Results from the MOFlearning project published in Matter
Published on
Artificial Intelligence and materials chemistry :
an enhanced MOF database to accelerate CO₂ capture
A major breakthrough involving François-Xavier Coudert, one of the leaders of the targeted MOFlearning project within the PEPR DIADEM program, was published in the journal Matter on June 4, 2025, in an article entitled “CoRE MOF DB: A curated experimental metal-organic framework database with machine-learned properties for integrated material-process screening.“
This international work, led by French, American, and South Korean scientists, is redefining standards in the field of porous materials, Metal-Organic Frameworks (MOFs), used for the selective capture of gases such as carbon dioxide. By integrating automated data extraction, cheminformatics filtering, and machine learning approaches, the experimental CoRE MOF database has been significantly updated and enriched.
The result : more than 40,000 experimentally synthesized MOF structures are now ready for large-scale simulations, with essential predictive properties (such as stability, heat capacity, and water affinity) to assess their performance under real-world conditions.
The study notably identified 34 MOFs that outperform CALF-20, the current benchmark material for CO₂ capture. Some of these new MOFs are effective even at very low CO₂ concentrations, paving the way for applications in direct air carbon capture (DAC).
This database aims to become the open-access reference source for the many AI-driven studies in the MOF field, promoting knowledge sharing. This breakthrough highlights the relevance of the MOFlearning project within the PEPR DIADEM program, demonstrating how artificial intelligence can accelerate the rational design of materials to tackle environmental challenges.
Link to the targeted MOFlearning project
Plus d'actualités
