High-throughput screening of single-atom catalysts on 1 T-TMD for highly active and selective CO2 reduction reaction: Computational and machine learning insights

The study of transition metals and various possible surface compositions has sparked great interest in low-cost materials, which exhibit high activity and selectivity in catalysis. While various single-atom catalysts loaded on transition metal dichalcogenide (TMD) substrates with excellent CO2 reduction performance have been identified, the relationship between catalytic activity and the intrinsic properties of TMD single-atom catalysts remains unclear. Hence, a high-throughput first-principle computational approach is proposed to screen 24 transition metals anchored on 8 TMD monolayers to determine their catalytic activity in CO2RR. The results show that Fe@CoS2, Pt@TiTe2 and Co@CoS2 exhibit exceptional performances with low CO2RR limiting-potentials of −0.045 eV, 0.75 eV, and 0.54 eV, respectively, showcasing selective pathways towards formic acid (HCOOH), methane (CH4), and methanol (CH3OH). Employing the Sure Independence Screening and Sparsifying Operator method(SISSO), key descriptors linking the performance of single-atom catalysts with their intrinsic features are identified, providing insights for the discovery of superior CO2RR catalysts. Moreover, it was observed that a feature of the anchored single atom, the difference between covalent radius and atomic radius (CR-R), is associated with multiple crucial reaction steps, exhibiting a strong linear relationship with the charge transfer of *COOH. This work not only identifies promising CO2RR catalysts but also establishes a predictive framework for screening catalysts based on their intrinsic properties, paving the way for future advancements in CO2 reduction research.