Discovering new materials for electrocatalytic energy conversion reactions is a key step toward energy sustainability. However, for catalysts to be viable in practice, they must perform in multiple, potentially conflicting objectives. We demonstrate this challenge for the acidic oxygen reduction reaction (ORR), where activity, stability, and material cost must be balanced. Using the continuous composition space of high-entropy alloys (HEAs) together with our established models for activity and dissolution, we identify a Pareto-optimal set of ORR catalysts within the Ag─Au─Cu─Ir─Pd─Pt─Rh─Ru system via multiobjective Bayesian optimization. Additionally, we introduce a fine-tuned machine learning model that predicts adsorption energies for alloys spanning 12 elements and 9 adsorbates. Our results show that alloying expands the hypervolume spanned by the Pareto front, consisting of low- to medium-entropy alloys composed primarily of Ag, Au, Cu, Pd, and Pt. We further propose an approach for analyzing the Pareto front by quantifying the loss in hypervolume when critical elements (Au, Pd, and Pt) are removed, clarifying their relative contributions to optimal performance. This work highlights the need to consider all relevant objectives in catalyst optimization and the advantage of HEAs as a powerful platform for multiobjective catalyst discovery.
Plenge et al. (Tue,) studied this question.