Design and screening of transition-metal doped chalcogenides as CO2-to-CO electrocatalysts

Abstract

CO2 electroreduction is one of the most potential ways to realize CO2 recycle and energy regeneration. The key to promote this technology is the development of high-performance electrocatalysts. However, the rational design of electrocatalysts with highly catalytic activity and products selectivity toward CO2 reduction reaction (CO2RR) remains a challenging task. Herein, we developed a machine learning (ML) model to achieve efficient exploration of electrocatalysts for CO2RR by combining with density functional theory (DFT). The results show that the electron numbers of the d orbital is the most important descriptor and the support vector regression (SVR) has the best predictive performance. The coefficient of determination (R2) and mean squared error (MSE) are 0.9193 and 0.0162, respectively. Based on the well-trained model, the overpotential of Ni@MoS2 is successfully predicted to be 0.477 V and it shows the best electrocatalytic performance for CO2RR. DFT calculation results show that *COOH → *CO is the potential-limiting step of CO2-to-CO electroreduction for Ni@MoS2. The DFT-calculated overpotential is 0.481 V, which is consistent with the ML-predicted results. This work provides a convenient machine learning model for the effective theoretical design and screening of CO2-to-CO electrocatalysts.