Lila Sciences finds palladium catalysts for green hydrogen via AI lab
Lila Sciences announced on September 25, 2026, that its AI-driven lab identified six palladium-based catalyst families for acidic oxygen evolution in green hydrogen production. The team screened 2,942 catalysts across 53 material systems in three months, defying prior assumptions that palladium would underperform in acidic conditions. The top composition matched ruthenium’s performance in…
Key points
- Lila Sciences’ AI lab screened 2,942 catalysts in three months, finding six palladium-based families for acidic OER
- Top catalyst matched ruthenium’s stability but uses palladium, easing supply constraints for green hydrogen
- AI-driven workflow automated 96 parallel tests per batch, cutting human effort by over 90%
The discovery came from Lila’s AI Science Factory (AISF), a fully automated workflow combining Bayesian reasoning, language models, and lab equipment. The system synthesized and tested 96 catalysts in parallel, reducing human effort by over 90%. While the catalysts are not yet commercialized, the team is testing durability at industrial scales and expanding the approach to other reactions like green ammonia production. Lila’s chief scientific officer, Rafael Gómez-Bombarelli, emphasized that the system’s automation exceeds 90%, though full autonomy remains pending robotics upgrades.
Lila Sciences’ AI Lab Uncovers Palladium Catalysts for Green Hydrogen
Unite.AI · 25 September 2026
Loading the full article…
This text was published by Unite.AI. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Researchers benchmark how LLMs handle political character attacks · 1 src
- arXiv study finds 30% of AI research agents hack rewards without instructions · 2 src
- Researchers release META, an episodic-memory trading agent for financial decisions · 1 src
- Epydemix Agent Framework automates epidemic modeling with AI agents · 1 src
- Researchers introduce tracer, a user simulator for AI behavior alignment · 1 src
Comments
via GitHub Discussions