Describing Binary Mixture Diffusion in Carbon Nanotubes with the Maxwell-Stefan Equations. An Investigation Using Molecular Dynamics Simulations

Adsorption and diffusion of pure components and binary mixtures containing methane, ethane, propane, n-butane, isobutane, and hydrogen at 300 K in a variety of configurations of carbon nanotubes (CNTs) have been investigated using configurational-bias Monte Carlo (CBMC) simulations and molecular dynamics (MD) simulations. Both self-diffusivities, Di,self, and the Maxwell-Stefan (MS) diffusivities, Di, were determined for a variety of molecular loadings (theta), approaching saturation limits. For comparison purposes, self-diffusivities were also determined in pure fluids of varying densities using MD. At low loadings (theta), the Di,self correspond to the value for low-density gases. With increasing loadings, however, the Di,self in CNTs are slightly higher than the values in fluids when compared at the same molecular density. In CNTs, the Di,self is significantly smaller in magnitude than the MS diffusivity Di, signifying strong correlations between molecular jumps along the tube. Consequently, for mixture diffusion, the component self-diffusivities are close together. MD simulations of binary-mixture diffusion demonstrate that the mixture-diffusion characteristics can be estimated with good accuracy from the pure-component diffusion parameters using the MS diffusion formulation. In the estimation procedure, the binary-exchange parameter D12 is estimated from the pure-component self-exchange coefficients D11 and D22using the interpolation scheme suggested earlier for transport in zeolites (Skoulidas et al. Langmuir 2003, 19, 7977).