A Temperature-Programmed Desorption and IR-Spectroscopic Study

Department of Inorganic and Physical Chemistry, Indian Institute of Science, Bangalore-560012, India. ReceiVed: April 11, 2005; In Final Form: June 16...
0 downloads 0 Views 130KB Size
J. Phys. Chem. B 2005, 109, 15417-15421

15417

Adsorption of n-Hexane in Zeolite-5A: A Temperature-Programmed Desorption and IR-Spectroscopic Study N. Sivasankar and S. Vasudevan* Department of Inorganic and Physical Chemistry, Indian Institute of Science, Bangalore-560012, India ReceiVed: April 11, 2005; In Final Form: June 16, 2005

The adsorption and desorption of n-hexane over Zeolite-5A has been investigated as a function of loading using simultaneous Fourier transform infrared (FTIR)-temperature-programmed desorption (TPD) measurements. The TPD profiles show a second peak developing at lower temperatures when loading exceeds 16 hexane molecules per Zeolite-5A unit cell or two molecules per R-cavity of the Zeolite-5A structure. The infrared spectra rule out two types of adsorption sites as the origin of the two peaks in the TPD. Changes in the conformation of the adsorbed hexane as a function of loading and temperature were followed by monitoring the position of the methylene stretching modes in the infrared spectra. With increasing loading, the adsorbed hexane adopts a stretched trans conformation. These changes occur at loading levels below 12 hexane molecules per Zeolite-5A unit cell. No change is observed above this loading, ruling out any conformational change at loadings where the second peak is seen in the TPD. The second peak in the TPD arises, therefore, from a combination of steric repulsion and loss of translational entropy.

Introduction The molecular sieving action of zeolite solids, with their welldefined channels and cavities, has been widely exploited in separation processes and in the selective cracking and reforming of hydrocarbons.1-6 The dimensions of the channel entrances effectively control entry of reactant molecules and departure of products to those whose molecular dimensions, at least for some conformations, are less than a critical size. Once within the zeolite, a guest molecule can, in principle, enjoy greater conformational freedom, depending on the topology and dimensions of the void space within the zeolite, and which has direct bearing on its subsequent catalytic conversion. Here, we report results of a temperature-programmed desorption (TPD) and spectroscopic study of n-hexane within the pores of Zeolite5A, a molecular sieve catalyst widely used for the separation of linear and branched hydrocarbons to enhance octane values. This zeolite consists of 0.4 nm channels punctuated at regular intervals by 1.14 nm diameter cavities (R-cages). Each unit cell of the Zeolite-A structure has eight such R-cages. The adsorbed hexane has, therefore, a choice of conformations from which to adopt, ranging, at one extreme, from a stiff extended alltrans state (molecular dimension 1.03 nm) to lying curled up within the cavities at the other. In fact, evidence for both possibilities based on computer simulations has been reported from different studies.7-9 The knowledge of the location, conformation, and energetics of n-alkanes in microporous zeolites is fundamental to an understanding of the key steps in various catalytic processes.3-5 In recent years, there has been a resurgence of interest in adsorption and diffusion of linear and branched chain alkanes in zeolites because they have a direct bearing on separation process in the petrochemical industries.6 A large number of theoretical and experimental studies on the adsorption of n-alkanes in zeolites including Zeolite-A have been reported. Adsorption measurements of n-hexane and n-heptane over * Corresponding author. E-mail: [email protected].

Zeolite-5A have shown that up to ∼22 hexane molecules can be accommodated per unit cell (mol/uc) of the zeolite.1,10 Adsorption isotherms were analyzed by the multisite-Langmuir model with inclusion of an attractive sorbate-sorbate interaction term.11 Within the zeolite, the hexane molecules are probably located in the R-cages of the Zeolite-A lattice, which can accommodate up to three molecules as long as the alkane is shorter than C8. The maximum loading, ∼22 hexanes/uc, would correspond roughly to three molecules per R-cage. 2H NMR and Molecular Mechanics/Molecular Dynamics studies of n-hexane in Zeolite-5A at a loading of 8 and 4 mol/uc showed that n-hexane molecules are located close to the wall of the R-cavity and that their conformation was all-trans with the bulky methyl groups pointing in the direction of the windows.7 The simulations also indicated that the n-hexane molecules tend to remain in the all-trans conformation even at higher temperatures. Configuration Bias Monte Carlo (CBMC) calculations8,9 also found that n-alkanes from butane to decane adsorb exclusively in the R-cages of Zeolite-A. The conformation, however, was found to be coiled rather than all-trans. In the literature, the majority of experimental studies on n-alkanes in zeolites have been on adsorption and diffusion measurements. Changes in conformation on adsorption or the conformation of adsorbed molecules have always been considered to be too difficult to study by experimental methods. Most studies of conformation of n-alkanes in zeolites have been by molecular dynamics and simulation techniques.6 This is surprising because vibrational spectroscopy, which has been used extensively in identifying surface species,12 is also a powerful tool for studying conformation. In a recent report,13 the use of FT-Raman spectroscopy to study conformation of n-hexane in zeolites has been highlighted. The authors were able to detect changes in conformation in different zeolites and as a function of temperature. An enhancement of the all-trans conformation on adsorption was observed. Infrared spectroscopy has been widely used for probing the molecular structure and structural changes in the adsorbed

10.1021/jp0518714 CCC: $30.25 © 2005 American Chemical Society Published on Web 07/21/2005

15418 J. Phys. Chem. B, Vol. 109, No. 32, 2005

Sivasankar and Vasudevan

state.12,14 It has also found extensive applications in studying conformation of alkyl chains in the native as well as confined environments.15-18 It is widely recognized that the methylene stretching modes in the infrared are sensitive to the conformation adopted by the chain.15,16 The positions of methylene symmetric, νsym (CH2), and asymmetric, νasym (CH2), modes shift to lower frequencies with increase in trans order. For an all-trans alkyl chain, as in the case of crystalline n-alkanes, the symmetric and antisymmetric stretching frequencies are in the ranges of 28462850 and 2916-2920 cm-1, respectively. With increasing number of gauche conformers, the range shifts to 2855-2865 and 2925-2935 cm-1, respectively. In the present study, we have investigated changes in the state of the adsorbed hexane as a function of loading within the zeolite using temperatureprogrammed desorption while simultaneously monitoring the infrared spectra. The methylene stretching modes in the infrared spectra have been used to monitor changes in conformation of the alkanes on adsorption within the zeolite. Experimental Section Infrared spectra were recorded on a Perkin-Elmer Spectrum2000 FTIR spectrometer in the diffuse reflectance mode using a DRIFT (P/N 19900 series) accessory with a cooled MCT detector. The sample chamber consisted of water-cooled stainless steel block with a zinc selenide window. Sample temperatures could be controlled between 273 and 773 K and could also be ramped at rates between 10 and 20 K/min for temperature-programmed desorption (TPD) measurements. The sample chamber was connected to a gas handling manifold and also on-line to a quadrupole mass spectrometer (SRS QMS300 series gas analyzer). TPD profiles were recorded by monitoring the intensity of the m/e ) 41 fragment of hexane as a function of temperature. Weighed quantities of the zeolite was activated in the infrared sample chamber by heating in a stream of helium at 673 K for 2 h, prior to dosing with known volumes of hexane. Infrared spectra were collected by co-adding 128 scans at a resolution of 4 cm-1. The zeolite, activated under conditions similar to those in the adsorption measurements, was used to record the background spectra. The infrared spectra reported here are, therefore, difference absorbance spectra of the zeolite with and without adsorbed hexane. For the variable-temperature measurements, the background spectra, too, were recorded at different temperatures. The TPD measurements and the infrared spectra were recorded with He carrier gas flow rates maintained at 30 cm3/min. Zeolite-5A was prepared by calcium exchanging Na-Zeolite-A (Fluka; Si/Al ) 1) with 1 M CaCl2 solution for 6 days at 350 K followed by filtering and washing with deionized water. The filtrate was dried for 24 h at 373 K.19 The crystallinity of the zeolite, post ion-exchange, was checked by recording the powder X-ray diffraction patterns (using a Shimadzu XD-D1 X-ray diffractometer). Results and Discussion Temperature-Programmed Desorption. The TPD profiles of n-hexane adsorbed on Zeolite-5A, as a function of loading from 1.7 to 22 mol/uc, are shown in Figure 1. The maximum loading of ∼22 mol/uc is close to the maximum values reported from adsorption measurements.1,10 At the lowest loading, 1.7 mol/uc, the TPD of n-hexane shows a single symmetric profile with the maxima at 542 K. With increase in loading, the TPD maximum shifts to lower temperatures, and at a loading of ∼14 mol/uc, the peak appears at 495 K. Above a loading of ∼15

Figure 1. Temperature-programmed desorption (TPD) of n-hexane from Zeolite-5A as a function of loading. The loading levels are indicated on the right of the figure.

mol/uc, a second peak appears at lower temperature (∼420 K), which grows in intensity with loading. With the appearance of the second peak, there is no further change in the intensity or shift in the position of the hightemperature maxima. The desorption maximum at lower temperature, however, shows a small shift reaching a value of 400 K at maximum loading. At maximum loading, the ratio of the areas under the two peaks is ∼1:2. The observation of two peaks in the TPD of hexane is not unique to the Zeolite-5A system; a similar observation has been reported for the adsorption of n-hexane and n-heptane over the zeolite ZSM-5.20,21 The TPD profile of n-hexane over Silicalite1, the siliceous form of ZSM-5, showed two maxima, 400 and 500 K, at saturated loading (8 mol/uc).21 The presence of the second peak in the TPD profile was attributed to a loss of entropy at higher loading. The two TPD maxima were shown to correspond to desorption from sites with different entropies of adsorption and that the choice of the adsorption site was determined primarily by differences in entropy rather than enthalpy.20,21 Adsorption was favored on sites where the alkane molecules were less constrained. The analyses of the TPD profiles of n-hexane in ZSM-5 zeolite by the single heating variable coverage procedure arrived at similar conclusions.22-24 A decrease in activation entropy of desorption of n-hexane with increasing coverage was observed, indicating that n-hexane molecules suffer a loss of freedom at high loading.22-24 A more dramatic explanation for the two peaks in the TPD of n-hexane and n-heptane over ZSM-5 as a function of loading comes from simulation studies.20,25 At low loadings of n-alkanes (C5-C8) in ZSM-5, the molecules were found to move freely through the pores of the zeolite, that is, the straight channels and intersections. At higher loadings, above 4 mol/uc, however, the C6 and C7 alkanes exhibit an absence of mobility with the molecules apparently “frozen” in the zigzag channels. This behavior is observed only for the C6 and C7 alkanes, and the reason offered by the authors was that the length of these alkanes is commensurate with length of the zigzag channel section of the zeolite ZSM-5. Such an argument based on “commensurate freezing” would, however, not be able to explain our TPD results because it requires that the length of the hexane molecule and the length of a structural unit of the zeolite be commensurate. Considering the structure of Zeolite-5A, it is unlikely that the origin of the two peaks in the TPD is due to desorption

Adsorption of n-Hexane in Zeolite-5A

Figure 2. IR spectra of n-hexane adsorbed on Zeolite-5A as a function of loading. The infrared spectra of n-hexane in solid as well as liquid phases have been given for comparison.

from adsorption sites with differing heats of adsorption. 2H NMR and Molecular Mechanics studies had shown that R-cavity is the preferred adsorption site for n-hexane in Zeolite-5A,7 a conclusion arrived at independently from adsorption isotherm measurements.8 The IR spectra (discussed in the subsequent section) rules out extraframework Ca2+ cations as alternate adsorption sites. The R-cavity can accommodate up to three hexane molecules. The saturation loading level of ∼22 hexane mol/uc corresponds, roughly, to all eight R-cages of the unit cell being fully populated with three hexane molecules. It is interesting to note that the lower temperature second peak in the TPD appears when loading levels exceed ∼15 mol/uc, that is, when occupancy exceeds two hexane molecules per R-cage, assuming uniform distribution among the R-cages. The ratio of the areas of the two peaks in the TPD at maximum loading, ∼1:2, too suggests that the origin of the second peak in the TPD is related to the occupancy of the R cage exceeding two. Infrared Spectra. The state of the adsorbed hexane during the TPD measurements was simultaneously monitored by FTIR spectroscopy. The infrared spectra of n-hexane adsorbed within Zeolite-5A were recorded as a function of loading and temperature. The quality of the spectra in the CH2 bending region is poor, and hence only the methylene stretching regions of the

J. Phys. Chem. B, Vol. 109, No. 32, 2005 15419 spectra are discussed. As was mentioned in the Experimental Section, the spectra were recorded as difference absorbance spectra; consequently, there is no contribution from the framework vibration of the zeolite to the spectra. The methyl and methylene stretching modes of n-hexane adsorbed in Zeolite-5A at different loadings are shown in Figure 2. For comparison, the spectra of n-hexane in the liquid and solid phases are also shown. The solid-phase spectra were obtained by condensing n-hexane vapors on to a KBr disk at 50 K. In this region, the modes, in order of decreasing wavenumber, are νasym (CH3), νasym (CH2), νsym (CH3), and νsym (CH2). In the solid phase, the methyl νasym stretching modes appear at 2960 (Bu) and 2952 cm-1 (Au) and the νsym at 2870 cm-1, while the corresponding methylene modes appear at 2920 and 2851 cm-1, respectively. It may be seen that the spectrum of the adsorbed hexane is intermediate between that of the spectra in the solid and liquid phases. With increasing loading, the methylene symmetric stretch moves to lower wavenumbers so that the methyl symmetric stretch is more clearly seen and the spectrum of the adsorbed n-hexane start resembling that of the solid. At maximum loading (22 mol/uc), the positions of the bands are 2965 (2967), 2928 (2937), 2870 (2879), and 2855 (2867) cm-1, respectively (the figures in parentheses are the corresponding gas-phase values). The infrared spectra show no significant change in the position, line-widths, or relative intensities of either the methyl or the methylene stretching modes at loadings at which the second peak appears in the TPD. It is, therefore, unlikely that the two peaks in the TPD are due to desorption of hexane molecules adsorbed on two types of site of chemically different nature. The fact that the frequencies are close to that of hexane solid/liquid rules out the possibility that the extraframework Ca2+ cations are the adsorption sites. Variable-temperature infrared spectra of n-hexane adsorbed on Zeolite-5A were recorded in the temperature range 298613 K. The spectra in the methyl and methylene stretching region are shown in Figure 3a. The loading at 298 K is 22 mol/ uc. The methylene modes, especially the symmetric stretch mode, show a shift to higher frequencies with increasing temperature and at higher temperatures are merged with the methyl symmetric modes. The positions of the methyl stretching modes show no significant change with temperature. The infrared intensities show a steady decrease with temperature and are completely absent by 613 K. This is a straightforward

Figure 3. (a) Variable-temperature IR spectra in the C-H stretching region of n-hexane adsorbed in Zeolite-5A in the temperature range 298-613 K. (b) Variation in the intensity of the νasym (CH2) stretching mode of n-hexane adsorbed in Zeolite-5A as a function of temperature. The derivative is shown as a dotted line.

15420 J. Phys. Chem. B, Vol. 109, No. 32, 2005

Figure 4. Lorentzian decomposition of the methylene and methyl bands of n-hexane adsorbed in Zeolite-5A. The individual components are shown as dotted lines.

reflection of the fact that the concentration of adsorbed hexane decreases with temperature due to desorption. The intensity, I, of the methylene asymmetric stretching (νasym) mode at 2928 cm-1 as a function of temperature is plotted in Figure 3b along with the differential, [dI/dT]. The derivative plot shows that the decrease in intensity is two-step. The temperatures, 400 and 495 K, are in reasonable agreement with the desorption maxima in the TPD profile for hexane adsorbed at maximum loading (Figure 1). The spectrum of hexane adsorbed in Zeolite-5A shows a shift in the position of the conformationally sensitive methylene symmetric (νsym) and antisymmetric (νasym) stretching modes to lower frequencies with loading and to higher frequencies with temperature. The positions of the methylene νsym and νasym modes for hexane adsorbed in Zeolite-5A at different loadings were obtained from a decomposition of the individual spectrum as a sum of Lorentzians (Figure 4). The peak shapes and positions of the methyl νsym and νasym modes of the adsorbed hexane were kept identical to that of hexane in the solid phase in the decomposition procedure. The rationale for doing so is that, unlike the methylene modes, the methyl modes are insensitive to the conformation of the alkane chain and in the solid-phase all modes are well resolved (Figure 2). The peak shapes and positions of the methyl νsym and νasym modes were obtained from the spectra of solid hexane by a similar decomposition procedure (see Supporting Information). The variation in the position of the fitted methylene stretching frequencies with loading is shown in Figure 5. It may be seen that the positions of methylene stretching (both νsym and νasym) modes shift to lower wavenumber with increased loading until a loading of ∼11 mol/uc, above which there are no changes in band positions with loading. This shift in the methylene stretching modes to lower wavenumbers indicates increased trans order with loading. The frequencies of the νsym and νasym modes of the adsorbed hexane at a loading of >12mol/uc, 2847 and 2924 cm-1, are close to that of solid hexane, 2851 and 2920 cm-1. It is interesting to note that at loading levels at which the second peak appears in the TPD profile (indicated by an arrow in Figure 5), the position of the methylene νasym and νsym modes shows no change. The analysis of the IR spectra clearly indicates that there are no conformational changes associated with the appearance of the second peak in the TPD at loading levels above 15 mol/uc. Changes in conformation of n-hexane adsorbed in Zeolite-5A do occur, but at a much lower loading, below ∼11 mol/uc.

Sivasankar and Vasudevan

Figure 5. Variation in the CH2 mode frequencies of adsorbed n-hexane with loading. The arrows indicate the loading level at which the second peak appears in the TPD.

Figure 6. Variation in CH2 stretching mode frequencies of adsorbed n-hexane in Zeolite-5A with temperature. The derivative of the intensity of the νasym (CH2) stretching mode with temperature is shown as a dotted line.

The explanation for these observations is that at low loadings sufficient “free” volume is available for hexane molecules in the R-cage of the Zeolite-5A lattice to adopt different conformations. At higher loading, the “free volume” available per hexane molecule is reduced and is forced to adopt a more “transordered” conformation. This increase in trans conformation is more significant at lower loading levels. The shift in the TPD maxima toward lower temperature with loading in the range 1-15 mol/uc is also an indication of hexane-hexane steric interactions. A similar analysis of the variable-temperature spectra of adsorbed n-hexane was carried out (Figure 6). The spectra were analyzed as a sum of Lorentzians, keeping the positions and profiles of the methyl stretching modes identical to that in the solid. The spectra could be analyzed only below 523 K because at high temperature the intensity drops considerably due to desorption and hence the quality of the fit is poor. The positions of the bands, as obtained from the fit, have been plotted as a function of temperature in Figure 6. The derivative of the intensity of the νasym (CH2) mode is also shown in Figure 6. It may be seen that up to 333 K, there is a decrease in frequencies, but above 333 K, once desorption commences, there is a shift to higher wavenumbers, which is an indication of an increase in gauche conformations. This is probably a consequence of the increased space available per hexane molecule and the higher temperature.

Adsorption of n-Hexane in Zeolite-5A Conclusions The adsorption and desorption of n-hexane over Zeolite-A has been studied as a function of hexane loading using simultaneous Fourier transform infrared (FTIR) and temperatureprogrammed desorption (TPD) measurements. The TPD profiles show systematic changes with loading. At loadings less than ∼16 mol/uc, which corresponds, if distributions were uniform, to a maximum of two hexane molecules per R-cages (there are eight such cages per unit cell of the Zeolite-A structure), the TPD shows a single maximum that shifts to lower temperature with loading. Above a loading of ∼16 mol/uc, a second maxima appears at lower temperature, and it is this feature that grows in intensity with loading. At the highest loading, the ratio of the area of the two features is ∼1:2. The infrared spectra show no evidence for adsorption of hexane on two different types of sites, ruling out this possibility as the origin of the two features in the TPD. The positions of the conformation sensitive methylene stretching modes in the infrared were used to monitor changes in conformation of the adsorbed hexane as a function of loading and temperature. With increasing loading, the adsorbed hexane adopts a more transordered conformation, but above a loading of ∼11 mol/uc, there is no change in the positions of the methylene stretching modes. At this loading, the frequencies are close to that of solid hexane, suggesting that the adsorbed hexane is in a trans stretched conformation. The infrared spectra rule out any significant change in conformation of the adsorbed hexane at the loading above which the second maximum appears in the TPD. Having ruled out multisite adsorption, the question remains as to the origin of the second peak observed in the TPD of n-hexane over Zeolite-5A. The second peak is, therefore, likely to be associated, at least in part, with a loss in the entropy of adsorption at higher loading and in part due to a decrease in enthalpy of adsorption due to steric repulsion. A decrease in the entropy of adsorption has indeed been proposed in the literature for modeling adsorption of C4-C8 n-alkane in silicalite-1.20,21,24,25 Any model would have to explain why the TPD profile shows a dramatic change when the occupancy of the R-cages exceeds two, whereas up to a loading of two hexanes per cage the change is more or less what may be anticipated the TPD maxima shifting to lower temperatures. Our results suggest that a possible reason changes in the TPD are more gradual at lower loadings is that part of the crowding effect is alleviated by the adsorbed hexane molecules adopting the stretched trans conformation that would allow hexane-hexane interactions to be more favorable. Analysis of adsorption isotherms of hexane over Zeolite-A has indeed invoked such interactions.10,11 At a loading of two hexane per R-cage, the infrared spectra indicate that all hexane molecules have attained this conformation and no further changes are observed. Further loss in the entropy of adsorption at higher loading would be

J. Phys. Chem. B, Vol. 109, No. 32, 2005 15421 due to loss of translational mobility. This is understandable; because the maximum permissible occupancy per R-cage is three hexanes, any cage having this occupancy would block all molecular traffic through it. A combination of increased steric repulsion along with loss of translational mobility is the origin of the second peak in the TPD of hexane over Zeolite-A at loadings exceeding two molecules per R-cage. Acknowledgment. This work was supported by the Department of Science and Technology, Government of India. N.S. thanks CSIR, New Delhi, for a fellowship. Supporting Information Available: (1) FTIR spectra of n-hexane adsorbed over Zeolite-A. (2) Lorentzian decomposition of the methylene and methyl bands of solid n-hexane. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Breck, D. W. Zeolite Molecular SieVes; John Wiley and Sons: New York, 1974. (2) Barrer, R. M. Zeolites and Clay Minerals as Sorbents and Molecular SieVes; Academic Press: New York, 1978. (3) Ruthven, D. M. Principle of Adsorption and Adsorption Processes; John Willey: New York, 1984. (4) Thomas, J. M.; Thomas, W. J. Principles and Practice of Heterogeneous Catalysis; VCH Publishers Inc.: New York, 1997. (5) Schirmer, W.; Fiedrich, G.; Grossmann, A.; Stach, H. First international Conference on Molecular sieVes; London, 1967. (6) Bates, S. P.; van Santen, R. A. AdV. Catal. 1998, 42, 1. (7) Zaborowski, E.; Zimmermann, H.; Vega, S. J. Am. Chem. Soc. 1998, 120, 8113. (8) Bates, S. P.; van Well, W. J. M.; van Santen, R. A.; Smit, B. J. Phys. Chem. 1996, 100, 17573. (9) Bates, S. P.; Well, W. J. M.; van Santen, R. A.; Smit, B. J. Am. Chem. Soc. 1996, 118, 6753. (10) Eberly, P. E. Ind. Eng. Chem. Res. 1969, 8, 140. (11) C. Silva, J. A.; Rodrigues, A. E. AIChE J. 1997, 43, 2524. (12) Ryczkowski, J. Catal. Today 2001, 68, 263. (13) Huang, Y.; Wang, H. Langmuir 2003, 19, 9706. (14) Tolstoy, V. P.; Chernyshova, I. V.; Skryshevsky, V. A. Handbook of Infrared Spectroscopy of Ultrathin Films; John Wiley and Sons: New Jersey, 2003. (15) MacPhail, R. A.; Straus, H. L.; Snyder, R. G.; Elliger, C. A. J. Phys. Chem. 1984, 88, 334. (16) Snyder, R. G.; Strauss, H. L.; Elliger, C. A. J. Phys. Chem. 1982, 86, 5145. (17) Venkataraman, N. V.; Vasudevan, S. J. Phys. Chem. B 2000, 104, 11179. (18) Venkataraman, N. V.; Vasudevan, S. J. Phys. Chem. B 2001, 105, 1805. (19) Pluth, J. J.; Smith, J. V. J. Am. Chem. Soc. 1983, 105, 1992. (20) van Well, W. J. M.; Wolthuizen, J. P.; Smit, B.; van Hoof, J. H. C.; van Santen, R. A. Angew. Chem., Int. Ed. Engl. 1995, 34, 2543. (21) Millot, B.; Methivier, A.; Jobic, H. J. Phys. Chem. B 1998, 102, 3210. (22) Chen, L. F.; Rees, L. V. C. Zeolites 1988, 8, 310. (23) Song, L.; Rees, L. V. C. J. Chem. Soc., Faraday Trans. 1997, 93, 649. (24) Olson, D. H.; Reischman, P. T. Zeolites 1996, 17, 434. (25) Smit, B.; Maesen, T. L. M. Nature 1995, 374, 42.