Machine learning enabled exploration of multicomponent metal oxides for catalyzing oxygen reduction in alkaline media

Abstract

Low-cost metal oxides have emerged as promising candidates used as electrocatalysts for the oxygen reduction reaction (ORR) due to their remarkable stability under oxidizing conditions, particularly in alkaline media. Recent studies suggest that multicomponent metal oxides, with their intricate compositions and synergistic effects, may outperform their single-metal oxide counterparts. However, exploring the considerable number of potential combinations of multicomponent metal oxides using experiments would be time- and cost-intensive. Herein, we analyzed 7798 distinct metal oxide ORR catalysts from previous high-throughput experiments, which included metal elements such as Ni, Fe, Mn, Mg, Ca, La, Y, and In. These catalysts were tested at different potentials, specifically 0.8 and 0.63 V vs. reversible hydrogen electrode (V_RHE). After feature engineering, we employed the XGBoost method to build the machine learning model and mapped the performance of unexplored compositions. Feature explanations suggested that for achieving high current density, attention should be paid to a high number of itinerant electrons (interant electron) and high configuration entropy. Finally, we identified promising regions within 15 different ternary metal oxides with higher catalytic activities for catalyzing the ORR at 0.8 and 0.63 V_RHE, respectively. We found that for the current density at 0.8 V_RHE, the ternary systems Mn–Ca–La, Mn–Ca–Y, and Mn–Mg–Ca show promising potential for further investigations, in particular for hydrogen fuel cells. Similarly, for the current density at 0.63 V_RHE, the Mn–Fe–X (X = Ni, La, Ca, and Y) and Mn–Ni–X (X = Ca, Mg, La, and Y) systems deserve close attention in the future, as they may contribute to the production of hydrogen peroxide (H₂O₂) as a commodity. This study highlights the significant potential of artificial intelligence in accelerating catalyst design and materials discovery, thereby paving the way for future advancements in sustainable energy technologies.