Machine learning-driven insights into self-healing silicon-based anodes for high-performance lithium-ion batteries.

In recent years, the utilization of silicon, rather than graphite, has emerged as a compelling alternative for anode materials in Li-ion batteries, promising higher energy density. However, a significant challenge lies in the degradation of silicon anodes due to volume fluctuations during charge and discharge cycles, resulting in a rapid decline in battery capacity. To tackle this issue, researchers are investigating the integration of self-healing polymers as binding agents in the anode structure through trial-and-error approaches, which is both time-consuming and expensive. Overcoming practical experimentation challenges, this study delves self-healing polymers through machine learning methods as a more practical approach. The role of structural features and functional groups within these polymers in maintaining anode integrity and prolonging battery capacity across multiple charge cycles were explored by utilization of random forest, ridge algorithms, support vector machines, and neural networks. Notably, the neural networks algorithm exhibits superior performance, achieving 96% accuracy for test data. SHAP analysis revealed that ether functional groups, donor and acceptor hydrogen bonds, and dual-interconnected rings have the most positive impact on preserving battery capacity. In this study, we introduce a set of design principles for selecting functional groups aimed at enhancing the self-healing capabilities and prolonging the lifespan of Si-based LIBs. This study has the potential to pave the way for the development of more efficient and enduring Li-ion batteries.