Overcoming the Trade-off Between the Activity, Stability, and Cost of OER Catalysts for Proton Exchange Membrane Water Electrolysis in a Self-Driving Laboratory

Abstract Proton exchange membrane water electrolysis (PEMWE) offers a sustainable route to hydrogen production, yet its practical deployment is limited by trade-offs among activity, stability, and precursor cost of acidic oxygen evolution reaction (OER) catalysts. Here, we present a self-driving laboratory (SDL) integrating automated batch synthesis, electrochemical half-cell testing, and AI-guided decision-making to simultaneously optimize these properties. A composition-constrained design space, clustering-based batch selection, and relaxed-percentile Pareto filtering enable efficient exploration of ternary Ir–Ru–X (X = Ni, Co, Fe) systems. Using this framework, the platform conducted 24 experiments per ternary system and identified Ir10Ru65Ni25/C as a Pareto-balanced composition exhibiting an overpotential of 211.8 mV at 10 mA cm–2, limited degradation after accelerated durability testing, and reduced Ir content. In PEMWE, Ir10Ru65Ni25 delivered 3.90 A cm–2 at 1.8 V and stable operation for 1000 h at 1.0 A cm–2, demonstrating SDL-driven discovery of multimetallic OER catalysts.