DFT and machine learning insights into productive versus self-metathesis selectivity in ruthenium-catalyzed ethenolysis

Ethenolysis, the catalytic cross-metathesis of carbon-carbon double bonds with ethylene, represents an important route for the production of high-value terminal olefins. In this work, density functional theory (DFT) calculations and parallel machine learning (ML) analyses were employed to investigate the factors that alter the reaction selectivity during ruthenium-catalyzed ethenolysis of ethyl oleate. In this context, the selectivity was evaluated as the competition between productive metathesis (PM) and self-metathesis (SM), with the thermodynamic preference for catalyst recovery emerging as the determinant process. In particular, the overall reaction energy associated with PM and, in turn, catalyst regeneration, consistently correlates with experimental selectivity trends. Complementary ML analyses, based on tree-based ensemble algorithms, further demonstrate that catalytic performance arises from multidimensional relationships between various steric and electronic descriptors rather than from isolated electronic properties. These findings establish a predictive framework for the rational design and evaluation of selective ruthenium catalysts for sustainable olefin metathesis applications.