Single atom embedded ZnO monolayers as bifunctional electrocatalysts for the ORR/OER: a machine learning-assisted DFT study

Abstract

Electrocatalysts that exhibit bifunctional activity for the oxygen reduction reaction (ORR) and the oxygen evolution reaction (OER) are essential for advancing the sustainability of clean energy. Using density functional theory (DFT) computations, we systematically investigated the catalytic performance of 17 transition metal single atoms embedded in two-dimensional ZnO for the ORR and OER. Our results indicate that these single atoms strongly interact with ZnO, forming stable single-atom catalysts (SACs). Among them, Ni–ZnO is identified as a promising bifunctional ORR/OER catalyst due to its low overpotentials (η^ORR = 0.42 V, η^OER = 0.54 V). Furthermore, employing the constant potential method, the η^ORR (0.32 V) and η^OER (0.31 V) values can be further reduced under acidic conditions. Machine learning (ML) analysis revealed that the number of outermost electron (N_e) and first ionization energy (E_i) are the two primary descriptors governing OER activity, while ORR activity is mainly influenced by E_i and the atomic radius (R_TM). This study provides theoretical guidance for designing low-cost, efficient bifunctional ORR/OER electrocatalysts and demonstrates the potential of ML in elucidating the relationship between intrinsic catalyst properties and their catalytic activity.