Silicalite alkanes

Vlugt T J H, Krishna R and Smit B 1999 Molecular simulations of adsorption isotherms for linear and branched alkanes and their mixtures in silicalite J. Phys. Ohem. B 103 1102-18... 

Fig. 8.22 Schetnatic structure of the zeolite silicalite showing the straight and zig-zag chaimels. (Figure adapted fron Smit B and JI Siepmann 2994. Simulating the Adsorption of Alkanes in Zeolites. Science 264 1118-1120.)...

Liu et al. [42] reported permeation of mixture of hexanes and octanes through silicalite membranes. It was found that the permeances of the mixture could not be predicted by the single-component data. In the separation of alkane isomers, the permeance of 2,2-DMB is significantly reduced in the presence of n-hexane resulting in a permselectivity much higher than the ideal separation factor [7]. 

Zhu, W., Kapteijn, F., and Moulijn, J.A. (2000) Adsorption of light alkanes on silicalite-1 reconciliation of experimental data and molecular simulations. Phy. Chem. Chem. Phys., 2,1989-1995. 

We now consider the diffusion of other hydrocarbons. Most calculations have been performed for n-alkanes, up to and including n-hexane, but alkenes and alkynes have also been considered. Calculations involving larger molecules have been mainly restricted to silicalite. 

Dumont and Bougeard (68, 69) reported MD calculations of the diffusion of n-alkanes up to propane as well as ethene and ethyne in silicalite. Thirteen independent sets of 4 molecules per unit cell were considered, to bolster the statistics of the results. The framework was held rigid, but the hydrocarbon molecules were flexible. The internal coordinates that were allowed to vary were as follows bond stretching, planar angular deformation, linear bending (ethyne), out-of-plane bending (ethene), and bond torsion. The potential parameters governing intermolecular interactions were optimized to reproduce infrared spectra (68). 

June et al. (85) presented united-atom calculations for butane and for hexane in silicalite, whereby the bond and dihedral angles of the alkanes were allowed to vary. In addition, the calculation of hexane took account of an additional intramolecular Lennard-Jones potential for nonbonded atoms more than three bonds apart (which prevents the alkane crossing over itself). The interaction parameters for the alkane molecules were taken from Ryckaert and Bellmans (3), and those governing the interaction of the alkanes with the zeolite from a previous study of the low-occupancy sorption of alkanes in silicalite (87). Variable loadings of alkanes were considered from 1 to 8 molecules per unit cell were considered, and calculations were allowed to run for 500 ps for diffusion at 300 K. 

A large proportion of the work described in this section refers to n-alkane sorbates however, other molecules such as alkenes and alcohols have also been considered. Most interest has been focused on the silicalite/ ZSM-5 system. 

The configuration-bias Monte Carlo (CB-MC) technique (112) has also been extensively applied to characterize the sorption of alkanes, principally in silicalite (111, 156, 168-171) but also in other zeolites (172-174). Smit and Siepmann (111, 168) presented a thorough study of the energetics, location, and conformations of alkanes from n-butane to n-dodecane in silicalite at room temperature. A loading of infinite dilution was simulated, based on a united-atom model of the alkanes and a zeolite simulation box of 16 unit cells. Potential parameters were very similar to those used in the MD study of June et al. (85). As expected, the static properties (heat of adsorption, Henry s law coefficient) determined from the CB-MC simulations are therefore in close agreement with the values of June et al. The... 

Fig. 10. Probabilities of finding an n-alkane in each of the three pore regions of silicalite as a function of carbon number, Nc. Reprinted with permission from Ref. 111. Copyright 1994 American Chemical Society.

The efficiency of the CB-MC technique has been used by Maginn et al. (769), who considered the low-occupancy thermodynamics of sorption of alkanes as long as C25 in silicalite. The locations of such long molecules are no longer correctly predicted by considering the end-to-end vector and the chain midpoint. To overcome this problem, a coarse-graining technique was used to describe both the adsorbate and the zeolite, allowing for accurate microscopic characterization. 

The CB-MC method has been used to simulate the adsorption isotherms of various alkanes in silicalite (170, 171). Using potential parameters that were fitted to obtain good agreement with experimental Henry s law coefficients, Smit and Maesen (170,171) have simulated the adsorption isotherms of straight-chain alkanes in silicalite. Good agreement was obtained for ethane and propane in comparison with the different type-I curves measured experimentally. The overall agreement with experimental isotherms was found to be satisfactory with hexane and heptane, and a kink is seen... 

Bates et al. (172-174) considered the energetics, locations, and conformations of alkanes ranging from butane to decane in a variety of different all-silica zeolites. Calculations similar to those described already were performed for alkanes in mordenite, zeolite rho, faujasite, ferrierite, and zeolite A. A linear increase in the calculated heat of adsorption with increasing carbon number was found for all zeolites. Less experimental information is available to compare with the calculated heats of adsorption, and thus the performance of the technique and parameters cannot be subjected to quite the same scrutiny as the results for silicalite (111). Nonetheless, where... 

Figure6.4 Simulated structure ofthe MFI-type zeolite Silicalite-1 (O atoms in dark gray and Si atoms light gray), projected on the be plane and showing the zigzag channels. The broken lines indicate the periodic cell boundaries. Alkane molecules (not drawn to scale) are indicated by black circles. Thanks to Dr. Merijn Schenk for the zeolite picture.

Krishna and Paschek [91] employed the Maxwell-Stefan description for mass transport of alkanes through silicalite membranes, but did not consider more complex (e.g., unsaturated or branched) hydrocarbons. Kapteijn et al. [92] and Bakker et al. [93] applied the Maxwell-Stefan model for hydrocarbon permeation through silicalite membranes. Flanders et al. [94] studied separation of C6 isomers by pervaporation through ZSM-5 membranes and found that separation was due to shape selectivity. 

Figure 1.6. Zero coverage energy of adsorption of n-alkanes versus carbon number Nc, on three microporous solids (and two others, for comparison). Molecular sieve carbon, after Carrott and Sing (1987) Silicalite after Canott and Sing (1986) NaX and macroporous silica, after Kiselev (1967) non-porous carbon, after Carrott and Sing (1987) and Avgul and Kiselev (1965).

The effect of the loading on the adsorption energy in silicalite is shown in figure 1, for the linear C1-C4 alkanes and also for isobutane. The adsorption energy for the linear alkanes raises slightly with increasing the loading. For isobutane it remains almost constant... 

Figure 2 shows the effect of the ZSM-5 framework relaxation on the adsorption energy of the linear alkanes. Contrary to what is observed for silicalite, the effect of the zeolite relaxation is noticeable even for the smaller alkanes. That implies the use of specific force fields to represent the ZSM-5 structure in order to avoid its colapse during the simulations. 

---