Boron Sites in Borosilicate Zeolites at Various ... - ACS Publications

J. Phys. Chem. B 2004, 108, 18535-18546

18535

Boron Sites in Borosilicate Zeolites at Various Stages of Hydration Studied by Solid State NMR Spectroscopy Son-Jong Hwang,*,† Cong-Yan Chen,‡ and Stacey I. Zones‡ DiVision of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, California 91125, and CheVronTexaco Energy Technology Company, Richmond, California 94802 ReceiVed: May 27, 2004; In Final Form: August 9, 2004

The local structures of framework boron atoms in borosilicate zeolites B-β, B-SSZ-33 and B-SSZ-42 have been studied in the course of hydration/dehydration by employing solid-state NMR methods. In particular, characterization of trigonal boron sites has been studied in great detail. 11B MAS NMR spectra showed that boron trigonally coordinated to the framework (B(OSi)3, denoted as B[3]) can be readily transformed to a defective trigonal boron site (B(OSi)2(OH), denoted as B[3]-I) as a result of hydration. The presence of B[3]-I sites was proven by utilizing a number of different NMR methods including 11B MAS NMR at two different fields (11.7 and 19.6 T), 11B MQMAS, 11B CPMAS, and 11B 2D HETCOR experiments. The B[3]-I species can be converted into B[3] upon dehydroxylation, but its presence can also be sustained even after very high-temperature treatment (at least up to 500 °C). The formation of deboronated species, B(OH)3, in distorted form was detected even under a mild hydration treatment. HETCOR NMR revealed that hydroxyl protons with chemical shifts at 2.4 and 3.3 ppm in 1H NMR are correlated with B[3] and B[3]-I sites, respectively. The presence of a new hydroxyl proton at 3.8 ppm in 1H NMR that showed selective correlation with B[3]-I in HETCOR NMR was also identified.

I. Introduction

SCHEME 1

In recent years, much attention has been given to the syntheses and characterization of borosilicate zeolites because their weaker acidity is suitable for certain catalytic reactions that require mild solid acids as catalysts.1-3 The borosilicate zeolites can also provide a unique post-synthetic route for preparing aluminosilicate or other isomorphous forms of zeolites at certain Si/M ratios (M represents trivalent metal ion such as Al and Ga) where the trivalent metal ions cannot be incorporated directly into the framework in synthesis.4 In the past two decades the change of coordination geometry at the boron centers (tetrahedral (B[4]) T trigonal (B[3])) upon hydration/dehydration treatments of borosilicate zeolites, especially for the proton form where the H+ becomes the charge compensating cation, has been studied in detail both experimentally5-9 and computationally.10-11 Such a transformation of boron coordination and subsequent variation of the local geometry were good rationales in explaining the weaker acidic borosilicate zeolites compared with the corresponding Al substituted sites. Experimentally, use of high resolution 11B solid-state NMR techniques has been most efficient in investigating changes of structures around boron sites by taking advantage of the element’s electric quadrupole moment (I ) 3/2).12 The B(OSi-)3 types of trigonal geometry render the quadrupolar coupling constant (Cqcc) as high as 2.6 MHz, which is large enough to yield second order quadrupole broadening even under a fast magic angle spinning (MAS) condition. The resulting anisotropic line shape is readily distinguished from that of tetrahedrally coordinated boron sites, where the Cqcc is measured to be an order of magnitude lower. * Corresponding author. Fax: 1-626-567-8743. E-mail: sonjong@ cheme.caltech.edu. † California Institute of Technology. ‡ ChevronTexaco Energy Technology Co.

The trigonal boron sites B[3] formed via dehydration from the conversion of tetrahedral boron sites B[4] of the proton form of borosilicate zeolites remain tightly anchored with the framework through three B-O-Si bonds in dehydrated environments. However, the B[3] sites are very susceptible to nucleophilic attack by water molecules when the materials become hydrated (see Scheme 1).9 This attack leads to the creation of B-OH bonds as found in B[4]-I species. Note that the negative charge on the boron centers in the B[4]-I species is balanced by coordination of OH2+. Successive hydrolysis steps will eventually replace all the B-OSi bonds with B-OH bonds, resulting in release of boron atoms from the framework as B(OH)4- species and leaving a silanol nest (four SiOH groups) in the framework. The deboronation reaction takes place quickly even at ambient temperature when a dehydrated material becomes rehydrated. For example, we found that the formation of B(OH)4- species can be readily observed in a 11B MAS NMR spectrum that is obtained after a drop of water is added to a dehydrated borosilicate zeolite powder containing trigonal boron sites. The vulnerability of B-OSi bonds to moisture is wellknown. The boron content was reported to be reduced up to 40% just by washing with the distilled water.9 To avoid the deboronation, ion exchange of the proton of borosilicate zeolites with alkali metal cations or NH4+ was employed because these bulkier cations were proved to protect B[4] sites from the coordination transformation even under dehydration conditions.8,9 A calcination process that uses ammonia gas was also

10.1021/jp0476904 CCC: $27.50 © 2004 American Chemical Society Published on Web 11/09/2004

18536 J. Phys. Chem. B, Vol. 108, No. 48, 2004 SCHEME 2

introduced to generate template free zeolites having all boron sites in tetrahedral coordination.9 In any cases, once the catalytically active H-form of borosilicate zeolites is prepared, the framework boron sites will be present in the B[3] form upon dehydration. These B[3] sites are susceptible to the progressive deboronation during a normal laboratory storage if great care is not taken. In the process of hydrolysis, a number of different types of boron species are expected to be formed as intermediates before the framework boron atoms are freed from B-OSi linkages. They include both trigonal and tetrahedral boron species, depending on the number of coordinating OH groups as sketched in the structural moieties shown in Scheme 2.7,9 Currently, much of the spectroscopic details of the species in Scheme 2 remain unresolved. As will be discussed next in this report, a number of high-resolution NMR methodologies were employed to elucidate the structural changes around boron atom centers during hydration/dehydration. First, the formation and transition of intermediates were investigated at different stages of the hydration/dehydration processes by acquiring 11B MAS NMR spectra after single pulse (single quantum) excitation of the central transition (-1/2T 1/2). Often, the second-order quadrupole interactions of trigonal species (B[3], B[3]-I, B[3]-II) give rise to a crowded 1D spectrum, especially when trigonal species coexist and render overlap of characteristic quadrupole powder patterns. 11B MAS NMR spectra obtained at a higher magnetic field (19.6 T) were used to unravel the problem. Second, a 11B 2D multiple quantum MAS (MQMAS)13-15 method was employed to obtain highly resolved isotropic line components free of quadrupole interaction. Elimination of the second order broadening effect in this technique is based on proper correlation of the radio frequency (rf) driven symmetric multiple quantum excitation with the single quantum coherence under the MAS condition. As originally demonstrated by Frydman et al.,13 the MQMAS NMR technique has been proven to be extremely useful in examination of the half-integer quadrupole nuclei and applied extensively to study amorphous and crystalline inorganic solid materials. A recent and thorough review on both theoretical and experimental aspects was reported by Amoureux and Pruski.15 Third, a systematic approach was attempted to correlate 11B and 1H NMR signals in this work. For this purpose, 1H-11B cross-polarization (CP) based heteronuclear correlation spectroscopy16 (HETCOR) was utilized. Use of high MAS rate (>10 kHz) as well as low rf power was essential for this technique in order to obtain efficient 1H-11B cross-polarization and 1H spectral resolution that is high enough to show individual correlation with 11B MAS powder patterns in the twodimensional contour display. Fild and Koller8,17 have reported that internuclear distances between 1H and 11B in a calcined and dehydrated B-β might be estimated using the rotational echo double resonance (REDOR) NMR method. Information about which hydroxyl groups are closely related to trigonal boron centers can then be obtained via REDOR method. Both HETCOR and REDOR techniques are used to probe heteronuclear dipole interactions under MAS conditions. Although HETCOR NMR spectroscopy lacks in precise examination of

Hwang et al. interatomic separations, it was proven to be more effective in making selective associations between hydroxyl groups and boron sites, especially when a number of different boron species were populated. Such an individual correlation allowed us to assign the hydroxyl group that is directly associated with the B[3] center, as a possible framework Brønsted acid center. Here we report our studies on the following three borosilicate zeolites at various stages of hydration using the NMR techniques discussed above: B-β (BEA*), B-SSZ-33 (CON), and B-SSZ-42 (IFR). The B-β sample was most heavily studied at each different dehydration level. II. Experimental Section 1. Material Synthesis. The borosilicate zeolites B-SSZ-33, B-SSZ-42, and B-β were synthesized by using the appropriate structure directing agents (SDAs) in hydroxide form, as outlined in our previous publications.18-21 Cab-o-sil M-5, sodium borate decahydrate, and NaOH were used as silicon, boron, and alkali metal sources, respectively. The compositions of these zeolites were determined by Galbraith Laboratories (Knoxville, TN) via elemental analyses using ICP methods. The molar Si/B ratios of these zeolites are 17.0 (B-SSZ-33), 25.9 (B-SSZ-42), and 15.4 (B-β). 2. Calcination. The as-made borosilicate zeolites were calcined in thin beds in a muffle furnace to remove the SDAs occluded in zeolite channels. The temperature program was as follows: 1 °C/min to 120 °C, hold for 2 h, 1 °C/min to 540 °C, hold for 4 h, 1 °C/min to 600 °C, and hold for 4 h. The calcinations were carried out in a steady flow of nitrogen containing just a slight bleed of air over the bed of zeolite. 3. Dehydration. Calcined materials were hydrated in several different manners after the calcinations, which will be noted individually. The calcined and hydrated borosilicate zeolites were first packed in a 4 mm ZrO2 NMR rotor, and the whole ZrO2 rotor was placed inside a 15 cm long 5 mm glass NMR tube. The glass NMR tube was then attached to a vacuum line and evacuated to 10-3 Torr while heated at 1 °C/min up to a target temperature (e.g., 120 °C) and then held at the temperature for 4 h. When the dehydration temperature was set to be higher than 120 °C, the sample was first held at 120 °C for 2 h in order to remove most of water from the framework and then further heated to the target temperature at 1 °C/min. Cooling to room temperature was allowed to occur naturally inside the heating furnace while the evacuation was maintained. Dry N2 gas was introduced to the sample before the rotor was exposed to air, and the rotor was closed immediately with the tight sealing kel-F cap. Such care was found to be efficient in avoiding introduction of moisture. 4. NMR Analysis. Most of the solid-state NMR spectra were collected using a Bruker DSX-500 spectrometer operating at 11.7 T with 500.2, 160.5, and 99.4 MHz for 1H, 11B, and 29Si nuclei, respectively, and using a Bruker 4 mm CPMAS probe. A typical spinning rate was 12 kHz and MAS spectra were recorded after applying 4 µs-π/2 pulse for 1H and 29Si and 1 µs-π/8 pulse for 11B nucleus. The chemical shifts were referenced to TMS and BF3‚O(CH2CH3)2. 11B MAS NMR spectra at 19.6 T with 11B resonance frequency of 266.1 MHz were obtained at the National High Magnetic Field Laboratory in Tallahassee, FL, without 1H decoupling and at a 10 kHz sampling spinning rate. The 11B 2D MQMAS experiments were performed using the standard z-filtered sequence22 at 13 kHz sample spinning. For the 1H-11B CPMAS and 2D HETCOR experiments, the spinlocking during cross-polarization transfer was maintained in the

Boron Sites in Borosilicate Zeolites

J. Phys. Chem. B, Vol. 108, No. 48, 2004 18537

Figure 1. 1H and 11B MAS NMR spectra at 11.7 T of calcined B-β at various levels of hydration: (a) sample A, freshly calcined and then vacuum-dried at 500 °C; (b) sample B, freshly calcined sample and then stored for one year in closed vial, no dehydration treatment; (c) sample C, sample after vacuum-drying of sample B at room temperature; (d) sample D, sample after vacuum-drying of sample C at 120 °C; (e) sample E, sample after vacuum-drying of sample D at 450 °C.

sudden-passage regime by applying very low rf pulse (1.3 kHz) for boron.23-25 The adiabaticity parameter R ()υ21s/υQυr) was estimated to be less than 0.0001 for trigonally coordinated boron atoms. The sample was spun at 10 kHz. III. Results and Discussion 1. Dehydration of Boron-β. Figure 1 shows 1H and 11B MAS NMR spectra of calcined B-β zeolite at various levels of hydration. Note that all spectra were obtained with the same number of transients and normalized to the sample weights used in the NMR measurements. This allows signals to be directly compared and the relative changes among the steps of dehydration to be measured. Both bottom spectra (see Figure 1a) from sample A, which was a freshly calcined and dehydrated sample, can serve as reference spectra in future comparisons. These spectra represent the local structure of boron atoms in well dehydrated B-β framework, as represented by B[3] shown in Scheme 1. The 1H signal at 2 ppm is assigned to the silanol group (Si-OH) while the 11B signal represents the trigonal boron sites coordinated to three OSi groups of the zeolite framework. Such assignments are in good agreement with previous reports.5-8 The 11B spectrum shows a MAS NMR line shape (10 ppm ∼ -2 ppm) of the typical powder pattern that originates from the second order quadrupole broadening of trigonally coordinated 11B nuclei in a planar configuration. Detailed NMR parameters for the quadrupole interaction will be discussed below. A small and narrow peak around at -4 ppm represents tetrahedral boron groups B[4] (see Scheme 1). The presence of B[4] groups can be attributed to Na+ ions that exist in small quantity in the sample as confirmed by 23Na MAS NMR and elemental analysis (not shown). The results also indicate that the B-β sample studied here is dominated by the H-form. Na+ ion is known to stabilize B[4] from coordination conversion upon dehydration.8,9 Further investigation is in progress to confirm the selective association of Na+ ion with the B[4] unit. Figure 1b depicts the spectrum collected from B-β sample B, which was a freshly calcined material and then hydrated while stored in a closed vial for 1 year. A stepwise dehydration was performed on the same sample, resulting in samples C-E (Figure 1, parts c-e, respectively). Rather drastically hydrated samples were also studied and will be discussed later. Moisture uptake by sample B is revealed by the presence of water peak (∼5 ppm) in the 1H NMR shown in Figure 1b. From the signal

intensity of 1H NMR, the water content was measured to be about 10 wt %. Concomitantly, structural modification of the boron sites in sample B is well displayed by the 11B NMR spectrum (see Figure 1b) upon hydration. Appearance of another broad line shape ranging from 10 to 17 ppm in 11B MAS NMR spectrum is eminent while the powder pattern of B[3] sites shows reduced intensity compared to that of Figure 1a of sample A. The broad line shape downfield (10-17 ppm) was also reported in the literature8,17 and was interpreted as a formation of nonframework trigonal boron species which bear no B-OSi links, such as B(OH)3. This implies that a mild hydration of calcined borosilicate zeolite would induce the deboronation, as also discussed in the Introduction section. Further investigation was made to confirm the assignment of B(OH)3 to the broad line shape by employing high-resolution NMR techniques and will be discussed below. It is also worthwhile to note that the increase of the -4 ppm peak indicates some conversion of B[3] to tetrahedral boron sites B[4] or B[4]-I upon hydration (see Scheme 1). The observation has been better analyzed by spectral simulation and 2D MQMAS method (see below). The spectra of B-β sample C resulting from the first step of dehydration, i.e., evacuation at room temperature, are shown in Figure 1c. 1H NMR shows that most of the mobile water molecules were readily removed and the silanol groups (2.0 ppm) become partly resolved. The broad line shape at 4.5 ppm represents hydrogen bonded H2O groups. The shape and intensity of the -4 ppm peak in the 11B MAS NMR spectrum of sample C (Figure 1c) appeared to be nearly the same as that of sample A (freshly calcined sample dehydrated in a vacuum at 500 °C, see Figure 1a). Partial recovery of the B[3] peak is noticeable in sample C, indicating that some tetrahedral boron sites which were formed in sample B upon hydration (Figure 1b) underwent coordination conversion back to B[3] sites even via such a mild dehydration treatment (also see Scheme 1). A rather striking observation at this point is the reduction of a broad peak (around 16 ppm) that was originally considered to be a part of powder pattern of nonframework boron species (see below). Such change became more eminent as further dehydration took place at higher temperatures. Spectra of sample D (dehydrated at 120 °C) in Figure 1d show the drastic decrease of hydrogen bonded water and complete disappearance of the 16 ppm peak in 11B NMR. Note that the broad resonance around 3 ppm shown in the 1H NMR spectrum of sample D in Figure 1d was noticeably reduced when the dehydration temperature

18538 J. Phys. Chem. B, Vol. 108, No. 48, 2004

Hwang et al.

Figure 2. Experimental 11B MAS NMR spectra of calcined B-β at 11.7 and 19.6 T along with simulated line shapes. (a) sample B, hydrated while stored for 1 year in closed vial, no dehydration treatment (see Figure 1b); (b) sample D, vacuum-dried at 120 °C (see Figure 1d). Arrow 1: B[3]* sites resolved in 19.6 T spectra; Arrow 2: B(OH)4- type species formed due to additional hydration; Arrow 3: 16 ppm peak (see text for details). Note that both samples were exposed to air during the sample packing for the measurement at 830 MHz spectrometer.

was increased to 450 °C for sample E (Figure 1e). The change was accompanied by an eminent reduction of a broad resonance at 11 ppm in the 11B NMR spectra. Dehydration at 450 °C (sample E) resulted in NMR spectra (Figure 1e) that are similar to those of sample A which was a freshly calcined sample after calcination at 500 °C (Figure 1a), where no noticeable difference between the 1H NMR spectra but some minor distortion in 11B NMR line shape were observed. These results have clearly demonstrated the reversibility of the coordination transformation of boron sites upon hydration/dehydration process. It reveals that 11B MAS NMR is well suited to show the formation of intermediate species during the hydration/dehydration process. To make structural identification of these intermediates, a closer NMR investigation was performed as follows. 2. Trigonal Boron Sites Other Than B[3]: Finding of B[3]*. To analyze in greater detail the 11B NMR line shapes of samples B and D (see Figure 1, parts b and d), a number of different approaches were taken as discussed below. First, numerical simulation of the line shape acquired at 11.7 T for sample B in Figure 1b was performed by using QUASAR26 and by taking into account the possible presence of B(OH)3 type nonframework trigonal boron species (denoted as [B]-nf) as proposed by Fild et al.8 Inclusion of the line shape for B[3]-nf (Cqcc ) 2.15 MHz, η ) 0.20, δiso ) 18.0 ppm) in addition to the main trigonal site B[3] (Cqcc ) 2.56 MHz, η ) 0.14, δiso ) 10.1 ppm) resulted in a reasonable fit of the experimental spectrum as shown in the bottom part of Figure 2a. The resulting parameters obtained with an iterative fitting process are consistent with the previous report.8 Note that besides the trigonal boron sites, two different lines for tetrahedral boron sites, B[4] and B[4]*, were introduced to fit the peak at around -4 ppm region. The position of B[4] sites is assigned to -4 ppm (also see Figure 1a). The downfield peak (-3.5 ppm) is marked with B[4]* temporarily, and its identity will be discussed later. Second, 11B MAS NMR spectra of samples B and D that were used for Figure 1, parts b and d were acquired using a 830 MHz spectrometer (19.6 T) at the National High Magnetic Field Laboratory (NHMFL) in Florida. However, when spectra were obtained at the higher field (19.6 T), the validity of the decomposition of the low field (11.7 T) spectrum which is depicted in the bottom part of Figure 2a appeared to be uncertain. As evidently shown in Figure 2a, the 11B MAS NMR spectrum of the same sample at higher magnetic field (19.6 T)

unveiled the existence of more boron species which could not be characterized by the spectrum at 11.7 T alone. The remarkable enhancement in resolution at the higher magnetic field is mainly due to the fact that the second order quadrupole coupling is inversely proportional to the strength of operating magnetic field.27 Direct observation of such hidden resonances at 19.6 T became a crucial factor in determining the number and type of spectral line components. Third, the differences between the two spectra obtained at two different fields were investigated next. Because the quadrupole induced shift is dependent on the magnetic field strength, the position of peaks is expected to be varied at the higher field. The comparison here was presented by employing a numerical simulation of spectra at 19.6 T. Note that the simulated spectrum underneath of the spectrum at 19.6 T in Figure 2a is simply a projected spectrum produced by using the fitting parameters reported above. The experimental spectrum at 19.6 T appeared to be blurred because 1H decoupling was not possible for the measurement. The direct comparison of the experimental spectrum with the simulated line shape allowed us to extract the newly observed and unidentified peaks that are marked with arrows 1 and 2 in Figure 2a. The projected line position of B[3]nf species at 19.6 T is marked by a vertical broken line. Figure 2b shows the exact same array of experimental and simulated spectra that were collected from B-β sample D (after a vacuum dehydration at 120 °C). As discussed in the previous section, the part of line shape of B[3]-nf species (namely, the 16 ppm peak) was found to disappear in the spectrum of sample D at 11.7 T (see Figures 1d and 2b). In the spectra at 19.6 T, however, it is interesting to observe the presence of a peak centered at 16 ppm (marked with the broken line and arrow 3 at the same position in Figure 2). The intensity of this peak of sample D (Figure 2b) appeared to be markedly reduced from that of sample B in Figure 2a. For the dehydrated sample D, it was necessary to introduce an additional powder pattern that could explain the broad shoulder at 11 ppm shown in the spectrum at 11.7 T and the presence of a peak (arrow 1) in the spectrum at 19.6 T. The bottom half of Figure 2b shows the newly constructed line components after an iteration using QUASAR that resulted in a good fit for the spectrum at 11.7 T. The projected spectrum onto the 19.6 T field appeared to reproduce the experimental line shape very well. In particular, the position of arrow 1 is now well anticipated due to the newly introduced powder pattern. The NMR parameters for this new

Boron Sites in Borosilicate Zeolites powder pattern estimated from the line fitting are Cqcc ) 2.46 MHz, η ) 0.29, δiso ) 14.6 ppm, strongly indicating boron in a trigonal coordination should be responsible for the line component. Here B[3]* is used to distinguish this newly found site from the other trigonal boron sites B[3] or B[3]-nf. Note that NMR parameters found for B[3] in Figure 2b were identical with those of B[3] in Figure 2a within experimental error range. When examining the spectra in Figure 2a, the peak marked with arrow 1 in the spectrum at 19.6 T now indicates that the B[3]* species is also present in the hydrated sample B. This result strongly suggests that the interpretation of the spectrum at 11.7 T (Figure 2a) should include contributions from the powder patterns of both B[3]-nf and B[3]* groups as well as that of B[3]. Koller recently also used this approach to explain 11B MAS spectral line shapes of a calcined B-β at different dehydration temperatures.17 We found from the spectrum at 19.6 T in Figure 2a that the two humps at 16 and 12.5 ppm (not the isotropic chemical shifts), temporarily assigned to be B[3]-nf and B[3]*, respectively, show roughly the same contribution in the signal intensity. However, a reasonable fitting of the experimental spectrum at 11.7 T in Figure 2a could not be obtained by taking into account equal or similar contributions from both B[3]-nf and B[3]* components. As a result, we came to the conclusion that the spectrum can be properly interpreted by excluding the contribution of the particular powder pattern of B[3]-nf and finding out a way to explain the line shape in the region between 13 and 20 ppm: a Gaussian type peak with its center at around 16 ppm. A new label on this peak is given as P1 for the sake of distinction from the previous assignment. At this moment, we can associate the identity of P1 with a boron species that is formed in a relatively highly hydrated environment and is readily transformed to a trigonal boron site (most likely B[3]) even under mild dehydration treatment. From the Gaussian type line shape, it is proper to speculate that the coordination geometry of this boron site is deformed from the typical plane structure of trigonal borons and is closer to tetrahedral coordination. In addition, the chemical shift of P1 (∼16 ppm) strongly indicates that the peak might be closely related to those defective tetrahedral boron sites B[4]-II or B[4]III (see Scheme 2). Note that chemical shifts of these sites are expected to be between the two extremes of -4 and 19.2 ppm of the isotropic chemical shifts for B(OSi)4- and B(OH)4- type structures, respectively. To better characterize the signal around P1, resolution enhancement was carried out by employing the 2D MQMAS NMR method. Figure 3 shows 11B 2D MQMAS spectra at 11.7 T of samples B and D discussed in Figure 2. As marked with arrows, the assignment of bands in the 2D contour plots can be readily made by matching up components found in the analyses of 1D MAS NMR spectra (see Figure 2). In the resolution aspect, there is no additional information gained from the 2D method. However, like other typical 2D MQMAS spectra, the shape and position of a band reveal valuable information about the quadrupole interaction of the responsible structure. For example, trigonally coordinated boron sites B[3] and B[3]* display bands dispersed widely along the anisotropic axis in agreement with the quadrupole parameters reported above while the tetrahedral boron sites B[4] and B[4]* show rather circular bands with minor elongation along the chemical shift axis because of nominally weak quadrupole interactions (Cqcc ∼0.2 MHz). Note that separation of two tetrahedral boron sites is clearly revealed in the 2D MQMAS spectra, which is well consistent with the spectral simulation shown in Figure 2. The appearance of B[4]* signal is eminently associated with a mild


Figure 3. 11B 2D MQMAS spectra of calcined B-β at 11.7 T: (a) sample B, hydrated while stored for 1 year in closed vial (Figures 1b and 2a); (b) sample D, vacuum-dried at 120 °C (Figures 1d and 2b). Note that the projection onto the isotropic axis for Figure 3b is omitted because of its similarity to that of Figure 3a. The spinning sideband is marked with an asterisk.

hydration of calcined B-β, as pointed out in the previous discussion on Figures 1b and 2a. The formation of B[4]* cannot be unambiguously correlated with the appearance of the P1 or the B[3]* signals. The slight downfield shift of its position from the position of B[4] could provide a clue in speculating its possible structure. De Ruiter et al.28 pointed out that tetrahedral boron species such as B(OH)x(OSi)y (x + y ) 4) could be responsible for downfield peak shifting. The isotropic chemical shifts of those entities are unknown to our best knowledge. The postulated species include the B[4] associated with H+(H2O)n as the counterions or the B[4]-I structure which is formed by replacing one of the B-O-Si bonds in B(OSi)3 with OH (see Scheme 1). The B[4]* site is easily removed in sample C (see Figure 1c) after room-temperature evacuation as learned from the previous discussion on Figures 1 and 2. Whichever is the identity of tetrahedral boron sites characterized by -3.5 ppm in 11B NMR, its flexibility on the transformation between trigonal and tetrahedral coordination is found to be remarkable. For P1, its band shape is not unambiguously distinguished from that of B[3]* in Figure 3a for the hydrated sample B. However, the disappearance of P1 in the dehydrated sample D is evidently demonstrated in Figure 3b. Comparison of these two MQMAS spectra allows us to conclude that, unlike the other two trigonal sites, the P1 site has a rather circular contour shape without much dispersion along the anisotropic axis. It is, therefore, more likely that P1 is not associated with part of the powder pattern of a trigonal boron site but represents a tetrahedrally coordinated boron species as we initially speculated (see above). As a matter of fact, the identity of P1 was better characterized when the zeolite was further hydrated (see below). By now,our most of the boron species in the weakly hydrated B-β (sample B) represented by 11B MAS spectrum in Figure 2a have been identified, except for the signal marked with arrow 2 in 11B MAS spectrum at 19.6 T (Figure 2). The peak at ∼19.0 ppm can be assigned as a B(OH)4- type species because the isotropic chemical shift matches well with that of boric acid in aqueous solution (19.2 ppm). It is quite rational to expect the formation of B(OH)4- at the end of the hydration steps, provided that the hydration level is sufficiently high. Its presence was not observed in the spectrum at 11.7 T (Figure 2a), indicating that the hydration level of sample B might have changed before experiments at 19.6 T were performed. A similar mismatch has already been observed by detection of a peak at 16 ppm (arrow 3) in the spectrum at 19.6 T of the dehydrated sample while any indication of its presence does not exist in the spectrum at


Hwang et al.

Figure 4. 11B MAS NMR spectra of calcined B-β zeolite at 11.7 T: (a) sample F, sample B (see Figure 1b) hydrated by exposing to air at room temperature; (b) sample G, sample F further hydrated by soaking in water (55 mg of distilled water were added to 61 mg of sample F); (c) sample H, sample G vacuum-dried at 120 °C; (d) sample I, sample H vacuum-dried at 450 °C. (e) 2D MQMAS spectrum of sample F.

11.7 T (see Figure 2b). We speculate that these two samples packed in NMR rotors most likely have adsorbed some moisture before the NMR measurement at 19.6 T at NHMFL in Florida, which may explain the presence of the unexpected peaks present in both spectra at 19.6 T (see Figure 2, parts a and b). 1H NMR spectra of these two samples at 19.6 T are not available to show the degree of hydration. As presented in the following section, the formation of B(OH)4- has been observed at 11.7 T when the calcined B-β samples were further hydrated. Although the phenomena can be explained on the basis of the known behavior of zeolite samples, it should still be pointed out that, because of the aforementioned differences in the exact hydration level of the samples, the direct comparison of two spectra obtained at different magnetic fields might not be completely accurate. 3. Dehydration after Deboronation. When the calcined B-β zeolite was exposed to ambient air, rapid adsorption of moisture took place. A further hydration treatment by soaking the sample in water resulted in significant dissolution of boron atoms from the framework. B-β sample B presented in Figure 1b was employed in the following hydration/dehydration course with an aim of investigating the structural changes in boron sites after being subjected to such severe hydration conditions. 11B MAS NMR spectra collected in this course are presented in Figure 4. After being exposed to ambient atmosphere overnight, the water content of the resulting sample F rose to about 2.5 times higher than that of the parent sample B, as quantified from 1H MAS NMR spectra (not shown). The resulting changes on boron sites are clearly revealed in Figure 4a. First of all, the 11B MAS NMR powder patterns that are characteristic of trigonal boron sites, B[3] and B[3]*, have disappeared. The coordination conversion to B[4] sites appeared to be markedly enhanced while a broad peak at 16 ppm showed a strong growth at the same time. The identity of this 16 ppm peak was discussed in the previous section and it was assigned as P1 (see Figure 3a). Note that right beside the 16 ppm peak there is a broad line shape widely stretched up to 5 ppm on the 11B chemical shift axis, covering the region where NMR line shape of B[3]* used to reside. Acquisition of a 2D MQMAS spectrum allowed us to unveil the spectroscopic line shape of P1 more clearly (see Figure 4e), despite additional complexities. Unlike our initial speculation (see discussion on Figure 3 above), P1 is not characterized by a Gaussian type line shape but a powder pattern showing sizable dispersion along the anisotropic axis. Figure 4e also shows the emergence of an additional resonance, which

position in the isotropic dimension (δiso) is at ∼21 ppm. The projected spectrum onto the axis also indicates the presence of additional component. Its appearance explains the formation of boron sites in different structural form as a result of the advanced hydration. Estimation of their quadrupole coupling parameters could be readily made by applying the following equations.14,15

17 δiso ) δF2 + (δF1 - δF2) , 27

x

SOQE ) Cqcc

1+

η2 ) 8.246 × 10-3νoxδF1 - δiso 3

where δF1 and δF2 stand for the centers of gravity that can be read from the 2D contour plot, SOQE is the second-order quadrupole effect, and νo is the Larmor frequency. The isotropic chemical shifts (δiso) were determined to be 16.7 and 19.2 ppm, respectively, for these two sites. Both show similar SOQE values of approximately 1.8 MHz. Consequently, the interpretation on P1 needs to be reconsidered to include not only the defective B[4]-II or B[4]-III but also trigonal sites. In any case, the responsible structure should be associated with a severely distorted form, i.e., with high asymmetry parameter (η). A closer investigation on the identity is currently underway. It should be noted that the signal intensities over the two regions (B[4] vs P1+B(δiso∼21 ppm)) were measured to be 45:55 at this particular hydration level. When sample F was further hydrated with the water vapor over 1 M aqueous KCl solution in a closed vessel, the water content of the resulting sample rose only about 20% more than that of sample F, implying that sample F was already well hydrated under the ambient conditions. The P1 resonance shows narrowing in line width and downfield shift of its position by 0.8 ppm, indicating that its structural change moves toward the formation of the deboronated species, B(OH)4-. After adding distilled water to sample F, the 11B MAS NMR spectrum of the resulting hydrated sample G (Figure 4b) clearly shows the formation of B(OH)4- (19 ppm) species, if not in the form of B(OH)3 present in aqueous solution,29 implying dissolution of boron species from the zeolite framework. In addition to the complete disappearance of the P1 peak, the reduction of the B[4] peak was also observed in Figure 4b. About 20% of total boron atoms was measured to remain in the framework as indicated by B[4] signal.


Figure 5. 29Si MAS NMR spectra of calcined B-β zeolite: (a) sample A, sample freshly calcined and then vacuum-dried at 500 °C (see Figure 1a); (b) sample F, sample B hydrated by exposing to air at room temperature (see Figure 4a); (c) sample G, sample F further hydrated by soaking in water (55 mg of distilled water were added to 61 mg of sample F; see Figure 4b); (d) sample I, sample H further vacuumdried at 450 °C (Figure 4d).

Sample G then underwent the similar dehydration process as described in Figure 1. Two subsequent spectra shown in Figure 4, parts c and d, represent the structural transformation of boron species that are present in the zeolite after such deboronation/dehydration treatments. Without further spectral analysis, the 11B MAS NMR spectra strongly suggest that boron atoms were incorporated back into the framework, at least in the form of B(OSi)3, i.e., B[3]. The formation of B[3]* species was observed in significant amounts as shown in Figure 4c. The powder pattern of B[3] after evacuation at 450 °C appeared to be broadened and distorted compared to that of sample E in Figure 1e, which could be attributed to the formation of B[3]* species as an intermediate in large quantity. The 29Si MAS NMR spectrum of each sample in Figure 5 reveals, in turn, the changes in B-OSi coordination. The very freshly calcined material after dehydration at 500 °C, sample A, shows the majority of its silicon signal at -111 ppm with a very small portion at -102 ppm, representing Si-(OSi)4 (Q4) and the silanol group HOSi-(OSi)3 (Q3) associated with the B[3] units and/or isolated silanol groups, respectively.28 The formation of SiOH groups observed by 29Si MAS NMR can serve as a direct measure of the hydrolysis of B-OSi bonds caused by hydration. Figure 5b shows a substantial growth of the peak at -102 ppm, indicating that significant amount B-OSi bonds underwent the

J. Phys. Chem. B, Vol. 108, No. 48, 2004 18541 hydrolysis in sample F by being exposed to moisture from the ambient atmosphere. For sample G, which was prepared by further hydration via soaking in water, there is no significant change found in 29Si MAS spectrum except slight decreasing of the Q4/Q3 ratio when deboronation was accomplished by soaking (see Figure 5c). The change of the ratio was attributed to the additional dissolution of B[4] units in further hydrated sample G. This result strongly suggests that the boron species giving rise to the P1 resonance in 11B MAS NMR (Figure 4a) should bear no B-OSi linkage. The identity of the P1 peak can now be postulated to be extraframework boron species such as B(OH)3, but with close association with the silanol nest and/ or (H2O)n in such a way that a mild dehydration can regenerate B-OSi bonds. Figure 5d shows that, with sample I, the dehydration at 450 °C nearly completes the recondensation of B-OSi bonds, as compared to the spectrum of Figure 5a for sample A (the freshly calcined sample which was then vacuumdried at 500 °C), although minor modification seems to take place in the framework silicon atoms. This is in good agreement with distorted line shape of B[3] sites in 11B MAS NMR (Figure 4d). 4. Boron SSZ-33 and SSZ-42. The formation of B[3]* species in the process of dehydration was not limited only to B-β. Two other borosilicate zeolites, B-SSZ-33 and B-SSZ-42, were examined after calcination, hydration and dehydration at 120 °C. Figure 6 shows experimental and simulated 11B MAS NMR spectra of these two different materials at 11.7 T. All of the NMR parameters acquired from numerical simulations of spectra for various boron species in these samples are compiled in Table 1. The numbers and types of the components involved here look very similar to those observed in B-β (see Figure 2b) except the contribution of B[4] species. Additional investigation with 23Na MAS NMR (not reported here) confirmed that the amount of sodium in borosilicate zeolites shows a good correlation with the signal intensity of B[4] species. The results support the previous findings that alkali metal ions protect B[4] sites from the coordination transformation to B[3] species.8,9,28 The relative amount of B[3]* vs B[3] seems to be proportional to the amount of B[4] species, but more cases should be examined before generalizing the trend. The asymmetric parameter η of B[3]* appears to be slightly larger for B-β than that of other zeolites, although it is not proper to take the difference seriously because of low accuracies in the determination due to the blurred line shape. Note that the P1 resonance was also observed with a B-SSZ-33 sample (see Figure 6a).

Figure 6. Experimental 11B MAS NMR spectra at 11.7 T along with simulated line shapes: (a) B-SSZ-33 and (b) B-SSZ-42. Both samples were dehydrated at 120 °C in a vacuum.


Hwang et al.

TABLE 1: Boron Species Found during the Hydration/Dehydration Treatment of Borosilicate Zeolites

zeolite B-β (BEA*)

chemical shift species (ppm) B[3] B[3]* B[4]-I B[4] B[3]-nf

B-SSZ-33 B[3] (CON) B[3]* B[4] B-SSZ-42 B[3] (IFR) B[3]* B[4] a

10.0 14.6 -3.5 -3.8 18.0 16.7a 19.2a 10.1 15.3 -3.2 -3.6 10.7 14.8 -3.4

quadrupole parameters Cqcc and η 2.53 MHz, 0.15 2.46 MHz, 0.29 2.15 MHz, 0.20 1.8 MHz (SOQE) 1.8 MHz (SOQE) 2.54 MHz, 0.13 2.46 MHz, 0.2 2.6 MHz, 0.18 2.4 MHz, 0.2

rel intens vs B[3]

ref

1 0.26 0.02 0.04 0.28

8, 17, 28 17

1 0.3 0.1 0.03 1 0.4 0.3

2 2

8, 17

Denoted as P1, the identity is yet unknown (see text).

The P1 peak disappeared when the material was evacuated at higher temperatures. When both zeolites were dehydrated at higher temperatures, no significant differences in either 1H or 11B MAS NMR spectra were observed up to the dehydration temperature of 500 °C in contrast to those of B-β sample (see Figure 1). It was generally observed for all three zeolites that the 11B NMR line shape of B[3]* species got smeared out to become a part of the dominating B[3] peak when the evacuation process took place over 350 °C. The results clearly imply the conversion of B[3]* to B[3]. However, the overall line shape of 11B MAS NMR spectra for samples evacuated above 350 °C appeared to be distorted and shows a broad tail on the left side of the powder pattern of B[3]. The change is well illustrated between parts d and e of Figure 1 and more evidently in Figure 4d. The line shape can be better explained with inclusion of another powder pattern which severely deviates from the planar structure (η > 0.6) and occupies about 10% of the main B[3] sites in the case of B-β sample E (Figure 1e). The results indicate that roughly 30% of B[3]* species could remain unconverted but significantly altered from its original geometry as a result of the high-temperature treatment. No noticeable change in the B[4] peaks at about -4 ppm was detected after the hightemperature dehydration. 5. 1H NMR and 11B CPMAS NMR: Characterization of B[3]*. A 1H MAS NMR spectrum of B-β zeolite was obtained at each step of the dehydration process and Figure 1 has already illustrated the extent of dryness of the samples as a function of the dehydration temperature. Besides the eminent spectral change showing the disappearance of surface adsorbed water, 1H MAS NMR spectra showed various changes of surface hydroxyl groups that could be correlated with the changes observed in 11B NMR spectra (see Figure 1). A closer examination of 1H NMR spectra is performed in this section. The spectrum of B-β sample D in Figure 1d is redisplayed in Figure 7a after deconvolution into five different components: 1.7, 2.0, 2.2, 3.0, and 5.0 ppm. The broad peak at 5.0 ppm was shifted to upfield (4.6 ppm) and disappeared completely at evacuation temperatures greater than 350 °C while the majority of upfield peaks showed no significant decrease in intensity or variation of chemical shift as the dehydration temperature was raised. Figure 7b shows a plot of intensity vs dehydration temperature for the decomposed lines. Note that the 3.0 ppm peak showed a minor upper field shift of about 0.2 ppm at higher temperatures and a gradual 66% loss of intensity by 450 °C. A 1H MAS NMR spectrum of a relatively well dehydrated β zeolite sample has been known to give rise to a

number of different lines due to hydroxyl groups in different environments.30 Regardless of the type of heteroatoms (Al or B), a 1H NMR spectrum would show the presence of mainly two classes of hydroxyl groups: bridging hydroxyls that are responsible for the Brønsted acid sites and isolated SiOH groups. It is also known that the weak acid sites of borosilicate zeolites is closely related to low values of isotropic chemical shifts (2-3 ppm) of the possible bridging hydroxyl groups in the NMR spectra.8,31 Unlike aluminosilicate zeolites, the proton form of borosilicate zeolites undergoes B[4] to B[3] transformation upon dehydration, which most likely precludes the existence of the bridging hydroxyl groups (B-O(H+)-Si) around B[3] units (see Scheme 1). In addition, B[4] units that are still present even after dehydration should be most likely charge compensated by Na ions, not H+. The decomposed lines in Figure 7 can be assigned thus as follows: strongly adsorbed surface water (5.0 ppm), different types of hydroxyl groups located around the trigonal boron sites B[3] and B[3]* (3.0, 2.2, and 2.0 ppm) and the isolated SiOH groups (1.7 ppm). From a similarly treated B-β zeolite, Fild et al.8 also reported three 1H NMR lines at 2.0, 2.3, and 3.0 ppm that were proved by REDOR NMR experiments to be closely located to the framework trigonal boron sites, i.e., B[3]- - -O(H)Si groups. In terms of position and behavior of individual lines upon dehydration, very similar experimental results were also obtained from 1H NMR measurements of B-SSZ-33 and B-SSZ-42 in this study. A noticeable difference in the relative intensities of individual peaks was observed among B-β, B-SSZ-33, and B-SSZ-42. For example, upon dehydration at 250 °C, at which most of the strongly bound surface water is removed, B-SSZ-42 showed a remarkably small amount (∼30%) of a hydroxyl peak at 2.2 ppm compared to that found in B-SSZ-33 or B-β samples. However, such variations in the distribution of individual hydroxyl groups could not be correlated with differences observed from 11B MAS NMR experiments (e.g., B-SSZ-42 contains the highest amount of B[4] units.). Unlike aluminosilicate zeolites, not much effort has been made to characterize 1H NMR signals of borosilicate zeolites. An obvious hurdle is related to the fact that the chemical shifts of the three lines corresponding to the hydroxyl groups around B[3] sites are not well distinguished from other types of hydroxyl groups present in zeolites. It is worthwhile to note that the 1H NMR spectrum of deboronated and dehydrated B-SSZ-33 exhibits deconvoluted line components very similar to those observed in parent B-SSZ-33 (see Figure 1S in the Supporting Information for data from deboronated and dehydrated B-SSZ-33). It is quite striking to observe lines at chemical shifts of 4.5, 3.0, 2.3, and 2.0 ppm that are exactly identical to those observed before deboronation, although the contributions of 4.5 and 3.0 ppm peaks are found to be particularly greater. These results support the finding that the presence of B[3] right next to SiOH groups makes almost invisible contributions in deterring the chemical shifts of the hydroxyl groups. Consequently, the 1H NMR signal of the B[3]- - -HOSi group can be only investigated through boron atoms, and 1H-11B double resonance NMR offers an excellent tool. Much of the detailed results of 1H NMR studies will be published elsewhere. In the present work, we focus on better characterization of the B[3]* species using 11B CPMAS experiments and 2D HETCOR NMR. The cross-polarization method was used in both one- and twodimensional fashions to show which hydroxyl sites are closely located around the two known trigonal boron species, B[3] and B[3]*. Since all three zeolites gave similar results, only the experimental spectra from B-SSZ-33 are presented in Figure



Figure 7. (a) 1H MAS NMR spectrum of B-β sample D (dehydrated at 120 °C, see Figure 1d) and its deconvoluted components. (b) Dependence of peak intensities on the dehydration temperature.

Figure 8. (a) 11B CPMAS NMR spectra of B-SSZ-33 at different cross-polarization contact times after the sample was dehydrated at 120 °C. (b) 2D HETCOR spectrum of B-SSZ-42 after dehydration (0.5 ms contact time).

8a. The cross polarizing efficiency was found to be higher for B[3]* as the corresponding powder pattern appears even at 0.2 ms of cross-polarization contact time. The growth of powder pattern for B[3] species becomes evident after increasing contact times to greater than 0.5 ms, and the spectrum at 5.0 ms resembles closely to the MAS spectrum shown in Figure 6a. Note that B[4] sites are also cross polarized from protons but with much lower efficiency, supporting the hypothesis that the BO4- unit could have Na+ as the countercation. The individual correlations can be mapped out efficiently by means of the 2D HETCOR method. Figure 8b shows a 2D HETCOR spectrum of dehydrated B-SSZ-42 zeolite. The spectrum was obtained using 0.5 ms contact time and a sample spinning rate of 10 kHz. Under restriction of the contact time used in this experiment (0.5 ms), a few observations are summarized below: (1) Surface bound water (∼4.2 ppm) and one type of hydroxyl group (3.0 ppm) show evident correlation with the B[3]* band, indicating that they are closely located to B[3]* units. (2) The hydroxyl group at 3.0 ppm is also responsible for cross-polarization of B[3] units. (3) B[3] units reveal a rather weak correlation with the 2.2 ppm peak, meaning that the corresponding hydroxyl group is weakly tied to the B[3] sites. (4) No correlation from the 2.0 and 1.7 ppm peaks was observed. From all these results, it can be concluded that the hydroxyl groups responsible for the 3.0 ppm resonance in 1H NMR are

most closely associated with the SiOH- - -B[3] sites while geometrically the distance from the hydroxyl proton to B[3]* seems to be shorter than the distance to B[3]. However, the 2D correlation contour maps exhibited significant variations at different contact times and different dehydration temperatures. When a 2D spectrum was acquired with a longer contact time (5 ms) for a borosilicate zeolite that was dehydrated at 450 °C, the 1H lines projected on the 1H spectrum axis (figure not shown here) appeared to be very close to those in the 1D MAS spectrum (Figure 1e). This is due to the nature of the crosspolarization process which makes the loosely correlated spins actively contact at longer contact periods. In this case of longer contact time, B[3] species were dominantly correlated with all the individual 1H lines. The contribution of the 1.7 ppm peak to the contour map has not yet been detected unambiguously. On the other hand, when examining the mildly hydrated B-β sample C (see Figure 1c for the corresponding 1H NMR spectrum), only the B[3]* powder pattern was correlated with a broad 1H line at 4-6 ppm region. A series of 2D HETCOR spectra were then obtained at each different stage of the dehydration process in order to closely investigate the hydroxyl groups around the triognal boron sites. B-β zeolite was used for this study (see Figure 9 for 2D spectra). The collection of projected lines onto the 1H axis is presented in Figure 10. The results are discuused below. i. Only B[3]* sites show correlation with the ∼4 ppm 1H resonance that is thought to represent surface bound water (4-6


Figure 9.

11

Hwang et al.

B 2D HETCOR spectra of calcined B-β evacuated at (a) room temperature and (b) 120, (c) 250, (d) 350, and (e) 450 °C.

Figure 10. 1H MAS NMR spectra of B-β observed via 2D 1H-11B HETCOR spectroscopy at different stages of dehydration (also see Figure 9). Evacuation temperatures were (a) room temperature and (b) 120, (c) 250, (d) 350, and (e) 450 °C. (f) Deconvolution of part b.

ppm) (Figures 9a and 10a). It is a fairly striking observation considering that hydrogen bonded water groups (H2O)n surround any hydroxyl groups and trigonal boron sites. If the mobility

of water groups or a proton exchange process prohibits the efficient cross-polarization to trigonal borons, there is no reason for B[3]* sites to be selectively cross polarized. Therefore, it is

Boron Sites in Borosilicate Zeolites probable that the broad 1H lines around 4-6 ppm might not only represent the surface bound water groups but also B-OH groups both in the trigonal and tetrahedral defective sites as shown in Scheme 2. In fact, the 1H spectrum in Figure 10a can be decomposed to two lines at 5.6 and 4.1 ppm with 2:3 ratio in intensity. The 5.6 ppm peak was removed when dehydration at 120 °C was performed while the presence of the 4.1 ppm peak was observed even after evacuation at 250 °C (see below). ii. Upon removal of most of the surface bound water (Figures 9b and 10b), two proton lines at 3.3 and 2.4 ppm show selective correlations to mainly B[3]* and B[3] sites, respectively. Correlation between 3.3 ppm and B[3] sites seems to be present to some extent but the strength appears to be less when compared to the case of B-SSZ-42 (Figure 8b). The 1H projection spectrum in Figure 10b can be decomposed to four lines at 3.8, 3.3, 2.4, and 2.0 ppm as shown in Figure 10f. The components are consistent with our previous fit of 1H spectra (see Figure 7a) except for the 1.7 ppm peak, which was assigned to the isolated SiOH groups. The result indicates that most of hydroxyl groups are closely associated with trigonal boron sites. Note that there is a difference of about 0.2-0.3 ppm in line positions measured with these two different methods, i.e., the direct 1H MAS NMR and the projected 1H spectrum of 2D HETCOR NMR. Despites of slight upfield shift caused by removal of surface bound water, the 3.8 ppm peak is believed to originate from the same hydroxyl proton that was observed at 4.1 ppm peak in Figure 10a. The line width of the 3.8 ppm was measured to be ∼ 2 ppm wide. Its contribution appears to be as high as 50% of the 1H spectrum. It is observed in the 2D spectrum (Figure 9b) that the 3.8 ppm shows strong correlation only with the B[3]* sites. The peak was no longer present in spectra for dehydration temperatures greater than 350 °C (Figure 10, parts d and e). This observation is in good agreement with Koller’s report on an additional proton site at 4.3 ppm observed from a calcined and dehydrated B-β using 1H{11B} REDOR NMR experiments.17 According to Koller, the 4.3 ppm resonance appears to show the strongest correlation with trigonal boron sites. Using the REDOR method alone is, however, not possible to resolve which of the two trigonal sites (B[3] and B[3]*) is correlated. The identity of the 3.8 ppm peak is yet to be explored. Again, the 2D HETCOR method proves to be remarkably instrumental because it provides detailed insights into the structures that are directly related to the 11B nuclei, which are especially in the trigonal geometry. iii. When the surface bound water is completely removed from the zeolite (Figure 10, parts d and e), the positions of hydroxyl groups in the projected spectra can be more clearly observed. The 3.8 ppm peak is no longer present. At 450 °C, measurable reduction of the 3.3 ppm peak is also detected, and at this point the selective correlation pairs, OH(3.3 ppm)- - -B[3]* and OH(2.4 ppm)- - -B[3], are more clearly represented in the 2D contour plot (see Figure 9e). The gradual depletion of the 3.3 ppm peak now can be directly correlated to the reduction in intensity of the B[3]* group after evacuation at 450 °C. Considering the fact that the B[3]* species appeared to be an intermediate formed in the process of hydration/dehydration of borosilicate zeolites, the hydroxyl group OH(3.3 ppm) is found to be a critical component of its formation and deformation. All of the spectroscopic evidence presented in this study strongly suggests that the identity of B[3]* species should be attributed to trigonally coordinated boron atoms with one or two of B-OSi bonds being replaced by B-OH(3.3 ppm). Having directly coordinated hydroxyl groups, it is rational to expect that B[3]* species have cross-polarization efficiency superior to B[3] sites

J. Phys. Chem. B, Vol. 108, No. 48, 2004 18545 SCHEME 3

even under hydrated circumstances, which agrees well with our experimental observations. Using 1H-27Al CPMAS NMR method, Roberge et al. have drawn a similar conclusion and proposed the presence of tetrahedral Al site bearing a direct Al-OH bond in their study of Al-β zeolite.32 Conversion of B[3]* to B[3] can now be understood as a consequence of dehydroxylation when dehydration takes place at higher temperatures (over 400 °C). Unresolved issues regarding the incomplete conversion of B[3]* to B[3] and severe alteration of B[3]* geometry can be explained as follows. For example, when two surrounding hydroxyl groups, Si-OH, are involved in forming Si-O-Si bond by ejecting a H2O molecule (Scheme 3), the B[3]* species would be isolated and could be stable at high-temperature such as 450 °C. In this case, a significant structural distortion around B[3]* sites might occur, resulting in a considerable change of the asymmetric parameter of B[3]* species. Note that the structure of B[3]* is depicted with one B-OH bond in Scheme 3 without determining how many OH groups were coordinated to B[3]* sites. However, if B[3]* bears two hydroxyl groups (B(OSi)(OH)2, see B[3]-II in Scheme 2), it is unlikely to have such structural distortion as the dehydroxylation proceeds. The B[3]* site can now be assigned to B[3]-I, a defective trigonal boron site (Scheme 2). At this moment, we have not obtained a firm experimental evidence that could conclusively determine whether B[3]* can be assigned to either B[3]-I or B[3]-II. Note that the assignment of B[3]* to B[3]-II was recently proposed by Koller.17 The formation of B[3]-I was not detected by Koller. It is also probable that the 3.8 ppm peak is responsible for hydroxyl groups in the boron geminal structure, B(OSi)(OH)2, and B[3]-I and B[3]-II are positioned in 11B NMR too close with each other to be distinguished in our measurements reported here. It is postulated that B[3]-II species be completely converted to either B[3]-I or B[3] species during dehydration below 350 °C. Further investigation is underway to make conclusive identification. iv. The association of the 2.0 ppm peak with any of trigonal boron sites was not conclusively observed in our current setup (0.5 ms contact time) although the OH(2.0 ppm) shows a correlation band with B[3] sites at a longer contact times (not shown here). This observation is in disagreement with previous reports8,17 which showed that 2.0 ppm peak is associated with the trigonal boron sites to a similar extent as the 2.4 ppm peak. As demonstrated in their 1H{11B} REDOR experiments using dehydrated B-β, however, the distances from the framework trigonal boron sites to different hydroxyl sites (3.0, 2.3, and 2.0 ppm) appear to be reasonably distinct, with OH(3.0 ppm) being the shortest while OH(2.0 ppm) is the longest. Discrimination between OH(2.4 ppm) and OH(2.0 ppm) could be investigated quantitatively by means of REDOR NMR. The presence of two different types of bridging hydroxyl groups in aluminosilicate zeolites have been known and exploited in versed methods.30,32-36 It is interesting to learn whether the two hydroxyl groups that are associated with B[3] sites in borosilicate zeolites would show acidic character similar to that of aluminosilicate zeolites.

18546 J. Phys. Chem. B, Vol. 108, No. 48, 2004 IV. Conclusions The formation and the evolution of a number of trigonal and tetrahedral boron sites during the hydration/dehydration process were investigated in detail in this work by employing multinuclear and multidimensional solid-state NMR methods. NMR evidence supports the presence of a B(OSi-)2(OH) species, B[3]-I, a defect site that showed stability up to 450 °C. Distorted B(OH)3 type species were detected while the formation of B(OSi-)(OH)2 species was not unambiguously concluded. Boron shows a great degree of flexibility in coordination conversion between trigonal boron, B[3], and tetrahedral boron, B[4], in the framework of calcined borosilicate zeolites upon change of the hydration level. Boron can be removed from the framework of the proton form of dehydrated borosilicate zeolites even under mild hydration treatment. The deboronation mechanism is related to successive substitutions of B-O-Si bonds with B-OH. Even after a great extent of deboronation via hydration, reoccupation of boron into the framework position can be achieved by dehydration treatment. 1H-11B crosspolarization and two-dimensional heteronuclear correlation spectroscopies provide powerful tools to selectively identify the proton resonances corresponding to hydroxyl groups in B[3]-I (3.8 and 3.3 ppm) and silanol groups near to B[3] sites (2.4 ppm). Acknowledgment. We thank ChevronTexaco Energy Technology Co. for supporting this work. Acquisition of NMR spectra at 19.6 T was made possible by help from Drs. Hyungtae Kwak and Z. Gan at the NHMFL in Florida. Their contribution is greatly acknowledged. Helpful discussion from Prof. Mark E. Davis at California Institute of Technology is also acknowledged. The NMR facility at Caltech was supported by the National Science Foundation under Grant 9724240 and partially supported by the MRSEC Program of the National Science Foundation under Award DMR-0080065. Supporting Information Available: Figure 1S, showing 1H MAS NMR spectra of SSZ-33 after deboronation and dehydration at 250 °C. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Millini, R.; Perego, G.; Bellussi, G. Top. Catal. 1999, 9, 13. (2) Chen, C. Y.; Zones, S. I.; Hwang, S. J.; Bull, L. M. Proceedings: 14th International Zeolite Conference; van Steen, E., Callanan, L., Claeys, M., Ed.; 2004; p 1547.

Hwang et al. (3) Chen, C. Y.; Zones, S. I. Proceedings: 13th International Zeolite Conferemce; Galarneau. A., Di Renzo, F., Fujula, F., Vedrine, J., Eds.; Elsevier: Amsterdam, 2001; paper 26-O-05 on CD-ROM. (4) Chen, C. Y.; Zones, S. I. Proceedings: 13th International Zeolite Conferemce; Galarneau. A., Di Renzo, F., Fujula, F., Vedrine, J., Eds.; Elsevier: Amsterdam, 2001; paper 11-P-16 on CD-ROM. (5) Scholle, K. F. M. G. J.; Veeman, W. S. Zeolites 1985, 5, 118. (6) Brunner, E.; Freude, D.; Hunger, M.; Pfeifer, H.; Reschetilowski, W.; Unger, B. Chem. Phys. Lett. 1988, 148, 226. (7) Coudurier, G.; Auroux, A.; Vedrine, J. C.; Farlee, R. D.; Abrams, L.; Shannon, R. D. J. Catal. 1987, 108, 1. (8) Fild, C.; Shantz, D. F.; Lobo, R. F.; Koller, H. Phys. Chem. Chem. Phys. 2000, 2, 3091. (9) de Ruiter, R.; Kentgens, A. P. M.; Grootendorst, J.; Jansen, J. C.; van Bekkum, H. Zeolites 1993, 13, 128. (10) Stave, M. S.; Nicholas, J. B. J. Phys. Chem. 1995, 99, 15046. (11) Valerio, G.; Plevert, J.; Goursot, A.; di Renzo, F. Phys. Chem. Chem. Phys. 2000, 2, 1091. (12) Engelhardt, G.; Michel, D. High-Resolution Solid-State NMR of Silicates and Zeolites; Wiley: Chichester, England, 1987. (13) Frydman, L.; Harwood, J. S. J. Am. Chem. Soc. 1995, 117, 5367. (14) Amoureux, J. P.; Fernandez, C. Solid State NMR 1998, 10, 211. (15) Amoureus, J. P.; Pruski, M. Encyclopedia of NMR; Grant, D. M., Harris, R. K., Eds.; John Wiley&Sons: Chichester, England, 2002; Vol. 9, pp 226-251. (16) Schmidt-Rohr, K.; Clauss, J.; Spiess, H. W. Macromolecules 1992, 25, 3273. (17) Koller, H. Proceedings: 14th International Zeolite Conference; van Steen, E., Callanan, L., Claeys, M., Ed.; 2004; p 1380. (18) Zones, S. I. U.S. Patent 4,963,337, 1990. (19) Zones, S. I. U.S. Patent 5,653,956, 1997. (20) Zones, S. I.; Hwang, S.-J. Microporous Mesoporous Mater. 2003, 58, 263. (21) Zones, S. I.; Yuen, L. T.; Toto, S. D. U.S. Patent 5,187,132, 1993. (22) Amoureux, J.-P.; Fernandez, C.; Steuernagel, S. J. Magn. Reson., Ser. A 1996, 123, 116. (23) Vega, A. J. J. Magn. Reson. 1992, 96, 50. (24) Vega, A. J. Solid State NMR 1992, 1, 17. (25) Amoureux, J.-P.; Pruski, M. Mol. Phys. 2002, 100, 1595. (26) Amoureux, J.-P.; Fernandez, C.; Dumazy, Y. 37th Rocky Mountain Conference, Denver, CO, 1995, Abstract No. 264. (27) Freude, D.; Haase, J. NMR Basic Princ. Prog. 1993, 29, 1. (28) de Ruiter, R.; Pamin, K.; Kentgens, A. P. M.; Jansen, J. C.; van Bekkum, H. Zeolites 1993, 13, 611. (29) Cotton, F. A.; Wilkinson, G. AdVanced Inorganic Chemistry, 6th ed.; Wiley-Interscience Publication: New York, 1999. (30) Mu¨ller, M.; Harvey, G.; Prins, R. Microporous Mesoporous Mater. 2000, 34, 281. (31) Scholle, K. F. M. G. J.; Kentgens, A. P. M.; Veeman, W. S.; Frenken, P.; van der Velden, G. P. M. J. Phys. Chem. 1984, 88, 5. (32) Roberge, D.; Hausmann, H.; Holderich, W. F. Phys. Chem. Chem. Phys. 2002, 4, 3128. (33) Beck, L. W.; White, J. L.; Haw, J. F. J. Am. Chem. Soc. 1994, 116, 9657. (34) Mu¨ller, M.; Harvey, G.; Prins, R. Microporous Mesoporous Mater. 2000, 34, 135. (35) Brunner, E.; Beck, K.; Koch, M.; Heeribout, L.; Karge, H. G. Microporous Mater. 1995, 2, 395. (36) Paze, C.; Zecchina, A.; Spera, S.; Cosma, A.; Merlo, E.; Spano, G.; Girotti, G. Phys. Chem. Chem. Phys. 1999, 1, 2627.