J. Phys. Chem. B 2002, 106, 4941-4946
4941
Quantitative Investigations of Acidity, and Transient Acidity, in Zeolites and Molecular Sieves† Xingwu Wang, Jeffrey Coleman, Xin Jia, and Jeffery L. White* Campus Box 8204, Department of Chemistry, North Carolina State UniVersity, Raleigh, North Carolina 27695-8204 ReceiVed: December 18, 2001; In Final Form: February 21, 2002
To demonstrate that transient acidity exists in certain solid acid catalysts, a convenient and robust 1H solidstate NMR spin-counting method is introduced for the accurate quantification of Bronsted acid sites in zeolites and molecular sieves. Poly(dimethylsiloxane) is used as an inert and easily handled spin-counting standard, allowing internal calibration of MAS NMR peaks arising from both acidic and nonacidic hydrogens in the catalyst. Results from example systems including H-ZSM5, H-ferrierite (H-FER), and SAPO-34 are presented, and the relevance of these measurements to zeolite synthesis conditions, postsynthetic treatments, and reaction mechanisms are discussed. Using this technique, we show the first direct spectroscopic proof that framework Bronsted acidity in SAPO-34 molecular sieves decreases with time following removal of the synthesis template molecules. This effect is similar to hydrothermally catalyzed dealumination in traditional zeolites, except that it occurs at ambient temperature and moisture levels. Our data indicates that these catalysts must be stored in a moisture-free atmosphere to preserve activity following template removal. These results are key to understanding hydrocarbon conversions in SAPO catalysts, particularly in the methanol-to-hydrocarbon chemistry area, where SAPO-34 is a leading candidate for commercialization.
Introduction The specific characteristics of zeolites and molecular sieves that are responsible for their ability to catalyze chemical reactions are their regular crystalline structure and their acidity. In many cases, acid site density is not constant for a given catalyst but changes with experimental conditions. Although X-ray diffraction methods are routinely used for quantifying crystallinity and elucidating structure types, there does not exist a reliable, routinely accessible method to measure the simplest aspect of zeolite acidity, i.e., the number of Bronsted acid sites. The most common techniques involve adsorption or temperature programmed desorption of NH3 and infrared spectroscopy. Each of these methods can present problems, as the NH3 TPD methods may suffer from multiple adsorbates per acid site and extreme sensitivity to subtle changes in the experimental parameters (flow rates, catalyst bed depths, temperature gradients, etc.). Furthermore, quantification by infrared methods is limited by a lack of knowledge of molar extinction coefficients for either direct observation of acidic hydroxyl groups or adsorbates such as pyridine. Several reports have highlighted the use of 31P MAS NMR of trimethylphosphine1 and trimethylphosphine oxide2 probe molecules for probing both Lewis and Brønsted acidity. Novel Na-poisoning experiments have also been effectively used to determine the number of catalytically active framework sites.3 Farneth and Gorte have recently reviewed the variety of methods used to characterize both the strength and number of acid sites.4 In this review, the preferred methods for quantifying Brønsted sites involve combined TPD and TGA analysis of reactive amines such as isopropylamine, * To whom correspondence should be addressed. E-mail: Jeff_L_White@ ncsu.edu. † Presented at the 43rd Rocky Mountain Conference on Analytical Spectroscopy, July 2001, Denver, CO.
because stoichiometric reaction with the catalyst produces one olefin and one ammonia molecule per acid site.5 Even so, a convenient, direct, and noninVasiVe method for acidity determination is necessary for the routine analysis of variations in framework acid site concentration with catalyst synthesis conditions, postsynthetic treatments, and catalyst modification methods (e.g., final Brønsted acid site concentration vs synthesis gel SiO2/Al2O3, synthesis pH,6 template type and amount,7,8 dealumination,9,10 surface titrations,11 and hydrothermal treatment12). Such a technique is desirable because no assumptions about active site accessibility by a probe molecule are necessary. The ability to independently measure acid site concentrations in different zeolite types allows quantitative comparisons of acid site number vs acid site strength data in many hydrocarbon reactions. For example, one could imagine preparing a certain Si/Al ratio zeolite and comparing its activity/selectivity to a second catalyst that was synthesized with a lower Si/Al ratio but had been dealuminated to the same framework acidity as the original material. This would address postulates concerning creation of “unequal” or “enhanced” Brønsted acid sites via dealumination methods.13 Many assumptions concerning zeolite chemistry could be tested if the experimentalist could easily and accurately determine the Brønsted acid site density in any zeolite using a routine noninvasive method. In this contribution, we demonstrate the first quantitative 1H solid-state NMR data proving that Brønsted acid sites in SAPO34, a silicoaluminophosphate catalyst used for methanol-toolefin hydrocarbon chemistries, are not stable in the presence of moisture, even at ambient temperatures. In our pursuit of an experimental strategy which provides unambiguous proof of this effect, we show that a combination of solid-state magic-angle (MAS) spinning at moderate spinning speeds and 1H spincounting NMR with an internal standard provides an easily accessible, noninvasive, and completely quantitative way to
10.1021/jp0145816 CCC: $22.00 © 2002 American Chemical Society Published on Web 04/17/2002
4942 J. Phys. Chem. B, Vol. 106, No. 19, 2002 determine the number of acid sites in a zeolite or molecular sieve. We recognize that spin-counting methods are not new. Pfiefer and co-workers have reported early 1H MAS NMR studies in which the total hydrogen concentration was determined via comparison to an external (i.e., a separate experiment) spin-counting standard or via adsorption of gaseous bases.14-16 The key advantages of our method are that (1) no probe molecules are required, (2) the internal spin calibration standard is completely inert, (3) the standard may be handled as a solid along with the catalyst, and (4) only one measurement is required, thereby ensuring identical experimental response for all species of interest. More importantly, no assumptions regarding the accessibility of a specific probe molecule to different cage or channel sites in the catalyst are required. Finally, no assumptions about “nondetectable” spin behavior, as with 27Al NMR studies of framework aluminum sites, are present in this approach. All acidic and nonacidic hydroxyl groups are measured accurately and simultaneously. To be clear, this method does not provide any data on the relative strengths of acid sites between zeolites or within the zeolite, only the number of acidic Brønsted sites. Our method involves the incorporation of an internal spin-counting standard, poly(dimethylsiloxane) (PDMS), inside the MAS rotor with the catalyst sample. Liu and Maciel have previously reported comparisons of the quantitative aspects of 1H CRAMPS vs MAS-only experiments, in which a commercial silicone rubber was used as the internal spin-counting standard.17 Although their primary focus was on the multiple-pulse CRAMPS experiment, the necessary details for quantitative 1H spin-counting MASonly experiments were outlined. We have used this method, with some modifications, to obtain the results reported in this paper for zeolite and silicoaluminophosphate catalysts. Our criteria for selection of a suitable spin-counting standard were similar to the work by Liu and Maciel, and include the following: (1) the standard must be a solid for ease of handling; (2) the standard must be inert, suitable even for air and moisture sensitive catalysts; (3) the standard must have a well-resolved chemical shift to minimize interference with signals of interest; and (4) the standard must have a narrow 1H MAS NMR signal for any MAS spinning speeds >1 kHz. In cases where there are signals of interest near 0 ppm, i.e., near the PDMS signal, a siloxane polymer like polydiphenylsiloxane (PDPS) may be used for the spin-counting standard because the signals for this standard are near 7 ppm. All data in this paper use PDMS as the standard. The experimental method is described in detail in the following sections, with relevant examples from both silica and zeolite chemistry. A final example provides the first direct spectroscopic proof of the hydrolytic destruction of active sites in SAPO-34, a catalyst that is currently under investigation by many research groups for methanol-to-olefin conversions. Experimental Section NMR data were obtained on a Bruker DSX 300 MHz spectrometer using MAS at speeds of 7-11 kHz with 4 mm zirconium oxide rotors. Silica samples were obtained from Grace-Davidson, and the Ferrierite and ZSM-5 zeolites were obtained from Zeolyst Corporation. The SAPO-34 sample was synthesized by the method of Lok.25 The PDMS standard was obtained from Rheometrics Scientific, Inc. Zeolite samples were prepared from the ammonium form by a slow, stepwise calcination in flowing air at final temperatures of 550 °C on a glass vacuum line. Dehydrated zeolite samples were prepared via stepwise temperature increase to 400 °C under vacuum.
Wang et al. Silica samples were dehydrated under inert gas atmosphere at atmospheric pressure. All samples were loaded into MAS rotors in the glovebox to prevent exposure to moisture, and the sample was enclosed with a top and bottom press-fit Teflon spacer, as well as the Kel-F cap. Dry nitrogen was used for spinning during the experiment, and the sample was stored in nitrogen until transferred immediately to the probe. In practice, we found that the sample could sit exposed to the atmosphere with this doublecap arrangement for several days before there was any detectable change in the spectrum. The 1H T1 values for the PDMS, silica, and zeolite protons were all less than 2.5 s. Therefore, the 30 s recycle time used in our experiments results in quantitative peak intensities. Typically, 16 scans were obtained for each spectrum shown. Before the results on zeolite acidity are presented, it is important to describe the experimental requirements necessary to achieve quantitative spin response in a MAS NMR experiment. As was discussed in great detail by Campbell and English, commercial solid-state NMR probes do not provide homogeneous radio frequency excitation/detection over the entire sample volume of the MAS NMR rotor.18 However, nonuniform r.f. field regions are minimized for smaller rotor volumes. Our experiments were performed using a 4 mm magic-angle spinning rotor in which the sample region was confined to the middle region of the rotor; that is, the sample volume element was completely contained within the middle 1/5 element of the total rotor length. For our 4 mm MAS rotors, this confined central volume was very nearly spherical, even though strictly speaking the volume was cylindrical. In other words, the dimensions of the confined sample volume were a cylinder of length of 4 mm and 4 mm diameter. Sample confinement was achieved using lower and upper Teflon spacers on either side of the sample volume. Because each MAS probe has different radio frequency homogeneity, calibration prior to use is required by spincounting a sample of known concentration, e.g., a hexamethylbenzene/PDMS or camphor/PDMS sample, and Varying the restricted sample Volume until consistent and accurate mass balance is achieVed. This is necessary to ensure that each spin from the sample and the spin-counting standard are equally excited and detected in the measured signal response and cannot be overlooked even for commercial probes that utilize some type of standard spacer arrangement. Our results for the calibration experiment, when run in triplicate, resulted in the detection of 98, 101, and 99% of the expected signal based on the known weights of the sample. In other words, a known amount of hexamethylbenzene and a known amount of PDMS were introduced into the confined sample volume in the center of the rotor, as described above, and the spectrum was deconvoluted to determine their relative amounts. Of course, all sideband intensities were included in the calculation. Within the restricted volume region of the rotor, identical results were obtained no matter where the PDMS standard was placed, e.g., on the rotor wall at the top vs bottom vs in the center, thereby further confirming that the r.f. excitation and detection was quantitative. In contrast, when the same experiment was run without sample volume reduction, i.e., using the standard rotor volume, only 85% of the expected signal was measured. In this way, the method is clearly shown to be detecting all spins, because the line width of HMB is much, much greater than either PDMS or any signals from the zeolite protons. If the ringdown time of the probe was too long to accurately detect all 1H spins, as might be a concern, then clearly this would manifest itself in the calibration experiment, because the HMB would be underestimated. This was not the case. Furthermore, even
Quantitative Investigations of Acidity though the experiments were run on the 1H channel of a doubleresonance probe, we routinely also run CRAMPS experiments with 1.8 µs 90° pulse widths and 3.5 µs sampling windows. In our single-pulse experiments, it must be remembered that the 1H T values of both the zeolitic and PDMS protons are 2 sufficiently long to allow quantitative detection, and with a 4 µs acquisition delay, all spins should be detected. Furthermore, comparitive studies of zeolites by CRAMPS and single-pulse experiments provided identical results, thereby ensuring that π/2 pulse, or Bloch decay, experiments are quantitative for zeolites.17 Our experimental procedure was as follows: (1) an empty MAS rotor and volume-restricting spacers were weighed using a Sartorius MC-5 microbalance with 1-2 µg accuracy and repeatability; (2) a small amount of PDMS was weighed using the microbalance, and then the PDMS was placed on the inside wall of the rotor just above the bottom spacer; typically, 50100 µg of PDMS was used; (3a) for the moisture-sensitive zeolite and dehydrated silica samples, the PDMS-containing rotor was filled with 4-8 mg of the sample of interest inside the glovebox; (4) the total sample weight was determined using the microbalance; (5) the 1H MAS NMR spectrum was acquired using a simple π/2 pulse; (6) the spectrum was deconvoluted, and spectral areas were integrated to provide quantitative PDMS: sample signal ratios for each signal in the spectrum. Spectra were deconvoluted and integrated using the commercial software package PeakFit from Jandel Scientific. Deconvolution results to determine spectral areas for each species were obtained using least-squares fitting via minimization of residuals. Mixed Gaussian-Lorentzian line shapes were used in the analysis, and all correlation coefficient (R2) values exceeded 0.98. In every case, only one line was used in the fit for each physically relevant species in the sample, and sideband intensities, when observed, were included in the calculations. We have found this procedure to be highly reproducible for the determination of zeolite acid site concentrations, with a standard deviation for multiple experiments of 3%. Results and Discussion As a control experiment, a silica sample containing known -OH group concentration was measured. The results are shown in Figure 1 as a function of dehydration temperature under N2 atmosphere. The reader should first recognize that the PDMS standard does not affect the spectra in any way, other than the addition of a single, narrow signal at 0.1 ppm. In addition to serving as an internal quantitation standard, the PDMS also functions as an internal chemical shift reference. Figure 1 parts a and b are the MAS 1H NMR spectra for silica gel containing small amounts of residual moisture because of the lower dehydration temperatures of 150 and 400 °C, respectively. Residual moisture exists at these dehydration temperatures because the dehydration was done under atmospheric pressure; dehydration under vacuum would lead to removal of water at these temperatures. The key point is that by comparison of Figure 1a-c (150 vs 400 vs 800 °C), we can quantitatively determine the amount of residual moisture and the concentration of hydroxyl groups from the silica. Silica is an important support for many types of chemistry, and the critical factor is the number of available surface hydroxyl groups. Spin-counting results for the silica sample in Figure 1a-c reveal that the hydroxyl group concentration is 0.58, 0.61, and 0.60 mmol/g of bulk SiO2, respectively. Because this is the same SiO2 sample in each case, and only the dehydration temperature is varied, one should expect to obtain the same [Si-OH] concentration. The residual moisture in the 150 and 400 °C samples of Figure 1a-b are
J. Phys. Chem. B, Vol. 106, No. 19, 2002 4943
Figure 1. 1H MAS spin-counting results on a sample of silica dehydrated at (a) 150, (b) 400, and (c) 800 °C under nitrogen atmosphere at atmospheric pressure. The broad downfield feature seen in a and b results from residual H2O. The asterisk in c denotes a transient signal unrelated to the sample. The quantities of silica hydroxyl groups and residual water are reported in the text, and the deconvolution results are shown in Figure 2.
Figure 2. Example showing deconvolution results for the sample in Figure 1a. The experimental spectrum is overlaid with the total fit in the top spectra; individual components are shown in the bottom frame along with their assignments. (The horizontal axis is expanded relative to Figure 1.)
0.89 and 0.65 mmol/g SiO2, respectively. We note that hydrogen-bonded silanols, if they exist, do give rise to signals in the same reagion as the physisorbed water. However, the hydroxyl group concentration is constant in each experiment, and no residual broadening of the silanol signal is observed in Figure 1c, suggesting that proximate, hydrogen-bonded hydroxyl groups are not present in this sample (in the absence of moisture). Also, these data agree very well with wet chemical determination of [Si-OH] via methyl-magnesium halide titrations; our experience has shown agreement within 10% in most cases between the two methods for silicas dehydrated over a range of conditions. Of course, the ability to get data without resorting to wet or gas-phase titrations is appealing. The silica control experiments, as well as the hexamethylbenzene/PDMS calibrations discussed above, indicate that all zeolitic hydrogens are detected in the data described below, because zeolites have hydrogen densities much less than hexamethylbenzene (and therefore a longer T2) and comparable to the dehydrated silicas. Figure 2 shows an example of the spectral deconvolution in which three Gaussian-Lorentzian peaks were used to fit each spectrum, thereby revealing the quantities of water, silica hydroxyl groups, and PDMS. Of course for the 800 °C sample, only two peaks were used in the fit, because all water was removed at this temperature. Excellent fits with correlation
4944 J. Phys. Chem. B, Vol. 106, No. 19, 2002
Wang et al.
Figure 3. 1H spin-counting MAS spectra of zeolite (a) HZSM-5, Si: Al ) 15, (b) HZSM-5, Si:Al ) 25, (c) H-Ferrierite, Si:Al ) 10, (d) H-Ferrierite, Si:Al ) 27.5, and (e) HSAPO-34, Si/Al ) 0.19 and Si/ Si+Al+P ) 0.095. Additional small signals seen in some of the spectra near 3 (a) and 6 ppm (d) are explained in the text. The narrow PDMS peak is adjusted off-scale for clarity in the catalyst peak regions; however, the integrated intensity ratios of the H+ and PDMS peaks are near unity in each case.
coefficients in excess of 0.98 were obtained using only these three physically relevant components. We also note by comparing Figure 1a-c that line widths for the deconvolved silica hydroxyl (SiOH) peak at 1.9 ppm decrease with increasing dehydration temperature. The full-width at half-maximum line widths are 0.62, 0.45, and 0.33 ppm in Figure 1-c, respectively. Such results are consistent with a decrease in the amount of silica hydroxyl groups that are hydrogen-bonded to residual water molecules. Spin-counting experiments on zeolite solid acid catalysts are particularly useful because one can simultaneously determine the concentration of acidic bridging hydroxyl groups, nonacidic terminal silanols, and residual template or ammonium ions. Shown in Figure 3 are 1H MAS spin-counting spectra for the acidic zeolites H-ZSM5 (Figure 3a-b) and H-Ferrierite (Figure 3-d), each with different SiO2/Al2O3 ratios. SAPO34, a synthetic chabazite analogue where each acid site is generated via silicon substitution for aluminum and phosphorus, is shown in Figure 3e. Although only three examples are shown in this report, we have conducted numerous experiments on a variety of zeolite types, including chabazites, Beta, and HY, and the results here are representative of the general applicability of this method. As has been reported many times in the literature, Brønsted acid site protons in zeolites generally appear between 3.8 and 5.0 ppm, and nonacidic silanol groups appear near 1.7-2.1 ppm.19-21 In addition, other signals between 2.5 and 3.5 ppm are often detected in the 1H MAS spectra of zeolites. Such signals arise from nonframework or amorphous alumina species bearing hydroxyl groups, e.g., AlOH, and have been commonly observed in ZSM-5 and Y-type zeolite spectra.18 Finally, the presence of any residual ammonium ions from the exchange process or from incomplete removal of oxidized template molecules is revealed by signals in the 6-7 ppm range. For the SAPO-34 sample in Figure 3e, the Brønsted acid site gives rise to a peak at 3.9 ppm, and the nonacidic Al-OH and P-OH groups may be seen in the 1-2 ppm range. As can be seen by inspection of the five example spectra in Figure 3, as well as the deconvolution results in Figure 4, each of these components are observed and may be quantified by the spincounting technique. Figure 3a-b shows HZSM-5 spectra for
Figure 4. (a) Fit of total spectrum deconvolution (top) and individual peaks (bottom) for the SAPO-34 sample in Figure 3e. The peaks from 1 to 2 ppm come from nonacidic terminal hydroxyl groups. (b) Similar data in the top and bottom frames are shown for the HFER sample from Figure 3c.
samples with Si:Al of 15 and 25, respectively. The experimentally determined Brønsted acid site concentrations were 0.49 and 0.34 mmol H+/g of zeolite, respectively, for the samples in Figure 3a-b. The experimentally determined Brønsted acid site concentration of the H-FER samples in Figure 3c-d were 0.78 and 0.44 mmol H+/g of zeolite, respectively. Also, the H-FER sample in Figure 3d contains a small amount of residual NH4+ ions as seen in the spectrum; the concentration determined from spin counting was 0.025 mmol H+/g of zeolite. Finally, for the SAPO-34 sample in Figure 3e, the [H+] ) 1.10 mmol/g of molecular sieve. One particularly powerful benefit of the spincounting method is the ability to compare the number of measured Brønsted acid sites to that expected based on elemental Si/Al ratios of the final catalyst. For example, the H-FER sample in Figure 3c has Si:Al ) 10, whereas the sample in Figure 3d has Si:Al ) 27.5. On the basis of the structure of ferrierite, one can calculate the number of Brønsted acid sites that would occur if all Al were incorporated in the framework. Such a calculation for the samples in Figure 3c indicates that only 78% of the elemental Al is in the framework, whereas 95% of framework Al occurs for the lower Si:Al sample in Figure 3d (total Bronsted H+ + NH4+). This is in everyway consistent with the data, because a significant amount of nonframework Al-OH species is observed in Figure 3c as the broad signal near 2-3 ppm and in the detailed fit shown in Figure 4b for this sample. Also, comparison of the high Al ZSM-5 in Figure 3a shows more nonframework Al-OH signal than the low Al catalyst in Figure 3b. It is unlikely that every nonframework Al atom has an -OH group attached to it, and therefore, one should not expect complete agreement based on the elemental analysis data. Our experience to date has shown
Quantitative Investigations of Acidity
Figure 5. 1H MAS spin-counting spectra for the SAPO-34 sample shown in Figure 4a as a function of elapsed time following removal of the tetraalkylammonium template molecule. (a) Immediately following template removal and dehydration, but with exposure to atmosphere for 3 h between the template removal and dehydration steps; (b) same as in part a, but after 7 days of storage in a nitrogen glovebox; (c) same as in part b, but after 10 days of storage in a glovebox; (d) same as in part c, but following reexposure to atmosphere for 3 h and a second dehydration step.
that as the Si:Al ratio decreases the percent of total Al that exists as framework Al decreases. In other words, the agreement between the Brønsted acid site concentration and total Al content decreases as the Al content increases. These results depend both on the ability to incorporate more Al from the synthesis gel into the framework during crystallization, as well as the hydrothermal stability of the framework Al during calcination and dehydration. For the SAPO-34 sample in Figure 4a, the experimentally measured value for the Brønsted [H+] concentration of 1.12 mmol/g is less than the 1.5 mmol/g expected based on Si elemental analysis (i.e., assuming each Si leads to an acid site). The 1.12 mmol H+/g value is in agreement with previously published values for this catalyst.26 As reported by Barthomeuf, there are clear inconsistencies for SAPO catalysts, in that several mechanisms exist for Si substitution in the ALPO framework. Some of these incorporate multiple Si sites in close proximity, thereby reducing the number of acid sites from what would be expected based on Si elemental analysis.22-24 An acid site is created when Si substitutes only for P; if two Si atoms substitute for an Al and a P, no acid site is formed. Previous work by Barthomeuf has shown that these acid-site generating isolated Si sites cannot exist in SAPO-34 above an atomic fraction of 0.10-0.11.23 Also, Si can be incorporated in amorphous SiO2 phases, and Al may be incorporated as amorphous Al2O3 phases, thereby rendering comparisons of elemental Si+P/Al ratios somewhat meaningless in relation to acidity. As such, there is a need to compare complete material balance for Si to the experimentally measured Brønsted acid site concentrations. The spin-counting method introduced here can address this problem, because the PDMS 29Si signal can also serve as an internal spincounting standard for the different Si sites, e.g., those with differing numbers of Al neighbors in the lattice, as measured using 29Si NMR.23 Therefore, in one sample preparation, we can collect quantitative 1H and 29Si NMR data with an internal reference peak in each spectrum for calibration. Of course, this will be a time-consuming experiment in the 29Si case, but the information will prove valuable for understanding Si incorporation in SAPO catalyts. Our experiments to date have clearly revealed that the acid site density in SAPO-34 is not a constant. Figure 5 shows 1H
J. Phys. Chem. B, Vol. 106, No. 19, 2002 4945 MAS spin-counting NMR data on the SAPO-34 sample shown in Figure 4 but as a function of sample exposure to various environmental conditions following removal of the tetraethylammonium template used in the synthesis. As described in the figure caption, we have acquired these data based on exposure times to either ambient atmosphere, i.e., moisture, or to a dry nitrogen glovebox environment. Visual inspection shows that the Brønsted acid site peak near 4 ppm decreases its intensity in Figure 5a-d. Quantitative analysis of this acidity loss, using the spin-counting method, reveals acid site densities of 1.12, 0.95, 0.87, and 0.42 mmol/g in Figure 5a-d, respectively. These results directly address the stability questions raised by Briend and co-workers, in which they observed different crystallinities and porosities in similar SAPO catalysts as a function of catalyst preparation and exposure. They attributed this behavior to attack by water of the framework bonds around tetrahedral atoms in the catalyst.22 Our results indicate unequivocally that this leads to a loss of acidity. The fact that the acidity loss is severely retarded by storing the catalyst in an inert atmosphere is strong evidence that framework hydrolysis is the active mechanism. The acidity loss appears to be irreversible, because subsequent vacuum heating/dehydration steps do not lead to any increase in acidity. Recall that immediately prior to analysis, the sample is dehydrated stepwise up to temperatures of 450 °C, all the while maintaining a vacuum at least at 10-3 Torr. Also, the nonacidic hydroxyl peaks between 1 and 1.6 ppm do not change in these experiments. Spin-counting of these species for each of the spectra in Figure 5 indicates a constant concentration of 0.45 ( 0.02 mmol total nonacidic hydroxyl groups per gram of catalyst, further showing that the spin-counting technique is very reproducible from sample to sample. Although these experiments quantify the acidity loss specifically after template removal, the fact that the largest acid site density that we have measured is only 70-75% of the theoretical limit suggests that acid site hydrolysis may also occur when the template is present, albeit at a much slower rate. The transient nature of the active sites in SAPO-34 must be considered when catalysis experiments involving this catalyst are carried out, particularly those involving methanol-to-hydrocarbon synthesis.26-28 In conclusion, we believe that this simple, accurate, and highly reproducible method for experimental determination of Brønsted acidity in essentially any zeolite or solid acid molecular sieve will prove extremely useful in a variety of catalyst science areas. Because of the high sensitivity afforded by 1H detection, the reader will recognize that the method is completely general and may be applied to any sample in which resolved 1H peaks are observed in the MAS spectrum (e.g., metal oxides, functionalized nanoparticles, thin films on surfaces, etc.). In particular, the ability to determine acidity as a function of catalyst synthesis conditions or postsynthetic modifications such as steaming, ionic stabilization, or acid site titrations will be most interesting. For example, we have done controlled steaming experiments on HZSM-5 in which the Bronsted acidity loss caused by dealumination is easily detected and quantified. The specific application of this method to SAPO-34 revealed, for the first time, that Brønsted acidity is lost when the calcined material is exposed to the moisture. Preliminary data also suggests that acid density decreases in SAPO-34 over time even prior to template removal, albeit at a much slower rate. We are conducting additional experiments in which a full multinuclear NMR strategy is being used in conjunction with the spin-counting work to reveal the mechanistic pathways for Si incorporation and acidity losses in SAPOs.
4946 J. Phys. Chem. B, Vol. 106, No. 19, 2002 Acknowledgment. The authors gratefully acknowledge support from the National Science Foundation (DMR-0137968) and from the NSF-REU program (Grant 0097485). References and Notes (1) Lunsford, J. H.; Rothwell, W. P.; Shen, W. J. J. Am. Chem. Soc. 1985, 107, 1540. (2) Sutovich, K. J.; Peters, A. W.; Rakiewicz, E. F.; Wormsbecher, R. F.; Mattingly, S. M.; Mueller, K. T. J. Catal. 1999, 183, 155. (3) Kotrel, S.; Rosynek, M. P.; Lunsford, J. H. J. Catal. 1999, 182, 278. (4) Farneth, W. E.; Gorte, R. J. Chem. ReV. 1995, 95, 615. (5) Parrillo, D. J.; Adamo, A. T.; Kokotailo, G. T.; Gorte, R. J. Appl. Catal. 1990, 67, 107. (6) Lechert, H. Microporous Mesoporous Mater. 1998, 22, 495. (7) de Moor, P.; Beelen, T. P. M.; van Santen, R. A.; Beck, L. W.; Davis, M. E. J. Phys. Chem. B 2000, 104, 7600. (8) Shantz, D. F.; Fild, C.; Koller, H.; Lobo, R. F. J. Phys. Chem. B 1999, 103, 10858. (9) Triantafillidas, C.; Vlessidis, A. G.; Evmiridis, N. P. Ind. Eng. Chem. Res. 2000, 39, 307. (10) Apelian, M. R.; Fung, A. S.; Kennedy, G. J.; Degnan, T. F. J. Phys. Chem. 1996, 100, 16577. (11) Weber, R. W.; Moller, K. P.; O’Connor, C. T. Microporous Mesoporous Mater. 2000, 35-6, 533. (12) van Bokhoven, J. A.; Konignsberger, D. C.; Kunkeler, P.; van Bekkum, H.: Kentgens, A. P. M. J. Am. Chem. Soc. 2000, 122, 12842.
Wang et al. (13) Ma, D.; Deng, F.; Fu, R.; Dan, X. W.; Bao, X. H. J. Phys. Chem. B 2001, 105, 1770. (14) Freude, D.; Hunger, M.; Pfeifer, H. Chem. Phys. Lett. 1986, 128, 62. (15) Pfeifer, H.; Freude, D.; Hunger, M. Zeolites 1985, 5, 274. (16) Freude, D.; Hunger, M.; Pfeifer, H. Chem. Phys. Lett. 1982, 91, 307. (17) Liu, C. C.; Maciel, G. E. Anal. Chem. 1996, 68, 1401. (18) Campbell, G. C.; Galya, L. G.; Beeler, A. J.; English, A. D. J. Magn. Reson. 1995, 112 A, 225. (19) Michel, D.; Engelhardt, G. High-Resolution Solid-State NMR of Silicates and Zeolites; John Wiley & Sons: New York, 1987. (20) Liu, H.; Kao, H.; Grey, C. P. J. Phys. Chem. B 1999, 103, 4786. (21) Beck, L. W.; White, J. L.; Haw, J. F. J. Am. Chem. Soc. 1994, 116, 9657. (22) Briend, M.; Vomscheid, R.; Barthomeuf, D. J. Phys. Chem. 1995, 99, 8270. (23) Barthomeuf, D. J. Phys. Chem. 1993, 97, 10095. (24) Briend, M.; Vomscheid, R.; Barthomeuf, D. J. Phys. Chem. 1994, 98, 9614. (25) Lok, B. M.; Messina, C. A.; Patton, R. L.; Gajek, R. T.; Cannan, T. R.; Flanigan, E. M. U.S. Patent 4,440,871, 1984. (26) Song, W.; Fu, H.; Haw, J. F. J. Am. Chem. Soc. 2001, 123, 4749. (27) Arstad, B.; Kolboe, S.Catal. Lett. 2001, 71, 209. (28) Song, W.; Fu, H.; Haw, J. F. J. Am. Chem. Soc. 2000, 122, 10726.