An empirical approach-based analysis for the exploration of ternary metal sulfide as an active and selective CO2 reduction electrocatalyst

In the quest for sustainable electrochemical carbon dioxide reduction reaction (CO 2 RR) strategies, developing efficient and selective electrocatalysts remains a paramount challenge. Metal sulfides offer diverse types of adsorption sites, leading to a promising avenue to overcome the drawbacks of conventional catalysts, including metals and alloys. Since there are limited references and discussions to study the trend of metal sulfide as a CO 2 RR electrocatalyst, here we developed a less burdensome empirical workflow. The point of the methodology lies in the straightforward learning from experimental data, and the utilization of high-throughput experimental tools is not compulsory. Using the workflow, we aim to clarify what properties we should be concerned about to predict and further obtain optimal electrocatalysts in this early stage of exploration. The methodology integrates a careful analysis of experimental data with material informatics, leveraging density functional theory (DFT) calculations and machine learning (ML). For the case study, we specifically target the ternary metal sulfide selective for syngas carbon monoxide (CO) production. By employing high-dimensional regression ML models trained on a dataset of 18 samples, our analysis underlines the importance of considering crystal structure beyond atomic composition as the catalyst design strategy. We identify that ternary metal sulfides with hexagonal lattice systems and containing cations among Zn/In/Cd are optimal for CO-selective electrocatalysts. Our study offers insights into exploring uncharted materials for a sustainable CO 2 RR with a versatile and burdenless workflow adaptable to various application fields.