Yannick Lorgouillouxab,
Mathias Dodina,
Enrico Mugnaiolic,
Claire Marichala,
Philippe Caulleta,
Nicolas Batsd,
Ute Kolbc and
Jean-Louis Paillaud*a
aEquipe Matériaux à Porosité Contrôlée (MPC), Institut de Science des Matériaux de Mulhouse (IS2M), UMR CNRS 7361 – UHA, ENSCMu, 3b rue Alfred Werner, 68093 Mulhouse Cedex, France. E-mail: jean-louis.paillaud@uha.fr
bLMCPA, EA 2443, UVHC, Pôle Universitaire de Maubeuge, Boulevard Charles de Gaulle, 59600 Maubeuge, France
cJohannes Gutenberg-Universität Mainz, Institut für Physikalische Chemie, Jakob-Welder-Weg 11, 55128 Mainz, Germany
dIFP Energies nouvelles, Direction Catalyse et Séparation, Rond-point de l'échangeur de Solaize – BP 3, 69360 Solaize, France
First published on 8th April 2014
The synthesis and the structure of IM-17, a new germanosilicate with a novel zeolitic topology, prepared hydrothermally with decamethonium as the organic structure directing agent, are reported. The structure of calcined and partially rehydrated IM-17 of chemical formula per unit cell |(H2O)14.4|[Si136.50Ge39.50O352] was solved ab initio using electron diffraction ADT data in the acentric Amm2 (setting Cm2m) space group and refined by the Rietveld method. This new zeolite framework type contains a 3D pore system made of intersecting 12, 10 and 8-ring channels.
In our search for new large pore zeolites in the (Si,Ge) system, we decided to engage decamethonium (Fig. 1) as OSDA by varying several synthesis parameters. This diquaternary ammonium cation has a quite long chain and is known to lead to the aluminosilicates Nu-8729 (NES) and ZSM-1230 (MTW), which are 10 and 12MR zeolites, respectively. Note that the magnesium-containing aluminophosphate DAF-1 (DFO), a 12, 10 and 8MR molecular sieve is also hydrothermally synthesised with the same OSDA.31 Our investigations led to the discovery of a new germanosilicate zeolite framework containing d4r units, IM-17,32 that was obtained among known zeolites. Here we present the synthesis, characterization and structure of this new molecular sieve. The structure was determined by a combination of automated electron diffraction tomography (ADT) and Rietveld refinement on powder X-ray diffraction data (PXRD). Electrons have a strong interaction with matter and can be easily focused in nanoprobes. These characteristics allow to perform single-crystal diffraction on single nanometric particles. Since the late 1980s, electron diffraction has been used together with other techniques for structure determination of porous materials. In particular, electron diffraction assisted PXRD for space group determination and peak deconvolution.33–37 Early encouraging examples also demonstrated that it is in principle possible to perform ab initio structure solution based on electron diffraction data alone.38,39 Nevertheless, due to dynamical scattering and experimental difficulties in data acquisition, only a partial structure can be normally achieved using conventional in-zone electron diffraction data, and there is a major concern in recognizing the correct phase scheme among the several proposals automatically determined by direct methods.40 In the last years, ADT method has allowed to collect more complete and off-zone (less dynamical) electron diffraction data and to reduce the exposure time and the beam damage on the sample. Electron diffraction is therefore becoming a routine method for structure determination of porous materials and diverse nanocrystalline phases that have been solved and refined with a combination of ADT and PXRD methods.41–44
The PXRD pattern of calcined IM-17 was indexed in a C-centred orthorhombic unit cell with the parameters a = 39.109(8) Å, b = 22.242(5) Å, c = 12.680(4) Å and V = 11029.6(45) Å3 by the Louër's DICVOL9145 indexing routine of the STOE WinXPow program package.46 The observed systematic extinctions suggested C222 (#21), Cmm2 (#35), C2mm (Amm2 (#38)) and Cmmm (#65) as possible space groups. For the structure determination the extracted intensities of the reflections up to 60°(2θ) were taken. From this set of data, the structure of calcined IM-17 was mainly solved by direct methods with the highest symmetry, i.e. in space group Cmmm using the EXPO2009 software.47 Some oxygen and two missing disordered framework silicon T sites were deduced from model building using Cerius2 environment together with difference Fourier maps during the Rietveld refinement.48 The Rietveld49 refinement was performed using the GSAS package50 on the complete pattern via the interface EXPGUI.51 The atomic coordinates of the whole framework atoms (i.e. T and O atoms with T = Si or Ge) were used as the starting model. All the atoms were refined isotropically. Soft restraints were placed on the bond lengths and angles of the framework (T–O = 1.68(4) or 1.62(4) Å and O–T–O = 109.5(30)°). Initially, for the Rietveld refinement, on each T atom site, a silicon atom and a germanium atom were both placed at the same position with the sum of their occupancy factors constrained to be 1. Then, as the refining progressed, some T sites were set as purely siliceous. After refinement of the framework, successive calculated Fourier difference maps revealed a scattering density inside the void volume attributed to the presence of adsorbed water molecules. Then, one oxygen atom has been placed on these positions and refined (position and occupancy factor). Further details on crystallographic and Rietveld refinement data are given in ESI, Table S1 and Fig. S1.† The final atomic parameters, selected bond distance and bond angles are listed in ESI, Tables S2 and S3.†
In order to verify if the space group Cmmm was the true one, a study by ADT (Automated electron Diffraction Tomography) has been undertaken. Thus, calcined sample no. 14 (Table 1) was dispersed in ethanol, sonified and sprayed on a carbon-coated cupper grid using a modified UIS250v Hielscher sonifier.52 ADT acquisition was performed with a FEI TECNAI F30 S-TWIN transmission electron microscope equipped with field emission gun and working at 300 kV, using the semi-automatic module described by Kolb et al.53 Crystal position was tracked in microprobe STEM mode and nano electron diffraction (NED) patterns were acquired sequentially in steps of 1°. STEM images were collected by a FISHIONE high angular annular dark field (HAADF) detector and acquired by Emispec ESVision software. NED patterns were taken with a CCD camera (16-bit GATAN ULTRASCAN4000, 4096 × 4096 pixels) and acquired by the Gatan Digital Micrograph software. A condenser aperture of 10 μm and mild illumination setting were used in order to produce a semi-parallel beam of 70 nm in diameter on the sample. ADT was coupled with electron beam precession in order to improve reflection intensity integration.52 Precession was performed with a NANOMEGAS DigitStar device keeping the precession angle at 1.2°.
Samples | Molar synthesis mixture compositions | Materials | ||||
---|---|---|---|---|---|---|
SiO2 | GeO2 | OSDA | HF | H2O | ||
a All syntheses were performed at 170 °C for 14 days.b Silica source is TEOS.c Silica source is Aerosil 200.d No solid synthesized.e Very small amount of MFI-type zeolite synthesized.f GeO2 dense phase argutite.g Pure phase but lower yield. | ||||||
1b | 0.6 | 0.4 | 0.25 | 0.5 | 8 | BEC |
2b | 0.6 | 0.4 | 0.5 | 0.5 | 8 | AST |
3c | 0 | 1 | 0.25 | 0 | 20 | d |
4c | 0.2 | 0.8 | 0.25 | 0 | 20 | IM-17 + MFI + 122B |
5c | 0.4 | 0.6 | 0.25 | 0 | 20 | IM-17 + MFI + 122B |
6c | 0.5 | 0.5 | 0.25 | 0 | 20 | IM-17 + 122B + εMFIe |
7c | 0.6 | 0.4 | 0.25 | 0 | 20 | IM-17 + 122B + εMFIe |
8c | 0.75 | 0.25 | 0.25 | 0 | 20 | 122B |
9c | 0.9 | 0.1 | 0.25 | 0 | 20 | 122B |
10c | 1 | 0 | 0.25 | 0 | 20 | 122B |
11c | 0.6 | 0.4 | 0.5 | 0 | 20 | IM-17 + MFI |
12c | 0.6 | 0.4 | 0.5 | 0 | 40 | IM-17 + 122B |
13c | 0.6 | 0.4 | 0.125 | 0 | 10 | MFI + argf + IM-17 |
14c | 0.6 | 0.4 | 0.25 | 0 | 10 | IM-17 |
15c | 0.6 | 0.4 | 0.25 | 0 | 5 | IM-17 |
16c | 0.6 | 0.4 | 0.125 | 0 | 5 | IM-17 |
17c | 0.8 | 0.2 | 0.25 | 0 | 3 | IM-17 |
14bisb | 0.6 | 0.4 | 0.25 | 0 | 10 | IM-17g |
ADT data elaboration, including three-dimensional diffraction volume reconstruction and visualization, cell parameter determination and reflection intensity extraction, was performed with ADT3D software, according with the algorithms developed by Kolb et al.52–55 Ab initio structure solution was performed by direct methods implemented in the program SIR2011,56 assuming the kinematic approximation I ≈ Fhkl2. The structure model obtained from ADT, (see IM-17 structure paragraph of the results and discussion part below), was refined by Rietveld refinement using GSAS as described above.
Samples | SiO2 | GeO2 | Synthesis time (day) | Stirringb | Materials |
---|---|---|---|---|---|
a All samples synthesized at 170 °C, with OSDA/(Si + Ge) = 0.25, H2O/(Si + Ge) = 20 and Aerosil 200 as the silica source.b Autoclave stirred at 15 rpm during thermal treatment.c Very small amount of MFI-type zeolite synthesized. | |||||
7 | 0.6 | 0.4 | 14 | No | IM-17 + 122B + εMFIc |
18 | 0.6 | 0.4 | 3 | No | 122B |
19 | 0 | 1 | 7 | No | 122B + IM-17 |
20 | 0.2 | 0.8 | 14 | Yes | 122B + IM-17 |
8 | 0.75 | 0.25 | 14 | No | 122B |
21 | 0.75 | 0.25 | 30 | No | 122B |
22 | 0.75 | 0.25 | 14 | Yes | 122B |
![]() | ||
Fig. 2 XRD patterns of samples 8 and 14 showing the very broad peaks of the 122B material and the sharp peaks corresponding to pure IM-17, respectively. |
As shown in Table 1, the three obtained phases were present in sample 7, the first sample that was synthesized in hydroxide medium. A parametric synthesis study was thus conducted in order to obtain a pure IM-17 sample. A selection of the most representative results is given in Table 1. Modification of the Si/Ge ratio had a great effect on the synthesized materials (samples 3 to 10). When the synthesis mixture was Si rich, i.e. when Si/Ge ≥ 3 and even in purely siliceous system (samples 8, 9 and 10), samples only contained the poorly-crystalline 122B material. IM-17 was the main product for intermediate Si/Ge ratios: 1 ≤ Si/Ge ≤ 1.5 (samples 6 and 7). When the GeO2 content was further increased, an MFI-type zeolite crystallized in a quite large amount along with IM-17 and small quantities of 122B.
For the syntheses leading to samples 11 to 16, the water and decamethonium amounts were varied. The results of these syntheses showed that IM-17 could be obtained as a pure material only when the water amount was decreased (samples 14, 15 and 16). However, the H2O/OSDA ratio also seems very important, as a concentrated synthesis mixture with a low amount of decamethonium (H2O/OSDA = 80, sample 13) did not lead to pure IM-17, but to a mixture containing also MFI-type zeolite and the GeO2 dense phase argutite. To synthesize pure IM-17, the reaction mixture thus needs water and OSDA contents such that H2O/(Si + Ge) ≤ 10 and H2O/OSDA ≤ 40. Sample 17 showed that with highly concentrated medium, the Si/Ge ratio in the mixture could be increased up to 4 and still lead to the pure IM-17 material, which shows that these media very poor in water are favourable for the synthesis of IM-17.
A few syntheses not described in detail here were conducted with TEOS as the silica source and/or with seeds, but they did not allow getting better results. Indeed, as reported in Table 1, sample 14bis prepared with TEOS instead of Aerosil 200 as silica source used for sample 14 leads also to IM-17 but with a yield three times lower as it was already reported for IM-16 (UOS).63
The influence of the synthesis conditions was also studied (Table 2). Samples 18, 19 and 7 showed that for the same synthesis mixture composition, the 122B material was obtained first (3 days). After longer time (7 days), IM-17 was also visible on the XRD pattern of the product, and this material became predominant after 14 days. It is thus possible that the 122B material transformed into IM-17 when the synthesis time increased, and 122B might be an early-stage disordered IM-17 analogue. Presence of a large amount of 122B solid in sample 20 may have been due to a much higher nucleation rate of 122B under stirring. As sample 8 contained only 122B, samples 21 and 22 were prepared to see if the same mixture composition could lead to IM-17 under other synthesis conditions, but no improvement was observed.
Bulk chemical analysis of sample 14, 15 and 14bis gave a mean Si/Ge molar ratio of 3.5, 2.8 and 5.3, respectively. These results prove also the influence of the water content and the chemical source of silica on the chemical composition of IM-17. According to the quantitative 1H liquid NMR analysis performed on sample 14, the OSDA is intact inside the structure of IM-17 (in ESI, Fig. S2†). A total weight loss of about 11.9% is observed on the TGA curve (not shown) of as-made sample 14. Accordingly, from quantitative 1H liquid NMR, TGA and chemical analyses and the present structural study on the calcined same sample, the chemical formula per unit cell of as-made IM-17 (sample 14) is close to |(C16H38N2)5.5 (OH−)11 (H2O)2|[Si136.50Ge39.50O352].
According to the preliminary structural study, 14 inequivalent crystallographic sites are present but are not resolved on the spectrum due to a distribution of environments. In order to enhance the signal corresponding to silanols, a 1H–29Si CPMAS experiment was performed. This experiment based on magnetization transfer from 1H to 29Si thanks to dipolar interaction is shown on Fig. 4b. A resonance is observed at −101.5 ppm corresponding to silanol species. Surprisingly, despite a very long acquisition time (34156 scans) a low signal to noise ratio was obtained suggesting that the sample contains only a very small amount of silanol groups. Decreasing the contact time to 1 ms (spectrum not shown) did not improve the signal to noise ratio. Furthermore, this resonance can not be detected on the 29Si MAS NMR spectrum (Fig. 4a) that was recorded in quantitative conditions suggesting that indeed only traces of silanol are present in this sample. As a consequence, the interrupted framework model is not possible.
![]() | ||
Fig. 5 Three-dimensional ADT reconstructed diffraction volume for IM-17: (a) view along a*, (b) view along b*, (c) view along c*. Space group Cm2m. |
No further extinction was detected in the ADT reconstructions, leading to extinction symbol ‘C---’ and related space groups C222, C2mm, Cm2m, Cmm2 and Cmmm. The short direction of growth of the platelets is always associated with the main crystallographic direction c*. As a consequence, this direction stands always vertical and falls in the missing cone of conventional ADT acquisitions.66,67 Because 00l reflections could not be sampled, extinction symbol ‘C--21’ could not be excluded, and therefore space group C2221 was also taken into account for structure solution.
Structure solution of IM-17 was obtained ab initio on the basis of electron diffraction intensities extracted from the best ADT acquisition. Structure refinement converged in acentric space group Cm2m (Amm2) with R = 27.19% (ESI, Table S4†), while it did not converge in all other considered space groups. Remarkably, in contrast with the initial PXRD study (in space group Cmmm), ab initio structure solution was not possible assuming the centrosymmetric space group Cmmm. The fact that the solution converged in this space group, whereas it did not converge in all the other considered space groups, suggests that this is the real symmetry of the structure and that IM-17 nano-domains crystallize coherently in Cm2m (Amm2) symmetry.
The potential map of the solution could be interpreted as a framework of 25 silicon and 32 oxygen positions. The 23 strongest peaks in the potential maps corresponded nicely to 12 equivalent silicon or germanium positions of the model refined in Cmmm space group (Si,Ge1-12 in ESI, Table S2†). The remaining 2 silicon atoms are associated with significantly weaker potentials, listed as positions Si17 and Si18 in ESI, Table S5.† These last two positions correspond to the two disordered silicon atoms in the Cmmm model (Si13 and Si 14 in ESI, Table S2†). It is important to note that these two Si atoms associated with disorder in the Cmmm model correspond in fact to weak and smeared peaks in the potential map obtained ab initio.
This is a strong hint that this part of the structure is affected by a certain disorder, probably associated with alternating nano-domains that crystallize with opposite orientation. The ADT acquisition performed with a nano-probe of 70 nm was able to sample a particle that was “mostly” made by a coherent domain (in ESI, Fig. S7†) allowing ab initio solution with a single crystal approximation.
The presence of one or more smaller domains differently oriented, whose reflections geometrically coincide with the ones of the main domain, produced an increment of the structural residual R values of the solution (ESI, Table S4†) and smeared the peaks associated with the part of the structure that more prominently violate centrosymmetric geometry. This kind of disorder was recognized and described by electron diffraction in the case of several microporous materials.44,68,69
Consequently, the starting model used for the definitive Rietveld refinement of IM-17 was then the complete structure generated in sub-space groups Amm2 after the PXRD study which corresponds to the incomplete ADT model. The final Rietveld refinement is plotted on Fig. 6 and details on crystallographic and Rietveld refinement data are given in Table 3. The atomic parameters are listed in Table S6.† In Table S7† are reported selected bond distances and bond angles, respectively. A CIF file is also available in ESI.†
a After the Rietveld refinement, the total number of extra framework oxygen atoms (Ow1–Ow5) is about 19.2 (see Table S6†) per unit cell i.e. 2.4 per asymmetric unit. However, the refinement did not take into account the scattering power of the hydrogen atoms of each water molecule. Consequently, the true number of water molecules is about 25% lower than the total number of extra framework oxygen atoms refined, i.e. about 1.8 water molecules per asymmetric unit as reported in the chemical formula of the table.b The definition of these residual values are given in the GSAS manual.50 | |
---|---|
Chemical formula per asymmetric unit | |(H2O)1.8|[Ge4.94Si17.06O44]a |
Space group | Amm2 |
λ (Å) | 1.5406 |
a (Å) | 12.679(2) |
b (Å) | 22.217(4) |
c (Å) | 39.058(6) |
V (Å3) | 11![]() |
Z | 8 |
Temperature (K) | 293 |
Number of data points | 5057 |
Number of observed reflexions | 3066 |
Number of structural parameters | 240 |
Number of profile parameters | 12 |
Number of restraints (bonds, angles) | 251 (101, 150) |
Number of constraints | 133 |
Rpb | 0.0532 |
wRpb | 0.0739 |
wRexpb | 0.0207 |
RFb | 0.0410 |
RF2b | 0.0488 |
χ2b | 9.64 |
Largest difference peak and hole (ē Å−3) | 0.761, −0.786 |
The asymmetric unit of calcined IM-17 contains 25 crystallographically non equivalent T-sites and 60 bridging oxygen atoms (ESI, Table S6 and Fig. S8†). The Rietveld refinement indicates that 14 T-sites are purely siliceous (Si10 and Si13 correspond to sites Si17 and Si18 of the ADT model or Si13 and Si14 of the Cmmm model), one is fully occupied by germanium (Ge24) and 10 are Si, Ge mixed sites. The refined chemical composition of the inorganic framework per unit cell is [Si136.5Ge39.50O352] (Si/Ge ≈ 3.46), which closely matches that obtained by chemical analysis (Si/Ge ≈ 3.5 for sample no. 14).
The refined total amount of adsorbed water molecules converged to about 14.4 molecules per unit cell (see Table 3 and S6†). They are distributed on 5 non equivalent crystallographic sites (Table S6†). Ow3 and Ow5 interact strongly with the framework (Fig. S9 and Table S7†), the shortest Ow–O distances being 2.19(12) and 2.33(12) Å, respectively.
The structure of the new zeolite IM-17 can be described in a manner already used for IM-1663 (UOS) and IM-2070 (UWY) which consists in using convenient tertiary building units which are made of composite building units or CBUs. Thus, as illustrated on Fig. 7a, T49-units containing 49 T atoms (T = Si, Ge) and consisting in 6 fused CBUs (…ats-mor-stf-mel-bre-mel…) sharing 4- and 5-rings with four additional T atoms connected to the mor CBU are stacked along the c-axis to form columns, each T49-unit is rotated by 180° relative to its neighbours according to a twofold-axis parallel to the c-axis (Fig. 7a). After connection via edges in the b,c-plane (Fig. 7b), each column being a mirror image relative to its neighbours, a periodic building unit or PerBU layer is formed (Fig. 7c). This connection mode generates new mel and pairs of fused lau and mtw CBUs. The full structure is obtained by connecting these layer PerBUs through the 4-rings of the lau and mtw CBUs via bridging oxygen atoms in the b,c-plane along the a-axis as illustrated in Fig. 7d. Three types of d4r CBUs are thus created which differ in their Si/Ge molar ratio. It is worthy to note that, as expected, 83.4% of the germanium atoms are concentrated at the vertices of these d4r CBUs. A tile representation of the IM-17 framework is given on Fig. 8.
The three-dimensional pore system of IM-17 is constituted of large 12MRs and small 8MRs pores along [100] (Fig. 9a) that intersect medium 10MRs pores along [001] (Fig. 9b). Due to the presence of intercalated d4r CBUs, no channel is visible along the b-axis (Fig. 9c). This channel system, for a better visualisation, has been computed with ZEOMICS71 and its representation by spheres and cylinders is reported on Fig. 10.
![]() | ||
Fig. 9 Projections of the IM-17 structure along (a) [100], (b) [010] and (c) [001]. The oxygen atoms have been omitted for clarity. |
![]() | ||
Fig. 10 The 3D channel system of IM-17 as computed by ZEOMICS.71 The orange cylinders are large pore, the yellow ones are medium pores and the blue spheres are medium cages. The small 8MRs pores have been omitted for clarity. |
As illustrated on Fig. 11, two types of large cages connect the large and medium channels. The [4464104122] cage is crossed by the 12MRs channel along [100] and two orthogonal 10MRs channels along [001] and the small 8MRs channel passes through the [44586682104] cage along [100] which is also crossed by the same 10MRs channels. Assuming a van der Waals radius for oxygen of 1.35 Å, the free diameter of the tri-axial ellipsoidal [4464104122] and [42586682104] cages are 5.3 × 9.9 × 15.7 Å and 6.1 × 7.8 × 15.8 Å, respectively. The free diameters of the elliptical 12MRs and 10MRs channels are 6.0 × 7.7 Å and 4.6 × 6.0 Å, respectively.
Finally, the measured pore volume of calcined IM-17 from N2 adsorption at 77 K (used kinetic diameter for N2 = 3.86 Å) is 0.22 cm3 g−1 (in ESI, Fig. S10†). It corresponds to a large-pore zeolite and to the computed accessible volume value of 0.222 cm3 g−1 by ZEOMICS71 from a probe diameter of 4 Å. These micropores volumes are intermediate to those obtained for the germanosilicates IM-2070 (0.16 cm3 g−1) and IM-1218 (0.26 cm3 g−1). The coordination sequences and vertex symbols calculated from TOTOPOL72 for IM-17 are listed in ESI, Table S8.†
Footnote |
† Electronic supplementary information (ESI) available: Tables and figures, a cif file. See DOI: 10.1039/c4ra01383b. Further details of the crystal structure investigation can be also obtained from the Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany, (fax: +49 7247 808 666; e-mail: crystaldata@-fiz.karlsruhe.de) on quoting the depository number CSD-427372. |
This journal is © The Royal Society of Chemistry 2014 |