Unlocking stable iodine capture in 2D COFs: insights from DFT combined with multiscale−descriptors−driven machine learning

As the reliance on nuclear energy increases, so does the need to address the environmental risks posed by radioactive iodine isotopes, making the development of efficient adsorbents critical. This study systematically investigates the adsorption mechanisms of radioactive iodine on two−dimensional covalent organic frameworks (2D COFs) using a combined approach of density functional theory (DFT) and machine learning (ML). We successfully predicted adsorption energies across a diverse range of COF structures and identified important electronic and structural factors that influence adsorption strength. The random forest model proved to be the most reliable predictor, while equations derived from the sure independence screening and sparsifying operator (SISSO) provided clear and interpretable structure–energy relationships. Our results indicate that strong adsorption is associated with reduced molecular orbital levels and activation of I2 molecule, which is facilitated by orbital hybridization and charge transfer. Additionally, we clarified the layer−dependent effects and the transition from physical to chemical adsorption through detailed analyses of electronic structures and ab initio molecular dynamics (AIMD). These results not only enhance our fundamental understanding of COF–iodine interactions but also establish a predictive framework to accelerate the discovery and design of highly efficient COF–based adsorbents for nuclear waste management.