Linear regression model for metal-organic frameworks with CO₂ adsorption based on topological data analysis

Metal-organic frameworks (MOFs), self-assembled porous materials synthesized from metal ions and organic ligands, are promising candidates for the direct capture of CO₂ from the atmosphere. In this work, we developed a regression model to predict the optimal component of the MOF that governs the amount of CO₂ adsorption per volume based on experimentally observed adsorption and structure data combined with MOF adsorption sites. The structural descriptors were generated by topological data analysis with persistence diagrams, an advanced mathematical method for quantifying the rings and cavities within the MOF. This enables us to analyze direct effects and significance of the geometric structure of the MOF on the efficiency of CO₂ adsorption in a novel way. The proposed approach is proved to be highly correlated with experimental data and thus offers an effective screening tool for MOFs with optimized structures.