UNSW Researchers Use Ai To Improve Green Ammonia Production

Scientists at UNSW Sydney have used artificial intelligence to dramatically improve the process of producing green ammonia, reducing laboratory testing from thousands of experiments to just a few dozen.

The team previously developed a method to produce ammonia using air, water and renewable electricity. Their new work, published in Small, uses machine learning to identify a highly efficient catalyst, significantly improving the energy conversion rate and feasibility of green ammonia as an alternative to fossil fuel-based production.

Reducing 8000 Experiments To 28

Ammonia plays a major role in agriculture, particularly in fertiliser, but the industrial processes that produce it contribute around 2 per cent of global greenhouse gas emissions. Traditional production requires high temperatures and pressure, making it energy-intensive and heavily reliant on fossil fuels.

UNSW School of Chemistry researcher, Dr Ali Jalili, said the team sought to improve their previous proof-of-concept method by finding a more efficient catalyst to lower energy requirements and increase output.

“To find that catalyst, we were looking at more than 8000 possible combinations,” said Dr Jalili. “Using Ai meant we only had to run 28 experiments in the lab.”

The team selected 13 metals known for their chemical properties related to ammonia synthesis. These included metals with high nitrogen or hydrogen absorption qualities. The challenge was to find the best combination among thousands of possibilities.

“We trained a machine learning model with data on how each metal behaves,” Dr Jalili said. “That gave us a shortlist of 28 promising candidates to test.”

A New Five-Metal Catalyst

The final result was a five-metal combination: iron, bismuth, nickel, tin and zinc. The catalyst achieved a sevenfold increase in the rate of ammonia production and delivered near-100 per cent Faradaic efficiency – a metric indicating how much energy is used in the desired chemical reaction.

“Almost all of the electrical energy we used went into making ammonia. Very little was wasted,” said Dr Jalili.

The system operates at 25°C, compared to the more than 400°C required by the traditional Haber-Bosch method. This makes the process not only less energy-intensive, but also more scalable and cost-effective.

Local Production And New Use Cases

The team is already working on real-world applications. One focus is decentralised production using shipping container-sized ammonia modules that can be leased or purchased by farmers. These modules include the Ai-optimised catalyst, a plasma generator and an electrolyser in one unit.

“Our approach breaks away from centralised ammonia production, which depends on billion-dollar facilities,” said Dr Jalili. “Instead, we’re enabling smaller, decentralised units that are cheaper and quicker to deploy.”

These mobile ammonia units could help farmers produce fertiliser onsite, reducing transport costs and further lowering the carbon footprint.

Hydrogen Storage And Energy Opportunities

Ammonia is also being explored as a hydrogen carrier. It contains more hydrogen per volume than liquid hydrogen, which could make it more suitable for energy storage and transportation in the emerging hydrogen economy.

“This same system doubles as a carbon-free hydrogen carrier,” said Dr Jalili. “It opens up new possibilities in energy markets and aligns with global efforts to build a clean hydrogen economy.”

The research team is now focused on deploying the Ai-discovered catalyst more broadly, refining its performance and supporting wider adoption of green ammonia in both agriculture and energy sectors.

Last Updated on June 20, 2025 by Nick Ross