Molecular dynamics computer simulations of binary Lennard-Jones fluid mixtures: Thermodynamics of mixing and transport coefficients

Abstract

Equilibrium (NVT) ensemble molecular dynamics are used to evaluate the transport coefficients of binary Lennard-Jones mixtures via Green-Kubo formulae. Time correlation and mean-square displacements are employed to determine the component self-diffusion and mutual-diffusion coefficients. Time correlation functions are used to evaluate the shear viscosity, thermal conductivity and the thermal diffusion coefficient (Soret-Dufour coefficient). The results are for the low fluid density regime (ρ → 0), where there are experimental data and analytic expressions for the transport coefficients based on kinetic theory. For Ar-Kr, the simulation transport coefficients extrapolate reasonably well to the experimental and the analytic expressions in the ρ → 0 limit. Agreement near the triple point with experiment and previous simulation is also good. The model Ar-Hg viscosity, thermal conductivity and mutual diffusion coefficient manifest more dramatic variations in the simulations as ρ → 0, suggesting deviations from first-order kinetic theory. The temporal evolution of small clusters (dimers and trimers) is investigated using number-population autocorrelation functions. They show that in the ρ → 0 limit in the model Ar-Hg mixtures there are dimers and trimers which are long-lived, in excess of 10 ps, possibly accounting for the deviations from kinetic theory. Partial molar enthalpy, volume and species chemical potential of these fluids are calculated using particle-insertion and particle-exchange techniques applied within (NPT) ensemble molecular dynamics of Ar-Kr and Ar-CH4 mixtures. As noted in earlier work, there is only limited agreement with experimental excess quantities using Lorentz-Berthelot mixing rules. In contrast, the partial molar quantities are accurately obtained and made use of in evaluating the transport coefficients of these mixtures.