and 10LPSOI-LS exhibit a pronounced high-intensity peak at approximately 57.7 eV, which can be attributed to overlapping contributions from Li₂P and Li₃P. This indicates the formation of a solid electrolyte interphase at the lithium interface. Notably, 10LPSOI-BM also shows an additional shoulder peak at approximately 56.1 eV, which can be assigned to Li–O bonding, possibly associated with the formation of Li₂O. This additional feature is consistent with chemical alteration of the 10LPSOI-BM interface under its specific cycling history.

The P 2p spectra of pristine 10LPSOI-BM and 10LPSOI-LS exhibit similar features, with contributions at approximately 131.8, 132.7, and 133.5 eV that are assigned to PS₄ ^3–, oxysulfide-related POS₃ ^3–, and PO₄ ^3– environments, respectively. These assignments are consistent with the ³¹P MAS NMR results shown in Figure . After the lithium metal stability test, 10LPSOI-BM exhibits clear signs of interfacial degradation, whereas the structure of 10LPSOI-LS remains largely unchanged. In 10LPSOI-BM, reduced phosphorus appears at approximately 130.7 eV, which can be attributed to Li _x P. , Simultaneously, an oxidized phosphate-like contribution emerges at around 135 eV, assigned to PO₃ ^–-type environments. The presence of Li _x P together with PO₃ ^–-related components suggests interfacial decomposition of PS₄ ^3– units upon contact with lithium metal. The formation of Li _x P likely results from reductive cleavage of P–S bonds, while oxygen-containing units may reorganize into phosphate-like structures. , This decomposition behavior is more pronounced in 10LPSOI-BM, indicating greater interfacial instability than in 10LPSOI-LS.

In the S 2p region, both pristine 10LPSOI-BM and 10LPSOI-LS exhibit peaks at approximately 162.6 and 163.8 eV, which are attributed to PS₄ ^3– units. A slight additional contribution at around 165 eV is observed in pristine 10LPSOI-BM, which can be assigned to S–S bonding. The presence of this S–S component may indicate a higher degree of structural disorder or residual amorphous phase in 10LPSOI-BM. After the lithium metal stability cycling process, both 10LPSOI-BM and 10LPSOI-LS show the formation of a Li₂S component at approximately 161.1 eV, indicating interfacial reduction of the sulfide framework and the formation of a solid electrolyte interphase. , In addition, 10LPSOI-BM exhibits an additional contribution centered at around 169 eV, which can be assigned to SO₄ ^2– with sulfur in the S⁶⁺ oxidation state. This feature is not observed in 10LPSOI-LS after cycling. Considering the strongly reducing environment of lithium metal, direct electrochemical oxidation to sulfate is unlikely. Instead, this contribution is more plausibly attributed to chemical oxidation of highly reactive sulfur intermediates generated during interfacial decomposition of 10LPSOI-BM. The sulfate-related contribution observed in the postcycled 10LPSOI-BM sample may reflect greater chemical alteration under its specific cycling history; however, because the BM and LS samples experienced different cycling durations and failure states, this difference should be interpreted qualitatively rather than as evidence of an intrinsically more defect-rich interphase.

The O 1s spectra of pristine 10LPSOI-BM and 10LPSOI-LS exhibit a dominant peak at approximately 531.1 eV, which can be assigned to Li–O–P bonding environments. This contribution likely originates from amorphous Li₃PO₄ and oxysulfide units, consistent with the ³¹P MAS NMR results. Interestingly, pristine 10LPSOI-LS shows a small shoulder at approximately 532.5 eV, which may be associated with an oxygen-containing phosphorus environment, potentially including P–O–P-type bonding; however, a specific oxygen-bridged configuration cannot be established conclusively from the present XPS data. In addition, 10LPSOI-LS presents a weak shoulder at around 529.6 eV corresponding to Li–O bonding. Although this feature could be related to Li₂O, no clear Li₂O signal is observed in the Li 1s spectrum, suggesting that the contribution may instead arise from lattice oxygen or a thin passivation layer at the surface of 10LPSOI-LS.

Figure S9 shows the C 1s spectra of both 10LPSOI-BM and 10LPSOI-LS. Both solid electrolytes display similar carbon-related features, although differences in intensity are observed. The C–C and C–H component in 10LPSOI-BM is approximately 44,000 CPS, which is lower than that of 10LPSOI-LS at approximately 85,000 CPS. The carbon detected in 10LPSOI-BM can be attributed to the conductive carbon tape used for XPS mounting, whereas the higher carbon signal in 10LPSOI-LS is likely derived from residual hydrocarbon fragments originating from the ethyl propionate solvent, consistent with previous reports. The C–H and C–C bonds of 10LPSOI-LS were not detected in the Raman spectra shown in Figure . This absence may be attributed to the limited penetration depth of the Raman laser, which may prevent detection of a very thin hydrocarbon passivation layer on the particle surface. In addition, no D band (∼1350 cm^–1) or G band (∼1580 cm^–1) was observed in the Raman spectra. This result indicates that no graphitic carbon structures were formed in either 10LPSOI-BM or 10LPSOI-LS. In contrast, XPS is a highly surface-sensitive technique with a probing depth of several nanometers and is capable of detecting trace amounts of surface contaminants. Therefore, the C 1s signal observed in Figure S9 is attributed to small amounts of hydrocarbon components derived from residual solvent fragments or adventitious carbon adsorbed on the particle surface. The coexistence of oxygen-containing surface species and carbon-related components in 10LPSOI-LS suggests a distinct surface chemical environment compared with 10LPSOI-BM. However, the present XPS data do not establish that these species form a continuous passivation layer or directly cause the improved lithium–metal stability. , After the lithium metal stability cycling process, the O 1s spectra of both 10LPSOI-BM and 10LPSOI-LS remain largely unchanged, indicating that the oxygen bonding environments are not significantly affected by cycling. Overall, the XPS results indicate that 10LPSOI-BM undergoes significant interfacial degradation after cycling, as evidenced by the formation of PO₃ ^– and SO₄ ^2– components. These products may reflect interfacial decomposition and the exposure of reactive surface species; however, their physical origin cannot be determined from the present XPS measurements alone. In contrast, the interface of 10LPSOI-LS remains comparatively stable, as no additional oxidation-related components are observed after cycling. The surface-sensitive XPS results indicate differences in the chemical environments of 10LPSOI-BM and 10LPSOI-LS after cycling. These differences may be associated with their distinct interfacial behavior; however, the formation, continuity, and direct stabilizing role of a hydrocarbon–oxygen passivation layer cannot be conclusively established from the present measurements.

Battery performance measurements were conducted to evaluate the compatibility of 10LPSOI with an oxide cathode in ASSLMB configurations. The cells were assembled using NMC₁₁₁ as the cathode material and Li–In as the anode, forming half-cell configurations, as illustrated in Figure S10. The cathode composite consisted of NMC₁₁₁, 10LPSOI as the ionic conductor additive, Li_5.5PS_4.5Cl_1.5 served as the ionic conductor for the separator between anode and cathode, and acetylene black as the electronic conductive additive, and the electrochemical performance was evaluated at 30 °C under a 0.1C rate within a voltage window of 2.0–3.7 V vs Li–In. As a starting point, the comparison of the half-cell performance using 10LPSOI-BM and 10LPSOI-LS is presented in Figure a and Figure b, respectively. The initial charge and discharge capacities of the 10LPSOI-BM cell with NMC₁₁₁ are 187.5 mAh g^–1 and 144.8 mAh g^–1, respectively. The first- and second-cycle discharge capacities were 144.8 and 144.9 mAh g^–1, respectively, indicating essentially unchanged discharge capacity between the two cycles. Because the first- and second-cycle discharge profiles are nearly identical, they substantially overlap in Figure a. An enlarged comparison of these two cycles is provided in Figure S11. In contrast, the 10LPSOI-LS cell exhibits a higher initial charge capacity of 198.5 mAh g^–1 but a slightly lower initial discharge capacity of 137.8 mAh g^–1. The higher charge capacity observed for 10LPSOI-LS may be attributed to its smaller particle size, which facilitates more effective Li⁺ transport pathways within the cathode composite. However, its lower bulk ionic conductivity may contribute to increased polarization during discharge, resulting in a reduced discharge capacity. At the second cycle, the charge and discharge capacities of the 10LPSOI-BM cell are 149.7 mAh g^–1 and 144.9 mAh g^–1, respectively, while those of the 10LPSOI-LS cell are 143.3 mAh g^–1 and 138.0 mAh g^–1, respectively. Upon further cycling, the 10LPSOI-BM half-cell exhibits more pronounced capacity fading compared to the 10LPSOI-LS half-cell, as shown in Figure c. By the 20th cycle, the 10LPSOI-LS half-cell catches up to the 10LPSOI-BM cell in both charge and discharge capacities, delivering 136.5 mAh g^–1 and 133.7 mAh g^–1, respectively. In contrast, the 10LPSOI-BM cell shows capacities of 136.2 mAh g^–1 and 132.9 mAh g^–1 at the same cycle. At the 50th cycle, 10LPSOI-LS maintains charge and discharge capacities of 121.5 mAh g^–1 and 118.7 mAh g^–1, respectively, whereas 10LPSOI-BM declines to 97.0 mAh g^–1 and 94.7 mAh g^–1. Moreover, the Coulombic efficiency of the 10LPSOI-BM half-cell at the 50th cycle is 97.6%, with a capacity retention of 65.4%. In contrast, the 10LPSOI-LS half-cell exhibits a slightly higher Coulombic efficiency of 97.7% at the 50th cycle, along with a significantly improved capacity retention of 86%. The superior capacity retention of the 10LPSOI-LS half-cell indicates enhanced electrochemical stability compared to 10LPSOI-BM, consistent with the XPS results. Furthermore, the smaller particle size of 10LPSOI-LS likely facilitates improved ion transport pathways and reduces interfacial voids between the solid electrolyte and electrode components. Finer solid electrolyte particles promote more uniform microstructures and enhance charge transport properties, thereby contributing to improved overall electrochemical performance in ASSLMBs.

8.

Charge–discharge capacity curves of Li–In | Li_5.5PS_4.5Cl_1.5 | 70NMC₁₁₁ + 30­(10LPSOI-BM or 10LPSOI-LS) + 3AB cells using (a) 10LPSOI-BM and (b) 10LPSOI-LS as ionic conductor additives in the cathode composite, measured at 30 °C and 0.1C with a voltage window of 2.0–3.7 V vs Li–In. (c) Cycling performance of 10LPSOI-LS and 10LPSOI-BM over 50 cycles at 0.1C and 30 °C.

Following the half-cell evaluation, in which 10LPSOI-LS demonstrated superior cycling stability compared to 10LPSOI-BM, lithium metal full-cell performance was further investigated. The lithium metal battery was assembled in the configuration of Li | 10LPSOI-M (M = BM or LS) | 70NMC₁₁₁ + 30­(Li_5.5PS_4.5Cl_1.5) + 3AB. In this configuration, 10LPSOI served as the primary ionic conductor in the separator layer between the anode and cathode, Li_5.5PS_4.5Cl_1.5 acted as the ionic conductor additive in the cathode composite, and lithium metal was used as the anode, as illustrated in Figure S12. The battery electrochemical performance was performed at 30 °C under a 0.1C rate within a voltage window of 2.6–4.2 V vs Li/Li⁺. Figure a and b depict the charge–discharge curves of full-cell ASSLMBs employing 10LPSOI-BM and 10LPSOI-LS, respectively. The initial charge and discharge capacities of the 10LPSOI-BM full cell are 166.1 mAh g^–1 and 125.5 mAh g^–1, respectively, corresponding to an initial Coulombic efficiency of 75.5%. After the first cycle, the second-cycle charge capacity decreases to 132.2 mAh g^–1, while the discharge capacity reaches 126.5 mAh g^–1, resulting in a Coulombic efficiency of 95.7%. At the same time, the 10LPSOI-LS full cell exhibits a slightly lower but comparable initial charge capacity of 145.3 mAh g^–1, with a discharge capacity of 104.8 mAh g^–1, corresponding to an initial Coulombic efficiency of 72%. After the first cycle, the second-cycle charge and discharge capacities of 10LPSOI-LS reach 116.0 and 106.7 mAh g^–1, respectively, with a Coulombic efficiency of 92.0%. The lower ionic conductivity of 10LPSOI-LS may contribute to its lower initial battery capacity compared with 10LPSOI-BM; however, differences in interfacial contact, particle morphology, and other synthesis-dependent characteristics may also influence the full-cell response. The charge–discharge capacities of the 10LPSOI-BM ASSLMB remain relatively stable up to the third cycle. In the fourth cycle, an abnormal increase in charge capacity to 226.3 mAh g^–1 is accompanied by a decrease in discharge capacity to 98.2 mAh g^–1, resulting in a Coulombic efficiency of 43.4%. This anomalous behavior indicates transient electrochemical instability in the 10LPSOI-BM full cell. Possible origins include changes in interfacial contact, local electrolyte degradation, or the development of a partial internal electronic pathway. By the sixth cycle, the apparent charge and discharge capacities return to 135.1 and 126.4 mAh g^–1, respectively, with a Coulombic efficiency of 93.5%. However, the present measurements do not allow the origin of this temporary behavior to be identified conclusively, and the possibility of microshort behavior cannot be excluded. The larger particle size of 10LPSOI-BM may contribute to less uniform interfacial contact; however, the present data do not isolate particle morphology as the origin of the transient instability. Previous studies have shown that grain boundaries and interfacial voids can influence nonuniform lithium deposition and short-circuit behavior. , As additional information, the charge–discharge performance of an ASSLMB with the configuration Li | Li_5.5PS_4.5Cl_1.5 | 70NMC₁₁₁ + 30­(Li_5.5PS_4.5Cl_1.5) + 3AB was evaluated, as shown in Figure S13a, with the corresponding cell schematic illustrated in Figure S13b. The initial charge and discharge capacities of the Li_5.5PS_4.5Cl_1.5 ASSLMB are 169.4 mAh g^–1 and 134.1 mAh g^–1, respectively. In the second cycle, the charge and discharge capacities reach 135.5 mAh g^–1 and 135.2 mAh g^–1, respectively. The capacities remain stable up to the fifth cycle, after which performance deteriorates, with significant capacity fading observed from the sixth cycle onward. By the 12th cycle, the discharge capacity declines sharply to approximately 12 mAh g^–1. The Li_5.5PS_4.5Cl_1.5 full-cell result is included as a reference comparison to provide additional context for the lithium–metal full-cell performance of 10LPSOI-BM using a sulfide solid electrolyte prepared by ball milling. Although the Li_5.5PS_4.5Cl_1.5 cell showed more rapid capacity degradation, the present comparison does not isolate the origin of this difference to a single structural feature. Instead, the improved cycling behavior of 10LPSOI-BM may reflect combined differences in composition, phase characteristics, local structure, and lithium–metal interfacial chemistry.

9.

Charge–discharge capacity curves of Li | 10LPSOI-M (M = BM or LS) | 70NMC₁₁₁ + 30Li_5.5PS_4.5Cl_1.5 + 3AB cells using (a) 10LPSOI-BM and (b) 10LPSOI-LS as the separator layer and ionic conductor between anode and cathode, measured at 30 °C and 0.1C within a voltage window of 2.6–4.2 V vs Li/Li⁺. (c) Cycling performance of 10LPSOI-LS and 10LPSOI-BM over 50 cycles at 0.1C and 30 °C.

Returning to the comparison between 10LPSOI-BM and 10LPSOI-LS, both solid electrolytes maintain ASSLMB performance up to the 50th cycle, with 10LPSOI-BM exhibiting a higher discharge capacity than 10LPSOI-LS. However, closer examination of the cycling behavior of 10LPSOI-BM reveals noticeable fluctuations in discharge capacity over repeated cycles. This irregular trend leads to variations in the Coulombic efficiency of the 10LPSOI-BM cell. Such fluctuations suggest that a fraction of Li⁺ may become trapped during the plating (charge) process and cannot be fully recovered during stripping (discharge). This irreversible lithium loss may result in the formation of dead lithium and dendritic growth. The underlying causes may include degradation of the solid electrolyte interphase, microcrack formation within the electrolyte layer, or inhomogeneous Li⁺ transport pathways and distribution. , On the other hand, the Coulombic efficiency of 10LPSOI-LS remains relatively stable throughout cycling, without noticeable capacity deterioration, reaching 94.7% at the 50th cycle. This value is higher than that of 10LPSOI-BM, which decreases to 90.8% at the same cycle. The capacity retention, calculated relative to the second-cycle discharge capacity, was 98.0% for 10LPSOI-BM and 99.3% for 10LPSOI-LS at the 50th cycle. The Coulombic efficiencies of the full cells using 10LPSOI-BM and 10LPSOI-LS are 90.8% and 94.7%, respectively, which are lower than those observed in the half-cell configurations (97.6% for 10LPSOI-BM and 97.7% for 10LPSOI-LS). The lower Coulombic efficiencies of the full cells may reflect additional irreversible processes associated with direct contact between lithium metal and the solid electrolyte. The postcycling XPS results from the lithium symmetric cells show Li₂S- and Li _x P-related interfacial products, indicating that solid-electrolyte decomposition can occur at the lithium interface; however, these XPS data do not directly quantify dead lithium or establish the origin of the charge–discharge capacity imbalance in the full cells. Nevertheless, 10LPSOI-LS shows a higher Coulombic efficiency compared to 10LPSOI-BM, indicating improved interfacial stability against lithium metal. The slightly higher capacity retention of 10LPSOI-LS indicates comparatively stable cycling over the measured 50 cycles. However, the present electrochemical data do not allow this behavior to be attributed specifically to continued electrode activation or to an unchanged solid electrolyte interphase. This behavior is consistent with the XPS results of 10LPSOI-LS, which indicate a more robust and chemically stable interfacial structure. These findings indicate that 10LPSOI-LS provides improved cycling stability and compatibility with lithium metal and NMC₁₁₁-based cathode composites under the tested conditions. The improved cycling stability of 10LPSOI-LS is consistent with the synthesis-dependent differences observed in particle morphology, local phosphorus environments, and surface chemistry. The finer particle morphology may contribute to improved interfacial contact and more uniform ion-transport pathways. In addition, XPS and ³¹P MAS NMR reveal distinct surface and local structural environments in 10LPSOI-LS compared with 10LPSOI-BM. However, the present measurements do not establish a direct causal relationship between a specific POS₃ ^3– configuration, a hydrocarbon-containing surface layer, and the improved full-cell performance. These observations support the potential of 10LPSOI-LS for further evaluation in all-solid-state lithium–metal battery configurations.

Conclusion

In summary, this study compares 10LPSOI solid electrolytes prepared by mechanical ball milling and a modified liquid-phase shaking method to clarify how the synthesis route influences structure, morphology, ionic transport, and lithium–metal interfacial behavior. Although 10LPSOI-LS exhibited a lower room-temperature ionic conductivity of 0.7 mS cm^–1 than 10LPSOI-BM at 3.4 mS cm^–1, it showed a substantially higher critical current density of 0.925 mA cm^–2 and improved long-term lithium plating/stripping stability under the tested conditions. In full-cell configurations, 10LPSOI-LS also exhibited more stable cycling behavior and a slightly higher capacity retention than 10LPSOI-BM over 50 cycles. These differences are correlated with synthesis-dependent variations in particle morphology, phase characteristics, local phosphorus environments, and surface chemistry. The finer particle morphology of 10LPSOI-LS may contribute to improved interfacial contact and more uniform lithium-ion transport, while the spectroscopic results indicate distinct local structural and surface environments compared with 10LPSOI-BM. However, the present results do not establish a direct causal relationship between a specific POS₃ ^3– configuration, hydrocarbon-containing surface species, and the improved electrochemical performance. Overall, the modified liquid-phase shaking method provides a promising route for tuning the interfacial properties of iodide–oxide–sulfide solid electrolytes and may offer potential advantages for scalable preparation of sulfide-based solid electrolytes.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.6c05347.

• XRD data, particle-size distributions, DC-polarization lithium–metal stability measurements, EIS and XPS data, charge–discharge results, and cell schematic illustrations (PDF)

R.F.I. conceptualized the study, conducted the main experimental work, analyzed the data, created the graphs, and handled data curation, as well as wrote the paper. M.T. conducted the experimental work. I.Y.M. conducted the experimental work. M.S. conducted the experimental work and analyzed the data. K.H. contributed to the study’s conception, data analysis, review, editing, and supervision. A.M. secured funding, analyzed data, reviewed and edited the manuscript, and supervised the study.

The authors declare no competing financial interest.