UNSW Sydney: Scientists use AI to make green ammonia even greener

Scientists and engineers at UNSW Sydney, who previously developed a method for making green ammonia, have now turned to artificial intelligence and machine learning to make the process even more efficient.

Ammonia, a nitrogen-rich substance found in fertiliser, is often credited with saving much of the world from famine in the 20th century. But its benefit to humankind has come at a cost, with one of the largest carbon footprints of all industrial processes. To produce it, industrial plants need temperatures of more than 400°C and extremely high pressures – more than 200 times normal atmospheric pressure. Such energy-intensive requirements have made ammonia production a major contributor to global greenhouse gas emissions, accounting for 2% worldwide.

But in 2021, a UNSW team discovered a way to make ammonia from air and water using renewable energy, at about the same temperature as a warm summer’s day.

Dr Ali Jalili, with UNSW’s School of Chemistry, says while the original proof-of-concept demonstrated that ammonia could be created entirely from renewable energy, at low temperatures and without emitting carbon, there was still room for improvement. For example, could it be produced more efficiently, using lower energy, less wasted energy and producing more ammonia?

To answer these questions, the team needed to find the right catalyst – a substance that speeds up the chemical reaction without being consumed by it. As they explained in the paper published today in the journal Small, opens in a new window, the team began by coming up with a shortlist of promising catalyst candidates.

“We selected 13 metals that past research said had the qualities we wanted – for example, this metal is good at absorbing nitrogen, this one is good at absorbing hydrogen and so on,” Dr Jalili says.

“But the best catalyst would need a combination of these metals, and if you do the maths, that turns out to be more than 8000 different combinations.”

Enter artificial intelligence

The researchers fed a machine learning system information about how each metal behaves and trained it to spot the best combinations. That way, instead of having to run more than 8000 experiments in the lab, they only had to run 28.