Applying machine-learning screening of single transition metal atoms anchored on N-doped γ-graphyne for carbon monoxide electroreduction toward C₁ products

Carbon monoxide electroreduction (COER) has been a key part of tandem electrolysis of carbon dioxide (CO₂), in which searching for high catalytic performance COER electrocatalysts remains a great challenge. Herein, by means of density functional theory (DFT) computations, we explored the potential of a series of transition metal atoms anchored on N-doped γ-graphyne (TM@N-GY, TM from Ti to Au) as the COER electrocatalysts. We found that the final product selectivity of these single-atom catalysts depended on the position of the metal atom in the periodic table, with metals in the front and middle of each periodic period exhibiting high selectivity for CH₄, while metals in the back producing CH₃OH. Machine learning (ML) found that metal atomic number was intrinsic to the difference in COER performance of these single-atom catalysts (SACs). The free energy changes showed that Mn@N-GY and Ni@N-GY exhibited outstanding COER catalytic performance for producing CH₄ and CH₃OH, respectively. Our results provide theoretical and experimental guidance for designing efficient COER catalysts to generate C₁ products.