Bulk-to-interface fluorination for stable and low-pressure all-solid-state lithium metal batteries

Abstract

All-solid-state lithium metal batteries are widely considered promising next-generation energy storage systems owing to high specific energy and enhanced safety. However, their practical deployment is hindered by high stack pressure and inferior electrochemical performance. Here, we exploit the fast thermodynamic diffusion of fluorine atoms to design a core-shell structured sulfide electrolyte, Li_5.4PS_4.4Cl_1.4F_0.2-0.2LiF, featuring 50 nm LiF nanoshell and F-enriched bulk. During electrochemical operation, fluorine atoms diffuse into the LiNi_0.83Co_0.12Mn_0.05O₂ positive electrode lattice, enhancing structural robustness and mitigating mechanochemical failure, while the LiF nanoshell stabilizes both the Li metal negative electrode and positive electrode interfaces through spontaneous fluorination diffusion. As a result, full cells demonstrate good electrochemical performance, including long cycle life, high-voltage stability, and robust operation across a wide temperature window. Furthermore, all-solid-state pouch cells operated under a low stack pressure of 2.5 MPa exhibit stable cycling over 350 cycles (1 C) with 85% capacity retention, and achieve a high specific energy of over 400 Wh/kg (based on solid electrolyte, Li metal, and positive electrode materials). This bulk-to-interface fluorination strategy effectively mitigates mechanochemical failures, offering an alternative pathway toward low-pressure, long-life, and high-energy all-solid-state batteries.

All-solid-state batteries have attracted widespread attention due to their potential for high specific energy and safety, but they struggle to operate stably under low pressure. Here, the authors reveal that fluorine atoms can spontaneously form stable bulk and interfacial structures through rapid diffusion kinetics, thereby improving the ability of all-solid-state batteries to operate stably under low pressure.

Introduction

All-solid-state lithium metal batteries (ASSLMBs) are regarded as promising next-generation energy storage technologies owing to their potential for high specific energy and intrinsic safety^1,2. By replacing liquid electrolytes with inorganic solid electrolytes and employing lithium metal negative electrodes, ASSLMBs could outperform conventional lithium-ion batteries in both energy and safety^3,4. Yet, their practical realization is still impeded by key challenges: the need for high external stack pressure to sustain electrode–electrolyte contact, instability at both negative electrode and positive electrode interfaces, and mechanical degradation of Ni-rich layered oxide positive electrodes^5–7. These factors lead to insufficient cycle life, poor rate performance, and limited specific energy at the cell level.

Considerable efforts have been devoted to addressing these challenges through solid electrolyte design (such as composite or heterogeneous multilayer electrolytes)^8–10, interfacial engineering (such as Al₂O₃-coated positive electrode and Mg₁₆Bi₈₄ negative electrode interlayer)^11,12, and electrode architecture optimization (such as Al-alternative positive electrodes and Ag-C negative electrodes)^13,14. Sulfide-based solid electrolytes, in particular, have emerged as promising candidates because of their high ionic conductivity (10⁻³ ~ 10⁻² mS cm⁻¹), favorable deformability, and scalable synthesis^15,16. However, they are chemically reactive against both lithium metal and high-voltage oxide positive electrodes, leading to unstable interphases, parasitic reactions, and accumulation of mechanical stress^17,18. Concurrently, Ni-rich layered oxides—the most practical high-energy positive electrodes—suffer from lattice oxygen release, phase transitions, and intergranular cracking under the harsh electrochemical and mechanical conditions of ASSLMB operation^19,20. Numerous strategies—including positive electrode coatings (e.g., fluoride)¹¹, electrolyte modified (e.g., F substitution or shell structure)^21–24, and elastomer interlayers²⁵—have been proposed to mitigate these issues. However. a unifying strategy that concurrently stabilizes both positive electrode and negative electrode interfaces and reinforces positive electrode integrity under low stack pressure has yet to be established.

Here we report a bulk-to-interface fluorination strategy that exploits the fast thermodynamic diffusion of fluorine atoms to overcome these coupled challenges. By designing a core–shell sulfide solid electrolyte, Li_5.4PS_4.4Cl_1.4F_0.2-0.2LiF (LPSCF-LiF), we combine an F-enriched bulk with a conformal 50 nm LiF nanoshell (Fig. 1a). During electrochemical cycling, fluorine atoms spontaneously diffuse into the Ni-rich positive electrode bulk, enhancing its structural robustness and suppressing mechanochemical failure, while the LiF nanoshell promotes the in situ formation of stable fluorinated interphases at both positive electrode and negative electrode interfaces (Fig. 1b). This multi-scale strategy not only enables ASSLMBs to operate stably at low stack pressures but also imparts good electrochemical performance, including long cycling stability (3000 cycles), high-voltage tolerance (188 mAh g⁻¹ at 4.6 V at 1 C), fast-charging capability (156 mAh g⁻¹ at 5 C), and wide-temperature adaptability (−40 to 80 °C). Furthermore, all-solid-state pouch cells under low stack pressure of 2.5 MPa exhibit stable cycling over 350 cycles with 85% capacity retention at 3 mAh cm⁻², and achieve a high cell-level specific energy of 409 Wh kg⁻¹ (based on solid electrolyte, Li metal, and positive electrode materials) under higher loading (5.4 mAh cm⁻²). Our findings highlight fluorine-mediated bulk and interfacial engineering as a powerful design principle for advancing practical ASSLMBs.

Fig. 1. Diagram of the formation of LPSCF-LiF electrolyte and their mechanism of action in batteries.

a Schematic diagram of LPSCF-LiF electrolyte design and synthesis. b Working principle of Li|LPSCF-LiF|Ni83 ASSLMBs.

Results

Structural origin of fluorinated electrolytes

To identify materials that remain stable against both lithium metal and high-voltage positive electrodes, we systematically compared the electrochemical stability windows of representative inorganic solid electrolytes (including sulfides-, halides-, and oxides-) and lithium binary compounds. As shown in Fig. 2a, none of the mainstream inorganic solid electrolytes (such as Li₆PS₅Cl, Li₃InCl₆, and Li₇La₃Zr₂O₁₂) can simultaneously satisfy the stability requirements at both the lithium negative electrode and high-voltage positive electrodes (0-4.3/4.6 V). Remarkably, among lithium binary phases, only lithium fluoride (LiF) meets this criterion, exhibiting a wide electrochemical window (0-6.4 V)²⁶. Furthermore, low surface water adsorption energy (0.16 eV) endows LiF with good hydrophobicity, thereby markedly enhancing environmental tolerance²³. Nevertheless, the intrinsically low ionic conductivity of LiF necessitates rational structural design to mitigate its adverse impact on overall electrolyte performance.

Fig. 2. Physical and chemical properties of LPSCF-LiF.

a Stable electrochemical window of typical inorganic solid electrolytes and lithium binary materials²⁶. b Diffusion coefficient of Li, S, Cl and F atoms with different structures based on AIMD simulation. c The structural diagram of LPSC₇F₁-F₁ electrolyte. d Thickness of LiF nanoshell, and ionic conductivity (under stacking pressure of 5 MPa) before and after air exposure of different solid electrolytes. e TEM, SAED, and EDS of LPSCF-LiF electrolytes. XPS of LPSCF-LiF electrolyte with 30 min etching: (f) F 1 s and (g) P 2p. h Three-dimensional distribution diagram of LPSCF-LiF electrolyte obtained by TOF-SIMS.

Ab initio molecular dynamics (AIMD) simulations can provide atomistic insights into how fluorine (F), introduced in different proportions and chemical states (Supplementary Data 1-5), modulates the structure and stability of Li₆PS₅Cl (LPSC₈) electrolytes (Fig. 2b). The results show that a minor (one-eighth) substitution of lattice chlorine (Cl) with F (LPSC₈ vs. LPSC₇F₁) exerts negligible influence on the diffusion coefficients and migration pathways of S, Cl and Li, indicating that such limited doping has minimal impact on the structural stability of the electrolyte (Supplementary Fig. 1a, b, and f). As expected from lower atomic mass, lattice F displays higher diffusivity than Cl and S. Increasing the degree of substitution (LPSC₆F₂) further accelerates F migration but destabilizes the Cl, S and Li sublattices, as excessive replacement with smaller F atoms perturbs the structural framework and significantly alters the ion-transport pathways (Supplementary Fig. 1c). In contrast, combining minor lattice F substitution with minor free F addition (LPSC₇F₁-F₁, Fig. 2c) preserves the stability of S, Cl and Li while maintaining larger difference in diffusion rate between fluorine and S/Cl atoms ((Supplementary Figs. 1d and f). Considering the larger diffusion coefficient of Li (2.14 × 10⁻⁵cm²s⁻¹) compared with other atoms (S: 7.2 × 10⁻⁷cm²s⁻¹, Cl: 6.5 × 10⁻⁷cm²s⁻¹, F: 6.18 × 10⁻⁶cm²s⁻¹), and the strong driving force between reductive Li and oxidative F, Li and F tend to migrate toward the surface to form a LiF nanoshell. With further free F incorporation (LPSC₇F₁-F₂), F diffusion accelerates but excessive F again destabilizes the lattice, markedly increasing S, Cl and Li mobility (Supplementary Fig. 1). These results highlight the necessity of precisely balancing lattice F and free F to simultaneously achieve structural stability and fast F dynamics diffusion.

Guided by AIMD insights, we designed a core–shell structured electrolyte via precise control over the relative proportions of lattice and free F, combined with tailored thermal treatment. Specifically, one-eighth of the Cl atoms in Li_5.4PS_4.4Cl_1.6 (LPSC_1.6, denoted LPSC) were substituted by F, yielding Li_5.4PS_4.4Cl_1.4F_0.2 (LPSC_1.4F_0.2, denoted LPSCF). The overall structure is only minimally affected, only as the smaller ionic radius of F leads to a slight lattice contraction, evidenced by the shift of XRD peaks to higher angles (Supplementary Figs. 2 and 3). Homogeneous lattice incorporation is confirmed, and no nanoshell formation is observed (Supplementary Fig. 4). Because elemental free F cannot be experimentally obtained, we introduced LiF as the source of free F. Upon adding a small amount of LiF (Li_5.4PS_4.4Cl_1.4F_0.2-0.2LiF, denoted LPSCF-LiF), a fraction of LiF is incorporated into the LPSC lattice, further reducing the lattice spacing (Supplementary Fig. 2b); whereas the remaining portion diffuses toward the particle surface during controlled slow cooling to form a ∼50 nm LiF nanoshell—consistent with the AIMD simulations that free F diffuses more rapidly than S and Cl owing to its lighter mass (Fig. 2d, e). Under slow cooling, this high diffusivity enables free F to migrate to the particle periphery. Notably, rapid quenching does not yield the nanoshell. At elevated temperatures, accelerated atomic thermal motion randomizes the positions of free fluorine (LiF), distributing it throughout the bulk and along the electrolyte boundaries. Upon quenching, LiF lacks sufficient time to migrate toward interfacial regions and becomes kinetically trapped at its high-temperature sites. This trapping preserves LiF-rich domains both in the interior and at the edges of the electrolyte, ultimately hindering nanoshell formation (Supplementary Fig. 5). A similar phenomenon was previously reported by Zhang et al.²⁷, showed that excess Cl (analogous to free Cl) with a relatively light mass can diffuse to the surface during slow cooling to form a LiCl nanoshell, whereas fast quenching prevents shell formation. F 1 s spectra of X-ray Photoelectron Spectroscopy (XPS) depth profiling verifies the coexistence of LiF nanoshell with internal Li-F bonds and Li-F-PS₄^3- linkages (Fig. 2f), while P 2p spectra provide further evidence for strong F-PS₄^3- interactions (Fig. 2g)²⁸. Cold-pressed pellets of LPSCF-LiF reveal uniform distribution of both nanoshell and lattice F over large dimensions (100 × 100 × 0.5 μm) by Time of Flight-Secondary Ion Mass Spectrometry (TOF-SIMS, Fig. 2h). Although nanoshell formation slightly reduces the initial conductivity, the electrolyte still retains high conductivity (3.25 mS cm⁻¹) and robust structural stability even after extended air exposure (Fig. 2d, Supplementary Figs. 6 and 7). Further F enrichment thickens the nanoshell (~200 nm) but induces severe conductivity loss and compromises structural integrity (Fig. 2d, Supplementary Figs. 1 and 8). Therefore, the LPSCF-LiF electrolyte with a LiF nanoshell and F-rich bulk uniquely integrates the dual structural features of bulk doping and artificial coating.

Negative electrode interfacial behavior of Li | LPSCF-LiF

The interfacial behavior at the negative electrode plays a decisive role in determining the performance of all-solid-state batteries. To assess the stability of the negative electrode–electrolyte interface, Li|SE|Li symmetric coin cells without stack pressure were assembled. At a current density of 1 mA cm⁻² (Fig. 3a), the plating/stripping lifetime of LPSCF-LiF (>7000 h) is markedly extended compared with that of LPSC (101 h) and LPSCF (813 h). Notably, both LPSC and LPSCF exhibited rapid rise in polarization voltage, ultimately leading to cell failure, indicating that electrolyte decomposition at the Li interface induces interfacial resistance buildup and polarization. In sharp contrast, LPSCF-LiF maintained stable voltage profile throughout, underscoring better interfacial stability. Electrochemical impedance spectroscopy (EIS) further corroborates that the resistance of the LPSC-based cell increases rapidly, accompanied by sharp rise in polarization voltage, leading to short-circuiting within 120 h, whereas the LPSCF-LiF-based cell shows no noticeable change (Supplementary Fig. 9).

Fig. 3. Negative electrode interfacial behavior of Li | LPSCF-LiF.

a Li plating/stripping curves of Li|SE|Li cells (coin cells). b Net charge transfer and P-S bond retention in different electrolyte after AIMD simulations (300 K, 10 ps). In situ Raman diagrams of electrolytes from (c) Li|LPSC|Li and (d) Li|LPSCF-LiF|Li cells (1 mA cm^-2, 0.5 mAh cm^-2, 30 °C). Cryo-TEM images of Li metal from (e) Li|LPSC|Li and (f) Li|LPSCF-LiF|Li cells (1 mA cm^-2, 0.5 mAh cm^-2, 30 °C). g TOF-SIMS of Li metal from Li|LPSCF-LiF|Li cells (1 mA cm^-2, 0.5 mAh cm^-2, 30 °C). FIB-SEM images of interface from (h) Li|LPSC|Li and (i) Li|LPSCF-LiF|Li cells (1 mA cm^-2, 0.5 mAh cm^-2, 30 °C). j XPS of electrolytes from Li|LPSC|Li and Li|LPSCF-LiF|Li cells after 30 min etching (1 mA cm^-2, 0.5 mAh cm^-2, 30 °C).

Multiscale simulations and characterizations, from atomistic insights provided by AIMD and Cryo-TEM (Transmission Electron Microscope) to mesoscale observations by Focused Ion Beam-Scanning Electron Microscope (FIB-SEM), in situ Raman spectroscopy and TOF-SIMS, enable more accurate understanding of the underlying origins of the pronounced discrepancies observed during the operation of symmetric cells at the macroscopic level. AIMD simulations reveal that (Fig. 3b and Supplementary Data 6, 7), at the Li|LPSC interface, the strongly reducing nature of Li drives substantial electron transfer into the electrolyte, resulting in P-S bond cleavage and the subsequent formation of Li₂S decomposition products, accounting for the rapid polarization observed in LPSC (Supplementary Fig. 10b). In situ Raman spectroscopy detects the formation of Li₂S within 20 h (Fig. 3c), while cryo-TEM further confirms the emergence of a heterogeneous interlayer arising from chemical instability (Fig. 3e). SEM-FIB analysis likewise verifies that chemical degradation leads to the development of cracks and voids, resulting in the loss of interfacial contact and mechanical failure (Fig. 3h). Strikingly, introducing a LiF interlayer between LPSC and Li suppresses electron transfer, preserves the integrity of the P–S bonds (Fig. 3b), and stabilizes the LPSC framework ((Supplementary Fig. 10d). This protection is corroborated by in situ Raman (Fig. 3d) and cryo-TEM (Supplementary Fig. 11), both of which show that the electrolyte and Li negative electrode remain structurally unaltered at the early stage, indicating that the LPSCF-LiF electrolyte markedly enhances the mechano-chemical stability at the Li interface.

Although LiF is thermodynamically stable against Li initially, long-term cycling drives F^- migration within nanoshell toward the Li negative electrode under the applied electric driven. The migrated F⁻ combines with Li⁺ to form a thin (~24 nm), uniform solid-electrolyte interlayer (SEI) after 500 h of plating/stripping, comprising LiF nanocrystals embedded within an amorphous matrix (Fig. 3f). Correspondingly, the LiF nanoshell in the LPSCF-LiF electrolyte becomes thinner due to F migration (Supplementary Fig. 12). Given the intrinsic thermodynamic and electrochemical stability of LiF against Li metal, we attribute the formation of this interlayer to diffusion rather than interfacial reactions. As described by Fick’s law, species migrate from regions of high to low concentration, and the small ionic radius of F^- enables rapid diffusion—consistent with its high mobility within the bulk electrolyte. Because diffusion proceeds more slowly than interfacial reactions, the SEI does not form at the initial stages but instead develops gradually during prolonged cycling. The diffusion-derived SEI also exhibits better uniformity, as evidenced by TOF-SIMS, which reveals a spatially homogeneous F-rich interlayer over a large area (Fig. 3g). This homogeneous and stable SEI ensures interfacial stability and improves wettability, as evidenced by the continuous and well-contacted interface revealed by FIB-SEM (Fig. 3i). XPS further confirms that this diffusion-formed SEI is dominated by F-rich species with negligible contributions from other components (Supplementary Fig. 13). Examination of the electrolyte interior shows that, unlike LPSC, which undergoes extensive decomposition of PS₄^3- groups, the LPSCF-LiF electrolyte largely retains its original framework, indicating that the LiF nanoshell effectively protects the bulk from degradation (Fig. 3j). The intrinsic robustness of the electrolyte, combined with in situ formation of LiF-rich SEI, endows dynamic interfacial stability over 7000 h plating/stripping.

Positive electrode Interfacial Behavior of LPSCF-LiF | Ni83

To probe the cathodic interfacial behavior, we assembled Li|LPSCF-LiF|Ni83 cells with different electrolytes in the positive electrode, while the separator employed the electrolyte LPSCF-LiF to minimize the influence of the negative electrode on the observed results. Distribution of relaxation times (DRT) analysis derived from in situ electrochemical impedance spectroscopy resolved four contributions: bulk electrolyte resistance (D1), interfacial resistance (D2), and cathodic charge-transfer and ion-diffusion resistance (Fig. 4a, b). During the initial cycle, both LPSC and LPSCF-LiF exhibited comparable D1 and D2 resistances. However, with increasing state of charge, LPSC showed pronounced rise in charge-transfer resistance, together with distinct separation of charge-transfer and diffusion peaks, indicative of structural degradation and emergence of additional impedance components. In sharp contrast, even at 4.6 V, LPSCF-LiF exhibited no discernible decomposition signatures, underscoring better mechano-chemical stability under high-voltage.

Fig. 4. Positive electrode interfacial behavior of LPSCF-LiF | Ni83 (1 C = 200 mA g⁻¹).

DRT analysis at different states of charge with (a) LPSC and (b) LPSCF-LiF electrolyte (1 C, 30 °C). SEM-FIB of positive electrode with (c) LPSC and (d) LPSCF-LiF electrolyte after 100 cycles (1 C, 30 °C). HRTEM of Ni83 positive electrode with (e) LPSC and (f) LPSCF-LiF electrolyte after 100 cycles (1 C, 30 °C). g. FIB-HADDF-STEM and EDS of Ni83 positive electrodes with LPSCF-LiF electrolyte after 3 cycles (0.1 C, 30 °C). TOF-SIMS of (h) surface and (i) bulk of Ni83 with LPSCF-LiF electrolyte after 100 cycles (1 C, 30 °C).

FIB-SEM morphological analysis provided direct structural evidence for the origin of impedance splitting: the Ni83-LPSC interface displayed severe deterioration, including interfacial voids and internal cracks caused by mechanical stress and heterogeneous interfacial strain (Fig. 4c). Conversely, the Ni83-(LPSCF-LiF) interface retained intimate interfacial contact without significant cracking (Fig. 4d). High-Resolution Transmission Electron Microscopy (HTREM) further reveals that in LPSC electrolyte, the Ni83 particles fragmented into petal-like features, indicative of catastrophic collapse and decomposition (Fig. 4e and Supplementary Fig. 14). By contrast, in the LPSCF-LiF electrolyte, Ni83 maintained structural coherence with a thin (6 nm) positive electrode electrolyte interlayer (CEI, Fig. 4f). XAFS analyses show that both the Ni-O and Ni-M coordination environments in Ni83 remain essentially unchanged before and after cycling (Supplementary Fig. 15). Collectively, these results demonstrate that the LPSCF-LiF electrolyte significantly improves the mechano-chemical stability of both the Ni83 interface and the bulk.

Further Focused Ion Beam–High-Angle Annular Dark Field–Scanning Transmission Electron Microscopy (FIB–HAADF–STEM) elucidates the interaction between the electrolyte and Ni83. Energy-dispersive spectroscopy (EDS) reveals that, after only three cycles, pronounced F signals appear both at the surface and within the interior of Ni83—despite the pristine positive electrode containing no fluorine (Fig. 4g and Supplementary Fig. 16). This observation indicates not only the formation of an F-rich CEI but also substantial F diffusion into the Ni83 bulk, which is further corroborated by Electron Energy Loss Spectroscopy (EELS), Supplementary Fig. 17. The primary source of F is the LiF nanoshell, whose thinning after cycling supports this conclusion (Supplementary Fig. 18). TOF-SIMS mapping over larger areas confirms the structural homogeneity: the CEI is enriched in F and Ni (Fig. 4h), with negligible amounts of P/S/Cl impurities from the electrolyte (Supplementary Fig. 19), and the Ni83 bulk is likewise F-rich (Fig. 4i). As noted by Wu et al.^29–31, in situ formation of LiF-rich CEI effectively mitigates electrolyte and positive electrode decomposition and suppresses parasitic reactions. Moreover, as reported by Wang et al.^12,32, electrochemically driven F^- migration into the Ni83 bulk further enhances positive electrode structural resilience; the high electronegativity of F stabilizes neighboring transition-metal and oxygen atoms, anchoring them and preventing dissolution or migration. By leveraging a LiF-rich nanoshell capable of F diffusion, our LPSCF-LiF electrolyte simultaneously achieves F-rich CEI and F-rich Ni83 bulk—yielding dual interfacial and bulk protection that collectively enhances the mechano-chemical stability of the positive electrode.

Electrochemical performances of Li | LPSCF-LiF | Ni83 ASSLMBs

To evaluate the electrochemical performance of all-solid-state batteries, we assembled Li | |Ni83 cells using LPSCF-LiF as the electrolyte separator in all configurations, while incorporating different electrolytes into the composite positive electrode. Under high-rate conditions (5 C charge, 1 C discharge) and comparatively low stack pressure (8 MPa), the LPSCF-LiF based cell delivering high specific capacity of 156 mAh g⁻¹ at first cycle and maintain 103 mAh g⁻¹ after 3000 cycles, corresponding to average capacity decay of only 0.011% every cycle (Fig. 5a and Supplementary Fig. 20c). In stark contrast, the LPSC based cell underwent rapid capacity, coulombic efficiency decay and polarization increases after merely ~100 cycles, indicative of structural collapse at the electrolyte-Ni83 interface (Supplementary Fig. 20a). Meanwhile, the LPSCF based cell maintained consistently high coulombic efficiency but displayed continuous capacity fading throughout cycling (Supplementary Fig. 20b).

Fig. 5. Electrochemical performances of ASSLMBs (1 C = 200 mA g^-1).

a Discharge capacity and coulombic efficiency of Li | |Ni83 cells (mold cells) with different positive electrode at 5C-1C (charge-discharge) after 3 cycles at 0.2 C (mold cells) at 30 °C. b Discharge capacity of Li | |Ni83 cells (mold cells) with LPSCF-LiF electrolyte at different temperature (−40, 0, 80 °C). c Discharge capacity of Li | |Ni83 cells (mold cells) with different positive electrode at 1 C under 2.8-4.6 V at 30 °C. d Discharge capacity and coulombic efficiency of pouch cells with LPSCF-LiF electrolyte at 1 C after 3 cycles at 0.2 C under 2.5 MPa (inset: corresponding relative pressure variation) at 60 °C. e Specific energy of pouch cells with high loading at 0.2 C under 2.8-4.6 V and 2.5 MPa (inset: cell real picture) at 60 °C. f Comparison of charge current density and cycle life between our cell and state-of-the-art cells reported in the literature^{12,13,33–42}.

The stability of the cells was further evaluated by examining their electrochemical performance under multiple operating conditions, including low stack pressure, high voltage, wide temperature range. Under low stack pressure (2.5 MPa) and high-rate conditions (5 C charge and 1 C discharge), the LPSCF-LiF electrolyte-based cell retains capacity retention of 85% after 1000 cycles, whereas the LPSC-based cell falls to ~40% within fewer than 200 cycles (Supplementary Fig. 21). Even with coated positive electrodes (Ni83@LNO), the cell with LPSC exhibits pronounced capacity decay, accompanied by extensive cracking within the Ni83 bulk. These results further demonstrate that the LPSCF-LiF electrolyte not only improves interfacial stability with Ni83 but also enhances the structural integrity of the Ni83 bulk. The LPSCF-LiF based cell also exhibited good cycling stability and high areal capacity across broad temperature range (-40 to 80 °C) under 2.5 MPa stack pressure, underscoring robust performance under multiple operating conditions (Fig. 5b). At an elevated cutoff voltage of 2.8-4.6 V under a stack pressure of 2.5 MPa at 1 C, the instability of the LPSC-based cell became even more pronounced, with the coulombic efficiency collapsing within only a few dozen cycles, accompanied by severe fluctuations in the charge–discharge profiles and rapid capacity decay (Fig. 5c and Supplementary Fig. 22a). Although the LPSCF-based cell maintained relatively high coulombic efficiency, it retained only ~10% of its initial capacity after 200 cycles (Supplementary Fig. 22b). In sharp contrast, the LPSCF-LiF-based cell exhibited high-voltage durability, delivering a high specific capacity of 188 mAh g⁻¹ and preserving 91% of its capacity over extended cycling (Supplementary Fig. 22c).

To evaluate practical feasibility, we further assembled pouch cells and operated under low stack pressure of 2.5 MPa. The electrolyte and composite positive electrode were each mixed with 1 wt% PTFE to form freestanding films. While the addition of PTFE slightly reduces the ionic conductivity of the electrolyte, its influence on cycling stability is negligible (Supplementary Fig. 23). When paired with 3 mAh cm^-2 positive electrode, Li foil, and solid electrolyte film, the cell retained 85% of capacity over 350 cycles at 1 C (Fig. 5d). During the first three cycles (0.2 C), pressure drop of ~0.03 MPa was observed due to activation and interfacial stabilization. After reapplying 20 MPa restored interfacial contact, almost no detectable pressure loss occurred under 2.5 MPa (1 C), confirming the structural robustness of this battery system under low-pressure conditions. Increasing the positive electrode loading (5.4 mAh cm^-2) and cut-off voltage (2.8-4.6 V) further enabled high cell-level specific energy of 409 Wh kg⁻¹, calculated on basis of the total mass of solid electrolyte, Li metal, and positive electrode materials (Fig. 5e and Supplementary Table 1). Moreover, the cell still maintains more than 250 Wh kg⁻¹ after 50 cycles. Notably, the performance of the LPSCF-LiF based cell is competitive with that of previously reported solid-state batteries (Fig. 5f, Supplementary Table 2)^{12,13,33–42}. The good cycling stability and high specific energy achieved in pouch cells under low stack pressure underscore the practical potential of this battery system.

Discussion

In summary, we developed a bulk-to-interface fluorination strategy to tackle the coupled challenges of low-pressure operation, interfacial instability, and positive electrode degradation in all-solid-state lithium metal batteries. By engineering a core–shell sulfide electrolyte, Li_5.4PS_4.4Cl_1.4F_0.2-0.2LiF (LPSCF-LiF), we combined an F-enriched bulk with a conformal LiF nanoshell. During electrochemical cycling, fluorine atoms in LiF nanoshell diffuse into the Ni-rich positive electrode bulk to enhance structural integrity, while the LiF nanoshell also induces stable fluorinated interphases at both Li metal and Ni-rich oxide interfaces. This synergistic mechanism effectively suppresses mechanochemical failures and enables stable cycling under stack pressures as low as 2.5 MPa. As a result, ASSLMBs based on the LPSCF-LiF electrolyte demonstrate good electrochemical performance, including long cycle life of 3000 cycles with minimal capacity decay, robust operation up to 4.6 V, fast-charging capability, and stability across a wide temperature window from -40 to 80 °C. Importantly, all-solid-state pouch cells deliver high cell-level specific energy (> 400 Wh kg⁻¹, based on solid electrolyte, Li metal, and positive electrode materials) under practical loading and low pressure conditions (2.5 MPa), highlighting the promise of this approach for real-world deployment. These findings establish fluorination-mediated bulk and interfacial engineering as an effective design principle for overcoming mechanochemical failures, offering a promising pathway toward the realization of low-pressure, high-energy, and durable solid-state batteries.

Methods

Materials preparation

To prepare argyrodite SEs fabrication, LiF (Aladdin, 99.9%), Li₂S (Aladdin, 99.9%), LiCl (Aladdin, 99.9%), and P₂S₅ (Aladdin, 99%) powders were weighted according to the stoichiometric molar ratio of Li_5.4PS_4.4Cl_1.6-x-yLiF. These mixtures with a total mass of 1 g were placed in a zirconia jar (50 ml) filled with zirconia balls (10 large balls with diameter 10 mm, and other small balls with diameter 5 mm) and mechanically milled at 500 rpm for 8 h using a planetary ball milling apparatus (YXQM, Changsha Miqi Instruments & Equipment Co.). The mass ratio between the zirconia balls and the mixtures was 70:1. The resultant powders were pressed into pellets in a quartz tube (KSL-1100X, HF-Kejing) for annealing at 500 °C for 8 h with ramping and cooling rate of 1 °C min^–1. The contrasting electrolyte was prepared via a quenching method: following the aforementioned heating and insulation conditions, the quartz tube was immediately immersed in cold water to induce quenching. All samples were protected in an argon atmosphere (O₂ < 0.1 ppm, H₂O < 0.1 ppm) throughout all procedures.

Materials characterization

X-Ray Diffraction (XRD) was conducted using Rigaku Ultima IV system (Cu Kα, scan rate: 10°/min, step size: 0.02°). For in situ Raman scattering tests, LabRAM HR Evolution instrument with 532 nm He-Ne laser, and home-made spectro-electrochemical cell obtained from Beijing Scistar Technology Co. Ltd. X-ray photoelectron spectroscopy (XPS; Axis Supra + , Shimadzu Corporation) with a monochromatic Al Kα radiation was used to investigate the chemical compositions of solid electrolytes and negative electrodes and solid electrolytes before and after cycles. The cycled solid electrolytes and negative electrodes for XPS characterization were peeled out from the disassemble battery pellets, with the aid of adhesive tape. All binding energies of the XPS spectra were calibrated by setting the adventitious carbon C 1 s peak to 284.8 eV. Cryogenic focused ion beam (Cryo-FIB) milling and scanning electron microscopy (SEM) analysis were performed on Thermo Scientific Aquilos SEM (10 kV). The Transmission Electron Microscope (TEM) test was conducted using Thermo Fisher Scientific., Spectra 300 (200 kV) and Thermoscientific/Krios G3i (200 kV). When performing Cryo-TEM test, The TEM grid was loaded onto the cryo-holder in liquid nitrogen atmosphere and quickly transferred to the TEM in cryogenic temperature. Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS) was conducted under PHI Nano ToF2. TOF-SIMS was used to investigate the chemical compositions of solid electrolytes and negative electrodes before and after cycles. The cycled negative electrodes for TOF-SIMS characterization were peeled out from the disassemble battery pellets, with the aid of adhesive tape. For the air-stability test, samples were exposed to ambient conditions (0.5 h, 30 °C, 30 ± 2% relative humidity), followed by drying at 150 °C for 2 h to remove surface-adsorbed moisture. Subsequent measurements, including ionic conductivity and related characterizations, were then conducted. X-ray absorption fine structure (XAFS) spectra were collected using a benchtop easyXAFS300+ system (easyXAFS, Inc., USA) equipped with a Johansson-type curved crystal monochromator (Si (5 5 1)) and a Mo X-ray tube. For all measurements, the specimens were loaded into air-tight sample holder and then transported from the Ar-filled glovebox (O₂ < 0.1 ppm, H₂O < 0.1 ppm) to the equipment used for measurements.

Electrochemical measurements

Electrochemical impedance spectroscopy (EIS) measurements were conducted by using SP-300 impedance analyzer (BioLogic). The EIS data was collected within the frequency range of 1 Hz to 7 MHz (10 data points for per decade of frequency) and at the amplitude of 20 mV for the applied potentiostatic signal under 30°C. Ionic conductivity was evaluated using CA | SE | CA (carbon-coated aluminum foil) cells under an external pressure of 5 MPa.The galvanostatic charge and discharge measurements were carried out by Land CT2001A and NEWARE CT-4008Tn-5V10mA-164 battery analyzer. Li-Li symmetric cells were assembled in a coin-cell configuration, whereas full cells were evaluated using mold-cell and pouch-cell formats. All cell fabrication processes were conducted in an Ar-filled glovebox (O₂ < 0.1 ppm, H₂O < 0.1 ppm) and all battery tests are carried out in the incubator (LRH-150, Shanghai Yiheng Scientific Instrument Co., Ltd.). All electrochemical measurements were repeated at least three times to ensure reproducibility, and the cell with the median performance was selected for presentation in the figures. All positive electrodes: 1 C = 200 mA g^–1. All the negative electrodes use Li foil (diameter: 10 mm, thickness: 20 μm, mass: ~1 mg, purity: >99.9%, China Energy Lithium Co., Ltd). About coin cell (Li symmetric cells) assembly and tests: first, the solid electrolyte (diameter: 12 mm, mass: ~60 mg cm^-2, thickness: ~250 μm) was co-pressed at 500 MPa for 10 min in the stainless-steel molds. Then, two lithium foils (diameter: 10 mm) were placed on either side of the electrolyte respectively, and were consolidated at 100 MPa for 1 min. Put the obtained symmetric cells into the coin battery mold (CR2032), the case and spring are stainless steel. The electrochemical performance was evaluated at 30 °C. About all-solid-state cell assembly and tests: all-solid-state cells were assembled by using Li foil as negative electrode, using LPSCF-LiF as solid electrolyte diaphragm, and using mold cell. The composite positive electrode consisted of Ni83 (from Canrd), carbon black (from Canrd), and the respective solid electrolytes in a weight ratio of 72:5:23 (wt%) using Hummer Acoustic Mixers (HAM100). The battery assembly utilized stainless steel molds as clamping fixtures, polyetheretherketone (PEEK) as the outer sleeve, and alumina ceramics as the inner liner (diameter: 10 mm). The assembly of mold batteries were performed through a layer-by-layer cold-pressing protocol. First, LPSCF-LiF solid electrolyte (diameter: 10 mm, mass: ~60 mg cm^-2, thickness: ~250 μm) was compacted within the alumina-ceramic inner liner at 100 MPa. Subsequently, the Ni89 composite positive electrode was then deposited onto the electrolyte surface and co-pressed at 500 MPa for 10 min. The Li foil was subsequently positioned against other side of the electrolyte. Finally, the cell was consolidated at 100 MPa for 1 min. The resulting pellet was sandwiched between stainless-steel current collectors and hermetically sealed within the mold. The potential window was set as 2.5 ~ 4.3 V (vs. Li⁺/Li) or 2.8 ~ 4.6 V (vs. Li⁺/Li) for batteries. The electrochemical performance was evaluated under 2.5/8 MPa at 30 °C. The temperature for the temperature variation test is 80, 0 and -40 °C. About pouch cells: the solid electrolyte is mixed with 1 wt% polytetrafluoroethylene (PTFE, Aladdin) and rolled to about 40 μm under 60 °C. The composite positive electrode is mixed with 1% PTFE and rolled to specific loading under 60 °C. The positive electrode and Li foil were cut into 2 × 3 cm and the solid electrolyte sheet was 2.1 × 3.1 cm. Roll the lithium foil, electrolyte and composite positive electrode sheets together under 60 °C, then encapsulate the single-layer battery in a pouch battery. The potential window was set as 2.5 ~ 4.3 V (vs. Li⁺/Li) or 2.8 ~ 4.6 V (vs. Li⁺/Li) for batteries. The electrochemical performance was evaluated under 2.5 MPa at 60 °C.

Computational details

All calculations were performed using density functional theory (DFT) with van der Waals (vdW) corrections, as implemented in the Vienna ab initio Simulation Package (VASP). The exchange–correlation interactions were described by the Perdew-Burke-Ernzerhof (PBE) functional within the generalized gradient approximation (GGA). The projector augmented wave (PAW) method was employed to treat the ion-electron interactions. A plane-wave cutoff energy of 400 eV was used, and structural relaxations were considered converged when the total energy reached 1 × 10⁻⁵eV. For interfacial models containing large numbers of atoms, only the K-point (1 ×1 × 1 k-point mesh) was sampled. Ab initio molecular dynamics (AIMD) simulations were conducted in the canonical (NVT) ensemble using a Nosé thermostat with a time step of 2 fs. The structural evolution was analyzed via radial distribution function (RDF) calculations using the Visual Molecular Dynamics (VMD) package. Some of the source data were processed and rendered as graphical images by VESTA software⁴³.

Author contributions

C.W., Z.S., Y.-B.H., X.S., and Z.Z. supervised the research and wrote the manuscript. J.S. completed most of the experimental research and wrote the manuscript. W.Y., S.J., S.G., and X.C., M.L. analyzed the partial data. All the authors discussed the results and commented on the manuscript.

Peer review

Peer review information

Nature Communications thanks Hiroyuki Ueda, Min-Sik Park and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The source data generated in this study are provided in the Supplementary Information or Source Data file. 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 10.1038/s41467-026-73012-4.