Catalytic CO−O coupling on high-entropy alloys: A composition optimization dependent on the reaction assumptions

Generating atomic-scale data for catalysis simulations is often not the primary bottleneck in accurately predicting catalytic processes. Instead, the reliability of these predictions is highly sensitive to the assumptions underlying the modeling approach. In this work, we demonstrate how modeling assumptions for the coupling reaction between surface-adsorbed CO and O influence the predicted optimal composition of high-entropy alloys. • Catalytic carbon monoxide and oxygen coupling is optimized on high-entropy alloys. • A single dataset is used to propose optimal compositions for two different scenarios. • The desired catalytic application guides the choice of reaction modeling assumptions. • Methanol oxidation and CO-tolerant hydrogen fuel cell have different optimal catalysts. • Experimental measurements can iteratively update modeling assumptions. Catalytically joining carbon monoxide and oxygen atoms is crucial for various environmental and industrial applications. However, the reaction environments for this process vary significantly across different applications. In this paper, we demonstrate how reaction environments and associated modeling assumptions influence the predictions of optimal disordered alloy catalysts for CO-O coupling. Our analysis is based on a dataset comprising several hundred *CO, *OH, and *O simulated adsorption energies on atomically disordered face-centered cubic (111) surfaces composed of Ag, Au, Cu, Pd, Pt, and Ru. This dataset is provided alongside this publication to support further machine learning-based searches for optimal CO-O coupling catalysts within this composition space for any CO-O coupling application. We illustrate the impact of different modeling assumptions on the identification of CO-O coupling catalysts through two key examples: electrochemical methanol oxidation and CO-tolerant hydrogen fuel cells.