Deciphering the Mechanism and Activity of Diatomic Catalysts for Urea Electrosynthesis with DFT and Machine Learning

Abstract Electrochemical urea synthesis under ambient conditions offers a sustainable alternative to the Haber‐Bosch process but is hindered by sluggish C─N coupling and poor selectivity. To address these challenges, we this study constructs a comprehensive library of diatomic catalysts (DACs) anchored on defect‐engineered nitrogen‐doped graphene and systematically investigate their mechanistic pathways and activity trends for urea synthesis through high‐throughput screening integrated with machine learning (ML). Among the candidates, QV2 and QV3‐typed DACs emerge as the most stable, with stability dictated by N coordination at carbon vacancies. Notably, QV3 exhibits superior activity by enabling a distinct C─N coupling mechanism, wherein the enlarged intermetallic spacing promotes facile N─N bond cleavage of the *NHNH intermediate, generating highly reactive *NH species that readily couple with *CO, thereby markedly reducing the coupling barrier compared with QV2. Furthermore, a unified potential‐determining‐step (PDS) criterion is established to accelerate the identification of highly active catalysts, providing a standardized metric for rapid screening and mechanistic evaluation. ML analysis further highlights two key descriptors, the metal‐N coordination distance (D TM‐N2 ) and the d‐electron parameter (θ d ), that govern the PDS. These findings offer experimentally accessible DAC targets and clear mechanistic insights, thereby establishing a rational basis for next‐generation urea electrocatalyst design.