Unraveling the separation mechanism of gas mixtures in MOFs by combining the breakthrough curve with machine learning and high-throughput calculation

In the field of metal - organic frameworks (MOFs) screening studies, the batch calculation of the mixed gas breakthrough time difference ( Delta T i ) in MOFs and its intricate correlation with various descriptors remain underexplored. This research undertook batch calculations of the breakthrough curves (BC) for different gases within a simulated natural gas environment, designating Delta T i as the performance metric for MOFs in gas separation. The separation performance of computation-ready experimental MOFs for CH 4 /C 2 H 6 and CH 4 /CO 2 mixtures was analyzed in depth utilizing machine learning (ML)-assisted high-throughput computational screening (HTCS) techniques. Then, five ML algorithms were used to quantify the relationship between MOF descriptors and performance, and the effect of the metal center site on the separation performance was further explored. Ultimately, the top ten MOFs were selected for each system. Combining HTCS, ML, and BC, this work provides fresh insights for understanding and designing MOFs with customized adsorption and separation properties.