A Machine Learning Shortcut for Screening the Spinel Structures of Mg/Zn Ion Battery Cathodes with a High Conductivity and Rapid Ion Kinetics

Abstract Spinel structures with various porosities are promising ion batte cathodes for enhancing the performance of electrode materials. Based on the machine learning method, we performed a comprehensive screening of all spinel structures from the periodic table and identified the best Mg/Zn ion battery cathode materials with a high conductivity and rapid ion kinetics, with a prediction accuracy of 91.2%. We used a target-driven XGBoost algorithm to accelerate the ab initio predictions and reported six new spinel structures (MgNi2O4, MgMo2S4, MgCu2S4, ZnCa2S4, ZnCu2O4, and ZnNi2O4) with high electronic conductivities, high ion diffusions (>1 × 10−9 cm2s−1), low volume expansions ( 10−4 S · cm−1) at room temperature. The proposed strategy shortens the research cycle of spinel screening for cathodes of Mg/Zn ion batteries and offers a solution toward the design of high-performance 3D electrode materials.