The single and mixed sorbed phase dynamics of methane (CH 4 ) and carbon dioxide (CO 2 ) in the two independent pore networks, the large cavity (LC) and sinusoidal channel (SC), of the ITQ-1 zeolite (MWW framework type) are studied by means of atomistic computer simulations. Equilibrium molecular dynamics predicts different loading dependences for the two gases, for both the self-and the collective diffusion coefficients. In particular, the self-diffusivity of pure CO 2 exhibits the decreasing trend with loading which is usually observed in nanoporous materials, whereas the self-diffusivity of pure CH 4 displays a peculiar loading dependence which goes through a maximum if we consider only the CH 4 sorbed within the LC pore network of the ITQ-1 unit cell. From a qualitative perspective, the different loading dependence of the self-diffusivity for the two pure sorbates is attributed to their different probability density distribution inside the supercages of the LC pore network. Besides, the composition dependence of the self-diffusivity for each sorbate of the binary mixture can be explained in terms of the sorption thermodynamic selectivity for CO 2 in the zeolite.The loading dependence of the collective diffusivity comes from the evolution with loading of the self-diffusivity of the individual sorbed molecules and the many-body nature of the center of mass dynamics.To gain insight in the phenomena leading to the maximum in selfdiffusivity for pure CH 4 inside the LC of ITQ-1, we develop a new "mesoscopic" stochastic model which can describe the diffusive behavior of a system of particles at equilibrium. The model is based on discretizing space into slabs by drawing equispaced parallel planes along a coordinate direc-
VII VIIIABSTRACT tion. A central role is played by the probability that a particle exits a slab via the face opposite to the one through which it entered (transmission probability), as opposed to exiting via the same face through which it entered (reflection probability). A simple second-order Markov process model invoking this probability is presented, leading to an expression for the selfdiffusivity, applicable for large slab widths, consistent with a continuous formulation of diffusional motion. The model is validated via molecular dynamics simulations in a bulk system of soft spheres across a wide range of densities.At this point, the concentration dependence of self-diffusivity within periodic media can be investigated by implementing the second-order Markov process model into a method, called diffusion through space discretization (DSD). Introducing the concept of minimum crossing surface, we obtain a unique decomposition of the self diffusion coefficient into two parameters with specific physical meanings. Two case studies showing a maximum in self-diffusivity as a function of concentration are investigated, along with two cases where such a maximum cannot be present.Finally, the DSD method is applied to the LC pore network of the ITQ-1 zeolite, for CH 4 and CO 2 , explaining the diffusivity...