Zeolites with Continuously Tuneable Porosity

Zeolites are important materials whose utility in industry depends on the nature of their porous structure. Control over microporosity is therefore a vitally important target. Unfortunately, traditional methods for controlling porosity, in particular the use of organic structure-directing agents, are relatively coarse and provide almost no opportunity to tune the porosity as required. Here we show how zeolites with a continuously tuneable surface area and micropore volume over a wide range can be prepared. This means that a particular surface area or micropore volume can be precisely tuned. The range of porosity we can target covers the whole range of useful zeolite porosity: from small pores consisting of 8-rings all the way to extra-large pores consisting of 14-rings.

PXRD patterns of UTL samples that have been hydrolysed at room temperature for 5 minutes to form IPC--1P. The peak positions are not dependent on the acidity except for the TMA--OH treated sample. However, the PXRD patterns of this sample do become similar to all the others after treatment for a longer time. The measured Si/Ge ratio for these samples is in the range 25--37 and shows no trend with conditions. Figure S2 Evolution of the position of the 200 reflection on treatment with 12 M HCl for 5 minutes at RT and then 1, 2, 8, 16 and 24 hours at 85 °C. The final peak position is moved to smaller 2θ values, indicating an increase in the interlayer spacing.

S2.1 Laboratory X--ray Diffraction
X--ray diffraction data were collected on a Panalytical Empyrean diffractometer using Cu Kα1 radiation operating in reflection geometry.  Figure S4 --PXRD patterns of Ge--UTL after hydrolysis at 95 o C using the indicated concentration of HCl, measured before (top) and after calcination of the products (bottom). Figure S5 -Correlation of the position of the d--spacing of the 200 reflection with molarity after hydrolysis and calcination.

S2.2 Synchrotron Powder X--ray diffraction of IPC--6
Powder diffraction data were collected on the Materials Science Beamline at the Swiss Light Source (SLS) in Villigen, Switzerland 1 (Table S1). Rietveld refinement was performed using the XRS−82 suite of programmes, 2 and the structure drawings were produced using CrystalMaker. 3 The profile plots were prepared with the programme ppp14. 4 The diffraction pattern of IPC--6 was indexed with the programme TREOR 5 implemented in the software CMPR 6 on a monoclinic unit cell with parameters a = 20.78, b = 13.89, c = 12.30 Å, b = 101.48º. The systematic absences were indicative of body centring (space groups I1, I2 or I2/m). This unit cell was transformed to the equivalent C--centred one (a = 21.94 Å, b = 13.89 Å, c = 12.30 Å, b = 112º) to facilitate comparison with the UTL--type framework 7 , from which it had been prepared, and the other UTL--related frameworks (PCR and OKO). The b and c parameters are similar to those of UTL, which is consistent with the UTL layers remaining intact after the disassembly and reassembly process. A Le Bail fit (no structural model) of the diffraction pattern was performed in the space group C2/m, and the zero correction, sample displacement, half widths and unit cell parameters refined. The h00 reflections were found to be significantly broadened, so anisotropic line broadening in that direction was also refined. From the d200 spacing, which is indicative of the interlayer spacing, it was thought that the framework structure of IPC--6 might be a mixture of the OKO and PCR framework types, composed of UTL layers stacked along a, connected either through a single--four ring (A) as in OKO or directly through an oxygen atom (B) as in PCR. Alternation of these two (AB) would result in a d200 spacing (10.18 Å) that lies between those of the OKO and PCR frameworks (11.37 Å and 9.13 Å, respectively). However, this model has a primitive unit cell. Inspection of the TEM images showed that the layers are not stacked in an ordered fashion, i.e. the sequence of interlayer distances observed in the electron microscopy images does not correspond to a regular (AB)(AB)(AB) sequence. The layer connections are arranged almost randomly (e.g. (BA)(AB)(BA)(AB)(AB)(BA)(AB)), but correspond to approximately 50% (AB) and 50% (BA) sequences. In other words, on average, a unit cell contains 0.5(AB) and 0.5(BA) connections. The average length of a does not change, because it is the same for both arrangements. However, the stacking disorder does cause the reflections along the h00 direction to broaden. A model for the average framework structure was built by superimposing two structures: one with the (AB) connection sequence (P2/m) and the second with the same structure shifted by (1/2,1/2,0) to give a (BA) connection sequence (also P2/m). This model is C--centred, as the indexing indicated, and describes a 50:50 mixture of the two connections. The occupancy parameters for all framework atoms were set to 0.5. With this relatively simple model, the positions of the superimposed layers allow either connection to occur between layers, and in this way a switch from (AB) to (BA) stacking can occur. However, the model implies that there are no extended connections of just one type. Geometric restraints were imposed on the bond distances and angles of the framework atoms corresponding to one of the two superimposed structures, and the second was constrained to retain a strict (1/2,1/2,0) shift. Neutral scattering factors were employed for all atoms. The crystallographic data are given in Table S2, the atomic parameters in Table S3 and the profile fit for this model is shown in Figure S3. The bond distances and angles are reasonable for an all--silica framework (Si−O distances between 1.58 Å and 1.63 Å, Si−O−Si angles between 133.8º and 177.4º and O−Si−O angles between 105.6º and 111.5º). The difference electron density map calculated for the final structure was featureless, indicating that the electron density is well--described by the model. However, there are still some significant differences apparent in the profile plot ( Figure S3). This was first thought to be an effect of a preferred orientation of the thin plate--like crystallites during the data collection, but no clear direction for the differences could be established. Therefore, it is assumed that the differences result from the fact that the diffraction data is complicated by a significant amount stacking defects. A more rigorous simulation of the stacking disorder is planned. 0.5820(3) 0.6652(8) 0.8938(8) a : Numbers in parentheses are the estimated 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. Isotropic displacement parameters (B) were set to 1 for Si and 2 for O. Figure S6. Observed (black), calculated (red) and difference (blue) profiles for the Rietveld refinement of IPC--6. The profiles in the inset have been scaled up by a factor of 6 to show more detail. Reflection positions are marked as vertical bars.

S2.2 Electron microscopy
The size, shape and elemental composition of zeolite crystals were examined by scanning electron microscopy (SEM) using a Jeol JSM 5600 SEM equipped with an EDX system. For the measurement, crystals were coated with a thin platinum layer by sputtering in vacuum chamber of a BAL--TEC SCD--050. The microstructures were investigated using high resolution transmission electron microscopy (HRTEM) on a Jeol JEM--2011 electron microscope operating at an accelerating voltage of 200 kV. The Jeol JEM--2011 electron microscope is equipped with an Oxford Link ISIS SemiSTEM EDX system, which was used for confirming chemical compositions of the samples. The HRTEM images were recorded using a Gatan 794 CCD camera. The camera length, sample position and magnification were calibrated using standard gold film methods. Figure S7 TEM images of particles of IPC--6 after hydrolysis using 1.5 M HCl and further calcination. Similar to the image of a different sample shown in the main paper (Figure 4), this image shows clearly the staged nature of the material with two distinct interlayer spacings of 1.1 nm (A) and 0.9 nm (B), together with occasional stacking faults (dotted yellow lines). Red arrows indicate the direction of the IPC--6 unit cell, and the inset shows the FFT of the image in 1, with the diffuse streak indicating the presence of the stacking disorder.
A " B " B " A " A " A " B " B " A " A " B " B " A " B "A " B " B " B " A " B " B " B " B " A " A " B " B " A " A " Figure S8. TEM image of the IPC--7 sample made using 5M HCl, followed by calcination. The images show very disordered materials with an average d--spacing of 1.24 nm. Local larger d--spacings (up to 1.4 nm) are detectable as marked (A). Local defects can be seen as mismatched layers (B) and where two layers merge into one, (C). Other local disordering is marked by circles. Figure S9. TEM image of the interlayer spacing in IPC--7 showing spacings of 1.4 nm (corresponding to D4R units between the layers, A) and 1.1 nm corresponding to S4R units between the layers, B. The image also shows a defect where the two different layer spacings coexist between two layers in different parts of the crystallite. Figure S10. Proposed average structure of IPC--7 showing 50% of the layers linked by D4R units and 50% linked by S4R units. This model is consistent with the TEM image in Figure S8 and with the diffraction pattern, which gives an indexed monoclinic unit cell of a = 12. 336 Å, b = 14.391 Å, c = 26.129 Å b = 99.41° The structural model consists of a two dimensional arrangement of 14--, 12--and 10-ring channels that is consistent with the adsorption experiments.

A"
A" B" B" B" B" Figure S11 Adsorption isotherms for all samples after calcination. Figure S12 -Correlation of BET surface area and micropore volume with molarity of calcined samples.

S2.5 Solid--state NMR
Solid-state 29 Si NMR spectra were recorded using a Bruker Avance III spectrometer equipped with a 9.4 T wide-bore superconducting magnet. Samples were packed into 4 mm zirconia rotors and rotated at the magic angle at a rate of 10 kHz. Quantitative spectra were recorded with signal averaging for between 440 and 592 transients for the different samples with a recycle interval of 120 s. A pulse of flip angle 90° was applied, with ω 1 /2π ≈ 90 kHz. Q 3 Si species were identified by cross polarisation (CP) from 1 H with a spin lock duration of 5 ms and high-power (ω 1 /2π ≈ 80 kHz) TPPM 1 H decoupling during acquisition. For the spectra shown in Figure S several experiment with different spin lock durations are shown by the different coloured spectra. For the CP experiments, signal averaging was carried out for 1024 transients with a recycle interval of 4 s. Figure S13 The direct (black) and { 1 H} 29 Si CP spectra (red, blue and green) for hydrolysed Ge--UTL sample recovered after 5 minutes room temperature hydrolysis with 0.1 M HCl (left) and 12 M HCl (right). Figure S14 The quantitative MAS 29 Si (black) and { 1 H} 29 Si CP spectra (red) spectra for samples hydrolysed at 1.5 M (left) and 6 M (right) after 17 hours at 95 °C. The Q3:Q4 ratios are 25.0:75.0 and 23.8:76.2 respectively for the 1.5 M and 6 M samples respectively, indicating the quite similar nature of the lamellar intermediates Spectra have been normalised, since the absolute intensities of the MAS spectra vary depending on number of transients averaged.