(574e) Can Molecular Simulations Predict the Binary Interaction Parameters of the Activity Coefficient Models?

Ability for predicting the thermodynamic and phase equilibrium properties of liquid mixtures is vital for the design of industrial chemical processes. In practice, molecular thermodynamic models such as non-random two liquid (NRTL) and UNIFAC models are widely applied for predicting these properties. The binary interaction parameters that are the input to these thermodynamic models, quantify the nature of interactions between different components of the mixture. These binary interaction parameters are usually calculated by regressing experimental data. In this work, we determine these interaction parameters by combining molecular simulations with theory, as such an approach is predictive and does not depend on the availability of experimental data. The thermodynamic model of interest in this work is the non-random two liquid (NRTL) activity coefficient model. Binary interaction parameters are calculated for the water-methanol, methanol-methyl acrylate, and water-methyl acrylate mixtures. For each of the binary mixtures, the effective interaction strength between the components is obtained from molecular dynamics simulations, which are in turn used in the calculation of the binary interaction parameters. We show that the binary interaction parameters calculated from molecular simulations agree reasonably well with those estimated from regressing the experimental data.