Machine learning-accelerated exploration on element doping-triggering material performance improvement for energy conversion and storage applications

Abstract

Element doping, as a crucial material modification strategy, can effectively regulate the electronic structure, crystal structure, and surface chemical properties of materials. The selection of doping elements and the precise control of doping conditions are key to determining the material's final performance, making doping strategies widely applicable across various fields. However, traditional experimental methods for optimizing doping conditions are often time-consuming and costly, while theoretical calculations, though insightful, tend to be resource-intensive, requiring significant time and expense with limited efficiency. Machine learning (ML) has emerged as a powerful tool to accelerate the development of element-doped materials by leveraging large datasets to predict optimal doping strategies. This review examines the application of ML in the design and screening of high-performance doped materials, with a focus on electrocatalysis, photocatalysis, and lithium batteries. ML techniques can accurately predict material performance, reduce experimental costs, and reveal complex relationships between doping and material properties. Despite notable progress, challenges such as data quality and multi-objective optimization persist. The review also highlights potential solutions to these issues. Looking forward, future research should prioritize advancing ML methodologies and improving material databases to further drive the discovery of next-generation doped materials for diverse applications.