Enantiomerically enriched, polycrystalline molecular sieves

Schmidt et al. (24) reported on the synthesis of STW using a computationally predicted OSDA. This work demonstrated the feasibility of a priori predicting chemically synthesizable monoquaternary, imidazolium OSDAs to create a specified, fluoride-mediated, pure-silica framework. Additionally, STW has been reported to form using diquaternary imidazolium-based OSDAs that are of sufficient size to conform to 10MR channel structure, implying more rigid, chiral analogs may be included in the framework of STW, and may potentially impart structural chirality (24–28). Here, we implemented the previously published computational method implemented by Schmidt et al. (24) and Pophale et al. (29) and used the molecular design constraints suggested by Davis and Lobo (15). The computational method was modified such that a given enantiomer of each potential OSDA molecule was simulated in both enantiomers of STW, with successful candidates producing a strong stabilization in only a single enantiomer. Based on our previous work with STW, we believed that a computed stabilization energy larger than −15 kJ⋅(mol Si)⁻¹ would be needed to form STW. Ultimately, a single OSDA candidate was selected (Table 1). Relative to the other predicted OSDAs for STW, 2 has the largest energy difference between enantiomers, making it the most suitable target for experimental evaluation. Coupled with energy predictions, this molecule was also selected after ensuring that the both enantiomers were synthetically attainable (24). A full description of the computational method is provided in SI Appendix.

Detailed procedures for the chiral resolution and reaction schemes used to synthesize OSDA 2 are given in SI Appendix, Figs. S1 and S2. Chiral resolution of the starting compound (trans-2-phenylcyclopropane-1-carboxylic acid, 1) was performed by successive crystallizations with either dehydroabietylamine or quinine as chiral derivatization agents to obtain the R (1a) or S (1a′) forms of the starting material (SI Appendix, Fig. S1). Polarimetry measurements were taken for the salt and free-acid forms of 1a and 1a′ to confirm enantiopurity, after the separation was complete. Further details of the synthesis of 2 are provided in SI Appendix (note: enantiomers of 2 will be denoted R-2 and S-2). (Λ,R)-BINPHAT tetrabutylammonium salt was used as a chiral shift reagent to detect the enantiopurity of 2 after executing the reactions outlined in SI Appendix, Fig. S2. As shown in Fig. 2, the neat product ¹H NMR resonances at δ 4.80 and 4.04 ppm are singular and distinct multiplets. Addition of BINPHAT to a racemic sample of 2 results in a twofold spectral change: (i) a distinct shift upfield of only the aforementioned peaks from their initial position and (ii) an observed peak split. Integration of the split peaks indicates that they maintain a 1:1 ratio, expected for a racemic compound. However, the two enantiomers of 2 show only a single peak. This method confirms the enantiopurity of 2 and demonstrates that no racemization occurs throughout the organic synthesis scheme.

Syntheses of STW were conducted with enantiopure samples of the R and S enantiomers of 2, as well as the racemic mixture of the two. These syntheses were performed at temperatures and H₂O/SiO₂ ratios that are typical for silica-enriched, fluoride-mediated syntheses. Because STW is composed of ∼80% double four-tetrahedral atom rings that are known to be stabilized by inclusion of germanium, addition of varying quantities of germanium to the synthesis gels was explored (30, 31). Following the work of Schmidt et al. (24), reagent molar ratios that led to the synthesis of STW in the shortest times were 1 SiO₂:x GeO₂:5 H₂O:0.5 HF:0.5 2 at 160 °C (where x varies from 0.05 to 0.5) using 10% (wt/wt) seeds produced from an achiral diquaternary OSDA with crystal sizes on the order of 1 μm. Aluminum-containing samples were also synthesized (in the presence of Ge) using aluminum isopropoxide. In these syntheses, initial gel Si/Al ratios were maintained above 50; otherwise, RTH impurities were observed in the resultant products (25). Aging the synthesis gels over the course of 24 h was found to reduce crystallization times. A complete summary of the synthesis results is provided (SI Appendix, Tables S1–S3) along with representative powder X-ray diffraction (PXRD) patterns (SI Appendix, Fig. S3). Additionally, ¹⁹F, ²⁹Si, and ²⁷Al solid-state NMR (SS NMR) spectra, a representative thermogravimetric analysis, and a selection of scanning electron micrographs are listed as supporting characterizations (SI Appendix, Figs. S4–S8). In general, the concentrations of germanium and aluminum were found to increase in the product materials relative to the synthesis gel. Representative results from energy-dispersive X-ray spectroscopy are given in SI Appendix, Table S4 for products from germanosilicate and aluminogermanosilicate synthesis gels. In addition to STW, numerous other phases were produced without significant variation in synthesis conditions, including LTA, RTH, IWV, CSV, as well as several phases that could not be identified or consisted of layered organosilicate materials. Synthesis of these microporous materials is not unexpected, because they have all been shown to be derived from similarly shaped imidazole-based OSDAs under crystallization conditions similar to those reported here (25–28, 32).

As-made STW crystals obtained using 2 were analyzed by ¹³C cross-polarization magic-angle spinning SS NMR to evaluate whether the OSDA remains intact and directs the formation of STW. Fig. 3 compares the neat OSDA liquid ¹³C NMR spectrum (Fig. 3A) with those obtained from analyzing the occluded organic (from SS ¹³C NMR) and the organic recovered by dissolution of the framework structure (liquid ¹³C NMR). Agreement between all spectra indicates that 2 is occluded intact within the STW framework and is the structure-directing agent. As such, STW is not formed from an organic decomposition product or as a consequence of spontaneous, directed crystallization from the seed crystals. Efforts to perform chiral shift experiments on recovered OSDA 2 collected from dissolution of the STW framework by hydrofluoric acid were not successful due to solubility issues (2, chiral shift reagent, and solvent combinations).

Although it has been speculated that chiral OSDAs (when directed toward a chiral structure) should necessarily lead to an enrichment in the chirality of framework, this concept has never been conclusively proven. To do so, the chirality of the organic occluded within the STW structure must be known, as must the chirality of the framework structure for a bulk, polycrystalline sample. CD provides an effective method of accomplishing the former for as-made STW samples. The results for R-, S-, and racemic STW samples are given in Fig. 4. As shown by the data provided in Fig. 4, the molecules occluded within R- and S-STW absorb light at 242 nm with equal and opposite polarities. No measured rotation of light is observed for the racemic sample because, statistically, any polarization effects are negated for a sample containing equivalent quantities of chiral crystals. These data collectively demonstrate that 2 remains chiral and enantiopure within the STW framework. Although it may be possible to analyze the structural chirality of a calcined STW structure, no adsorption in the 200- to 300-nm range was detected from our germanosilicate samples. Incorporation of UV-active heteroatoms beyond those that have been used in this work may provide a pathway for future investigation via CD studies.

In previous reports involving the enrichment of polymorph A in *BEA (which does not guarantee enantioenrichment), characterization of any true structural enrichment of polymorph A has been limited to transmission electron microscopy (TEM) analysis (PXRD analyses are fraught with problems) of the polymorph domains and demonstration of enantioenrichment by measuring some function (e.g., catalysis or adsorption) (15, 16). A distinct issue with this methodology, however, is that the synthesis of these materials lacks proper controls with regard to the enantiomorphs obtained in the product materials, and therefore it is not possible to perform appropriate analytical and experimental controls to confirm chirality. The methods we have developed and used here are not subject to these problems. We are able to control the synthesis of a chiral OSDA (and obtain the R-, S-, and racemic forms thereof) to yield bulk, polycrystalline STW samples composed of individual crystals of a single chirality. As such, characterization and functional results that are capable of probing the bulk chirality of the two enantioenriched samples of opposing direction must demonstrate enantiomeric excess (ee) that is approximately equal and opposite, with the racemic preparation revealing no ee. The CD experiments for the R-, S-, and racemic STW samples provide an illustration of how these types of controls can be exploited to derive meaningful conclusions.

Characterization of the structural chirality of STW is complicated. An examination of the literature yields few methods for effectively determining this property for siliceous microporous materials beyond the demonstration of enantiomeric excesses derived from reactions or adsorption experiments. Analytical difficulty partially stems from the small crystal sizes (as is the case for STW synthesized with OSDAs used here) that prohibit the use of single-crystal XRD. Rojas et al. (31) point toward the possibility of examining the optical activity to quantify enantiomeric excess within a bulk, polycrystalline sample. SEM and high-resolution TEM (HRTEM) have proven to be highly advantageous methods for analyzing crystallite morphologies. HRTEM has been used previously to analyze the polymorph domains in *BEA (16, 20). However, specific chiral space groups (i.e., structural enantiomers) within the detected polymorph A domains were not able to be determined. Fundamentally, the difficulty in unequivocally distinguishing between the space groups of a chiral material using traditional TEM techniques arises from properties inherent to both structural chirality and experimental limitations. Specifically, on a given 2D plane (as is observed in a typical TEM experiment) the structures of the two enantiomers are superimposable and indistinguishable. A method was therefore developed to perform 3D HRTEM experiments by Ma et al. (21) to effectively characterize chiral space groups. The projection of the STW framework shifts after rotation along the screw axis as schematically illustrated in Fig. 5A. The rotation direction can be either clockwise (to the right) or counterclockwise (to the left). A zeolite crystal [that has been deposited with gold nanoparticles (diameter ∼5 nm) that serve as reference points with the microporous structure] is first aligned to [2$\overline{1}\overline{1}$0] zone axis and a through-focus series of HRTEM images are taken from a thin area. The crystal is then tilted continuously to the right or to the left by 30° about the screw axis. Upon rotation, the [1$\overline{1}$00] zone axis is observed if the tilting is clockwise. Based on simulated results, the crystal is right-handed if the shift direction is downward (Fig. 5 B and C). Upon rotation, the [1$\overline{1}$00] or [10$\overline{1}$0] zone axes are observed if the tilting direction is to the right or left, respectively. A series of through-focus images are then taken again along either zone axis. Two images along [2$\overline{1}\overline{1}$0] and [1$\overline{1}$00] are obtained by structure projection reconstructions using the through-focus series of HRTEM images and are aligned based on the positions of gold nanoparticles (Fig. 5 D–G and SI Appendix, Figs. S9–S11). By comparing the aligned images from the two zone axes there is an observable shift between the two projections from which the space group of the crystal can be assigned. Additional images are provided in SI Appendix.

The results collected from using this method to analyze STW crystals selected from bulk, polycrystalline samples synthesized using the R-, S-, and racemic versions of 2 are given in Table 2. Out of the six crystals analyzed for both the R- and S-OSDA–derived samples, five were determined to possess the P6₁22 and P6₅22 space groups, respectively, demonstrating (within this dataset) notable enantioenrichment. Moreover, these experimental results are consistent with computational predictions whereby R-OSDA 2 is expected to yield crystals with P6₁22 space group and, similarly, S-OSDA 2 is anticipated to result in crystals with the P6₅22 space group (Table 1). Analysis of the racemic sample resulted in an equal number of crystals from each space group, as expected.

Ideally, for a given sample synthesized using a chiral OSDA, HRTEM analysis would subsequently demonstrate only crystals from the expected space group. However, as shown in Table 2, a crystal from both chiral OSDA-derived samples was determined to be from the unexpected space group (e.g., P6₅22 from the R-OSDA). Possible reasons for incomplete purity include that (i) the crystal of the opposite space group was a seed crystal, (ii) there was OSDA degradation that occurred that yielded a racemic portion of the polycrystalline sample, and (iii) the inherent nature of the crystallization process does not synthesize an enantiopure, but only an enantioenriched, polycrystalline sample. Although currently the number of crystals that have been analyzed is not sufficient to draw a statistically conclusive picture of the bulk chirality of the samples, these HRTEM data (coupled with the distinct differences detected between R- and S-OSDA synthesized materials) are useful to demonstrate enantioenrichment of a bulk, polycrystalline sample of a molecular sieve.

CD and HRTEM characterizations demonstrate that a given enantiomer of OSDA 2 maintains its respective chirality in the occluded state and is capable of producing an enantioenriched, molecular sieve framework. Although there may exist crystallization processes wherein achiral molecules lead to enantioenrichment of product solids, such as by spontaneous chiral symmetry breaking by self-catalyzed crystallization (33), such systems would not allow for the directed synthesis of specific enantiomers of crystals, as we are able to demonstrate in this work. Thus, it follows that the initial hypothesis by Davis and Lobo (15) that any bulk, polycrystalline sample having enantioenrichment (not obtained solely from chiral seeds) must be derived from a chiral OSDA with a particular set of properties seems valid. Additionally, this synthesis methodology ensures that all forms (R-, S-, and racemic) are attainable, thus allowing for appropriate controls to be performed.

The chirality in STW is defined over the specific distance within the helical STW pore structure (Fig. 1) (34). As a consequence, adsorption and catalysis that show ee’s will involve molecules that are of sufficient size to effectively experience the chirality of the structure, yet still be able to pass through the limited size of the pores. As such, we expect that any measured ee’s will be greatly dependent on the selection of molecules used to test for a function. Moreover, the external surfaces of the crystallites may behave nonspecifically, resulting in diminishing ee’s.

To examine whether the enantioenriched STW samples are capable of performing enantioselective functions, catalysis and adsorption experiments were conducted. Epoxide ring-opening reactions on 1,2-epoxyalkanes were selected for catalysis, because they have been shown to take place within molecular sieves at relatively low temperatures (35–37). Such low-temperature probe reactions are particularly desirable because symmetry-breaking dispersive interactions with the catalyst surface are expected to be relatively large contributors to the transition state free energies compared with high-temperature scenarios. Reaction and analysis procedures are detailed in SI Appendix. The results from these reactions are reported in Table 3. For epoxides shorter than C₈, enantiomeric excesses that are not significantly different from experimental error are observed. However, as the chain length is further extended (to approximate the size of OSDA 2) the magnitude of the ee’s become significantly different from experimental error. Results show that approximately the same magnitude of the ee’s are obtained, but in opposite directions, from the R-STW and the S-STW, as should be. Trans-stilbene oxide was also used as a substrate and minimal ring opening was observed. Because this substrate is too large to enter the pores of STW, this control experiment demonstrates that the rate of reaction outside the pore structure (i.e., on the surface of the crystals) is negligible, and that any measured ee’s do not occur as a consequence of surface or solvent effects. No reactions were observed using uncalcined racemic STW.

Vaporsorption experiments were performed using the pure R-, S-, and racemic solutions of 2-butanol. Previous computational studies suggested that germanosilicate STW is capable of selectively adsorbing R- and S-glycidol into the helices of the respective enantiomers of the framework, with increasing discrimination between enantiomers from a racemic mixture with decreasing system temperatures (38). Experiments were therefore performed at 278 K (which allows for a vapor pressure great enough to collect sufficient adsorption data). Fig. 6 illustrates the adsorption isotherms obtained for R-, S-, and racemic 2-butanol in germanosilicate R- and S-STW. For R-STW, R-2-butanol is selectively adsorbed relative to S-2-butanol. The racemic 2-butanol isotherm lies at the approximate average of the isotherms from the enantiomerically enriched sample 2-butanol isotherms. An equivalent, but inverse, result is observed from S-STW.

A full list of chemicals used in this study and their sources can be found in SI Appendix. The full synthesis procedure for the chiral SDA used in this study as well as the respective ¹H and ¹³C NMR characterization data are described in detail in SI Appendix.

The hydrothermal synthesis procedures are given in SI Appendix, along with a comprehensive list of starting material composition ranges and the resultant products.

The powdered sample was crushed and dispersed into ethanol using ultrasonic processing. Several drops of the dispersion were dropped onto a carbon-film–supported TEM grid. Gold particles were deposited onto the sample through sputter coating.

Electron diffraction patterns and HRTEM images were obtained using a 200 kV using JEOL JEM-2100F (Cs = 0.5 mm, Cc = 1.1 mm). Each projection image was reconstructed from a through-focus series of 20 HRTEM images acquired with a constant focus step of 53 Å. The structure projection reconstruction was done using software Qfocus (41).

In a typical epoxide ring opening reaction, 20 mmol of an epoxide-functionalized substrate and 5 g of methanol (as a nucleophilic solvent) were added to 20 mg of Al-containing STW (Si/Al = 30 ± 10). Reactions were carried out at 323 K in thick-walled crimp-sealed glass reactors (VWR) submerged in a temperature-controlled oil bath. The enantiomeric excess of aliquots was determined by chiral gas chromatography. Detailed descriptions of procedures followed during reaction testing can be found in SI Appendix.