Computational Screening of Transition Metal-Nitrogen-Carbon Materials as Electrocatalysts for CO2 Reduction

Atomically dispersed M-N-C catalysts are a promising, cost-effective class of materials for reducing CO₂ to value-added products through the CO₂ reduction reaction (CO₂RR). However, complex multi-objective optimization of several properties including catalyst stability, activity, and selectivity for target products are necessary to make CO₂RR more efficient with this class of catalysts. We systematically investigate activity and selectivity for carbon monoxide, formic acid, and hydrogen evolution pathways on model M-N₄C₁₀ active sites for 26 transition metal species. Our work shows that under acidic conditions, all the considered M-N₄C₁₀ sites except M=Fe, Co, Cr, Cd, and Pt should have CO₂RR onset potentials lower than the hydrogen evolution reaction. We identify the transition metal active sites that should catalyze the CO pathway, leading to gaseous CO production, CO poisoning, or reduction to further products. To understand the reasons for predicted activity and selectivity, we furthermore correlate atomic features for the transition metals with the calculated onset potential of each pathway, showing moderate correlation between both electronegativity and atomic radii with the CO₂RR onset potentials. The high-throughput and feature-based approach in this work not only serves as a guide for present experimental efforts but can also serve as a starting point for machine learning efforts to accelerate active site modeling and catalyst discovery.