.png){kind=icon}
.png){kind=icon}
Return
Stable acidic oxygen-evolving catalyst discovery through mixed accelerations
.png){kind=icon}
PRE
AI
DOI：10.1038/s41929-025-01463-x.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Ruthenium oxides (RuOx) are promising alternatives to iridium catalysts for the oxygen-evolution reaction in proton-exchange membrane water electrolysis but lack stability in acid. Alloying with other elements can improve stability and performance but enlarges the search space. Material acceleration platforms combining high-throughput experiments with machine learning can accelerate catalyst discovery, yet predicting and co-optimizing synthesizability, activity and stability remain challenging. A predictive featurization workflow that links a hypothesized catalyst to its actual single- or mixed-phase synthesis and acidic oxygen-evolution reaction properties has not been reported. Here we report a hierarchical workflow, termed mixed acceleration, integrating theoretical and experimental descriptors to predict synthesis, activity and stability. Guided by mixed acceleration through 379 experiments, we identified seven ruthenium-based oxides surpassing the Pareto frontier of activity and stability. The most balanced composition, Ru0.5Zr0.1Zn0.4Ox, achieved an overpotential of 194 mV at 10 mA cm−2 with a ruthenium dissolution rate 12 times lower than that of RuO2.
Proton-exchange membrane water electrolysers rely on iridium to catalyse their anodic reaction, and while ruthenium is a less costly alternative due to its similar activity, it is not as stable. Now, a hierarchical machine-learning catalyst discovery workflow, termed mixed acceleration, is put forward to predict catalyst synthesis, activity and stability, and identify promising RuOx-based water oxidation catalysts.
Keywords:
ruthenium oxides
oxygen-evolution reaction
proton-exchange membrane water electrolysis
machine learning
catalyst stability