Article Cite This: Chem. Mater. 2018, 30, 888−897
pubs.acs.org/cm
Elementary Steps of Faujasite Formation Followed by in Situ Spectroscopy Sebastian Prodinger,† Aleksei Vjunov,† Jian Zhi Hu,† John L. Fulton,† Donald M. Camaioni,† Miroslaw A. Derewinski,*,† and Johannes A. Lercher*,†,‡ †
Institute for Integrated Catalysis, Pacific Northwest National Laboratory, P.O. Box 999, Richland, Washington 99352, United States Department of Chemistry and Catalysis Research Institute, TU München, Lichtenbergstrasse 4, 85748 Garching, Germany
‡
S Supporting Information *
ABSTRACT: Ex situ and in situ spectroscopy was used to identify the kinetics of processes during the formation of the faujasite (FAU) zeolite lattice from a hydrous gel. Using solidstate 27Al magic angle spinning (MAS) nuclear magnetic resonance (NMR), the autocatalytic transformation from the amorphous gel into the crystalline material was monitored. Al Xray absorption near-edge structure shows that most Al already adopts a tetrahedral coordination in the X-ray-amorphous aluminosilicate at the beginning of the induction period, which hardly changes throughout the rest of the synthesis. Using 23Na NMR spectroscopy, environments in the growing zeolite crystal were identified and used to define the processes in the stepwise formation of the zeolite lattice. The end of the induction period was accompanied by a narrowing of the 27Al and 23Na MAS NMR peak widths, indicating the increased level of long-range order. The experiments show conclusively that the formation of faujasite occurs via the continuous formation and subsequent condensation of intermediary sodalite-like units that constitute the key building block of the zeolite.
■
INTRODUCTION The formation of crystalline metastable tectosilicates such as zeolites is one of the most complex inorganic transformations. Despite decades of intense research,1,2 only a small fraction of the potentially stable zeolites have been synthesized. Macroscopically, it can be described classically by layer-by-layer growth models;3,4 however, the elementary processes on an atomistic scale are still being debated.5 Reviews by Cundy, Cox, and Guth et al. summarize the numerous proposed formation mechanisms.5−7 Barrer initially postulated the formation of zeolites to occur via linking secondary building units (SBUs), consisting of tetrahedral and polyhedral rings.8 This solutionmediated process was later supported by the results of Kerr and Zhdanov.9,10 At the same time, however, Flanigan and Breck proposed crystal growth via transformation of the solid hydrogel.6,11,12 The addition of organic structure-directing agents allowed the focus to be on such assembly beyond the kinetically most preferred linking of preformed building units.13,14 The variety of mechanisms proposed for zeolites of different compositions highlights the need to experimentally follow and in turn understand the elementary steps of the formation to expand the currently available synthesis space in an insight-directed manner.15 The challenges lie not only in the design of new pore structures but also in syntheses strategies that lead to new pathways for stable zeolites in aqueous media. For example, low defect concentrations have been identified as being critical for stability in the condensed aqueous phase,16−18 © 2018 American Chemical Society
a potential use for catalyzing environmentally benign organic transformations. Conversions in aqueous media resemble the conditions of hydrothermal syntheses and pose, therefore, significant challenges to structural stability. To understand the reaction steps in the formation of zeolites at an atomistic level, one must explore the chemistry in the reactive environment. The typically used temperatures of ≤250 °C and the corresponding autogenous pressures, as well as the highly alkaline reaction medium, make this a challenging task. While in the past experimental limitations have impeded in situ characterizations of the nucleation and crystallization steps, newly developed capabilities allow in situ measurements during synthesis.15,19−21 To observe the subtle stabilization of zeolites, several experimental techniques must be applied simultaneously. The most basic approach is ex situ characterization by continuous sampling of the gel fraction or by dedicated experiments with varying crystallization durations. The reproducibility in the parallel experimentation and chemical alteration during sampling as well as the separation of the solid product from the liquid and the subsequent drying are major causes of induced artifacts.19 Received: October 30, 2017 Revised: January 7, 2018 Published: January 12, 2018 888
DOI: 10.1021/acs.chemmater.7b04554 Chem. Mater. 2018, 30, 888−897
Article
Chemistry of Materials
and depolymerize.5,10,29 Such reversible polymerization and depolymerization allow the possibility that crystallites form by addition of atoms, ions, or SBUs to existing domains. Once a domain exceeds a certain diameter, X-ray detectable crystallites (>10 nm)30,31 are formed, as deduced from the first appearance of FAU diffraction peaks. This initiates the second stage of the crystallization process involving the rapid, autocatalytic formation of larger crystal domains, yielding a relative crystallinity of 85% within 6 h. The high concentration of Si and Al monomers and small SBUs in the gel aid the formation of the zeolite at this point. They enable additional condensation reactions, leading to the self-accelerating, autocatalytic S curve (e.g., Figure 1). Consumption of nutrients eventually reduced the crystallization rate and initiated the third stage ranging from 6 h until the end of the experiment. Helium ion microscopy (Figure S2) shows a large number of polydisperse particles, suggesting nucleation and crystal growth occur simultaneously, with the slow heating rate favoring the formation of a high concentration of nuclei.32 In the diffractograms of the samples obtained during the initial 6 h, diffraction peaks in addition to the broad anomalous diffraction signal associated with the amorphous material (2θ = 29°) were identified (Figure S1b). The peak positions agree well with those of a variety of sodium aluminosilicates (e.g., sodalite and natrosilite). As these silicates were not observed upon investigation of the aged gel in a capillary (Figure S1c), we conclude that the aluminosilicate species are an artifact induced by the workup procedure (heating of residual occluded gel in the solid). In Situ Al XAFS Analysis. To analyze size domains below the required threshold for X-ray diffraction,30,31 the synthesis mixture was studied by in situ X-ray absorption spectroscopy, probing the electronic and geometric structure of Al species. Let us first turn to the Al K-edge XANES varying as a function of temperature (Figure 2). The intense peak at 1565.5 eV is assigned to tetrahedrally coordinated Al.33 The pre-edge peak at 1562.4 eV is sensitive to both the bond length and the bond angle distortions of the Al−(O−)4 tetrahedron, while the near-edge feature at 1563.8 eV is primarily attributed to variations in structure of the second and higher shells (Figure 2a).34 Most of the changes occurred within the first 3 h of synthesis, suggesting the presence of a FAU precursor species that is subsequently transformed into the growing FAU crystallite. Continuous polymerization and depolymerization of precursor units (e.g., polyhedrals) at the gel−liquid interface29 are hypothesized to stabilize the gel (secondary amorphous phase).10 Note in this context that Navrotksy et al. reported some dissolution of the amorphous solid prior to the onset of crystallization.32 This process of increasing order is to be monitored best with the pre-edge peak in the Al XANES. Figure 2b expands the region between 1568 and 1572 eV. There is a constant decrease in the intensity of this band during the course of the synthesis. Changes in this region are tentatively assigned to the alteration of T-sites in higher shells or the T-site proximity of Na+ or water in the forming FAU crystallite. Note that the region of this band spans only 4 eV around 1570 eV. This is consistent with the presence of tetrahedrally but not octahedrally coordinated species. The latter species are characterized by an absorption over a broader region between 1569 and 1577 eV.35 Changes in this region suggest varying degrees of localization of Na+ balancing the Al tetrahedra.36
We report in this work, therefore, the combination of X-ray diffraction with in situ X-ray absorption spectroscopy and magic angle spinning (MAS) nuclear magnetic resonance (NMR) spectroscopy to probe their sensitivity and specificity in the mechanism of formation of zeolites. The synthesis of faujasite (FAU) was chosen as an example for following the kinetics of zeolite crystallization. The ordered structure of FAU consists of sodalite cages connected by double six-membered rings (D6R),22 forming cavities with an inner diameter of 1.2 nm (“supercage”) that can be accessed by windows of 12 tetrahedra with diameters of ∼0.7 nm. Depending on the Al content, these materials are denominated as zeolite X (Si/Al < 1.5) and zeolite Y (Si/Al > 1.5). This differentiation is characterized by the possible exchange positions of Na cations.23 At such low Si/Al ratios, ordering of the Si and Al atoms in the framework becomes important.24−27 Using 29Si MAS NMR, Melchior et al., for example, proposed that the formation of FAU depends on the ordering of subunits such as sodalite27 and D6R units.28 This was based on limiting the number of possible orientations of the subunits by excluding Al−O−Al bonds (Lowenstein rule) and minimizing of Al−Al next nearest neighbors (Al pairs, Al− O−Si−O−Al).27 The high abundance of both Na and Al allows the elegant probing of the chemistry during the crystallization. We employ 27 Al and 23Na MAS NMR, taking advantage of the fast relaxation times and high natural abundance of 27Al and 23Na. In addition, in situ Al XAFS enabled characterization of the first coordination spheres surrounding Al during synthesis. Ex situ characterization using XRD complements the information gained from in situ measurements.
■
RESULTS AND DISCUSSION Crystallization Kinetics. The crystallinity of FAU (Si/Al = 1.1; zeolite X) was assessed by X-ray diffraction during synthesis. The time dependence of the concentration of the FAU phase as a function of synthesis time is shown in Figure 1. Three stages are identified in the S-shaped curve. When the Si and Al sources were initially mixed, a dense amorphous hydrogel was formed. No crystallinity associated with FAU was observed during the initial 3.8 h, known as the induction period. During the induction period, this hydrogel and other dissolved nuclei are hypothesized to continuously polymerize
Figure 1. Crystallization curve for faujasite determined by XRD measurements of samples characterized ex situ. The time-resolved diffractograms can be found in Figure S1b. The color coding is reported in the legend. 889
DOI: 10.1021/acs.chemmater.7b04554 Chem. Mater. 2018, 30, 888−897
Article
Chemistry of Materials
Figure 3. K2-weighted Al EXAFS Img[χ(R)] spectra acquired at different temperatures during the FAU synthesis experiment followed in situ. Vertical bars are added to simplify the comparison of peak positions. The temperature color coding and the respective reaction times are reported in the legend.
being useful for monitoring kinetic changes by NMR spectroscopy. Thus, 27Al and 23Na, having 100% natural abundance and fast relaxation times, were used to monitor changes during synthesis.37 In situ spectra for the Al environment during formation of FAU are shown in Figure 4 (see also Figure S4a). Two species were observed in a region associated with tetrahedral Al. The significantly larger and broader species (Al[Fr]) is attributed to tetrahedral Al in a solid, and the smaller and narrower peak at 76 ppm is attributed to tetrahedral Al in the liquid phase. This latter assignment was supported by the analysis of the clear liquid above the gel after centrifugation (Figure S4b). The slight variation in the chemical shift of Al(OH)4− (typically 80 ppm) is attributed to partial replacement of hydroxyl groups by siloxy groups. For the sake of simplicity, however, it will be named Al(OH)x−. Increasing the temperature had a pronounced effect on the broad peak of tetrahedral Al, significantly reducing its intensity, which is attributed to the loss of spin magnetization (Figure S4a).21,38 This reversible effect is less pronounced for liquid Al(OH)x− species, a consequence of the spins being in a more homogeneous environment. The intensities and line widths of both peaks as a function of crystallization time and temperature are depicted in panels b and c of Figure 4. The impact of crystallization on the chemical shift is shown in Figure S5a. The line width can be used to characterize the crystallization and the structural order of the probed Al (Figure 4c). As the number of coherent domains increases with crystallinity, the Al environment becomes more structurally ordered, leading to a narrower line width.19 After approximately 8 h, the line width for the framework Al species reached its minimum. Comparing the rate of crystallization via changes in the line width and changes in the coherent lattice domains [XRD (Figure 1)] shows the perfect agreement between both methods (Figure S5b). At the same time, the line shape of the liquid species did not significantly change. Its decreasing concentration shows the gradual consumption of the dissolved species and its incorporation into the framework (Figure 4c and Figure S15). This is in line with the work of Navrotsky et al., observing a marked and sudden decrease in
Figure 2. Normalized Al XANES acquired at different temperatures during the FAU synthesis experiment followed in situ (c). Panels a and b demonstrate the minor changes in the spectral shape in the pre-edge region (1562−1565 eV) and in the ∼1570 eV region, respectively. The inset in panel a shows that most prominent change occurs within 3 h. The XANES spectra were recorded in a temperature interval from 25 to 75 °C over a time interval of 7 h.
The intensity of this band decreased continuously during the entire time period, suggesting a changing Na+ association as the FAU crystallite grows. In the zeolite, Na+ interacts with two or more oxygens of Al T-sites because of their proximity, whereas in the precursor, Na+ is more strictly localized at one of the oxygens of an Al site. The higher intensity of this band suggests, thus, that the alumina tetrahedra in the precursor must have Na+ associated with the T-sites that resembles sodium aluminate.36 The geometry of the Al species during synthesis was monitored via Al EXAFS shown as a temperature series in Figure 3. We note that significant changes in the peak shape or position up to 4.5 Å from the absorber Al atom were not observed during the course of the reaction. This suggests that the majority of primary building blocks, i.e., tetrahedral Al with a single shell of Si [Al(OSiOH)4]−, are formed during the gel aging procedure. The continuous zeolite framework assembly during the synthesis reaction hardly affects the nature of this coordination. The crystallinity upon completion of the experiment assessed by XRD (diffractogram shown in Figure S1a) was identical to that observed by intermittent X-ray diffraction analysis. In Situ 27Al MAS NMR. To obtain detailed information with respect to longer-range ordering, in situ MAS NMR has been used. From the elements present in the synthesis mixture (Si, Al, O, H, and Na), the low natural abundance and long relaxation time of 29Si and 17O isotopes excluded both from 890
DOI: 10.1021/acs.chemmater.7b04554 Chem. Mater. 2018, 30, 888−897
Article
Chemistry of Materials
by the growing Si−O framework, causing adjustment of the cation charge and, as a consequence, decreasing the shielding of the aluminum nucleus. In Situ 23Na MAS NMR. Investigating the Al nucleus via XAFS showed that the Al environment is established during the aging period followed by a reorganization of the gel during the induction period (≤3.8 h). 27Al MAS NMR then showed the narrowing of the tetrahedral Al peak as the continuous attachment of aluminosilicate species resulted in the formation of the zeolite. Let us now turn to in situ 23Na MAS NMR spectroscopy to more closely probe the elementary steps of zeolite formation during the induction period and crystallization. The potential of this approach was demonstrated by Engelhardt et al. differentiating the typical quadrupolar line shapes of Na+ in fully dehydrated zeolites, which in turn allowed us to identify specific ion exchange sites (Figure 5).39 The population of the specific
Figure 5. Identified cation positions occupied by Na+ in the fully crystalline FAU structure: purple for SI (D6R), red for SI′ and SII′ (sodalite cage), and yellow for SII and SIII (supercage).
exchange sites in zeolite X (Si/Al = 1.1) explored here does not stabilize Na+ in the D6R (SI), with the Na+ cations being distributed between sodalite cage (SI′ and SII′) and the supercage (SII and SIII) in a 30:70 ratio (see also Figure S11).39 With in situ 23Na MAS NMR spectroscopy (Figure 6a), however, only one slightly asymmetric peak was observed (see also Figure S6). The hydration sphere of Na+ in the case of aqueous systems or a hydrated state results in a downfield shift as well as a narrowing of the peaks.40−42 The suppression of the typical quadrupolar line shapes by the hydration sphere does not allow attribution to individual exchange sites. Line fitting showed, however, two peaks (Figure S6); the broad peak contributing to the asymmetry of the overall peak is assigned to distorted species with a higher quadrupolar coupling constant (QCC).42,43 We attribute it to Na+ interacting with the solid material in the gel. This would be the case for Na+ terminating Si−O− in, e.g., an amorphous gel or Na+ within already formed pore structures. The narrow peak is assigned to octahedrally coordinated Na+ in the aqueous phase constituting the majority of the signal.43 This assignment was confirmed by varying the pulse width with the amorphous gel at room temperature (Figure S7). In contrast to the solid species, a 2 μs longer pulse
Figure 4. (a) In situ 27Al MAS NMR spectra showing the changes during the synthesis of FAU. The initial 3 h also contains the heating period, which is also shown separately in Figure S4a. Spinning side bands are marked with an asterisk. Deconvolution of the spectra (including the heating stage, shaded in gray) led to changes in the peak area (b) and line width (c) being observed for liquid Al(OH)x− and the solid tetrahedral Al (Al[Fr]). Chemical shift changes are reported in Figure S5a. The color coding is reported in the legend.
the concentration of dissolved Al as the crystallization of FAU was initiated.32 The change in the chemical shift from 60 to 62 ppm (Figure S5a) is attributed to the formation of the zeolite framework. The shift to more parts per million indicates a more electronegative environment surrounding the nuclei induced 891
DOI: 10.1021/acs.chemmater.7b04554 Chem. Mater. 2018, 30, 888−897
Article
Chemistry of Materials
The relative integrated areas and chemical shifts of the deconvoluted peaks are depicted as a function of synthesis time (Figure 6b,c). The change in line width is compiled in the Supporting Information (see Figure S9). Heating the aged gel reduced the signal intensity and resulted in an upfield shift of the peaks (Figure S8). It is hypothesized that increasing temperatures increase the mobility of the Na+ ions, affecting their chemical environments and, hence, the chemical shift. Further changes were not detected until the end of the induction period (∼3.8 h). With the onset of crystallization, the concentration of Na+ in the aqueous phase increases whereas the amount of Na+ bound to solid, e.g., amorphous aluminosilicate and silica precursors, decreases (Figure 6b). As the zeolite framework is constructed, the resulting negative framework charge becomes delocalized over several atoms, mobilizing some of the Na+ that is released into the aqueous phase. Simultaneously with these changes, the broad peak showed a downfield shift by several parts per million, indicative of a more electronegative environment surrounding Na+, as is the case in a zeolite framework (Figure 6c). The line width of the broad peak also significantly narrowed over the duration of the synthesis, indicative of an increased level of long-range order toward the end of the synthesis (Figure S9). Note that the changing line width and chemical shift are in line with changes induced by crystallization and formation of the zeolite and observed via 27Al MAS NMR. Therefore, we hypothesize the broad peak to be a superposition of several resonances of Na+ species associated with solid material. This is corroborated by ex situ 23Na MAS NMR of hydrated and dehydrated samples showing the presence of several Na species linked to solid material (Figure S10). Interestingly, the similarity between the hydrated and dehydrated 23Na MAS NMR spectra of the initial gel (Figure S10) suggests a symmetric, environment of the Na species having a low quadrupolar constant in contrast to that of the fully crystalline zeolite. This is in agreement with our observation of a changing solid Na + environment as crystallization proceeds as indicated by in situ experiments (Figure 6). The superposition of liquid Na+ and solid Na+ signals (Figure 6) prevents insight into the elementary steps of FAU formation. However, high spinning frequencies to minimize second-order quadrupolar interactions resulted in the generation of spinning side bands that allow closer investigation of the solid species in the synthesis gel (Figure 7). The side bands are present as two well-resolved peaks at −20 and −22 ppm shifted upfield from their isotropic positions at 2.8 and 0.8 ppm, respectively. Attributed to the satellite transition,44 they are indicative of at least two distinct species of Na+ in solid environments. As Na+ is not present in the D6R units (SI),39 we assign the peaks to the other possible positions in the FAU framework: the peak at −20 ppm to Na+ in the sodalite cage type environment (SI′ and SII′ sites) and the peak at −22 ppm to Na+ in the FAU supercage (SII and SIII). The overall intensity increased in the isothermal stage as the zeolite is formed (Figure 7a). Plotting the changes in crystallinity [27Al MAS NMR line width (Figure 4)], together with the two spinning side band peaks (Figure 7a) as a function of time, we find the parallel transformations are clearly visible (Figure 7b). It is important to note that the maximum yield of the sodalite units is obtained prior to that of the supercage. While both units are formed initially, the rate of formation for the FAU supercage increased once a threshold concentration of sodalite
Figure 6. (a) In situ 23Na MAS NMR spectra collected during FAU synthesis. (b and c) Changes in the peak area and chemical shift for Na+(aq) (black) and Na+ interacting with solids (green) during in situ 23 Na MAS NMR spectroscopy, respectively. Changes to the peak width are reported in Figure S9. The shaded area corresponds to the heating period.
width is required to maximize the intensity of the peak signal of the liquid species, a result of the spin’s higher mobility in the liquid. Consequently, the short pulse width (π/4, 1.5 μs) results in an underestimation of the liquid concentration, with elemental analysis of the liquid fraction showing the fraction of Na in solution to be as high as 90%. 892
DOI: 10.1021/acs.chemmater.7b04554 Chem. Mater. 2018, 30, 888−897
Article
Chemistry of Materials
the formation of sodalite units precedes the formation of the supercage. It must be noted that the used techniques were unable to explore whether a fully formed sodalite cage is required prior to the growth of the FAU supercage. We tend to exclude this requirement, as it would lead to a formation mechanism involving solely the reaction of several sodalite units into supercage structures, which we consider to be entropically challenging. Instead, the condensation of partially formed, halfopen sodalite units and other SBUs that results in the growth of the complex FAU structure is proposed. This is structurally directed by the different Na+ environments (see Scheme 1).
■
CONCLUSIONS The crystallization of faujasite starts with the aging process, when most SiO2 particles in the Si sol are depolymerized by the presence of hydroxide anions. Silica units form a gel upon being mixed with the Al solution. This primary amorphous phase (Figure S16), consisting of a variety of primary building units such as small silica and aluminosilicate oligomers, is then structurally rearranged into a secondary amorphous phase (monitored by in situ Al XANES). The subtle changes in the pre-edge peak in the Al XANES, attributed to the length and degree of distortion in the Al−O bond, increased only during the initial heating period as the short-range ordered FAU precursors (secondary amorphous phase) are formed. Al EXAFS (Figure 3) shows that the Al coordination does not change, in agreement with the constant MAS NMR and X-ray diffraction results during the induction period. These observations appear to be in contrast to a recent report by Rimer et al. in which, despite the similar gel composition, the addition of SiO2 particles to the alkaline Al solution at room temperature resulted in a predominately monodisperse gel of spherical undissolved SiO2 particles with a thin Al coating.45 Consequently, a heterogeneous gel was formed with most Al in solution, favoring a nonclassical nucleation on the exterior of these core−shell particles. This demonstrates that the method of gel preparation has a profound impact on the state of the primary amorphous phase and the subsequent mechanism of formation as shown recently by Ivanova et al. in the case of BEA zeolite.15 By continuous polymerization and depolymerization, these aluminosilicate nuclei eventually reach a critical size at which crystal growth and propagation of the nuclei are energetically feasible, ending the induction period.46,47 While we do not investigate the location of origin for nucleation, the more homogeneous nature of our gel (amorphous aluminosilicate vs core−shell particles) allows us to conclude that nucleation and consequently crystal growth occur throughout the bulk solid rather than the exterior.45 Crystallization is best followed by MAS NMR techniques using 27Al and 23Na as probe atoms (Figures 4 and 6). The narrowing of the Al line width corresponds to the increasing level of order attained during crystallization. Simultaneously, the relative concentration of Na +(aq) increases as the aluminosilicate solid is transformed. As the zeolite is formed, the environment of the framework charges changes, monitored with in situ 23Na MAS NMR (Figures 6 and 7). Initially, the negative charge is localized on Al(OH)x− as well as terminating silica tetrahedra. In both cases, Na+ acts to balance the negative local charge. With the assembly of the zeolite framework and the geometric arrangement of the individual building units into subunits
Figure 7. Changes in the spinning side band associated with solid Na+ material as a function of synthesis time. A high-field peak and a lowfield peak were identified at −22 and −20 ppm, respectively. The initial heating period is not reported (see Figure S12). Panel b shows the kinetic transformation of amorphous material (27Al MAS NMR line width) into crystalline FAU as directed by the speciation of Na+ ions [plotted as formed fraction of the final concentration of sodalite (− 20 ppm) and the supercage (− 22 ppm)]. Changes during the induction period are not considered (approximated to 0% conversion). See also Figure S13 for additional trends (chemical shift and peak width). The color coding is reported in the legend.
units was achieved. The generation of sodalite units provides the necessary substrate for the production of the FAU supercage (Figure 5). Therefore, we hypothesize that the formation of the FAU supercage critically depends on the concentration of sodalite subunits. As the synthesis progressed and the level of structural order increased, the −22 ppm peak narrowed significantly, while the peak at −20 ppm hardly changed in line width (Figure S13). Using the intensity ratios of the −22 and −20 ppm peaks toward the end of the synthesis (Figure S13), the distribution between Na+ in the supercage (− 22 ppm) and Na+ in the sodalite cage (− 20 ppm) was estimated to be 70:30. This is in agreement with the population study conducted by Engelhardt et al. showing a combined population of 67 Na per unit cell in SII and SIII sites and 24 Na per unit cell in SI′.39 For the 96 atoms in the FAU unit cell, this corresponds to a 70:30 ratio. Thus, overall, we conclude that 893
DOI: 10.1021/acs.chemmater.7b04554 Chem. Mater. 2018, 30, 888−897
Article
Chemistry of Materials
Scheme 1. Several Stages of Zeolite Formation Can Be Observed Using the Combination of in Situ Techniques Described Hereina
a
(a) The aged gel consists mainly of an amorphous solid aluminosilicate (gray). (b) The structure-directing role of Na+ leads to the generation of sodalite-like units (red lines indicate possible bonds formed). (c) Formation of the FAU supercage by arrangement of sodalite units. (d) Fully crystalline FAU. The color coding is as follows: green for Na+(aq), blue for Na+(amorph), red for Na+(sod), and yellow for Na+(FAU). Red dashed lines show possible bonds formed between precursor species.
■
(e.g., sodalite cages), the negative framework charge is distributed over several nearby atoms. This transition can be observed by following the chemical shift and line width in 23Na MAS NMR (Figure 6) indicating a weaker coordination. In the zeolite, the Na+ cations occupy specific locations such as those in the sodalite units (SI′ and SII′) and supercages (SIII and SII). This differentiation of the Na+ environment is being observed with 23Na MAS NMR (Figure 7). First, the concentration of Na+ in sites of the sodalite cage type environments is maximized (i.e., SI′ and SII′ sites represented by −20 ppm side band peak in Figure 7). Once a certain number of sodalite units has been formed, further growth of the FAU supercage is facilitated (i.e., Na+ in SII and SIII exchange sites, represented by the −22 ppm side band peak in Figure 7). The changes in the spinning side bands observed with 23Na MAS NMR (Figure 7) strongly suggest the sodalite cage to be a critical prerequisite to the subsequent formation of the supercage. Investigating the synthesis across different time and domain scales allowed us to gain important insight into the formation of FAU. Most of these reactions occur within the first 3−4 h after the end of the induction period. At the same time, the longrange order as assessed by XRD is at 85% completion. While the crystallization of zeolites can be described macroscopically by a layer-by-layer approach,3,4 we show here that at the atomistic scale, the gradual conversion into a FAU framework can be understood by an atom-by-atom growth model, where the structure-directing role of Na+ cations is exploited to first form sodalite cage subunits that are critical to the growth of the FAU zeolite (Figure 7). This and other recent work15 show the successful probing of the zeolite formation mechanism using in situ spectroscopies of the elemental building blocks in zeolites and, thus, promise to open new and interesting pathways for understanding and designing nanoporous aluminosilicates.
EXPERIMENTAL PROCEDURES
Synthesis Procedure. Classical Synthesis. Faujasite was synthesized using the procedure described by Navrotsky et al.32 First, 1.38 g of NaOH (Sigma-Aldrich) was dissolved in 5.14 g of H2O (Milli-Q), followed by the addition of 2.2 g of the silica source (Ludox HS-40, Sigma-Aldrich) to the hot and alkaline solution. It was stirred until it became clear. Separately, an Al solution was prepared by dissolving 0.81 g of NaAlO2 (Sigma-Aldrich) in 6.17 g of H2O (Milli-Q). The Al solution was then added to the Si solution under intensive stirring (600 rpm) and aged while being stirred at room temperature for 30 min. The gel composition is 5.5:1:3.5:175 Na2O:Al2O3:SiO2:H2O. Once aging was complete, the gel was placed inside Teflon-lined autoclaves and then inside an oven, heated to 70 °C at a rate of 0.25 °C/min under static conditions, and kept at this temperature for 25 h. At set intervals, autoclaves were removed from the oven and cooled to room temperature, and the synthesized material was separated from the liquid phase via centrifugation (8000 rpm for 8 min). The solid residue was washed several times with distilled water and then dried at 70 °C overnight before undergoing further characterization. Faujasite Synthesis Followed in Situ via Al K-Edge XAFS. The Al K-edge XAFS experiments were performed at Phoenix II, the elliptical undulator beamline at the Swiss Light Source (SLS) at the Paul Scherrer Institute (PSI) (Villigen, Switzerland). Energy calibration was achieved by setting the inflection point of an Al foil spectrum to 1559.6 eV. The double-crystal monochromator employed a set of KTiOPO4 (011) crystals to provide an energy resolution of ∼0.6 eV over a scan range for the Al K-edge from 1500 to 2150 eV. Two Nicoated mirrors were set at an angle of 1.45° to provide a cutoff of higher harmonics. An unfocused 1.0 mm × 1.0 mm beam having a flux of ∼109 photons/s was used. Measurements were performed in fluorescence mode. I0 was measured as the total electron yield signal taken from a 0.5 μm thin polyester foil, which was coated with 50 nm of Ni. This I0 detector was held in a miniaturized vacuum chamber (2.9 × 10−6 mbar), which is separated by a thin Kapton foil from the measurement chamber itself. The X-ray fluorescence was detected using a four-element Vortex Si-drift diode detector. ATHENA48,49 was used for background processing necessary to extract the χ(k) data from the background function. A Fourier filter cutoff distance, Rbkg, of 1.0 Å was used. The XAFS data were weighted by k2 and truncated using a Hanning window with dk = 1.0 Å−1 in the range of 1.5 Å−1 < k < 8.0 Å−1. 894
DOI: 10.1021/acs.chemmater.7b04554 Chem. Mater. 2018, 30, 888−897
Article
Chemistry of Materials 27
Al MAS NMR. Hydrated spectra were recorded on a Varian 500 MHz spectrometer using a 4 mm HX probe. 27Al experiments were conducted using a 0.6 μs pulse width for a π/20 pulse angle. Recycle delays of 1 s were applied to fully relax the spins; 5000 scans at a spinning rate of 16 kHz were collected and analyzed using MestreNova software. The peaks were referenced externally to a 1.0 M Al(NO3)3 solution set to 0 ppm. To fully hydrate all the Al, the samples were stored in a desiccator with a Ca(NO3)2 solution for a minimum of 48 h. Upon completion of a 27Al experiment, the 23Na spectrum was recorded immediately. A pulse width of 1.4 μs was used to simulate a π/20 pulse; 5000 scans with a pulse delay of 1 s were recorded at a spinning rate of 16 kHz. The signal was referenced to a 1.0 M NaCl solution at 0 ppm. 23 Na MAS NMR. In addition to the hydrated samples, the material obtained was also investigated with solid-state NMR in a dehydrated state. Dehydration was achieved by heating the sample to 400 °C (5 °C/min) and keeping it at this temperature for 10 h. Experiments were conducted at different field strengths. A Bruker 500 MHz spectrometer with a triple-resonance wide-bore probe was used. The resonance frequency was set to 132.3 MHz, and the sample was spun at 10 kHz. A pulse width of 1.4 μs was used corresponding to a π/20 pulse; 5000 scans with a pulse delay of 1 s were recorded at a spinning rate of 16 kHz. The signal was referenced to solid NaCl set to 0 ppm.
The faujasite synthesis gel was prepared following exactly the same recipe as described above and then aged for 30 min while being stirred. The gel was loaded in the EXAFS cell (see Figure S3). The cell was loaded in the vacuum chamber, and the synthesis was followed in situ from 25 to 75 °C. The temperature ramp was set to 0.12 °C/min. Once the set point temperature was reached, the gel was allowed to react for an additional 1 h to achieve maximum crystallinity. Upon completion, the product was filtered and washed several times with distilled H2O. In Situ Synthesis: NMR. Typically, 300 mg of gel with the same composition as described for the classical synthesis was loaded in the high-temperature and high-pressure MAS rotor for in situ MAS NMR experiments. The temperature was increased at a rate of 0.25 °C/min to 70 °C and kept at this temperature for 15 h. In situ 23Na and 27Al MAS NMR measurements were taken on a Varian 500 MHz NMR spectrometer using a 7.5 mm HX MAS probe with a spinning rate of 3 kHz at resonance frequencies of 132.3 and 130.3 MHz, respectively. 23 Na MAS NMR was additionally measured on a Varian 300 MHz spectrometer with a 7.5 mm HX MAS probe, spinning at 3 kHz and 81.8 MHz resonance frequency to confirm the nature of the spinning side bands (Figure S14). Variable-temperature experiments were conducted with the commercially available heating stack provided by Varian Co., and the real temperature in the rotor was calibrated with ethylene glycol. 23Na MAS NMR spectra were recorded using a pulse width of 1.5 μs for a π/4 pulse, and 128 scans were accumulated with a 1 s recycle delay. The chemical shifts were externally referenced to a 1.0 M NaCl aqueous solution at 0 ppm. 27Al MAS NMR experiments were performed using a pulse width of 1 μs for a π/4 pulse, 128 scans, and a 1 s recycle delay. The spectra were externally referenced to a 1.0 M Al(NO3)3 aqueous solution. Over the course of the experiment, 400 spectra were obtained during the isothermal stage in addition to those measured at varying temperatures. The spectra were analyzed with MestreNova software. Initially, an exponential apodization function (100 Hz) was applied to the time domain free induction decay. In addition, the 400 spectra obtained during the isothermal stage were averaged to 40 and 20 spectra for 27Al and 23Na, respectively. The different Al and Na species were assessed by deconvolution of the peaks with symmetric peaks of Lorentzian/Gaussian line shape, which was kept constant for the individual species. All other parameters were optimized to obtain the smallest error. An example of a line fitting can be seen in Figure S6. The pulse widths were chosen to be optimized for solid materials [requiring shorter pulse widths (see Figure S7)]; thus, the concentrations of liquid species observed with 23Na and 27Al are underestimated. Corresponding concentrations of Na and Al in the solution are reported in the Figure S15. Ex Situ Characterization. Samples obtained from classical and in situ syntheses were additionally characterized ex situ using the following methods. X-ray Diffraction (XRD). XRD patterns were typically collected with a Rigaku Mini Flex II benchtop X-ray diffractometer using a Cu Kα radiation of 0.154056 nm (30 kV and 15 mA). The step size was 2°/ min ranging from 5 to 65°. Helium Ion Microscopy (HIM). HIM images were obtained using 30 keV He ions with a 1.0 pA beam current at normal incidence. Secondary electrons were detected using an Everhart−Thornley detector. For HIM imaging, a very thin layer of carbon (
