Stabilizing Ru in Multicomponent Alloy as Acidic Oxygen Evolution Catalysts with Machine Learning-Enabled Structural Insights and Screening

Developing active, stable, and cost-effective acidic oxygen evolution reaction (OER) catalyst is a critical challenge in realizing large-scale hydrogen (H₂) production via electrochemical water splitting. Utilizing highly active and relatively inexpensive Ru is generally challenged by its long-term durability issue. Here, we explore the potential of stabilizing active Ru sites in Ru_<i>x</i>(Ir,Fe,Co,Ni)_1-<i>x</i> multicomponent alloy by investigating its phase formation behavior, OER performance, and OER-induced surface reconstruction. The alloy exhibited a multiphase structure composed of major face-centered cubic (fcc) and minor hexagonal close-packed (hcp) phases at near equimolar concentration. Machine-learned interatomic potential (MLIP) coupled with replica-exchange molecular dynamics was utilized to describe the atomic scale mixing behavior of the Ru_<i>x</i>(Ir,Fe,Co,Ni)_1-<i>x</i> catalysts and other RuIr-based alloys. The model supports our experimental findings of the well-mixed bulk fcc phase and provides an indication of the minor hcp phase formation. The optimized Ru_0.20(Ir,Fe,Co,Ni)_0.80 catalyst exhibited improved OER activity with an average overpotential of ∼237 mV measured at 10 mA cm^-2 and enhanced stability with a low activity degradation rate of ∼1.1 mV h^-1 in 24 h of operation. The acidic OER conditions induced the formation of a thin RuIr-rich oxide shell layer with a trace amount of 3d metals, where Ru was found to be relatively stabilized near the surface of the evolved nanoparticles. The machine learning-accelerated high throughput simulation protocol was further employed to screen other potential RuIr-containing quinary alloys based on expected phase stability. This work highlights the opportunity of stabilizing Ru in a multicomponent alloy matrix with improved activity and stability.