J. Phys. Chem. 1995, 99, 1076-1079
1076
Multinuclear NMR Studies Reveal a Complex Acid Function for Zeolite Beta Larry W. Beck and James F. Haw* Department of Chemistry, Texas A&M Universio, College Station, Texas 77843-3255 Received: October 28, 1994; In Final Form: December 2, 1994@
Recently developed NMR methods reveal that the acid sites of zeolite Beta are complex. Variable-temperature 'H and 'H{ 27Al} spin echo double-resonance magic angle spinning measurements showed that the protons associated with aluminum gave a narrow resonance at 4.5 ppm at 123 K, but upon warming to 298 K, this intensity was redistributed into a narrow resonance at 4.1 ppm and a broader resonance at 5.5 ppm. 13C shifts of adsorbed acetonitrile and acetone suggest a Bronsted acid strength intermediate between zeolites HZSM-5 and HY, as well as possible complexation with an unusual aluminum site associated with the framework that was observed by 27Al NMR.
Introduction Rapid progress is being made in the understanding of the acid sites in zeolites as a result of the maturation of theoretical and experimental methods including spectroscopic probes of acidity. The most studied materials by far have been zeolites HZSM-5 (MFI) and HY (FAU). NMR studies of weakly basic probe molecules including ketones' and acetonitrile2 indicate that these zeolites are strong acids ( < l o o % H2S04) and not superacids ( > 100% H2S04) as was previously believed. There is also some evidence that the local structures and properties of Bronsted sites may be more diverse than previously believed. We recently demonstrated that the 'H magic angle spinning (MAS) spectrum of HZSM-5 is deceptively simple at room temperature; in addition to the familiar peak at 4.3 ppm due to the known Bronsted site, there is a broad resonance at 6.9 ppm that sharpens at lower temperat~res.~ Given these unexpected observations about two of the best characterized of zeolites, it is likely that some of the other known aluminosilicate zeolites are holding surprises as well. Furthermore, it is prudent to apply some of the recently developed methods of zeolite acidity characterization to other materials in order to assess whether or not they are of general value. We therefore performed an NMR characterization of the acid function of another important material, zeolite Beta (BEA).4The selection of this material was based on several criteria. Beta is generally believed to have an acid strength intermediate between those of HZSM-5 and HY;5,6yet, Beta shows unique selectivity in some acid-catalyzed reactions, and this could motivate its widespread use. Beta has a three-dimensional channel structure with pore diameters ranging between 0.56 and 0.75 nm,' and these dimensions again put it as intermediate with respect to the two well-characterized materials (although the topologies of different zeolites are not easily compared). The Beta framework forms through the more-or-less random stacking of two different polymorphic layers, and this leads to a structure described as f a ~ l t e d ;thus, ~ . ~ more opportunities may exist for local structural diversity. Indeed, there is evidence that some of the aluminum may be converted from tetrahedral to octahedral when the ammonium form of as-synthesized material is calcined to make the acid form and then hydrated.8-10 In other zeolites, this would be attributed to the formation of extraframework aluminum (dealumination), but for Beta, the tetrahedral-
* To whom correspondence should be addressed. AbFtract published in Advance ACS Abstracts, January 15, 1995
0022-3654/95/2099- 1076$09.00/0
octahedral interconversion appears to be reversible with mild treatment conditions.* This has been used to argue that the octahedral aluminum is associated with the lattice in analogy to the octahedral aluminum in some silicon-substituted aluminophosphate molecular sieve^.^ Although 29Siand 27Alspectra of Beta have been p ~ b l i s h e d , ~we ~ " are ~ ~ ~unaware of any previous 'H studies or any NMR studies of probe molecules or in situ reactions. Our final rationale for selecting Beta for this investigation was the availability of two well-characterized Beta samples of independent origin that gave essentially identical results. Variable-temperature IH MAS spectra of both Beta samples revealed a single, relatively sharp Bronsted signal with a chemical shift of 4.5 ppm at 123 K. But when the sample temperature was raised to 298 K, the Bronsted site environments became more diverse as reflected by the partial conversion of this intensity into a second, broader peak centered at 5.5 ppm. Both features were due to protons close to aluminum as proven by 'H{ 27Al}spin echo double-resonance experiments. 27Aland 'H experiments support the formation of octahedral aluminum that is different from the extraframework aluminum seen in other zeolites. I3C studies of the probe molecule acetonitrile support an intermediate acidity for Beta, and an in situ study of reactions of acetone showed a high selectivity for the formation of isobutylene and acetic acid. Acetone also exhibited multiple coordination environments in Beta, including one very different from the Bronsted complexes seen in HZSM-5 and HY.
Experimental Section Materials. Two samples of Beta were characterized. SMR62302 was manufactured by Davison and was provided to us by Dr. Robert Beyerlein. It had a reported surface area of 690 m2/g and a reported SUA1 ratio of 14. This material was highly crystalline as confirmed by X-ray diffraction. Our measurements of the SUA1 ratio of the as-received material were 12.2 by elemental analysis (Galbraith Laboratory) and 11.3 by 29Si NMR. The material, as received, had been synthesized with organic template and treated with NH4N03. The zeolite was calcined under flowing air to a final temperature of 810 K, and then washed with 2.0 M NH&l followed by deionized water. Elemental analysis showed a Si/Al of 14.0, but the standard interpretation of the 29Siintensities suggested a value of closer to 20. The second Beta sample was provided by Dr. Ramesh Borade and Dr. Abraham Clearfield of Texas A&M and was identical to the material thoroughly characterized as sample A
0 1995 American Chemical Society
J. Phys. Chem., Vol. 99, No. 4, 1995 1077
Letters
523 K
Jl
/\.",
298K
A
- A
b wf 27Ai pulse
I"" I"" I " " I " " I ' " ' I " " 25 20 15 10 5 0
I""
I""1
-5 -10 -15 PPm
Figure 1. Variable-temperature 'H MAS NMR spectra of the dehydrated proton form of zeolite Beta. Note the temperature dependence of the downfield signals.
in ref 13. Our 'H and 27AlMAS results were identical for the two materials, and all results reported are for the Davison sample. [13C2]Acetonitrile(99% 13C) and [13C2]acetone (99% 13C) were obtained from Cambridge Isotope Laboratory. All adsorptions were performed at room temperature using shallow-bed CAVERN devices.14 NMR Spectroscopy. All spectra reported were obtained using a Chemagnetics CMX-300. Active spin-speed control was used, and all spectra were measured in 7.5 mm zirconia rotors at a rotational frequency of 4000 f 3 Hz. The 299.6 MHz 'H MAS spectra were obtained using a composite pulse to excite magnetization within the coil but not weak background signals from the spinning m0du1e.l~ The x/2 pulse width was typically 4.0 ps, and pulse delays of 10 s were used to ensure quantitation based on T1 measurements for this and other zeolites. Each proton spectrum is the average of 32 scans. Liquid acetone (2.11 ppm with respect to TMS) was used as an external chemical shift reference. *'A1 MAS spectra were obtained at 78.3 MHz using a nonselective x118 pulse width of 1 ps and were referenced to 0.1 M aqueous aluminum nitrate (0.0 ppm). 13C MAS spectra were acquired using pulse widths of 4.5 ps and pulse delays of 1 s.
Results and Discussion Dehydrated zeolite samples yielded well-resolved MAS spectra without the application of multiple pulse line narrowing. Figure 1 reports representative 'H MAS spectra of zeolite Beta over a temperature range of 123-523 K. At 123 K, the spectrum was well described by two isotropic peaks: the 2.0 ppm resonance assigned to silanol groups for ZSM-5 and Y type zeolites and a peak at 4.5 ppm, which is similar in appearance to the 4.3 ppm peak observed at 298 K for Bronsted sites in HZSM-5. Raising the sample temperature resulted in a reversible redistribution of the downfield signal intensity. At 298 K, there was a reduced intensity narrow line (shifted to 4.1 ppm), but the spectrum also showed at least one broad
1 2 1 0
8
6
-2 PPm Figure 2. Spin echo 'H spectra of calcined zeolite Beta permitting assignment of protons dipolar coupled to (and hence spatially close to) aluminum. Results are shown at both 298 and 123 K; see ref 3 for a detailed description of the spectroscopic method. Three spectra are shown in each case corresponding to (a) an echo spectrum with no *'A1 irradiation, (b) a spin echo double-resonance (SEDOR)spectrum, and (a -b) a difference spectrum emphasizing protons coupled to aluminum. The 5 periods were set to be exactly equal to one rotor period, as must be done for spin echo experiments on rotating solids. 4
2
0
component further downfield. The total intensity of all signals downfield of 3 .O ppm was approximately independent of temperature (correcting for the Curie law), and the integrated intensity of these signals relative to the silanol(2.0 ppm) signals was also approximately independent of temperature. Thus, we conclude that the dramatic line shape change at 298 vs 123 K reflects a redistribution of the intensity at 4.5 ppm in the lowtemperature spectrum to a diversity of resonance frequencies at 298 K. This interpretation is very different from the case of HZSM-5, for which the 6.9 ppm signal is present but very broad at 298 K and sharpens upon cooling to 123 K without affecting the intensity of the 4.3 ppm Figure 1 indicates that the downfield proton signals coalesced into a single Lorenzian line at 523 K, consistent with chemical exchange. Figure 1 shows that the downfield region of the 'H spectrum of zeolite Beta is strongly temperature dependent, but what evidence is there that any of these signals are associated with aluminum sites and hence acidity? Figure 2 shows representative results from the application of the recently developed 'H("A1) double-resonance assignment techniques3 The spin echo difference spectra in Figure 2 show only those 'H signals from nonmobile protons very close in space (tenths of a nanometer) to 27AIsites. Thus, the silanol signals are completely nulled in the difference spectra, but nearly all of the downfield intensity is seen, proving spatial correlation with aluminum for the latter proton signals. The difference spectrum measured at 298 K clearly shows two signals: the narrow resonance at 4.1 ppm and a single broad line of comparable intensity centered at 5.5 ppm. Both proton environments are proximal to aluminum and hence should be considered related to the acid function. The spinning sidebands observed in these spectra (which can be seen with the broader spectral region shown for
1078 J. Phys. Chem., Vol. 99, No. 4, I995
A
Letters
h
a
a
A
b b 1
300
1
1
1
I
250
l
l
1
1
200
I""I""I""I""I""I""I""I""1
150 125 100 75
0
-25 -50 PPm Figure 3. 27AlMAS NMR spectra (512 scans) of hydrated Beta samples. The top spectrum (a) is of the as-received zeolite. Spectrum b was obtained after calcination to generate the acid form of the zeolite and rehydration to relax the lattice sufficiently to obtain a well-resolved 27Alspectrum.This procedure is seen to convert some of the tetrahedral aluminum (54 ppm) to octahedral aluminum (-0.3 ppm). 50
25
1
1
1
1
150
1
1
1
1
100
1
1
1
1 1 1 1
50
0 PPm
Figure 4. '!C MAS spectra of acetone-2-'?C on calcined zeolite Beta acquired at 298 K. Spectrum a (1024 scans) is from 2.0 equiv of acetone after heating to 473 K. Spectrum b (2048 scans) is from 0.25 equiv of acetone after heating to 373 K. See text for assignments. Asterisk
denotes spinning sidebands.
temperature; a more complicated behavior was observed on HY samples.* Ab initio computations were used to rationalize this downfield shift in terms of a novel protonation mechanism. The behavior of acetonitrile on Beta (spectra not shown) was similar Figure 1) are due primarily to the 1H-27A1 dipolar coupling with a smaller contribution from lH chemical shift anisotropy. to that on HZSM-5 with the exception that the magnitude of Using methods similar to those of Pfeiffer and c o - w ~ r k e r s ~ ~ , the ~ ~ shift at any given temperature was not as large for Beta. and adopting their value of the 'H CSA of HZSM-5 for Beta, For example, 0.8 equiv of acetonitrile yielded nitrile carbon we were able to calculate a lH-*'Al internuclear distances of shifts of 119, 124, and 130 ppm for temperatures of 298, 373, and 523 K, respectively. Based on the interpretation in ref 2, 0.245 and 0.257 nm for Beta at 123 and 298 K, respectively. the acetonitrile results suggest that the Bronsted acidity of Beta These values are virtually indistinguishable from 0.250 nm is indeed intermediate between those of HZSM-5 and HY in reported for HZSM-5 at room temperature.16 The breadth of strength. Acetontrile is dynamic on these zeolites, so discrete the sidebands of the 5.5 ppm isotropic peak precluded a sufficiently accurate measure of intensity for quantification of signals were not seen for the various adsorption sites. and only the internuclear distance for that site, but the intensities were an overall assessment of acid properties was obtained. Hutchings et aLZ0recently reported that the conversion of qualitatively similar to those for the narrow resonance, implying acetone on Beta was highly selective for isobutylene compared that the 1H-27A1 distance is also short for the 5.5 ppm site. to the reaction on HZSM-5. Those workers proposed that Our physical interpretation of the results in Figures 1 and 2 cracking of mesityl oxide (an aldol product of acetone) yielded is that all of the Bronsted sites in Beta can adopt a restricted the olefin selectively. We previously described in situ NMR range of low-energy conformations that results in a narrow distribution of 'H resonance frequencies for the Bronsted protons studies of aldol reactions in HZSM-5 and HY and reported the magnitudes of protonation shifts for mesityl oxide and acetone at 123 K. We suggest that some of these sites are converted on those zeolites.' Figure 4a shows a 13CMAS spectrum from into different, perhaps distorted, local structures at higher an in situ study of acetone-2-13C on Beta. Mild heating temperature, and this is manifested by a redistribution of some converted much of the acetone to acetic acid-l-13C (181 ppm) of the signal intensity into the broad 5.5 ppm signal at higher and aliphatic carbon signals similar to those seen previously in temperature. the oligomerization of isobutylene-2-13Con zeolite HY.?' We Figure 3 shows representative 27Alspectra of as-prepared Beta compared with calcined and then rehydrated Beta. As previpreviously observed cracking of mesityl oxide to acetic acid on HY. Further heating of acetic acid on acidic zeolites formed ously reported by others,8-10 this process converted some of the tetrahedral aluminum (54 ppm) to octahedral aluminum CO2, the coproduct in Hutching's flow reactor studies. (-0.3 ppm). Our elemental analysis results indicate that this A small peak at 233 ppm is also seen in Figure 4a. This aluminum was not easily washed out, as extraframework peak was more obvious when a lower loading of acetone-2-13C aluminum might be. Other investigations have shown that was adsorbed as in Figure 4b. Even with a 0.25 equiv loading protons associated with extraframework aluminum resonate of acetone, however, the largest signal was at 221 ppm, a value less than that for acetone on Bronsted sites in HZSM-5 and between 2.5 and 3.0 ppm.18,19Figure 2 very clearly shows that our calcined, dehydrated materials have no signals correlated greater than the average value for HY. The NMR spectra in with 27Alin this shift range. Thus, our results would seem to Figure 4 do not permit unambiguous assignment of the support the interpretation that the aluminum sites created by coordination environment of the smaller, 233 ppm peak. This calcination are not extraframework sites and may be aluminum feature might reflect interaction with a stronger Bronsted site, still associated with the framework. As noted previously for although the acetonitrile shift would seem to contraindicate that other zeolites, we were unable to obtain a well-resolved 27Al explanation. A second possibility is that the 233 ppm peak MAS spectrum of dehydrated Beta samples, due to the extremely corresponds to coordination of acetone to the frameworklarge quadrupole coupling constant. associated aluminums that are capable of octahedral coordinaWhat can NMR say about the strength and reactivity of the tion. Regardless of the detailed assignment of the 233 ppm acid sites in calcined Beta? Protonated acetonitrile has a peak, Figure 4b is further evidence of a more complex acid Hammett acidity of -10.3, and the I3C shift of acetonitrile function for Beta, and Figure 4a supports the claim of different moves upfield in liquid superacids, consistent with protonation. selectivity for acid-catalyzed reactions on Beta. We recently reported that the 13C resonance of acetonitrile In summary, NMR methods recently developed for charactermoved downfield in HZSM-5 to an extent that increased with izing acidity on HZSM-5 and HY appear to be generally
J. Phys. Chem., Vol. 99, No. 4, 1995 1079
Letters applicable to other zeolites, as suggested by their extension to a third framework type. The properties of the acid sites of zeolite Beta are more complex than HZSM-5 or HY. These properties include temperature-dependent structural distortions of some of the Bronsted sites and the possibility of valences other than four for some of the framework-associated aluminum, which could imply combined Bronsted-Lewis properties for some sites.
Acknowledgment. This work was supported by the Department of Energy (DE-FG03-93ER14354). References and Notes (1) Xu, T.; Munson, E. J.; Haw, J. F. J. Am. Chem. SOC. 1994, 116, 1962-72. (2) Haw, J. F.; Hall, M. B.; Alvarado-Swaisgood, A. E.; Munson, E. J.; Lin, Z.; Beck, L. W.; Howard, T. J. Am. Chem. SOC. 1994,116,730818. (3) Beck. L. W.: White. J. L.: Haw. J. F. J. Am. Chem. SOC.1994. 116,'9657-61. (4) Treacv. M. M. J.: Newsam. J. M. Nature 1988. 332. 249-251. ( 5 ) Hedgi,'S. G.; Kumar, R.; Bhat, R. N.; Ratnasamy, P. kolifes 1989, 9, 231-7. (6) Maache, M.; Janin, A.; Lavalley, J. C.; Joly, J. F.; Benazzi, E. Zeolites 1993, 13, 419-26. (7) Higgins, J. B.; LaPierre, R. B.; Schlenker, J. L.; Rohrman, A. C.; Wood, J. D.; Kerr, G. T.; Rohrbaugh, W. J. Zeolites 1988, 8, 446-52.
(8) Bourgeat-Lami, E.; Massiani, P.; Renzo, F. Di; Espiau, P.; Fajula, F. Appl. Catal. 1991, 72, 139-52. (9) Jia, C.; Massiani, P.; Barthomeuf, D. J. Chem. Soc., Faraday Trans. 1993, 89, 3659-65. (10) Kiricsi, I.; Flego, C.; Pazzuconi, G.; Parker, Jr., W. 0.;Millini, R.; Perego, C.; Bellussi, G. J. Phys. Chem. 1994, 98, 4627-34. (1 1) Fyfe, C. A.; Strobl, H.; Kokotailo, G. T.; Pasztor, C. T.; Barlow, G. E.; Bradley, S. Zeolites 1988, 8, 132-6. (12) Perez-Pariente, J.; Sanz, J.; Fomes, V.; Coma, A. J. Catal. 1990, 124, 217-23. (13) Borade, R. B.; Clearfeld, A. J. Phys. Chem. 1992, 96, 6729-37. (14) Munson, E. J.; Murray, D. K.; Haw, J. F. J. Catal. 1993,141,733736. (15) White, J. L.; Beck, L. W.; Ferguson, D. B.; Haw, J. F. J. Magn. Reson. 1992, 100, 336-41. (16) Hunger, M.; Freude, D.; Frenzke, D.; Pfeiffer, H. Chem. Phys. Left. 1992, 191, 391-5. (17) Fenzke, D.; Hunger, M.; Pfeiffer, H. J. Magn. Reson. 1991, 95, 477-83. (18) Freude, D.; Frohlich, T.; Hunger, M.; Pfeiffer, H. Chem. Phys. Lett 1983, 98, 263-6. (19) Hunger, M.; Freude, D.; Pfeiffer, H. Catal. Today 1988, 3, 50712. (20) Hutchings, G. J.; Johnston, P.; Lee, D. F.; Williams, C. D. Cafal. Lett. 1993, 21, 49-53. (21) Lazo, N. D.; Richardson, B. R.; Schettler, P. D.; While, J. L.; Munson, E. J.; Haw, J. F. J. Phys. Chem. 1991, 95, 9420-5. JP942933L