Researchers Use Machine Learning to Screen Dual-Atom Catalysts for Urea Electrosynthesis

A New Blueprint for Clean Urea Production

Researchers have developed a powerful new framework that combines quantum chemistry and machine learning to design catalysts for the electrochemical production of urea, a vital fertilizer, without creating harmful byproducts.

The electrochemical conversion of waste gases like carbon monoxide (CO) and nitrogen oxides (NO) into urea is a promising green chemistry goal. However, the reaction is notoriously difficult to control.

"The challenge is that when CO and NO react on a catalyst, they usually don't form urea. Instead, they tend to make unwanted by-products such as ammonia or hydrocarbon compounds. This makes selective urea production very difficult."
— Co-lead author Professor Qin Li, Griffith University

To solve this, a team from Griffith University and Queensland University of Technology focused on dual-atom catalysts—pairs of metal atoms positioned on carbon edges.

The team simulated 90 catalyst designs with high-accuracy quantum methods. They then used this data to train a machine learning model that screened over 1,400 additional catalyst candidates.

The critical discovery was a single, powerful predictor: co-adsorption energy, which measures the combined binding strength of CO and NO on the catalyst surface.

Only catalysts with a moderate co-adsorption energy successfully drive the reaction toward urea.

"We found a very narrow 'sweet spot' for this energy. If CO and NO bound too weakly, they fell off the surface. If they bound too strongly, the gases got over-reduced and formed the unwanted side products. Only moderate binding strength favoured urea formation."
— Co-lead author Dr Yun Han

The machine learning model, which incorporated atomic properties and carbon edge structural information, was able to rapidly predict the co-adsorption energy for each candidate. This dramatically narrowed the field, reducing a pool of 1,458 potential catalysts down to just 259 promising candidates, a few of which were then validated by further simulation.

"This approach dramatically accelerates catalyst discovery. This study provides a clear design rule for making urea catalysts, and shows how machine learning and chemistry can solve complex reaction problems."
— Co-lead author Professor Aijun Du, Queensland University of Technology

The research, titled "Machine Learning-Assisted Design Framework of Carbon Edge-Dominated Dual-Atom Catalysts for Urea Electrosynthesis," was published in ACS Nano.