Introduction

To achieve the goal of high-energy-density and low-cost lithium-ion batteries, nickel-rich and cobalt-free layered oxide positive electrode materials have attracted widespread attention1,2,3,4. Among these oxides, LiNiO2 (LNO) offers a high theoretical specific capacity of 275 mAh g-1 and eliminates reliance on expensive and toxic cobalt5,6. However, the widespread application of LNO is hindered by thermodynamic instability, which leads to parasitic side reactions with organic liquid electrolytes at highly delithiated states (~4.2 V vs. Li/Li+), increasing the risk of thermal run-away and potential explosion

All-solid-state lithium-ion batteries (ASSLIBs) with nonflammable solid electrolytes (SEs) have emerged as a promising solution to address the safety challenges of LNO8,9,10. Among various SE candidates, sulfide SEs have garnered extensive attention due to their high ionic conductivity (up to 10-2 S cm-1) and favorable ductility and elasticity11. Nevertheless, LNO suffers from interfacial incompatibility with the sulfide SEs. This incompatibility triggers parasitic reactions involving oxidation and decomposition of the sulfide SEs, especially at high voltages, resulting in increased interfacial impedance and rapid capacity decay12,13,14.

Constructing a protective layer on the LNO has proven to be an effective strategy to mitigate interfacial incompatibility with sulfide SEs. For instance, LixAlyZnzOδ and LiNbO3 coating layers have been demonstrated to suppress continuous side reactions at LNO|Li6PS5Cl (LPSCl) interface and improve the cycling stability of LNO-based ASSLIBs, achieving capacity retention of 83.1% after 200 cycles and 74% after 400 cycles, respectively8,15,16. Although significant progress has been made through interfacial modification, long-term cycling stability of LNO in sulfide-based ASSLIBs is still hindered by its intrinsic structural and mechanical fragility. The hexagonal 2-hexagonal 3 (H2-H3) phase transition during cycling triggers lattice collapse and internal stress, driving the formation of microcracks. As the SE cannot penetrate fractured regions, permanent ionic contact disruption occurs in the particles, generating electrochemically inactive domains and capacity loss. Meanwhile, the drastic bulk lattice changes under the rigid solid-solid contact between the positive electrode and SE also lead to interfacial contact loss5,6,7,17,18,19. Therefore, the coupled effect between bulk phase structural collapse and interfacial degradation constitutes the fundamental cause that limits the cycling life of LNO-based ASSLIBs. Achieving long-term stable operation requires addressing both issues simultaneously.

Here, we design a universal surface-to-bulk heterophase reconstruction (HPR) strategy achieved through differences in the diffusion kinetics of dopant elements: high-valence dopants tend to accumulate on the surface, forming intermediate phases that improve interfacial compatibility and stability; whereas low-valence dopants readily diffuse into the bulk, thereby synergistically enhancing both interfacial and bulk stability. The LiNi0.964Al0.03W0.006O2 (LNAW) is a representative example. W6+ tends to accumulate at the surface to form a W-enriched spinel phase, which effectively improves the interfacial compatibility with sulfide SE and exhibits enhanced ionic conductivity and interfacial Li+ diffusion kinetics. The inactive Al3+ diffuses into the bulk to form stable TM−O bonds, effectively reducing Li+/Ni2+ mixing, maintaining a robust layered bulk structure. More importantly, the synergy between Al and W induces a HPR-derived stable layered-spinel framework, which significantly suppresses the detrimental H2-H3 phase transition and enhances both structural and mechanical stability. We demonstrate that suppressing interfacial side reactions is the prerequisite for achieving high specific capacity, while enhancing bulk structural stability is the key to mitigating rapid capacity degradation, thereby achieving a balance between high capacity and long-term stability during extended cycling. As a result, under an elevated cut-off voltage of 4.5 V, the LNAW delivers an initial specific capacity of 187.6 mAh g-1 at 0.5 C and sustains long-term stability of 720 cycles with a high retention of 65% in sulfide-based ASSLIB. More importantly, the proposed HPR strategy shows good universality and has been further validated in Al/Mo, B/W, and B/Nb systems. This strategy offers a promising pathway toward cobalt-free positive electrodes for safe, high-energy-density and long-life ASSLIBs.

Results

Degradation mechanisms of LNO in sulfide-based ASSLIBs

Elucidating the impact of structural and interfacial evolution on capacity fade of LNO in sulfide-based ASSLIBs is critical for rational material design. Here, we conducted comprehensive electrochemical and spectroscopic analyses on cycled composite electrodes. As shown in Supplementary Fig. 1a, b LNO exhibits rapid capacity degradation in sulfide-based ASSLIBs with a 64% capacity retention after 100 cycles. Cyclic voltammogram (CV) of LNO revealing a sequential phase transition from hexagonal 1 to monoclinic (H1-M), followed by a transition to hexagonal 2 (M-H2), and finally to hexagonal 3 (H2-H3) (Supplementary Fig. 1c)20. As shown in Supplementary Fig. 1d, when the cut-off voltage is limited to 4.15 V to avoid the H2-H3 phase transition, the capacity retention reaches 96.6% after 80 cycles, which is significantly higher than the value (75.3% after 80 cycles) under the cut-off voltage is 4.3 V. This result implies that the H2-H3 phase transition is largely responsible for rapid capacity degradation. This is because the H2-H3 phase transition is accompanied by poorly reversible structural evolution and significant lattice strain, leading to the formation of intragranular microcracks over extended cycling, and thereby accelerating the degradation of LNO21.

To further elucidate the impact of irreversible H2-H3 phase transition, the cycled composite positive electrodes were disassembled for focused ion beam scanning electron microscopy (FIB-SEM) and scanning transmission electron microscopy (STEM) analyses. As shown in Supplementary Fig. 2a, FIB-SEM reveals the formation of significant microcracks in the cycled LNO particles. As the SE cannot penetrate these fractured particles, it results in the formation of isolated and electrochemically inactive regions. Furthermore, high-resolution STEM was employed to examine the cycled composite positive electrode at the atomic scale. As shown in Supplementary Fig. 2b, a ~ 5 nm thick rock-salt phase was observed on the surface of LNO. The formation of the rock-salt phase is caused by the migration of Ni2+ into the Li layer under highly delithiated states, leading to the Li+/Ni2+ mixing. Such disorder hinders Li+ diffusion within the bulk structure, thereby impairing electrochemical performance22,23.

X-ray photoelectron spectroscopy (XPS) of the cycled composite positive electrode was analyzed to investigate the interfacial compatibility between LNO and LPSCl. As shown in Supplementary Fig. 2c, severe interfacial parasitic reactions between the LNO and the LPSCl were demonstrated by the formation of polysulfides and sulfite byproducts (blue and green lines). These byproducts show poor ionic conductivity to increase the interfacial resistance of ASSLIBs, thereby significantly damaging the battery performance with low specific capacity and poor cycling stability24,25. Herein, we demonstrated that the fast capacity fading of LNO in sulfide-based ASSLIBs is not solely attributed to the accumulation of passivation byproducts from severe LNO|LPSCl interfacial parasitic reactions. The formation of particle microcracks and detrimental rock-salt phases also plays a critical role.

Material design and structural characterizations of LNAW

To address these challenges, a surface-to-bulk HPR strategy was developed to introduce Al and W into LNAW via a one-step solid-state sintering method (Methods section). As illustrated in Fig. 1a, electrochemically inactive Al stabilizes the bulk layered structure, while W facilitates surface reconstruction, leading to the formation of a robust W-enriched surface spinel phase. The chemical composition of LNAW was determined by inductively coupled plasma-optical emission spectroscopy (ICP-OES, Supplementary Table 1), showing that the atomic ratio of elements closely matches the designed stoichiometric ratios. The synchrotron X-ray diffraction (SXRD) patterns and Rietveld refinements demonstrate that all samples adopt a hexagonal R-3m structure, as shown in Fig. 1b and Supplementary Figs. 3–55,26. The structural site occupancy refinement for all atoms in LNO, LiNi0.97Al0.03O2 (LNA) and LNAW is provided in Supplementary Tables 2-4. The degree of Li+/Ni2+ mixing can be estimated using Rietveld refinements, which is an important indicator of structural stability27. As shown in Supplementary Table 3, the LNA and LNAW exhibited a lower Li+/Ni2+ mixing (2.92% and 2.2%, respectively) compared to 4% in LNO, supporting the role of Al in stabilization of the bulk layered structure.

Fig. 1: Materials design and formation mechanism of LNAW.
Fig. 1: Materials design and formation mechanism of LNAW.The alternative text for this image may have been generated using AI.
Full size image

a The schematic illustration of the heterophase reconstruction concept for the LNAW sample. b SXRD and Rietveld refinement results for the LNAW sample. The theoretical Bragg reflections correspond to the layered R-3m phase. c The Ni K-edge XANES spectra of LNO and LNAW samples, with the inset showing the first derivative of XANES spectra. Normalization was performed using standard procedures in Athena software. XPS spectra of d W 4 f and e Al 2p in LNO and LNAW materials. f The migration energy barriers for the diffusion of Al and W to the Li sites and Ni sites, respectively. g Calculated the Li+/Ni2+ exchange energies for Al and W cations substituted in LNO.

X-ray absorption spectroscopy (XAS) and XPS were performed to investigate the coordination environment and valence states. The Ni K-edge X-ray absorption near edge structure (XANES) of LNO and LNAW are presented in Fig. 1c. The first derivative of the XANES spectrum (inset) showed a shift toward lower energy for LNAW, demonstrating a decrease in the average Ni valence state compared to LNO. As shown in Fig. 1d, e, XPS spectra of Al and W elements confirm their successful incorporation in LNAW. The W 4 f spectrum displays two peaks at 35.3 and 37.4 eV, corresponding to W 4f7/2 and W 4f5/2 of the W6+28,29,30. The Al 2p spectrum exhibits a peak at 72.3 eV, corresponding to the Al3+−O bond31. The formation of such strong Al−O bonds strengthen the lattice rigidity and enhances structural stability, as Al3+ is electrochemically inert and does not participate in redox reactions32.

To investigate the formation mechanism of this unique structure, density functional theory (DFT) calculations were performed using the structural models illustrated in Supplementary Fig. 6 and Supplementary Data 1-3. As shown in Fig. 1f, the calculated migration energy barriers reveal that Al has a lower migration barrier than W, resulting in faster Al diffusion into the LNO bulk, while W tends to localize at the surface. In addition, the formation energies of Al and W substitution at Ni and Li sites (Supplementary Fig. 7) indicate that Al and W preferentially substitute at Ni sites rather than Li sites, which is in good agreement with the results from SXRD Rietveld refinement. The Li+/Ni2+ exchange energy profiles shown in Fig. 1g reveal that Al doping increases the energy barrier for Li-Ni cation exchange, thereby suppressing Li+/Ni2+ mixing. In contrast, W incorporation lowers this barrier, promoting Li+/Ni2+ mixing at the surface. This disorder may facilitate heterophase reconstruction by promoting the formation of a spinel phase on the surface33.

The lattice structures of LNO and LNAW particles were investigated at the atomic scale using high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM). As seen in Fig. 2a, b, and Supplementary Fig. 8, the particle size of LNAW (400 ~ 500 nm) is smaller than that of LNO (1 ~ 2 μm), which can be attributed to the lowered surface energy resulting from W6+ doping. This observation is further supported by SEM images (Supplementary Figs. 9,10), where W clearly inhibits excessive LNAW particle growth. Such a phenomenon has been widely observed in hexagonal LiCoO2 material34,35. The energy-dispersive spectroscopy (EDS) mappings (Fig. 2c, Supplementary Fig. 11 and line scanning (Supplementary Fig. 12) of LNAW reveal that Al, Ni, and O are homogeneously distributed across the crystalline bulk regions of the LNAW particles. In contrast, W is primarily accumulated at the particle surfaces and grain boundaries. The atomic structures of LNO and LNAW were analyzed using HAADF-STEM to validate the structural design of HPR. Figure 2d reveals that the LNO sample exhibits a well-defined layered structure. In contrast, LNAW exhibits a distinct ~2 nm surface spinel layer atop a layered bulk structure (Fig. 2e). This was further confirmed by corresponding Fast Fourier Transform (FFT) patterns. According to our DFT calculations, W preferentially substitutes for Ni in surface layers, promoting local Li+/Ni2+ mixing and initiating the formation of a thin spinel phase36. Figure 2f presents the atomic-scale EDS spectrum of the LNAW surface, confirming W enrichment at the surface and Al incorporation throughout the bulk lattice.

Fig. 2: Lattice and surface structure of LNAW.
Fig. 2: Lattice and surface structure of LNAW.The alternative text for this image may have been generated using AI.
Full size image

a,b STEM image of LNO and LNAW samples. c LNAW corresponding EDS mappings of Ni, O, Al and W elements. HAADF-STEM images and corresponding atomic structure of the phase boundaries observed in d LNO and e LNAW materials. FFTs from marked regions are also presented. f Line-profile analysis utilizing STEM-EDS.

Electrochemical properties

The electrochemical performance of LNO and LNAW was evaluated in sulfide-based ASSLIBs. In which, LPSC serves as SE, with ionic conductivity of 3.9 ms cm-1 (Supplementary Fig. 13). Figure 3a and Supplementary Fig. 14 show the initial charge-discharge curves of LNO and LNAW within 2.4–4.3 V vs. Li/Li+. LNO delivers an initial capacity of 173.9 mAh g-1 at 0.1 C, while LNAW achieves a significantly higher capacity of 206.3 mAh g-1. The Supplementary Fig. 15 shows the cycling stability of LNO, LiNi0.994W0.006O2 (LNW), LNA, and LNAW in sulfide-based ASSLIBs at 0.1 C. The initial discharge capacities follow the order LNAW > LNW > LNO > LNA. This is mainly because the interfacially compatible LNW enables deeper delithiation (higher charge state), thereby achieving a higher initial capacity. However, the deeper Li (de)intercalation induces cumulative lattice strain and mechanical fatigue in the bulk structure during cycling, leading to the rapid capacity degradation of LNW. By contrast, LNA sacrifices some initial capacity but effectively strengthens structural stability, achieving a high capacity retention of 85% after 80 cycles. This result reveals that the distinct dominant roles of the interfacial compatibility and bulk structural stability in determining the electrochemical performance of LNO-based ASSLIBs; the former is the prerequisite for achieving high initial capacity, while structural/mechanical stability is the primary cause of long-term cycling. In this work, the synergistic effect arisen from HPR simultaneously optimizes interfacial compatibility and bulk structural stability, achieving an improved balance between capacity and stability. Furthermore, the CV curves in Fig. 3b and Supplementary Fig. 16 show that the H2-H3 phase transition is most effectively suppressed in LNAW, and its reversibility is significantly improved.

Fig. 3: Electrochemical performance of Li-In | LPSCl | LNAW and Li-In | LPSCl | LNO at 27 °C.
Fig. 3: Electrochemical performance of Li-In | LPSCl | LNAW and Li-In | LPSCl | LNO at 27 °C.The alternative text for this image may have been generated using AI.
Full size image

a Initial charge-discharge curves of LNO and LNAW at 0.1 C. b Cyclic voltammetry curves of LNAW at 0.05 mV s-1. c,d Comparison of the rate capabilities of LNO and LNAW. e Long-term cycling performance of LNO and LNAW at 0.5 C rate. f Corresponding discharge curves of (e). g Long-term cycling performance of LNAW under cut-off voltage of 4.5 V. h Rate performance of LNAW under cut-off voltage of 4.5 V. i Rate performances of reported LNO in ASSLIBs. j Cycling stabilities and capacities after cycling at 0.5 C of reported LNO in ASSLIBs.

Figure 3c, d compare the rate capabilities of LNO and LNAW in sulfide-based ASSLIBs from 0.2 to 10 C, with corresponding discharge curves shown in Supplementary Fig. 17. LNAW delivered higher reversible capacities than LNO at all tested specific currents. Even at the high C-rates (2, 5, and 10 C), the LNAW delivered reversible capacities of 136.0, 94.2, and 56.7 mAh g-1, exceeding those of LNO (73.9, 41.4, and 16.7 mAh g-1, respectively). The long-term cycling performance of LNO and LNAW in sulfide-based ASSLIBs was further assessed, as shown in Fig. 3e, f. The LNO showed a low capacity of 115.8 mAh g-1 at 0.5 C and rapid capacity degradation with 55.8% retention capacity after 200 cycles. In sharp contrast, the LNAW exhibited significantly improved stability with a high reversible capacity of 173.9 mAh g-1 and a high retention of 84% after 200 cycles. Supplementary Fig. 18 shows that LNAW maintains 81.1% capacity retention after 200 cycles at 1 C, further highlighting its good electrochemical stability under practical operating conditions. To validate the effectiveness of the HPR strategy in enhancing both structural and interfacial stability of LNAW, the cut-off voltage of LNAW was elevated to 4.5 V in sulfide-based ASSLIBs. As shown in Fig. 3g and Supplementary Figs. 19, 20, the LNAW delivered a discharge capacity of 187.6 mAh g-1 at 0.5 C and sustained long-term cycling over 700 cycles, retaining 65% of the initial capacity. Figure 3h and Supplementary Fig. 21 further demonstrate the high-rate performance of LNAW, with capacities of 132.5 and 85.8 mAh g-1 at 2 C and 5 C, respectively. The achieved performance can be attributed to several critical factors. Firstly, the W-enriched surface spinel phase constructs a stable LNAW|LPSCl interface that mitigates interfacial side reactions while facilitating favorable ion transport kinetics. Secondly, the integrated surface-to-bulk spinel-layered lattice framework further strengthens the bulk crystal structure, maintaining its dynamic reversibility even at a highly delithiated state. Figure 3i, j and Supplementary Table 5 provide a comprehensive comparison of electrochemical performance relative to previously reported LNO-based ASSLIBs. As a result, LNAW demonstrates high rate capability and long-term cycling stability under high voltage. These data indicated that LNAW achieves performance levels comparable to those reported in the literature, underscoring the effectiveness of the HPR design.

To elucidate the origin of the pronounced performance difference between LNO and LNAW, we first compared the intrinsic ionic and electronic conductivities of these pristine materials. Continuous and interconnected pathways for both ionic and electronic transport are essential for achieving high-capacity solid-state composite positive electrodes, as any limitation in either channel inevitably degrades the overall electrochemical performance. As shown in Supplementary Fig. 22 and Supplementary Tables 6,7, both LNW and LNAW exhibit enhanced electronic and ionic conductivities compared with LNO and LNA, suggesting that the W-enriched spinel phase not only facilitates electron transport within the composite positive electrode but also promotes Li+ migration, thereby improving overall reaction kinetics.

Subsequently, in situ galvanostatic electrochemical impedance spectroscopy (GEIS) measurements combined with distribution of relaxation time (DRT) analysis were conducted to further elucidate the Li+ transport kinetics and the interfacial stability at the LNAW|LPSCl interface37. Supplementary Fig. 23 presents the initial-cycle GEIS spectra of LNO and LNAW, where two main regions were identified: Rct representing the charge transfer resistance at the interface and Rd corresponding to solid-phase diffusion within the particles38. It is noted that in ASSLIBs, due to limited solid-solid contact paths and restricted ion migration, diffusion tends toward finite-length or quasi-diffusion behavior. In such cases, the Nyquist plot may not show a visually recognizable Warburg tail. As shown in Supplementary Fig. 23c, Rct at the LNO|LPSCl interface increases continuously during the initial charge and discharge process, particularly after the H2-H3 phase transition at 4.2 V (Point 3), while Rd remained nearly unchanged. This suggests that the sharp increase in interfacial resistance is the main factor hindering ion transport kinetics in LNO during initial cycling. In contrast, the Rct of LNAW|LPSCl interface remained lower, demonstrating that the HPR strategy effectively stabilizes interfacial ion transport pathways and suppresses resistance buildup.

Supplementary Fig. 24 tracked the impedance evolution of the two composite positive electrodes throughout long-term cycling, showing that the impedance of LNAW composite positive electrode was maintained at 400 Ω, whereas the impedance of LNO composite positive electrode sharply increased to 900 Ω after 200 cycles. Additionally, CV measurements and galvanostatic intermittent titration technique (GITT) measurements (Supplementary Figs. 25, 26) reveal that LNAW exhibits a higher Li+ diffusion coefficient than LNO, indicating enhanced reaction kinetics and high rate capabilities. The impedance and kinetic analyses confirm that LNAW possesses significantly improved interface compatibility with LPSCl, offering stable ion transport and minimized resistance over long cycling.

Interfacial compatibility between LNAW and LPSCl

To investigate the interfacial compatibility including chemical and electrochemical stability between LNAW and LPSCl, we further investigated LNO and LNAW composite positive electrodes at different charge/discharge states using soft XAS (sXAS) and XPS. Figure 4a, b show the Ni L-edge total electron yield (TEY) sXAS spectra at various charge/discharge states. The Ni valence-state change can be qualitatively assessed by deconvolving the high-energy (L3, high) and low-energy (L3, low) states, where the L3, high/L3, low intensity ratio is positively correlated with the Ni valence state39. At open-circuit voltage (OCV), a distinct decrease in the L3, high/L3, low ratio was observed for LNO compared to its pristine state, indicating a reduction in Ni valence due to inherent chemical incompatibility with LPSCl. This irreversible Ni reduction arises from interfacial side reactions5,40. In contrast, LNAW maintained a nearly unchanged L3, high/L3, low ratio at OCV, attributed to the protection from W-enriched surface spinel phase.

Fig. 4: Interfacial compatibility between LNAW and LPSCl.
Fig. 4: Interfacial compatibility between LNAW and LPSCl.The alternative text for this image may have been generated using AI.
Full size image

a TEY Soft XAS spectra of Ni L-edge. Normalization was performed using standard procedures in QANT software. b Corresponding Ni L3, high/L3, low ratios of LNO and LNAW composite positive electrode at states of pristine, OCV, after 1st cycle and after 200 cycles at 4.3 V. c FT-EXAFS of Ni of LNO and LNAW at pristine state and after 200 cycles at 4.3 V. S 2p XPS spectra for d LNO, e LNAW, f LNW and g LNA composite positive electrodes after the 1st cycle and long-term cycled at 4.3 V. All cells were disassembled after 200 cycles at 0.5 C under 4.3 V for testing.

To assess electrochemical stability, Ni L2, 3-edge sXAS was performed on cycled LNO and LNAW. After the first cycle, LNO exhibited an increased L3, high/L3, low ratio compared to the OCV state, indicating an increase of Ni oxidation state. Ideally, the Ni valence should revert to its OCV state at the end of a complete charge-discharge cycle. However, in LNO, the increase in Ni valence suggests incomplete reduction, which can be attributed to low coulombic efficiency and particle cracking-induced interfacial contact loss, limiting the full reduction of Ni ions. During subsequent cycling, the continuous accumulation of interfacial side products and progressive structural and interfacial degradation lead to an overall decrease in Ni valence. The non-monotonic evolution of the L3, high/L3, low ratio during cycling is attributed to the competing effects of interfacial side reactions, irreversible Li consumption, and local structural changes, which collectively modulate the Ni valence at different cycling stages. In contrast, LNAW maintained a nearly identical Ni valence state compared to that at OCV, highlighting its electrochemical stability with the LPSCl. This suggests that the structural and interfacial integrity of LNAW is preserved throughout the cycling process. After 200 cycles, LNAW retained a largely unchanged Ni valence state, indicating minimal degradation and no accumulation of interfacial by-products. These findings are further supported by Fourier-transformed extended X-ray absorption fine structure (FT-EXAFS) analysis, as shown in Fig. 4c and Supplementary Fig. 27. The Ni−O and Ni−Ni coordination distances in LNO shifted after prolonged cycling, reflecting structural degradation and poor interfacial stability. In contrast, the corresponding interatomic distances in cycled LNAW showed negligible change, confirming the preservation of stable Ni local coordination environments during extended cycling.

To further identify the interfacial byproducts after cycling, surface-sensitive XPS was performed. Figure 4d, e and Supplementary Fig. 28 present the S 2p XPS spectra of the LNO and LNAW composite positive electrodes after the 1st and long-term cycles. In Fig. 4d, the pristine LPSCl electrolyte exhibits a characteristic S 2p doublet with the S 2p3/2 peak at 161.7 eV (red component) attributed to sulfur in the argyrodite phase41. After the 1st cycle, a noticeable decrease in the intensity of the argyrodite LPSCl peaks (red component) in the LNO composite positive electrode indicates the onset of side reactions. These reactions lead to the formation of electrochemically inactive byproducts polysulfide (P2Sx) and sulfite (SO32-), which are identified by the appearance of two additional peaks at 163.5 eV (blue component) and 166.7 eV (yellow component), respectively24,25. These byproducts are known to possess high resistivity and poor Li+ conductivity, thereby impeding interfacial ion transport and degrading electrochemical performance. After 200 cycles, the intensities of these byproduct peaks exceed that of the original LPSCl signal, confirming the continuous side reactions at the LNO|LPSCl interface. In contrast, as shown in Fig. 4e, the main argyrodite peak remains dominant even after 200 cycles, with only minimal contributions from polysulfide and sulfite byproducts. These results indicate that the surface-to-bulk HPR strategy effectively suppresses LPSCl decomposition and enhances interfacial compatibility. In addition, to decouple the individual effects of Al and W, the LNA and LNW composite positive electrodes were investigated. As shown in Fig. 4f, g, the cycled LNW|LPSCl composite positive electrode exhibits significantly reduced decomposition products compared with LNA|LPSCl composite positive electrode, confirming that the W-enriched spinel surface effectively mitigates SE degradation and improves interfacial compatibility.

Structural and mechanical stability of LNAW

To further correlate the interfacial stability with the bulk structural evolution, SXRD was performed to quantitatively assess the impact of the H2-H3 phase transition on the structural stability of LNO, LNA, LNW and LNAW. As shown in Fig. 5a, b, the shift of the (003) peak serves as an indicator of changes in the c-axis lattice parameter (Δc) with quantitative values derived from SXRD data (Fig. 5c, Supplementary Fig. 29 and Supplementary Table 8)42. As shown in Fig. 5a, a 0.04° shift in the (003) peak was observed after the first discharge compared with OCV, corresponding to a significant 0.6% increase of Δc. After cycling, the peak shift increased to 0.06°, indicating continuous structural degradation (Δc~8%). In comparison, after the first cycle, the Δc of LNA and LNW are approximately 0.26% and 0.32%, respectively, while the (003) peak of LNAW almost returned to its original 2θ position due to its good phase-transition reversibility. After prolonged cycling, LNAW still maintains the smallest Δc (0.26%), the Δc of LNA is 0.54%, whereas the Δc of LNW is up to 0.86%, approaching the degradation level of LNO (0.8%). These results clearly indicate that despite LNW exhibiting improved interfacial stability, its deeper charge states induce severe lattice distortion during cycling, accelerating bulk structural fatigue. In contrast, LNA partially suppresses lattice expansion and delays structure degradation, while the stable layered-spinel lattice skeleton of LNAW further alleviates lattice strain during long-term cycling and effectively suppresses cumulative structural degradation. FIB-SEM was performed on the composite positive electrodes to reveal mechanical degradation. As shown in Fig. 5d, e, LNO exhibits pronounced microcracks and interfacial contact loss due to severe H2-H3 phase transitions. These detached fragments lose mechanical contact and can no longer participate in the electrochemical reaction, resulting in rapid capacity decay. Severe microcracks and particle fragmentation are also obvious in the LNW electrode (Fig. 5f, g), whereas they are partially suppressed in the LNA electrode (Fig. 5h, i), indicating that LNA contributes to enhanced mechanical stability. LNAW exhibited high morphological integrity with no observations of microcracks due to the suppression of H2-H3 phase transition (Fig. 5j, k). Moreover, even at 4.5 V high voltage, the additional FIB-SEM analyses (Supplementary Figs. 30, 31) further demonstrate that LNAW maintains good mechanical stability even after long-term cycling.

Fig. 5: Structure and morphology evolution of LNAW.
Fig. 5: Structure and morphology evolution of LNAW.The alternative text for this image may have been generated using AI.
Full size image

The selected SXRD patterns of the H2-H3 transformation region for cycled a LNO and b LNAW electrodes. c The variation of the c-axis parameter in pristine and cycled four electrodes. Cross-sectional FIB-SEM images of cycled d, e LNO; f, g LNW; h, I LNA; and j, k LNAW composite positive electrodes. Atomic resolution HAADF-STEM images of cycled l, LNO and m, LNAW samples. All cells were disassembled after 200 cycles at 0.5 C under 4.3 V for testing.

Combined with the aforementioned XPS interfacial analysis, this further confirms that interfacial stability is a prerequisite for achieving high initial capacity, while structural/mechanical stability is crucial for long-term cycling. Interfacial chemical stability and bulk mechanical integrity are interdependent, and only HPR with a W-enriched surface spinel phase and Al-enriched bulk layered phase can simultaneously optimize both capacity and cycling stability, further highlighting the critical significance of HPR design in composite positive electrodes in ASSLIBs (Supplementary Fig. 32). The atomic-scale structural evolution of LNO and LNAW was examined using HAADF-STEM. As shown in Fig. 5l and Supplementary Fig. 33, the initially ordered surface lattice of cycled LNO exhibited significant distortion, with the formation of a rock-salt phase extending to a depth of 5 nm. This indicates that prolonged cycling induced undesirable side reactions, leading to charge inhomogeneity and hindering lithium-ion and electron transport within the electrode. In contrast, LNAW showed significantly less rock-salt formation, effectively suppressing surface structural damage while preserving the integrity of the bulk layered framework (Fig. 5m). These findings underscore the effectiveness of the HPR design in mitigating structural degradation and enhancing microstructural stability under long-term cycling.

General design principles for HPR

To verify the universality of the HPR strategy, DFT calculations were conducted on representative low-valence inactive ions (Al3+, B3+) and high-valence ions (Nb5+, Mo6+, W6+) to evaluate migration barriers and Li/Ni exchange energies. The results in Supplementary Fig. 34 reveal that the migration barriers increase with the oxidation state, suggesting that high-valence ions are energetically unfavorable for bulk diffusion and tend to accumulate on the surface, while low-valence ions are more likely to diffuse into the bulk phase. Meanwhile, low-valence dopants exhibit higher Li/Ni exchange energies, effectively suppressing cation mixing, while high-valence dopants are more prone to induce local surface reconstruction (Supplementary Fig. 35). Based on these insights, we propose a general HPR design principle: coupling low-valence bulk-stabilizing ions with high-valence surface-regulating ions achieves multiscale stabilization from bulk to interface. Specifically, inactive dopant (Al3+/B3+) diffuses into the bulk phase to enhance structural integrity and mitigate bulk fatigue, whereas Nb5+/Mo6+/W6+ segregates at the surface to induce reconstruction layers that suppress SE decomposition and interfacial side reactions. To validate this concept experimentally, three additional systems (Al/Mo, B/W, and B/Nb) were synthesized with molar ratios (~0.03:0.006). SXRD analyses (Supplementary Figs. 36–38) confirmed the designed phase structure, and electrochemical results (Supplementary Fig. 39) demonstrated enhanced discharge capacity and cycling stability due to the synergistic stabilization of surface and bulk regions. In summary, the HPR strategy establishes a generalized design guideline for Ni-rich layered oxides in ASSLIBs: low-valence dopants strengthen the layered lattice and suppress cation disorder, while high-valence dopants induce controllable surface reconstruction that improves interfacial compatibility. The synergistic regulation of bulk and surface stability enables simultaneous improvement of structural robustness and interfacial kinetics, providing a practical pathway toward high-capacity and long-life Ni-rich positive electrode materials.

Discussion

In summary, this work elucidates the coupled interfacial and bulk failure mechanisms that fundamentally limit the electrochemical stability of LNO-based ASSLIB. We propose a universal HPR strategy that simultaneously stabilizes both regions through the valence- and diffusion-governed distribution of dopant ions. In the representative LNAW material, low-valence Al diffuses into the bulk to reinforce the layered lattice and suppress cation mixing, while high-valence W enriches at the surface to form a protective spinel phase that enhances interfacial compatibility and Li+ transport. Benefiting from the synergy of W and Al, a stable layered-spinel framework is constructed, which effectively eliminates lattice strain, suppresses the formation of microcracks, and significantly improves chemical and mechanical stability. As a result, under a high cut-off voltage of 4.5 V, LNAW exhibits a high initial capacity of 187.6 mAh g-1 at 0.5 C and retains 65% of its capacity after 720 cycles. More importantly, the HPR concept, governed by the interplay of dopant valence and diffusion kinetics, has been validated in multiple systems (LiNi0.964Al0.03Mo0.006O2, LiNi0.964B0.03W0.006O2, and LiNi0.964B0.03Nb0.006O2), demonstrating its universality. This work offers insights and a viable path forward for realizing long-life, high-energy ASSLIBs.

Methods

Synthesis of positive electrode materials

This section describes the synthesis of LiNiO2 (LNO), LiNi0.97Al0.03O2 (LNA), LiNi0.994W0.006O2 (LNW), LiNi0.964Al0.03W0.006O2 (LNAW), LiNi0.964Al0.03Mo0.006O2 (LNAMo), LiNi0.964B0.03W0.006O2 (LNBW), and LiNi0.964B0.03Nb0.006O2 (LNBNb). All materials were synthesized by a facile one-step solid-state method using LiOH (Sigma-Aldrich, 99.9%), Ni(OH)2 (lab-scale preparation) as raw materials and Al(NO3)3·9H2O (Sigma-Aldrich, ≥ 98%) as Al source, WO3 (Sigma-Aldrich, 99.9%) as W source, B2O3 (Sigma-Aldrich, 99.99%) as B source, Nb2O5 (Sigma-Aldrich, 99.99%) as Nb source, MoO3 (Sigma-Aldrich, 99.97%) as Mo source. For each material, stoichiometric amounts of Ni(OH)2, LiOH (5% excess), Al(NO3)3·9H2O, WO3, B2O3, Nb2O5 and MoO3 were mixed using an agate mortar or a planetary ball-milling (FRITSCH, pulverisette 7) at 300 rpm for 1 h with a ball-to-powder weight ratio of 10:1 using 5 mm ZrO2 balls. The resulting powder mixtures were then subjected to a multi-step calcination process under an O2 atmosphere at 400 °C for 3h, 690 °C for 3 h, and 900 °C for 1 h, with a heating rate of 5 °C/min. After naturally cooling to 27 °C, four materials were obtained, respectively.

Materials characterization

The SXRD data (with a wavelength λ = 0.5903 Å), the synchrotron-based NEXAFS of Ni L-edge (data transformation directly processed by Athena software), and the synchrotron-based EXAFS of Ni K-edge were performed on beamlines at the Australian Synchrotron (Clayton), part of ANSTO. The HADDF-STEM analysis was conducted using a FEI Titan G2 80–300 microscope at 300 kV equipped with a probe corrector. The ICP-OES analysis was conducted using an Agilent 5110 instrument. Powder samples were digested in acid, diluted to a fixed volume, and subsequently analyzed. The XPS measurements were performed on a Thermo ESCALAB 250 system with a monochromatic Al-Kα (1486.6 eV) X-ray source (energy step size 0.1 eV) to investigate the relative content and the chemical state of elements. The XPS data were fitted with the asymmetric Gaussian-Lorentzian sum function in Advantage software. Direct current (DC) polarization technique: 55 mg of LNO, LNW, and LNAW powders were compacted in a 10 mm PEEK all-solid-state battery mold under a pressure of 360 MPa for 1 min. Stainless-steel rods (part of the PEEK all-solid-state battery mold) were used as blocking electrodes. The DC polarization measurement was performed at 27 °C under a constant voltage of 10 mV for 0.8 h using an EC-Lab workstation. The pellet thickness was measured after testing to calculate the conductivity. Ionic conductivity measurement: Ionic conductivity of the as-synthesized SE was probed via electrochemical impedance spectroscopy. Samples were prepared by densifying 160 mg of SE in a 10 mm diameter PEEK all-solid-state battery mold by 360 MPa for 3 min. With stainless steel rods as contacts, the measurement was performed in a frequency range from 1 MHz to 1 mHz with an excitation amplitude of 10 mV. Symmetric cells were prepared by pressing 80 mg Li3InCl6 (LIC) and 30 mg positive electrode material under 320 MPa. The electron-blocking configuration was LIC | LNO | LIC.

Electrochemical characterizations

Laboratory-scale all-solid-state batteries (10 mm inner diameter) were fabricated using composite positive electrodes composed of the synthesized materials, LPSCl (≤5 μm, Ganfeng Lithium Co., Ltd), and vapor-grown carbon fiber (VGCF, Resonac China) as the conductive additive. The composite was prepared by mixing the active material, LPSCl, and VGCF in a weight ratio of 70:27:3 using an agate mortar for 1 h. All assembly processes were conducted in an argon-filled glove box. A total of 100 mg of LPSCl SE was used as the separator layer. First, a dense SE pellet was formed by pressing 100 mg of LPSCl under 240 MPa for 60 s. Then, 5–7 mg of the prepared composite powder was evenly spread on one side of the SE pellet and pressed under 360 MPa for 10 s to form the positive electrode layer. On the opposite side of the SE pellet, a 50 μm thick indium foil (10 mm diameter) was placed. A 30 μm thick lithium foil (8 mm diameter) was then attached to the indium to serve as the Li source. The full cell assembly was encapsulated in a stainless-steel plate case and subjected to a constant pressure of ~20 MPa to ensure interfacial contact. For all post-cycling measurements, the cells were disassembled in an argon-filled glove box, the positive electrodes were collected, and transferred for XPS/FIB-SEM/XRD analysis without air exposure. Galvanostatic charge-discharge tests and cycling performance of assembled ASSLIBs were evaluated at 27  °C using a standard battery testing instrument (Neware battery test system, CT-4008T-5V50mA-164, Shenzhen, China). All tests were performed at 1.8 ~ 3.7 V and 1.8 ~ 3.9 V vs. Li-In. The C-rate of 1 C corresponds to 200 mA g-1. The GEIS was performed by applying an alternating current voltage of 10 mV over the frequency range of 1 mHz to 1 MHz.

Calculation of diffusion coefficient

The chemical diffusion process was analyzed using GITT, where the relaxation behavior in positive electrode materials arises from Li+ diffusion. The diffusion coefficient was determined using the following equations43:

$${D}_{{{{\rm{GITT}}}}}=\frac{4}{\uppi \uptau }{\left(\frac{{{{m}}}_{{{{\rm{B}}}}}{V}_{{{{\rm{M}}}}}}{{M}_{{{{\rm{R}}}}}{{{\rm{S}}}}}\right)}^{2}{\left(\frac{\Delta {E}_{s}}{\Delta {E}_{\tau }}\right)}^{2}$$
(1)

where VM is the molar volume, MR is the relative formula mass, mB is the active mass of the electrode, S is the contact area between electrolyte and sample.

Computational details

All calculations in this work were performed using the VASP software package44,45. The electronic interactions were described by the plane wave method generalized gradient approximation (GGA) of the Perdew-Burke-Ernzerhof (PBE) functional46,47. The cutoff energy for the calculations was set to 550 eV. In all calculations, a 3 × 3 × 1 supercell was used, and the Monkhorst-Pack (MP) method was applied to generate a 3 × 3 × 1 k-point mesh48. To accurately simulate the localized 3 d electron orbitals of Ni, the GGA + U method was adopted, with a U value of 6.2 eV for Ni49. Additionally, the D3 van der Waals correction was included to accurately describe the interlayer interactions50. Spin polarization was also considered. For convergence criteria, the energy was converged to 10–5 eV, and the force was converged to 0.01 eV/Å to ensure precision in the results. To investigate the diffusion energy barrier and diffusion path for Al and W, the climbing image nudged elastic band (Cl-NEB) method was used, with five images and a force convergence criterion of 0.02 eV/Å51.