Facile mechanical refinement of sulfide solid electrolytes for stable all-solid-state Li metal batteries

Abstract

All-solid-state lithium metal batteries offer high energy density and enhanced safety, yet mechanical instability at both positive electrode/solid electrolyte and Li metal/solid electrolyte interfaces severely limits cycling stability and rate performance, particularly at a low stacking pressure. Here, we report a facile, industry-compatible strategy to refine polycrystalline Li_5.5PS_4.5Cl_1.5 grains using Al₄C₃ abrasives. The resulting Al₄C₃-engineered Li_5.5PS_4.5Cl_1.5 exhibits reduced average particle size and a narrowed size distribution. The refined electrolyte grains enable a high relative density in both anolytes and composite positive electrodes. Beyond particle refinement, Al₄C₃ in the anolyte stabilizes the Li/electrolyte interface through its high modulus, low electronic conductivity, and favorable interfacial mechanics, enabling a high critical current density/capacity of 3.2 mA cm^‒2/3.2 mAh cm⁻². Leveraging our designed electrolyte as both anolyte and catholyte, the all-solid-state lithium metal batteries demonstrate stable cycling performance under a low stack pressure of 4 MPa and a positive electrode loading of 1.4 mAh cm⁻², sustaining over 1500 cycles at 1.4 mA cm^‒2. This study provides a potentially simple and scalable approach to optimize solid electrolyte particles and interfacial mechanics.

All-solid-state lithium metal batteries are promising for safe, high-energy storage, but interfacial instability limits performance. Here, authors use Al₄C₃-engineered Li₅.₅PS₄.₅Cl₁.₅ electrolytes to improve interfacial mechanics, enabling stable low-pressure operation and over 1500 charge-discharge cycles.

Introduction

The pursuit of high-energy-density energy storage devices has sparked intense research interest in lithium metal batteries (LMBs) coupled with layered oxide or conversion-type positive electrodes, which are capable of potentially doubling the energy density of state-of-the-art lithium-ion batteries (LIBs)^1,2. However, the integration of electrodes with such high specific energy in LMBs raises safety concerns stemming from the inherent flammability of organic liquid electrolytes^3–5. Fortunately, breakthroughs in solid superionic conductors offer a transformative solution by enabling the construction of all-solid-state Li metal batteries (ASSLMBs) using non-flammable inorganic solid electrolytes (ISEs)^6–8. This paradigm shift heralds a new era in next-generation energy storage systems.

Nevertheless, the transition from liquid-state to solid-state architectures introduces unprecedented challenges in the mechanical aspect that critically limit both cycling stability and rate performance. On the positive electrode side, the absence of fluidity in ISEs creates persistent difficulties in establishing and maintaining intimate solid-solid contact with active materials, leading to high interfacial resistance and compromised charge transfer kinetics^9–11. On the negative electrode side, the deposition of lithium metal induces large hydrostatic stress, which positively correlates with the Li plating overpotential¹². When this stress exceeds the mechanical tolerance of ISEs, the deposited lithium metal can penetrate the brittle ISE layer through Griffith-like crack extension, ultimately leading to battery failure^13,14. Furthermore, during lithium stripping, each Li ion that traverses the Li/ISE interface leaves behind an electron and a vacancy in the negative electrode. When the rate of vacancy injection surpasses its bulk diffusion rate, the accumulated vacancies form interfacial voids at the Li/ISE boundary^15–17. These voids reduce the effective Li/ISE interfacial contact area and create localized current hotspots, which, in turn, promote lithium intrusion during subsequent plating cycles^18–21. To address these issues, it is essential to engineer favorable interfacial mechanics of ISEs, including improved electrode/ISE contact and enhanced resistance to lithium intrusion, particularly under practical low stack pressures, where interfacial contacts are more prone to deterioration^22,23. However, this aspect remains largely underexplored, as most previous research on ISE engineering has primarily focused on regulating their (electro)chemical behavior^24–26.

Recently, growing recognition of the mechanical characteristics of ISEs has led to the development of design methodologies aimed at tuning particle size, phase composition, and microstructure. For example, Liu et al. employed ball-milling followed by sintering to homogenize the microstructure of Li_5.3PS_4.3ClBr_0.7 (LPSCB), achieving electrolyte pellets with high relative density²⁷. Others have enhanced the relative density and mechanical properties of Li_6 − xPS_5 − xCl_1 + x (x = 0–0.7, LPSC) via thiol-based treatments^28–31. Despite these advances, existing techniques for engineering the mechanical properties of ISEs are often energy-intensive, laborious, or rely on environmentally unfriendly chemical reagents³². Thus, there is an urgent need for a facile, industry-compatible approach to engineer ISEs with optimized interfacial mechanics, enabling stable operation of ASSLMBs under practical stack pressures.

In this work, we propose a scalable mechanical strategy to engineer polycrystalline Li_5.5PS_4.5Cl_1.5 (LPSC) particles by introducing Al₄C₃ abrasives through a simple mechanical mixing process. Al₄C₃ is selected as the abrasive owing to its high Vickers hardness, (electro)chemical stability toward electrolytes/electrodes, and weak interfacial affinity with sulfide electrolytes. These attributes enable Al₄C₃ to efficiently fragment and disperse the relatively soft LPSC particles during mixing, reducing the particle size distribution from 1.17 to 20.38 μm in pristine LPSC to 1.49–4.53 μm in Al₄C₃-processed LPSC (AlC-LPSC), with a decreased average particle size of 1.99 μm. Beyond particle refinement, AlC-LPSC exhibits the following advantages for suppressing lithium intrusion when employed as the anolyte. (1) The reduced grain size promotes high pellet relative density, while the incorporation of high-modulus Al₄C₃ enhances pellet mechanical robustness; (2) The electrically insulating nature of Al₄C₃ contributes to reduced electronic conductivity; (3) Al₄C₃ is stable to Li metal, and the Li/Al₄C₃ interface possesses a high product of interfacial energy (γ) and bulk modulus (E), along with a high work of adhesion (W_ad), which favorably regulate lithium plating and stripping behaviors. As a result of these combined benefits, symmetric Li cells employing AlC-LPSC as the anolyte exhibit a high critical current density (CCD) of 3.2 mA cm⁻². When employed as a catholyte paired with the layered oxide positive electrode material LiNi_0.83Co_0.12Mn_0.05O₂ (NCM83), the reduced grain size of AlC-LPSC facilitates more intimate contact between the active material and ISE, while simultaneously minimizing overall deformation within the composite positive electrode. These effects collectively enhance charge carrier transport kinetics. The favorable interfacial mechanics of AlC-LPSC contribute to marked improvements in both the cycling stability and rate capacity of ASSLMBs. Notably, the Li||NCM83 cell utilizing AlC-LPSC as both anolyte and catholyte achieves stable cycling over 1500 cycles at a current density of 1.4 mA cm⁻² under a low stack pressure of 4 MPa. This work offers valuable guidance for the rational design of ISE interfacial mechanical properties.

Results and discussion

Scalable grain refinement engineering via Al₄C₃ abrasive

Currently, LPSC electrolytes are widely employed as interlayers in sulfide-based ASSLMBs, owing to their high ionic conductivity and kinetic stability against lithium metal^33–36. Despite these advantages, polycrystalline LPSC particles synthesized via conventional routes typically exhibit a relatively large average particle size^37,38. As a consequence, the stacking of large LPSC grains leads to a low relative density of the ISE interlayer after plastic deformation, increasing the risk of mechanical failure during lithium deposition³⁹. Moreover, when mixed with positive electrode active materials, the large particle size of LPSC results in an inhomogeneous distribution of ion-conductive phases within the composite positive electrode, thereby impairing the construction of an efficient ion transport network⁴⁰. To address the aforementioned challenges, a grain refinement approach for LPSC using Al₄C₃ abrasives is proposed (Fig. 1a). This refinement process can be readily achieved via low-speed mechanical mixing, offering scalability for practical implementation. As evidenced by scanning electron microscopy (SEM) and the corresponding energy-dispersive spectroscopy (EDS) mapping (Fig. 1b, c), pristine LPSC synthesized via the conventional route exhibits an average particle size of 4.19 ± 3.00 μm (mean diameter ± SD, n = 319), with a fraction of particles exceeding 10 μm (Fig. 1d and Table S1). In contrast, LPSC particles refined by Al₄C₃ abrasives (AlC-LPSC) predominantly fall within the 1.58–2.28 μm range, with an average particle size of 1.99 ± 0.57 μm (mean diameter ± SD, n = 301). Laser diffraction measurements further confirm the particle size reduction in AlC-LPSC, exhibiting a smaller D₅₀ (2.897 μm) compared to that of LPSC (21.69) (Fig. S1).

Fig. 1. Morphology and component characterization of AlC-LPSC.

a Diagram depicting the mechanical processing of Al₄C₃ within the LPSC matrix and the consequent improvement in interfacial mechanics when the AlC-LPSC is utilized as both catholyte and anolyte. SEM images and EDS mappings of LPSC (b) and AlC-LPSC (c) particles. d Particle diameter range of LPSC and AlC-LPSC particles and the corresponding mean diameter (μm) ± SD. XRD patterns (e) and Raman spectra (f) of LPSC and AlC-LPSC electrolytes.

To elucidate the distinct fracture and dispersion effects of Al₄C₃, we systematically processed LPSC with or without Al₄C₃ abrasive under different mechanical treatments (Tables S2 and S3). Conventional ball-milling LPSC with merely zirconia media is inefficient and often leads to re-agglomeration of fine particles (Fig. S2a–d) due to the stress-induced re-adhesion behavior of fractured sulfide particles. In contrast, the introduction of Al₄C₃ transforms this process, yielding finer and more homogeneously dispersed electrolyte powders with reduced energy input (Fig. S2e–h). The high fracture efficiency and anti-agglomeration effects of Al₄C₃ can be attributed to the following factors. First, its high hardness (Vickers hardness of ~25 GPa) and small size (~3 μm) provide a frequent and localized stress transfer during milling. Substances with lower Vickers hardness (LiF: ~1 GPa; Li₃N: ~4 GPa) failed to replicate the refinement effect under identical mild milling conditions (Fig. S3). Second, the weak interfacial adhesion between Al₄C₃ and LPSC, evidenced by the distinct interface separation in a double-layered cold-pressed pellet (Fig. S4), prevents the newly created LPSC surfaces from re-agglomerating. Consequently, the AlC-LPSC composite achieves optimal refinement and dispersion with minimal processing time and energy, enabling the subsequent fabrication of a dense electrolyte/composite positive electrode.

The LPSC grain-refinement process using Al₄C₃ abrasives is governed solely by physical interactions. As shown in Fig. 1e, the X-ray diffraction (XRD) pattern of the AlC-LPSC composite retains the characteristic diffraction peaks of both pristine LPSC and Al₄C₃, with no evidence of new phase formation. This chemical compatibility is further supported by Raman spectroscopy, where the AlC-LPSC sample exhibits distinct vibrational modes at 495 and 985 cm⁻¹ corresponding to Al₄C₃, along with the characteristic vibrational bands of PS₄³⁻ units in LPSC at 193, 254, 427, and 577 cm⁻¹ (Fig. 1f). Moreover, the identical characteristic peaks of LPSC phases before and after the ball milling process in the XRD patterns indicate the well-retained electrolyte structure during the mechanical mixing process (Fig. S5). The purely physical nature of the particle refinement ensures that the intrinsic ionic conductivity of LPSC is largely preserved. Electrochemical impedance spectroscopy (EIS) measurements reveal that the ionic conductivity of the composite electrolyte only shows a slight, expected decrease due to the inert nature of Al₄C₃, but the electronic conductivity of the composite electrolyte is simultaneously suppressed, which is beneficial for resisting Li depositing inside the electrolyte to cause a short circuit (Fig. S6).

AlC-LPSC anolyte for Li intrusion suppression

The refinement of electrolyte particles and uniformly distributed Al₄C₃ abrasives in AlC-LPSC enable the fabrication of highly dense and mechanically reinforced electrolyte pellets. Both top-view and cross-sectional SEM images reveal the densification effects. The pristine LPSC pellet exhibits considerable pores and heterogeneous electrolyte clusters of varying sizes (Figs. 2a and S7a, b), whereas the AlC-LPSC pellet demonstrates enhanced densification, characterized by a smooth and densely packed microstructure (Figs. 2b and S7c, d). Further microstructural characterization was conducted using SEM coupled with time-of-flight secondary ion mass spectrometry (SEM-TOF-SIMS), which mapped the distribution of LPSC and Al₄C₃ by detecting Li⁺ and Al⁺ secondary ion fragments within the selected area. Correlative analysis of the Al⁺ signal and the corresponding SEM microstructural features confirms that the observed dark regions in the electrolyte pellet originate from Al₄C₃ filling rather than structural defects, such as pores, thereby validating the densification of AlC-LPSC (Fig. 2c). Atomic force microscopy (AFM) topography analysis further confirms the densification effect. As depicted in Fig. 2d, there are height variations with prominent local protrusions for the LPSC pellet surface. In contrast, consistent with the smooth surface morphology observed via SEM, a uniform distribution of height profiles is observed for the AlC-LPSC pellet surface (Fig. 2e). The selected areas for height acquisition were further used for the modulus measurement. The pristine LPSC electrolyte delivers an uneven modulus distribution, with the mean Young’s modulus (E) determined to be 11 GPa, whereas the AlC-LPSC pellet displays a higher and uniform modulus, averaging 16 GPa (Fig. 2f, g). To decouple the effect of particle size reduction from the intrinsic reinforcement provided by Al₄C₃, we prepared the LPSC electrolyte pellet using small-size LPSC (denoted ss-LPSC, whose average particle size is similar to that of AlC-LPSC). The ss-LPSC pellet shows an average modulus of 21 GPa (Fig. S8). This result indicates that particle refinement can contribute to modulus enhancement, primarily through improved pellet densification. However, increasing the Al₄C₃ content to a ratio of 1:3 in AlC-LPSC can markedly increase the electrolyte average modulus to 31 GPa (Fig. S9), surpassing the theoretical modulus of LPSC (~22 GPa). These results indicate that both the particle refinement effect and the high modulus of Al₄C₃ contribute to the high average modulus of AlC-LPSC.

Fig. 2. Benefits associated with the use of AlC-LPSC anolyte.

SEM images of LPSC (a) and AlC-LPSC (b) pellets. c SEM-TOF-SIMS images of the AlC-LPSC pellet. Height distribution signals of LPSC (d) and AlC-LPSC (e) pellet. Young’s modulus of LPSC (f) and AlC-LPSC (g) pellet. h XRD pattern of molten Li/Al₄C₃. Ionic (i) and electronic conductivity (j) of blocking cells using LPSC electrolyte and AlC-LPSC/LPSC/AlC-LPSC sandwiched structure.

To evaluate the mechanical resistance of AlC-LPSC against lithium penetration, we performed an in situ pressure-driven lithium intrusion test in Li||Cu cells (Fig. S10). The cell voltage was monitored under stepwise increasing stack pressure until a short circuit (voltage drop to 0 V) occurred, indicating complete Li penetration through the electrolyte. As shown in Fig. S10a, the cell with pristine LPSC short-circuited immediately once the pressure reached 25 MPa. In contrast, the cell using AlC-LPSC/LPSC/AlC-LPSC maintained a normal voltage under 60 MPa and sustained functionality at this high pressure for over 2.5 h before eventual failure (Fig. S10b). This result demonstrates that the high mechanical robustness imparted by AlC-LPSC can suppress lithium metal penetration.

Compared with the pristine LPSC pellet, the densified and mechanically reinforced AlC-LPSC electrolyte exhibits a higher critical stress for suppressing Li intrusion. Beyond this role, the Al₄C₃ abrasives also contribute to suppressing Li intrusion by serving as a stable component of the solid electrolyte interphase (SEI), effectively regulating Li plating/stripping behavior. XRD analysis of molten Li/Al₄C₃ mixtures reveals only the characteristic peaks of Li and Al₄C₃, confirming the chemical inertness of Al₄C₃ toward Li metal (Fig. 2h). This inertness is further corroborated by post-cycling characterization of the Li/AlC-LPSC interface. As shown in Fig. S11, X-ray photoelectron spectroscopy (XPS) reveals that the P 2p and S 2p spectra of the Li/AlC-LPSC interface only demonstrate the signals from PS₄³⁻ and the decomposition phase of LPSC, while the Al 2p spectrum exclusively displays signals corresponding to Al₄C₃, with no evidence of reduction products, verifying the chemical stability of Al₄C₃ during electrochemical cycling. Complementarily, XRD patterns of the cycled AlC-LPSC electrolyte retain the characteristic peaks of Al₄C₃ and LPSC phases, with no emergence of new crystalline phases (Fig. S12). These characterization results indicate that, upon partial reduction of LPSC during cycling, the dispersed Al₄C₃ abrasives in AlC-LPSC are incorporated into the SEI. Under the idealized conditions assumed in our density functional theory (DFT) calculations, compared to the reduction products of LPSC (Li₂S, Li₃P, and LiCl), Al₄C₃ displays a higher product of interface energy (γ) with Li and a bulk modulus (E) value of 105.44 J GPa m⁻² (Figs. S13, S14, Table S4, and Supplementary Data 1). This suggests a potential advantage in resisting Li intrusion during Li plating, although these simulations represent simplified interfaces and may not fully capture the behavior under experimental conditions. Moreover, Al₄C₃ possesses a higher work of adhesion (W_ad) towards Li, which may indicate its favorable interfacial contact with Li metal during Li stripping and thereby lower the local current density in the subsequent deposition process^41,42.

According to the fracture mechanics hypothesis, reducing the overpotential of lithium deposition can alleviate the stress generated during Li plating, thereby suppressing lithium intrusion^43,44. From this perspective, the relatively low ionic conductivity of AlC-LPSC is considered a drawback. To overcome this limitation, a rational design strategy is employed in this work: AlC-LPSC, with its high densification and capability to effectively regulate Li plating/stripping behavior, is used as the anolyte in direct contact with the Li metal, while pristine LPSC serves as the bulk electrolyte interlayer (Fig. S15). Subsequent tests were conducted on symmetric cells constructed with a sandwich configuration of AlC-LPSC/LPSC/AlC-LPSC. The measured ionic (σ) and electronic conductivities (σ_e) of this sandwich structure are presented in Fig. 2i, j. As expected, with such an optimized electrolyte structure, only a slight compromise on the ionic conductivity of AlC-LPSC/LPSC/AlC-LPSC is displayed. Concurrently, benefiting from the intrinsically low electronic conductivity of Al₄C₃ (3.2 × 10⁻¹⁰ S cm⁻¹, Fig. S16), a reduced electronic conductivity is observed in the AlC-LPSC/LPSC/AlC-LPSC compared to that of pristine LPSC (5.23 × 10⁻⁹ S cm⁻¹, Fig. S17). Based on the electronic conductivity hypothesis of Li intrusion, the low electronic conductivity of the sandwiched electrolyte structure theoretically migrates the direct Li deposition inside the ISE and reduces the parasitic reactions at the Li/ISE interface.

Then, Li symmetric cells using LPSC and AlC-LPSC/LPSC/AlC-LPSC sandwiched structure were assembled and tested. The optimal ratio of Al₄C₃/LPSC in the AlC-LPSC interlayer was investigated by acquiring critical current density (CCD) values under step-increased current densities tests. As shown in Fig. 3a, the symmetric cell without AlC-LPSC anolyte exhibits an inferior CCD of 0.8 mA cm⁻². After inserting the AlC-LPSC anolyte with varying Al₄C₃/LPSC ratios (1:3, 1:5, and 1:7), improvements in CCD performance are exhibited, with the highest value of 3.2 mA cm⁻² achieved at the ratio of 1:5 (Figs. S18 and 3b). This performance shows high competitiveness when benchmarked against the reported studies (Table S5). Such enhanced CCD and lithium-penetration suppression achieved by AlC-LPSC cannot be attributed to high relative density alone, but is also strongly governed by interfacial thermomechanical descriptors, most notably a high interfacial energy/Young’s modulus product and a high work of adhesion. By decoupling the effects of densification, particle refinement, and interfacial configuration, we demonstrate that anolytes with comparable or even higher relative density (ss‑LPSC) fail to deliver equivalent performance (Fig. S19 and Table S6). Furthermore, replacing Al₄C₃ in the anolyte with materials possessing only a single favorable interfacial property—such as LiF (high γE) or Li₃N (high W_ad)—yields CCD values comparable to those of ss‑LPSC, despite their lower densification. However, neither alternative achieves the high CCD observed in cells using AlC‑LPSC anolytes, indicating that a single interfacial attribute is insufficient to replicate its performance (Table S7 and Fig. S20). The practical advantage of the AlC-LPSC anolyte becomes even more pronounced under demanding conditions: when electrolyte thickness is reduced, cells using AlC-LPSC anolytes achieve higher CCDs than their counterparts using LPSC (Fig. S21). These results underscore the critical contribution of Al₄C₃’s interfacial thermomechanical properties to the effective suppression of lithium intrusion.

Fig. 3. Suppressing Li intrusion by AlC-LPSC/LPSC/AlC-LPSC sandwiched structure.

Voltage profiles of Li|LPSC|Li (a) and Li|AlC-LPSC/LPSC/AlC-LPSC|Li (b) cells during galvanostatic discharge/charge tests at step-increased current densities under a stack pressure of 4 MPa. EIS evolution of Li|LPSC|Li (c) and Li|AlC-LPSC/LPSC/AlC-LPSC|Li (d) cells during galvanostatic discharge/charge tests at 0.5 mA cm⁻² under 30 °C. Galvanostatic cycling performance of Li symmetric cells using LPSC and AlC-LPSC/LPSC/AlC-LPSC sandwiched structure at 0.5 mA cm⁻²/0.5 mA h cm⁻² (e) and 1 mA cm^‒2/1 mA h cm⁻² (f) under a stack pressure of 4 MPa. Cross-sectional SEM images of Li/LPSC (g) and Li/AlC-LPSC (h) interface after Li plating/stripping. SEM images of plated Li on SS in Li|LPSC|SS (i) and Li|AlC-LPSC/LPSC/AlC-LPSC|SS (j) half-cells.

Adopting the AlC-LPSC with optimized ratio as the anolyte, the chemical stability between AlC-LPSC/LPSC/AlC-LPSC and Li was first evaluated by resting EIS and in situ EIS during Li plating/stripping, respectively. As displayed in Fig. S22a–c, during 32 h of resting, the Li symmetric cell using LPSC experiences a continuous growth in impedance, whereas the Li symmetric cell using the AlC-LPSC/LPSC/AlC-LPSC sandwiched structure maintains stable interfacial resistance (Fig. S22d–f). The interfacial stability was further corroborated by in situ EIS during Li plating/stripping at 0.5 mA cm⁻², where the Li symmetric cell using LPSC shows progressive resistance escalation, contrasting with the stable interfacial resistance of that using AlC-LPSC/LPSC/AlC-LPSC (Fig. 3c–e). The S 2p spectra also verify the suppressed interfacial side reactions achieved by the AlC-LPSC anolyte, as evidenced by the reduced content of Li₂S at the Li/AlC-LPSC interface than that at the Li/LPSC interface (Fig. S23). The Li symmetric cells were further subjected to the rate test with a fixed capacity of 0.1 mA h cm⁻². As shown in Fig. S24, the Li symmetric cell using AlC-LPSC/LPSC/AlC-LPSC displays smaller overpotentials at progressively elevated current densities of 0.2, 0.5, and 1 mA cm⁻², indicative of the enhanced chemical stability of the Li/AlC-LPSC interface.

Furthermore, long-term Li plating/stripping tests under different current densities and cut-off capacities were conducted to evaluate the Li stability of the AlC-LPSC. The symmetric cells using AlC-LPSC/LPSC/AlC-LPSC sandwiched structure stably operate over 1700 and 2300 h at 0.2 mA cm⁻² with different capacities of 0.2 and 0.5 mA h cm⁻², respectively (Figs. S25 and S26). Under a moderate current density of 0.5 mA cm⁻² and capacity of 0.5 mA h cm⁻², the Li symmetric cell using AlC-LPSC/LPSC/AlC-LPSC sandwiched structure also demonstrates high stability during the 4500 h Li plating/stripping process (Fig. 3e). In stark contrast, the overpotential of the Li symmetric cell using LPSC undergoes a rapid increase from 10 to 45 mV after 832 h of Li plating/stripping, followed by a short circuit indicated by the abnormal drop of overpotential. Under a higher current density of 1 mA cm⁻² and a capacity of 1 mA h cm⁻², the performance disparity becomes even more pronounced, with the Li symmetric cell using LPSC short-circuiting after only 6 h, whereas the Li symmetric cell using AlC-LPSC/LPSC/AlC-LPSC demonstrates stable Li plating/stripping behavior for 550 h (Fig. 3f), which is competitive to some recent relevant reports as shown in Table S8. To evaluate the practical viability of the interface, the Li symmetric cells were subjected to conditions that intensify interfacial degradation. Under a much lower stack pressure of 0.9 MPa at 0.2 mA cm⁻²/0.2 mA h cm⁻², the cell using AlC-LPSC anolytes maintained stable operation for 1200 h. In contrast, the cell using LPSC demonstrates a sharp increase in overpotential and failed after 215 h of Li plating/stripping (Fig. S27). Additionally, under a demanding condition of elevated temperature (60 °C) and higher current density (1 mA cm⁻²/1 mA h cm⁻²), the AlC-LPSC-based cell exhibited higher Li plating/stripping stability than its LPSC-based counterpart (Fig. S28).

Post-mortem SEM analyses were further performed to reveal the evolution of the Li/ISE interface and the morphology of Li deposition. As shown in Fig. 3g, considerable holes/pores can be observed both at the interface and the electrolyte bulk for the cell using LPSC. On the contrary, intimate electrode/electrolyte contact at the Li/AlC-LPSC interface as well as higher structural integrity were observed for the cell using AlC-LPSC (Fig. 3h). Furtherly, Li||stainless-steel (SS) half-cells employing LPSC and AlC-LPSC/LPSC/AlC-LPSC sandwiched structure were assembled to capture the deposition morphologies of Li on the SS electrode. As illustrated in Fig. 3i, the deposition of Li across the LPSC electrolyte exhibits a non-uniform distribution, characterized by varying sizes and localized clustering. For the situation that uses AlC-LPSC anolytes, the Li deposition presents a homogeneous morphology (Fig. 3j), indicating the favorable role of the AlC-LPSC anolyte in regulating Li plating. These collective results manifest the enhanced Li/ISE stability achieved by the strengthened interfacial mechanics and regulated Li plating/stripping behavior associated with the use of AlC-LPSC anolytes.

AlC-LPSC catholyte for improving ion transport kinetics

When pristine LPSC is employed as the catholyte and layered oxide NCM83 is used as the positive electrode active material (PEAM), severe phase segregation between LPSC and NCM83 is observed (Fig. S29a, b). The cold-pressed LPSC/NCM83 composite electrode exhibits localized LPSC aggregation and regional electrolyte depletion, indicating the formation of an inefficient ionic percolation network within the positive electrode (Figs. 4a and S30a, b). Upon cycling, the LPSC/NCM83 electrode suffers from severe structural degradation and aggravated interfacial delamination, further impairing ion transport across the electrode (Fig. 4b). In contrast, AlC‑LPSC enables a more homogeneous distribution of catholyte and PEAM particles throughout the composite electrode powders (Fig. S29c, d), without inducing any detectable interfacial reaction between Al₄C₃ and NCM83 (Fig. S31). After cold pressing, the AlC-LPSC/NCM83 composite positive electrode exhibits an improved compaction and a more uniform spatial distribution of electrode components (Figs. 4c and S30c, d). This microstructural optimization is quantitatively substantiated. The AlC-LPSC/NCM83 positive electrode exhibits a lower porosity ($\epsilon$), evidenced by both a higher relative density (82.0% vs. 79.6% for LPSC/NCM83, Table S9) and a reduced surface void fraction (5.01% vs. 10.56% for LPSC/NCM83, Fig. S32). Moreover, tortuosity is reduced from 3.80 for the LPSC-based positive electrode to 1.64 for the AlC-LPSC-based positive electrode (Fig. S33 and Table S10). Benefited from the optimized microstructure, after extended cycling, the AlC-LPSC/NCM83 positive electrode retains its dense architecture and well-preserved catholyte coverage (Fig. 4d), thereby sustaining favorable ion transport kinetics throughout the cycling process.

Fig. 4. Ion transport enhancement in the positive electrode using AlC-LPSC.

SEM images and EDS mappings of LPSC/NCM83 composite positive electrode before (a) and after cycling (b). SEM images and EDS mappings of AlC-LPSC/NCM83 composite positive electrode before (c) and after cycling (d). e Time-voltage profile for EIS. EIS evolution (f) and the corresponding DRT contour map (g) of LiIn||LPSC/NCM83 cell during the charge and discharge process at 0.09 mA cm⁻² under 30 °C. EIS evolution (h) and the corresponding DRT contour map (i) of LiIn||AlC-LPSC/NCM8 cell during the charge and discharge process at 0.09 mA cm⁻² under 30 °C. j Rate performance of LiIn||LPSC/NCM83 and LiIn||AlC-LPSC/NCM83 cells. GITT tests (k) and the D_Li (l) of LiIn||LPSC/NCM8 and LiIn||AlC-LPSC/NCM8 cells. The stack pressure was 100 MPa.

Impedance evolution analysis of LiIn||NCM83 cells employing LPSC and AlC-LPSC as catholytes offers critical insights into their interfacial mechanics and ion transport kinetics (Fig. 4e–i). EIS was performed at various charge and discharge states (Fig. 4e). Although both cells show the general trend of increasing interfacial impedance during charging and decreasing impedance during discharging, the LiIn||NCM83 cell utilizing AlC-LPSC catholyte exhibits a lower initial resistance and more moderate impedance variation throughout cycling, indicative of enhanced charge transfer kinetics (Fig. 4f, h). Distribution of relaxation time (DRT) curves and the corresponding contour maps of the EIS results further delineate the contributions of individual electrochemical processes to the overall impedance. Four dominant relaxation peaks are consistently observed across all cells, among which, τ₁ and τ₂ located at time constants of ~10⁻⁶ s and 10⁻⁵ s, are related to the bulk and grain boundary impedance, respectively, while the relaxation peaks τ₃ in 10⁻⁴–10⁻² s are the interfacial impedance at positive electrode/ISE and negative electrode/ISE, and τ₄ at 10⁻² s–10⁻¹ s refers to the charge transfer⁴⁵. Notably, the cell with pristine LPSC exhibits pronounced changes in the 10⁻⁴–10⁻² s domain, reflecting sluggish Li⁺ transport kinetics at the LPSC/NCM83 interface due to deteriorated interfacial contact (Figs. S34a and 4g). After cycling, there is an intensity growth of the relaxation peak in this timescale, confirming irreversible degradation of the LPSC/NCM83 interface. In contrast, the AlC-LPSC/NCM83 positive electrode demonstrates only a marginal impedance shift within the same relaxation window, with no significant peak intensification after a full charge/discharge cycle (Figs. S34b and 4i). This behavior highlights the improved interfacial ion transport and redox reversibility enabled by the AlC-LPSC catholyte.

Subsequently, the electrochemical performances of LiIn||NCM83 cells using either pristine LPSC or AlC-LPSC as catholytes were evaluated. As shown in Figs. 4j and S35, the rate capability tests reveal that the LiIn||NCM83 cell with AlC-LPSC delivers higher discharge capacities of 174.9, 163.3, 155.9, 146.5, 136.5, and 131.4 mA h g⁻¹ at 0.1, 0.2, 0.3, 0.5, 0.8, and 1 C (1 C = 1.4 mA cm⁻²), respectively, than the cell using pristine LPSC catholyte. This improvement highlights that the mechanically optimized interface achieved by AlC-LPSC enhances ion transport kinetics within the positive electrode. Further insights into lithium-ion diffusion were obtained by galvanostatic intermittent titration technique (GITT) measurements of LPSC/NCM83 and AlC-LPSC/NCM83 composite positive electrodes after 2 cycles of activation at 0.11 mA cm⁻² at 30 °C. The voltage profiles indicate reduced polarization for AlC-LPSC/NCM83 (Fig. 4k). Additionally, the AlC-LPSC/NCM83 positive electrode exhibits a lower average overpotential during both charge (0.007 V) and discharge (0.009 V) processes compared to LPSC/NCM83 (0.012 V for charge and 0.014 V for discharge), as shown in Fig. S36. Moreover, the average Li-ion diffusion coefficients (D_Li), calculated from GITT data, are higher for the AlC-LPSC/NCM83 positive electrode during both charging (5.36 × 10⁻⁹ vs. 2.97 × 10⁻⁹ cm² s⁻¹) and discharging (4.48 × 10⁻⁹ vs. 3.30 × 10⁻⁹ cm² s⁻¹) compared to LPSC/NCM83 (Fig. 4l), further confirming the enhancement of ion transport kinetics imparted by AlC-LPSC.

Performance of ASSLMBs using AlC-LPSC anolyte/catholyte

ASSLMBs were assembled using pristine LPSC or AlC-LPSC as anolyte/catholyte, paired with a layered oxide NCM83 positive electrode (areal capacity: 1.4 mA h cm⁻²). In both configurations, the employed central electrolyte interlayer was LPSC. All electrochemical measurements were conducted under a stack pressure of 4 MPa at 30 °C. As shown in Fig. 5a, the cell using only LPSC exhibits severe capacity degradation with increasing current density. In contrast, the Li||NCM83 cell using AlC-LPSC anolyte/catholyte demonstrates improved rate capability, delivering discharge capacities of 175.8, 150.3, 136.3, 119.3, 102.8, and 94.9 mA h g⁻¹ at rates of 0.1 C, 0.2 C, 0.3 C, 0.5 C, 0.8 C, and 1 C (1 C = 1.4 mA cm⁻²), respectively. Throughout the rate tests, the Li||NCM83 cell using AlC-LPSC anolyte/catholyte also shows lower voltage polarization compared to its counterpart using only LPSC (Fig. 5b, c). Subsequently, cycling stability was examined at 0.2 C following the rate test. After 400 cycles, the Li||NCM83 cell using only LPSC retains merely 55% of its initial discharge capacity. In contrast, the Li||NCM83 cell using AlC-LPSC anolyte/catholyte demonstrates a high retention of capacity (82% of its initial discharge capacity), highlighting its improved long-term stability.

Fig. 5. Electrochemical performance of ASSLMBs under a stack pressure of 4 MPa.

a Rate performance of Li||NCM83 cells using LPSC and AlC-LPSC as anolyte/catholyte. Voltage curves of Li||NCM83 cells cycled at different rates using LPSC (b) and AlC-LPSC (c) as anolyte/catholyte. d Long-cycling performance of Li||NCM83 cells using LPSC and AlC-LPSC as anolyte/catholyte at 0.7 mA cm⁻². dQ/dV curves of Li||NCM83 cells using LPSC (e) and AlC-LPSC (f) as anolyte/catholyte at 0.7 mA cm⁻². g Long-cycling performance of Li||NCM83 cells using LPSC and AlC-LPSC as anolyte/catholyte at 1.4 mA cm⁻². h Long-cycling performance of LiSi||NCM83 cells using AlC-LPSC as anolyte/catholyte at 0.7 mA cm⁻² under a stack pressure of 100 MPa.

To further assess the electrochemical reversibility, cyclic voltammetry (CV) was performed. Within three cycles, the Li||NCM83 cell using AlC-LPSC anolyte/catholyte exhibits a higher degree of redox peak overlap than the cell using only LPSC, indicative of improved reversibility (Fig. S37). EIS was subsequently conducted for the Li||NCM83 cells after 50 cycles at 0.2 C. As shown in Fig. S38a, the Li||NCM83 cell using AlC-LPSC anolyte/catholyte demonstrates lower overall impedance. Corresponding DRT analysis further confirms the reduced electrode/electrolyte interfacial resistance and enhanced ionic transport kinetics in the cell configuration using AlC-LPSC anolyte/catholyte (Fig. S38b).

To evaluate long-term cycling behavior under more stringent conditions, both the Li||NCM83 cell using LPSC and AlC-LPSC anolyte/catholyte were activated at 0.14 mA cm⁻² for three cycles and then tested at 0.7 and 1.4 mA cm⁻², respectively. As illustrated in Fig. 5d, the Li||NCM83 cell using AlC-LPSC anolyte/catholyte delivers a high initial discharge capacity of 125 mA h g⁻¹ at 0.7 mA cm⁻² and retains 86% of this capacity after 800 cycles. In contrast, the Li||NCM83 cell using only LPSC suffered rapid capacity decay and short-circuited after just 115 cycles. Differential capacity (dQ/dV) analysis during the initial 100 cycles at 0.7 mA cm⁻² reveals a noticeable increase in voltage polarization for the Li||NCM83 cell using only LPSC (Fig. 5e), whereas the Li||NCM83 cell using AlC-LPSC anolyte/catholyte exhibits highly overlapped dQ/dV curves throughout the entire cycling period, indicative of stable and reversible redox reactions (Figs. 5f and S39). When the current density is increased to 1.4 mA cm⁻², the Li||NCM83 cell using AlC-LPSC anolyte/catholyte retains a high capacity retention of 93% after 1500 cycles (Figs. 5g and S40). By contrast, the Li||NCM83 cell using only LPSC failed after just 15 cycles, as evidenced by the unstable voltage profile during charging (Fig. S41). Furtherly, the cycling performance of the cells with high PEAM loadings were assessed at 0.1 C (0.57 mA cm⁻²). As shown in Fig. S42a, b, the cell using LPSC, with PEAM loading of 26.8 mg cm⁻² (5.35 mA h cm⁻²), delivers an initial discharge capacity of 164.89 mA h g⁻¹. However, it suffers from rapid failure, encountering a short circuit after only 19 cycles. In contrast, the cell using AlC-LPSC anolyte/catholyte with PEAM loading of 28.5 mg cm⁻² (5.70 mA h cm⁻²) exhibits an initial discharge capacity of 170.19 mA h g⁻¹ and can stably function for over 100 cycles with a capacity retention of 92.7%. In addition, the low-temperature performance of the Li||NCM83 cell was evaluated at 0 °C. Benefiting from the satisfactory bulk ionic conductivity (Fig. S43), the robust Li/ISE interface that resists Li intrusion (Fig. S44), and the efficient ion-transport network within the positive electrode (Fig. S45), the cell employing AlC-LPSC anolyte/catholyte exhibits enhanced low-temperature performance. As shown in Fig. S46, the AlC-LPSC-based cell maintains a high discharge capacity of 136.5 mA h g⁻¹ (initial: 141.2 mA h g⁻¹) after 150 cycles at 0.1 C (0.09 mA cm⁻²), whereas the LPSC-based cell short-circuits after only 19 cycles. To evaluate the general applicability of our proposed strategy in solid-state batteries, a long-term cycling test was also conducted for the LiSi||NCM83 cell using AlC-LPSC. As shown in Fig. 5h, the cell delivers a high discharging capacity of 146.5 mA h g⁻¹ at 0.7 mA cm⁻², with a capacity retention of 85.7% after 1000 cycles.

In summary, we propose a facile, industry-compatible strategy to produce grain-refined AlC-LPSC electrolytes via Al₄C₃-assisted ball-milling process. The resulting AlC-LPSC exhibits reduced and homogenized particle size, enabling the fabrication of densified and mechanically robust anolyte pellets as well as uniform positive electrode composites. Beyond particle refinement, the incorporation of Al₄C₃ regulates Li plating/stripping, suppressing lithium intrusion and enhancing reversibility. With AlC-LPSC as the anolyte, the Li symmetric cell exhibits a remarkably high CCD value of 3.2 mA cm⁻² and achieves stable Li plating/stripping for 550 h under a demanding condition of 1 mA cm⁻²/1 mA h cm⁻². As a catholyte paired with NCM83, the uniform AlC-LPSC fine grains accelerate ion transport and mitigate chemo-mechanical degradation, enabling durable long-term cycling. When used as both anolyte and catholyte in Li||NCM83 cells, AlC-LPSC enables stable cycling performance, achieving 93% capacity retention after 1500 cycles at 1 C (1.4 mA cm⁻²) under a low stack pressure of 4 MPa. This work demonstrates a simple yet effective approach to engineer interfacial mechanics in ASSLMBs, providing insights for mitigating mechanical degradation under practical operating conditions.

Methods

Synthesis

Li_5.5PS_4.5Cl_1.5 (LPSC) electrolyte was synthesized via a mechanochemical ball-milling process followed by thermal annealing. Specifically, raw materials of Li₂S (99.98%, Sigma-Aldrich), P₂S₅ (99.99%, Macklin), and LiCl (99%, Aladdin) were weighed and loaded into a 45 mL zirconia ball-milling jar with 90 g of zirconia balls (diameter: 10 mm) and ball-to-powder ratio of 60:1. The mixture was ball-milled at 500 rpm for 20 h (with 5-min rest after each 5-min ball milling) using a PULVERIZER 80 (Nanjing Chi Shun Technology Development Co., Ltd). Subsequently, the obtained electrolytes were annealed at 450 °C for 10 h (with a heating rate of 5 °C/min, followed by furnace cooling), yielding LPSC. The Al₄C₃-Li_5.5PS_4.5Cl_1.5 (denoted as AlC-LPSC) electrolyte was obtained by a ball-milling process of LPSC and Al₄C₃ (99.9%, Macklin) in different ratios at 110 rpm for 30 min. All the synthesized processes were performed in inert atmosphere.

Material characterization

X-ray diffraction (XRD) was conducted on the Rigaku SmartLab instrument with a 2θ range of 10°–80°, a scan rate of 10° per minute, and a step size of 0.02°. Raman spectra were performed on a JDbin-yvon LabRamHR800 spectrometer equipped with a 532 nm laser excitation source. Scanning electron microscope (SEM) images and energy dispersive spectroscopy (EDS) mappings were captured on a Zeiss GeminiSEM560 with a beam voltages of 3 and 15 kV, respectively. X-ray photoelectron spectroscopy (XPS) was performed on a Thermo Scientific K-Alpha with an Al-Kα source (1486.6 eV). XPS samples were prepared inside an Ar-filled glovebox (H₂O < 0.1 ppm, O₂ < 0.1 ppm). To avoid air exposure, the samples were transferred from the glovebox to the XPS analysis chamber using a vacuum transfer module. All spectra were charge-corrected by referencing the adventitious C 1s peak to 284.8 eV. Scanning electron microscope coupled with time-of-flight secondary ion mass spectrometry (SEM-TOF-SIMS) was conducted on ThermoFisher SCIENTIFIC under positive ion mode sputtered by the Ga⁺ ion beam (30 keV). Topographies and Young’s modulus of electrolytes were conducted through atomic force microscope (AFM) measurements on Bruker IconXR under a protective nitrogen atmosphere.

Conductivity tests

Ionic and electronic conductivities of electrolytes were measured on a Princeton PARSTAT MC-1000 potentiostat and a CHI 660E (under low-temperature measurement) by assembling blocking cells equipped with stainless-steel (SS) disks as the electrodes. For SS|LPSC|SS blocking cells, the LPSC interlayer was obtained through pressing 150 mg of LPSC powder in a poly(ether ether-ketone) (PEEK) mold under 360 MPa. For the SS|AlC-LPSC/LPSC/AlC-LPSC|SS configuration, 140 mg LPSC was first pressed into the pellet under 100 MPa, and 5 mg of AlC-LPSC powder was evenly spread on both sides of the LPSC pellet and pressed under 360 MPa for 2 min. Finally, two pieces of SS foils (10 mm in diameter) were attached on both sides of the electrolyte pellets. Ionic conductivity measurements were conducted by potentiostatic electrochemical impedance spectroscopy (PEIS) with a frequency range of 1 MHz–10 Hz and an amplitude of 20 mV. Electronic conductivity was measured by the direct current (DC) polarization method with applied voltage ranging from 0.2 to 0.5 V (with a voltage interval of 0.1 V and an elapsed time of 3600 s), and the results of electronic conductivities are calculated by Ohm’s law.

Density functional theory calculation

The interfacial energy between phases A and B was computed using the theoretical framework of Lepley et al.⁴⁶. First, the interfacial formation energy (F) is determined via Eq. 1, where E(AB) is the total energy of the A/B interface, n_A and n_B denote the numbers of each phase, and E(A) and E(B) represent their respective bulk energies. The strain-corrected interface energy (γ) is then derived from Eq. 2, where σ is the strain energy. The work of adhesion (W_ad) is calculated based on Eq. 3, with S_A and S_B being the surface energies of the individual phases.

$$F=E\left(AB\right)-n_{A}E\left(A\right)-n_{B}E\left(B\right)$$ 1

$$F=\gamma +n_A\sigma$$ 2

$$W\mathrm{ad}=S_A+S_{\mathrm{B}}-\gamma$$ 3

All density functional theory (DFT) computations were carried out with the Vienna Ab-initio Simulation Package (VASP, version 5.4.4). The interactions between ions and electrons were described using the projector augmented-wave (PAW) approach, while the Perdew–Burke–Ernzerhof (PBE) functional within the generalized gradient approximation (GGA) was adopted to treat exchange-correlation effects^47,48. For interface/surface energy calculations, we adopted a 3 × 3 × 1 k-point mesh and a cut-off energy of 400 eV. Bulk material properties were computed with a finer 12 × 12 × 12 k-point mesh and an increased cut-off energy of 600 eV. Structural relaxations were conducted via the conjugate gradient algorithm, with convergence criteria set to 10⁻⁵ eV for total energy and 0.01 eV/Å for atomic forces.

Cell assembly and electrochemical tests

Li symmetric cells were assembled by pressing 150 mg of electrolyte to obtain a electrolyte pellet (~1 mm), following with attaching 100 μm-thick Li foils (10 mm in diameter) on both sides of the pellet and pressing under 25 MPa for 1 min for adequate contact. Critical current density (CCD) and long-term Li plating/stripping tests were performed on Li symmetric cells using a Neware CT-4008 battery test system at 30 °C under 4 MPa. Potentiostatic electrochemical impedance spectroscopy (EIS) measurements were conducted on Li symmetric cells utilizing a Princeton PARSTAT MC-1000 potentiostat. The frequency range was from 1 MHz to 1 Hz with an amplitude of 10 mV, and data were acquired at a resolution of 10 points per decade. The distribution of relaxation times (DRT) was derived through the deconvolution of electrochemical impedance spectroscopy (EIS) data, implemented in MATLAB with the DRTtools. For Li||SS half cells, the assembly process mirrored that of the symmetric cells, except one lithium foil was substituted with a stainless steel (SS) disk measuring 10 mm in diameter. The plating capacity was 0.2 mA h cm⁻². Cyclic voltammetry (CV) for Li||NCM83 cells was conducted on a CHI 660E, with a scan range of 2.5–4.25 V vs. Li⁺/Li and at a scan rate of 0.1 mV s⁻¹.

LiIn||LiNi_0.83Co_0.12Mn_0.05O₂ (NCM83) cells utilizing the LPSC or AlC-LPSC as catholyte were assembled as below. First, the NCM83 active material and LPSC or AlC-LPSC catholyte (7:3 in mass ratio) were ball-milled at 110 rpm for 1 h to yield LPSC/NCM83 and AlC-LPSC/NCM83 composite positive electrodes. To assemble LiIn||NCM83, 110 mg of LPSC electrolyte was put into the PEEK mold and pressed under 100 MPa to obtain an electrolyte pellet (~0.8 mm). Subsequently, ~7 mg cm⁻² of the composite positive electrode material was evenly coated onto one surface of the electrolyte and compressed under 360 MPa for 2 min. A 100 μm-thick LiIn foil was then applied to the opposite side. The LiIn foil was fabricated through a repeated winding-rolling process of Li (China Energy Lithium Co., Ltd.) and In metal (Alfa Aesar). Galvanostatic charge-discharge cycling of the cells was performed using a Neware CT-4008 battery test system at 30 °C, with a voltage range of 1.9–3.65 V (vs. Li⁺/LiIn). Galvanostatic intermittent titration technique (GITT) measurements for LiIn||NCM83 cells were conducted on a Neware CT-4008 battery test system at 30 °C, with one data point acquired per second. The stack pressure of LiIn||NCM83 cells was 100 MPa.

Li||NCM83 cells utilizing AlC-LPSC as anolyte/catholyte were assembled as below. 110 mg of LPSC electrolyte was put into the PEEK mold and pressed under 100 MPa to obtain an electrolyte pellet (~0.8 mm). Then, 5 mg of AlC-LPSC electrolyte was evenly spread on the negative electrode side of the LPSC electrolyte pellet and pressed under 100 MPa. Subsequently, ~7 mg cm⁻² of the AlC-LPSC/NCM83 positive electrode material was uniformly coated onto the opposite side of the electrolyte and subjected to a pressure of 360 MPa for 2 min. A 100 μm-thick lithium foil was then applied to the remaining side and compressed at 10 MPa to ensure sufficient interfacial contact. For the cell employing high positive electrode loading, 1 wt% of vapor-grown carbon fiber (VGCF) is additionally added, and Li@Cu foils (with a 50 μm-thick Li layer) were utilized as the negative electrodes. Galvanostatic charge-discharge performance was evaluated using a Neware CT-4008 battery test system at temperatures of 0 and 30 °C within a voltage range of 2.5–4.25 V, under a low stacking pressure of 4 MPa. The capacity retention is calculated by dividing the capacity of the last cycle by the capacity of the first cycle (under the same current conditions). The electrochemical data presented in this work were obtained from single representative cells.

The assembling procedure of LiSi||NCM83 cells utilizing AlC-LPSC as anolyte/catholyte follows the same protocol as that used for Li||NCM83 cells, while the Li foil was replaced with LiSi. The LiSi negative electrode was fabricated according to the following steps: first, micro-sized silicon (μSi) powder was thoroughly mixed with polytetrafluoroethylene (PTFE) binder at a mass ratio of 98:2. The resulting mixture was then rolled into a uniform membrane using a precision-controlled rolling machine. Subsequently, the silicon membrane and Li foil were sequentially attached on the electrolyte pellet and subjected to 100 MPa to form the final LiSi alloy negative electrode. The LiSi||NCM83 cell was tested under 100 MPa at 30 °C within a voltage range of 2.5–4.25 V. All the assembling procedures were operated in the glove box with H₂O and O₂ content below 0.1 ppm.

Author contributions

Y.Z. and J.P.T. conceived the project and designed the experiment. J.E.K. and H.S. carried out the experiments. H.S. performed the theoretical study. J.Q.Z. and X.L.W. participated in the discussion of the data. J.E.K. and H.S. wrote the initial manuscript with the input of all authors. Y.Z. and J.P.T. revised the manuscript and directed the work.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. [A peer review file is available].

Funding

This work is supported by “Pioneer” and “Leading Goose” R&D Program of Zhejiang (No. 2025C01181), National Natural Science Foundation of China (Nos. U20A20126 and 52472109) and Central Government Guidance Fund for Local Science and Technology Development (No. 2024ZY02005).

Data availability

Source data are provided with this paper. All other relevant data that support the findings of this study can be obtained from the corresponding authors upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains Supplementary material available at https://doi.org/10.1038/s41467-026-76505-4.