A fundamental understanding of the processes affecting fluid behavior in nanoscale confinements is crucial to numerous existing and emerging applications in nanotechnology, materials science, biology, adsorption and membrane transport, as well as a host of other areas. In the last two decades a vast array of new nanoporous materials such as templated periodic MCM-41 silicas, carbon nanotubes as well as various aluminophosphates have been developed, all considered to hold promise for a variety of novel applications. The infiltration of fluid mixtures into the nanopores in these materials is a common feature of most applications being investigated, and this has catalyzed several new developments in the understanding of fluid equilibrium and transport at the nanoscale, most of which have, however, been devoted to pure component systems, with little attention to mixture transport.

For long, models of mixture transport have been based on highly respected approaches such as the dusty gas model, or statistical mechanical treatments such as that of Bearman and Kirkwood. However, they have failed to provide satisfactory solutions, with no definitive treatment even for a classical experiment such as the Stefan tube. We present here a novel theory of mixture transport in nanopores, which represents wall effects via a species-specific friction coefficient determined by its low density diffusion coefficient. This low density diffusion coefficient for each species may be obtained experimentally, or from simulation. Further, the treatment uses a frame of reference based on the individual species velocity, rather than the commonly used mass averaged mixture velocity. Good agreement of the predicted Onsager coefficients with those from molecular dynamics simulations is demonstrated in the mesopore range, while considering inhomogeneity of the density distributions. It is found that the commonly used assumption of a cross-sectionally uniform density in the momentum balance is in serious error, as also the traditional use of a centre of mass based frame of reference. The theory also shows how wall effects can be unambiguously considered through the friction coefficient, avoiding the commonly used arbitrary superposition of viscous and diffusive contributions.