Machine-learning-guided prediction of Cu-based electrocatalysts towards ethylene production in CO2 reduction

Abstract

Cu-based materials are the most commonly used electrocatalysts for CO2 reduction to ethylene. The selectivity of copper-based catalysts is affected by many complicated and coupled factors, such as composition, additive and morphology. Therefore, developing highly selective copper-based catalysts for ethylene production is still a significant challenge. This study constructs a CO2 reduction catalysis database using published experimental data. Machine learning (ML) models are developed to study the importance of various factors on the CO2 reduction activity of Cu-based materials. The ML model predicts that the needle-like structured Cu2O (110) composited with copper hydroxide, N-doped carbon black would benefit the Faradaic efficiency of ethylene production in KOH electrolyte. This data-guided ML framework provides a facile alternative method for the quick screening of active Cu-based catalysts towards CO2 reduction to ethylene.