Bifunctional Oxygen\nReduction/Evolution Reaction Activity\nof Transition Metal-Doped T‑C<sub>3</sub>N<sub>2</sub> Monolayer:\nA Density Functional Theory Study Assisted by Machine Learning

Designing efficient and cost-effective bifunctional electrocatalysts\nfor the bifunctional oxygen reduction reaction (ORR)/oxygen evolution\nreaction (OER) is crucial for sustainable and renewable energy technologies.\nIn this study, we systematically investigate the potential of single\ntransition metal (TM)-doped T-C<sub>3</sub>N<sub>2</sub> as bifunctional\nORR/OER electrocatalysts using density functional theory and machine\nlearning. The results reveal that TM atoms can be stably incorporated\ninto the N vacancy (TM<sub>N</sub>) and the central hexagonal hole\n(TM<sub>i</sub>) of T-C<sub>3</sub>N<sub>2</sub>, creating various\ncoordination environments for the TM atoms, which can influence the\nORR/OER electrocatalytic performance. The TM atom embedded in the\ncentral hexagonal hole (Cu<sub>i</sub>) is a robust bifunctional ORR/OER\nelectrocatalyst due to its low overpotentials (0.53 V for ORR and\n0.52 V for the OER) and superior thermodynamic stability. The ORR/OER\ncatalytic performance of Cu<sub>i</sub> maintains well under the biaxial\nstrain (−1% to +6%), as the ORR and OER overpotentials of Cu<sub>i</sub> change slightly with the biaxial strain. Nevertheless, the\nORR and OER overpotentials increase sharply once the biaxial compressive\nstrain exceeds −1%. Hence, substrates with lattice constants\nequal to or larger than T-C<sub>3</sub>N<sub>2</sub> are required\nto obtain good bifunctional ORR/OER activity in experimental equipment.\nLastly, we employ the machine learning method with a gradient-boosted\nregression model to determine the origin of ORR and OER activity.\nThe results indicate that the charge transfer of TM atoms (<i>Q</i><sub>e</sub>) is the dominant descriptor for ORR activity,\nwhile the d-electron counts (<i>N</i><sub>e</sub>) and the\nd-band center (ε<sub>d</sub>) are critical descriptors for OER.\nOur research highlights the efficiency of TM atom-doped T-C<sub>3</sub>N<sub>2</sub> as bifunctional electrocatalysts and offers valuable\ninsights for developing electrocatalysts for future clean energy conversion\nand storage applications.