The Assembly-Disassembly-Organization-Reassembly Mechanism for 3D-2D-3D Transformation of Germanosilicate IWW Zeolite

Hydrolysis of germanosilicate zeolites with the IWW structure shows two different outcomes depending on the composition of the starting materials. Ge-rich IWW (Si/Ge=3.1) is disassembled into a layered material (IPC-5P), which can be reassembled into an almost pure silica IWW on treatment with diethoxydimethylsilane. Ge-poor IWW (Si/Ge=6.4) is not completely disassembled on hydrolysis, but retains some 3D connectivity. This structure can be reassembled into IWW by incorporation of Al to fill the defects left when the Ge is removed.


Experimental section 3
Characterization 4 Structure Analysis 5

List of Tables and Figures:
Table S1 Examples of conditions used for hydrolysis of Ge-rich IWW (Si/Ge ratio 3.1)

Table S2
Textural properties of IWW samples valuated from argon and nitrogen sorption       The observed (black), calculated (red) and difference (blue) profiles for the Rietveld refinement of Ge-rich IWW. The data in the inset have been scaled up by a factor of 6 to show more detail at high angles. The tick mark indicate the positions of the reflections.

Fig. S4
Background correction for the powder diffraction pattern of the Ge-poor IWW. It is an interpolation of points selected by eye and adjusted manually during the course of the refinement. The humps under the two peaks at ca. 6 º correspond to an unidentified impurity that was included in the background so it does not disturb the refinement.

Fig. S9
Adsorption isotherms of nitrogen measured at 77K (left) and argon measured at 87K (right). Parent calcined IWW zeolite (Si/Ge 3.1) (1), hydrolysed and calcined IPC-5 (2), surfactant-treated and calcined IPC-5SW calc (3), and surfactant-treated-intercalated and calcined, restored IWW (4). The open points denote desorption. Although the N2 and Ar isotherms show the same trend for each sample, the values of their BET surface area, micropore volume and others can slightly differ, which is caused by different parameters of both adsorptives (see results in Table S2).

Experimental section
The preparation of IWW followed the procedure in the reference 1 Table S1. Typically, 0.1g of solid was stirred with 100-250 ml of acid solution for 16-48 hours. The solid was centrifuged and dried at 60-100 °C overnight. A part of the hydrolysed material was calcined at 550 °C for 6 hours.
Intercalation of hydrolysed IWW Swelling: Swelling was performed using a mixture of C16TMA-Cl (25%) and C16TMA-OH (25%) with the ratio for 1 g of hydrolysed IWW (IPC-5P) 6.7ml of C16TMA-Cl (25%) and 13.3ml of C16TMA-OH (25%). C16TMA-OH (25%) was prepared from 25% solution of C16TMA-Cl by ion-exchange using resin AG 1-X8 (Bio-Rad). The slurry was stirred in the ice-cooled bath for 7 hours. The product was isolated by centrifugation, properly washed with water and dried at 60° C. A part of the swollen sample was further calcined at 650 °C for 6 hours.
Intercalation of diethoxydimethylsilane (DEDMS): The hydrolysed IPC-5P or the surfactanttreated IPC-5P was treated with DEDMS in 1M HNO3 solution (for 1 g zeolite, 0.5 g of DEDMS and 10 g of acid). The reaction was carried out in the autoclave at 175 °C for 16 hours. The solid was filtered, washed with water and calcined at 650 °C for 6 hours.
Reaction with octylamine: Hydrolysed IPC-5P was stirred with octylamine (for 1g used 6.5g of octylamine) at 75 °C for 16 hours. The solid was separated by centrifugation, dried at 40 °C and finaly calcined at 650 °C for 6 hours.
Alumination: Alumination was performed by reacting 0.2g of AlCl3 with 0.1g of hydrolysed IWW in 50 ml of 0.05M HCl solution under reflux for 3 days. The solid was isolated by centrifugation, washed with water, dried overnight and calcined at 550 °C for 6 hours.

Characterisation
The structure and crystallinity of zeolites under study were determined by X-ray powder diffraction using a Bruker AXS D8 Advance diffractometer equipped with a graphite monochromator and a position sensitive detector Våntec-1 using CuKα radiation in Bragg-Brentano geometry or STOE diffractometer under CuKα radiation in Debye-Scherrer mode.
For the structure analysis of IWW powder diffraction data were collected on the Materials Science-Powder X04SA Beamline at the Swiss Light Source (SLS) at Paul Scherrer Institut in Villigen, Switzerland 2 . Data collection details are given in Table S3.The Rietveld refinement was performed using the program package XRS−82 3 , and the structure drawings were produced using the program CrystalMaker 4 . The plots with observed, calculated and differences patterns were prepared with the programme ppp14 5 .
The chemical analysis was obtained by the Energy Dispersive X-Ray Spectroscopy (EDX) on a Jeol JSM 5600 instrument. Solid-state NMR spectra were acquired using Bruker Avance III 600 MHz spectrometer equipped with a widebore 14.1 T magnet. Powdered samples were packed into conventional 4 mm ZrO2 rotors. 29 Si MAS NMR spectra were obtained using single-pulse experiments, with a recycle interval of 180 s. The MAS rate was 10 kHz. For the 27 Al MAS NMR the MAS rate of 14 kHz was used. The IR spectra were collected by a Perkin Elmer Spectrum GX IR spectrometer from the wave number range of 400-4000 cm-1. Adsorption isotherms of nitrogen (at -196˚C) and argon (at -186˚C) were measured on a Micromeritics ASAP 2020 static volumetric instrument. In order to attain sufficient accuracy in the accumulation of the adsorption data, the ASAP 2020 was equipped with pressure transducers covering the 133 Pa, 1.33 kPa and 133 kPa ranges. Prior to the sorption experiments, samples were outgassed at 110 °C under turbomolecular pump vacuum until the residual pressure of 0.5 Pa was obtained. After further heating at 110 °C for 1 h the temperature was increased (1 °C/min) until the temperature 300 °C was achieved. This temperature was maintained for 6 h. The specific surface area (SBET) was evaluated by BET method 6 using adsorption data in the p/p0 range of 0.05-0.20. The adsorbed amount at relative pressure p/p0 = 0.98 reflects the total adsorption capacity (Vtot). For adsorption isotherms of nitrogen, the t-plot method 7 was applied to determine the volume of micropores (Vmic). For adsorption isotherms of argon, the DFT algorithm (using standard Micromeritics software for cylindrical pores for Argon on Oxides at -186 °C) was used to calculate the volume of micropores (Vmic) and pore-size distribution for pores less than 5 nm in diameter. The volume and pore-size distribution of mesopores with the size from 5 to 20 nm was calculated from the desorption branch of the isotherm using BJH method 8 with Halsey equation.

Structure analysis
The diffraction pattern of Ge-rich IWW (Si/Ge molar ratio 3.6) and Ge-poor IWW (Si/Ge molar ratio 6.4) correspond to zeolite ITQ-22 (structure type IWW). They were indexed with the programme TREOR 9 implemented in the software CMPR 10 in an orthorhombic unit cell with parameters a = 42.01, b = 12.95, c = 12.64 Å, for Ge-rich IWW, and a = 42.01, b = 12.96, c = 12.61 Å, for Ge-poor IWW. The systematic absences were indicative of the extinction symbol Pba− (space groups Pbam and Pba2). Structure refinement of the diffraction data of Ge-rich IWW was started in Pbam, in which the structure was initially solved1, using those coordinates as the initial model structure. Geometric restraints were imposed on the bond distances and angles of the framework atoms throughout the refinement, and their weight relative to the diffraction pattern was decreased as the refinement progressed. Neutral scattering factors were employed for all atoms. All the tetrahedrally coordinated atoms (T atoms) were initially refined as pure Si. At the first stages of the refinement, the T−O bond distances corresponding to the T sites that form the D4R were significantly longer than those of the rest of T sites, which was already an indication of the presence of germanium, since the Ge−O distance (1.76 Å) is longer than the Si−O distance (1.61 Å). The electron density clouds on T sites T1-T4 observed in the difference Fourier map (Fig. S2) showed unequivocally that Ge was located in the D4R. The occupancy factors of the four mixed Si/Ge sites refined to 89 %, 88 %, 27 % and 44 % respectively, which corresponds to Si/Ge= 4.7. This led to a significant improvement of the R values from RF = 0.122 and Rwp = 0.500 (Rexp = 0.018), in the pure silicon structure, to RF = 0.111 and Rwp = 0.388.
The mirror plane perpendicular to the c-axis divides the d4R into two symmetrically equivalent halves, and therefore the percentage of Ge in four of the T sites is necessarily equal to that of their four respective symmetrically equivalent T sites. To check if the Si/Ge fraction is actually different in each T positions of the D4R, the mirror plane was removed, reducing the symmetry to Pba2. This resulted in an increase in the number of T sites in the asymmetric unit from 16 to 28. After relaxing the structure using a distance-least-squares procedure, refinement of the occupancy parameters of the T sites (again initially modelled as pure Si) showed that Ge is located exclusively at the eight T sites (T1-T8) that form the D4R in Pba2. It was reasoned that Ge might tend to avoid Ge−O−Ge bonds, and therefore, to get a set of starting values for the Ge occupancy factors in the D4R, those of the four Ge atoms that are not connected through a Ge−O−Ge bond were constrained to have the same value. However, when they were subsequently refined freely, they yield values that indicate the existence of these bonds, as will be discussed later. Refinement of the framework coordinates without restraints led to T−O distances for T1-T8 that are consistent with Ge population, i.e. the T sites with higher percentage of Ge have longer T−O distances. At this stage the difference Fourier map still showed some electron density clouds on mixed Si/Ge atoms, suggesting that Ge population may have to be higher; however they did not refine to higher values. A closer inspection of the map revealed the presence of some "holes" close to the electron density peaks, which indicated that the position of these Ge atoms was not correct yet. Bearing in mind that the Ge−O distance is significantly longer than the Si−O distance, these discrepancies in the Fourier map could mean that the position of the T atom is slightly different depending on whether there is Si or Ge on that particular case, since a single (average) T position is being used. A few more cycles of refinement of the structure led to a cleaner Fourier map (the peaks and holes were just 1.5 electron/Å 3 high or deep). Two nonframework-atom positions were found, and they were included as water molecules (modelled as oxygen atoms). Ow(1) is located at the intersection between the 8-ring and 10ring channel (3.5 Ow(1) per unit cell), within hydrogen-bonding distance from oxygen O19 (Ow(1)−O19 = 2.23 Å). Ow(2) lies in the 12-ring channel (4 Ow(2) per unit cell, 2 per channel), making two close contacts with two framework oxygens (Ow(2)−O1= 2.54 Å and Ow(2)−O37=3.09 Å). This makes a total of 7.5 water molecules per unit cell, probably adsorbed after calcination.
All framework atoms were refined isotropically, and to keep the number of parameters to a minimum, only one displacement parameter was refined for oxygen and one for the T atoms. The thermal parameters of Si/Ge and O refined to values close to 0.01 and 0.02, respectively, so they kept fixed at those values. Geometric restraints were used on the bond distances and angles of the framework atoms through the refinement, but their relative weight with respect to the powder diffraction data was reduced during the course of the refinement. For the mixed Si/Ge sites, the bond distances restraints were adjusted to reflect the amount of Ge at each site. No evidence of Ge with a coordination higher than 4 was observed. The crystallographic data are given in Table S4, atomic parameters in Table S5 and the fit of the profile calculated for this model to the experimental data is shown in Fig. S3. Selected interatomic distances and bond angles are given in Table S6 and Table S7. A similar procedure was followed for the Rietveld refinement of the diffraction data of Gepoor IWW, with lower Ge content. The pattern contains two broad peaks with low intensity that do not belong to IWW, which were included in the background (Fig. S4) so they do not disturb the refinement. Similarly to what was found for Ge-rich IWW, the longer T−O bond distances for T1-T8 and the electron density clouds on these T sites were indicative of the presence of Ge in the d4R's. Their occupancy factors refined to a Si/Ge ratio of ca. 6, in good agreement with that found in the chemical analysis. No extraframework species were found. The crystallographic data are given in Table S8, atomic parameters in Table S9 and the fit of the profile calculated for this model to the experimental data is shown in Fig. S5. Selected interatomic distances and bond angles are given in Table S10 and Table S11.
In both samples the refined T−O distances are consistent with Ge population, being the T−O average distances for T sites in the d4R (T1-T8, see Fig. S1) longer than the typical Si−O bond length, and longer for T sites with higher Ge occupancy. The T−O−T bond angles are, on average, ca. 141° if at least one of the T sites is a mixed Si/Ge position, and ca. 151° otherwise. In both samples Ge atoms are located exclusively at the d4R's. This is the preferential location for Ge found in ITQ-22 (Si/Ge = 3.2) when it was first synthesized and solved1, although in that case small amounts of Ge (occupancies less than 11 %) were found in 6 other T sites, and the substitution of Si by Ge in the T sites of the D4R is around 55 % (4 T sites because it was solved and refined in Pbam). A significantly higher percentage of Ge was found in the D4R in Ge-rich IWW (see Table S5). Interestingly, one of the 4-rings in the D4R (T1-T4 atoms) is formed almost exclusively by Ge (> 90% Ge at each T site), whereas ca. 50 % of the Si has been replaced by Ge in the other 4-ring (T5-T8 atoms in Fig. S1). The eight Ge-containing T sites yield a total of 0.94+0.94+0.90+0.98+0.67+0.57+0.69+0.30 = 6.0 Ge sites in the asymmetric unit (i.e. 6.0 × 4 = 24.0 Ge per unit cell), which gives a Si/Ge = 3.7 nicely corresponding to Si/Ge = 3.6 found by the chemical analysis. In Ge-poor IWW, there is a more even distribution of Ge atoms over the 8 T sites of the d4R (see Table S9), being T1 and T4 the ones with the highest percentage of Ge. These two T sites are in the same 4-ring that is almost entirely Ge−O−Ge in Ge-rich IWW. There are 15.8 Ge atoms per unit cell, yielding a Si/Ge = 6.1, in a good agreement with the chemical analysis (Si/Ge = 6.4).
The composition of the D4R´s in Ge-rich and Ge-poor IWW samples is [6Ge,2Si] and [4Ge,4Si], respectively, which involves the existence of Ge−O−Ge bonds in the D4R's. They have also been observed in IWW in Ref. 1, with Ge occupancy factors bigger than 50 % in six of the eight T sites. The incorporation of Ge is known to favor the crystallization of zeolites that contain d4R's because the increase of the average T−O bond distance and decrease of the T−O−T angle relax the structure and makes it more stable. However, large distortions of the d4R's caused by a high Ge content are expected to decrease the overall stability of the zeolite. Indeed, a linear correlation has been observed between the relative total energy versus the number of Ge−O−Ge linkages present in the corresponding distribution. This means that Ge atoms, from an energetic point of view, tend to locate far from each other 11 .
For Ge-rich IWW, a complete 4-ring can be expected to be dissolved (the almost purely Ge one) upon treatment in acid solution, as well as part of the other 4-ring composed by mixed T positions Si/Ge. By contrast, for Ge-poor IWW all T sites in the D4R are mixed Si/Ge positions and consequently this cage would be partially removed during the acid treatment, i.e. the extraction of some of the 8 T atoms rather than disassembly of a complete 4-ring would occur.
Germanium has been found to progressively incorporate into the D4R in ITQ-17 12 (BEC) to reach four Ge atoms per cage (50 % replacement of Si by Ge in T1, the only T site in the d4R), and a further increase of the Ge content favors the substitution of Si atoms at crystallographic sites not in the d4R 12 . This is in agreement with 19 F MAS NMR spectroscopy results, which provide evidence about the composition of the distribution of Ge among the D4R's, since a low field shift of the 19 F signal assigned to F-in the D4R has been observed with increasing n values [nGe,(8-n)Si] in that D4R ,12,13 . Besides, computational calculations on ITQ-17 showed that at high Ge content (Si:Ge 1.7 : 1), where the D4R's contain 4 Ge and 4 Si atoms as shown by the Rietveld refinement results, configurations with Ge at the opposite sides of the d4R are preferred to avoid the formation of Ge−O−Ge bonds. Configurations with five Ge atoms in one D4R where found to be higher in energy, and consequently, less probable, because they involve three Ge−O−Ge linkages 12 . However, a different behavior has been observed for zeolite ITQ-2111. In this case the T1 position in the D4R is also preferred, and T2 is only occupied for higher Ge content. However, T3 sites remain purely siliceous even for Si/Ge ratios as low as 1.  16,17,18 From the synthesis and structure analysis of AST-type framework with different Si/Ge ratio, including the Si-and Ge-end members, Wang et al. 19 concluded that Ge is first inserted in the D4R's ordering in a Lowenstein avoidance rule-like pattern, being the 6:2, 4:4 and 2:6 Si:Ge ratios preferred, apart from all Si and all Ge D4R's.
The two samples studied in this work show T−T distances within the d4R longer than the 3.1 Å typically found for purely siliceous zeolites like octadecasil (AST) 20 , ITQ-7 (ISV) 21 or ITQ-17 (BEC) 12 as expected from the Ge's larger ionic radii. The distances for Ge-poor IWW are ca. 3.26 Å except those involving one of the T sites with the highest Ge population (T1 and T4), which are up to 3.34 Å. An increase in Ge content produces a lengthening of the T−T distances in Ge-rich IWW (up to 3.37 Å). They are slightly longer than the 3.2-Å distances of Ge-containing zeolites ITQ-22 (Si/Ge= 4)1 and ITQ-24 (Si/Ge= 7.6, IWR) 22 and the pure Ge AST 19 . Refinement of the structures in the space group Pba2 allowed differences between the two 4-rings (formed by T atoms T1-T4 and T5-T8, see Fig. S1) that compose the D4R to come to light, since the mirror plane that made them equivalent in Pbam has been removed. Apart from an independent occupancy factor for Ge in each of the eight T sites, subtle structural differences were also apparent, because the T−T distances that were equivalent in Pbam are not forced to be equal in Pba2. This is more evident for the D4R's face diagonal T1−T4 compared to T2−T3, which differ in ca. 0.20 Å for both samples, and for T6−T3 compared to T5−T4 and T7−T4 compared to T3−T8 in Ge-rich IWW, which also differ in 0.20 Å.   T is used to designate the mixed Si/Ge T sites. The multiplicity of all atoms is 1. b : Numbers in parentheses are the standard deviations (esd's) in the units of the least significant digit given. Each restraint was given a weight equivalent to the reciprocal of its esd. Values without an esd were not refined. c : The % of Si at each T site is the difference to 100 %. The rest of the T sites have been found to be purely siliceous. d : Occupancy factors of the water molecules.             Table S1.   IWW (4). The open points denote desorption. Although the N2 and Ar isotherms show the same trend for each sample, the values of their BET surface area, micropore volume and others can slightly differ, which is caused by different parameters of both adsorptives (see results in Table S2).