基于机器学习的二氧化碳电化学还原制备甲酸盐研究

Investigation of electroreduction of carbon dioxide into formate based on machine learning

系统研究了不同双金属中心催化剂催化二氧化碳电化学还原制备甲酸盐。借助机器学习,确定了反应中心金属原子序数、电负性和电离能等特征对双金属中心催化剂表面二氧化碳还原具有主要的影响。基于这些特征,通过高通量机器学习快速预测了105种双金属中心催化剂二氧化碳电还原制甲酸盐及其主要竞争反应的Gibbs自由能变,筛选出29种双金属中心催化剂更倾向于二氧化碳还原得到甲酸盐,是潜在的转化二氧化碳为甲酸盐的高性能催化材料。运用类似的方法预测了105种双金属中心催化剂表面二氧化碳还原中间体的结构,发现中间体吸附能与其吸附构型具有显著的相关关系。

Electrocatalytic reduction of carbon dioxide to high value-added chemical products has provided a new route to alleviate greenhouse effect and other global problems, attracting intensive attention from both industry and academia. However, it still remains a great challenge to develop electrocatalysts with high performance for practical applications. As one of the major products from carbon dioxide electroreduction, formate is of key importance due to its stability, portability, and high volumetric energy density. In this work, we systematically studied the performance of electrochemical reduction of carbon dioxide to formate on various dual-metal-site catalysts using machine learning in conjunction with density functional theory(DFT) calculations. It was determined that the atomic number,electronegativity and ionization energy of metal atoms in the reaction center are major factors influencing the adsorption of formate intermediates over dual-metal-site catalysts. Based on these features, we predicted the adsorption free energy change for the electroreduction of carbon dioxide to formate and its main competitive reaction, hydrogen evolution reaction. 29 out of 105 dual-metal-site catalysts were identified as potential catalysts for formate production from carbon dioxide electroreduction. A similar method was used to predict the structure of the carbon dioxide reduction intermediates on the surface of 105 dual-metal-site catalysts, and it was found that the adsorption energy of the intermediates has a significant correlation with their adsorption configuration.