CO2 uptake prediction of metal-organic frameworks using quasi-SMILES and Monte Carlo optimization

Metal-organic frameworks (MOFs) are organic-inorganic hybrid crystalline porous materials with high specific surface areas that have revolutionized materials science and adsorbent development. The research aims to investigate and develop QSPR (quantitative structure-property relationship) analysis of MOFs that applies the quasi-SMILES parameters such as BET (Brunauer, Emmett and Teller) specific surface area and pore volume, pressure, and temperature for CO2 uptake prediction of MOFs for the first time. The total data set, including 260 quasi-SMILES features of MOFs, were randomly split into training, validation, and test sets thrice. Here, six QSPR models have been constructed using two target functions based on quasi-SMILES descriptors. The significance of different eclectic descriptors of CO2 increase and decrease uptake capacity of MOFs is presented. Mechanistic interpretation of the effective descriptors for the model is also offered. Based on the model interpretation results, adding basic N- and O-containing, and double-bond containing functional groups to the surfaces of organic linkers of MOFs plays a significant role in improving CO2 uptake properties. The satisfactory statistical quality of the three proposed models based on TF2 shows that the generated models can be efficient for predicting the CO2 capture capacity of MOFs.