Machine-Learning assisted screening of double metal catalysts for CO₂ electroreduction to CH₄

Electrochemical CO₂ reduction reaction (CO₂RR) has become a promising application in addressing energy challenges and environmental crises. However, the scaling relationship between the reaction intermediates constrains the successful deep reduction of CO₂. Dual-metal-site catalysts (DMSCs) have emerged as potential electrocatalysts for CO₂RR by breaking the scaling relationship due to their more adaptable active sites. Herein, this study aims to investigate the correlation between the adsorption energies of essential intermediates in CO₂RR catalysis with double transition metal atoms anchored on graphdiyne monolayer (TM₁-TM₂@GDY) through machine-learning (ML) assisted density functional theory (DFT) calculations. The results reveal the important descriptors of CO₂RR catalyzed by TM₁-TM₂@GDY, and demonstrate that the heteronuclear TM₁-TM₂@GDY have great potential for deep CO₂ reduction. Especially, Co-Mo@GDY and Co-W@GDY show low limiting potential (-0.60 V and −0.39 V, respectively) and high selectivity on the reaction from CO₂ to CH₄ based on the free energy diagrams. This study indicates that the two TM atoms on GDY act cooperatively for the catalysis of CO₂RR. Notably, utilizing ML eliminates the need to calculate all transition metal combinations by DFT, which is a great boost in quickly investigating catalytic performance and high screening for excellent catalysts.