J. Phys. Chem. 1992,96, 10914-10917
10914
Study of ZSM-5 and ZSM-11 Zeolites by '*'Xe NMR: Effect of SVAI Ratio Q.Chen,+ M. A. Spdnguel-Hueft J. Fraissard,*.t Matthew L. Smith,* David R. Corbin,l and Cecil Dybowski*~* Laboratoire de Chimie des Surfaces, CNRS URA 14 28, Universitl Pierre et Marie Curie, 4 Place Jussieu, 75252 Paris Cedex 05, France, Department of Chemistry and Biochemistry and Center for Catalytic Science and Technology, University of Delaware, Newark, Delaware 1971 6, and Central Research and Development,I Experimental Station, Du Pont Company, Wilmington, Delaware 19880-0262 (Received: May 7, 1992; In Final Form: September 8, 1992)
The NMR spectroscopy of xenon sorbed in pentasil zeolites of the MFI (ZSM-5) and MEL (ZSM-11) structures is different from that of xenon adsorbed in the faujasite structure. The dependence of the xenon NMR chemical shift and spin-lattice relaxation rate on Si/Al ratio at 298 K points out a discontinuity in the surface properties at around 2 Allunit cell and then a nonrandom distribution of aluminum atoms in the lattice.
Introduction The NMR of xenon in porous solids has been extensively studied and, in particular for zeolites, the dependence on properties such as location of compensating cations, short-range crystallinity, size of supported metal particles, and porous free volume is now well established.' It has been shown that, in the case of the faujasite structure, the effect of Si/Al ratio on the chemical shift, 6, of adsorbed xenon is very small at 298 K for decationized zeolites or zeolites with Na+ cationsS2The chemical shift extrapolated to zero xenon concentration decreases by about 4 ppm as Si/Al ratio increases from 1.2 to 54. By extrapolating this small effect to zaolita of various structures, it has been possible to relate the shift quantitatively to Xtswface interactions, and especially the X t c a g e (or Xe-channel wall) interactions, Le., to the geometric characteristics of the free volume. This small effect of the Si/Al ratio has recently been confumed for faujasitic mlites and ZSM-5 ~ e o l i t e s . ~In * ~this J paper, we report NMR chemical shifts and relaxation times for '29Xe adsorbed in ZSM-5 and ZSM-11 zeolites and how these depend on Si/Al ratio.
Experimental Section Zeolite Synthesis. Na,TPA-ZSM-5 and Na,TBA-ZSM- 1 1 [where TPA = tetrapropylammonium and TBA = tetrabutylammonium] were prepared by the method of Rollman and VolyocsikS6The crystallinity of the materials was verified by X-ray powder diffraction using a Phillips APD 3600 or APD 3720 diffractometer. Scanning electron micrographs showed that all samples of ZSM-5 have estimated crystallite sizes from 1 to 2.5 Mm. Si/Al ratios were determined by inductively coupled plasma spectrometry.' Na,TPA-ZSM-5 and Na,TBA-ZSM-1 1 were calcined in flowing air at 60 K/h to 823 K for 10 h to remove the organic templates. In this manner, samples of Na-ZSM-5 with Si/Al ratios of 24, 34, 67, and 490 and Na-ZSM-11 with Si/Al ratios of 23,33,63, 128, and 680 were prepared. Samples of NH4-ZSM-5 were made by ion exchange of the sodiated materials. Calcination of the NH4-ZSM-5 samples under the calcination conditions described above produced H-ZSM-5 samples with Si/Al ratios of 24, 128, and 490. All samples will be denoted in the text by the names followed by the Si/Al ratio. Adsorption. The samples were carried through the following procedure: After an initial outgassing to 1 X Torr at 298 K for 24 h, each sample was outgassed for 40 h at 673 (f3) K. Then xenon adsorption isotherms were determined at 298 K on a chemical volumetric apparatus. The parameter n is the number of atoms taken up per gram of zeolite. NMR Specbwopy. We use the extrapolated shift of bulk xenon at zero density as the reference for these measurements.* Universitd Pierre et Marie Curie. *University of Delaware. Du Pont Co. Contribution No. 5867.
All resonances shifted to higher frequency than this reference are positive. Spectroscopy was performed with a Bruker CXP 100 and WM 250 NMR spectrometers operating at 24.9 and 69.19 MHz. The pulse delay ranged between 0.5 and 2.0 s, depending on the relaxation time of xenon in the sample. The uncertainty in the reported shifts is estimated to be f0.5 ppm. The relaxation data was acquired using the inversion-recovery experiment and the data were fit to a three-parameter singleexponential function using a simplex algorithm. The influence of low compression of pure zeolites on xenon chemical shift 6 is negligible.9 We have taken care to maintain the samples in powder form at similar packing densities. On the other hand, the influence of temperature on 6 is sometimes very important.1° We have evaluated the dependence of chemical shift at 298 K on Si/Al ratio. We have tested the influence of temperature on the signal shape for detecting some structural defects.
Results Isotherms. The isotherms for ZSM-5 expressed in logarithmic coordinates (Figure 1) are af the classical form for xenon adsorption in zeolites: a linear increase of log n with log P at small values of the pressure P, the uptake tending toward saturation at pressures approaching 1000 Torr. As with NaY zeolites, for P I 1000 Torr, the amount of xenon adsorbed increases with aluminum concentration (i.e., with increasing Na+ concentration). All samples except ZSM-5-1950 show the same saturation value of 1.4 X lozoatoms g-I. The isotherms for ZSM-11 are very similar to those for ZSM-5 (Figure 2). They reach saturation at about the same value of n. The effect of the Si/Al ratio on the isotherms can be seen by plotting the variation of the uptake at 5 or 10 Torr versus the number of aluminum atoms per unit cell, [AI],,, as in Figure 3. Such plots are linear with a sharp discontinuity at about [All,, = 2. NMR Spectroscopy. ZSM-5. The Iz9XeNMR spectra for ZSM-5-based samples consist of a single signal under all conditions, except for ZSM-5-67. The spectra of ZSM-5-67 show a single signal for T I268 K, but a shoulder appears at high pressures for T > 268 K,being increasingly well resolved at higher temperatures. The width of the signal, Av,depends on [Al],,. It increased slightly with n for [All,, < 2. For [AI],, > 2, Av is much greater at low xenon concentrations and decreases as n increases in the range 1 X lozo < n < 1.1 X loz1(Figure 4). The Iz9XeNMR chemical-shift variation with n is similar for all ZSM-5 samples, except ZSM-5-1950 (Figure 4). The main difference among the other samples is a slight difference in the extrapolated chemical shift at infinitely low xenon uptake, 6, (Figure 5). 6, increases linearly with [All,,, with a sharp discontinuity at about [AI], = 2. For sample ZSM-5-1950, the value of 6, fits with the other samples; however, the slope of the curve of chemical shift versus n is much steeper than for the other materials.
-
0022-3654/92/2096-109 14$03.00/0 0 1992 American Chemical Society
The Journal of Physical Chemistry, Vo1. 96, No. 26, 1992 10915
IZ9XeNMR of ZSM-5 and ZSM-11 Zeolites
-
Xe atomlg
ZSM 5
t
PPm
150 140 130
Av PPm 120 10
1
110
PXe (Torr)
100
.30
Figure 1. Uptake of xenon in Na-ZSM-5 zeolites at 298 K. Si/AI ratios: ( 0 )24; (A) 34; (V) 67; (B)120; (*)490; (+) >1950.
100
Xe atomlg
Figure 4. '29Xechemical shift (full line) and line width (dotted line) as a function of n in Na-ZSM-5 at 298 K. Si/Al ratios: (0)24; (A) 34; (V) 67; (0)120; (+) 490; ( 0 )>1950.
6s PPm
1019l
1
.
/
l...ll.l
.
. ....--.. .
. . . . . I . .
10
Pxe CroW
100
Figure 2. Uptake of xenon by Na-ZSM-11 at 298 K. Si/A1 ratios: (0) 30; (A) 40; (*) 80; (W) 160; (*) 01.
Xe at.!
-1
Figure 5. The limiting chemical shift, I,, at low xenon concentration as a function of [All,, for Na-ZSM-5 (0)and Na-ZSM-11 (A)at T = 298 K. The error in the shift is h0.5 ppm.
1
I 5
OO
sorbed in H-ZSM-5 as a function of n. Si/Al ratios: (+) 490.
U.C.
Figure 3. Xenon uptake versus [AI],, for ZSM-5 (0)and ZSM-11 (A) at T = 298 K. Full line: Px, = 10 Torr. Dotted line: Px, = 5 Torr.
The uptake-dependent parts of the xenon chemical shift may be characterized by a virial expansion in n 6(n,T) - 6, = Bln +
15
20
1020 Xe atomslg Figure 6. Uptake-dependent relaxation rate, 1/T, - l/Tl0, of 129Xe
0 1.0 2.0 3.0 AI I
10
+ ...
(1)
where bI(r ) and 62( 2") are virial coefficients. From the variation of the chemical shift, we find the following parameters: a1(298) = 1.5 (f0.1) X (ppm &/atom. and ti2(298) = 1.2 (10.2) X (ppm gz)/atomzfor all samples except ZSM-5-1950. The fact that the reduced chemical shift has a common dependence on n indicate that xenon-xenon collisionsaffect the NMR spectral parameters similarly in all of these samples, regardless of Si/Al ratio or identity of the cation (H+ or Na+).
(W)
24; (A) 120;
TABLE I: Limiting Relaxation Rates, l/Tlo, of 12%enon in H-ZSM-5 at L a w Concentration [All,, 1/Tm (s-')
0.20 0.79 3.84
2.18 (PO.05) 2.10 (h0.05) 1.73 (h0.05)
The relaxation rates of IZ9Xein H-ZSM-5-24,H-ZSM-5-120, and H-ZSM-5-490depend strongly on n, as can be seen from the data in Figure 6 taken at 298 (11) K. This behavior is quite similar to the dependence of chemical shift on n, so we fit it to a virial expansion in n l/T1 - l/Tlo = (l/Tll)n + (l/Tlz)nZ+ ...
(2)
where l/Tlo, l/Tll, and 1 / T I 2are coefficients in the virial ex-
10916 The Journal of Physical Chemistry, Vol. 96, No. 26, 1992 ZSM-11 ppm 1601
2 0.
.-
10.
‘%:gSe.-
e . a , c . u . , ~ ~ ; ; ~ ~ ~, ~ , ~ i i ,
1O2O
102’
Xe atcms/g
Figure 7. 129Xechemical shift (full line) and line width (dotted line) as a function of n in Na-ZSM-11 at 298 K. Si/AI ratios: (0)30; (A) 40; (V) 80; (0)160; (+) m . The error in the shift is *0.5 ppm.
pansion of the relaxation rate. The extrapolated relaxation rate, 1/T10(298K), has a slight dependence on [All,, (Table I). ZSM-11. The spectra of xenon adsorbed in ZSM- 1 1 consist of a single signal for all values of n. The dependence of 6 on n is very similar to that of ZSM-5 (Figure 7). The dependence of extrapolated chemical shifts, 6,, and line widths, Au, on [All,, are quite similar to those of ZSM-5 (Figure 5 ) . In particular, the discontinuity at [All,,, = 2 is observed. However, for the same values of n and [All,,, Au(ZSM-11) is always less than Au(ZSM-5) by about a factor of 2. Relaxation rates were not determined for these samples.
Discussion At low xenon pressures, the amount of xenon adsorbed depends on the aluminum content of the framework. The higher the aluminum content, the more xenon is adsorbed. This dependence on the aluminum content results from electrostatic interaction between the cation and the induced dipole of the xenon that augments the van der Waals forces with the lattice. The strong dependence of 6 on n is independent of [All,,. The sharp variation of the slope at n = (8 - 9) X 1020atoms g-l is not due to a change in the presence of aluminum in the lattice or to the presence of different amounts of cation. There are two possible reasons for this characteristic. 1. At high enough xenon concentrations, the probability of multiple collisions along channels (and especially at their intersections) is no longer negligible. In the bulk gas where there is an isotropic distribution of collisions, the increase of three-body collisions and highersrder collisions reduces the growth of 6 with n. It is unlikely that this effect would be the same for diffusion in the anisotropic environment of ZSM-5. The difference in the virial coefficients” of xenon in ZSM-5 [a1(298K) = 0.08 (*0.01) ppm/amagat; 6?(298 K) = 3.3 (k0.4) X lo4 ppm/amagat2] as compared to the bulk [i$(300 K) = 0.548 ppm/amagat; 1S~(300 K) = 1.69 X lo4 ppm/amagat*] reflect this difference. 2. A second explanation for the variation in slope is the possibility of forming X t X e dimers, which depends on the diffusivity of xenon and its concentration. In the case of NaY zeolite, Chen and FraissardlO have shown that, at low temperature, the slope of plots of 6 versus n is small as n 0. It then increases rapidly with n as the average concentration approaches 1 atom per supercage. Beyond this value, it changes again. In ZSM-5 and ZSM-11, the narrow channels result in restricted diffusion of xenon. The probability of forming Xe-Xe dimers only becomes large at sufficiently high concentrations. Relaxation of xenon adsorbed in ZSM-5 is much more effective than the relaxation of xenon in the bulk gas phase.12 The de-
-
Chen et al. pendence of relaxation rate on n is strikingly similar to the dependence of chemical shift on n for ZSM-5 zeolites, although the change of slope is not so pronounced. The relaxation rate at low uptake, l/Tlo, decreases as [All,, (or cation content) increases. The strong increase of the relaxation rate with n (independent of [All,,) indicates that xenon-xenon interactions play a substantial role in enhancing the relaxation process. This contrasts strongly with the relaxation behavior of xenon in NaY zeolite, in which the relaxation rate decreases with uptake of xenon.” For gas-phase xenon, Torrey’sI4model for relaxation by atomic collisions predicts the relaxation rate, l/Tl, to be proportional to the chemical shift, u. Comparison of Figures 4 and 6 shows that a similar result seems to hold for xenon sorbed in the pores of the ZSM-5 structure. At high pressures and sufficiently elevated temperatures, the spectrum of ZSM-5-67 exhibits a second line at a lower chemical shift. The fact that this line is less well resolved at low temperature and that it disappears for T < 268 K suggests that it can be attributed to a structural defect having larger pore volume rather than to an inhomogeneous distribution of aluminum. Adsorption in small pores is favored at low temperatures. ZSM-5- 1950 is a sample consisting of a well-crystallized phase, characterized by a narrow NMR signal that extrapolates to the “correct” 6, based on [All,,, and an amorphous phase whose presence is responsible for the rather stronger dependence of 6 on n. All other things being equal, the slope is proportional to the inverse of the free volume of the well-crystallized phase. Finally, the slope is not the result of the absence of aluminum, since all well-crystallized samples of silicalite have the same value of 6, and have dependences on n similar to those of the other members of the series investigated here.I5 6, decreases with [All,,, from 111.5 to 102.6 ppm (or about 10%) when the silicon-to-aluminumratio increases from 24 to 50. This change is slightly greater than for NaY zeolite but is similar in relative value. It is logical that the effect of the chemical nature of the surface should be more important if the residence time of xenon at the surface is longer. The most remarkable feature of xenon NMR spectroscopy of ZSM-5 and ZSM-11 is the change in the Xe-surface interactions at [All,, c 2, as seen in the discontinuities of 6, and Au. Quantum mechanical calculations have shown that the aluminum atoms are preferentially located at certain crystallographic sites in MFI.I6 Studies of 29SiNMR relaxation in MFI in the presence of templates reveal a change in behavior for [All,, = 2.” Auroux et a1.I8 noticed large variations in the heat of adsorption of NH3 on acidic MFI for this aluminum concentration. Finally, studies of the parallel adsorption of n-hexane and cyclohexane on MFI indicate a sharp decrease in the ratio K = [n-hexane]/[cyclohexane] in the adsorbed phase at this aluminum concentration. All of these results are consistent with the ”Xe results. There is an abrupt change in the surface properties of ZSM-5 and ZSM-11 for [All,, 2, which can only be due to a particular distribution of the A1 atoms in the lattice. This point confirms the results of Freude,I9 who showed by 29SiMAS NMR spectroscopy that aluminum is not statistically distributed in the MFI structure. This difference has been observed systematically for all the MFI and MEL zeolites. Hence, it is possible to use this characteristic to distinguish pentasil zeolite with [All,, > 2 from those with [All,, > 2. It appears that there is also a slightly different aluminum distribution in ZSM-5 and ZSM-11 for [All,, > 2. Indeed, 6, is identical for [All,, < 2, but is noticeably different for [All,, > 2.
-
Conclusions The NMR spectroscopy of xenon adsorbed in penmil zeolites of the ZSM-5 and ZSM-11 forms is sensitive to the local environment. The dependence on structure expresses itself as a slight dependence of 6, on [All,. The variation with aluminum content is, however, small. The dependence on n indicates that xenonxenon interactions are similar in all of these materials, independent of aluminum content. One may still usc 6, to determine the mean
J. Phys. Chem. 1992’96, 10917-10922
free path imposed by the structure, and thereby the rough physical characteristics of the free volume. All the results (both chemical shifts and relaxation rates) indicate there is a discontinuity in the surface properties at around [All, 2. We believe this expresses a nonrandom distribution of aluminum atoms in the lattice, in agreement with other observation of similar materials.
-
Acknowledgment. This work was supported in part by the Sponsors of the Center for Catalytic Science of the University of Delaware, by Grant 19011-ACS of the Petroleum Research Fund of the American Chemical Society, and by the National Science Foundation under grant CHEM 9013926. C.D. gratefully acknowledges a grant-in-aid from Sun Refining and Marketing Co. We acknowledge Ronald Nickle (deceased) and Penrose Hollins for technical assistance. Registry No. Xenon-I 29, 13965-99-6.
References and Notes (1) Fraissard, J.; Ito, T. Zeolites 1988, 8, 350, and references therein. (2) Ito, T.; Fraissard, J. In Proceedings of rhe Fifh International Conference on Zeolltes; Rees, L. V. C., Ed.; Heyden: Frankfurt; 1980; p 510.
10917
(3) Chen, Q.Dissertation, Universitl Pierre et Marie Curie, Paris, 1990. (4) Chen, Q.; Springuel-Huet, M. A.; Fraissard, J. Zeocat 90: Catalysis and Adsorption by Zeolites; Elsevier: Amsterdam; 1991; p 129, and refmnccs therein. ( 5 ) Alexander, S.;Coddington, J. M.; Howe, R. F. Zeolites 1991, 1 1 , 368. (6) Rollman, L. S.;Volyocsik, G. E. Inorg. Synrh. 1983, 22, 61-68. (7) Corbin, D. R.; Burgess, B. F., Jr.; Vega, A. J.; Farlec, R. D. Anal. Chem. 1987, 59, 2722. (8) Jameson, A. K.;Jameson, C. J.; Gutowsky, H. S.J . Chem. Phys. 1970, 53, 2310. (9) Chen, Q.;Fraissard, J. P. J . Phys. Chem. 1992, 96, 1816. (10) Chen, Q.;Fraissard, J. P.J . Phys. Chem. 1992, 96, 1809. (1 1) The vinal coefficients were converted to a common basis using a value for the amagat, the density of an ideal gas at STP, of 2.69 X IOl9 atoms an-’. (12) Hunt, E. R.; Carr, H. Y . Phys. Reo. 1963, 130, 2302. (13) Smith, M. L.; Dybowski, C. J . Phys. Chem. 1991, 95,4942. (14) Torrey, H. C. Phys. Reu. 1963, 130, 2306. (15) Fraissard, J. Unpublished results (16) Derouane, E. G.; Fripiat, J. J. Zeolites 1985, 5, 165. , (17) Debra, G.; Gourgue, A,; B”agy, J.; de Clippeleir, G. Zeolites 1986, 6, 161. (18) Auroux, A.; Gravelle, P. C.; Vedrine, J. C.; Rekas, M. In Proceeding8 of the Fifrh International Conference on Zeolites; R m , L. V. C., Ed.; Heyden: Frankfurt, 1980; p 433. (19) Freude, D. Srud. Surt Sci. Catal. 1989, 52, 169.
Mixed Valence Oxide-Dispersion- Induced Micropore Filling of Supercritical NO Z.-M. Wang: T. Suzuki: N. Uekawa,t K. Asakura,t and K. Kaneko*vt Department of Chemistry, Faculty of Science, Chiba University, Yayoi, Chiba-shi, Japan, and Department of Chemistry, Faculty of Science, The University of Tokyo, Tokyo, Japan (Received: May 11, 1992; In Final Form: September 23, 1992)
Ultrafine Ti(1V)-doped a-FeOOH particles were dispersed on the high-surface-area carbon fibers of uniform micropores in order to elucidate the role of the surface electronic factor in the micropore filling of supercritical NO. EXAFS and XANES showed that dispersed Ti-doped a-FeOOH particles are ultrafine and have the same local structure as that of bulk a-FeOOH particles. The microporosity of the samples did not change significantly by the doping of Ti(1V) in dispersed a-FeOOH on the basis of Nz adsorption. Also, the surface acidic site concentration measured by the irreversible NH3 adsorption was almost constant regardless of the Ti doping. However, doping of Ti(1V) in the dispersed a-FeOOH enhanced the micropore fdling of supercritical NO by 40%, at the maximum. The enhancement of the NO micropore filling by Ti doping was presumed to come from the increase of quasi-free electrons in the dispersed a-FeOOH fine particles due to mixed valence formation, which is enhanced by the surface segregation of Ti(1V) from the XPS examinations.
Introduction Pores of less than 2 nm in width are termed micropores according to IUPAC classification,’ although recent statistical mean-field theories predict that the critical width between the micropore and mesopore is 1.3-1.7 nmS2s3In a slit-shaped micropore, the surface potentials from the opposite pore walls are overlapped to enhance the molecular adsorption in low relative pressure; to enhance the molecular adsorption in low relative pressure; such enhanced adsorption is called micropore filling. As micropore filling is an enhanced physical adsorption, it is not a predominant proceap for a supercriticalgas (a gas above the critical temperature) but is for a vapor. Accordingly, great amounts of vapor molecules are sufficiently adsorbed even at low-pressure regions by microporous solids owing to the micropore-filling Zeolites, aluminophosphatcs, and activated carbon fidem (ACFs) are representative microporous solids. Zeolites and aluminophosphatesl2have cylindrical micropores inherent to their crystal structures, while ACFs have uniform slit-shaped micropores whose pore wall is composed of microcrystallites of graphite structure. ACFs have much greater micropore volume than zeolites and can provide great nanospace available for a specific reaction.I3 ‘Chiba University. *TheUniversity of Tokyo.
0022-3654/92/2096- 10917$03.00/0
Adsorption of supercritical gases by microporous solids has been studied by using a high-pressure adsorption techniq~e.’~ Supercritical CH, has especially gathered much attention and interest from both experimental and theoretical researcher~.~~-l’ For example, Gubbins and T a d 5 and Myers and KaraviasI2 have studied, theoretically, CH4adsorption on model carbon microw. Talu et al.” tried to correlate the high-pressure CH4 adsorption with their micropore structure of zeolites. Thus, studies on the adsorption of a supercritical gas are limited to high-pressure adsorption. There has not been sufficient research on the adsorption of a supercritical gas by microprous solids at subambient pressure regions. A N O gas near room temperature is a supercritical gas because its critical temperature is 180 K. NO cannot be adsorbed by microporous solids with the aid of a microporefilling mechanism, although development of a good adsorbent for NO is expected from the field of atmospheric environmental science. N O has an unpaired electron and tends to be dimerized in the micropore to exhibit diamagnetism.’* The dispersion of oxides or oxyhydroxidesof Fe(II1) with great magnetic moments due to their five unpaired electrons on ACF’s induced marked micropore filling of NO through the dimerization on the surface.I9 Application of an external magnetic field enhanced the NO micropore filling over ACF‘S.~O*~~ Not only the magnetic perturbation but also a weak chemisorptive mechanism should be associated with the enhancement of the NO micropore filling.22 0 1992 American Chemical Society
