Machine learning accelerates the screening of single-atom catalysts towards CO2 electroreduction

Abstract

With the gradual increase of global warming and energy crisis, electrocatalytic reduction of CO2 is necessary to alleviate atmospheric contamination and produce value-added fuels and chemicals effectively. As promising heterogeneous candidates, single-atom catalysts (SACs) are prospective for CO2 reduction with high atomic efficiency and unique electronic structure. However, the underlying structure-performance relationship of single-atom electrocatalysts in machine learning (ML) perspectives is also urgent to be explored. Herein, reviews emphasize how to design efficient single-atom electrocatalysts for reducing CO2 by performing ML, with attention on strategies in selecting active sites, tuning coordination environment, and regulating synergistic effects. Subsequently, recent advances in the catalytic performance of diversified SACs towards the CO2 reduction reaction are discussed with the assistance of ML and density functional theory. Finally, challenges and prospects in CO2 reduction are prospected for this emerging field. This review provides an advanced overview of the recent progress and future development of SACs by rapid and low-cost ML methods to present theoretical insights for rationally designing highly efficient electrocatalysts.