Machine Learning Aided Design of Reversible MXene Electrocatalysts for Li-Air Batteries

Abstract

Li-air batteries (LABs) are noted for their remarkable potential to deliver high energy density. However, their commercialization faces challenges due to slow kinetics in the oxygen reduction reaction through Li oxide formation and the oxygen evolution reaction through Li oxide decomposition, as well as significant charge/discharge overpotentials and issues with irreversibility. MXenes, known for their tuneable properties, show great promise in overcoming the challenges associated with LABs. However, the extensive variety of potential MXene structures complicates the rapid identification of the most suitable materials. In this study, we integrate density functional theory (DFT) with machine learning (ML) to predict M1M2XT₂-type MXenes (M: transition metal, X: C/N, T: terminal groups) that can enhance key reaction steps while preserving the reversibility of charging and discharging. Our results demonstrate that a finely tuned Extra Trees Regression model, optimized with the ideal number of features, can effectively learn, allowing us to identify promising functionalized MXenes with superior energetics than the usually explored O-terminated MXenes. This study underscores the potential of combining ML with DFT to expedite the discovery of advanced materials for next-generation energy storage technologies.