| Mar 02, 2026 |
Researchers used artificial intelligence to redesign hydrogen fuel cell catalysts, boosting performance and durability while cutting costs for clean transport.
|
|
(Nanowerk News) Hydrogen cars promise a clean alternative to fossil fuels, but one stubborn problem has held them back: the platinum catalyst at the heart of every hydrogen fuel cell costs too much and wears out too fast. Now, a team of Korean researchers has used artificial intelligence to crack a molecular puzzle that could change that.
|
|
Scientists at KAIST and Seoul National University have developed an AI tool that predicts how metal atoms arrange themselves inside a catalyst. Their discovery? Zinc plays a surprisingly decisive role in helping platinum and cobalt atoms snap into a high-performance structure, one that works better and lasts longer than today’s commercial catalysts.
|
|
Professor EunAe Cho of KAIST’s Department of Materials Science and Engineering led the research alongside Professor Won Bo Lee of Seoul National University’s School of Chemical and Biological Engineering. The work is published in Advanced Energy Materials (“Machine Learning-Guided Design of L1₀-PtCo Intermetallic Catalysts: Zn-Mediated Atomic Ordering”)
|
 |
| Schematic diagram of AI-based atomic alignment prediction. (Image: KAIST)
|
|
Think of it like solving a jigsaw puzzle. Instead of trial and error, the AI calculates in advance which arrangement of pieces will complete the picture fastest. In this case, the “pieces” are metal atoms, and the AI figured out the optimal path to organize them.
|
|
The problem the team set out to solve is well known in fuel cell research. Platinum-cobalt alloy catalysts deliver strong performance, but forcing their atoms into a regular pattern, known as an “intermetallic” structure, requires extreme heat. That process causes particles to clump together and destabilizes the material, making it impractical for real fuel cells.
|
|
The researchers turned to machine learning-based quantum chemistry simulations to map exactly how atoms move and settle inside the catalyst. The AI revealed that adding zinc acts as a mediator: it helps platinum and cobalt atoms find their correct positions more easily, producing a more sophisticated and stable structure without the destructive high-temperature processing.
|
|
When the team synthesized a zinc-platinum-cobalt catalyst based on the AI’s predictions, the results confirmed the model’s accuracy. The new catalyst outperformed commercial platinum catalysts in both activity and long-term durability. It proved that a virtual blueprint designed by artificial intelligence could translate directly into a superior material in the lab.
|
 |
| Synthesis process of Zinc-introduced Platinum-Cobalt catalyst. (Image: KAIST)
|
|
The implications stretch well beyond passenger cars. Researchers expect the technology to help extend catalyst lifespans and cut manufacturing costs across a range of carbon-neutral industries, including hydrogen trucks built for long-distance hauling, hydrogen-powered ships, and energy storage systems.
|
|
Professor EunAe Cho stated, “This research is a case of utilizing machine learning to predict the atomic arrangement tendency of catalysts in advance and implementing this through actual synthesis,” and added, “AI-based material design will become a new paradigm for the development of next-generation fuel cell catalysts.”
|
