The Synergistic Effect between Metal and Sulfur Vacancy to Boost CO Reduction Efficiency: A Study on Descriptor Transferability and Activity Prediction.

Both metal center active sites and vacancies can influence the catalytic activity of a catalyst. A quantitative model to describe the synergistic effect between the metal centers and vacancies is highly desired. Herein, we proposed a machine learning model to evaluate the synergistic index, , which is learned from the possible pathways for CH production from CO reduction reaction (CORR) on 26 metal-anchored MoS with and without sulfur vacancy. The data set consists of 1556 intermediate structures on metal-anchored MoS, which are used for training. The 2028 structures from the literature, comprising both single active site and dual active sites, are used for external test. The XGBoost model with 3 features, including electronegativity, d-shell valence electrons of metal, and the distance between metal and vacancy, exhibited satisfactory prediction accuracy on limiting potential. Fe@Sv-MoS and Os@MoS are predicted to be promising CORR catalysts with high stability, low limiting potential, and high selectivity against hydrogen evolution reactions (HER). Based on some easily accessible descriptors, transferability can be achieved for both porous materials and 2D materials in predicting the energy change in the CORR and nitrogen reduction reaction (NRR). Such a predictive model can also be applied to predict the synergistic effect of the CORR in other oxygen and tungsten vacancy systems.