Molecular dynamics simulation of solute diffusion in Lennard-Jones fluids

We performed a molecular dynamics (MD) simulation for a system of 5 solute molecules in 495 solvent molecules interacting through the Lennard-Jones (LJ) 12-6 potential, in order to study solvent density effects on the diffusion coefficients in supercritical fluids. The effects of the size of the solute and the strength of the solute-solvent attractive interaction on the diffusion coefficient of the solute were examined. The diffusion coefficients of the solute molecules were calculated at T = 1.5 (in the LJ reduced unit), slightly above the critical temperature, from rho = 0.1 to rho = 0.95, where rho is the number density in the LJ reduced unit. The memory function in the generalized Langevin equation was calculated, in order to know the molecular origin of the friction on a solute. The memory function is separated into fast and slow components. The former arises from the solute-solvent repulsive interaction, and is interpreted as collisional Enskog-like friction. The interaction strength dependence of the collisional friction is larger in the low- and medium-density regions, which is consistent with the 'clustering' picture, i.e., the local density enhancement due to the solute-solvent attractive interaction. However, the slow component of the memory function suppresses the effect of the local density on the diffusion coefficients, and as a result the effect of the attractive interaction is smaller on the diffusion coefficients than on the local density. Nonetheless, the solvent density dependence of the effect of the attraction on the diffusion coefficient varies with the local density, and it is concluded that the local density is the principal factor that determines the interaction strength dependence of the diffusion coefficient in the low- and medium-density regions (rho < 0.6).