We present and discuss here results concerning the implementation of molecular dynamics simulations with
the Green−Kubo formalism to predict transport coefficients in SF6/N2 mixtures under conditions of practical
interest. SF6 was modeled using a flexible molecular force field recently proposed [Olivet, A.; Vega, L. F. J.
Chem. Phys. 2007, 126, 144502], while the force field of Galassi and Tildesley was used for N2 modeling
[Galassi, G.; Tildesley, D. J. Mol. Simul. 1994, 13, 11]. The influence of mixture composition on the mutual
diffusion coefficients and shear viscosities was investigated by a series of molecular dynamic simulations of
SF6/N2 mixtures with different compositions at 300 K and 1 MPa. Temperature and pressure dependencies of
the mutual diffusion coefficient were also investigated for a fixed composition of SF6/N2, covering the
temperature range from 260 to 340 K and two pressure values (1 and 2 MPa). Simulations with these anisotropic
force fields led to mutual diffusion coefficients that diminish when the SF6 concentration in the mixture is
increased. This behavior differs from the hypothesis of no composition dependence assumed for several
estimation methods available in the literature for gaseous binary mixtures at low pressures. Such a composition
dependence is explained on the basis of the deviations from ideality of SF6/N2 mixtures. The effect of the
molecular flexibility considered in the SF6 molecular model is analyzed by comparing the transport coefficients
estimated with the aforementioned force fields to those estimated with simple Lennard-Jones potentials.
Differences between predictions with the anisotropic force fields and the simpler Lennard-Jones potentials
were more significant in the case of the shear viscosity; while viscosities estimated with the Lennard-Jones
potentials showed no dependence on composition, the estimations based on the anisotropic force fields exhibited
a maximum value for the viscosity, which is the same behavior described by an empirical method. Results
presented here are pure predictions, thus covering part of the lack of information existing about the transport
coefficients of these mixtures, which is of relevance for several industrial processes.