Distribution of Tetrahedral Aluminium Sites in ZSM-5 Type Zeolites: An

Dec 12, 1996 - Distribution of Tetrahedral Aluminium Sites in ZSM-5 Type Zeolites: An 27Al (Multiquantum) Magic Angle Spinning NMR Study. Priit Sarv*...
0 downloads 0 Views 219KB Size
J. Phys. Chem. 1996, 100, 19223-19226

Distribution of Tetrahedral Aluminium Sites in ZSM-5 Type Zeolites: An (Multiquantum) Magic Angle Spinning NMR Study

19223 27Al

Priit Sarv* Institute of Chemical Physics and Biophysics, Akadeemia 23, EE0026 Tallinn, Estonia

Christian Fernandez and Jean-Paul Amoureux Laboratoire de Dynamique et Structure des Mate´ riaux Mole´ culaires, CNRS URA801, UniVersite´ des Sciences et Technologies de Lille, F-59655 VilleneuVe d’Ascq Cedex, France

Kari Keskinen Neste Oil, Catalysis Research, Technology Centre, P.O. Box 310, FIN-06101 PorVoo, Finland ReceiVed: August 15, 1996; In Final Form: October 16, 1996X

Up to now the distribution of Al atoms in the zeolite lattice could be monitored mainly by 29Si MAS NMR. The data about the nonequivalent aluminium T sites, present in the 27Al MAS NMR spectra, was obscured by second-order effects of quadrupolar interaction. By applying the five-quantum 27Al MQMAS NMR method to ZSM-5 type zeolites, we were able to distinguish at least two nonequivalent aluminium T sites in the H-ZSM-5 and establish the relation between 27Al, 29Si, and 1H NMR data. Comparison with 29Si MAS NMR spectra gives information about the distribution and siting of aluminium in the zeolite framework.

Introduction Zeolites are porous aluminosilicates made up from corner and edge-sharing SiO4 and AlO4 tetrahedra. In commercial zeolites, the ratio of Si/Al is usually in the range of 1-100 and distribution of Al and Si over the tetrahedral sites (T sites) is disordered. Over the past decade high-resolution solid-state NMR spectroscopy has established itself as a powerful technique for characterization of zeolites and related materials with respect to structure elucidation, catalytic behavior, and mobility properties.1-3 In the case of low Si/Al ratio zeolites, the 29Si magic angle spinning (MAS) NMR spectra of the simplest systems show five resonances corresponding to the five local silicon environments: Si[4Al], Si[3Al, Si], Si[2Al, 2Si], Si[Al, 3Si], and Si[4Si].4 In the more complex systems (ZSM-5, mordenite, ferrierite), resonances belonging to crystallographically nonequivalent T sites can be resolved, but because of the lack of Si/Al ordering the resonances are relatively broad, about 3 ppm. Removal of all the aluminium from zeolites produces completely siliceous structures, which are now perfectly ordered, and the 29Si MAS NMR spectra show very sharp resonances (hwhh < 0.5 ppm), whose number correspond to crystallographically nonequivalent sites in the unit cell and whose relative intensities reflect the population of these sites.5 The chemical shift (CS) of Si[4Si] sites and the average T-O-T bond angle of these sites have a linear relationship.6 The primary information one can get from 27Al MAS NMR spectra is related to coordination of Al atoms.1 Usually the tetrahedral lattice aluminium in zeolites gives only one line with the CS in the range of 50-65 ppm. Lippmaa et al. have shown that in analogy to silicon T-O-T sites there is a linear relationship between the CS and the mean Al-O-Si bond angle of tetrahedral Al atoms in aluminosilicates.7 Accordingly, one should be able to resolve crystallographically nonequivalent sites also in 27Al MAS NMR spectra of zeolites. For example, in the 29Si MAS NMR spectrum of ZSM-5 we can see a -113 ppm peak and a shoulder at -115 ppm, but in the 27Al MAS X

Abstract published in AdVance ACS Abstracts, November 15, 1996.

S0022-3654(96)02519-1 CCC: $12.00

NMR spectrum there is only one peak, assigned to framework aluminium.1 So far, only in zeolite omega8 and lately also in MCM-229 the crystallographically nonequivalent framework sites of aluminium could be detected. The reason for this is the overlap of lines caused by the second-order broadening and line shift of 27Al resonances due to quadrupolar interaction.7 Recently a two-dimensional (2D) multiquantum MAS (MQMAS) NMR experiment of quadrupolar nuclei was proposed by Frydman and Harwood10 and further developed by Fernandez and Amoureux.11,12 In this experiment it is possible to separate the contributions from the two interactionssCS and quadrupole interaction (QI), and one can get information about the distribution of CS of 27Al (nonequivalent sites) and electric-field gradient (EFG) on these sites respectively. In this Letter we study what new knowledge can we get by applying this MQMAS method to various ZSM-5 type zeolites. Experimental Section The MAS NMR spectra were measured on Bruker AMX500 spectrometer using a homemade (3.5 mm o.d. rotor) probe head. The spinning speed was approximately 15 kHz. AMX500 was equipped with an additional 250 W power amplifier to ensure the radio-frequency (rf) field strength of 120 kHz. KAl(SO4)2·12H2O was used to measure the rf power and to set the magic angle. 29Si MAS NMR spectra were measured with a 30° flip angle and with an 8 s repetition delay. 1H MAS NMR spectra were measured with a 30° flip angle and with a 10 s repetition delay. 29Si spectra were deconvoluted using Gaussian lines, and 1H spectra were deconvoluted using Lorentzian lines.13 Deconvolution was done with a Bruker 1D WIN-NMR program. To perform five-quantum MQMAS experiments on zeolite samples, a precise setting of experimental parameters is needed. We give a brief description of the experiment, but more details can be found elsewhere.11,12,14-17 The MQMAS experiments were performed with a two-pulse sequence. The first pulse has the length optimized to obtain the best efficiency for the multiquantum ((5Q) coherence creation. The second pulse is © 1996 American Chemical Society

19224 J. Phys. Chem., Vol. 100, No. 50, 1996

Letters

Figure 1. Calculated signal intensity vs SOQE for the MQMAS experiment, νrf ) 120 kHz, η ) 0.

designed to transfer these two symmetrical coherences with the same efficiency into observable (-1)Q single-quantum coherence. A proper phase cycling is done to select only the desired coherence transfer pathways (0)((5)(-1). Prior the experiments the numerical calculations with the special computer program PULSAR17,18 were performed, to see if the 120 kHz rf power is enough to excite the necessary (5Q coherence (Figure 1). The typical quadrupole coupling constants (Cq) for framework aluminium sites in hydrated zeolites are of the order of 1-2 MHz.19 As one can see from Figure 1, 120 kHz of rf power can excite the (5Q coherence even if Cq is more than 2 MHz, but with a decreasing efficiency with the increasing Cq. This was verified also experimentally by acquiring the (5Q) MQMAS spectrum of hydrated AlPO-21 (spectrum not shown), where there are two five-coordinated aluminium sites with a Cq of 5.9 and 7.4 MHz.20 The flip angles of the first and second pulses were found from the numerical simulation. Finally we used 4.5 and 2.1 µs for the first and second pulse lengths, respectively. The delay t1 between the two pulses is incremented regularly, using the TPPI method.12,21 The 27Al (5Q) MQMAS experiment involved the accumulation of 512(t2) × 256(t1) data points, where the number of scans was typically 2000 with the repetition delay of 0.1 s and dwell time in t1 was 8 µs. A 2D Fourier transform with respect to t1 and t2 leads to pure-absorption 2D spectra.12 For a single site in a powder with second-order quadrupolar broadening, all resonances corresponding to different orientations of crystallites are located along a single axis (A) with a slope given by

ν2 ) (-25/12)ν1 (in Hz)

(1)

A high-resolution isotropic spectrum where the second-order quadrupolar broadening is completely averaged to zero can be obtained from a skew projection parallel to the A axis. Residual broadening, however, remains due to CS and/or Cq distributions, but that should give information about the nature of the distribution.11 We studied the following ZSM-5 samples: NaZSM-5 (Si/ Al ) 36), HZSM-5 (Si/Al ) 36) and CaHZSM-5 (25% of H+ sites Ca2+ exchanged, Si/Al ) 46). Si/Al ratios were determined by 29Si MAS NMR. The amount of Ca2+ was determined by elemental analysis and H+ exchange rate by 1H MAS NMR. The ZSM-5 zeolites were synthesized by methods described by C. J. Plank et al.22 CaHZSM-5 was prepared from a HZSM-5 (Si/Al ) 46) by a conventional ion-exchange procedure using calcium acetate solution. Prior to NMR measurements all samples were kept at 75% relative humidity for at least 48 h. Before the 1H NMR experiments all samples were dehydrated at 400 °C in a 10-3 Torr vacuum, and rotors were filled in dry atmosphere of nitrogen and oxygen. Results and Discussion Most tetrahedral sites in high and medium Si/Al ratio zeolites are occupied by silicon atoms. Usually the replacement of

Figure 2. Sheared 2D (5Q) MQMAS NMR spectrum of H-ZSM-5 (A), Na-ZSM-5 (B), and CaH-ZSM-5 (C). The actual larmour frequency in the ν1S dimension is ν0(p - R), where p ) 5 and R ) -25/12.

silicon atoms by aluminium atoms does not change the overall crystal structure. We also know that the isotropic chemical shifts of 27Al and 29Si atoms are linearly correlated with the average T-O-T bond angle,6,7 and in some aluminosilicate sodalites there has been established a linear relationship between the 27Al and 29Si chemical shifts of framework atoms.23 Therefore, assuming uniform distribution of aluminium between the T sites, we conclude that in a first approximation (all aluminium sites have the same Cq) the isotropic spectrum of 27Al should be similar to the spectrum of 29Si. We can resolve two nonequivalent T sites in the 29Si MAS NMR spectrum of ZSM-5 (-113 and -115 ppm peaks), and therefore at least the same kind of distribution is expected also for 27Al MAS NMR spectrum. The sheared 5Q 2D (MQ)MAS NMR spectra of ZSM-5 samples are given in Figure 2. Shearing of the spectrum aligns the ν2 axis parallel to A axis, and a projection to the ν1S axis

Letters

J. Phys. Chem., Vol. 100, No. 50, 1996 19225

Figure 3. Isotropic projection of the 2D (5Q) MQMAS NMR spectrum of H-ZSM-5 (A), Na-ZSM-5 (B), and CaH-ZSM-5 (C) plus the deconvolution.

(isotropic axis) gives us a 1D high-resolution spectrum of aluminium without quadrupolar second-order broadening. The second-order quadrupolar shifts, which can be calculated using formulas given in ref14, are still present in the spectrum. But since in our case the quadrupolar shifts are the same for all sites (ca 0.6 ppm in the ν1S dimension), then also the quadrupolar coupling is about the same for all the aluminium sites (SOQE ) 1.7 MHz, where SOQE ) Cq((1 + η2)/3)1/2) and different sites have been detected to the same extent in the 2D 5Q MQMAS NMR spectrum (see Figure 1). Therefore the projection to the isotropic axis reflects the actual distribution of the aluminium sites in the sample (Figure 3). Although two lines are clearly resolved on most spectra, the best fit was achieved by using three Gaussian lines for simulation. The line width of the lines within a single simulation was kept equal. The isotropic projection of the aluminium spectrum of sample HZSM-5 shows two main peaks with isotropic chemical shifts 57.1 ( 0.5 ppm (peak bAl) and 54.5 ( 0.5 ppm (peak bAl′) with the average line width of 2.8 ppm and approximate intensity ratio of 2:1 (Figure 3A). Remarkably the 29Si MAS NMR spectrum shows the similar distribution of peaks for the Si(0Al) T sites: -112.3 ppm (peak bSi) and -115.4 ppm (peak bSi′) with the average line width of 2.9 ppm and intensity ratio 2:1 (Figure 4A). The relation between the isotropic chemical shift and the average T-O-T bond angle is given by Alδ ) j - 25.44, for 27Al7 and -0.500R j + 132 and Siδ ) -0.5793R 29Si,6 respectively. By applying these formulas, we get the following relations between NMR peaks and the average T-O-T bond angles: peak bAl, (150 ( 1)°; peak bAl′, (155 ( 1)° and peak bSi, (150 ( 1)°; peak bSi′, (155 ( 1)°. Consequently the two peaks in the 27Al MAS NMR spectrum of HZSM-5 represent the corresponding tetrahedral framework silicon sites replaced by aluminium. Moreover, taking also into account the similar relative intensity of peaks bAl versus bAl′ and bSi versus bSi′ in both spectra and also similar line width of the peaks, we may conclude that aluminium is randomly

Figure 4. 29Si MAS NMR spectrum (A) and 1H MAS NMR spectrum (B) of sample H-ZSM-5 together with the deconvolution and the difference spectrum.

distributed over the lattice. This result supports the findings from defect energy calculations, predicting that there would be little deviation from a random distribution of aluminium over the possible framework sites.24 At this point it would be interesting to speculate about the possible locations of the b′ sites in the framework. Since the Si/Al ratio of our H-ZSM-5 sample is 36, we expect it to be in the orthorhombic phase.25 Using the assignments of Fyfe et al. for the 29Si MAS NMR spectrum of the highly siliceous zeolite ZSM-5,26 we find the possible locations for the bSi′ sites (and also for the bAl′ sites) to be: T2, T3, T4, T8, and T11. These are the locations whose 29Si MAS NMR resonances fall into the 2.9 ppm range around -115.4 ppm and because of the lack of Si/Al ordering form a shoulder in the 29Si MAS NMR spectrum of H-ZSM-5 (Figure 4A). So in the given limits of resolution (fwhh > 2.5 ppm), we can speak about two different aluminium sites in zeolite H-ZSM5. Regarding the considerable difference of the CS of these sites, we expect to find also at least two different Bro¨nsted acid sites there. Indeed, it is known that low-temperature 1H MAS NMR shows the existence of two different Bro¨nsted sites in H-ZSM-5: site b′ (chemical shift 6.9-7.0 ppm) and site b (chemical shift 4.2-4.3 ppm), with the approximate intensity ratio of 1:2 respectively.27-29 The 1H MAS NMR spectrum from sample H-ZSM-5, taken at room temperature (Figure 4B), gives the line at 4.2 ppm (site b) and a shoulder at 5.4 ppm, corresponding to site b′.29 The relative intensity of line b′ with respect to the total intensities of lines b and b′ is 32% ( 5%, which is in line with the previous studies.29 It is interesting to note that the relative intensity of the high-field shoulder from the 27Al spectrum (peak bAl′) correlates with the relative intensity of the low-field shoulder from the 1H spectrum (peak b′). A

19226 J. Phys. Chem., Vol. 100, No. 50, 1996

Letters structure of the tetrahedral line in 27Al MAS NMR spectrum: there are at least two resolved peaks belonging to tetrahedral framework atoms of HZSM-5 type zeolites. The line widths, relative intensities, and also average T-O-T bond angles calculated from the CS of these peaks agree with those deduced from 29Si MAS NMR spectra. This implies an almost random substitution of silicon atoms by aluminum atoms in the framework of the H-ZSM-5 type zeolite. The two aluminium peaks were assigned to two different Bro¨nsted sites, known from the 1H MAS NMR experiments. The fine structure of the tetrahedral aluminium line is sensitive to changes in surrounding of aluminium atom (cation exchange), enabling the direct characterization of the chemical manipulations of the catalytically active site. Acknowledgment. This research was supported by the Copernicus Grant CIPA CT94-0184 and the Estonian Science Foundation Grant No. 2302. References and Notes

Figure 5. 1H MAS NMR spectrum of CaH-ZSM-5 before (A) and after (B) 25% Ca2+ exchange.

tentative explanation is that the increase of electron density at the aluminium site has led to the decrease of electron density at the hydrogen site, leading to more acidic Bro¨nsted site. Indeed, calculated proton affinity versus crystallographic T-O-T angles shows lower proton affinities for sites with T-O-T angle 155° compared to sites with T-O-T angle 150°.30 Therefore we assign the bAl′ peak (isotropic CS equal 54.5 ppm in the 27Al MAS NMR spectrum) to more acidic Bro ¨ nsted sites (b′) (6.9 ppm peak in low-temperature 1H MAS NMR spectrum) and the bAl peak (isotropic CS equal 57.1 ppm) to less acidic Bro¨nsted sites (b) (4.2 ppm peak in 1H MAS NMR spectrum) and in the following denote them as b and b′ sites. We can also see that cationic exchange has a noticeable effect on the 27Al (5Q) MQMAS spectrum of zeolite ZSM-5 (Figures 2 and 3), reflecting the change in the distribution of T-O-T bond angles. The peak at 55.1 ppm is decreased if we have Na+ instead of H+ as a cation (Figure 3), and a pronounced peak at 59.4 ppm (T-O-T angle 146°) has emerged. At the same time if we exchange 25% of H+ to Ca2+, the relative intensity of the 54.5 ppm peak remains the same, but the 57 ppm peak decreases and the 59 ppm peak increases. Upon Ca2+ exchange the line width of the peaks decreases from 2.8 to 2.4 ppm and upon Na+ exchange increases to 3.3 ppm. This effect can be explained by structural changes at the aluminium site induced by increased repulsion (both steric and electrostatic) between cations in case of Na-ZSM-5 and decreased repulsion in case of CaH-ZSM-5. Similar decrease of the line width can be seen also if we compare the 1H MAS NMR spectra of CaHZSM-5 before and after Ca2+ exchange (Figure 5). We think that the explanation lies in the ionic radius of the cations: Ca2+, 0.99 Å and Na+, 0.97 Å. One Ca2+ cation induces less structural changes than two Na+ cations in the same volume. Conclusion A novel (5Q) MQMAS NMR method gives us the possibility to separate the broadening effects due to quadrupolar interaction, leading to isotropic spectra. We demonstrated the utility of the method for the zeolite characterization by revealing a fine

(1) Engelhardt, G.; Michel, D. High-Resolution Solid-State NMR of Silicates and Zeolites; John Wiley & Sons: Chichester, 1987. (2) Sto¨cker, M. Stud. Surf. Sci. Catal. 1994, 85, 429. (3) Bell, A. T., Pines, A., Eds.; NMR Techniques in Catalysis; Marcel Dekker, Inc.: New York, 1994. (4) Lippmaa, E.; Ma¨gi, M.; Samoson, A.; Tarmak M.; Engelhardt, G. J. Am. Chem. Soc. 1981, 103, 4992. (5) Fyfe, C. A.; Feng, Y.; Grodney, H.; Kokotailo, G. T.; Gies, H. Chem. ReV. 1991, 91, 1525. (6) Thomas, J. M; Kennedy, J.; Ramdas, S.; Hunter, B. K.; Tennakoon, D. T. B. Chem. Phys. Lett. 1983, 102, 158. (7) Lippmaa, E.; Samoson, A.; Ma¨gi, M. J. Am. Chem. Soc. 1986, 108, 1730. (8) Fyfe, C. A.; Gobbi, G. C.; Kennedy, G. J.; Graham, J. D.; Ozubko, R. S.; Murphy, W. J.; Bothner-By, A.; Radok J.; Chesnick, A. S. Zeolites 1985, 5, 179. (9) Hunger, M.; Ernst, S.; Weitkamp, J. Zeolites 1995, 15, 188. (10) Frydman, L.; Harwood, J. S. J. Am. Chem. Soc. 1995, 117, 5367. (11) Fernandez, C.; Amoureux, J. P. Chem. Phys. Lett. 1995, 242, 449. (12) Fernandez, C.; Amoureux, J. P. Solid State Nucl. Magn. Reson. 1996, 5, 315. (13) Fenzke, D.; Gerstein B. C.; Pfeifer, H. J. Magn. Reson. 1992, 98, 469. (14) Medek, A.; Harwood, J. S.; Frydman, L. J. Am. Chem. Soc. 1995, 117, 12779. (15) Fernandez, C.; Amoureux, J. P.; Delmotte, L.; Kessler, H. Microporous Mater. 1996, 6, 125. (16) Fernandez, C.; Amoureux, J. P.; Chezeau, J. P.; Delmotte, L. Microporous Mater., in press. (17) Amoureux, J. P.; Fernandez, C.; Frydman, L. Chem. Phys. Lett., in press. (18) Amoureux, J. P.; Fernandez, C.; Dumazy, Y. J. Chim. Phys. 1995, 2, 1939. (19) Freude, D.; Haase, J. NMR 1993, 29, 1. (20) Jelinek, R.; Chmelka, B. F.; Wu, Y.; Grandinetti, P. J.; Pines, A.; Barrie, P. J.; Klinowski, J. J. Am. Chem. Soc. 1991, 113, 4097. (21) Ernst, R.; Bodenhausen, G.; Wokaun, A. Principles of NMR in one and two dimensions; Oxford University Press: New York, 1987. (22) Plank, C. U.S. Patent 3 926 782, 1970. (23) Jacobsen, H. S.; Norby, P.; Bildsøe, H.; Jakobsen, H. J. Zeolites 1989, 9, 491. (24) Schro¨der, K.-P.; Sauer, J.; Leslie, M.; Richard, C.; Catlow, A. Zeolites 1992, 12, 20. (25) Hay, D. G.; Jaeger, H. J. Chem. Soc., Chem. Commun. 1984, 1433. (26) Fyfe, C. A.; Feng, Y.; Grodney, H.; Kokotailo, G. T. J. Am. Chem. Soc. 1990, 112, 8812. (27) Beck, L. W.; White, J. L.; Haw, J. F. J. Am. Chem. Soc. 1994, 116, 9657. (28) Brunner, E.; Beck, K.; Koch, M.; Pfeifer, H.; Staudte, B.; Zscherpel, D. Stud. Surf. Sci. Catal. 1994, 84, 357. (29) Freude, D. Chem. Phys. Lett. 1995, 235, 69. (30) Redondo, A.; Hay, P. J. J. Phys. Chem. 1993, 97, 11754.

JP962519G