Introduction

All-solid-state batteries (ASSBs) offer significant potential as next-generation energy storage solutions due to their higher energy density and enhanced safety1,2,3,4. However, their practical deployment is hindered by several key challenges5, including the development of suitable solid-state electrolytes (SSEs) and the optimization of interfacial evolution. Typically, viable SSEs must exhibit high lithium-ion conductivity at room temperature and possess robust electrochemical stability6,7. Additionally, interface issues, such as interface chemical8 and electrochemical reactions9, poor contacts10, and mechanical instability11, contribute to inefficient lithium-ion transport, which can significantly impact the battery’s electrochemical performance and safety12,13. A critical focus of our research is the suboptimal physical contact between the SSEs and the positive electrode active material by developing and employing suitable solid electrolyte materials.

Typically, the inorganic SSEs used in ASSBs are ceramic materials with intrinsic high hardness and brittleness14,15, which makes it difficult to buffer stress at the solid-solid contact interface. Moreover, the anisotropic nature of SSEs, characterized by disparate chemical expansion coefficient, diffusion coefficient, stiffness, and crystal orientation relative to the positive electrode material, exacerbates interfacial mismatch, leading to contact failure and impeded ion transport. While contemporary research on ASSBs interfaces predominantly emphasizes chemical and electrochemical reactions16,17, such as surface modifications of electrodes and electrolytes to mitigate side reactions, it is imperative to concurrently investigate and optimize the mechanical behavior of electrolyte materials to comprehensively understand battery performance and durability18. This emphasis on reactions primarily stems from early-stage comparisons of irreversible side reactions between electrolyte materials and positive or negative electrodes. Recent advancements in highly stable halide solid electrolytes, like Li3YCl619 and Li3InCl6, etc.20,21, showcase remarkable compatibility with the positive oxide electrodes. This promising development offers a unique avenue for in-depth investigation into the underlying mechanisms governing their mechanical behavior within battery systems.

During battery operation, the SSE is subjected to stresses from lithium deposition within electrolyte defects, the expansion and contraction of positive electrode particles, and external forces applied during assembly22,23,24. These adverse stresses can deform or even fracture the electrolyte particles, impairing the contact between the electrolyte and positive electrode particles and significantly hindering lithium-ion transport. The mechanical-electrochemical coupling behavior in SSEs is crucial while always neglected for the development of inorganic SSE materials. Mechanically robust electrolytes can undergo reversible deformation without fracturing25,26,27, maintaining consistency between the solid electrolyte layer and the positive electrode, enhancing mechanical stability28, ion transport performance29, and long-term cycling stability30. Thus, optimizing strain engineering in electrolyte materials is indispensable and requires meticulous consideration of their mechanical properties. Key mechanical metrics, such as elastic modulus, strength, and toughness, are vital for assessing the load-bearing capacity, deformation characteristics, and failure modes of materials in ASSBs. For example, Hu et al.31 discovered a new class of viscoelastic inorganic glass solid electrolytes that combine the high ionic conductivity of inorganic crystalline solid electrolytes with the flexibility of organic polymer solid electrolytes. Ruan et al.32 demonstrated that SSEs with high fracture toughness are more resistant to volume changes induced by lithium-ion deposition and stripping, thereby reducing crack formation and propagation, and improving the stability and cycle life of the battery. Eric et al.33 found that thinner samples of Li7La3Zr2O12 exhibited higher elastic modulus, hardness, and fracture toughness compared to thicker samples, thus offering better resistance to deformation caused by external forces. The research team of Jeff Sakamoto34 indicated that SSEs and positive electrode particles generate uneven stress. When this stress exceeds a certain threshold, both the SSEs and positive electrodes may experience fracture. Therefore, the mechanical properties of inorganic ceramic electrolytes, often characterized by brittleness, significantly influence battery performance due to their direct contact with the negative and positive electrodes. Consequently, investigating strategies to enhance the mechanical resilience of these materials is imperative.

Herein, we propose a dispersed defect-enhanced toughness strategy to significantly improve the mechanical robustness of halide electrolytes. By carefully controlling the quenching process during synthesis, we strategically introduce a high density of dispersed defects within the crystal structure of the electrolyte. These defects promote enhanced interactions between dislocations, as well as between dislocations and other crystal imperfections, effectively hindering dislocation motion and improving the material’s resistance to deformation. Moreover, the dispersed defects alleviate stress concentration at interfaces by mitigating dislocation pile-up, thereby suppressing crack propagation. This mechanism enhances the fracture toughness of the positive electrode–electrolyte composite under cyclic stress, contributing to the long-term mechanical stability of the ASSBs. Furthermore, the dispersed defects introduced through the quenching process can enhance the Young’s modulus of the material, enabling the electrolyte to better accommodate the strain imposed by the cathode during battery cycling, which is associated with lithium insertion and extraction. This enhanced strain accommodation capability contributes to improved crack resistance and fracture toughness, ultimately leading to enhanced mechanical performance and longevity in ASSBs. Using Li2+xYxZr1-xCl6 as a model system, we systematically investigated the influence of different cooling processes on the formation and dispersion of defects within the electrolyte structure as well as their impact on mechanical behavior in battery cycling. Synchrotron and laboratory powder X-ray diffraction (XRD), Cryo-transmission electron microscopy (Cryo-TEM), and electron paramagnetic resonance (EPR), etc., were used to investigate the phase and microstructure of the samples. Results indicated that the ion transport properties and phase structure of the Li2+xYxZr1-xCl6 electrolyte remained largely unaffected by different cooling conditions. However, quenching introduced additional defects. Quenched samples exhibited higher Young’s modulus, as determined by atomic force microscopy (AFM) and nanoindentation measurements. The Williamson-Hall method was used to calculate the internal stress, revealing higher microstrain in the quenched material. Additionally, synchrotron X-ray computed tomography (CT) analysis further confirmed the impact of the two electrolytes on battery performance. A comparison of the porosity of positive electrode composite materials before and after cycling showed that quenched materials had a greater capacity for mitigating volume changes in the positive electrode material. In conclusion, the dispersed defect-based toughening strategy, implemented via quenching, yielded unique microstructures and mechanical properties in Li2+xYxZr1-xCl6, which could contribute to the development of mechanically robust electrolytes for high-performance ASSBs. This research elucidates the underlying mechanisms governing the mechanical behavior of electrolyte materials and their impact on force-electrical coupling failure in ASSBs. These findings offer valuable insights for strategically modifying electrolytes to enhance their performance and reliability.

Results

Ion transport properties of electrolytes

By adjusting the Y/Zr ratio, a series of Li2+xYxZr1-xCl6 (x = 0.4, 0.5, 0.6, and 0.7) halide electrolytes were synthesized via a melting method. The ionic transport conductivities at different temperatures were evaluated through temperature-dependent electrochemical impedance spectroscopy (EIS) measurements, as shown in Fig. 1a. The highest ionic conductivity, 1.75 ± 0.05 × 10−3 S cm−1, was achieved for Li2+xYxZr1-xCl6 (x = 0.5). Considering the goal of minimizing energy consumption during production, Li2.5Y0.5Zr0.5Cl6 (LYZC) was heated for varying durations (0.5, 1, 2, and 12 h). The physical state and temperature-dependent ionic conductivity were measured, and as shown in Fig. 1b, the highest conductivity was observed after 2 h of heating, where the sample fully melted into the liquid phase. Therefore, 2 h of heating time was selected. Post-treatment of the melted LYZC involved two cooling methods: quenching (YZr-Q) and slow cooling (YZr-N). Temperature-dependent ionic conductivity measurements, presented in Fig. 1c, showed minimal differences between the two samples, with YZr-Q and YZr-N exhibiting conductivities of 1.69 × 10−3 S cm−1 and 1.75 × 10−3 S cm−1, respectively. This suggests that the cooling method has a minimal impact on ionic transport. Figure 1d presents a comparison of the ionic conductivities of Li2+xYxZr1-xCl6 (x = 0.4, 0.5, 0.6, and 0.7) samples and LYZC at various holding times (0.5, 1, 2, and 12 h). Additionally, X-ray diffraction (XRD) patterns of Li+xYxZr1-xCl6 Li2+xYxZr1-xCl6 (x = 0.4, 0.5, 0.6, and 0.7) are presented in Fig. 1e. Li2+xYxZr1-xCl6 (x = 0.4, 0.5, and 0.6) samples adopt an orthorhombic crystal system, with a space group of Pnma. When x = 0.7, the trigonal phase of Li3YCl6 begins to emerge, resulting in a slight change in the intensity of Li2.7Y0.7Zr0.3Cl6. The Rietveld refinement was carried out based on the orthorhombic two-phase Li2.5Er0.5Zr0.5Cl6 7 and the trigonal Li3YCl635 as structural references. The corresponding refined spectra are shown in Supplementary Fig. 1, and the structural information is provided in the Supplementary Tables 1 and 2. Other electrolyte properties, such as the electrochemical windows (Supplementary Fig. 2) and electronic conductivity (Supplementary Fig. 3), showed negligible differences between YZr-Q and YZr-N.

Fig. 1: Ion transport properties and XRD characterization of electrolytes.
figure 1

a The Arrhenius plots of Li+xYxZr-xCl6 Li2+xYxZr1-xCl6 (x = 0.4, 0.5, 0.6, and 0.7) samples co-melted at 550 °C for 2 h. b The Arrhenius plots of LYZC samples at different holding times (0.5, 1, 2, and 12 h). c The Arrhenius plots of quenched sample YZr-Q and slowly cooled sample YZr-N. d Comparison of ionic conductivities of Li2+xYxZr1-xCl6 (x = 0.4, 0.5, 0.6, and 0.7) samples co-melted at 550 °C for 2 h (top) and comparison of ionic conductivities of LYZC samples at different holding times (0.5, 1, 2, and 12 h) (bottom). Inset: Photograph of LYZC in the molten state after 2 h of holding. e XRD patterns of the Li2+xYxZr1-xCl6 (x = 0.4, 0.5, 0.6 and 0.7) samples.

Full size image

Electrochemical performance

A series of electrochemical measurements was conducted to compare the performance of YZr-Q and YZr-N samples in ASSBs. The schematic of the assembled ASSBs is shown in Fig. 2a. Figure 2b shows the voltage-capacity curves for the 1st and 100th cycles at a 1 C rate with a cutoff voltage of 4.3 V (vs. Li|Li+). As the cycling progressed, the batteries with YZr-N positive electrode composites displayed significant polarization, likely due to poor physical contact between YZr-N and the LiCoO2 (LCO) positive electrode material, resulting in slower lithium-ion migration and subsequent polarization. The particle sizes of both electrolytes were analyzed using a scanning electron microscope (SEM) to confirm that the physical contact differences were not caused by variations in particle size, as shown in Supplementary Fig. 4, where the particle size distribution of YZr-Q and YZr-N was found to be similar. Figure 2c displays the cycle number-capacity curves of ASSBs at different rates with a cutoff voltage of 4.3 V (vs. Li|Li+). The ASSB with YZr-Q positive electrode composites exhibited better rate performance, maintaining a specific capacity of 80 mAh g−1 at a 5 C rate. Even after high-rate cycling, the specific capacity remains stable upon returning to a 0.1 C rate, retaining its initial capacity. As shown in Fig. 2d, the long-term cycling stability of ASSBs tested at 1 C between 2.5 V and 4.3 V (vs. Li|Li+) demonstrated that the battery with YZr-Q positive electrode composites retained 78.37% of its initial capacity, with a charge capacity of 86.6 mAh g−1 after 500 cycles. In contrast, the battery with YZr-N positive electrode composites retained only 23.7% of its capacity, with a charge capacity of around 29 mAh g−1, indicating that YZr-Q possesses greater interfacial stability at this voltage. It is important to investigate whether the observed cycling behavior was caused by electrochemical reactions during charge/discharge, such as redox reactions of the electrode materials or electrolyte decomposition—X-ray photoelectron spectroscopy (XPS) analysis was performed on the electrode–electrolyte interface before and after cycling (the 1st, 20th, and 300th cycles), as shown in Supplementary Figs. 5 and 6. No significant peak shifts or changes in the chemical states of YZr-Q and YZr-N were observed, indicating that no side reactions occurred at the electrode–electrolyte interface and both electrolytes remained chemically stable after different thermal treatments. Additionally, we conducted battery tests under low-temperature conditions (−30 °C) and high-voltage operation (2.6 to 4.6 V vs. Li|Li+) for batteries assembled with the two samples. As shown in the Supplementary Figs. 7 and 8, the YZr-Q-based battery consistently exhibited stable electrochemical cycling stability.

Fig. 2: Full-battery electrochemical performance of electrolytes at 150 MPa stack pressure.
figure 2

a Schematic diagram of LYZC-based ASSBs. b Cyclic voltage-capacity curves of YZr-Q and YZr-N in ASSBS at different turns and magnifications. c Rate cycling curves of YZr-Q and YZr-N batteries. d The cycle performance of ASSBs at 1 C (1 C = 140 mAh g⁻¹ for LCO), 25 °C, and 4.3–2.5 V. DRT calculated from EIS measurements at different cycle numbers: e ASSBs assembled with YZr-Q; f ASSBs assembled with YZr-N.

Full size image

The distribution of relaxation times (DRT) was performed on the assembled batteries with the two electrolytes. In the high-frequency range of τ = 10−6 to 10−5 s, the impedance corresponds to the contact resistance at current collectors and electrode interfaces, and electrode particles. The range of 10−5 to 10−2 s is attributed to both the grain boundary impedance of the solid electrolyte and ion diffusion within the positive electrode material, while the range of 10−2 to 100 s is related to charge transfer resistance within the anode and at the anode-SSE interface. And the range of 100 to 101 s represents solid-state diffusion at the LCO cathode36,37,38. From the DRT data of the full battery after 400 cycles, as shown in Fig. 2e, f, it can be seen that in the range of 10−4 to 10−2 s, the interface resistance between the YZr-N electrolyte and LCO cathode increases with cycling. This indicates that, as cycling progresses, partial contact failure may occur between the LCO and YZr-N electrolyte, leading to an increase in the contact resistance between the electrolyte and positive electrode. The Nyquist plots of the EIS and fitting methods are shown in Supplementary Fig. 9, and the values for all elements in the equivalent circuit are listed in Supplementary Table 3. The semicircle is contributed by the interfacial resistance between LCO and SSEs. The interfacial impedance of the YZr-N assembled battery increased with cycling, while the YZr-Q-assembled battery maintained some interfacial stability. And the actual circuit diagram is shown in Supplementary Fig. 10.

To further verify the differences in electrochemical performance between the two electrolytes, we conducted tests under high loading and low stacking pressure conditions. The cycling performance of the battery with a high areal capacity loading (~2.85 mAh cm⁻2) and 150 MPa stack pressure is presented in Supplementary Fig. 11. After 200 cycles, the battery assembled with YZr-Q electrolyte still exhibits a higher capacity retention compared to the battery assembled with YZr-N electrolyte. The low stacking pressure battery is shown in Supplementary Fig. 12. At 0.1 C and a stacking pressure of 2 MPa, the contact area between the positive electrode material and solid electrolyte is reduced24,36, which limits ion transport properties and causes a noticeable capacity decrease in both samples. However, compared to the YZr-N sample, the YZr-Q sample still demonstrates significantly better capacity retention. To verify whether the differences in electrochemical performance are influenced by the negative electrode, we prepared a positive electrode composite material by mixing Li3InCl6 electrolyte with LCO and carbon black, which was then used as the cathode for both YZr-Q and YZr-N electrolytes, as shown in Supplementary Fig. 13. The capacity retention of the ASSBs assembled with both materials was almost identical after 500 cycles. Therefore, we conclude that the differing electrochemical performance of the two electrolytes is primarily influenced by the positive electrode material.

Microstructural inhomogeneity

To further validate the hypothesis and explore the effects of different heat-treated electrolytes on cathode material densification and microstructure, experiments were conducted at the BL16U2 and BL13HB beamlines of the Shanghai Synchrotron Radiation Facility (SSRF) using synchrotron X-ray computed tomography (CT) studies with sufficient contrast and resolution. Composite cathode materials with the same composition (70:30:1, LCO: LYZC: carbon black) and at different charge/discharge cycles (initial, 50th, and 100th) were analyzed to quantify the pore microstructure within the cathode. Using a machine-learning threshold segmentation model, composite cathode images (195 × 117 × 26 μm) were analyzed, revealing the distribution of three components: LCO (green), LYZC (red), and pores (blue). Voids formed during cycling block the ionic and electronic percolation networks in the cathode composite, leading to slower ion transport and uneven electrode reaction kinetics1,24,37. As shown in Fig. 3, with increasing cycling, the porosity of the cathode composite using YZr-Q electrolyte slightly increased from 11.74% (pre-cycling) to 12.6% after 50 cycles, and to 16% after 100 cycles (Fig. 3a–c). This slight increase may be due to stress cracking during the cycle of SSE and LCO or a small number of pores that appear in partial contact. In contrast, the cathode composite mixed with YZr-N electrolyte exhibited a significant increase in porosity, from 10.9% (pre-cycling) to 18.6% after 50 cycles and 29% after 100 cycles (Fig. 3d–f). The rapid increase in porosity can be attributed to contact failure between particles under stress, accompanied by YZr-N particle fracture. We collected tomographic images along the YZ direction from three positions of the sample before charging and after 100 cycles to highlight the contact between different particles. The pre-cycling slices are shown in Supplementary Figs. 14–16, where the average porosity of YZr-Q is 12.1%, and YZr-N is 9.2%. The initial porosity of the YZr-Q cathode composite is higher, which is due to the lower Young’s modulus of YZr-N compared to YZr-Q, causing more deformation and higher compaction under the same pressure. The post-cycling slices are shown in Supplementary Figs. 17–19. In contrast, the YZr-N composite showed larger, interconnected pores, while YZr-Q exhibited smaller, isolated pores, as can be seen in the 3D pore maps after cycling. These findings reveal that the YZr-Q electrolyte provides better rigidity to resist crack propagation and improved toughness to withstand cyclic stress induced by the expansion of LCO particles. In contrast, YZr-N, due to its inferior mechanical properties, is more prone to particle fragmentation under stress, leading to increased porosity in the cathode composite material. This impairs lithium-ion transport and ultimately causes differences in electrochemical performance between the two electrolytes.

Fig. 3: CT volume rendering of cathode composite and volume ratio of each material.
figure 3

The 3D images and volume fraction of a–c YZr-Q and d–f YZr-N composite cathodes.

Full size image

Mechanical properties of electrolytes

Atomic force microscopy (AFM) experiments were conducted to investigate the macroscopic mechanical properties of the electrolytes. During scanning, an AFM probe was applied to the sample surface with a constant force. The interaction between the probe and the sample resulted in significant changes in the force-displacement curves, which are related to the Young’s modulus of the material38. By fitting these curves, the average Young’s modulus of different regions of the sample can be calculated. The 3D AFM topographies of YZr-Q and YZr-N are shown in Fig. 4a and b, respectively, with a scanned area of 500 × 500 nm. The results reveal that the Young’s modulus of YZr-Q is significantly higher than that of YZr-N. Specifically, as shown in Fig. 4c, the Young’s modulus of YZr-Q ranges from ~6 to 15 GPa, with an average of 11.5 GPa, while YZr-N shows a range from ~2 to 12 GPa, with an average of 6.7 GPa. Additionally, the load-separation curves obtained from AFM, as shown in Fig. 4d, align with the divergence slope observed in the Young’s modulus distribution in Fig. 4c, confirming the higher Young’s modulus of YZr-Q compared to YZr-N. We also used nanoindentation to evaluate the mechanical properties, as shown in Supplementary Fig. 20. The load-depth curves of the two samples measured by the nanoindentation instrument similarly indicate a higher Young’s modulus for the YZr-Q sample. The specific values are provided in the Supplementary Table 4. Further, we employed Williamson-Hall (W-H) analysis from XRD measurements to quantify the internal strain (ε) of the samples, which reflects the influence of internal stress in materials subjected to different heat treatments. Synchrotron powder XRD analysis of both electrolytes revealed that while their characteristic peaks were largely consistent, peak shifts were observed, as shown in Fig. 4f, g. These shifts indicate lattice distortions induced by different heat treatments, leading to changes in diffraction peak positions. As shown in Supplementary Fig. 21, the XRD patterns also clearly exhibit peak shifts. We also performed structural refinement on the XRD patterns of both samples, with the detailed refinement profiles shown in Supplementary Fig. 22, and the structural information provided in the Supplementary Tables 5 and 6. The refinement results indicate differences in atomic occupancy between YZr-Q and YZr-N, as well as octahedral distortion in the quenched samples. To further investigate this, we employed continuous symmetry measure (CSM) calculations. By comparing the actual structure of lithium coordination sites with a perfectly symmetric structure, we quantified the presence and degree of lattice distortion39,40. As shown in Supplementary Fig. 23, YZr-Q exhibits a relatively higher average CSM value, indicating more highly distorted lithium sites, whereas YZr-N has a lower average CSM value, suggesting a coordination environment closer to an ideal symmetric structure with lower distortion. Additionally, the broadening of diffraction peaks was used to analyze particle size and intrinsic strain, as shown in Fig. 4e. The ε for YZr-Q and YZr-N was calculated to be 3.06 × 10−3 and 2.77 × 10−3, respectively, indicating that quenching introduces greater internal microstrain41. These results demonstrate that YZr-Q, obtained via quenching, possesses a higher Young’s modulus at the macroscopic level and greater internal microstrain. Consequently, YZr-Q is more resistant to deformation under external stress and, due to its higher internal microstrain, is better able to resist external deformation forces compared to YZr-N.

Fig. 4: Macroscopic and microscopic mechanical properties of electrolytes.
figure 4

The three-dimensional AFM topography of the YZr-Q (a) and YZr-N (b) samples. c The quantitative distribution of Young’s modulus of YZr-Q and YZr-N samples. d The Force-Separation curves of YZr-Q and YZr-N samples. e W-H analysis of the YZr-Q (top) and YZr-N (bottom) samples, Synchrotron X-ray powder diffraction pattern (f), and local magnification pattern (g) for YZr-Q and YZr-N.

Full size image

Origin of strain

Understanding the underlying mechanisms that lead to different mechanical properties in materials with identical chemical compositions is crucial. It is believed that during the solidification process from the high-temperature liquid phase, rapid cooling induces fast solidification, introducing dispersed, high-density defects within the material. In contrast, slow cooling allows more time for crystal growth and arrangement. The defect density was estimated from the slope of the deformed W-H plot (Supplementary Fig. 24), which revealed that the quenched sample YZr-Q exhibited a higher defect density. Cryo-transmission electron microscopy (Cryo-TEM) and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) were used to analyze YZr-Q and YZr-N samples, as shown in Fig. 5a, b. In the cryo-TEM lattice fringe images, representative crystalline regions of LYZC were identified. The images further indicated that rapid cooling in YZr-Q resulted in smaller crystal sizes and the formation of more amorphous (glassy) material (Fig. 5a)42. This is attributed to the fast-cooled rate, which did not allow sufficient time for atoms to diffuse and form larger crystals. In contrast, slow-cooled promoted crystal growth, resulting in larger and more regular crystals (Fig. 5b). HAADF-STEM images revealed that the diffraction spots of YZr-Q (Fig. 5a, Supplementary Fig. 25a) were elongated into elliptical shapes, in contrast to the circular spots in YZr-N (Fig. 5b, Supplementary Fig. 25b). This deformation indicates the presence of more defects in the YZr-Q sample. The EDS mapping spectra in Supplementary Fig. 26 showed that Y, Zr, and Cl elements were uniformly distributed across the particles, indicating that the different cooling rates did not significantly affect the distribution of chemical elements. To further confirm the difference in defect content between the quenched and slowly cooled samples, and to qualitatively analyze the defect density, electron paramagnetic resonance (EPR) was used to characterize the samples43. EPR detects material defects by interacting with unpaired electrons in the material. As shown in the EPR spectra in Fig. 5c, the intensity of the EPR signal was stronger in YZr-Q than in YZr-N, indicating a higher number of unpaired electrons and confirming that YZr-Q has more defects. Therefore, it is proposed that for the quenched sample YZr-Q, the increased defect density inhibits atomic slip within the material, resulting in a higher Young’s modulus and making the material more resistant to deformation. Additionally, the dispersed defects alleviate stress concentration at interfaces by mitigating dislocation pile-up44,45, effectively slowing crack propagation and enhancing the fracture toughness of the positive electrode material under cyclic stress. As illustrated in Fig. 5d, when LCO particles expand in the positive electrode composite, YZr-Q electrolyte can effectively cope with stress changes, maintaining good physical contact between the electrolyte and positive electrode particles. Consistent conclusions were observed in the focused ion beam-scanning electron microscopy (FIB-SEM) images of positive electrode composites after 500 cycles (Fig. 5e). In contrast, the YZr-N electrolyte fails to resist stress when facing the volume expansion of LCO particles, as shown in Fig. 5f. This results in fragmentation of the electrolyte and an increase in porosity within the positive electrode composite, gradually deteriorating the physical contact between the electrolyte and positive electrode particles, which eventually impairs ion transport. In the FIB-SEM images of YZr-N (Fig. 5g), large voids between LCO and YZr-N were observed, with YZr-N pulverized, leading to contact failure between LCO and YZr-N. Supplementary Fig. 27a, b shows the YZr-Q positive electrode composite material before cycling and after cycle for 500th, while Supplementary Fig. 27c, d displays the YZr-N positive electrode composite under the same conditions. Before cycling, both electrolyte-mixed composite positive electrodes exhibited good contact between the SSE and the LCO. However, after cycling, a noticeable increase in porosity between particles is observed in the YZr-N sample, with a deterioration in the contact between the SSE and LCO. Additionally, significant particle fracture is evident in the LCO of the YZr-N sample. In contrast, due to the improved toughness and strength of YZr-Q, the external stress is more effectively mitigated, allowing the SSE and LCO to maintain good mechanical contact even after cycling¼¼.

Fig. 5: Electrolyte strain origin.
figure 5

Cryo-TEM and HAADF-STEM images of YZr-Q (a) with the corresponding SAED pattern in the top-right inset, and YZr-N (b) with the corresponding SAED pattern in the top-right inset. c EPR characterization plots of YZQ and YZN. The schematic diagram of YZr-Q (d) and YZr-N (f) positive electrode composite material after cycling, and the FIB-SEM diagram of YZQ (e) and YZN (g) positive electrode composite material after 500 cycles of 1 C.

Full size image

Discussion

In this study, a dispersed defect-based toughening strategy was employed to develop mechanically robust Li2.5Y0.5Zr0.5Cl6 halide electrolytes. This approach significantly enhances the mechanical strength and toughness of the electrolyte material without compromising its ionic conductivity. It is observed that increasing the cooling rate results in higher Young’s modulus and defect density in the quenched samples. Specifically, the quenched samples exhibit a higher Young’s modulus compared to their slow-cooled counterparts, enhancing their resistance to external pressures. Furthermore, the defect toughening effect enables quenched samples to exhibit enhanced energy absorption capacity, delaying the onset of plastic deformation or fracture. These findings indicate that both the electrolyte and positive electrode materials maintain uniform mechanical properties and microstructures, facilitating continuous and effective mechanical contact between the electrolyte and positive electrode materials. Such enhanced contact is conducive to prolonged battery cycling performance. This study represents an experimental investigation into the influence of microstructure and mechanical properties of SSEs on electrochemical performance. To manage the cyclic mechanical stresses inherent in long-term batteries, reversible deformation and toughening mechanisms are essential. These mechanisms are crucial for regulating positive electrode strain induced by lithium deintercalation, thereby preserving the lattice coherence. This research offers valuable insights and directions for addressing mechanical challenges in ASSBs.

Methods

Materials preparation

Li2+xYxZr1-xCl6 (x = 0.4, 0.5, 0.6, and 0.7) samples were prepared by high-temperature melt quenching. Stoichiometric mixtures of Lithium chloride (LiCl, Macklin, ≥99.95%), Yttrium chloride (YCl3, Macklin, ≥99.95%), and Zirconium chloride (ZrCl4, Macklin, ≥99.95%) were mixed in an Ar-filled glove box and sealed in a quartz ampoule under vacuum, then fired at 550 °C for different time (0.5, 1, 2, and 12 h) with a ramping rate of 4.5 °C/min. In this instance, the YZr-Q sample is the Li2.5Y0.5Zr0.5Cl6 sample heated in a muffle furnace for 2 h and then removed and cooled to room temperature in water (cooling rate of 1000 °C/min), and the YZr-N sample is the LYZC sample heated in a muffle furnace for 2 h and then cooled to room temperature at a cooling rate of 1 °C/min. Both samples were then ground to powder form for 10 min each in a glove box using an agate mortar (H2O, O2 ≤ 1 ppm).

Williamson–Hall analysis

The Williamson-Hall (W-H) method is a procedure to analyze stress and strain derived from X-ray diffraction. The Williamson-Hall (W-H) method is an X-ray diffraction technique for analyzing the microstructure of materials and is based on the Williamson-Hall relationship, which describes the effect of grain size and microscopic stress on the peak width of X-ray diffraction. It provides information on both grain size and microscopic stresses, which is useful for studying mechanical properties, heat treatment effects, and microscopic defects in materials.

The sine of the diffraction peak width versus the diffraction angle (or its reciprocal) is plotted according to the Williamson-Hall relationship.

Based on Scheller’s formula

$${\beta }_{D}=\frac{K\lambda }{Dcos \theta }$$
(1)

The XRD broadening due to the microstrain is given by

$${\beta }_{\varepsilon }=4\varepsilon tan \theta$$
(2)

In XRD data, the broadening (\({\beta }_{T}\)) of the peaks is due to the combined effect of crystallite size (\({\beta }_{D}\)) and microstrain (\({\beta }_{\varepsilon }\)).

$${\beta }_{T}={\beta }_{D}+{\beta }_{\varepsilon }$$
(3)

K is the shape factor (K = 0.94), λ is the wavelength of the X-ray source (λ = 0.15405 nm), and is the peak position in radians.

$${\beta }_{T}=\frac{K\lambda }{Dcos \theta }+4\varepsilon tan \theta$$
(4)

It could also be written as

$${\beta }_{T}cos \theta=\frac{K\lambda }{D}+4\varepsilon sin \theta$$
(5)

The values of grain size and microscopic stress can be obtained by linearly fitting the W-H diagram. The slope of the fitted line is related to the grain size, and the intercept is related to the microscopic stress. By analyzing the fitting results, it is possible to determine the grain size and microscopic stress distribution of the material, which is essential for understanding the microstructure and properties of the material.

The equations used to extract the dislocation density values are also based on the improved Williamson-Hall method. The dislocation density \({\rho }_{d}\) is defined as the total length of the dislocation line per unit volume of the crystal, given by the following formula46:

$$\Delta K=\frac{0.9}{d}+{\left(\frac{\pi {A}^{2}{B}_{D}^{2}{\rho }_{d}C}{2}\right)}^{\frac{1}{2}}K$$
(6)

The modified Williamson-Hall equations ΔK and K are equal to ΔΚ = (Δ2θ) \(\cos {\theta }_{B}/\lambda\) and Κ = 2\(\,\sin {\theta }_{B}/\lambda\), respectively. The wavelength of Cu Kα is \(\lambda\) = 1.5406 Å. The full width, maximum (fwhm), or integral width (β) of the corresponding diffraction peak at θB is denoted by Δ2θ, where θB is the diffraction Angle at the exact Bragg position. The parameter A is determined by the effective external cutoff radius of the dislocation (A = 2), BD is the Burger vector, and the constant C is the mean contrast coefficient of the dislocation, with values ranging from 1 to 4, depending on the properties of the material47.

Characterizations

The powder XRD patterns of Li2+xYxZr1-xCl6 were measured using a PANalytical Empyrean diffractometer with Cu-Kα radiation (\(\lambda\) = 1.54178 Å). The sample preparation procedure was conducted in an Ar-filled glove compartment to prevent air exposure, and the sample was sealed with a Kapton film on a handmade sample table. Samples were selected to obtain better crystallinity, followed by Rietveld refinement using the GSAS-II software package to optimize structural parameters. The samples’ morphologies were examined with a scanning electron microscope, which was investigated by SEM (ZEISS Crossbeam Laser). X-ray photoelectron spectroscopy (XPS) was performed by a monochromatic Al K(alpha) source (25 mA, 15 kV) (Krotos AXIS Ultra Spectrometer). The mechanical properties of cold-pressed SE pellets were analyzed using Atomic Force Microscopy (AFM) in an argon-filled glove box with low levels of H2O and O2 (<0.1 ppm). The modulus distribution of SE pellets was obtained using the Dimension Icon AFM (Bruker) in peak force quantitative nanomechanical mapping mode. Nanoindentation tests the mechanical properties of the sample. Due to the air sensitivity of the sample, the sample is tested in a liquid pool filled with mineral oil. The sample does not show any visible color change or opacity throughout the indentation experiment and is not exposed to air (especially moisture) throughout the installation and measurement process. Cryo-transmission electron microscopy (cryo-TEM) characterizations were carried out using a microscope (JEOL JEM-F200) under cryogenic temperatures (−180 °C) at 200 kV. The electrolyte powder was sprinkled on lacey carbon grids and then transferred onto a TEM holder in an Ar-filled glove box. The Synchrotron-based X-ray powder diffraction patterns were collected using the Brockhouse High Energy Wiggler beamline at the Canadian Light Source (CLS) with a wavelength of 0.3497 Å. CT characterization was conducted at the Shanghai Synchrotron Radiation Facility (SSRF) on the BL16U2 and BL13HB beamlines. The energy of the X-ray beam is 33 keV. A total of 1080 projections were acquired over a 0°–180° angular range. Each projection was captured with an exposure time of 350 ms, yielding a pixel resolution of 0.65 µm. The samples were polymerized using UV light before testing to prevent exposure to air and moisture.

Electrochemical characterization

The ionic conductivity of Li2+xYxZr1-xCl6 was measured using the alternating current (AC) impedance method by the Bio-Logic electrochemical working station. The measurements were performed at temperatures ranging from −60 to 60 °C, with a frequency range of 1 Hz to 7 MHz and a constant voltage of 10 mV. The SE powder was cold-pressed into pellets at a pressure of 300 MPa using a 10 mm diameter customized mold cell. The electrical conductivities were determined using the direct current (DC) polarization method, with externally applied voltages ranging from 100 to 500 mV. For the Linear sweep voltammetry (LSV) test, approximately 60 mg of YZr-Q or YZr-N SE powder was cold-pressed into a pellet. A mixture of 20 mg halide SSE and carbon black with a mass ratio of 1:1 was uniformly coated on one side of the electrolyte pellet as the electrode, while a Li metal was attached to the other side as both the counter and reference electrode. To mitigate the interfacial incompatibility between the halide electrolyte and lithium metal, a Li6PS5Cl (LPSC, China Automotive Battery Research Institute Co. Ltd, Ionic conductivity of 5 × 10−3 S cm−1) interlayer was employed. The LSV measurements were performed using a Bio-Logic electrochemical workstation, with a forward scan range from open-circuit voltage (OCV) to 6 V, a reverse scan range from OCV to 0 V, at a scan rate of 1 mV s⁻¹, and a test temperature of 25 ± 0.5 °C.

Assembly and electrochemical characterizations of ASSBs

The battery assembling was carried out in a glove box filled with Ar (H2O, O2 ≤ 1 ppm) at 25 ± 0.5 °C. Commercial LiCoO2 (China Automotive Battery Research Institute Co. Ltd), halide SSE, and carbon black (Canrd Technology Co. Ltd.) were mixed manually in a mortar for 10 min with a mass ratio of 70:30:1. The Li metal (99.9%, 50 μm, Cellithium) and In foil (99.99%, 200 μm, Santech) were punched into 10 mm-diameter discs using a stainless steel die. The Li-In negative electrode was then prepared by combining In and Li at a mass ratio of 110:2. The composite positive electrode, a double-layer solid electrolyte, and a Li-In negative electrode were then layered and pressed into a customized mold cell. First, Li6PS5Cl powder for 40 mg was placed into the customized mold cell with a diameter of 10 mm and pressed at a pressure of approximately 200 MPa to create the first layer. Subsequently, 40 mg of synthetic LYZC halide SSE was evenly spread on the upper layer of Li6PS5Cl and similarly pressed at about 200 MPa to form a bilayer electrolyte pellet. Subsequently, the mixed positive electrode material was spread on top of the LYZC halide electrolyte layer. To enhance the contact between the electrolyte and the positive electrode, a pressure of about 300 MPa was applied for 5 min. Then, an In foil was attached to the surface of the Li6PS5Cl side, followed by affixing a lithium metal sheet onto the In foil side. The pressure was restricted to 100 MPa. The assembled battery liner was placed into a customized stainless steel casing and subjected to a pressure of approximately 200 MPa for 10 min. Finally, the screws of the steel mold were tightened to apply a final stacking pressure of 150 MPa to the battery. The battery was then subjected to charge and discharge tests in the voltage range of 2.5–4.3 V and 2.6–4.6 V at 25 ± 2 °C using the LAND battery test system. The battery tests under the temperature of −30 °C were realized by a freezer manufactured by Thermo Fisher Scientific.