1. Introduction
Zeolites are crystalline aluminosilicate minerals, known for their microporosity and pervasive use as molecular sieves, adsorbents and catalysts [
1]. One of their most renowned uses is as catalysts for fluid catalytic cracking in the petrochemical industry [
2]; however, there is ongoing research into using zeolites for hydrogen fuel storage [
3], carbon dioxide capture [
4] and selective extraction of radionuclides [
5]. Their structure consists of a three-dimensional framework built from silica and alumina TO
4 tetrahedra (T = Si or Al) that are connected via a shared apical oxygen atom. Within the structure, it can be recognised that the TO
4 tetrahedra arrange themselves into regular geometric subunits, which are referred to as secondary building units (SBUs) [
6].
Energetically, zeolites are recognised as being metastable and it is the stabilisation of the open framework cavities that is an integral factor dictating the crystallisation process. Species that provide such a structure stabilising effect are known as structure directing agents (SDAs) [
7,
8]. In natural occurring zeolite formation this stabilisation is achieved by metal cations which use electrostatic interactions to stabilise the voids and guide the assembly of certain SBUs [
9]. For the preparation of synthetic zeolites, we can employ the use of organic additives that use a mixture of electrostatic, van der Waals and hydrogen-bonding interactions to direct structure [
10,
11,
12]. Although the available non-bonding interactions between organic additives and the growing zeolite framework are known, there is little clarity on how exactly this impacts the zeolite crystallisation process. This is particularly important if we are to rationally traverse the metastable zeolite synthesis landscape and prepare new zeolite architectures for intended applications. The prevalent use of organic additives is best illustrated by the fact that less than 30 of the 255 zeolitic frameworks recognised by the International Zeolite Association can be synthesised in the absence of an organic additive [
13,
14].
Davis and Lobo [
15,
16] suggest that the potential behaviour of organic additives can be divided into three classifications: true templates, organic structure directing agents (OSDAs) and space-filling species. True templates are seen to employ strong interactions and imprint their symmetry onto the growing framework. The discrepancy between OSDAs and templates is unclear, but the former is attributed by weaker interactions and an absence of symmetric or geometric matchup. Lastly, space-filling species are recognised by weak interactions and occupy the framework void spaces rather than being forthrightly involved in directing assembly. Despite these recognised categories, additional additive roles such as “structure-blocking” [
17,
18] and influence on the free-energy landscape of crystallisation [
19,
20] have been suggested but have received little research. Currently, computational calculations and synthetic approaches only consider the templating roles of organic species to rationally predict the ideal additives to prepare new zeolites [
21,
22,
23]. However, a much wider appreciation of the structure directing capabilities of organic additives is required to take full advantage of their use in zeolite design.
18-crown-6 ether (18C6) is an example of an organic additive that can be used in the synthesis of multiple zeolites with a variety of topologies [
24,
25,
26,
27]. Previously, it has been observed that 18C6 can undertake differing roles in the synthesis of the polymorphic
EMT and
FAU-type zeolites. [
20,
28] Although in both the 18C6 exists as a ((18C6)Na
+) macrocation, it is a true geometric template for the synthesis of
EMT-type zeolites and a space-filling species in the synthesis of
FAU-type zeolites [
29,
30,
31,
32,
33]. This is important, as it demonstrates the integral role of 18C6 in the synthesis mechanism to differentiate between the two polymorphs. Zeolites RHO and ZK-5 (
KFI) are two zeolites that can also be prepared using 18C6 [
24,
26]; however, the role, identity, and location of the 18C6 species has not been determined.
Figure 1 shows the framework structure of these two zeolites, in addition to the constituent SBUs. Both zeolites consist of α-cages in a body-centred cubic arrangement and it is these α-cages that the 18C6 is believed to occupy [
33,
34]; however, this has not been explicitly proven. Due to the lack of structural information, there is currently no understanding of how the 18C6 molecule is involved in the synthesis of these two zeolites in comparison to the
FAU and
EMT-type zeolites.
For the first time, we report the crystal structure of zeolites RHO and ZK-5 with the 18C6 species occluded within their zeolite frameworks. We present experimental evidence that the 18C6 molecule occupies the α-cage cavities in both zeolites, in addition to the cation the molecule is coordinated to. Using these structures, we have gleaned the role of the 18C6 organic additive throughout the crystallisation of both zeolites.
2. Methods
2.1. Sample Preparation
The zeolite RHO and ZK-5 samples were synthesised using the verified procedures reported by Chatelain et al. [
24,
26,
27,
34] using 18-crown-6 ether (18C6) as organic additive. The molar batch compositions used in the synthesis of each zeolite are shown in
Table 1. The materials used for sample preparation were sodium hydroxide (NaOH), potassium hydroxide (KOH), strontium nitrate (Sr(NO
3)
2), caesium hydroxide solution (50 wt% CsOH in water), 18-crown-6 ether (C
12H
24O
6, 18C6), sodium aluminate (NaAlO
2), aluminium hydroxide (Al(OH)
3), colloidal silica (LUDOX
® HS-40, 40 wt% SiO
2 in water) and deionised water. Aside from the deionised water, all the materials used were purchased from Sigma-Aldrich (Gillingham, UK).
2.2. Zeolite RHO
Zeolite RHO was prepared as follows. First, the sodium and caesium hydroxide were added to the deionised water, followed by the addition of the 18C6. Upon complete dissolution the sodium aluminate was added, and the mixture permitted to stir until it was homogeneous. Next the colloidal silica was slowly poured into the solution under stirring, to negate any rapid gelation. The produced hydrogel was then aged under stirring at ambient conditions for 24 h. After aging, the hydrogel was transferred to a Teflon cup within a stainless-steel autoclave and heated at 110 °C for 8 days. Following heating, the autoclave was removed from the oven and left to cool to ambient temperature. The crystallised zeolite product was subsequently separated from the mother liquor via Buchner filtration and washed with deionised water until the pH of the filtrate was neutral. The product was then dried at 90 °C, before being ground prior to further calcination and dehydration.
2.3. Zeolite ZK-5
The zeolite ZK-5 precursor hydrogel was prepared by first dissolving the potassium hydroxide in a portion of the deionised water. The aluminium hydroxide was added to this solution, and then dissolved under stirring and heating to near 110 °C. After dissolution, the solution was cooled to ambient temperature and replenished with any water loss via evaporation during heating. In a separate vessel the 18C6 and strontium nitrate were dissolved in the second portion of deionised water. The colloidal silica was then poured into this solution and stirred until homogeneous. Next, the previously prepared alumina solution was rapidly poured into the silica solution under stirring to produce the hydrogel, which was aged under stirring for 30 min. After aging, the hydrogel was transferred to a Teflon cup within a stainless-steel autoclave and heated at 150 °C for 5 days. Following heating, the autoclave was removed from the oven and cooled to ambient temperature. To retrieve the as-synthesised zeolite ZK-5 sample the same process of separation, washing and drying performed for zeolite RHO was used.
2.4. Calcination and Dehydration
The as-synthesised sample of each zeolite was divided into two halves, to produce the filled and empty analogues. The filled analogue was synthesised with the 18C6 intact, and the empty analogue was calcined to remove the occluded 18C6.
Calcination was achieved by heating the relevant analogue in a tube furnace under air to 450 °C for 6 h at a ramp rate of 1 °C·min−1. Throughout this heating cycle the temperature was held static at 100 °C, 200 °C and 300 °C for 1 h each. After calcination the sample was cooled at a rate of 1°C·min−1 to ambient conditions, including a static stage at 200 °C for 1 h.
Based on sample conditions on the respective diffractometers, only the zeolite RHO empty and filled samples were dehydrated in preparation for measurements. This was performed by heating the samples in a tube furnace under vacuum. The heating cycle consisted of a ramp rate of 1°C·min−1, with static temperatures stages of 100 °C for 1 h and 200 °C for 6 h. Following this, the sample was then cooled to ambient conditions at a rate of 1 °C·min−1, including a static stage at 100 °C for 1 h during this cooling cycle. The dehydrated sample was then transferred under vacuum to a dry argon glove box and sealed into a glass vial under argon gas.
2.5. Thermogravimetry
The 18C6 and water mass content of the empty and filled zeolite analogues were studied using thermogravimetry. This was conducted on samples which had not been dehydrated. A Setaram Setsys Evolution TGA 16/18 (KEP Technologies Group, Sophia Antipolis, France) was used, with samples loaded into open 170 µL alumina crucibles. The samples were heated from 30 °C to 600 °C at a ramp rate of 5 °C·min−1, under 20 mL·min−1 flow of air. Instrumental errors were accounted for by correcting with a thermogravimetric scan of an empty crucible.
2.6. Elemental Analysis
The framework Si/Al ratio was estimated by using both energy dispersive X-ray (EDX) spectroscopy and solid-state magic angle spinning (SS MAS) nuclear magnetic resonance (NMR) spectroscopy.
EDX spectroscopy was performed using an Oxford INCA X-ray analyser (Oxford Instruments, Abingdon, UK) equipped to a JEOL SEM6480LV microscope (JEOL GmbH, Freising, Germany). Multiple sites on a variety of crystals were analysed, with an average and standard deviation recorded. The spectral line intensities measured were compared to a calibration standard.
SS MAS NMR spectroscopy of the
29Si nuclei was performed at the formerly EPSRC-sponsored SS NMR service at Durham University (Durham, UK). A Varian VNMRS spectrometer (Varian Inc., Palo Alto, CA, USA) was used, operating with a 9.4 T magnet, and equipped with two magic-angle spinning probes. Neat Si(CH
3)
4 was used as a chemical shift reference. Measurements were taken under an inert nitrogen atmosphere, with direct excitations using a frequency of 79.435 MHz, a pulse duration of 4.0 µs, a 5000 Hz spin rate and a 20.0 ms acquisition time.
1H cross-polarisation was performed at the same conditions, with the addition of two-pulse phase-modulated decoupling frequencies of 40 and 55 kHz. From this it was seen that there was a negligible concentration of Si-OH silanols present. The Si/Al ratio was determined using the conventional method of Lippmaa et al. [
35].
2.7. High Resolution Powder X-ray Diffraction
Both the empty and filled analogues of zeolites RHO and ZK-5 were analysed using high-resolution powder X-ray diffraction. The diffraction patterns for the dehydrated zeolite RHO samples were recorded on the ID22 beamline at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France. The detector parameters were calibrated with silicon and the incident X-ray radiation had a wavelength of 0.49598 Å. The samples were loaded into 1 mm diameter borosilicate capillaries within an inert nitrogen atmosphere to prevent any sample hydration. The capillaries were subsequently sealed with vacuum grease and wax, before being mounted onto the diffractometer and cooled to 100 K using an Oxford Cryosystems Cryostream 700 Plus. (Oxford Cryosystems, Oxford, UK). A scan range of 2θ = 0.5–45.0° (binning size of 0.002°) was used, equating to a scan length of approximately 22 min. Throughout scanning the capillaries were spun perpendicular to the X-ray beam to reduce preferred orientation effects.
The hydrated zeolite ZK-5 samples were analysed using a STOE STADI MP diffractometer (STOE & Cie GmbH, Darmstadt, Germany), equipped with a MYTHEN2 1K silicon strip detector (Dectris AG, Baden-Daettwil, Switzerland). The incident X-ray radiation was pure Cu Kα1, of wavelength 1.5406 Å, obtained using a Cu X-ray tube and a Ge(111) monochromator. A scan range of 2θ = 3–100°, step size of 0.015° and scan length of 13.6 h were used. The diffraction measurements were obtained in the Debye–Scherrer mode, using a transmission sample holder between two foil insets with 3 mm masks. The diffraction patterns were recorded at ambient temperature.
The patterns obtained from both diffractometers were used for structure determination, using the Rietveld method in the TOPAS Academic software (Coelho Software, Brisbane, Australia) [
36]. The starting structures used were the crystal structures of zeolite RHO in the
symmetry and the synthetic (Cs, K)-ZK5 zeolite, both reported by Parise et al. [
37,
38]. For the latter ZK5 zeolite structure, the Cs sites were substituted with K sites prior to the refinement process. First the structure of the empty analogues of zeolites RHO and ZK-5 were determined, and these models used as the starting structures for the respective filled analogues.
The Rietveld refinements were performed using the 2θ = 2.0–45.0° and 2θ = 5.0–100.0° data ranges for the zeolite RHO and ZK-5 samples accordingly. During the refinement process 100,000 iterations were used, in addition to a χ
2 convergence criteria of 0.001. The quality of the Rietveld fit was evaluated using the goodness of fit factor
G, equal to
. [
39] The average mean-square displacement in all directions,
Beq, was used to assess the degree of atomic displacement [
40]. The 18C6 oxyethylene chain was modelled as a rigid body, using the Cartesian coordinates of a geometry-optimised 18C6 species used previously by Nearchou et al. [
33]. The occupancy of the C and O oxyethylene chain sites were constrained to be equivalent, and translations and rotations of the rigid body refined. A summary of the refinement process and profile functions are included in the
Supplementary Materials.
4. Discussion
In zeolites RHO and ZK-5 it is seen that the occluded 18C6 species occupies the α-cage, as has been expected [
33,
34] but not explicitly proven. In both zeolites the 18C6 species is disordered amongst several possible orientations due to the inherent symmetry of the cubic system, whereby all three crystallographic axes are indistinguishable. Consequently, the periodic structure is an average of all these possible permutations, producing the average structure reported herein. The observation that the occluded 18C6 species fails to match up with the framework symmetry, as seen in zeolite EMC-2 [
29], would suggest that the 18C6 species does not behave as a geometric template in the crystallisation of zeolites RHO and ZK-5. However, as both the 18C6 molecule in the D
3d conformation and the α-cage S6R faces share three-fold symmetry, we speculate that the 18C6 species may adsorb to these faces on the crystal surface, promoting the growth of an adjacent α-cage. Crystal growth along this (111) direction would consequently lead to cubic close packing of the α-cages.
Interestingly, both zeolites RHO and ZK-5 can be prepared without the need for 18C6, [
27,
51,
52,
53] in contrast to zeolite EMC-2 where it appears to be a necessity [
20,
28]. For zeolite RHO it is the Cs
+ cations that direct the assembly of the D8Rs [
54] and in zeolite ZK-5 the K
+ cations are involved in the formation of the t-pau units [
48]. In both cases, this behaviour must be sufficient to assemble the relevant framework structure, without the need for an organic additive. Instead, it is expected that the 18C6 influences the kinetics of crystallisation. We believe that the 18C6 species must play a role as a space-filling species that stabilises the formation of the α-cage, which is supported by its presence in this cavity in both crystal structures. For zeolite RHO it is the non-complexed 18C6 molecule and in zeolite ZK-5 it is the ((18C6)K
+) macrocation that performs this stabilisation role. In both cases, the stabilising would involve non-bonding van der Waals interactions with the growing framework, with the ((18C6)K
+) macrocation having the addition of electrostatic interactions for zeolite ZK-5. The influence of this stabilisation on the actual synthesis of zeolite RHO is that it becomes increasingly facile and expands the crystallisation field [
55,
56,
57,
58] Therefore, we anticipate a similar impact on the synthesis of zeolite ZK-5. Such behaviour would suggest that it influences the free-energy of crystallisation, as we have seen for
FAU and
EMT-type zeolites [
20].
5. Conclusions
Herein, we present the crystal structure of the as-synthesised, 18C6-containing and calcined analogues of zeolites RHO and ZK-5. We have successfully elucidated the identity of the occluded 18C6 species, which in zeolite RHO is an isolated 18C6 molecule, and in zeolite ZK-5 is a coordinated ((18C6)K+) macrocation. In both zeolites, we observe that the relevant 18C6 species occupies the α-cage cavities. According to the fractional occupancies and thermogravimetric data, we predict that there are 0.54 molecules per cage in the former, and 0.66 macrocations per cage in the latter. Furthermore, the occluded 18C6 species can be present in a number of different orientations, with no periodic ordering between adjacent unit cells. The structures observed are thus an average of the available permutations.
Due to the lack of symmetry matchup between the 18C6 species and the zeolite framework it is deemed that in both cases the relevant 18C6 species does not behave as a true geometric template. Instead, we believe that the role of the 18C6 species is better described as a structure directing or space-filling agent. Rather than imprinting symmetry during crystallisation, the 18C6 species stabilise the open void of the α-cage as it is being assembled.
Using these crystal structures, a more coherent understanding of the role of 18C6 as an organic additive in the synthesis of zeolites can be reached. This presents the need for greater appreciation of the multiple available behaviours of organic additives that can be applied to prepare new zeolites.