Accurate computation of gas binding in the nanoscale porous organic cage CC3 via coupled cluster theory

This study investigates the binding of seven gas molecules-N₂, CH₄, C₂H₂, CO₂, H₂O, SF₆, and CHCl₃-within the central cavity of the nanoscale porous organic cage CC3, using a high-level local coupled cluster method that accounts for single, double, and perturbative triple excitations, extrapolated to the complete basis set limit. This results in the formation of the CC3@7 dataset, which presents unique challenges due to the need for accurate descriptions of confinement effects and many-body interactions that contribute to binding. The CC3@7 dataset is used to evaluate a variety of lower-cost computational approaches. Among the methods tested for accurately predicting the binding order for all seven gas molecules, the recommended MP2-based approach is MP2+<i>ai</i>D(CCD), which achieves a mean absolute error (MAE) of 0.4 kcal mol^-1. For density functional theory (DFT) methods, B97M-V+<i>E</i>^ABC, B97M-V, M06-L-D3, B97M-rV+<i>E</i>^ABC, PBE0+D4, and PBE+D4 are recommended, with MAEs ranging from 0.3 to 0.4 kcal mol^-1. Additionally, r²SCAN-3c and<i>ω</i>B97X-3c are identified as low-cost options, with MAEs of approximately 1 kcal mol^-1. Considering both accuracy and stability, PBE0+D4 is recommended for investigating nanoscale host-guest bindings when only DFT methods are feasible. Furthermore, PBE0+D4 has been successfully applied to study the binding of additional atoms and hindered solvent molecules, demonstrating the flexibility of the CC3 cage to accommodate larger molecules that exceed its cavity size.