CO₂ capture: more than 100,000 membrane combinations evaluated, seconds per candidate

Researchers at Koç University are combining molecular simulation and machine learning to sort out hybrid metal-organic/polymer membranes to separate carbon dioxide.

Separation of CO2 from an industrial flux per membrane costs much less energy than conventional absorption processes. The lock is somewhere else: you need a material that lets the gases pass quickly and which separates them well, two properties that vary in the opposite direction. To seek this compromise by synthesis and measurement, composition after composition, takes years.

A team from Koç University in Istanbul proposes an alternative: a framework that links molecular simulations and machine learning models that are trained on their results. It covers membranes with mixed matrix, where metallo-organic networks are dispersed in a polymer. More than 100,000 combinations were evaluated, with a promising candidate's performance predicted in seconds where experimental characterization would take weeks.

The procedure does not replace laboratory validation; it moves the bottleneck, reducing to a few dozen the number of materials that are worth synthesizing. The work is published in Communications Materials.