Document not found! Please try again

“Nanoblocks” in the Clear Solution Synthesis of ... - ACS Publications

Colin A. Fyfe,* Richard J. Darton, Celine Schneider, and Franziska Scheffler. Department of Chemistry, R.300 Wesbrook Building, UniVersity of British ...
0 downloads 0 Views 340KB Size
80

J. Phys. Chem. C 2008, 112, 80-88

Solid-State NMR Investigation of the Possible Existence of “Nanoblocks” in the Clear Solution Synthesis of MFI Materials Colin A. Fyfe,* Richard J. Darton, Celine Schneider, and Franziska Scheffler Department of Chemistry, R.300 Wesbrook Building, UniVersity of British Columbia, VancouVer, British Columbia V6T 1Z3, Canada ReceiVed: October 1, 2007

The structure of the intermediate species in the clear solution synthesis of the MFI framework (zeolite ZSM 5) has been investigated using the characteristic 13C, 14N, 15N, and 2D spectra of the tetrapropylammonium (TPA) template ions as probes as well as the 29Si spectra of the silicate species present in samples that can be isolated by centrifugation. Comparison with the corresponding spectra of the final products that can be characterized by X-ray diffraction indicates that there is no evidence for the involvement of nanospecies, as has been proposed, and that crystallization is most probably from an amorphous gel. This conclusion is supported by the lack of deuterium rotational echo double-resonance dephasing of the 29Si spectra by deuterated TPA of the earliest intermediate species obtained, while it is clearly observed in the final product. These observations indicate that any TPA ions present in the gel phase are not in intimate contact with the silicon nuclei as they would be if in the local MFI environment. This is supported by the very low amount of TPA found by 14N NMR, which is much less than needed for the proposed nano intermediates and the fact that the TPA present can be removed by simple re-suspension in water and recovery.

Introduction Zeolites are important microporous materials that have been used extensively as catalysts and adsorbents due to their high size and shape selectivities to small organic molecules.1,2 The usual synthesis is hydrothermal from a gel containing the appropriate metal oxides at high temperature under very basic conditions.3 More recently, zeolite syntheses from clear solutions have been reported (LTA,4 MOR,5 FER,6 MFI7), which can yield small (ca.100-200 nm) crystals at much more moderate temperatures. Possible intermediates and the mechanism of formation of these final “nanocrystals” have been studied by various researchers,8-15 most extensively by the group of Martens,16-24 using a wide variety of physical techniques. They have proposed that a 33-atom SiO4 tetrahedral building block is formed under the influence of the teterapropylammonium (TPA) ion template, which is retained inside the unit, which has dimensions of 1.3 × 1.0 × 1.3 nm, Figure 1A. Four of these are postulated to join together to form “nanoslabs” of dimensions 1.3 × 4 × 4 nm3, containing channel intersections, each of which is occupied by a template ion as shown in Figure 1B. These are then proposed to further combine, forming “nanoblocks”, Figure 2B, which then further combine to generate the extended framework of the MFI crystal, Figure 1C. However, this proposal has been severely criticized by Lobo and co-workers25-33 who cite low-angle X-ray and neutron diffraction experiments as evidence for the formation of amorphous colloidal silica particles and the subsequent crystallization of the MFI nanocrystals. The amorphous silica particles are believed to have some trapped TPA ions but are without any significant well-defined ordered structure. Further, Knight and Kinrade34,35 have criticized the initial 29Si solution NMR experiments used to assign the structures of both the basic * To whom correspondence should be addressed. E-mail: fyfe@ chem.ubc.ca.

Figure 1. Schematic representations of precursor containing TPA cation and nanoblock with included templates. Reproduced with permission from refs 18 and 19.

Figure 2. Schematic representation of the formation of Silicalite-1 material from clear solutions at elevated temperatures. Reproduced with permission from ref 19.

building unit itself shown in Figure 1A and of the proposed precursor species to it, as Martens and co-workers did not use 29Si-enriched materials, and thus Si-Si connectivity information could not be obtained. Very recently, Knight and Kinrade have carried out such experiments using 29Si enrichment on the intact clear solution reaction mixtures and suggest that the resonances assigned to the basic building unit and its precursors by Martens

10.1021/jp7095955 CCC: $40.75 © 2008 American Chemical Society Published on Web 12/13/2007

Solid State NMR Investigation

Figure 3. Schematic representation of the MFI channel structure showing the position and orientation of the MFI template at the channel intersection, as viewed down the straight channel. The dashed red lines indicate the location of the zigzag channel.

and co-workers are, in fact, due to other, previously known, structural units.35 Nanomaterials, in general, represent a unique challenge in terms of solid-state structure determinations as Bragg reflections in X-ray diffraction (XRD) broaden significantly below particle dimensions of 50-100 nm and eventually disappear completely. In contrast, solid-state NMR (SSNMR) is independent of particle dimensions, although the structure of the particle may well change, and this technique therefore represents a general approach to bridging this gap both in the present case and also in other systems. Thus, a system may be characterized by SSNMR for large crystals where X-ray diffraction is also viable and where a single-crystal structure can be obtained. These data can then be used as benchmarks and the various NMR spectral characteristics tracked to smaller and smaller particles, eventually overlapping with high resolution solution NMR. The progress of syntheses and species involved in them can thus be monitored if solid samples can be obtained at intermediate times. It is known that the presence of organic molecules and template ions alters the framework structure of MFI36,37 and that dipolar interactions between sorbate nuclei and the framework nuclei can be used to determine the structures of organic molecule-zeolite framework complexes.38-42 We reasoned that the reverse should also be true; that is, the conformations of flexible template ions must be determined by the framework structure, and the NMR spectral parameters of their nuclei should be diagnostic of the formation of the surrounding zeolite framework. In the case of MFI, the TPA ion resides at the intersection of the straight and zigzag channels with the four propyl groups along the channels, two in opposite directions along the straight channel and two in separate zigzag channels (Figure 3), and its conformation should provide a diagnostic probe of the formation of this basic structural arrangement that could be documented from the study of well-characterized MFI materials. It should also reflect the presence of the proposed nanoslabs and all subsequent aggregates, where exactly this arrangement is postulated. Further, the through-space dipolar couplings between the nuclei on the template and framework silicon nuclei should provide confirmatory evidence for the close-packed structural arrangement between them that this structure involves. In the present work, we present the results of applying this approach using 29Si, 2H, 27Al, 15N, 14N magic angle spinning (MAS) NMR, 1H/29Si, 1H/13C cross polarization (CP) MAS, and 1H/29Si/2D CP rotational echo double-resonance (REDOR) NMR experiments together with data from complementary techniques to the characterization of the intermediate species isolated from the clear-solution synthesis of nano MFI. Experimental Syntheses followed the general procedure of Schoeman.14 In a typical synthesis, tetraethylorthosilicate (TEOS, 3.75 g, 18

J. Phys. Chem. C, Vol. 112, No. 1, 2008 81 mmoles) was added dropwise to a stirred solution of aqueous TPA hydroxide (6.95 g, 1M, 6.9 mmoles) in a 50-mL conical flask over the period of 1 h. The mixture was then stirred at room temperature for 24 h before being equipped with a reflux condenser and heated to 90 °C. The mixture was held at 90 °C under constant stirring, and samples were taken at regular intervals, with the progress of the reaction being monitored by low-angle capillary XRD and particle size measurements. Aluminum-containing samples were synthesized in a similar manner except that aluminum nitrate was substituted for a small amount of the TEOS. Samples for SSNMR were obtained by both centrifugation at a speed of 5000 rpm followed by vacuum drying and by drying samples alone under vacuum. 15N-labeled TPA iodide was synthesized by the reaction of 15N-labeled ammonia gas (Cambridge Isotope Laboratories) with 1-iodopropane in a sealed system. The specifically 2H-labeled TPA ion was synthesized by the reaction of n-propylamine with d7-labeled 1-iodopropane (C/D/ N Isotopes) in propan-1-ol under reflux conditions. Both the 15N- and the 2H-labeled TPA ions were subsequently converted from the iodide salts to the corresponding hydroxides using Ambersep-900 OH ionexchange resin (Fluka). SSNMR spectra were obtained using a Bruker Avance 400 spectrometer operating at frequencies of 400.13 MHz for 1H and 79.494, 104.261, 100.622, 40.560, 61.423, and 28.915 MHz for 29Si, 27Al, 13C, 15N, 2H, and 14N, respectively. All the experiments were carried out using either Bruker 4- or 7-mm MAS probes at ambient temperature. 29Si chemical shifts were referenced to tetramethylsilane (TMS) with Q8M8 (the octamer Si8O12[OSi(CH3)3]8) as external secondary reference and 13C chemical shifts to TMS with adamantane as an external secondary reference. 14N and 15N chemical shifts were referenced to ammonium chloride, 27Al shifts to aqueous acidic aluminum nitrate, and 2H chemical shifts to the isotropic shift of solid deuterated hexamethylbenzene. Low-angle XRD experiments were performed on 1.0 mm diameter capillary samples of the reaction mixtures using a Bruker D8 Discover Diffractometer fitted with a General Area Detector. Further details of specific experiments are given in the figure captions. Scanning electrom microscopy was performed using a Hitachi S-2300, and particle size measurements were carried out using a Malvern Zetasizer 3000HS fitted with a He-Ne laser at a wavelength of 633.0 nm. The instrument was calibrated with a suspension of polystyrene particles of 100 nm diameter. Results and Discussion As indicated in the introduction, the key feature of the mechanism of the clear solution synthesis proposed by Martens and co-workers is the postulate that the TPA ion is involved at a very early point in the mechanism, exhibiting a templating function to produce nanounits (nanoslabs). In the present work, we have probed the environment of the TPA ion during the synthesis by characterizing its structure, local environment, and interactions with the silicate species in the mixture. These were compared to the features of the template in its (known) final channel intersection environment in the MFI framework using several benchmark samples of high quality and crystallinity. Experiments were carried out using the nuclei indicated in structure 1 and appropriate analogs of the TPA ion incorporating stable isotopes were synthesized as described in the Experimental Section in order to achieve this. Due to the complexity of this system and the subtleties of the distinctions that have to

82 J. Phys. Chem. C, Vol. 112, No. 1, 2008

be made, many different experiments have been carried out. For clarity of presentation, these are discussed under different headings. Isolation Procedure and Characterization of the Reaction. Previous studies involved isolation of intermediate species during the reaction by simple evaporation or by acidification with HCl and extraction by t-butanol and then evaporation/freeze drying.17 Indeed, one of the initial reports16 indicated that TPA ions were present in isolated material as evidenced by a solidstate 13C NMR spectrum, although it was not subsequently published. However, both of these procedures are problematical due to the very large excess (6.7 fold) of TPA ion used in the synthesis. Samples prepared by both techniques will show the presence of TPA, the former directly from the excess of template and the latter because various TPA salts have finite solubilities in the large excess of t-butanol used in the extraction. Given these problems, it was decided to investigate solid materials isolated by centrifugation (5000 rpm was used) as it was felt that this would involve least perturbation, and any TPA ions present as either counterions for framework charges or enclosed in channel intersections should be unaffected and precipitate with the silicate framework species. The earliest reaction times at which solid samples could be obtained were at 30-40 h, at which times the particle sizes were ca. 50 nm (see below). Information on the overall progress of the reaction was obtained from particle size measurements (Figure 5). At zero time at 90 °C (after 24 h at room temperature after mixing), only particles less than 10 nm diameter are present. At 20 h, 10 nm diameter particles are also present in a bimodal distribution with 3 nm diameter particles, while after 40 h, only larger particles are present whose average size distribution at 90 °C gradually changes with time from 50 to 80 nm diameter. The observation of very small particles at the shortest times is in good agreement with the literature, while the final dimensions agree with those obtained from electron microscopic measurements on the final crystalline product (not shown). Low-angle XRD scattering from capillary samples of the reaction mixture was also used to monitor the progress of the reaction to ensure it was reproducible and was carried out on all preparations. Figure 4 shows typical patterns that are in good agreement with previous literature.21 Under the reaction conditions described in the Experimental Section, at 90 °C, only a single broad peak at ca. 3° 2θ is observed at zero time (after 24 h at room temperature) that gradually moves to higher 2θ values and broadens with time. This peak has been interpreted in terms of X-ray scattering from the nanoslab (1 × 2 × 2 nm3)19 species, but a similar pattern is seen in cases where TPA is not the counterion27 and also in colloidal silica. After 40 h at

Fyfe et al.

Figure 4. Low-angle XRD measurements of the intact reaction mixture at 90 °C at the times indicated.

Figure 5. Particle size measurements at the times indicated on the 10-fold diluted reaction mixture.

Figure 6. Comparison of the powder XRD patterns for as-made F-MFI and the final 80-nm MFI nanocrystals obtained after 200 h.

90 °C, only a slight shift in the peak maximum is seen, and it is now possible to obtain the first sample by centrifugation for solid-state NMR analysis. At this stage size measurements show only a single particle size distribution centered about 50 nm diameter and the corresponding X-ray diffraction pattern shows no evidence for long-range order. At 60 h and longer times, there is the appearance of Bragg reflections, diagnostic of the formation of the MFI framework. These increase in intensity, coupled with further shifting of the broad peak to higher 2θ values and substantial loss of its maximum peak intensity together with further substantial broadening. Further samples were collected during this time period for study. The powder X-ray pattern of the isolated final product (after 200 h) agrees exactly with that of large particle MFI (Figure 6). Characterization of Reference Materials and Final Product. Figure 7 shows the 1H/29Si CPMAS, 1H/13C CPMAS, 1H/ 15N CPMAS, and 14N MAS spectra obtained for reference

Solid State NMR Investigation

J. Phys. Chem. C, Vol. 112, No. 1, 2008 83

Figure 8. 14N MAS spectra of F-MFI obtained at spinning rates of (a) 15 kHz (1024 scans), (b) 5 kHz (1024 scans), and (c) 2.8 kHz (10240 scans). Spectra were collected with a 9-µs pulse, 5-s recycle time, and 50-Hz line broadening.

Figure 7. Characteristic spectra of as-made F-MFI. (a) 29Si{1H} CP MAS spectrum at 3 kHz and a contact time of 3 ms (2048 scans, 3-s recycle time). (b) 13C{1H} CP MAS spectrum at 8.5 kHz and a contact time of 2.5 ms (512 scans, 2-s recycle time). (c) 15N{1H} CP MAS spectrum at 3 kHz and a contact time of 20 ms (16 scans, 2-s recycle time). (d) 14N MAS spectrum at 15 kHz with a 9-µs pulse, 5-s recycle time, and 50-Hz line broadening.

samples of F-MFI, chosen because this is the most perfectly ordered MFI framework obtainable in the as-synthesized form containing template. The relevant experimental parameters are given in the figure caption. The silicon spectrum of this sample shows no Q3 signal and sharp, relatively well-resolved Q4 resonances, reflecting the high degree of crystallinity. The cross polarization process is very efficient due to the short proton spin-lattice relaxation time (T1) and the intimate contact between the template protons and the framework silicons. A similar spectrum is obtained from single pulse experiments but in a much longer time due to the much longer 29Si T1 value (∼60 s). It is noteworthy that, in both cases, all the silicon signals are due to Q4 units, indicating the absence of defects. The 1H/13C CPMAS spectrum shows clearly resolved signals for the three different carbons in the propyl groups, diagnostic of the TPA ion location at the channel intersection with the chains extending into the straight and zigzag channels is the splitting of the methyl resonance into two signals.44 A smaller but identifiable splitting is seen here for the middle methylene carbon due to the high crystallinity but is lost in other samples. These observations are in good agreement with the results of a recent single crystal study.45 The 15N spectrum was obtained using an enriched TPA (100%) as it was anticipated that the 15N chemical shift anisotropy would be a diagnostic feature of the environment of the TPA ion at the channel intersection. However, the anisotropy is very small, indicating little deviation from perfect tetrahedral symmetry around the nitrogen. The isotropic shift is, however, diagnostic although the range of values for reference TPA samples with different counterions is small. Although 15N spectra were obtained for samples at all the different time points, it was found that the 14N MAS spectra were much more useful. These spectra, shown in Figures 7d and 8 for F-MFI, yield quadrupolar parameters of CQ ) 31.5 kHz and ηQ ) 0.9, in agreement with those from a static spectrum, although it is much more time consuming to obtain the latter spectrum. Thus, the 13C and 14N spectra of the template seem the most diagnostic of the TPA environment, and these were obtained for all samples isolated.

Figure 9. Characteristic spectra of as-synthesized MFI without fluoride ions. (a) 29Si{1H} CP MAS spectrum at 3 kHz with a contact time of 11 ms (1024 scans, 4 s recycle time). (b) 13C{1H} CP MAS spectrum at 8.5 kHz with a contact time of 1 ms (1024 scans, 4 s recycle time, 10 Hz line broadening). (c) 14N MAS spectrum at 15 kHz with a 9-µs pulse (6056 scans, 5-s recycle time, 50-Hz line broadening).

The F-MFI sample is the most highly crystalline material obtainable, but it contains a fluorine atom bonded to the framework.45 However, it is also possible to obtain assynthesized samples of almost as high a quality with template but no framework fluoride by standard alkaline synthesis, and one such sample was studied in order to rule out any major effects of the fluorine on the spectra. The spectra of this sample are shown in Figure 9 and are identical except for slight line broadening in some cases reflecting the somewhat less perfect ordering. The 29Si spectrum shows almost no Q3 signals and some resolution (it can be simulated almost exactly by applying 60 Hz line broadening to the corresponding spectrum of F-MFI in Figure 7a) and the 13C and 14N spectra are identical to those in Figure 7. These spectra are thus considered diagnostic of the TPA environment at the channel intersection of the MFI framework. Investigation of the Materials Formed During the Reaction. Figures 10 and 11 show the 1H/13C and 1H/29Si CPMAS spectra, respectively, of the final product and intermediate samples taken after the reaction times at 90 °C indicated. The final nanocrystal product (200 h) clearly shows the characteristic “doublet splitting” of the methyl resonance diagnostic of the TPA environment in the MFI framework as do all of the other

84 J. Phys. Chem. C, Vol. 112, No. 1, 2008

Figure 10. 13C CP MAS NMR spectra of samples isolated by centrifugation after reaction at 90 °C for the times indicated. Spectra were collected at 8.5 kHz with a 2.5 ms contact time (10240 scans, 5-s recycle time, no line broadening).

Figure 11. 29Si MAS NMR spectra of samples isolated by centrifugation after reaction at 90 °C for the times indicated. Spectra were collected at 8.5 kHz with a 10-ms contact time (1024 scans, 5-s recycle time, no line broadening).

spectra after 60 h. Since all of the spectra were obtained using the same parameters and since approximately equal quantities of sample were used, a rough crystallization curve can be obtained (not shown). The spectrum of the sample after 40 h reaction time showed much less intense carbon signals for the same number of acquisitions. There are signals that approximate the chemical shift values of propyl groups and additional ones due to small amounts of ethanol and tetraethyl orthosilicate, but there are no dominant methyl signals showing the doublet pattern characteristic of TPA at the special channel intersection location. Thus, there are few, if any, ordered MFI environments or crystalline material present, in agreement with the powder XRD results in Figure 4. The 29Si CPMAS spectrum of the final crystalline material (200 h) shows a Q3/Q4 ratio of 0.21, typical of highly ordered MFI. There is a shoulder at high field, reflecting the orthorhombic as-synthesized structure with broadening due to the presence of some defects (Figure 11). When the sample is calcined, a quantitative single 29Si spectrum shows a large Q4 signal as expected but only a very small Q3 resonance (∼4%), indicating very few defects are now present and that the crystals

Fyfe et al. have surface areas appropriate to their sizes (we estimate that (12 × 12 × 80) nm3 crystals should have about 20% Q3 and 100 nm crystals ∼2.5% Q3). Thus they cannot be agglomerates of very small (∼10-20 nm) individual crystals, as has been claimed to be the case for the final products from some syntheses carried out under similar conditions.46 At shorter reaction times, the Q3/Q4 ratios from the 29Si CPMAS spectra begin at a value of ca. 7 (40 h), similar to those of authentic silica gel materials where the CP source is the Q3 hydroxyl groups and gradually changes through subsequent samples to reach the limiting value of the final product. In fact, subtracting a contribution from the final spectrum weighted with the appropriate percentage of crystalline material obtained from the 13C spectra yields in each case a spectrum like that of the 40-h sample, which, as noted above, is similar to that of a typical silica gel.47 A sample containing a small amount of aluminum was also studied. The spectra (not shown) from single-pulse 27Al experiments, using small pulse angles to ensure quantitative reliability, showed a signal characteristic of tetrahedrally coordinated aluminum for all samples. There were small chemical shift differences between the peak maxima of the sample isolated at 40 h (-58 ppm) and the final 200-h product (-54 ppm), and the spectra at intervening times, similar to the corresponding 29Si and 13C spectra discussed above, could be deconvoluted in terms of these two limiting cases, but we do not feel that the differences are large enough to draw any definite conclusions from these data. Previous work has shown that the T2 relaxation time of tetrahedral aluminum in an amorphous environment was shorter than when it is in an ordered framework,48 and it was thought that this might also be used as a probe for the presence of the MFI framework. Spin echo spectra of the final MFI crystals and the sample obtained at 40 h showed that the latter relaxed much faster than the crystalline final product, but again it was felt that the result did not definitively prove that the initial sample was amorphous, although it is in agreement with the behavior expected. Figure 12 presents the 14N spectra of samples obtained at the times indicated and show that the TPA ions are very mobile up to 40-h reaction time, after which the spectra show the presence of material with much reduced mobility, the final spectrum being identical to that of completely crystalline MFI (shown in Figures 8 and 9.) Further structural information on the species present may be obtained from Figure 12, as the spectra were obtained under identical conditions on weighed amounts of material. This makes it possible to calculate the N/SiO2 ratio, which should be invariant if the 40-h and subsequent samples contain “nanoslabs” as these have been postulated to have one template per channel intersection, exactly the ratio as in the final as-synthesized MFI material. Such a plot is shown in Figure 13, which shows the N/SiO2 ratios of the samples normalized to that of the final crystals. It is clear that the ratio of ∼0.75 of the 40-h sample precludes the sample being composed of any “templated” species as proposed by Martens et al. It should be emphasized that particle size measurements at this reaction time show a single distribution centered at 50 nm, with no evidence at all for species smaller than ∼35 nm, and that XRD shows no evidence of crystalline material. It thus appears that the sample is amorphous silica gel with a small amount of associated template present from entrainment of some of the very large excess of TPA ions during the centrifugation process. Indeed, repeated resuspension and centrifugation of the 30h material from distilled water (three

Solid State NMR Investigation

J. Phys. Chem. C, Vol. 112, No. 1, 2008 85

Figure 14. 2D MAS NMR spectra at 3 kHz for (a) as synthesized F-MFI, (b) final 80-nm crystals (200 h), and (c) first isolated sample (30 h). All spectra were obtained using a 6-µs, 90° pulse with a recycle time of 1 s and 20-Hz line broadening.

Figure 12. 14N MAS spectra of samples obtained after reaction at 90 °C for 40, 72, 80, 100, 120, and 200 h. Spectra were collected with 5120 scans at 12.5 kHz with an 8-µs pulse, 7-s recycle time, and 30Hz line broadening.

Figure 15. Experimental 2D/29Si CP REDOR difference curve for assynthesized F-MFI.

protons from water in the atmosphere and with water and hydroxyl protons on surfaces. For this reason, we have opted for a more complex, but completely unambiguous, experiment, where the TPA was deuterated and the silicon magnetization dephased by a 2D REDOR experiment. To take advantage of the short 1H T1 values of protonated TPA to create the 29Si magnetization, three of the propyl groups were deuterated, and one was left protonated to provide a source for magnetization transfer, structure 2. This is because the dephasing experiment Figure 13. Plot of reaction time vs relative 14N intensity.

times) yielded over a 10-fold decrease in the calculated N/SiO2 ratio, indicating that the TPA ions are very easily removed. Investigation of the Possible Interactions between the TPA Ions and Silicate Species. In the previous experiments, the interaction between the TPA cations and the silicate species present were probed indirectly by their effect on the NMR characteristics of the templating cation. In this section, we attempt to address more directly these interactions by probing the distance dependent dipolar-based magnetization transfer between nuclei on the cation and the silicon nuclei of all silicate species present in its immediate environment. In principle, it would be possible to use the protons of the TPA and experiments such as CP, CP drain,49 or REDOR50 for this purpose. However, these and other such experiments directly involving magnetization transfer to or from protons are complicated by possible transfers involving the OH groups of Q3 silicons present. Although these can be deuterium exchanged, our previous experience with silica gels has been that it is very difficult to obtain quantitative exchange and to avoid back exchange with

is not quite as efficient as it would first appear, as the silicon nuclei most easily polarized by the protons on a given TPA ion will not be the same as those most easily dephased by the deuteriums on the same ion, although a net gain is expected and is indeed observed. All samples were studied using both single-pulse excitation and also 1H/29Si CP. Proton decoupling was applied during acquisition in all experiments. Figure 14 shows the deuterium MAS NMR spectra of the reference MFI material, the final (80-nm) crystals from the synthesis, and the first sample obtained by centrifugation after 40 h of reaction time. All the spectra show evidence of molecular motions of the propyl groups, which are more extensive in the case of the earliest sample, in agreement with the 14N spectra shown in Figure 12. Figure 15 shows the results of CP REDOR experiments with dephasing by deuterium on the reference F-MFI sample prepared with the partially deuter-

86 J. Phys. Chem. C, Vol. 112, No. 1, 2008

Fyfe et al.

Figure 16. 29Si{1H} CP REDOR experiments on the F-MFI sample with 2D dephasing for 32 rotor periods (left) and 48 rotor periods (right). Reference spectra shown in blue; dephased spectra shown in red. Spectra were collected with a spinning rate of 8.5 kHz, with an initial 29Si{1H} CP contact time of 7.5 ms (1344 scans, 4-s recycle time, 5-Hz line broadening).

Figure 17. 29Si{1H} CP REDOR experiments with 2D dephasing for 32 rotor periods (RIGHT) and 256 rotor periods (left) on the 80-nm nano crystals (200 h reaction at 90 °C. Reference spectra shown in blue; dephased spectra shown in red. Spectra were collected with a spinning rate of 8.5 kHz, with an initial 29Si{1H} CP contact time of 7.5 ms (1024 scans, 4-s recycle time, 5-Hz line broadening).

ated template. The 1H/29Si cross polarization was quite efficient; there was a relatively slow decay of the reference spectrum, and significant dephasing by deuterium was observed. Although previous work51,52 indicated that it was often best to use a composite 180° pulse for inversion of deuterium magnetization, it was found that it was possible to get good inversion of the deuterium resonance for these samples by a simple 180° pulse, and this was used in the middle of the REDOR sequence, with all other pulses applied to the silicon nuclei. Null experiments, where the match condition was deliberately misset, were carried out to check that the observed signal reduction was not due to any spurious effects of the additional circuitry for the triple tuning. Such effects are also ruled out by the complete lack of dephasing seen for some samples. Figure 16 shows the reference and dephased spectra for the F-MFI sample for two different numbers of rotor cycles at a spinning rate of 8.5 kHz. The complete curves showed the expected behavior and dephasings of approximately 20 and 40%, after 32 and 48 rotor cycles, respectively, were observed from the difference curves, and these were used as benchmarks establishing the behavior of “perfect” MFI to investigate the other samples. Thus, it is expected that substantial dephasing should be observed for proximities, distances, motions, and interactions similar to those present in the MFI structure. Figure 17 shows the spectra for the final nanocrystals where similar dephasings of 20% and 50% were observed after somewhat larger numbers of rotor cycles (64 and 256). The difference may be due to somewhat increased template motions in the less perfect structure of the final crystalline material. Certainly, some dephasing should be observed for any local TPA environment that is at all like that in MFI. In clear contrast, Figure 18 shows the corresponding spectra for the first isolated product (in this case at 36 h), where, after 256 rotor cycles, no dephasing at all is observed, indicating minimal interactions and contact between the TPA cations and

Figure 18. 29Si{1H} CP REDOR experiment with 2D dephasing for 256 rotor periods on the first isolated sample (30 h reaction at 90 °C). Reference spectrum shown in blue; dephased spectrum shown in red. Spectra were collected with a spinning rate of 8.5 kHz, with an initial 29 Si{1H} CP contact time of 3 ms (40960 scans, 4-s recycle time, 10Hz line broadening).

the silica present. Even after 512 rotor cycles, the longest dephasing time that could be used due to signal-to-noise constraints, no dephasing at all was observed. This could, at least in part, be due to the higher degree of motion of the ions which will reduce the heteronuclear dipolar interactions, but in turn, the motions themselves must be due to the ion being surrounded by substantial free volume rather than a constrained local environment like that of MFI. It cannot be argued that the 50-nm particles present at this stage of the reaction consist of nanoblocks that are somehow mobile, as the 14N data (Figures 12 and 13) show that the template ions are present at a concentration which is too low by more than an order of magnitude than that needed for the nanoblock structures in Figure 2. We thus conclude that at the earliest time at which solid material can be obtained by centrifugation, where the particles are approximately 50 nm in diameter, there is no evidence for

Solid State NMR Investigation specific interactions between any TPA cations and the silica species present, which appears to be simple silica gel. Those cations present may simply be entrained in the gel with no specific role at all or at most as charge-balancing species. A number of solid-state NMR experiments have been reported in a relatively recent publication,24 and these results should be considered in light of the present data. There are no complementary results reported in the study from particle size measurements, and the reaction temperatures are both higher and lower than that used in the present work, so it is difficult to ascertain the reaction progress at different times, and it is not definite that the samples studied are absolutely identical to those reported on here. However, some comparisons can be made. The synthesis mixture was kept at room temperature and then heated at 120 °C for 24 h to produce the final crystals, which were isolated by centrifugation. Two other samples were taken, one immediately after mixing by freeze drying at liquid nitrogen temperatures, thought to contain the 33-36 T-site precursor intermediate containing template and the second after 24 h at room temperature, assumed to contain the 1 × 4 × 4 nm3 template-containing nanoblocks. The nature of the final product is not in doubt, and the closest of the other two samples to those studied in the present work is that isolated after 24 h at room temperature (called sample 2). The data from this sample could be compared fairly to those of the first sample we were able to isolate after 30 or 40 h at 90 °C, since we conclude that our material is still amorphous at this stage, while sample 2 is said to consist of the nanoblocks that are highly ordered with the template in the MFI environment even at this shorter reaction time. Unfortunately, most experiments were done on samples 1 and 3. 1H/29Si REDOR experiments were carried out on all three samples, and dephasing is observed. However, these are similar in some ways to the CP experiments previously reported; they will yield similar information and, importantly, will have similar problems if the magnetization source (CP) or source of dephasing (REDOR) cannot be unambiguously identified. As indicated above, we feel that there may well be back exchange of the deuterated hydroxyl groups from various proton sources and that this can compromise the assumption that the TPA alkyl protons are the sole magnetization source and for this reason chose a dephasing experiment based on the deuterium nuclei of deuterated TPA. Indeed, the authors indicate this has happened at various points in the text from the H/C ratios from NMR “indicating the presence of inorganic HDO” and when discussing the proton spectra that “not all silanol groups have been replaced by deuterons or some back exchange has occurred on exposure of the sample to air”. Interestingly, a very comparable experiment was reported but not presented or described in the Experimental Section: This was a 29Si REDOR experiment with dephasing by 13C. It is not clear whether proton decoupling or CP was used, but “significant effects” are reported for the final product, but they were “unable to record significant effects for sample 1”. These latter data would seem to be in agreement with our findings although the authors report a fast decay of the echo signal limited their measurements. Summary and Conclusions All of the NMR experiments carried out in the present work are consistent with a mechanism for the “clear solution” synthesis of MFI that involves the hydrolysis of TEOS under basic conditions to form amorphous silica particles of ca. 80 nm in size. Heating at 90 °C causes the gradual appearance of fully crystalline MFI initiated between 40 and 60 h, corresponding to the reaction solution becoming cloudy and the appearance of Bragg reflections.

J. Phys. Chem. C, Vol. 112, No. 1, 2008 87 Whether the amorphous particles themselves crystallize or whether they act only as a nutrient source for the crystallization cannot be deduced from the data, but the final crystals have similar dimensions. Although there is no information on the oligomeric species formed early in the reaction at room temperature, there is no evidence that the first samples isolated after 30-40 h at 90 °C contain TPA ions in similar environments to that occupied by the template ion in the MFI framework. There is therefore no necessity to postulate the intervention of the “nanoblocks” or their condensation products in the formation of the final crystalline product as proposed by Martens and co-workers. The most informative NMR experiments in terms of detecting the presence of the MFI framework appear to be 13C, 14N, 2D, and the 29Si observe CP REDOR with dephasing by 2D. When combined with particle size measurements and powder XRD techniques, they provide self-consistent description of this crystalline material. There are a number of other MFI syntheses including the dense gel53 and dry gel54 syntheses that should also be amenable to study by the general approach described in the present work. Direct extensions of the present work may be made to other porous framework systems and, in fact, to the investigation of “nano” precursors in materials syntheses in general, providing information on structures not easily accessible by diffraction techniques. Acknowledgment. The authors would like to thank the Natural Sciences and Engineering Research Council (NSERC) of Canada for the award of Discovery and Major Equipment grants to C.A.F. and the Alexander von Humboldt Foundation for the award of a Feodor Lynen Fellowship to Dr. F. Scheffler. References and Notes (1) Breck, D. W. Zeolite Molecular SieVes; Academic Press: London, 1978. (2) Barrer, R. M. Zeolites and Clay Minerals as Sorbents and Molecular SieVes; Wiley: New York, 1978. (3) Robson, H. Verified Syntheses of Zeolitic Materials, 2nd ed.; Elsevier, Amsterdam, 2001. (4) Mintova, S.; Olson, N. H.; Valtchev, V.; Bein, T. Science 1999, 283, 958-960. (5) Yang, S.; Li, Q.; Wang, M.; Navrotsky, A. Microporous Mesoporous Mater. 2006, 87 (3), 261-267. (6) Majano, G.; Mintova, S.; Ovsitser, O.; Mihailova, B.; Bein, T. Microporous Mesoporous Mater. 2005, 80 (1-3), 227-235. (7) Yang, S.; Navrotsky, A.; Wesolowski, D. J.; Pople, J. A. Chem. Mater. 2004, 16 (2), 210-219. (8) Schoeman, B. J. Zeolites 1997, 18, 97. (9) Watson, J. N.; Iton, L. E.; Keir, R. I.; Thomas, J. C.; Dowling, T. L.; White, J. W. J. Phys. Chem. B 1997, 101, 10094. (10) Regev, O.; Cohen, Y.; Kehat, E.; Talmon, Y. Zeolites 1994, 14, 314. (11) de Moor, P.-P. E. A.; Beelen, T. P. M.; van Santen, R. A. Microporous Mater. 1997, 9, 117. (12) Dokter, W. H.; van Garderen, H. F.; Beelen, T. P. M.; van Santen, R. A.; Bras, W. Angew. Chem. 1995, 34, 73. (13) Beelen, T. P. M.; Dokter, W. H.; van Garderen, H. F.; van Santen, R. A. In Syntheses of Porous Materials, Zeolites, Clays and Nanostructures; Occelli, M. L., Kessler, H., Eds.; Dekker Inc.: New York, 1997; p 59. (14) Schoeman, B. J. Stud. Surf. Sci. Catal. 1997, 105, 647. (15) Ravishanker, R.; Kirschhock; Schoeman, B. J.; Vanoppen, P.; Grobet, P. J.; Storck, S.; Maier, W. F.; Martens, J. A.; De Schryver, F. C.; Jacobs, P. A. J. Phys. Chem. B 1998, 102, 2633. (16) Ravishanker, R.; Kirschhock; Schoeman, B. J.; Vos, D. D.; Grobet, P. J.; Jacobs, P. A.; Martens, J. A. Proceedings of the 12th International Zeolite Conference; Treacy, M. M. J., Marcus, B. K., Bisher, M. E., Higgins, J. B., Eds.; Materials Research Society: Pennsylvania, 1999; Vol. III., p 1825. (17) Ravishankar, R.; Kirschhock, C. E. A.; Knops-Gerrits, P.-P.; Feijen, E. J. P.; Grobet, P. J.; Vanoppen, P.; De Schryver, F. C.; Miehe, G.; Fuess, H.; Schoeman, B. J.; Jacobs, P. A.; Martens, J. A. J. Phys. Chem. B 1999, 103, 4960.

88 J. Phys. Chem. C, Vol. 112, No. 1, 2008 (18) Kirschhock, C. E. A.; Ravishankar, R.; Van Loovren, L.; Jacobs, P. A.; Martens, J. A. J. Phys. Chem. B 1999, 103, 4972. (19) Kirschhock, C. E. A.; Ravishankar, R.; Jacobs, P. A.; Martens, J. A. J. Phys. Chem. B 1999, 103, 11021. (20) Kirschhock, C. E. A.; Buschmann, V.; Kremer, S.; Ravishankar, R.; Houssin, C. J. Y.; Mojet, B. L.; van Santen, R. M.; Grobet, P. J.; Jacobs, P. A.; Martens, J. A. Angew. Chem., Int. Ed. 2001, 40, 2637. (21) Kirschhock, C. E. A.; Kremer, S.; Grobet, P. J.; Jacobs, P. A.; Martens, J. A. J. Phys. Chem. B 2002, 106, 4897. (22) Kirschhock, C. E. A.; Ravishankar, R.; Vespeurt, F.; Grobet, P. J.; Jacobs, P. A.; Martens, J. A. J. Phys. Chem. B 2002, 106, 3333. (23) Houssin, C. J. Y.; Kirschhock, C. E. A.; Magusin, P. C. M. M.; Mojet, B. L.; Grobet, P. J.; Jacobs, P. A.; Martens, J. A.; van Santen, R. M. Phys. Chem. Chem. Phys. 2003, 5, 3518. (24) Magusin, P. C. M. M.; Zorin, V. E.; Aerts, A.; Houssin, C. J. Y.; Yakovlev, A. Y.; Kirschhock, C. E. A.; Martens, J. A.; van Santen, R. M. J. Phys. Chem. B 2005, 109, 22767. (25) Kragten, D. D.; Fedeyko, K. M.; Sawat, K. R.; Rimer, J. D.; Vlachos, D. G.; Lobo, R. F.; Tsapatsis, M. J. Phys. Chem. B 2003, 107, 10006. (26) Ramanan, H.; Kokkoli, E.; Tsapastis, M. Angew. Chem., Int. Ed. 2004, 43, 4558. (27) Fedeyko, J. M.; Rimer, J. D.; Lobo, R. F.; Vlachos, D. G. J. Phys. Chem. B 2004, 108, 12271. (28) Rimer, J. D.; Kragten, D. D.; Tsapatsis, M.; Jeong, H.-K.; Lee, Y.; Vlachos, D. G. Stud. Surf. Sci. Catal. 2004, 154A, 317. (29) Fedeyko, J. M.; Vlachos, D. G.; Lobo, R. F. Langmuir 2005, 21, 5197. (30) Rimer, J. D.; Vlachos, D. G.; Lobo, R. F. J. Phys. Chem. B 2005, 109, 12762. (31) Rimer, J. D.; Lobo, R. F.; Vlachos, D. G. Langmuir 2005, 21, 8960. (32) Cheng, C.-H.; Shantz, D. F. J. Phys. Chem. B 2006, 110, 313. (33) Provis, J. L.; Vlachos, D. G. J. Phys. Chem. B 2006, 110, 3098. (34) Knight, C. T. G.; Kinrade, S. D. J. Phys. Chem. B 2002, 106, 3329. (35) Knight, C. T. G.; Wang, J.; Kinrade, S. D. Phys. Chem. Chem. Phys. 2006, 8, 3099. (36) (a) Olson, D. H.; Kokotailo, G. T.; Lawton, S. K.; Olson, D.; Meier, W. J. Phys. Chem. 1981, 85, 2238. (b) van Koningsveld, H.; Jansen, J. C.;

Fyfe et al. van Bekkum, H. Acta. Cryst. 1987, B43, 127. (c) van Koningsveld, H.; Jansen, J. C.; van Bekkum, H. Zeolites 1990, 10, 235. (37) (a) Fyfe, C. A.; Kennedy, G. J.; De Schutter, C. T.; Kokotailo, G. T. J. Chem. Soc., Chem. Comm. 1984, 541. (b) West, G. W. Aust. J. Chem. 1984, 37, 455. (c) Fyfe, C. A.; Stobl, H.; Kokotailo, G.; Kennedy, G. J.; Barlow, G. E. J. Am. Chem. Soc. 1988, 110, 3373. (38) Diaz, A., Ph.D. Dissertation, The University of British Columbia, Vancouver, Canada, 1998. (39) Fyfe, C. A.; Diaz, A. C.; Grondey, H.; Lewis, A. R.; Forster, H. J. Am. Chem. Soc. 2005, 127 (20), 7543. (40) Brouwer, D. H., Ph.D. Dissertation, The University of British Columbia, Vancouver, Canada, 2003. (41) Fyfe, C. A.; Brouwer, D. H. J. Am. Chem. Soc. 2006, 128, 11860. (42) Fyfe, C. A.; Brouwer, D. H. Can. J. Chem. 2006, 84 (2), 345. (43) Reischman, P. T.; Schmitt, K. D.; Olson, D. H. J. Phys. Chem. 1988, 92, 5156. (44) Nagy, J. B.; Gabelica, Z.; Derouane, E. G. Chem. Lett. 1982, 1105. (45) Lewis, A., Ph.D. Dissertation, The University of British Columbia, 1998. (46) Aguado, J.; Serrano, D. P.; Escola, J. M.; Rodriguez, J. M. Microporous Mesoporous Mater. 2004, 75, 41. (47) (a) Maciel, G. E.; Sindorf, D. W. J. Am. Chem. Soc. 1980, 102, 7606. (b) Klur, I.; Jacquinot, J.-F.; Brunet, F.; Charpentier, T.; Virlet, J.; Schneider, C.; Tekely, P. J. Phys. Chem. B 2000, 104, 10162. (48) Do, T. O.; Nossov, A.; Springuel-Huet, M. A.; Schneider, C.; Bretherton, J. L.; Fyfe, C. A.; Kaliaguine, S. J. Am. Chem. Soc. 2004, 126, 14324. (49) (a) Schaefer, J.; McKay, R. A.; Stejskal, E. O. J. Magn. Reson. 1979, 34, 443. (b) Stejskal, E. O.; Schaefer, J.; McKay, R. A. J. Magn. Reson. 1984, 57, 471. (50) Gullion, T.; Schaefer J. J. Magn. Reson. 1989, 81, 196 (51) Gullion, T.; Vega, A. J. Prog. Nucl. Magn. Reson. Spectrosc. 2005, 47 (3-4), 123. (52) Gullion, T. J. Magn. Reson. 2000, 146 (1), 220. (53) Burkett, S. L.; Davis, M. E. J. Phys. Chem. B 1994, 98, 4647. (54) Xu, W.; Dong, J.; Li, J.; Li, J.; Wu, F. Chem. Commun. 1990, 755.