Using AI To Capture Carbon Dioxide Before It Reaches The Atmosphere

Prof. Dr. Seda Keskin Avcı and her master’s student Feride Neva Yüngül from our Department of Chemical and Biological Engineering have developed a new AI-powered method that can identify high-performance membrane materials for carbon capture within seconds.

In the study, published in the Nature Portfolio journal Communications Materials, the researchers paired 8,683 different metal-organic frameworks (MOFs) with 12 polymers and evaluated a total of 104,196 potential material combinations.Their AI model reduced an assessment process that could take months in the laboratory to a matter of seconds, identifying the material pairs with the greatest potential to separate carbon dioxide from other gases.

The findings showed that many MOF-polymer combinations could outperform conventional membranes currently in use. The researchers have made their AI model and dataset openly available to the scientific community, potentially accelerating the development of more efficient membranes for carbon capture, natural gas purification and hydrogen production.