AI-guided lab finds longer-lived iridium-free oxygen catalyst
A human-supervised, more-than-90-percent automated platform combined materials synthesis, screening, machine-learning optimization and language-model reasoning to evaluate 2,942 catalysts across 53 material systems. The preprint reports that InMnPdOx kept overpotential below 0.5 volts for more than 1,000 hours in acidic testing at 10 mA per square centimetre.
Why it matters: The work joins AI selection to a high-throughput physical loop and surfaces a catalyst that could reduce dependence on scarce iridium or ruthenium in proton-exchange-membrane electrolysis.