MOLECULAR-DYNAMICS SIMULATION OF PROPANE AND METHANE IN SILICALITE

We have investigated the diffusion of propane and methane in the molec ular sieve silicalite by computer simulation using energy minimization and molecular dynamics techniques. We present heats of adsorption and self-diffusion constants, calculated using four sets of nonbonded int eractions, and compare them to experimental values. Extensive simulati on results for large ensembles of methanes in a rigid-molecule approxi mation are presented. Methane is studied at infinite dilution and at l oadings of 2, 4, 8, 12, and 16 molecules per unit cell. Theoretical se lf-diffusion constants range from 11.5 x 10(-5) to 2.0 X 10(-5) cm2/s at 300 K, in excellent agreement with pulsed field-gradient spin-echo nuclear magnetic resonance measurements. Simulations of propane allowe d free movement of all the internal coordinates of the molecule and in corporated large ensembles to achieve accurate representations of bulk properties. Propane is studied at infinite dilution and loadings of 4 and 12 molecules per unit cell. The corresponding theoretical self-di ffusion constants are 2.3 x 10(-5) and 6.0 x 10(-7) cm2/s at 300 K. Th ese simulated diffusion rates are also in excellent agreement with NMR measurements. Center-of-mass time distributions were calculated and e nergy minimizations of the molecules within the zeolite lattice were d one. These analyses show that the zigzag channels are the favored resi dence sites for both methane and propane. The calculated isosteric hea ts of adsorption of methane and propane are -5.8 and -10.3 kcal/mol, r espectively, in good agreement with experimental values. Both adsorbat e-silicalite and adsorbate-adsorbate interactions are shown to have an effect on the packing of the molecules in the zeolite. In addition, d ynamic effects related to the anisotropy of diffusion in the silicalit e lattice also have a role in packing. The microscopic diffusive behav ior of the molecules is dicussed and compared to the classical jump di ffusion model.