Prediction of Alkanolamine pK_a Values by Combined Molecular Dynamics Free Energy Simulations and ab Initio Calculations

Knowledge of aqueous protonation constants (pK_a) of chemical species is of significant importance in CO₂ reactive absorption system design. Their theoretical prediction has mainly relied on implicit solvent models, and the performance of explicit solvent simulations based on classical force fields have rarely been studied. In this paper, we report the results of simulations in explicit TIP3P water with the General Amber Force Field (GAFF) and with the SMD continuum solvent method for the deprotonation pK_a values of 29 conformationally diverse alkanolamine species commonly used in CO₂ capture. In both cases, we employ the Tissandier value for the hydration free energy of the proton (“The proton’s absolute aqueous enthalpy and Gibbs free energy of solvation from cluster–ion solvation data”, Tissandier, M.D. et al., J. Phys. Chem. A, 1998, 102, 7787–7794). The ideal–gas reaction free energies and their uncertainties were obtained from electronic structure calculations using five different compound methods (CBS-QB3, CBS-APNO, G3, G3B3, G4). The hydration free energies of the neutral and protonated forms of the alkanolamines were calculated using the semiempirical AM1-BCC charge method, in addition to several partial atomic charge sets based on the RESP fitting method using electrostatic potentials computed at different ab initio theory/levels in the gas phase as well as in the presence of the solvent reaction field. We incorporated the Galvani surface potential of the ions in the (pK_a) calculations. Although the individual species hydration free energies show significant sensitivity to the charge model, the resulting pK_a values from different charge models are quite similar. Moreover, we found that the protonated amine hydration free energies show slightly less sensitivity to the partial charge method than in the case of the neutral amine. While the predicted pK_a values based on the RESP charges yield reasonable agreement with the experimental data, they are prone to occasional disagreement for molecules of complex geometry. The best performance was achieved using the semiempirical AM1-BCC charges, which showed a mean absolute error of less than 0.73 pK_a units in comparison with experimental data. Our results suggest that the AM1-BCC charge method may be used to model electrolyte solutions encountered in the CO₂ reactive absorption process.