It is vital to search for highly efficient bifunctional oxygen evolution/reduction reaction (OER/ORR) electrocatalysts for sustainable and renewable clean energy. Herein, we propose a single transition-metal (TM)-based defective AlP system to validate bifunctional oxygen electrocatalysis by using the density functional theory (DFT) method. We found that the catalytic activity is enhanced by substituting two P atoms with two N atoms in the Al vacancy of the TM-anchored AlP monolayer. Specifically, the overpotential of OER(ORR) in Co- and Ni-based defective AlP systems is found to be 0.38 (0.25 V) and 0.23 V (0.39 V), respectively, showing excellent bifunctional catalytic performance. The results are further presented by establishing the volcano plots and contour maps according to the scaling relation of the Gibbs free-energy change of *OH, *O, and *OOH intermediates. The d-band center and the product of the number of d-orbital electrons and electronegativity of the TM atom are the ideal descriptors for this system. To investigate the activity origin of the OER/ORR process, we performed the machine learning (ML) algorithm. The result indicates that the number of TM-d electrons (Ne), the radius of TM atoms (rd), and the charge transfer of TM atoms (Qe) are the three primary descriptors characterizing the adsorption behavior. Our results can provide a theoretical guidance for designing highly efficient bifunctional electrocatalysts and pave a way for the DFT-ML hybrid method in catalysis research.
Keywords: first-principles calculation; machine learning; oxygen evolution reaction; oxygen reduction reaction; two-dimensional AlP.