(197bg) Automated Reaction Exploration of Solid Electrolyte Interphase

Solid electrolyte interphase (SEI) is a thin layer on the surface of an anode in a lithium-ion battery (LiB), serving as a protective layer that prevents the reactions between the electrolyte and anode. A well-formed SEI layer is essential for the stability and performance of LiBs. However, SEI degradation is a common problem in LiBs that causes reduced performance and safety issues. To understand SEI layer, SEI degradation has been extensively studied by experimental and computational methods. Due to the limitations of current experimental techniques, the exploration of reactions within SEI layer relies on computational methods, such as kinetic Mote-Carlo (KMC) method, and density functional theory (DFT). However, compared to experimental data, these reactions did not comprehensively cover all the products during SEI degradation. On the other hand, a more complicated reaction network by Persson was constructed by a machine learning model, which was trained by 5193 species and 86,001,275 reactions. However, this model still could not identify plausible species and reactions in SEI degradation. Here, an advanced reaction exploration program, Yet Another Reaction Program (YARP), is applied to explore the reaction network of LiBs. Different from using any filter based on chemistry knowledge in machine learning models, the graph-based elementary reaction step is utilized to explore all possible reactions in LiBs. Then, the transition states (TSs) are approximately located by the growing string method (GSM) with GFN2-xTB semi-empirical method (xTB). Finally, these TSs are refined by Berny optimization and intrinsic reaction coordinate (IRC) calculation to construct the reactions. With the idea of YARP, the reaction network with N-depth (N elementary reaction steps) can be constructed and the feasibility of all reactions in this network is guaranteed by quantum chemistry. The depth four exploration in ethylene carbonate (EC) shows that YARP is able to locate lithium ethylene mono-carbonate (LEMC), lithium ethylene di-carbonate (LEDC), and other byproducts that are consistent with experimental data.