MIT Develops AI Method to Predict Catalysts for 200 Million Tons of Greener Ammonia
Updated
Updated · MIT News · Aug 20
MIT Develops AI Method to Predict Catalysts for 200 Million Tons of Greener Ammonia
1 articles · Updated · MIT News · Aug 20
Summary
MIT researchers built a machine-learning method to identify promising metal nitride alloy catalysts for electrochemical ammonia production, aiming to replace years of trial-and-error screening across millions of possible combinations.
The work targets a major industrial problem: ammonia production consumes up to 2% of global energy and generates about 1.5% of greenhouse-gas emissions, while more than 90% of fertilizer ammonia still comes from fossil-fuel-based Haber-Bosch plants.
By linking catalytic activity to key electronic, chemical and structural properties, the model seeks alloys that lower energy use and improve selectivity for ammonia over unwanted side reactions.
The study, published Aug. 11 in EES Catalysis, remains theoretical; the next step is building a reaction cell to test the predicted materials under real operating conditions.
If validated, the approach could help make low-emissions ammonia competitive at the roughly 200 million metric tons the world uses each year.