Vapour-phase-transport rearrangement technique for the synthesis of new zeolites

Owing to the significant difference in the numbers of simulated and experimentally feasible zeolite structures, several alternative strategies have been developed for zeolite synthesis. Despite their rationality and originality, most of these techniques are based on trial-and-error, which makes it difficult to predict the structure of new materials. Assembly-Disassembly-Organization-Reassembly (ADOR) method overcoming this limitation was successfully applied to a limited number of structures with relatively stable crystalline layers (UTL, UOV, *CTH). Here, we report a straightforward, vapour-phase-transport strategy for the transformation of IWW zeolite with low-density silica layers connected by labile Ge-rich units into material with new topology. In situ XRD and XANES studies on the mechanism of IWW rearrangement reveal an unusual structural distortion-reconstruction of the framework throughout the process. Therefore, our findings provide a step forward towards engineering nanoporous materials and increasing the number of zeolites available for future applications.

Z eolites are crystalline porous materials used for many applications, including gas separation and catalysis [1][2][3] . Although millions of thermodynamically stable structures have been predicted under ambient conditions, until now, the zeolite community has only recognised~250 different zeolite topologies 4 . Such discrepancy between the numbers of proposed zeolite topologies and those produced via traditional hydrothermal approaches has prompted the development of alternative strategies for zeolite synthesis. These new strategies include both direct (e.g., using phosphorus-containing cations 5,6 , metal complexes 7 or proton sponges 8 as structuredirecting agents) and post-synthesis (3D-3D transformation at high pressures >1 GPa 9 ; 3D-2D-3D transformation known as ADOR 10 ) methods. However, most of these approaches use a trial-and-error tactic, except for ADOR 11 . ADOR is a unique approach because the topology of new zeolites can be easily predicted based on the knowledge of the parent structure. ADOR strategy was successfully applied for the structural reconstruction of several germanosilicate zeolites such as UTL 10 , UOV 12 and *CTH 13 (three-letters codes are assigned to established structures of zeolites that satisfy the rules of the IZA Structure Commission). However, attempts to transform other potential structures (e.g., IWW, ITH and ITR, among others) using this approach have been unsuccessful thus far because "open framework" zeolite layers containing pores perpendicular to the layer plane are highly labile, resulting in the complete degradation of the zeolite layers or, at best, in the reconstruction of the initial zeolite framework with a chemical composition different from that of the parent material 14 .
Here, we report a straightforward strategy to construct new zeolites through non-contact vapour-phase-transport (VPT) rearrangement at room temperature. This method offers the opportunity to prepare new zeolite topologies otherwise inaccessible by both hydrothermal and conventional ADOR synthesis routes. We have used a combination of in situ (XRD, XANES) and ex situ (NMR) techniques to follow the rearrangement process and to study the mechanism of VPT transformation involving intermediate structures. Diffraction, adsorption and microscopy data unambiguously showed that the transformation of IWW germanosilicate into the new zeolite phase (designated as IPC-18) proceeds without degradation of zeolite layers and thus affects only interlayer structure units without defects from structural rearrangement.

Results
3D-2D-3D zeolite transformation. The unsuccessful application of the ADOR strategy for the 3D-2D-3D transformation of specific zeolites, including IWW 14 , prompted us to develop alternative approaches for controllable disassembly-reassembly of appropriate zeolite frameworks containing labile structural units. The key step in such transformations is the hydrolysis of labile bonds (typically, Si-O-Ge or Ge-O-Ge) in double-four-ring (D4R) units connecting zeolite layers 15 . Severe hydrolysis conditions and inappropriate chemical composition of parent materials are among the most common reasons for the partial deterioration or even full collapse of zeolite frameworks during the disassembly 16 . Three types of germanosilicates with general formula Si x Ge 1-x O 2 and with chemical compositions (Si/Ge = x/ (1 − x)) similar to those used for conventional ADOR transformations 16 were chosen to implement the VPT 3D-2D-3D rearrangement: UTL (Si/Ge = 4.2), UOV (Si/Ge = 3.1) and IWW (Si/ Ge = 3.7). The structures of these initial germanosilicates are schematically shown in Fig. 1 as the combinations of dense (for UTL) or porous (UOV, IWW) layers and double-four-ring (D4R) building units connecting the layers. Because UTL and UOV zeolites have been already transformed into respective daughter zeolites by conventional ADOR strategy 12,17 , they were chosen in this study as reference materials to evaluate any differences between VPT and conventional ADOR approaches for known {parent zeolite-daughter zeolite} pairs. For the VPT transformation of the selected zeolites, a thin layer of zeolite powder was placed in the reactor containing a membrane permeable to vapours of low-molecular mass reactants located at the bottom part of the reactor (Fig. 1). The rearrangement of the zeolite was initiated by the interaction between acid vapour (12 M water solution of HCl with p H2O = 10 mm Hg and p HCl = 23 mm Hg at 303 K 18 ) and the frameworks{- Volatile germanium chloride (bp GeCl4 = 359 K, for pure GeCl 4 p GeCl4 = 94 mm Hg at 300 K 19 ) resulting from this interaction is adsorbed/dissolved and hydrolysed in the water solution (GeCl 4 + nH 2 O → Ge(OH) n Cl 4-n + nHCl), at the bottom of reactor. Such an approach allows unrestricted mass transport of the species formed by destruction of the most labile Ge-rich connecting units but avoids the deep reconstruction of the germanosilicate framework mediated by water.
Condensation of silanols ( , formed during the hydrolysis, resulted in the formation of the respective daughter structures from parent zeolites (Fig. 1). The extent of the structural rearrangement differs among the zeolites used. Treatment of UTL leads to the formation of the known IPC-7 material (Supplementary Fig. 1) with alternating connectivity between layers through single-four-ring (S4R) and D4R units 17 , whereas transformation of UOV results in the formation of the recently reported IPC-12 zeolite (Supplementary Fig. 1) with direct connectivity between layers, instead of the D4R units that are present in the initial framework 12 . The transformation of IWW results in the formation of a structure in which the layers are exclusively connected through S4R units (see discussion of this structure below). Therefore, the building units formed through VPT rearrangement depend on the structure of initial zeolite: the presence of pores in the layers of UOV and IWW zeolites results in facilitated mass transport and thus in the formation of S4R (for IWW) or O-bridge (for UOV) units, whereas treatment of UTL with non-porous layers produces material with alternating D4R/S4R units, under the same conditions. IPC-7 (derivative of UTL) and IPC-12 (derivative of UOV) materials obtained using liquid-phase ADOR method don't have any major difference in properties with their analogues synthesised by VPT. In contrast, the open framework of layers in IWW zeolite cannot withstand the conventional ADOR treatment. Nevertheless, at room temperature, VPT yields an intermediate derivative (denoted as IPC-18P, Supplementary  Fig. 2) that can be condensed at high temperatures to form the highly crystalline IPC-18 zeolite. The formation of S4R, but not D4R + S4R or O-bridge connections, between layers of IPC-18 can be explained as follows. A simulation study 20 predicted the IWW derivative with S4R connections to have lower formation enthalpy (by 3.5 kJ mol −1 ) than the structure with O-bridges. Conversely, under the conditions applied, mass transport can be inhibited during the transformation. Low-molecular weight siliceous products of hydrolysis of D4R units can remain in the zeolite pores and subsequently participate in condensation to form S4R units, as observed in some ADORable zeolites, such as OKO 21 . As expected, the final IPC-18 material is significantly Gedepleted (Si/Ge decreased from 3.7 to 65.0), thus indicating that germanium atoms are successfully removed from the IWW framework through the formation of volatile species (presumably GeCl 4 ).
Structural features of IPC-18 vs parent IWW. The XRD pattern of IPC-18 matches well that of the theoretically predicted framework (Fig. 2a). Based on Rietveld refinement of the final structure (Fig. 2b, c), the space group was identified as P2 1 /c, which differed from the parent IWW (Pbam). The following parameters of the cell were determined: a = 9.606(4) Å, b = 12.7280(21) Å, c = 40.717(7) Å, α = 90.0°, β = 94.97°, γ = 90.0°. The parent IWW zeolite has a 3D pore system with 8-and 12ring channels, which are both intersected by sinusoidal 10-ring channels. After VPT rearrangement, 8-and 12-ring pores located along the 001 projection remain intact, whereas 10-ring pores become eight-ring pores due to the transformation of D4Rs into S4Rs. This transformation of the pore system resulted in decreased micropore volume and average pore size (Fig. 2d, e). The micropore volume of IPC-18 is 1.7 times lower than that of the initial IWW sample (0.172 vs. 0.104 cm 3 g −1 , respectively). The average channel diameter decreases from 0.63 nm, for IWW, to 0.58 nm, for IPC-18. Despite differing in micropore volume and size, both parent and daughter zeolites show the same {mesopore + interparticle} volume (0.025 cm 3 g −1 ) and similar  Fig. 3). This result indicates that no significant fraction of additional pores, which would influence the textural properties of IPC-18, is created during the VPT transformation. This is an important conclusion because the state of framework silicon atoms is significantly changed in the course of the structural transformation. Solid-state 29 Si MAS NMR analysis of IWW, intermediate IPC-18P and IPC-18 shows the evolution of the degree of condensation of silanols in this process (Fig. 2f). The spectrum of the starting IWW is characterised by the presence of a signal at −115 ppm, which corresponds to Q 4 Si atoms. VPT treatment results in the hydrolysis of most of Si-O-Ge bonds and formation of the IPC-18P precursor, which increases the intensity of silanol groups (Q 3 atoms, −101 ppm) in the spectrum. The spectrum of IPC-18P also contains a low-intensity peak at −93 ppm, which can be related to Q 2 silicon atoms. This is an unusual feature of the zeolitic intermediate obtained by disassembly because a recent in situ NMR study revealed only Q 3 Si among deficient atoms during a top-down structural transformation of germanosilicate 15 . Subsequent thermal treatment of IPC-18P decreases the intensity of the signal attributed to silanol groups due to topotactic condensation of the zeolite layers into a 3D structure, with no evidence of additional defects in IPC-18 due to the reconstruction of the pore system in IWW. However, the NMR spectrum of the final IPC-18 still contains quite intense (25-30 % of overall area fraction) Q 3 signal (Fig. 2f). IPC-18 synthesis (IPC-18P reassembly) temperature lower than optimal can be the reason for relatively high content of silanols in final material. In addition, some Ge atoms are still remained in IPC-18 framework as Si(OGe)(OSi) 3 domains that give a peak, which can overlap with the Q 3 peak.
Structural differences and similarities between the parent IWW and its derivative can be directly observed through spherical aberration corrected (C s-corrected) Scanning Transmission Electron Microscopy (STEM) analysis of the respective materials (Fig. 3). Images collected from the top view of the layers (Fig. 3a, c) along the [100] orientation show the same member rings and the same pore arrangements, thus confirming their preservation during the VPT treatment, as well as topotactic condensation. Along this orientation, the D4Rs that are affected by the hydrolytic process are not visible in these top views. Therefore, after collecting data from the side view along [010], the interlayer distances significantly decreased due to changes in D4Rs which have now turned into S4Rs. The projected measured distance between "T" atoms for a D4R is~3.1 Å. Thus, initially, a decrease of~3 Å should be expected. The distances between layers directly measured on high-resolution images are 12.23 Å for IWW and 9.11 Å for IPC-18, in very good agreement with the   transformation of the linking units. Selected area electron diffraction (SAED) and/or Fourier diffractograms (FD) measurements were also collected along different orientations (see Fig. 3 insets), which clearly showed, first, the good crystallinity of the materials, with no additional structural defects, and, second, the formation of S4Rs. The analysis of HRTEM images collected in three projections of IWW crystals ( Supplementary Fig. 4a-c) and IPC-18 ( Supplementary Fig. 4d-f) confirms the aforementioned conclusion about the direction in which the anisotropic structural transformation occurs. For one projection [001], d-spacing is the same (12.0 Å) for both parent and daughter zeolites. The major difference between IWW and IPC-18 is the connectivity between layers ( Supplementary Fig. 4).
Mechanism of VPT zeolite rearrangement. Characterisation of the starting and final zeolite structures is relatively simple because they are crystalline materials with well-ordered frameworks. In contrast to the relatively stable IPC-18P layered material, any intermediate structures formed during the IWW-to-IPC-18P transformation (hydrolysis) or during the IPC-18P-to-IPC-18 condensation (reassembly) are hardly detectable without in situ methods as both processes are completed within 1 h. We used in situ XRD to trace the diffraction peaks with different Miller indices hkl-related to different directions in the structures of the treated materials (hkl (l ≠ 0) interlayer peaks, with varying positions over time, and intralayer hk0 reflections, which should maintain the positions) to shed the light on the whole mechanism of IWW-to-IPC-18 transformations (a general scheme is presented in Fig. 4e). In addition, we performed quantitative lattice analysis based on XRD patterns and followed the local state of the Ge atoms by in situ XANES during the VPT rearrangement to identify the relationship between the nature of leaving atoms and framework features at each stage of the structural transformation. Immediately after starting the treatment (<1 min), the positions of 00l reflections shifted to higher angles (Fig. 4a), indicating a decrease in d-spacing caused by disassembly of D4R connecting units. Surprisingly, the positions and intensities of hk0 reflections (110 and 400 are the most representative, Fig. 4a) are also significantly changed already at the first step of hydrolysis (0-2 min under conditions used). Analysis of the correlation between unit cell parameters and treatment time provided further insight into the processes that occur during the first step of the VPT (Fig. 4c). Not only the interlayer c parameter but all lattice constants (a, b, c) considerably change during this step, indicating partial distortion of the structure of IWW layers (illustrated on Fig. 4e). Simultaneously, the β angle, which is indicative of the shift of layers relative to each other, remains unchanged in the first step. Thus, at this stage, the layers become flexible but are still stacked, most likely due to ionic interactions and hydrogen bonding between interlayer species-primary products of hydrolysis. In situ XANES (Fig. 4d) revealed the major change in the nature of Ge atoms during this period since the Ge K-edge (E 0 ) steeply and linearly decreased over time at the same stage (0-2 min) because most Ge-OT IV (T IV = Si or Ge) bonds predominantly located at D4R units were broken. The intensity of some XRD reflections (related to both inter-and intralayer planes, e.g., 110 and 111) decrease in the first step of the VPT transformation, suggesting a complete loss of structural ordering in the respective directions of the lattice.
During the second step of hydrolysis (from 2 to 10-12 min), the aforementioned reflections appear at slightly shifted positions, indicating at least a partial recovery of the respective framework ordering. At first sight, the positions of the other hkl (l ≠ 0) reflections continue the gradual right-shift in the second stage of the rearrangement (Fig. 4a). However, the behaviour of the respective lattice constant (c) is more complex (Fig. 4c): after a series of minor changes (at~½, 2, 3 min when c parameter is either slightly decreased or increased), c starts to decrease, with a clear exponential decay trend (Fig. 4c, blue line) until the end of the procedure. The parameters a and b monotonically decrease during the second step of hydrolysis, whereas β peaks around 90.7°, indicating a slight, transient shift of the layers relative to each other during the structural reorganisation of the layers. These processes are accompanied by a linear change in E 0 as a function of time in XANES spectra, which is slower than the first step (Fig. 4d). Additional peaks around E = 11120 and 11125 eV appear in XANES spectra at this stage ( Supplementary Fig. 7), indicating an accumulation of hydrolysed Ge species extracted from the framework. Overall transformations during this period can be attributed to a slow hydrolysis of isolated Ge-O bonds remaining after primary decomposition of the D4Rs (major fraction) and extraction of Ge atoms located in zeolite layers (minor fraction because the layers are almost purely siliceous). Completion of the D4R disassembly and removal of intercalated species allows the partial recovery of layer organization as the parameters b and β return to the original values after 12 mins of treatment.
At the final, third, step of the VPT rearrangement (12 + min), most XRD reflections maintain their positions and intensities, and only a few hkl (l ≠ 0) reflections, such as 001 and 313, show changes in their intensity after 12 min (Fig. 4a), similarly to the lattice parameters (Fig. 4c) and to the Ge state, which remains virtually unchanged during the final stage of hydrolysis (Fig. 4d).
Heating the intermediate IPC-18P material up to~500 K has no effect on the positions or intensities of intralayer hk0 reflections, highlighting a complete layer rearrangement during the hydrolysis step of the VPT transformation (Fig. 4b). In contrast, the position of the 001 reflection in relation to the interlayer distance gradually shifts in the range of T = RT − 350 K, being almost constant at temperatures of 350-450 K. Further increasing the temperature results in a gradual collapse of the IPC-18 zeolite framework, accompanied by a shift in all XRD reflections at T = 450-650 K and in a decrease of their intensities at T >650 K (Fig. 4b). Based on our in situ XRD and XANES data, we propose the following mechanism of IWW-to-IPC-18 transformation, as illustrated in Fig. 4e  (3) slow rearrangement of the species remaining in the interlayer space until reaching the equilibrium state at RT (>10-12 min) (4) layers condensation (reassembly) at slightly increased temperatures (up to 350 K) with no effect on the structure of the layers.  The VPT rearrangement technique was developed and first exemplified for the ADOR transformation of different germanosilicate zeolites characterised by the unidirectional location of Geenriched D4R units (i.e., UTL, UOV, IWW). The VPT approach overcomes the limitations of the classical ADOR approach for the rational design of zeolites. Although VPT treatment of UTL and UOV germanosilicates produces known ADORable IPC-7 and IPC-12 zeolites of high crystallinity, VPT rearrangement of the IWW framework enabled preparation and structure refinement of the previously predicted, yet thus far inaccessible, IPC-18 zeolite via existing synthesis approaches. In situ XRD and XANES studies on the VPT rearrangement of IWW revealed the acidinduced reorganisation of not only interlayer but also intralayer framework domains, resulting in the temporary disordering of the structure followed by its reconstruction to a well-ordered material with a different framework from the parent zeolite. The successful application of VPT rearrangement to germanosilicate zeolites of different topologies highlights the potential of this technique for the 3D-3D transformation of crystalline materials with labile frameworks collapsing upon contact with the solvent (e.g., ordered organic-free germanates or some metal-organic frameworks). In combination with other methods for post-synthesis alteration of 3D frameworks, this approach makes it possible to manipulate the structures of anisotropically labile materials.
Synthesis of IWW zeolite. IWW zeolite samples was performed according to Ref. 22  VPT treatment. In total, 0.1 g of calcined zeolite was placed in the reactor containing polytetrafluoroethylene (PTFE) membrane over 10 ml of 12 M hydrochloric acid solution (p. a., Penta) at 298 K for τ = 5 min-48 h. Caution: the VPT procedure uses concentrated HCl. Standard laboratory practice for dealing with highly corrosive liquids (including appropriate personal protective equipment) should be used.
Basic characterisation. The structure and crystallinity of zeolites were determined by powder X-ray diffraction using a Bruker AXS D8 Advance diffractometer with Cu Kα radiation in Bragg-Brentano geometry. The chemical composition was determined on an ICP/OES (ThermoScientific iCAP 7000). Adsorption experiments were performed using nitrogen as a probe molecule. Ad-/desorption isotherms were measured using an ASAP 2020 (Micrometeritics) static volumetric apparatus at liquid nitrogen temperature (−196°C). Before the sorption measurements, all samples were degassed with a turbomolecular pump at 523 K for 4 h. The micropore volume was estimated by application of the t-plot method, and the pore size distribution was obtained using the NLDFT method. The morphology of the samples was determined under a scanning electron microscope (SEM) TES-CAN Vega. Solid-state 29 Si NMR spectra were recorded on an Agilent DD2 500WB spectrometer at a resonance frequency of 99.30 MHz. MAS NMR measurements were carried performed with a commercial 3.2 mm triple resonance MAS probe. The chemical shifts of 29 Si are referenced to tetramethylsilane at 0 ppm.
Synchrotron XRD. The in situ synchrotron XRD was performed at the BL14B and BL19U2 stations of the Shanghai Synchrotron Radiation Facility (SSRF). The wavelength of incident X-ray photons was 0.1240 nm, and the diffraction patterns were recorded using two-dimensional X-ray detectors. To monitor the hydrolysis process in real-time, the sample was sealed in a homemade cell with Kapton windows. Diffraction patterns from the sample were continuously recorded with an exposure time of 10 s per frame and with an interval of 2 s between adjacent frames. The beginning of reaction was controlled by an electric pump, which injected concentrated hydrochloric acid into the cell to initiate the hydrolysis. For the in situ heating experiment, the sample was placed on a heating stage. The temperature was increased from room temperature to 723 K at a rate of 1.5 K per minute and was then kept at 723 K for 1 h. Diffraction patterns were continuously recorded at an exposure time of 100 s per frame and at an interval of 20 s between adjacent frames.
Quantitatively analysis for the lattice parameters. For the XRD patterns obtained during the in situ hydrolysis step of the VPT transformation, the time-dependent peak positions of reflections (001), (320), (400), (120), (311), (310) Fig. 4 In situ data supporting the assumed mechanism of VPT. a In situ XRD demonstrating changes during the hydrolysis of the parent IWW zeolite; b in situ XRD collected during thermal treatment of IPC-18P; c variation of lattice parameters (lengths in Å and angle in degree) as a function of time during the VPT treatment; d evolution of the primary line extracted from in situ XANES spectra representing changes in Ge state during hydrolysis steps; e proposed mechanism (Si and Ge atoms are not distinguished for clarity and only structural, but not chemical, changes are shown). Full in situ XRD data, lattice parameters and XANES profiles are shown in Supplementary Figs. 5, 6 and 7, respectively. Analysis of the variation of lattice parameters and position of primary line in XANES spectra allows to distinguish several steps during VPT. Unexpectedly, the disassembly of Ge-rich D4Rs was accompanied by partial disorder of the layers, which become more crystalline on the second step of hydrolysis. The topology of resulting layers only slightly differs from that of starting zeolite. Following deintercalation of interlayer debris and condensation allows the formation of new IPC-18 zeolite and (211), as shown in Fig. SI-6, are used to calculate the quantitative lattice information with the fsolve function in Matlab, assuming a monoclinic crystalline structure for the materials.
Synchrotron XANES. Ge K edge XANES measurements were performed at the beamline 1W1B of the Beijing Synchrotron Radiation Facility (BSRF). The spectra were collected in the transmission Q-XAFS mode to monitor the hydrolysis reaction, with a rate of 37.6 s per spectrum. During the experiment, the sample was sealed in the homemade cell, as described above.
Electron microscopy. Electron microscopy analyses were performed in a coldFEG JEOL Grand ARM 300 electron microscope. The microscope was equipped with a double corrector from JEOL assuring a spatial resolution of 0.7 Å when operated at 300 kV. Before observation, the samples were deeply crushed using a mortar and pestle and dispersed in ethanol. A few drops of the suspension were deposited on a holey carbon copper microgrids. For imaging, both Annular Dark Field and Annular Bright Field detectors were used simultaneously, acquiring data in both detectors.