Ana
Palčić
*a,
Bartłomiej M.
Szyja
b,
Maja
Mičetić
c,
Tomaž
Čendak
d,
Mariame
Akouche
e,
Krunoslav
Juraić
c,
Marija
Čargonja
f,
Darko
Mekterović
f,
Vitomir
Vušak
g and
Valentin
Valtchev
e
aDivision of Materials Chemistry, Ruđer Bošković Institute, Bijenička 54, 10000 Zagreb, Croatia. E-mail: ana.palcic@irb.hr
bFaculty of Chemistry, Wrocław University of Science and Technology, Gdańska 7/9, 50-344 Wrocław, Poland
cDivision of Materials Physics, Ruđer Bošković Institute, Bijenička 54, 10000 Zagreb, Croatia
dDepartment of Inorganic Chemistry and Technology, National Institute of Chemistry, Hajdrihova 19, SI-1001 Ljubljana, Slovenia
eNormandie Univ, ENSICAEN, UNICAEN, CNRS, Laboratoire Catalyse et Spectrochimie, 6 Marechal Juin, 14050 Caen, France
fUniversity of Rijeka, Department of Physics, Ulica Radmile Matejčić 2, 51000 Rijeka, Croatia
gPliva Croatia Ltd., Prilaz baruna Filipovića 25, 10000 Zagreb, Croatia
First published on 8th July 2019
A typical zeolite synthesis reaction mixture involves a framework atom source, a solvent, a mineraliser, and a structure directing agent. However, these components often have a dual function, and their impact on the nucleation/crystallization process is difficult to establish straightforwardly. The present study deals with the structure directing effect of zinc on the preparation of zincosilicate RUB-17 (RSN-type material). First, an attempt was made to introduce several other heteroatoms (Al, B, Ge, Mg, Li, and Zr) into the RSN framework under the same reaction conditions as those for Zn-RSN. Second, different Zn sources have been employed in order to study their impact on RUB-17 intermediates and the final product properties. The obtained series of samples were analysed by XRD, XRF, GISAXS, SEM, TEM, and TG as well as by Raman, NMR, infrared and solid state UV/VIS spectroscopy. The physicochemical analysis of the obtained solids was complemented by DFT modelling. The results show that a RSN-type material can be solely prepared in the Zn-containing systems, where zinc plays a structure directing role besides its function as a framework cation. Namely, the lov-unit, a basic building unit of the RSN framework, is formed at the very beginning of the crystallization no matter the zinc source employed. Yet, the differences in the crystallization of different systems and relatively long time needed for full transformation of the precursor indicate that there is some other limiting step that accounts for the formation of zeolites.
In order to enhance the efficacy and activity of zeolite materials, further studies should be conducted. Preparation and detailed characterization of new materials and improving the properties of “classical” zeolites (materials of desired size, morphology, acidity, ion-exchange capacity, hierarchization, etc.) are the two principal axes of research in zeolite science.10–16 Understanding the mechanisms of formation of a particular material could enable the preparation of materials having the same structure but variable chemical composition and consequently different properties. Herein, the zincosilicate material RUB-17, which is so far the only known RSN-type zeolite material, is investigated.17 Its structure is built of 3-, 4-, 5-, and 6-membered rings forming a tridimensional system of channels 9(3.3 × 4.4 Å) × 9(3.1 × 4.3 Å) × 8(3.4 × 4.1 Å). 3MR frameworks are of particular importance since it has been theoretically predicted that the structures composed of small rings (3MRs and 4MRs) should exhibit lower framework density.7 More open microporous frameworks offer new possibilities to overcome diffusion limitations and process bulkier molecules. Moreover, the 9-membered ring containing materials are expected to have interesting performance in separation processes.
It should be stressed that in total there are 24 3MR containing zeolitic framework types, many of them having only one representative type material. Some of them are not silicates (e.g. gallogermanates) and still there is only one known 3MR aluminosilicate zeolite material (MEI-type).18 Furthermore, heteroatoms such as Be, Zn, Li and Ge were incorporated into the 3MR silicate materials (OSB-1, VPI-9, RUB-29, ITQ-40), where Be and Zn are present in many zeolitic materials, particularly natural zeolites (LOV-, NAB- and VSV-type materials). In addition, most of the Be and Zn zeolites are built of a lov-unit which comprises two 3MRs sharing one vertex (spiro center).18 For this study zincosilicate zeolitic material RUB-17 was chosen due to the availability and non-toxicity of zinc. The results of the investigation of the impact of the employed zinc source on the formation of the RUB-17 material and its properties are presented. The data will provide new findings on the processes affecting the synthesis of the 3MR-containing zincosilicate materials and present a basis for the design of zeolite materials with novel functions.
Molar oxide composition of the starting reaction mixture | Starting chemical | t c (days) | Final product |
---|---|---|---|
1SiO2:0.1ZnO:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | ZnO | 11 | RUB-17 |
1SiO2:0.05Al2O3:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | Powder Al | 12 | Feldspare type aluminosilicates |
1SiO2:0.05Al2O3:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | Al(OH)3 | 12 | |
1SiO2:0.05Al2O3:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | NaAlO2 | 12 | |
1SiO2:0.05B2O3:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | B2O3 | 20 | No precipitation |
1SiO2:0.05B2O3:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | B(OH)3 | 12 | Boron phyllosilicate |
1SiO2:0.1GeO2:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | GeO2 | 12 | Mixture of germanium silicates |
1SiO2:0.1MgO:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | MgO | 12 | Layered material |
1SiO2:0.1ZrO2:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | ZrOCl2·8H2O | 25 | Amorphous |
1SiO2:0.1ZrO2:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | Zr(IV) isopropoxide | 20 | 3D zirconium silicate |
1SiO2:0.05Li2O:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | LiOH·H2O | 60 | No precipitation |
1.1SiO2:0.5NaOH:0.5KOH:0.08TEAOH:44H2O | TEOS | 30 | No precipitation |
Thermogravimetric analysis of the solid samples was performed with a Setaram Setsys TGA instrument. The heating rate was 5 °C min−1 and the samples were heated up to 800 °C in air flow. FT-IR spectra of the solid samples were recorded on a Bruker TENSOR 37 spectrometer using KBr pellets, in the range from 400 to 4000 cm−1. Raman spectra were acquired by employing a PD-LD LS2 laser source (784.2 nm, 450 mW) with optical fibers. The Raman probe (BWTEK) was coupled with a Maya 2000 Pro spectrometer (Oceanoptics). A Shimadzu UV-3600 UV/Vis/NIR spectrometer was used to obtain solid state UV-Vis spectra of the studied samples mixed with BaSO4 powder supplied by Wako Pure Chemical Industries Ltd. The weight percentage of the samples was 1–5 wt%. The spectra were recorded at 20 °C in the range from 1200 to 200 nm at a resolution of 1 nm. The instrument was linked to an integrating sphere. The background was recorded using pure BaSO4.
The Zn/Si ratio of the solid samples was determined by employing an Energy Dispersive X-Ray Fluorescence (ED-XRF) spectrometer. Powder samples were pressed into pellets with a 7 mm diameter and attached to a carbon tape prior to analysis. A rhodium X-ray tube (X-Ray Optical Systems, model X-Beam) was used for ionisation under 50 kV and 0.04 mA. Each sample was irradiated for 30 min with a beam collimated to a diameter of 2 mm and perpendicular to the sample. Characteristic X-rays were collected with a silicon-drift detector (Amptek X-123SDD, resolution of 145 eV for Fe-Kα line) positioned at 45° with respect to the incoming beam. The measurements were performed in air, with 10 mm of source to sample distance and 25 mm of target to detector distance. The resulting X-ray spectra were fitted using the Quantitative X-ray Analysis System (QXAS) software developed by the International Atomic Energy Agency (IAEA) in order to obtain characteristic line intensities.21 The Si/Zn ratios were calculated from the ratio of these intensities. The detailed description of the calculation procedure is presented in the ESI.†
1H–29Si CPMAS NMR spectra were recorded on a 600 MHz Varian NMR System equipped with a 3.2 mm Varian MAS probehead. Larmor frequency for 29Si nuclei was 119.088 MHz. Chemical shifts of the 29Si nuclei are reported relative to the signal of 29Si in tetramethylsilane. All measurements were performed at a sample rotation frequency of 20 kHz, with a relaxation delay of 2 s and CP contact time of 5 s. For the crystalline RUB-17 samples prepared using ZIF-8 and Zn(NO3)2·6H2O as zinc source chemicals 10000 scans were accumulated with an acquisition time of 40 ms and 80 kHz XiX decoupling. For the rest of the samples, 28000 scans were accumulated with an acquisition time of 15 ms and 90 kHz XiX decoupling.
Structural properties of the materials including the size and arrangement of nanovoids were analysed by grazing incidence small angle X-ray scattering (GISAXS). GISAXS measurements were carried out at Elettra synchrotron Trieste (SAXS beamline), using an X-ray beam with an energy of 8 keV and a 2D Dectris photon detector (Pilatus 1M). The measured data were simulated and analysed using GisaxStudio software.22,23 The powder materials were pressed into a thin tile before the GISAXS measurements.
In order to obtain a meaningful result, several parallel runs need to be executed to account for random initial velocities. For the needs of this work 20 runs have been carried out for each system; however due to the small differences in the models and similar constraints imposed – all runs have been observed to occur identically, and they led to the same results. The structure representing a computational model has been obtained from the crystal structure recovered from a .cif file available at the International Zeolite Association website.18 It consists of a 3-member ring, with none (pure silica) or one heteroatom (Al, Zn) present. It is basically half of the symmetric lov-unit. The position of the heteroatom is at the vertices of the lov-unit, not in the spiro position. The building unit was embedded in the framework represented by 4 other T sites. In order to maintain the geometry of the cluster constructed this way, the Si atoms of the framework have been saturated with hydrides, and the positions of these hydrides were kept frozen during the simulations. The model is shown in Fig. S1.† Si atoms are represented by yellow spheres, O – smaller red spheres, H – smallest white spheres. The heteroatom is represented by a pink sphere.
The model represents the fully formed building unit, at the end of the process. In other words, the reaction that is carried out in this computational work is reversed and simulates the decomposition of the ring. This approach allows for better control of the process because it always leads to the desired product, i.e. the initial geometry in the MD simulation. At the same time, the transition state structures are the same as if the process was reversed, thus allowing estimation of the energetics of the reaction. This is “de facto” a kinetic approach – under the assumption that the thermodynamics of the process is always the same because from the same reactants (open ring) the same products (closed ring + water) are obtained. The thermodynamic stability of different structures (pure silica and two with heteroatoms) cannot be compared directly because of different compositions of those systems, but for each process, ΔH can be calculated and compared with that of other systems. For each of the systems – pure silica, Al heteroatom, and Zn heteroatom, two kinds of simulations were carried out: one with the water cleaving a bond between the heteroatom and Si atom in the same position or the Si atom in the top vertex of the 3-ring. This approach allowed identifying whether one of these bonds has a different activation barrier – thus whether it is easier to cleave. This approach also allows assessing if the heteroatom is built in a vacancy (“nest”) already formed by Si sites, or actively drives the process towards this particular structure.
Raman spectroscopy is a commonly used method to identify the presence of various structural features within zeolite (silicate) frameworks. It is especially convenient for detecting the small ring building units since three-membered rings exhibit bands in the region between 550 and 620 cm−1, four-membered rings from 470 to 530 cm−1 and five-membered rings from 370 to 430 cm−1. Bands in the range 290–410 cm−1 are attributed to six-membered rings while larger rings have bands below 290 cm−1.31,32 The Raman spectra of the solids prepared by employing various T atom sources are depicted in Fig. S3.† The three-membered rings characteristic of the RSN-type material were not detected in the samples prepared in Al, B and Mg containing systems. Aluminium-containing products exhibit nearly identical spectra with main bands below 550 cm−1. In the spectrum of the B(OH)3 synthesized product, there is a band with a maximum at 555 cm−1, while the MgO product has only one band at 683 cm−1. The ZrOCl2·8H2O product spectrum does not have well defined bands in the framework ring region. There is a band at 602 cm−1 in the spectrum of GeO2-containing systems. This peak could be attributed to the 3MRs in the Ge product which is not a surprise considering a large number of germanium-containing zeolite materials with 3MRs in the structure.18 Nevertheless, this system did not yield an RSN-type material. Furthermore, when Zr(IV) isopropoxide is employed, the end product features a band at 586 cm−1. This kind of material comprises 3MRs as deduced on the grounds of the reported structural data.29 However, in this silicate material, the zirconium has octahedral coordination and cannot be considered as a zeolitic type material.
Clearly, under given conditions only zinc used as a potential zeolite framework T atom together with silicon resulted in the formation of a zeolitic three-membered ring containing a RSN-type material. Thus, our attempt to replace the Zn in the RSN-type structure was not successful, which reveals its important structure-directing effect. In the following sections, we will present the results of the detailed study of the impact of the starting Zn compound on the properties of RUB-17 precursors and the final product.
Fig. 1 XRD patterns (A), Raman spectra (B) and NMR spectra (C) of the final products obtained after 11 days of hydrothermal treatment using different Zn sources. |
In the studied reaction systems, the solid and liquid phases are visibly separated as the solid is precipitated at the bottom of the autoclave. Moreover, the solids are highly aggregated forming a hard monolith. The SEM micrographs of selected representative samples reveal different crystal sizes and morphology depending on the starting Zn chemical (Fig. S5†). In the samples synthesised from ZnO and ZnCl2 the crystals have a wide size distribution ranging between 100 and 1500 nm. Nearly all of the crystals grow together building up complex aggregates of different sizes. In ZnO and ZnCl2 systems, the crystal faces are well-defined, and the edges are sharp. Their morphology is prismatic, ranging from short to long prisms. The crystals in the ZIF-8 containing system differ significantly. There are two types of crystals – very small (∼130 nm) with rounded edges and larger (∼800 nm) having sharper edges and more pronounced crystal faces. These results suggest that there is a difference in the crystal growth processes in the studied series of reaction systems.
Fig. 2 Powder XRD patterns of the reaction intermediate taken after 8 h of hydrothermal treatment of reaction mixtures prepared from various Zn sources at 180 °C. |
The chemical analysis of the solid phases was performed by XRF in order to provide information on zinc distribution in the precursors. The molar Si/Zn ratios of the precursors taken after 8 h of the hydrothermal treatment (Table 2) indicate a higher content of Zn in the solid phases of the systems prepared with ZIF-8, Zn(OAc)2·2H2O and Zn(acac)2·H2O than in the water-soluble Zn salts. This result points out that the Zn from these compounds reacts immediately with the silica species. The precursor in which ZnO was used has a higher amount of Zn as well. In this case ZnO is directly added to the reaction mixture, not precipitated from the solution of Zn salts so less ZnO was dissolved. In the systems containing water soluble zinc salts (Zn(NO3)2·6H2O, ZnCl2 and ZnBr2) there is less Zn in the solid phase in spite of ZnO precipitation. Nevertheless, the yield of the solid phase recovered after 8 h and at the end of the crystallization (Table 2) has a reciprocal trend. The soluble zinc salts yield about 5 wt% of the solid phase while ZnO and complex compounds result in an almost half lower quantity of the obtained material. Clearly, in the studied systems the precipitation of the solid phase is affected by the nature, concentration, and distribution of zinc species in the liquid phase. On the grounds of these data, it can be concluded that Zn bound with organic ligands is more prone to react with Si species. Furthermore, in the complex compound systems, the solubility of zincolilicate species is higher. Finally, it can be conveyed that, similarly to Al, Ge, Zr and Mg, under the studied reaction conditions, Zn promotes the solid phase precipitation and that the final solid phase yield is low in all systems because of the high alkalinity of the system. The fact that no precipitate has been formed in the all silica system corroborates this reasoning.
Starting chemical | Si/ZnXRF – 8 h | Solid phase yield – 8 h (wt%) | Solid phase yield – 11 d (wt%) | TG release – 8 h (wt%) | dTG minimum – 8 h (°C) | R (Å) | σ R (Å) |
---|---|---|---|---|---|---|---|
ZnO | 2.01 | 2.07 | 2.31 | −14.6 | 92.4; 117.3 | 10.2 | 12.5 (2.85) |
Zn(NO3)2·6H2O | 4.49 | 4.92 | 4.83 | −13.5 | 96.8; 119.6 | 15.6 | 4.1 (7.7) |
ZnCl2 | 4.31 | 5.12 | 5.55 | −13.9 | 97.3; 121.5 | 8.9 | 1.53 (7.2) |
ZnBr2 | 3.99 | 5.00 | 4.45 | −13.2 | 96.3; 121.6 | 11.6 | 2.6 (7.01) |
ZIF-8 | 3.11 | 2.95 | 2.87 | −12.5 | —; 117.0 | 20.5 | 8.65 (6.98) |
Zn(OAc)2·2H2O | 2.63 | 2.69 | 2.58 | −12.8 | 95.0; 116.7 | 17.6 | 9.85 (5.6) |
Zn(acac)2·H2O | 2.76 | 2.90 | 2.91 | −12.1 | 96.1; 120.3 | 11.6 | 2.7 (7.05) |
Fig. 3 Raman spectra of the 8 h hydrothermally treated solids prepared by employing different Zn sources. |
According to the 29Si {1H} CP NMR spectra of the 8 h precursors depicted in Fig. 4 there is a distribution of Si species in the solid phase of the samples. Different zinc sources exhibit an impact on the Si species distribution since the shape of the curves is a function of relative contributions of different species. The spectra range from −69 to −118 ppm which indicates that in all of the samples taken in the initial stage of the reaction there is a certain amount of Si which is located in the spiro position of the lov-unit. Namely, taking into account that the 3MRs have been detected in the Raman spectra of some of the samples and that in a completely crystalline material the Si at the spiro position exhibits a peak at −82 ppm it can be deduced that the three-membered rings connected to build the lov-units contribute to the NMR spectra of the 8 h precursors in the region around −80 ppm.
Fig. 4 Normalized 29Si {1H} cross-polarization NMR spectra of the samples taken after 8 h of hydrothermal treatment of the reaction mixtures prepared by employing different Zn sources. |
Fig. S6† displays the FTIR spectra of the studied RUB-17 precursors and the crystalline RUB-17. The bands from the 1300 to 760 cm−1 region are generally assigned to the asymmetrical O–T–O stretching mode, the symmetrical O–T–O stretching mode and the T–O bending mode of the TO4 tetrahedra.34 The peak positions in the spectra of the precursors coincide with the fully crystalline RUB-17 in this region. However, these peaks are less sharp in the spectra of the precursors. Furthermore, the peaks are shifted towards higher wavenumbers with respect to RUB-17. Clearly, the types of internal tetrahedra T–O chemical bonds are identical in both the crystalline material and its precursors. Still, a significant degree of disorder is present in the precursor materials. Additional structural information (external tetrahedral linkage vibrations) can be obtained from the bands in the region below 750 cm−1 in the infrared spectra. The band around 550 cm−1 is attributed to five membered rings in the structure.34 RUB-17 does exhibit a weak band at 557 cm−1. Moreover, there are three well resolved bands at 520, 445 and 408 cm−1. Any of these bands are present in the spectra of the precursors suggesting once again that there are no well-defined five membered rings, i.e. the parts of the RSN structure larger than the lov-unit have not yet been formed. However, in this region the precursors exhibit broad noisy bands ranging from 540 to 400 cm−1 which indicates that there are intertetrahedral bonds similar to the ones in the crystalline RUB-17 material.
UV/Vis spectral curves of the solid 8 h precursors (Fig. 5) have the same trend as the commercial zinc oxide used as a starting compound in the ZnO-containing system. Likewise, they all exhibit a maximum at 365 nm and have an absorption edge value at 390 nm. However, the intensities of the spectra are significantly lower than those of the pure commercial ZnO. Besides, the spectra of the precursors differ substantially from the spectra of the starting Zn compounds both in the curve shape and intensity (Fig. S7†). RSN-framework materials prepared in this study have an absorption edge position at 390 nm similar to both the precursors and zinc oxide. Although the bands in the spectra of the end products appear in the same wavelength range, there is a difference in their intensity which has been attributed to measurement execution conditions. The ZnO end product exhibits a solid state UV/Vis curve oriented into the opposite direction to the precursor materials with the maximum positioned at 250 nm. These sets of data suggest that the coordination of Zn differs in crystalline RUB-17 from the Zn coordination in the 8 h precursors. The surroundings of Zn in the precursors are more similar to the zinc oxide material than RUB-17. Furthermore, this represents a strong indication that zinc undergoes three states of coordination during RUB-17 synthesis: (1) the coordination in the starting compound; (2) ZnO-like coordination; (3) coordination in the RSN-type framework. It seems that the second coordination is essential for the formation of RUB-17 since it has been detected in all systems. In addition, the infrared and UV/VIS spectroscopy findings suggest that indeed there are similar chemical bonds (building units) in all of the studied 8 h precursors thus substantiating the conclusion that there are lov-units present in these samples.
Fig. 6 TEM images of the 8 h precursor samples prepared using different Zn sources. Arrows denote the dimensions of the voids. |
In addition, the applied model enabled calculating the pore size distribution for each sample (Fig. 8). The data indicate that the distribution curves differ depending on the employed zinc source. Clearly, the overall chemical composition of the reaction mixture determines the structure of the gel particles.
Fig. 8 Calculated spherical void radii distribution curves for the 8 h precursors prepared using various zinc sources. |
Two options have been considered for the formation of the lov-unit by the direct ring closure – one route (route 1) depends on the fully formed ste-unit (defined by the IZA Structure Commission),18 and assumes that the oxygen bridge is already formed between two Si sites in the ste-unit, and ring closure yields the lov-unit. The other route (route 2) assumes the opposite – the lov-unit is already formed, and upon the reaction, it gets embedded on the surface yielding a ste-unit. For the sake of comparison, the pure silica system, with a Si atom in place of the heteroatom, has also been considered and will be discussed first.
The energy profile of the reaction with the 5-coordinated Si atom at the ste (route 2) with the water molecule attached is not shown, yet the formation of this state is via the negligible barrier of 10 kJ mol−1, and the subsequent step of water release is spontaneous and stabilizes the system by 68 kJ mol−1 with respect to the 5-coordinated intermediate. Note that this intermediate is additionally stabilized by the hydrogen bond between the water and the oxygen of the top Si atom. The final product, the 3MR, is thermodynamically stable. Expectedly, it is similar to the route 1 pathway, because both these systems contain 5-coordinated Si atoms, thus are chemically similar and differ only in the constraints. This result leads to the conclusion that once the ste-unit is formed, the ring closure is a relatively easy and feasible process.
The situation is very similar for the Al system; however for pathway 1, the reaction is thermoneutral. There is no significant thermodynamic driving force, and additionally, the kinetic barrier is relatively high, compared to other investigated systems (87 kJ mol−1). This allows us to conclude that the formation of the lov-unit with the Al atom is unlikely. On the other hand, the formation of the ste-unit with the Al heteroatom should be considered an easy process. The process is exothermic by 57 kJ mol−1 and is accompanied by the barrier of only 47 kJ mol−1. The only problem in this process is that it assumes that the lov-unit is already formed in a reaction similar to pathway 1, which is much more difficult. The formation of the 3-membered ring with the Al atom in one T-site is thus the limiting step of the process of Al-containing RSN-type material formation.
The system containing zinc behaves similarly in both investigated pathways as shown in Fig. S9,† and in contrast to all other systems reaction occurs in one concerted step (indeed, the water release in Si and Al systems is very easy, but still it is an activated process). Both pathways are only slightly exothermic (approximately 25 kJ mol−1) with a low activation barrier of 40 kJ mol−1. In addition, in contrast to the Al and Si systems, the 5-coordinated Zn atom has not been observed, and rather the Zn–O bond cleavage occurs easier, leading to the Zn in the planar geometry.
All results suggest that the Zn significantly facilitates the formation of 3MRs. However both routes – either the embedding the 3MR into the growing crystal and 3MR closure at the surface – are equally probable. On the other hand – the small barriers suggest that the preparation process should occur much faster than that observed experimentally. This suggests that there is another rate limiting step during the synthesis.
There is a difference between the 8 h precursors prepared from the soluble zinc salts and complex zinc compounds in terms of their phase composition as revealed by XRD and Raman spectroscopy. Namely, zinc oxide precipitates upon the addition of hydroxide into zinc salt solutions, while in the complex compound systems zinc is coordinated with different ligands which prevent ZnO precipitation. This further reflects in the chemical composition of the solid and liquid phase as well as in the solid phase yield. When ZnO is formed, the silicate precipitate is favoured and consequently the Si/Zn ratio and the quantity of the solid phase are higher. Using complex zinc compounds leads to precipitation of less material of higher zinc content. As the reaction advances, the Si/Zn ratio of the solid changes until it reaches the final value of 3.5 in the crystalline material. However, the amount of the materials recovered at the end of the reaction is very similar to that in the 8 h samples. These data suggest that in the preparation of RUB-17 the properties of the Zn source and the equilibrium between the Zn species in the liquid phase and solid phase play an important role in the overall reaction mechanism. It can be assumed that both Zn species from both the solid and liquid phases react with Si species to form RUB-17, but it seems that the liquid phase mediated processes are the dominating ones.
Based on the spectroscopic measurements (NMR, IR, Raman, UV/Vis) and TG analysis it can be deduced that there are similar chemical bonds in the studied series of samples, yet the distribution of the present species is different. The results suggest that there are three-membered rings connected to form lov-units in all 8 h precursor samples. These data indicate that although lov-units are formed during the initial phase of the zeolite formation (8 h), the subsequent processes of establishing a long range order and assuming needed Zn coordination are rather complex and consequently it takes a longer time for the formation of an entirely crystalline material (11 d).17 A possible explanation could be an early, yet rapid precipitation of the solid phase which has a limited surface area exposed to the liquid phase. Namely, after 8 h of hydrothermal treatment at 180 °C the solid yield is quite low (less than 6 wt% in all of the systems – Table 2) and there are already basic building units present but the reaction is still long. Obviously the frequency of successful chemical reactions in terms of RSN-type framework building is low once the solid phase gets precipitated. This once again leads to the conclusion that in the studied system the solution mediated processes prevail over solid phase rearrangement.
TEM images of the selected samples revealed the presence of spherical voids within the precursor particles. The voids were further probed by GISAXS. Despite not being quantitative, the pore size distribution varies with the employed zinc source thus showing that the chemical composition of the reaction mixture directly affects the gel structure. The core–shell structure of zeolite precursor particles with a porous core was observed in aluminosilicate synthesis systems with a high concentration of alkali cations.35 It was concluded that the mother liquor is entrapped inside the pores due to the impermeable shell and that the onset of zeolite nucleation takes place at the shell part. Consequently, the external surface of the initial gel particles governs the entire crystallization process. If this reasoning is applied to the zincosilicate system studied herein, it can explain the formation of highly intergrown crystals in the end product. Taking into consideration that the reaction is completed after a long period of time, in spite of the structure building units formed in the early stage of the reaction, the external surface area of the precursor particles exposed to the liquid phase should be low or the number of RSN-type nuclei able to grow is small. Finally, the zincosilicate zeolite precursors have a structure similar to aluminosilicate zeolite precursors and obviously the events taking place at the molecular level direct the final product.
The experimental findings in this work are corroborated with the theoretical investigations. Namely, the results of the DFT modelling suggest that the limiting step in the system where Al would be located in one of the T sites is the three-membered ring formation. On the other hand, Zn strongly favours the 3MR formation regardless of the process pathway. Moreover, the low barrier energy values indicate that the overall RUB-17 crystallization should take place more rapidly than it, in fact, does which once again leads to the conclusion that there is some still unknown process which determines the RUB-17 formation. Since a variety of zinc sources has been studied herein in terms of the zinc state and chemical bonds, it can be deduced that the observed effect does not arise due to the differences in the availability of Zn neither various anions present in the system. Nevertheless, considering that there are lov-units in the initial stage of the reaction, yet Zn has ZnO-like coordination, it can be argued that the Zn coordination change is the limiting step here. However, the finding of the exact reason will be pursued in further studies.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9qi00433e |
This journal is © the Partner Organisations 2019 |