Researchers at TCD have used supercomputers to find the ‘holy grail’ of catalysis needed to create hydrogen fuel.
Hydrogen fuel is promised to play a major part in global carbon emissions reduction by offering an alternative fuel source for some of the most polluting sectors such as transport and industry.
It is produced by splitting water into oxygen and energy-rich hydrogen, which is then collected and stored in fuel cells. The only problem is that this process requires a huge amount of electricity, which means polluting natural gas has so far been the affordable option of choice for producers.
In order to become a zero-emission fuel, this production needs to be powered by renewable electricity close to wind farms or solar farms. To then make this feasible on an industrial scale, affordable catalysts that can limit the ‘overpotential’ of oxygen in the production process are needed.
Now, researchers from Trinity College Dublin (TCD) have combined powerful supercomputers and leading chemistry research to take a major step towards the ‘holy grail’ of catalysis. While elements such as ruthenium or iridium are effective but too expensive for commercial use, Prof Max García-Melchor and the researchers made a crucial discovery.
Hoovering up a haystack
Writing in Nature Communications, the team said that science has so far underestimated the activity of some of the more reactive catalysts and, as a result, the dreaded overpotential hurdle now seems easier to clear.
Furthermore, in refining a long-accepted theoretical model used to predict the efficiency of water-splitting catalysts, they have made it immeasurably easier for people – or supercomputers – to search for the elusive ‘green bullet’ catalyst. A problem that once looked like an empty canvas, they said, is now more like a paint-by-numbers affair.
Looking to the future, the researchers hope to use AI to digitally test a large number of Earth-abundant metals and ligands (the glue that holds them together to generate the catalysts) and see which of the near-infinite combinations yield the greatest promise.
“It is no exaggeration to say that before now such a hunt was akin to looking for a needle in a haystack,” said García-Melchor. “We are not over the finishing line yet, but we have significantly reduced the size of the haystack and we are convinced that AI will help us hoover up plenty of the remaining hay.”