P-NCA CAMs with a wide range of Ni contents were prepared to comprehensively investigate the mechanism responsible for capacity fading in ASSBs that feature Ni-rich cathodes, which include Li[Ni0.80Co0.17Al0.03]O2 (P-Ni80), Li[Ni0.85Co0.13Al0.02]O2 (P-Ni85), Li[Ni0.89Co0.10Al0.01]O2 (P-Ni90), and Li[Ni0.95Co0.04Al0.01]O2 (P-Ni95)). The chemical compositions of the CAMs were determined by inductively coupled plasma optical emission spectroscopy (ICP-OES) (Supplementary Table 1), which revealed that the Ni, Co, and Al fractions in the P-NCA CAMs are consistent with the expected compositions. X-ray diffractometry (XRD) revealed that all CAMs have hexagonal α-NaFeO2-type layered structures within the R\(\stackrel{-}{3}\)m space group (Supplementary Fig. 1), and scanning electron microscopy (SEM) showed that the prepared CAM particles are spherical, monodispersed, and have an average diameter of 10 µm (Supplementary Fig. 2).
S-NCA (S-Ni80, S-Ni85, S-Ni90, and S-Ni95) CAMs were prepared by coating 0.5 wt% boric acid onto the P-NCA CAMs to minimize degradation caused by surface reactions at the CAM/SE interface15. The time-of-flight secondary-ion mass-spectrometry (TOF-SIMS) depth-profiles presented in Fig. 1a reveal the presence of boron compounds (represented as BO− and LiBO3−) resulting from the reaction of boric acid with residual lithium during heat treatment of S-NCA CAMs. Furthermore, the 3D-rendered TOF-SIMS images of the S-Ni95 CAM (Fig. 1b,c) reveal that boron compounds are mainly present on the particle surface and less distributed in its interior. The boron compounds coated on the CAM surface provide a physical shielding barrier that prevents direct contact between the CAM and the SE, thereby suppressing parasitic side reactions at the interface; it also acts as a buffer layer that promotes Li diffusion across the space-charge layer, thereby minimizing interfacial issues that occur in an ASSB16–18.
M-NCA (M-Ni80, M-Ni85, M-Ni90, and M-Ni95) CAMs were prepared by doping with 1 mol% Nb during the lithiation process to minimize capacity fading due to the isolation of interior primary particles associated with microcracking, which limits their participation in electrochemical reactions16. Additionally, SM-NCA (SM-Ni80, SM-Ni85, SM-Ni90, and SM-Ni95) CAMs were prepared by coating M-NCA CAMs with 0.5 wt% boric acid. High-valence elements, such as Nb, effectively suppress particle coarsening during lithiation and facilitate the development of a microstructure consisting of radially oriented rod-shaped primary particles19, which is well known to suppress microcrack formation in CAM particles by effectively dissipating the localized strain caused by anisotropic lattice volume changes during charging and discharging20. Cross-sectional SEM was used to highlight the differences in the primary particle of the morphology modified CAM particles (Fig. 1d and Supplementary Fig. 3). The S-NCA CAM particles are composed of polygonal primary particles that decrease in size with increasing Ni content, which is ascribable to the optimum CAM lithiation temperature decreasing with increasing Ni content1. In contrast, the SM-NCA CAM particles consist of tightly packed small but long rod-shaped primary particles that are radially aligned at all Ni contents. The CAM microstructures were quantified by measuring the widths and lengths of the primary particles at the periphery of each secondary particle (Fig. 1e and Supplementary Fig. 3). The primary particles of the P- and S-NCA CAMs exhibit a wide size distribution, with widths that range from 100 to 800 nm and lengths in the 200–1500 nm range. Meanwhile, those of the M- and SM-NCA CAMs, whose primary particles were modified into fine rod shapes by Nb doping, were homogeneously distributed with a narrow size distribution, with widths within 300 nm and lengths of 400–1000 nm. Therefore, the primary particles in the P- and S-NCA CAMs have low average aspect ratios (approximately 2), whereas those of the primary particles in the M- and SM-NCA CAMs are high (approximately 5). Furthermore, the angles between the longitudinal axes of the primary particles and the diametric lines passing through the primary-particle centers were measured to quantify the orientations of the primary particles (Fig. 1f,g). The SM-NCA CAMs (Fig. 1g) contain primary particles that are more radially oriented than those in the S-NCA CAMs (Fig. 1f). These quantitative data confirm that Nb doping modifies the CAM microstructure, resulting in primary particle morphologies that are radially aligned and elongated20.
The electrochemical properties of the prepared CAMs, such as their initial charge–discharge voltage profiles and cycling stabilities, were characterized in ASSBs with an argyrodite Li6PS5Cl (LPSCl) SE. Supplementary Fig. 4a-d display initial charge–discharge voltage profiles for the P-NCA, S-NCA, M-NCA, and SM-NCA cathodes with Ni contents of 80–95%, respectively, cycled at a constant current density of 0.1 C (18 mA g− 1). Among these cathodes, the P-NCA series exhibited the lowest reversible capacities and poor Coulombic efficiencies across the range of Ni contents. The S-NCA cathodes exhibited slightly higher Coulombic efficiencies than the P-NCA cathodes, but it still remained around 80%. In contrast, the M-NCA and SM-NCA cathodes exhibited much higher Coulombic efficiencies and initial discharge capacities than the P-NCA and S-NCA cathodes at the same Ni content, which suggests that a modified radially oriented primary particle morphology facilitates Li+ diffusion in an ASSB21. The SM-NCA cathodes exhibited the highest Coulombic efficiencies and initial capacities compared to others with the same Ni content owing to the synergy of surface and morphology modification.
The prepared cathodes were cycled at a constant current density of 0.5 C (90 mA g− 1) for 100 cycles, the results of which are shown in Fig. 2a-d and Supplementary Fig. 4e-h. The P-NCA cathodes experienced rapid capacity deterioration upon cycling at all Ni contents, which was increasingly aggravated with increasing Ni content. After being modified, both the S-NCA and M-NCA cathodes exhibited improved initial capacities and cycling stabilities that resulted in the SM-NCA cathodes exhibiting synergy. However, the surface and morphological modifications affected cycling stability in a manner that depended on the Ni content of the CAM. The S-Ni80 cathode exhibited a higher capacity retention than the M-Ni80 cathode after 100 cycles (Fig. 2a). In contrast, the M-NCA cathodes showed higher capacity retentions than the S-NCA cathodes at Ni contents above 85% (Fig. 2b-d), which suggests that the capacity fading mechanism depends on the Ni content of the CAM, which is discussed in detail later.
The capacity retentions of the various cathodes are plotted as functions of the initial discharge capacity at 0.1 C to summarize their electrochemical performance (Fig. 2e). Increasing the Ni content in the P-NCA CAM tends to increase the initial capacity but decrease cycling stability, as observed for liquid electrolyte systems22. After surface coating, the S-NCA cathodes delivered higher initial capacities and capacity retentions than the P-NCA cathodes, which demonstrates that the surface coating effectively improves lithiation/delithiation reversibility by suppressing parasitic reactions at the CAM/electrolyte interface23–25. After modifying the primary particle morphology, the M-NCA cathodes exhibited higher initial capacities and enhanced cycling stabilities than the P-NCA and S-NCA cathodes, especially at high Ni contents, which demonstrates that the primary particles, which are radially aligned, long, and rod-shaped, facilitate the diffusion of Li+ from the surface to the interior, thereby effectively enhancing cycling stability21. Furthermore, the SM-NCA cathodes exhibited superior electrochemical performance compared to other cathodes at the same Ni content owing to synergy between the stabilized CAM-SE interface and the well-designed primary particle morphology. However, the extent to which surface and morphological modifications improve cycling performance depends on the Ni content of the CAM. Therefore, how each modification contributes to cell-performance enhancement needs to be quantified by comparing CAMs with and without each modification while considering their Ni contents.
Sulfide-based SEs have a narrow (1.71–2.31 V vs. Li/Li+) electrochemical stability window that does not match the operating voltages of Ni-rich cathodes, resulting in SE decomposition that hinders Li+ diffusion10. The capacity fading factor associated with surface reactions occurring at the CAM/electrolyte interface in an ASSB was investigated. The cycled P-NCA and S-NCA cathodes tested in Fig. 2 were disassembled and analyzed by XPS to gain insight into the changes in chemical bonding and decomposition products in the cathode after cycling23,26–28. The S 2p XPS spectra of the cycled cathodes were deconvoluted into four doublets (Fig. 3a); the trivial doublet (161.4/160.2 eV) corresponds to the Li2S impurity, the primary doublet (162.7/161.5 eV) corresponds to the PS43− tetrahedral unit in LPSCl, and the minor doublets observed at 163.8/162.6 and 164.7/163.5 eV correspond to P–[S]n–P (n = 1 or 2) and –S0–, respectively23,26. The P–[S]n–P-type bonds found in structures such as P2S74− (n = 1) and P–S–S–P (n = 2) are formed through the oxidative polymerization of the anionic units in LPSCl27,28, while the –S0– bond at higher binding energy is found in the homomonocyclic octasulfur compound, which is a thermodynamically stable oxidation product of LPSCl27,29. The degree of SE decomposition can be determined by comparing the combined area of the doublets originating from these P–[S]n–P-type and–S0– bonds with that of the primary doublet corresponding to the PS43− tetrahedral unit in LPSCl (Fig. 3d). This relative areal ratio was observed to increase with increasing Ni content in the CAM of the cycled P-NCA cathode, which demonstrates that a higher fraction of unstable Ni4+ in the highly charged state accelerates the degradation of the SE at the CAM/SE interface, leading to capacity fading. In contrast, the S-NCA cathodes exhibited consistently low relative areal ratios of 20% or less, irrespective of the Ni content. Oxidative SE decomposition was suppressed during cycling because the stable protective surface coating layer on the CAM prevents direct contact between the Ni-rich CAM and the SE (Fig. 1a-c). The electrochemical-testing (Fig. 2) and XPS (Fig. 3d) results show that the surface coating, which acts as a buffer layer between the Ni-rich CAM and the SE, is a prerequisite for suppressing capacity fading in Ni-rich ASSB cathodes.
The capacity fading factor associated with inner-particle isolation was investigated by comparing morphology-modified SM-NCA CAMs with S-NCA CAMs in ASSBs, under conditions in which degradation by surface reactions is minimized through surface coating with boron compounds. Figure 3b and Supplementary Fig. 5 shows cross-sectional SEM images of S-NCA and SM-NCA CAM particles charged to 4.3 V before and after 100 cycles. The charged S-NCA particles with only surface modification exhibited noticeable cracks before and after cycling at all Ni contents tested. In-situ XRD data (Supplementary Fig. 6) for the various CAMs, acquired using liquid-electrolyte-based half-cells, reveal that the degree of anisotropic lattice change during the charging process depends on the Ni content of the CAM, irrespective of morphology or surface modification; the contraction in the c-axis lattice parameter becomes increasingly abrupt with increasing Ni content in the CAM22. The CAM particles developed more microcracks with increasing Ni content when the S-NCA cathodes in the ASSBs were charged to 4.3 V, with more cracks generated after 100 cycles owing to the accumulation of irreversible damage caused by repetitive charging and discharging. Unlike an LIB, which uses a liquid electrolyte, the microcracks in an ASSB result in permanent ionic-contact disruption leading to electrochemically inactive interior particles and capacity loss13. The extent of this electrochemical isolation was measured by dividing each microcrack area after cycling by the entire particle area as observed by cross-sectional SEM (Fig. 3b) and then plotting the results as functions of Ni content (Fig. 3e). The electrochemical-isolation fraction induced by the microcracks was observed to increase in proportion to the Ni content, highlighting the limitations of using Ni-rich CAMs in ASSBs. However, the SM-NCA CAMs showed negligible microcracks in the charged state both before and after cycling, consistent with the morphologically modified primary particles effectively suppressing electrochemical isolation. Although SM-Ni95 exhibited some visible cracks owing to its high Ni fraction (Fig. 3b), consistent with the lower capacity retention of SM-Ni95 compared to the other SM-NCA cathodes (Fig. 2e), almost zero electrochemical isolation was observed at all Ni contents. The electrochemical testing results (Fig. 2) and microcrack data (Fig. 3e) demonstrate that the morphology modification of the primary particles was effective in suppressing the capacity fading in an ASSB, especially for CAMs with Ni contents above 85%. Therefore, minimizing isolation-related capacity fading is crucial for the use of Ni-rich cathodes in ASSBs.
Severe lattice volume changes in Ni-rich CAMs during cycling not only result in the isolation of the inner CAM primary particles but also the detachment of CAM particles from the SE14. We analyzed cross-sectional SEM images of the SM-NCA cathodes after 100 cycles charged to 4.3 V to investigate the capacity fading factor related to the detachment of CAM particles from the SE. Figure 3c and Supplementary Fig. 7 shows gaps between the CAM particles and the SE resulting from CAM particles detaching from the SE due to anisotropic lattice volume contraction of CAM. The gaps and contact loss caused by these detachments disrupt Li+ conduction between CAM particles and the SE, thereby contributing to capacity fading during cycling. Furthermore, the lattice volume changes in CAM particles also affected the composite cathode as a whole, causing not only detachment but also some pores inside the SE. The relationship between degree of detachment and CAM Ni content (Fig. 3f) reveals that the cathode containing 80% Ni (SM-Ni80) exhibits a somewhat less-developed extent of detachment, with a value of 0.99% recorded. However, the extent of detachment was observed to increase with increasing Ni content and was much more severe (9.13%) for the 95%-Ni-containing cathode (SM-Ni95). In a similar manner to the in-situ XRD analysis shown in Supplementary Fig. 6, which reveals that the extent of lattice volume contraction increases with increasing Ni content, irrespective of morphology engineering, the extent of detachment was also observed to increase with increasing Ni content independently of the engineered morphology. While engineering the primary particle morphology effectively inhibits the isolation of inner CAM particles by suppressing microcrack formation, it does not contribute to suppressing the detachment of CAM particles from the SE because lattice volume change is an intrinsic CAM property. Despite the modified surface and well-designed morphology of the SM-NCA cathode, which suppresses surface reactions and microcracking, the CAM still exhibited capacity fading in proportion to the Ni content (Fig. 2e). Therefore, the detachment of CAM particles from the SE needs to be suppressed through other approaches that maintain contact between the CAM particles and SE, even during repeated lattice volume changes, to further overcome capacity fading in SM-NCA cathodes.
The investigation of capacity fading factors revealed that surface reactions (Fig. 3a,d), isolation (Fig. 3b,e), and detachment (Fig. 3c,f) are the three factors that affect capacity fading in ASSBs with Ni-rich cathodes. To quantify their contributions to degradation mechanisms of ASSBs featuring Ni-rich cathodes, cycling results of appropriately modified CAMs (Fig. 2) were compared with those of the P-NCA and S-NCA cathodes to examine the influence of surface reactions, and the S-NCA and SM-NCA cathodes to investigate the effect of inner-CAM primary particle isolations. Figure 4a shows the capacity retentions of CAMs containing 95% Ni (P-Ni95, S-Ni95, and SM-Ni95). A comparison of P-Ni95 and S-Ni95, in which the only difference is the presence or absence of the surface coating, revealed that the surface modification improved the capacity retention from 60.4–72.9%, which indicates that surface reactions affect the cycle life of the Ni95 CAM by approximately 12.5%. Similarly, comparing S-Ni95 and SM-Ni95 revealed that the effect of isolation is 13.1%, which is related to the development of microcracks, because the two CAMs only differ in the morphologies of their primary particles, as shown in Fig. 3b. Furthermore, the proportion of detachment associated with capacity fading was quantified by comparing the SM-Ni50 cathode (with a 50% Ni content) with other SM-NCA cathodes, because SM-Ni50 exhibited minimal surface reactions, isolation, and detachment owing to its low surface reactivity and lattice volume changes (Supplementary Fig. 8,9). The SM-Ni95 cathode, in which surface reactions and isolation are suppressed, exhibited a capacity retention of 86.0%; therefore, detachment appears to contribute 10.3% compared with that of the SM-Ni50 cathode (Supplementary Fig. 10).
Summing the reduced capacity retention owing to each degradation factor provides the difference in the capacity retentions of the P-NCA and SM-Ni50 cathodes. Figure 4b shows the proportion of each factor that affects the cycling life of the CAM at each Ni content. The values within each bar represent the proportion of each factor, with the difference in capacity retention between the pristine cathode and the SM-Ni50 cathode considered to be 100% (a detailed comparison is provided in Supplementary Fig. 11). For Ni 80%, the proportion of surface reaction is dominant compared to isolation and detachment, suggesting the side reactions between the CAM and SE are mainly responsible for capacity fading. However, the contributions of isolation and detachment gradually increased with increasing Ni content because the mechanical integrity of the CAM deteriorates as the extent of lattice volume variation increases. Although the occurrence of surface reactions increased slightly, their relative proportions decreased with increasing Ni content, which indicates that the mechanical properties of the CAM resulting from lattice volume changes are the dominant factor that determines battery performance for Ni contents above 85%.
Figure 5 summarizes the three factors (surface reactions, isolation, and detachment) that contribute to the capacity fading observed for the composite cathodes in ASSBs. As shown in Fig. 3a,d, increasing the Ni content of the CAM leads to more surface reactions, which can be mitigated by coating the CAM surface. Furthermore, as shown in Supplementary Fig. 6, an increase in the Ni content of the CAM led to more abrupt anisotropic lattice volume contraction during charging, resulting in the interior particles becoming electrochemically isolated owing to microcrack formation (Fig. 3b,e) and detachment of the CAM particles from the SE, resulting in contact loss (Fig. 3c,f); degradations associated with mechanical stability are severe for CAMs with Ni contents above 85% (Fig. 4b). While modifying the morphology of primary particles can suppress capacity fading caused by isolation, capacity fading caused by detachment still remains challenging. Therefore, developing an ideal composite cathode for use in ASSBs requires suppressing the detachment in the CAM by developing zero-strain CAMs that experience minimal lattice volume changes or elastic SEs that continuously maintain contact between the CAM and SE during cycling30–32.
To verify the current achievements obtained by modifying the surface and morphology of a Ni-rich CAM for ASSB applications under practical conditions, we evaluated long-term cycling performance by fabricating ASSBs using an SM-Ni90 cathode and poly(1,1,2,2-tetrafluoroethylene) (PTFE) as a dry-electrode binder. For the electrode with areal capacity of 2.42 mAh cm− 2 (the active material loading is 12.3 mg cm− 2), the ASSB with pressurizing Ti rods retained 83.3% of its initial capacity after 500 cycles (Fig. 6a). To verify the performance of SM-Ni90 under higher loading and practical conditions (areal capacity is 7.10 mAh cm− 2 and the active material loading is 32.0 mg cm− 2), a pouch-type full-cell with C/Ag anodeless electrode was fabricated33. The pouch-type full-cell retained 80.2% of its initial capacity after 300 cycles (Fig. 6b), which is superior cycling stability with highest specific capacity among previously reported ASSBs using layered cathodes with Ni contents of 90% (Supplementary Table 3). Supplementary Fig. 12 shows XPS and CP-SEM results for the SM-Ni90 cathode disassembled in its charged state after 500 cycles. The disassembled cell was fabricated and cycled under the same conditions as that evaluated in Fig. 6a. These results reveal that surface reactions at the CAM/electrolyte interface and intra-particle isolation were largely absent, with only some of the CAM particles observed to be detached from the SE. Therefore, the capacity fading of the SM-Ni90 cathode observed in Fig. 6 is primarily ascribed to the detachment of the CAM particles from the SE. A new strategy for overcoming this detachment issue needs to be developed to facilitate the use of Ni-rich cathodes in ASSBs while simultaneously delivering high energy densities and cycling stabilities.