Machine Learning Identifies Catalyst ‘Sweet Spot’ for Greener Urea from Waste Gases

Researchers have used machine learning to identify a precise catalyst condition that enables the selective production of urea from waste gases, offering a promising path toward more sustainable fertilizer manufacturing. The breakthrough centers on a single measurable property—co-adsorption energy—that predicts whether a catalyst will drive carbon monoxide (CO) and nitrogen oxides (NO) to form urea instead of unwanted byproducts like ammonia or hydrocarbons.

Why Sustainable Urea Production Matters

Urea is a critical component of nitrogen-based fertilizers, supporting global food production. However, conventional urea synthesis is energy-intensive and relies heavily on fossil fuels, contributing significantly to greenhouse gas emissions. As demand for fertilizers grows with the world’s population, finding cleaner production methods has turn into a priority for both environmental and industrial stakeholders.

The Challenge of Selective Urea Synthesis

Electrochemical urea production using waste gases such as CO and NO presents a greener alternative, but it faces a fundamental obstacle: when these gases interact with a catalyst, they often fail to bond in the way needed to form urea. Instead, they tend to produce ammonia or hydrocarbon compounds, reducing efficiency and requiring costly separation steps.

As Professor Qin Li from Griffith University explained, “The challenge is that when CO and NO react on a catalyst, they usually don’t form urea. Instead, they tend to create unwanted by‑products such as ammonia or hydrocarbon compounds. This makes selective urea production extremely difficult.”

How Machine Learning Revealed the Catalyst ‘Sweet Spot’

To overcome this challenge, a research team from Griffith University and the Queensland University of Technology combined quantum chemistry simulations with machine learning. They began by modeling 90 dual-atom catalyst designs—pairs of metal atoms anchored on carbon material edges—using high-accuracy computer simulations to study how these materials interacted with CO and NO simultaneously.

From this initial dataset, they trained machine learning models to rapidly screen over 1,400 additional catalyst candidates. The analysis revealed that the key to selective urea formation lies not in how strongly CO or NO bind individually to the catalyst, but in how tightly they are held together when both are adsorbed at the same time.

This insight led to the identification of a single predictive metric: the “co‑adsorption energy.” As co-lead author Dr. Yun Han stated, “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, they poisoned the catalyst or favored side reactions. Only within a specific range did the gases remain in the optimal configuration to form a carbon-nitrogen bond and produce urea.”

Implications for Green Fertilizer Production

By enabling rapid, accurate screening of catalyst materials based on co-adsorption energy, this approach accelerates the discovery of efficient, selective catalysts for electrochemical urea synthesis. When paired with renewable electricity and waste gas feedstocks, such systems could significantly reduce the carbon footprint of fertilizer production.

The method similarly exemplifies how artificial intelligence can be applied to solve complex problems in materials science and green chemistry—using data-driven models to uncover non-obvious relationships that guide experimental design.

Conclusion

The identification of a narrow co-adsorption energy ‘sweet spot’ represents a meaningful step toward decarbonizing urea manufacturing. By focusing on the synergistic behavior of reactant gases on catalyst surfaces, researchers have uncovered a design principle that could enable cleaner, more efficient fertilizer production. As machine learning continues to integrate with quantum simulations, such interdisciplinary approaches are likely to play an expanding role in developing sustainable industrial processes.

Sources: Phys.org, National Tribune, SciMex