Machine learning high-throughput screening of rare earth SACs with different coordination environments for the HER

Abstract

High-throughput screening of all rare earth SACs was conducted using ML to evaluate their HER performance. DFT calculated the ΔG_*H data of 100 groups of catalysts, and the model trained by the GBR algorithm exhibited the highest accuracy, with R² and RMSE values of 0.970 and 0.157, respectively. The study also identified three potential HER catalysts (|ΔG_*H| < 0.20 eV).