Machine learning-assisted exploration of the intrinsic factors affecting the catalytic activity of ORR/OER bifunctional catalysts

Abstract

Oxygen reduction reaction and oxygen evolution reaction are pivotal in energy conversion. Herein, we systematically studied the catalytic performance of Pt-doped dual transition metal (DTM) Janus-MXenes as single-atom catalysts (SACs) using first-principles calculations and established machine learning (ML) models to explore the physical and chemical properties affecting the catalytic overpotential. Specifically, the stability of Janus-MXenes was first explored through cohesive energies, phonon dispersion and ab initio molecular dynamics simulations. The electronic properties of Pt-doped Janus-MXenes were investigated next. Some excellent catalysts, including Pt-VO-MnTiCO2 (ηORR/OER = 0.24/0.38 V) and Pt-VO-PdTiCO2 (ηORR/OER = 0.33/0.36 V), were found because of their ultralow overpotential. It’s attributed to the tuning of the electronic properties of SACs by DTM. Importantly, ML models were used to reveal the importance of descriptors affecting overpotential, thereby uncovering the origin of the catalytic activity of SACs. Our work significantly reduces the research costs of SACs and serves as a guide for designing high-performance catalysts.