Synthesis, Isotopic Enrichment, and Solid-State NMR Characterization of Zeolites Derived from the Assembly, Disassembly, Organization, Reassembly Process.

The great utility and importance of zeolites in fields as diverse as industrial catalysis and medicine has driven considerable interest in the ability to target new framework types with novel properties and applications. The recently introduced and unconventional assembly, disassembly, organization, reassembly (ADOR) method represents one exciting new approach to obtain solids with targeted structures by selectively disassembling preprepared hydrolytically unstable frameworks and then reassembling the resulting products to form materials with new topologies. However, the hydrolytic mechanisms underlying such a powerful synthetic method are not understood in detail, requiring further investigation of the kinetic behavior and the outcome of reactions under differing conditions. In this work, we report the optimized ADOR synthesis, and subsequent solid-state characterization, of 17O- and doubly 17O- and 29Si-enriched UTL-derived zeolites, by synthesis of 29Si-enriched starting Ge-UTL frameworks and incorporation of 17O from 17O-enriched water during hydrolysis. 17O and 29Si NMR experiments are able to demonstrate that the hydrolysis and rearrangement process occurs over a much longer time scale than seen by diffraction. The observation of unexpectedly high levels of 17O in the bulk zeolitic layers, rather than being confined only to the interlayer spacing, reveals a much more extensive hydrolytic rearrangement than previously thought. This work sheds new light on the role played by water in the ADOR process and provides insight into the detailed mechanism of the structural changes involved.

Introduction

Zeolites are inorganic porous compounds made up of fully cross-linked frameworks of corner-sharing SiO4 and AlO4 tetrahedra.[1] Owing to their crystalline structure and porosity, these materials have high surface areas, and so great potential for catalytic activity. They also possess a narrow range of pore sizes of molecular dimensions, leading to their well-known shape selectivity.[2−10] Such a unique combination of structural features accounts for their vast success and variety of applications over the years, and this has kept the targeting of new framework types at the forefront of research.[11−20] The recently introduced assembly, disassembly, organization, reassembly (ADOR) method,[21−25] shown schematically in Figure , has proven to be a feasible approach to achieve such a goal, transforming the way new, stable, and active materials can be synthesized. This unconventional synthetic approach is based on the chemically selective disassembly of hydrolytically unstable parent frameworks and their subsequent controlled reassembly into solids with predetermined structures. The key feature characterizing the parent zeolite required for the process is the presence of a hydrolytically sensitive dopant element incorporated within the framework at specific sites, which allows the chemically selective removal of the units that contain the dopant. Germanosilicates are excellent candidates for this process as Ge, which has greater flexibility in its coordination environment than Si, is known to locate preferentially within certain zeolite subunits and, in particular, occupies tetrahedral-atom positions within the double four rings (d4rs).[26−29] Moreover, germanosilicate zeolites have been shown to be much more sensitive to hydrolysis than silica or aluminosilicate materials.[29−31] Using germanosilicates that adopt the UTL[32,33] framework structure as the parent zeolite, ADOR has enabled the preparation of two new zeolite materials, IPC-2 and IPC-4,[21,24] whose topologies have been given the three-letter codes OKO and PCR, respectively, by the International Zeolite Association. It has also allowed the preparation of other new zeolite structures IPC-6, IPC-7, IPC-9, and IPC-10,[23,24,34] and some of these are likely to be difficult, or even impossible, to prepare using standard hydrothermal methods.[35,36]

Figure 1

General schematic representation of the ADOR process when using large volumes of hydrolysis solution. A is the assembly step for the preparation of the parent Ge-UTL zeolite. D is the full hydrolysis of the parent material to leave the layered IPC-1P intermediate. Over time, there is an organization (O) process that rearranges the structure to produce a new intermediate material, IPC-2P. Reassembly (R) by calcining the intermediates leads to the fully connected IPC-2 and IPC-4 zeolite structures. Note that the structures of IPC-1P, IPC-4, and IPC-2 are well established in the literature but that IPC-2P is heavily disordered, and the model shown is idealized. Full details of the structures and conditions used for the processes can be found in the literature.[21,24,37] Color key: blue = Si, red = O, except for UTL where blue = Si or Ge.

Controlling the acidity of the solution used to disassemble the parent zeolite in the ADOR process enables different final materials to be effectively targeted and their porosity tuned over a wide range.[24] Such a control over pore volume/surface area is difficult to achieve following traditional approaches to zeolite synthesis, highlighting the notable advance achieved by the development of the ADOR process. However, a better understanding of the hydrolytic mechanisms controlling the ADOR process is required, and, in particular, a more detailed insight of the interaction of the zeolitic framework with water would be beneficial. Thanks to its sensitivity to the local, atomic-scale environment, and its element-specific nature, solid-state NMR spectroscopy can complement other solid-state characterization techniques, such as powder X-ray diffraction (PXRD) and N2 adsorption, used for the investigation of zeolitic frameworks.[1,38−42] For most silica-based zeolites, 17O and 29Si are the two NMR-active nuclei of most interest. Despite its low natural abundance of 4.7%, 29Si (I = 1/2) is routinely employed for the NMR characterization of zeolites. However, 17O (I = 5/2) is less commonly studied, owing to the anisotropic quadrupolar broadening of the spectral line shapes, very low natural abundance (0.037%), and moderate gyromagnetic ratio.[43,44] To enable a complete and high-resolution spectroscopic investigation of the zeolitic structures of interest in each step of the ADOR process, isotopic enrichment is required. Here, we show how hydrolysis of a 29Si-enriched Ge-UTL zeolite using H217O can shed new light on the role played by water in the ADOR process and provide insight into the structural changes involved in the mechanism. The work shows that the detailed mechanistic pathway, and so the final products, of the ADOR process depends heavily on the conditions used, including the volume of the solution used to hydrolyze the initial parent zeolite.

Experimental Methods

Experiments on unenriched zeolites were carried out on materials from a single starting batch of Ge-UTL (Si/Ge ratio of 4.4), which was synthesized and calcined following the procedure in ref (24), but using a shorter reaction time (under static heating conditions) of 7 days. A similar procedure was followed to synthesize 29Si-enriched Ge-UTL, starting from 18% 29Si-enriched TEOS, with a longer overall reaction time of 14 days. The unenriched calcined Ge-UTL starting material was hydrolyzed with water (natural abundance) for different lengths of time. Low-volume reactions were carried out using a 10 mL round-bottomed flask topped with a condenser in refluxing conditions at 95 °C in 6 M HCl (freshly produced from 1.2 mL of natural-abundance water and 1.2 mL of 12 M HCl) for reaction times ranging from 4 to 48 h. For 17O-enriched and doubly 17O- and 29Si-enriched zeolites, hydrolysis was carried out using 6 M HCl (produced from 1.2 mL of water 41% enriched in 17O and 1.2 mL of 12 M HCl) for 16 h. Materials were washed with only a small volume (2.4 mL) of natural-abundance water to minimize any loss of 17O (see the Supporting Information for more detail). Calcination of the hydrolyzed products was carried out to remove any remaining water from hydrolysis and allow the condensation of the hydrolyzed layers in the framework. Typically, the zeolite was heated to 575 °C at a rate of 1 °C/min, held for 6 h, and cooled to room temperature at a rate of 2 °C/min under air.

29Si solid-state NMR spectra were acquired using a Bruker Avance III spectrometer, equipped with a 9.4 T wide-bore superconducting magnet, at a Larmor frequency of 79.459 MHz. Samples were packed into 4 mm ZrO2 rotors that were rotated at a rate of 10 kHz, using a 4 mm HX probe. Magic angle spinning (MAS) spectra were acquired using a radiofrequency (rf) field strength of ∼83 kHz, with a repeat interval of 120 s. The Q4/Q3 ratio was determined using DMFit,[45] with errors estimated from multiple fits. For cross-polarization (CP)[44,46] experiments, transverse magnetization was transferred (from 1H) using a contact pulse of 5 ms (ramped for 1H), with two-pulse phase modulation (TPPM) 1H decoupling (using a rf field strength of ∼70 kHz) during acquisition. The commercial probes used for these experiments exhibited a small 29Si background signal (estimated to be ∼8% of the total signal intensity). No correction has been made to the Q4/Q3 ratios plotted in Figure to account for this.

Figure 3

Plots of the Q4/Q3 intensity ratio in 29Si (9.4 T, 10 kHz MAS) NMR spectra of Ge-UTL zeolite as a function of hydrolysis time under low-volume conditions for samples (a) as prepared and (b) after calcination. In (a), the red and blue dashed lines show the Q4/Q3 ratios for idealized IPC-1P (2.5) and IPC-2P (7), respectively. Error bars have been estimated from multiple fits.

17O solid-state NMR spectra were acquired on Bruker Avance III spectrometers, equipped with 14.1 or 20.0 T wide-bore superconducting magnets, at Larmor frequencies of 81.331 and 115.248 MHz, respectively. Samples were packed into 3.2 mm ZrO2 rotors and rotated at a rate of 20 kHz. MAS NMR spectra were acquired using a rf field strength of ∼70 kHz, with a repeat interval of 1 s, using continuous wave (cw) 1H decoupling (∼100 kHz) where necessary. For CP experiments, transverse magnetization was transferred (from 1H) using a contact pulse of 1 ms (ramped for 1H), and cw 1H decoupling (∼100 kHz) was applied during acquisition. Variable-temperature CP experiments were carried out at the UK DNP MAS NMR Facility at the University of Nottingham. The neat 17O-enriched solid was packed in a sapphire rotor, a contact pulse of 1 ms was used, and swept-frequency TPPM decoupling (∼100 kHz) was applied during acquisition. Repeat intervals of 1 and 4.3 s were used for experiments recorded at 298 and 108 K, respectively. 17O MQMAS experiments were carried out using a triple-quantum z-filtered pulse sequence[47,48] and are shown after a shearing transformation.[49] Where desired, cw 1H decoupling was applied. The indirect dimension is scaled and referenced according to the convention described in ref (50). Two-dimensional 17O–29Si heteronuclear correlation spectra were acquired at 20.0 T using a D-HMQC pulse sequence (with 2–8 loops of SR421 recoupling),[51,52] central-transition selective pulses for 17O and cw 1H decoupling (∼100 kHz).

Chemical shifts are shown relative to TMS for 29Si and 2H (using Q8M8 (δ(OSi(OMe)3) = 11.5 ppm) and C2D2O4·2D2O (δ(CO2D) = 16.5 ppm) as secondary references) and H2O for 17O. See figure captions for further details.

PXRD data were acquired with a PANalytical Empyrean instrument operated in reflection, Bragg–Brentano, θ-2θ mode, and equipped with a Cu X-ray tube, a primary beam monochromator (CuKα1), and X’celerator RTMS detector. Typically, a 5–50° 2θ range was investigated over 1 h. For all isotopically enriched samples, powders were sealed in capillaries, and data were collected on a STOE STADIP instrument operated in Debye–Scherrer mode equipped with a Cu X-ray tube, a primary beam monochromator (CuKα1), and a scintillation position-sensitive linear detector. Typically, 5–50° or 5–40° 2θ ranges were investigated over 1.5 h or overnight, respectively. N2 volumetric adsorption data were acquired at −196 °C with a Tristar II 3020. Samples were degassed at 300 °C for 3 h prior to the adsorption experiment.

Results and Discussion

Samples of unenriched Ge-UTL were hydrolyzed, and the resulting products calcined and then characterized using 29Si MAS NMR, PXRD, and N2 adsorption. Owing to the cost of H217O and, therefore, the need to reduce the amount of water required, the standard hydrolysis procedure (as reported in ref (24)) needed to be scaled to use a low volume of water (i.e., 2.4 mL) for hydrolysis to enable 17O-enriched materials to be produced economically. Hydrolysis reaction times of 4, 8, 12, 16, 24, and 48 h were investigated, initially using unenriched water, as described in the Experimental Methods section. A striking difference between these studies and the large volume studies previously reported is that the formation of the fully hydrolyzed layered intermediate known as IPC-1P (see Figure ) is not observed.[21,24,37] Under previously reported conditions, IPC-1P can rearrange, with intercalation of extra silicon, into a material called IPC-2P where the layers are connected through O–Si–O links. In the low-volume experiments reported here, there is no evidence of IPC-1P, and at all stages, the intermediates have PXRD patterns with reflection positions consistent with IPC-2P (see the Supporting Information). Calcination of IPC-2P at 575 °C produces materials with similar PXRD patterns to IPC-2, and N2 adsorption measurements on the calcined samples are consistent with this result (see the Supporting Information), which also demonstrates that the small amounts of washing water used had not affected the porosity of the framework. The observation that IPC-1P is never formed points to a different mechanism in low volumes of hydrolysis solutions compared to that observed in higher volumes. The NMR experiments described in detail below are consistent with these observations.

Figure a shows 29Si MAS NMR spectra of the starting Ge-UTL zeolite containing only Q4 (Si(OSi)(OGe)4–) species. CP MAS spectra of this sample contain no signal, confirming no Q3 species are present. Figure b shows the spectrum after 16 h hydrolysis, where both Q4 (Si(OSi)4) and Q3 (Si(OSi)3(OH)) species are present, suggesting the formation of Si–OH groups. The spectral assignment was confirmed using a CP MAS experiment, shown by the red line in Figure b, where selective enhancement of the peak at δ = −102 ppm is observed. After calcination, the relative intensities of the resonances corresponding to Q4 and Q3 species vary slightly, as shown in Figure c, suggesting some Si–OH remain as defects (and confirmed using CP experiments). Unlike the diffraction measurements, the 29Si MAS NMR spectra of the zeolites do show a change with hydrolysis time, revealing an ongoing hydrolysis and rearrangement process, which is difficult to see in the average picture provided by PXRD (see the Supporting Information). Figure a shows a plot of the ratio of the intensities of the Q4 and Q3 species in the hydrolyzed samples as the reaction proceeds. An initial hydrolytic stage is observed with an increase in the amount of Q3 (i.e., Si–OH) sites, and a subsequent decrease in the Q4/Q3 ratio, until 12 h of reaction. However, Q4/Q3 never reaches the ideal number expected for the IPC-1P intermediate (Q4/Q3 = 2.5, red dashed line in Figure a) instead reaching a minimum Q4/Q3 of just over 4. This is consistent with the XRD results described above. After this point, a rearrangement process (similar to that observed in previous work)[53] begins, resulting in a Q4/Q3 ratio that increases until it reaches the expected value for the idealized structure of IPC-2P (Q4/Q3 = 7, blue dashed line in Figure a). This suggests that, under the low volume conditions described here, the initial hydrolysis process never reaches its conclusion before the rearrangement phase of the reaction begins. The relatively constant PXRD patterns for all reaction times after 12 h of reaction indicate that the interlayer spacing does not change significantly after this time, but the variation in Q4/Q3 ratio means that there are still significant changes occurring in the local structure throughout the process. Experiments carried out with larger volumes of washing water proved the repeatability and robustness of these results (see the Supporting Information).

Figure 2

29Si (9.4 T, 10 kHz MAS) NMR spectra of (a) calcined Ge-UTL starting zeolite, (b) after 16 h hydrolysis, and (c) after subsequent calcination. In (b), the 29Si (9.4 T, 10 kHz MAS) CP NMR spectrum is also shown (in red).

Figure b shows a plot of the ratio of the intensities of the Q4 and Q3 species after calcination of the samples shown in Figure a. All samples show the expected PXRD patterns for IPC-2. At the shorter time scales the Q4/Q3 values are approximately the same as the hydrolyzed samples, but the rearrangement process (from 12 h onward) leads to calcined IPC-2 samples that have successively fewer defects (higher Q4/Q3 values). We denote the disordered/defective nature of the zeolites formed at shorter times by an asterisk – IPC-2P* for the intermediate and IPC-2* for the final zeolite after reassembly.

In order to improve sensitivity, enable multidimensional spectroscopic characterization, and provide insight into the ADOR mechanism, two isotopically enriched zeolites were produced. The first involved the hydrolysis (i.e., disassembly) of Ge-UTL using small volumes of (41% enriched) H217O, as described in the Experimental Methods section. In addition, a zeolite sample doubly enriched in 17O and 29Si was prepared by a 16 h hydrolysis (using H217O) of a starting Ge-UTL zeolite synthesized with (18%) 29Si-enriched TEOS.

These materials were confirmed as being IPC-2P by PXRD (as shown in the Supporting Information). Successful 17O enrichment was also demonstrated using 17O MAS NMR. The 17O MAS NMR spectra of the hydrolyzed 17O-enriched sample, shown in Figure , reveal that structural changes occur over a long period after synthesis, with variation in the line shape still apparent after 30 days (during which period the sample remained packed within the NMR rotor at room temperature). After this point, no further change was observed. This indicates that the hydrolysis/rearrangement process continues even at room temperature. Two chemically different types of oxygens are present in the material: “bulk” Si–O–Si species and Si–OH species found in the interlayer regions after hydrolysis. The Si–O–Si species can be further categorized by their local environment as described below. It might be expected that the enrichment level of the interlayer Si–OH groups would be much higher as they are formed by direct hydrolysis in the disassembly step of the ADOR process. The use of enriched H217O provides the opportunity to gain further insight into the mechanism of hydrolysis if the position and proportion of enrichment can be determined and/or quantified.

Figure 4

17O (14.1 T, 20 kHz MAS) NMR spectra of Ge-UTL hydrolyzed with 17O-enriched H2O for 16 h, acquired 2 (black line), 16 (red line), and 30 (green line) days after synthesis.

Owing to the second-order quadrupolar broadening in the 17O (I = 5/2) MAS NMR spectrum, it is difficult to resolve the different species that contribute to the line shape.[43,44] Resolution can be improved using MQMAS experiments,[47,48] where resonances are separated in the δ1 dimension (after appropriate processing) on the basis of their isotropic chemical shifts and quadrupolar shifts (and, therefore, their quadrupolar coupling). As the second-order quadrupolar broadening is inversely proportional to field strength, the resolution observed will vary as B0 changes.[43,44]Figure shows 17O MQMAS NMR spectra of the 17O-enriched zeolite, acquired at 20.0 T, without and with 1H decoupling. (MAS and MQMAS spectra acquired at different B0 field strengths are shown in the Supporting Information.) A resonance is observed in both MQMAS spectra at δ1 ≈ 25 ppm, with ⟨PQ⟩ ≈ 5.3 MHz, and ⟨δiso⟩ ≈ 39 ppm, although a distribution in parameters is apparent. (Note that PQ(43,44) is a combined quadrupolar parameter (see Supporting Information) and is given by CQ(1 + ηQ2/3)1/2 and δiso is the isotropic chemical shift.) These values are also supported by the fitting of cross sections extracted from the MQMAS spectra, as discussed in the Supporting Information. The decoupled MQMAS spectrum (Figure b) shows an additional, lower intensity, resonance at δ1 ≈ 13 ppm, assumed to result from Si–OH species. It is difficult to extract accurate NMR parameters for this resonance, owing to the low sensitivity and lack of any characteristic quadrupolar broadening, but the position of the center of gravity would suggest that ⟨PQ⟩ ≈ 3.4 MHz and ⟨δiso⟩ ≈ 23 ppm. The sharp peak at δ = −5.1 ppm in the MAS spectrum (also shown in Figure ), which does not appear in the MQMAS spectrum, can be attributed to water. The position of this signal varies very little with the B0 field, suggesting a negligible quadrupolar interaction, as discussed in the Supporting Information.

Figure 5

17O (20.0 T, 20 kHz MAS) triple-quantum NMR spectra of Ge-UTL hydrolyzed with 17O-enriched H2O for 16 h, acquired using a z-filtered pulse sequence (a) without and (b) with cw 1H decoupling. The total acquisition times were (a) 7 and (b) 22 h. Also shown (above) is the 17O MAS NMR spectrum. The position of the signal seen in CP experiments is highlighted in green.

To confirm the assignment of the 17O MAS and MQMAS NMR spectra, CP experiments were performed. This approach should edit the spectrum on the basis of spatial proximity to 1H, preferentially enhancing the Si–OH species (as observed for 29Si in Figure b). However, CP to quadrupolar nuclei is a much more challenging experiment than its spin I = 1/2 counterpart, with multiple match conditions of differing intensity and with sensitivity also depending crucially on the spin-lock efficiency (determined by the adiabaticity parameter, α, which depends on the rf field strength of the spin-lock pulse, the quadrupolar coupling, and the MAS rate).[54−56] Consequently, the CP match condition was initially optimized using a model system (amorphous SiO2 enriched in 17O), as described in detail in the Supporting Information. From these experiments, a low-power match condition (enabling spin-locking in the sudden regime, where α ≪ 1) was selected.

As shown in Figure , 17O CP MAS NMR spectra (red lines) of the hydrolyzed 17O-enriched sample do show selective enhancement of the Si–OH signal, with the position of the peak maximum observed at 20.0 T in good agreement with the weak signal resolved in the 1H-decoupled MQMAS spectrum in Figure b (as shown by the green box in Figure ), thus confirming its assignment to Si–OH species. The poor sensitivity of these experiments, even after lengthy acquisition times (3 days at 14.1 T and 13 h at 20.0 T), possibly suggests rapid T1ρ relaxation and/or poor spin-locking behavior. However, as shown in the Supporting Information, the 1H T1ρ relaxation is certainly sufficient for CP at the contact times used, and the Si–OH species exhibit better 17O spin-locking efficiency at the rf fields applied than the Si–O–Si or water species, confirming this is not responsible for the poor CP efficiency observed. Given these observations, it is most likely that the reduced efficiency of CP results from reduction of the dipolar interaction that mediates the magnetization transfer, most probably as a result of motion. Evidence for H dynamics in the hydrolyzed zeolite structure is provided via 2H NMR, as shown in Figure and discussed further in the Supporting Information.

Figure 6

17O (20 kHz MAS) CP NMR spectra (red) of Ge-UTL hydrolyzed with 17O-enriched H2O for 16 h, acquired at (a) 14.1 T and (b) 20.0 T. Corresponding MAS spectra are also shown (black) for comparison.

Figure 7

2H (9.4 T, 10 kHz MAS) NMR spectra of a deuterated 17O-enriched hydrolyzed (16 h) Ge-UTL zeolite, with an expansion to show the broad spinning sideband manifold.

The 2H MAS spectrum is dominated by a narrow, isotropic line shape, suggesting rapidly reorienting D2O is present between the layers. A broad spinning sideband manifold, corresponding to Si−OD in the interlayer space, is also observed, although this cannot be simulated with just a single quadrupolar line shape, suggesting it is also affected by dynamics, and possibly suggesting H/D exchange with water. From the 2H MAS NMR spectra, it can be estimated that the intensity ratio of the Si−OD:D2O signal is ∼1:4 (suggesting a 1:2 ratio of silanol groups to molecular water). It seems reasonable to assume this water results from D2O in the interlayer space and not from surface water (given that samples are dried prior to study and CP to the interlayer Si–OH species appears to be affected by dynamics). However, we note that it is difficult to rule out completely the presence of any strongly bound surface water. The levels of water in the hydrolyzed zeolites vary with hydrolysis duration and storage time and conditions. From the samples studied in this work, the Si–OH:H2O ratio is estimated to be between 1:2 and 1:4. Further support for the modulation of the dipolar interaction being responsible for the poor CP efficiency is provided by the low-temperature (108 K) CP spectrum in Figure . A significant increase in CP efficiency is observed at the lower temperature, with a corresponding change in line shape resulting from the restriction or removal of any dynamics.

Figure 8

17O (14.1 T, 12.5 kHz MAS) CP NMR spectra of 17O-enriched hydrolyzed Ge-UTL zeolite, hydrolyzed for 16 h, acquired at 108 K (black line) and 298 K (red line).

From the cross sections extracted from MQMAS experiments (for Si–O–Si species) and the line shapes observed in CP and spin-lock experiments, where primarily Si–OH species are observed, it is possible to estimate the relative proportions of different O species present in the hydrolyzed Ge-UTL from fitting a quantitative (i.e., short flip angle) 17O MAS NMR spectrum. As shown in the Supporting Information, the ratio of Si–17O−Si:Si–17O−H:H217O species is ∼8:1:1. The amount of water present varies between samples hydrolyzed for different durations and those stored for different times under different conditions. One can broadly categorize the oxygen atoms in IPC-2P into four groups (Figure ): the Si–OH oxygen atoms and three different types of Si–O–Si units; those in the interlayer region, those in the layers but which are accessible to the pore spaces, and those in the layers that are not accessible to the pores. Using the models shown in Figure , one can calculate the expected ratio of Si–17O−Si:Si–17OH based on the distributions shown. This is a rather crude calculation as it takes no account of any difference in probability of any particular site being occupied by 17O preferentially over another, but one thing is clear, the observed Si–17O−Si:Si–17OH ratio of ∼8:1 cannot be explained by incorporation into only the interlayer oxygen sites (expected ratio = 2.5:1), and, at the very least, the organization/rearrangement step in the ADOR process involves oxygen atoms being introduced into the layers (for example, the model shown in Figure c). This observation is unexpected and, even when considering that any back exchange may be more rapid for Si–OH species, suggests a much more substantial rearrangement process during hydrolysis than previously thought.[21−25] The identification of these resonances also provides insight into the changes seen previously in the 17O MAS spectrum in Figure , with the Si–O–Si signal reversibly hydrolyzed to form Si–O–H. Although the low intensity of the Si–OH signal in the MQMAS experiments prevents accurate analysis of the relative proportions of each signal at each stage, it is possible to estimate the relative proportions of different O species present in the three 17O MAS spectra of hydrolyzed Ge-UTL shown. As discussed in the Supporting Information, the proportion of Si–OH increases with time, while that of Si–O–Si decreases, confirming low-level hydrolysis continues even at room temperature, most likely as a result of a small amount of acid remaining between the layers, owing to the reduced volume of (unenriched) washing water used.

Figure 9

A schematic showing hypothetical models of the most likely 17O incorporation patterns into idealized IPC-2P after hydrolysis with 17O-enriched water. The models are (a) where only the Si–OH groups contain 17O; (b) where 17O is incorporated into the Si–OH and Si–O–Si units in the interlayer units (expected Si–17O−Si:Si–17OH ratio is 2.5:1); (c) where 17O is incorporated into Si–OH units, Si–O–Si in the interlayer unit, and into the first layer of tetrahedra in the layers (expected Si–17O−Si:Si–17OH is 11.5:1); and (d) where 17O is equally likely to be incorporated into all possible oxygen sites in the structure (expected Si–17O–Si:Si–17OH is 16.5:1). Green spheres are oxygen atoms with a high probability of 17O incorporation, and red spheres indicate a low probability of 17O incorporation. Silicon is shown in blue. Note that experiments also indicate water is present in the interlayer region, but this is not shown.

This surprising conclusion was investigated further by the preparation of an IPC-2P sample doubly enriched in both 17O (during a 16 h hydrolysis) and 29Si (from the initial Ge-UTL synthesis). Figure shows 17O–29Si HMQC correlation spectra of this doubly enriched sample (acquired at 20.0 T, with SR421 recoupling of the Si–O dipolar interaction).[51,52] At short recoupling times, the spectrum contains signal only from Si–O–Si coordinated to Q4 Si species, while at longer times, the same 17O species correlate with the Q3 Si (i.e., Si–OH) groups, confirming these lie in the bulk of the zeolite layers and at a greater distance to the interlayer spaces. Although little signal is seen from any Si–O–Si coordinated to Q3 Si species within the interlayer regions, there are proportionately fewer of these (and fewer still in the more defective IPC-2P* material), and these might also experience more rapid relaxation, owing to their proximity to the dynamic interlayer water.

Figure 10

17O–29Si (20.0 T, 20 kHz MAS) D-HMQC correlation spectra of 29Si-enriched Ge-UTL hydrolyzed with 17O-enriched H2O for 16 h, acquired using (a) τ = 600 μs and (b) τ = 2400 μs of SR421 recoupling. Also shown (for comparison) are the 17O and 29Si MAS NMR spectra. The dashed green line denotes the position of maximum signal intensity for the Si–OH species identified using CP experiments.

In both spectra, there is no correlation of the 17O signal at δ ≈ – 6.4 ppm with any Si species, confirming it as nonstructural water. Interestingly, there are also no signals in the spectrum in the region where the CP experiments indicated the Si–OH species could be seen. This probably also results from rapid relaxation during the transfer step, as suggested above. Even if some back exchange of Si–OH species does occur when a sample is stored under ambient conditions, perturbing the relative intensities of the species, Figure , confirms a significant proportion of 17O resides in the bulk zeolite layers, rather than being restricted only to the interlayer regions.

Conclusions

The successful synthesis of isotopically enriched IPC-2P ADOR intermediate and IPC-2 zeolite has been demonstrated, and mechanistic studies provide new insight into the hydrolytic rearrangement that underpins the ADOR philosophy. The use of a lower volume of hydrolysis for the reaction to enable economic isotopic enrichment varies the reaction rate and the products obtained at each step, demonstrating the sensitivity of the ADOR process to a variety of experimental conditions and, therefore, the fine control of the reaction progression that may be possible. The work allows us to draw a modified mechanism (Figure ) suitable for these low hydrolysis volume situations that differs from that shown in Figure in that the hydrolysis to IPC-1P never goes to completion before the organization step begins. Both 29Si and 17O NMR spectroscopies have demonstrated that the hydrolysis and rearrangement process occurs over a much longer time scale than seen by diffraction (where a constant d spacing is observed at very short durations). 29Si MAS NMR spectra showed (through a consideration of the Q4/Q3 intensities) that for the first ∼12 h of reaction, hydrolysis was the dominant process, with subsequent rearrangement still occurring up to ∼48 h. However, 17O MAS NMR spectra of zeolites enriched in 17O during the hydrolysis step showed changes of the spectral line shape over a much longer time scale (∼30 days), reflecting a low level of ongoing hydrolysis even at room temperature, most likely as a result of the small amount of acid remaining between the layers, owing to the reduced volume of (unenriched) washing water used.

Figure 11

Modified ADOR process suitable for situations where there is a lower volume of hydrolysis solution and where full hydrolysis of the solution to IPC-1P never goes to completion before the organization step begins. Instead, the defective IPC-2P* is formed, which slowly transforms to a more ideal form of IPC-2P. The steps in which isotropic enrichment is introduced are also shown. Color key: Si = blue, O = red.

2H MAS NMR suggested that there are around 2–4 D2O molecules for every Si−OD species in the interlayer region, although this amount does vary with reaction duration and storage conditions. Two main signals were observed in 17O MAS NMR spectra of the enriched zeolites, attributed to Si–O–Si and Si–OH species (along with signal for H2O itself). MQMAS, CP, and spin-locking experiments were able to provide information on the spectral line shapes of the different components, enabling their relative proportions (8:1 for Si–17O−Si:Si–17OH) to be determined from the MAS spectrum.

The unexpectedly high proportion of Si–O–Si species enriched in 17O (even allowing for faster back exchange of Si–OH groups) highlighted a much more extensive hydrolytic rearrangement than previously thought. Heteronuclear (17O–29Si) correlation experiments confirmed that a substantial amount of 17O was incorporated into a bulk of the layers of the IPC-2P zeolite, rather than being confined only to the hydrolyzed interlayer regions. The ability to exploit isotopic enrichment (in both 29Si and 17O) of these materials has not only enabled a more detailed spectroscopic investigation but also has provided new insight into the ADOR mechanism and the possible ways in which zeolite structures could be more accurately targeted in the future.