Advancing CO2RR with O-Coordinated Single-Atom Nanozymes: A DFT and Machine Learning Exploration

Electrochemical CO2 reduction reaction (CO2RR) offers a promising route toward zero-carbon emissions. Recently emerged single-atom nanozymes (SANs), which combine the advantages of single-atom catalysts (SACs) and nanozymes, show potential in CO2RR electrocatalysis applications. Herein, we designed 260 carbon-supported TMO4 SANs, introducing heteroatoms outside the coordination sphere to modulate the interaction between the first-coordination sphere and the support structure. Using a collaborative workflow of the density functional theory (DFT) and machine learning (ML), we assessed these SANs for the CO2RR performance. Among them, 185 SANs demonstrated high stability. Our model, based on extreme gradient boosting regression (XGBR) and trained on 100 labeled SANs, successfully predicted the limiting potentials (UL) for 85 SANs. From these, we distinguished 33 SANs with superior activity, good specific product selectivity, and greater hydrogen evolution reaction (HER) suppression compared with benchmark TMN4-C catalysts. Particularly, MoO4 supported on external H-doped graphene (Mo-I) and FeO4 supported on external N,O-doped graphene (Fe-VI) exhibited remarkable enzyme-mimicking characteristics, with Mo-I showing a formate dehydrogenase-like behavior and Fe-VI mimicking CO dehydrogenase with UL values of −0.05 and −0.06 V, respectively. Through ML feature engineering and electronic structure property analysis, we determined the first-coordination sphere–support interaction (FCSSI) as a key descriptor in regulating the catalyst performance. We hope that these insights may provide valuable guidance for exploring potential SANs for CO2RR.