Hydration Free Energies of Multifunctional Nitroaromatic Compounds

Nitroaromatic compounds (NACs) are used as energetic materials, reagents, and pesticides; however, they are potentially hazardous for the environment and human health. To predict the environmental distribution of these compounds, the vapor pressure, aqueous solubility, and Henry's law constant are important properties, as is the solvation free energy in water from which the latter two can be computed. Here, we have calculated the hydration free energies for a set of nine nitroaromatic compounds containing one, two, and three nitro groups using the expanded ensemble molecular dynamics simulation method with TIP3P water and the GAFF, CGenFF, OPLS-AA, and TraPPE force field parameters and the RESP (gas phase), CHELPG (gas phase), and CM4 (aqueous phase) partial atomic charges calculated here. Also, we have computed hydration free energies using the reported default partial atomic charges of the OPLS-AA force field and using the semiempirical AM1-BCC charges with GAFF parameters. The effect of water model flexibility on the computation of hydration free energy is examined with CGenFF/(CHELPG+SPC-Fw) model. All the force fields studied generally led to less accurate predictions with increasing numbers of nitro groups. The average unsigned errors (AUE) show that 6 of 16 force-field/(charge+water) models used perform approximately equally well in predicting measured hydration free energies: these are CGenFF/(CHELPG+TIP3P), CGenFF/(CM4+TIP3P), OPLS-AA/(CHELPG+TIP3P), OPLS-AA/(CM4+TIP3P), TraPPE-UA/(CHELPG+TIP3P), and TraPPE-UA/(CM4+TIP3P). When using the default atomic charges, the OPLS-AA force field was the most accurate, though using CHELPG and CM4 charges led to better predictions. Our analyses indicate that not only the charges but also the van der Waals interaction parameters for the nitro-group nitrogen and oxygen atoms in the force fields are partly responsible for the performance variations in predicting solvation free energies. We also compared the force field-based simulation results with the predictions from the SM6 solvation model and Abraham linear solvation energy relationship (LSER) method. With an appropriate choice of theory and basis set both for geometry optimization and computation, which unfortunately is not known a priori, the SM6 model hydration free energy predictions for the NACs are comparable to the simulation results here. The Abraham LSER predictions with descriptors obtained from the Platts method are of reasonable accuracy. A useful addition to this paper is the Supporting Information that contains a compiled and evaluated list of the hydration free energies of the NACs studied here assembled from the literature.