Adsorption in zeolites: intermolecular interactions and computer simulation

The input to a simulation is the potential energy model. In microporous solids, the adsorbate adsorbent interaction is the most significant part of the total potential energy. The full scale semi-empirical PN model is summarised as an example of a state of the art potential model. The potential includes damped two-body dispersion interactions, three-body interactions, induced interactions (often negligibly small) and long-range electrostatic interactions. Repulsive interaction is modelled as a Born-Mayer function. All the contributions can be calculated from a knowledge of four parameters that depend on partial charges at sites. Of these the dipole polarizabilities and excitation energies can be calculated from independent information. A guide to the magnitude of the repulsive parameters A and b comes from quantum mechanical calculations, and these are then tuned against experiment. They maintain a high degree of transferability through combination rules. Several systems are discussed as examples. For argon adsorption in silicalite at 77 K it is found that a transition, similar in magnitude to that observed experimentally, occurs in the simulation, but only at very much higher pressure. Detailed evidence suggests that adsorbent lattice distortion, not permitted in the simulation, may occur in the real system. A similar conclusion applies to nitrogen in silicalite at 77 K, where the first, but not the second, of the experimental transitions is observed in simulation. In AlPO(4)-5, both argon and methane adsorption have been studied, good agreement with experiment can be obtained when methane is modelled as a 20-6, rather than a 12-6, molecule, and it seems improbable that lattice distortion occurs in these systems. A third zeolite studied using the PN potential is NaY. Here xenon, methane and p- and m-xylene adsorption have been simulated. Good agreement with experiment is attained for the first two adsorbates, but it is again found to be necessary to use a more realistic model than the 12-6 potential for methane interactions. A notable conclusion is that Na migration may occur at high loading of p-xylene. In many of these examples, comparison with the frequently used oxygen-only potentials are made. It is concluded that full-scale modelling of the PN type is beneficial in the characterisation of microporous zeolites. Nevertheless, full transferability cannot be claimed for the PN model in its present form. (C) 1998 Elsevier Science B.V. All rights reserved.