Effective Screening Descriptors of Metal–Organic Framework-Supported Single-Atom Catalysts for Electrochemical CO2 Reduction Reactions: A Computational Study

Metal–organic framework-supported single-atom catalysts (SACs@MOF) show considerable promise in CO2 reduction reactions (CO2RR). However, efficiently screening and designing optimal catalysts is hindered by the lack of effective descriptors for encoding the complex chemical microenvironments in SAC@MOF systems. Herein, through combining an intuition-guided dimensionality reduction strategy with machine learning (ML), we identified critical descriptors based on atomic features and the SAC's constrained coordination geometry, which capture the effects of complex chemical microenvironments on electrochemical CO2RR activity and selectivity for UiO-66-supported SACs. With these descriptors, accurate ML models were developed to predict the limiting potentials for producing HCOOH, CO, and CH4/CH3OH on 48 SACs@UiO-66-X (X = H, NH2, and Br). Moreover, the transferability of the developed descriptors and ML models was demonstrated on 48 additional systems with X = CH3, OH, and NO2. The accuracy of the predicted activity trends for specific SACs combined with different linker groups and the selectivity of the top-performing catalysts were validated through additional DFT calculations. This study provides an effective framework for understanding and modulating chemical microenvironments, enhancing the design and development of MOF-supported SACs for the CO2RR.