The significant influence of interfacial resistance on the overall
mass transfer performance, especially for catalysis with zeolites,
has attracted a lot of attention, but the understanding on entering-pore
processes is still very limited in comparison with that on intracrystalline
diffusion processes. Evaluating entering probabilities still depends
frequently on the expressions derived from hard sphere (HS) simulations
where substantial approximations exist. In this work, molecular dynamics
simulations have been performed to elucidate the interfacial barriers
involved in methane and p-xylene (PX) molecules entering
the ZSM-5 zeolite. The entering probabilities have been derived accordingly,
along with the occupancy distributions and the residence time distributions
of the incident molecules. The results have been compared with those
of HS simulations in the literature as well. It is observed that the
occupancy distribution of these two species exhibits peaks along the
entering paths, reflecting the apparent resistance due to adsorption,
which cannot be revealed through HS simulations. Bimodal distributions
have also been observed, which can be attributed to the coexistence
of the separate adsorption and entering-pore processes. With the increase
in temperature, such peaks tend to level off, approaching the results
of HS simulations. The entering probabilities of these two species
approach the results from HS simulations as well at high temperatures.
Because the entering probabilities of the two species show different
decreasing slopes with temperature, at low temperatures, the entering
probability of PX molecules can be higher than that of methane molecules.
However, the entering rate of the PX molecules is always quite lower
than that of the methane molecules because of its much longer residence
time. Such phenomena have been explained on the basis of entropic
and energetic effects. Based on a pseudo-reaction model, the results
of the entering probabilities have been well fitted, and an expression
has been proposed accordingly, which is expected to be applied to
large-scale models. Because such an expression is developed on a more
sophisticated ground than HS simulations, its applications might lead
to more satisfactory results.