Atomistic Prediction of Sorption Free Energies of Cationic Aromatic Amines on Montmorillonite: A Linear Interaction Energy Method

Abstract

Molecular dynamics (MD) simulations were performed to calculate free energies of sorption (ΔGsorb) of cationic aromatic amines to Ca-montmorillonite. We applied the linear interaction energy (LIE) method, well-established in the biochemistry field, to derive ΔGsorb. We obtained a mean average error of 0.3 kcal mol–1 within the compound training set and an error of 0.41 kcal mol–1 for a validation test set. We were able to reproduce absolute ΔGsorb values for a variety of compound structures, including the zwitterionic antibiotic oxytetracycline. MD simulations also provided atomistic level insights into the underlying driving forces that modulate sorption. Importantly, our approach provides a compelling alternative to polyparameter linear free energy relationship methods, which have shown limited success in capturing the sorption of ionogenic compounds with polar and/or charged moieties. We conclude that the LIE method can be used as a robust and tractable method to predict ΔGsorb within families of organic cations bound to aluminosilicate clay minerals.