Impact of solid-electrolyte interphase reformation on capacity loss in silicon-based lithium-ion batteries

Silicon (Si)-based anodes are promising for high-energy lithium-ion batteries due to their high theoretical specific capacity, low cost, and abundance. However, the substantial volume changes (up to 300%) during lithiation and delithiation cause insufficient lifetime and capacity fading. Various strategies have been developed to address volume expansion, but industrial upscaling remains challenging due to a lack of understanding of the degradation mechanisms. Si-based anodes typically consist of a porous matrix with graphite particles and Si active material embedded in a carbon binder domain (CBD), often incorporating nanoparticles or carbon nanotubes for conductivity. The ionic conductivity is determined by the porous network, and cycling stability is influenced by the interaction of the active material with the fluorine-containing electrolyte, leading to SEI formation. Volume rearrangement leads to cracking, CBD detachment, and SEI evolution, which has been studied in specific cell configurations. However, a comprehensive understanding of the structural and chemical evolution of the Si interface in full-cell configurations remains crucial for industrial upscaling. This study employs multi-scale characterization techniques to investigate the correlated structural and chemical evolution of the Si-nanocomposite anode interface upon cycling, aiming to improve anode architectures suitable for industrial applications.

The literature extensively discusses the challenges posed by silicon anodes in lithium-ion batteries, primarily focusing on the significant volume changes during cycling and the resulting capacity fade. Studies have explored various strategies to mitigate these issues, including the use of porous structures, nano-sized silicon particles, carbon nanotubes, and optimized binders to enhance conductivity and mechanical stability. The formation and evolution of the solid electrolyte interphase (SEI) layer on the silicon surface are also widely recognized as a major factor influencing battery performance. Previous research has examined SEI growth in half-cell configurations, highlighting the progressive SEI growth on void surfaces and the subsequent destruction of Si nanowire integrity. However, the impact of SEI reformation on capacity loss in industrially relevant full-cell configurations remains less well understood, necessitating the detailed investigation presented in this study. The influence of different C-rates on SEI growth and battery performance has also been noted in the literature. Finally, studies have shown structural changes at the active material interface in full-cell configurations, hypothetically linked to SEI reformation, but lacking experimental proof of the structural and chemical evolution at the Si interface upon cycling in a full-cell configuration.

This research developed a correlated multi-scale workflow to analyze the structural and chemical modifications of Si-nanocomposite anodes during cycling. The workflow integrated several techniques: 1. **X-ray Microscopy (XRM):** Used to select a representative volume of interest (VOI) at the micrometer scale, providing a large field of view to capture the overall microstructure of the anode. This initial analysis allows for the precise selection of smaller regions for more detailed characterization. 2. **Focused Ion Beam-Field Emission Scanning Electron Microscopy (FIB-FESEM) Tomography:** Utilized to obtain high-resolution (12 × 12 × 12 nm³) three-dimensional images of the selected VOI, revealing the nanoscale structural features of the anode. 3. **Energy Dispersive X-ray Spectroscopy (EDS) Tomography:** Correlated with FIB-FESEM, EDS was used to perform elemental mapping (50 × 50 × 250 nm³ resolution) of elements such as Si, Fe, O, and C, providing chemical information about the different phases present in the anode. 4. **Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS):** Used to obtain detailed elemental maps (12 × 12 nm²) of elements like ⁷Li, ¹⁶O, and ¹⁹F, providing insights into lithium distribution and SEI composition. 5. **Machine Learning (ML) Assisted Segmentation:** A convolutional neural network (CNN) with a U-Net architecture was trained to segment the FIB-FESEM tomography data. This automated segmentation process efficiently identified four phases: graphite, Si/FeSi₂, pores, and the combined CBD and SEC. The accuracy of the segmentation was around 90%. 6. **Electrochemical Modelling:** A simplified half-cell configuration was used for electrochemical modelling, incorporating the segmented microstructure data obtained from the CNN model to simulate the discharge behavior. This facilitated the analysis of the impact of SEC growth on lithium-ion transport and capacity. 7. **Chemomechanical Modelling:** A chemo-mechanical model was employed to investigate the stability of the SEI/silicon reaction front and the influence of material properties on the SEC evolution. This model was based on the continuum solid mechanics framework and incorporated the concept of the chemical affinity tensor. The Cut Finite Element Method (CutFEM) was used for numerical simulations of the reaction front kinetics. A full-cell configuration with a technology-relevant anode (Si-nanocomposite with a-Si and c-FeSi₂ domains in a porous CBD and graphite matrix) and an NMC cathode was used for electrochemical cycling. The electrochemical performance of the cell was evaluated using a Maccor battery tester. The samples (pristine, 3-cycles, and 300-cycles) were carefully prepared for the various microscopy and spectroscopy analyses.

The correlated multi-scale analysis revealed several key findings: 1. **Significant Si Volume Loss:** A substantial decrease (approximately 50%) in the volume fraction of the Si-nanocomposite core domains was observed from the pristine state to 300 cycles. This loss was correlated with the growth of a silicon electrolyte composite (SEC) phase in the vicinity of the Si domains. 2. **SEC Growth and Composition:** The initially thin SEI layer evolved into a micron-sized SEC structure during cycling. This SEC phase was characterized by a high concentration of oxygen and fluorine, indicating a reaction between the fluorine-containing electrolyte and the active silicon material. The presence of lithium within the SEC suggested Li-trapping, and the detection of Li, O, and F isotopes supported the formation of LiₓSi, SiOₓ, and lithium oxide. 3. **Lithium Trapping and Active Material Underutilization:** Electrochemical modelling, informed by the experimentally derived microstructure, showed that the growth of the SEC phase leads to the depletion of Li⁺ ions in the SEC regions. This depletion results in underutilization of the active material, which directly translates to capacity loss. The median state of Si lithiation decreased from 7.4% to 5.1% between 3 and 300 cycles, reflecting reduced capacity. 4. **Chemomechanical Model Insights:** The chemo-mechanical modelling indicated that the stability of the SEI/silicon reaction front depends critically on the ratio of Young's moduli of the SEI and Si phases. An unstable reaction front, where the modulus ratio exceeds 1, can lead to an inhomogeneous stress distribution, potentially contributing to further instability and SEC growth. The model showed that inhomogeneities like pores or cracks in the vicinity of the reaction front can accelerate instability. 5. **Capacity Retention and Si-core Integrity:** Despite the significant changes in the microstructure, the Si-composite core remained intact even after 300 cycles, indicating a capacity retention of approximately 70% at a C-rate of C/2. This core integrity suggests potential for further optimization of anode design to minimize the adverse effects of SEC formation.

The findings of this study address the long-standing challenge of capacity fade in silicon-based lithium-ion batteries by providing a comprehensive understanding of the structural and chemical evolution at the Si interface during prolonged cycling. The observed loss of Si volume from the core domains and the associated SEC growth are directly linked to capacity loss through Li⁺ ion depletion and active material underutilization. The chemo-mechanical modelling highlights the crucial role of material properties in influencing the stability of the SEI/Si reaction front, thus providing insights into how anode architecture can be optimized to mitigate SEC growth and capacity fade. The relatively high capacity retention at 300 cycles, despite significant SEC formation, suggests that further improvements are possible through targeted material design. Specifically, tailoring the mechanical properties (such as Young’s modulus) of the Si and SEI phases could improve stability and reduce unwanted SEC growth.

This study provides critical insights into the capacity loss mechanisms in silicon-based lithium-ion batteries. The correlated workflow combining advanced microscopy techniques, deep learning-based image analysis, and sophisticated modelling revealed the complex interplay between structural changes, chemical reactions, and electrochemical performance. The results emphasize the importance of the Si/SEI interface and highlight the significant impact of SEC growth on active material utilization and capacity fade. Future research should focus on designing anode materials with tailored properties, particularly the Young’s modulus of the Si and SEI phases, to enhance interface stability and reduce detrimental SEC formation. Exploring novel electrolyte formulations to minimize SEI formation and reformation is also an important avenue for future research.

While this study provides a detailed analysis of SEI reformation in silicon-based anodes, several limitations should be noted. The electrochemical modelling was simplified to a half-cell configuration, which may not perfectly capture all aspects of the full-cell behavior. The resolution of FIB-FESEM, while high, still limits the precise identification of individual SEI layers, especially at early cycling stages. The accuracy of the CNN-based segmentation is around 90%, indicating potential minor inaccuracies that may affect the quantitative analysis of phase volumes. Furthermore, the sample size for the microscopy analysis was limited, so the results may not be completely generalizable to all silicon-based anodes. Future studies could benefit from extending the modelling to full-cell configurations, developing higher-resolution imaging techniques, and increasing the statistical sample size.