Introduction

Zeolites, with structural robustness and the advantage of low-cost production, have been widely applied in chemical industries as catalysts and sorbents1. Conventionally, molecular sieving in zeolites can be achieved due to their well-defined pore size. However, this process becomes more difficult when molecules in the mixture have similar kinetic diameters. Recently, unexpected separation performance was achieved for alkyne/olefin separation2,3 and CO2 sieving4,5 by tuning the pore interior of zeolites with appropriate counter ions, implying that the well-defined pore interior environment may also play a significant role in separation using zeolites. Moreover, studies found that the extra-framework silica (EFSi; Si(OH)x) in zeolite micropores could enhance interaction between the zeolite and organic alkyl/aryl groups to achieve a higher reaction rate of protolytic cracking of n-pentane6. The silanol nests in siliceous zeolite were found to efficiently enrich olefin molecules through hydrogen-π interactions to enhance the hydroformylation reaction on the adjacent rhodium nanoparticles7. Therefore, a separation strategy considering both the pore size and pore interior environment (especially silanols) may uncover previously unidentified separation properties for zeolite materials, helping to solve challenging industrial purification tasks (i.e., cyclohexane/benzene separation).

Cyclohexane, an important organic solvent with wide utilization, serves as essential feedstock for the industrial production of resins, nylon fibers and pharmaceutical intermediates8. The production of cyclohexane in industries mainly relies on the hydrogenation of benzene, with millions of tons of cyclohexane produced globally each year based on this reaction9. In addition, this reaction often remains incomplete and a purification process is needed before any further applications10. However, the differences in boiling points and molecular sizes of 5.96 Å and 6.08 Å11, respectively, between cyclohexane and benzene are very small12, with state-of-the-art separations in industries mostly based on azeotropic and extractive distillations, which are highly energy-consuming accompanied by a complex process and a high carbon footprint13,14,15. Thus, achieving efficient cyclohexane-benzene separation with a simple and low-carbon process remains a challenge. Recently, emerging sorbent materials16,17,18,19,20,21,22, notably metal-organic frameworks (MOFs)11,12,23,24,25,26,27,28, have shown a preferential adsorption ability in benzene/cyclohexane mixtures and an alternative adsorption-based purification ability for cyclohexane. Moreover, there is significant interest in how these materials could function in mild, low-cost operational conditions with the advantages of regeneration and reuse. However, these processes have yet to be commercialized due to the inherently limited stability and high production costs of MOFs. Additionally, the remarkable separation performance of MOFs has been mainly realized in the liquid phase, while a large portion of cyclohexane in industries is produced by the hydrogenation of benzene in the vapor phase10,29,30.

Herein, a strategy of combining pore size and silanols in an -SVR-type zeolite was proposed to achieve efficient dynamic benzene/cyclohexane separation (Fig. 1a). Specifically, SSZ-74 zeolite with an -SVR framework was synthesized using an organic template with a strong structure directing ability under fluoride-free conditions. The unique ordered silanols inside the zeolite demonstrated strong but reversible interactions with benzene through SiOH…π interactions, which enabled the complete removal of benzene from cyclohexane (benzene <1 ppm) under dynamic conditions.

Fig. 1: Strategy of molecular sieving through the synergy of pore size and silanols.
Fig. 1: Strategy of molecular sieving through the synergy of pore size and silanols.The alternative text for this image may have been generated using AI.
Full size image

a schematic representation of traditional size-dependent molecular sieving and combining pore size with silanols for separating molecules with different and similar sizes; b a portion of the -SVR framework structure showing the ordered silanols and 10-MR pores, along with details regarding the pore sizes and kinetic diameters of benzene and cyclohexane. The -SVR framework is shown in grey stick, the ordered silanols are labeled by red atoms and the pore apertures along a, b, c are marked by yellow, blue and pink, respectively.

Results

Fabrication and structural characterization of -SVR zeolite

An exploration of the zeolite framework database showed that most 10-member ring (MR) zeolites had a pore size of 5.4−6.4 Å31, which matched the kinetic diameters of benzene and cyclohexane. This investigation also revealed that the -SVR framework served as a suitable candidate for this purpose, as it contained three-dimensional channels with 10-member ring pore openings and ordered silicon vacancies32,33,34,35 (Supplementary Fig. 1). The pore size of 10-MR in the -SVR framework (5.7−5.9 Å) was close to the molecule sizes of benzene and cyclohexane (5.95 and 6.08 Å, respectively), and the ordered silanols around the 10-MR pore aperture were easily accessible (Fig. 1b). Subsequently, 1,6-bis(N, N-dimethylethylammonium)-hexane (denoted as 6BDMEA2+) was designed as an organic structure directing agent (OSDA) (Supplementary Fig. 2) for the synthesis of the SSZ-74 zeolite with -SVR framework (Fig. 1b). The stabilization energy of several reported/designed OSDAs was calculated by molecular dynamics (MD) simulations to evaluate their structure directing ability. Compared to reported OSDAs (Supplementary Fig. 3), 6BDMEA2+ had stronger stabilization energy, indicating that it was more suitable for the synthesis of SSZ-74 zeolite. High-crystalline products could be obtained with different gel compositions (Supplementary Fig. 4 and Supplementary Table 1) under fluoride-free conditions (yield, up to 80%) without the addition of seeds. Specifically, SSZ-74 zeolite synthesized with the silica-to-aluminum ratio (Si/Al) set 100 (SSZ-74-100) demonstrated good thermal stability. The powder X-ray diffraction (PXRD) patterns of SSZ-74-100 verified that the synthesized sample was highly crystallized and the crystallinity was preserved even after the removal of OSDA molecules through calcination (Supplementary Fig. 5). The relative intensities of reflections at low angles in the PXRD patterns increased due to the removal of organic templates, and the -SVR framework possibly suffered from partial structure collapse under heating conditions due to the presence of ordered silanols35. Moreover, the N2 sorption isotherm showed a type IV curve (Supplementary Fig. 6), with Brunauer-Emmett-Teller (BET) surface area, external surface area, and micropore volume of 415 m2/g, 42 m2/g, and 0.15 cm3/g, respectively, suggesting the maintenance of most micropores and the existence of few intracrystalline mesopores36 (Supplementary Table 2).

As demonstrated in the scanning electron microscope (SEM) and transmission electron microscope (TEM) images, the samples displayed a uniform nanoplate-like morphology with a lateral size of around 500 nm (Supplementary Fig. 7). Three-dimensional electron diffraction data (3D ED) were collected from the as-made and calcined crystals, from which monoclinic unit cells were determined to as a = 20.14 Å, b = 13.13 Å, c = 19.45 Å and β = 103.4°. The reflection conditions extracted from the 3D ED data (Supplementary Fig. 8) suggested two possible space groups, Cc or C2/c, and the framework structures of the as-made and calcined samples could be successfully solved using the Cc space group (Supplementary Figs. 9, 10 and Supplementary Table 3, 4). All the Si and O atoms in the framework could be resolved, along with ordered silanols. The crystal structure was also verified using high-resolution annual dark field imaging utilizing the Cs-corrected scanning transmission electron microscope (Fig. 2b). Furthermore, to accurately verify the position of the organic template in zeolite, Rietveld refinement against the PXRD data of SSZ-74-100 was performed (Fig. 2a and Supplementary Table 5). The obtained difference Fourier map revealed that the residual electron density fit the approximate shape of the organic template37 (Supplementary Fig. 11). Based on this, the positions of the organic template molecules were determined to be located at the center of the undulating 10-MR channels along the a-axis by applying a simulated annealing algorithm (Supplementary Fig. 12). The final refinement results showed good fit with the experimental data, with agreement factors of Rwp = 0.0277 and Rp = 0.0204. This also implied that ~3.6 OSDA molecules were located in each SSZ-74 zeolite unit cell, which was consistent with the thermogravimetric analysis results (Supplementary Fig. 13; ~3.9 OSDA molecules per unit cell) and calculated structures from the MD simulations (Supplementary Fig. 3). Moreover, Rietveld refinement against the PXRD data of calcined SSZ-74-100 was also performed (Supplementary Fig. 14, 15 and Supplementary Table 6), where the agreement factors of the final refinement results were Rwp = 0.0709 and Rp = 0.0522.

Fig. 2: Structural characterization of the -SVR zeolite.
Fig. 2: Structural characterization of the -SVR zeolite.The alternative text for this image may have been generated using AI.
Full size image

a Rietveld refinement of the as-made SSZ-74-100 sample against the PXRD data; b Cs-corrected annual dark field image of the as-made SSZ-74-100 sample, overlapped with the projection of -SVR framework along the a-axis; c 1H MAS NMR spectrum of the calcined SSZ-74-100 sample; d 29Si MAS NMR and CP/MAS NMR spectra of the calcined SSZ-74-100 sample. e 2D 1H-1H DQ MAS NMR spectrum of the calcined SSZ-74-100 sample.

The 13C magic-angle-spinning (MAS) nuclear magnetic resonance (NMR) spectrum (Supplementary Fig. 16) also provided evidence for the existence of OSDA inside the as-made sample. The silanols in SSZ-74 zeolite were also investigated by the 1H and 29Si MAS NMR spectra (Fig. 2c, d). In addition to the terminal silanols at ~1.9 ppm, the calcined sample showed obvious and main 1H signals at 2.3−6.6 ppm, which could be assigned to various silanols with different hydrogen bond interactions7, serving as a typical feature of silanol nests. Two-dimensional (2D) 1H-1H double-quantum (DQ) MAS NMR experiment was further performed to verify the spatial proximity of various silanols in the calcined sample. As shown in Fig. 2e, the autocorrelation signal at (2.3, 4.6) ppm on the diagonal indicated that the silanols with weaker hydrogen bond interactions were in close proximity to each other. A similar phenomenon was also observed for silanols with stronger hydrogen bond interactions, as demonstrated by the appearance of the autocorrelation signal at (3.6, 7.2) ppm. In addition, the off-diagonal peak pair at (6.6, 9.4) and (2.8, 9.4) ppm indicated a spatial correlation between the silanols with different hydrogen bond interactions. These results confirmed the complex hydrogen bond network of silanol nests in calcined SSZ-74. Furthermore, the 29Si MAS NMR spectrum assisted in the rough evaluation of high Q3 percentage (Q3/(Q3 + Q4) ~15.1%) (Supplementary Fig. 17). Comparison between the 29Si MAS NMR and 29Si cross-polarization (CP)/MAS NMR spectra of the calcined sample in Fig. 2d indicated that abundant hydrogen bonded silanols could enhance the Q3 signals in the 29Si CP/MAS NMR spectra38. The above characterization results suggested that most of the micropores in the calcined SSZ-74-100 sample were preserved, and the ordered silanols inside were well-maintained and possibly existed in the form of silanol nests, therefore, further investigations about its performance could rely on these features.

Benzene and cyclohexane separation performance

Several zeolites (Supplementary Figs. 18, 19 and Supplementary Table 7) with different framework structures together with SSZ-74 were investigated to illustrate the synergetic effect of pore size and host-guest interaction on the separation performance. These consisted of the 8-MR zeolite including 4 A (LTA framework) and high silica SSZ-13 (CHA framework), 10-MR zeolite including high silica ZSM-22 (TON framework), ferrierite (FER framework) and ZSM-5 with and without abundant defects (ZSM-5-OH/ZSM-5-F, MFI framework), and 12-MR zeolite including Beta (*BEA framework), and high silica Y and NaY (FAU framework). Moreover, the sample prepared by treating SSZ-74 zeolite with (NH4)SiF6 solution (SSZ-74-Si) was included to appraise the effect of silanols. The adsorption capacities of benzene and cyclohexane for these zeolites were evaluated through the vapor adsorption isotherms of benzene and cyclohexane (Supplementary Fig. 20 and Supplementary Table 6). At 298 K, SSZ-74-100 showed visibly higher adsorption capacity of benzene than cyclohexane at a P/P0 of 0.01 (0.95 against 0.57 mmol/g) and 0.9 (2.12 against 1.49 mmol/g), demonstrating its potential for capturing trace benzene from cyclohexane. In addition, ZSM-5-OH, ZSM-5-F, SSZ-74-Si, Beta, and NaY possessed high adsorption capacities of benzene both at P/P0 of 0.01 and 0.9, while ferrierite and ZSM-22 only showed a higher adsorption capacity of benzene at low pressure and high silica Y showed high adsorption only at high pressure. Meanwhile, ferrierite and ZSM-22 showed a lower benzene capacity than other 10-MR zeolites, which was possibly due to their smaller micropore volumes (Supplementary Table 2). The micropores of the 4 A and SSZ-13 zeolites were too small for capturing benzene and cyclohexane molecules, therefore, they mainly adsorbed these molecules at high pressure on the outer surfaces. Notably, SSZ-74-Si possessed lower adsorption capacities of benzene and cyclohexane compared to SSZ-74-100, verifying that the presence of silanols played an important role in the adsorption capacity.

The ability of SSZ-74-100 and other zeolites to separate benzene/cyclohexane mixtures under dynamic conditions was evaluated by breakthrough experiments (Fig. 3a, Supplementary Fig. 21 and Supplementary Table 8). According to the breakthrough curves, the 4 A, high silica SSZ-13 (8-MR) and Beta (12-MR) zeolites showed essentially no separation ability. All 10-MR zeolites showed a certain dynamic uptake of benzene and could produce 95% pure cyclohexane steams (breakthrough point) at the outlet of the fixed beds at 298 K. The dynamic benzene uptake calculated from the breakthrough curves of ZSM-22 and ferrierite zeolite (0.346 and 0.118 mmol/g) were significantly smaller than other 10-MR zeolites (> 0.7 mmol/g), similar to their vapor adsorption results, and both zeolites presented low 95% pure cyclohexane mass-based productivity (denoted as 95%-productivity, related to the retention time). These results suggested that the impact of benzene capacity and dynamic uptake was significant for the separation performance, especially for 95%-productivity, which was also verified by comparing with high silica Y and NaY zeolites (Fig. 3c). Figure 3b shows the dynamic benzene/cyclohexane selectivity of typical zeolite samples, among which SSZ-74-100 exhibited a high selectivity of 9.48. Notably, all 10-MR samples (except for ferrierite zeolite) possessed a selectivity of > 2, and in the remaining zeolites, only NaY demonstrated a similar selectivity (~2.27) with ZSM-22 (~2.21), indicating that the 10-MR pore aperture was more suitable than 8- and 12-MR for enhancing the purification efficiency. Moreover, the Beta and high silica Y zeolites showed large uptake for both benzene and cyclohexane, while their dynamic selectivity for benzene was close to 1, suggesting that they are good sorbents instead of separation agents. In addition, the ZSM-5-F demonstrated the highest selectivity ~26, which was possibly caused by its quite low cyclohexane capacity (Supplementary Fig. 20g). Notably, the NaY zeolite showed the highest 95%-productivity and dynamic uptake, and the 95%-productivity was two or three times larger than that of SSZ-74-100 or ZSM-5-OH. While for high silica Y zeolite, its 95%-productivity and dynamic uptake were much lower. This was consistent with previous reports, which demonstrated that the interaction between cations in NaY with π bonds in benzene could enhance the separation of the benzene/cyclohexane mixture39. Notably, the 95%-productivity of silanol-less SSZ-74-Si and ZSM-5-F was lower than that of silanol-abundant SSZ-74-100 and ZSM-5-OH, respectively, even though they demonstrated similar dynamic uptake. This could be attributed to the absence of silanol nest inside the zeolite. Therefore, even with certain dynamic selectivity, sole physisorption without strong interaction between the sorbents and benzene could not guarantee the efficient purification of the cyclohexane stream. These results experimentally implied the practicability of efficient benzene/cyclohexane separation mainly based on the synergy of pore size and silanols.

Fig. 3: Separation performance of different zeolites.
Fig. 3: Separation performance of different zeolites.The alternative text for this image may have been generated using AI.
Full size image

a breakthrough curves for the recycle benzene/cyclohexane separation experiments of SSZ-74-100; b dynamic selectivity of different zeolite samples; c comparison of benzene dynamic uptake with 95% pure cyclohexane mass-based productivity for different zeolite samples; d comparison of benzene relative uptake with ultrapure cyclohexane mass-based productivity for different zeolite samples.

Critically, in terms of the production of ultrapure cyclohexane steam (benzene <1 ppm), the silanol nests inside the zeolites demonstrated the best performance (Fig. 3d) among the test samples. The SSZ-74-100 sample showed the highest ultrapure mass-based productivity of cyclohexane (denoted as ultrapure-productivity) under dynamic conditions (16.3 L/kg) with a high relative uptake of ~90.45%. By contrast, NaY with the highest 95%-productivity was unable to produce ultrapure cyclohexane (~10−7 L/kg), and possessed a notable low relative uptake. Meanwhile, samples with certain ultrapure-productivity all consisted of 10-MR zeolites with high relative uptake, suggesting that the suitable pore apertures possibly enhanced the ultrapure purification ability. Notably, the relative uptake of SSZ-74-100 was comparable to MOFs with excellent performance (Supplementary Fig. 22 and Supplementary Table 9). Moreover, the ultrapure-productivity of ZSM-5-F (0.49 L/kg) was far below that of ZSM-5-OH (8.3 L/kg), which further suggested that silanol nests inside sample could enhance the separation ability, especially for ultrapure-productivity, as it was difficult for benzene to breakthrough due to the presence of strong interactions. Notably, the ultrapure-productivity of SSZ-74-100 was about two times greater than that of ZSM-5-OH, even though the samples contained similar silanol amounts, indicating that the ordered silanols showed advantages over the disordered ones as they are located in the 10-MR channels, and they might be more readily accessed by guest molecules. These results further revealed the potential of SSZ-74 zeolites for the dynamically adsorptive removal of benzene to produce ultrapure cyclohexane from industrial benzene/cyclohexane streams, and the synergy of suitable pore size and silanols inside channels was of great significance to excellent performance. In addition, recyclability remains a requirement for practical sorbents. After four cycles of benzene/cyclohexane separations with the SSZ-74-100 zeolite, no obvious decline was observed in the retention time, and the sorbents could be regenerated at 473 K after each cycle (Fig. 3a and Supplementary Fig. 23). A breakthrough experiment using a benzene/cyclohexane mixture (molar ratio 1:99) was also performed on SSZ-74-100 (Supplementary Fig. 24), where a dynamic selectivity of 75.2 and an ultrapure cyclohexane mass-based productivity of 140.8 L/kg were obtained. This further confirmed the synergistic effect of pore size and that the silanols works well for this separation process.

Mechanism discussion

Previous reports indicated that the cations in NaY could interact with the π bonds in benzene and further enable efficient separation39,40,41. Moreover, hydrogen-π interactions between the silanol nests inside zeolites with olefin molecules were found in catalysis reactions6,7,42, which implied possible interactions between the ordered silanols and π bonds of benzene. In this study, Fourier transform infrared (FTIR) analysis was applied to SSZ-74-100, SSZ-74-Si, ZSM-5-OH and ZSM-5-F samples to investigate the interactions between the zeolites and benzene. Supplementary Fig. 25 presents the FTIR spectra of these samples, which confirmed the state of the silanols. SSZ-74-100 exhibited signals at 3742 and 3724 cm−1 and around 3500 cm−1, and these signals could be assigned to the terminal silanols (3742 cm−1) and silanol nest (3724 and 3500 cm−1) inside zeolite43,44, respectively, which was consistent with the above NMR characterization results. For the SSZ-74-Si sample, a much lower Q3 percentage (Q3/(Q3 + Q4) ~ 6.2%) in the 29Si MAS NMR spectra (Supplementary Fig. 26) suggested a decrease in silanols. The repairing of silanol nests in SSZ-74-Si could be also confirmed by the FTIR spectra with the disappearance of a broad signal around 3500 cm−1. Signals at 3662 cm−1 in the FTIR and 27Al MAS NMR spectra (Supplementary Fig. 27) revealed the formation of extra-framework Al after repair. ZSM-5-OH showed 3742, 3724 cm−1 signals and a broad signal around 3460 cm−1, indicating the existence of terminal silanols and silanol nests7. This was also confirmed by the 1H and 29Si MAS NMR spectra (Supplementary Fig. 28). In comparison, ZSM-5-F only showed signals at 3742 cm−1 belonging to terminal silanols, with low silanol content, as confirmed by the 29Si MAS NMR spectra (Supplementary Fig. 29). Thereafter, in situ FTIR spectra were acquired to trace the adsorption and desorption processes of each sample. Figure 4a presents the in situ benzene-adsorption FTIR spectra of SSZ-74-100, in which a slight red shift was observed for broad signals around 3500 cm−1 during the adsorption process. Furthermore, the intensity of the signal at 3724 cm−1 gradually decreased with benzene capture, due to the interactions between the silanol nests with benzene molecules. The redshifts for the broad signal around 3460 and 3724 cm−1 signals were also observed in the benzene-adsorption FTIR spectra of the ZSM-5-OH samples, while for the SSZ-74-Si and ZSM-5-F samples, only the intensity around 3000 cm−1 increased, which was assigned to the adsorption of benzene (Supplementary Fig. 30). The in situ FTIR spectra of the desorption process were also acquired at different temperatures to assess the interaction strength between zeolites and benzene (Fig. 4b and Supplementary Fig. 31). The intensities of signals at 1479 cm−1 assigned to the C = C stretching vibration of adsorbed benzene was used to estimate the interaction strength. The desorption process of SSZ-74-100 completely ended at 200−250 °C, while the desorption temperature for SSZ-74-Si is 130−150 °C. The higher desorption temperature suggested stronger interactions between the SSZ-74-100 framework and benzene. Similarly, the desorption temperature of ZSM-5-OH (250−300 °C) was higher than ZSM-5-F (130−150 °C). These results confirmed that the existence of silanols inside zeolite could enhance its interaction with benzene molecules.

Fig. 4: Investigation of the separation mechanism.
Fig. 4: Investigation of the separation mechanism.The alternative text for this image may have been generated using AI.
Full size image

a in situ FTIR spectra for the benzene adsorption process of SSZ-74-100; b in situ FTIR spectra for the benzene desorption process of SSZ-74-100 at different temperatures; c projection of -SVR framework with benzene from DFT calculation along [010], and a portion of the structure showing the benzene position, with the -SVR framework shown as grey sticks, the ordered silanols labeled by red atoms, and the green molecule representing benzene; d distribution of the minimum C…H distance between the guest molecules and silanols in -SVR framework.

The interaction energy between the -SVR framework and guest molecules (i.e., benzene and cyclohexane) was also evaluated based on DFT calculations45,46,47 (Fig. 4c, Supplementary Fig. 32, 33; Supplementary Table 10, 11). The results suggested that the adsorption energies were −96.6 and −86.2 kJ/mol for benzene and cyclohexane, respectively. Moreover, as shown in Fig. 4c, the absorption sites of guest molecules might be around the ordered silanols, which implied that the stronger interactions between benzene and ordered silanols resulted in preferential absorption. Furthermore, to clarify the impact of ordered silanols in SSZ-74 zeolite on separating benzene from cyclohexane, ab initio molecular dynamics (AIMD) simulations were carried out (Fig. 4d). The results showed that the minimum C…H distance between the H atoms in silanols and C atoms in benzene (2.0−2.5 Å) was shorter than that in cyclohexane (2.2−3.5 Å), which further suggested that the silanols demonstrated a stronger interaction with benzene molecules. The relatively slower diffusion of benzene due to stronger interactions could enable the separation of benzene/cyclohexane mixture within the -SVR zeolite framework. Moreover, the undulating channels in the -SVR framework might benefit the access of benzene to the silanols and enhance beneficial adsorption as the benzene possessed a planar structure while the cyclohexane possessed a chair/boat conformation48. The above results further validated our strategy, where the interaction between the ordered silanols inside zeolite and benzene was beneficial to the superior separation performance of zeolite.

Discussion

In summary, we reported the case of achieving efficient separation of benzene/cyclohexane based the on synergy of pore size and silanols in zeolites. A highly crystalline SSZ-74 zeolite was synthesized using a designed OSDA with strong structure directing ability under fluoride-free conditions. Breakthrough experiments showed that the SSZ-74-100 sample enabled the efficient complete removal of benzene from cyclohexane (benzene <1 ppm) under dynamic conditions with the highest ultrapure-productivity (~2 times that of the widely used ZSM-5), reinforcing its potential in direct vapor phase purification for industrial cyclohexane. Along with the in situ FTIR spectra, DFT calculations and AIMD simulations verified the strong and reversible interactions between the silanols inside SSZ-74 zeolite and benzene. Therefore, this study suggested that excellent separation performance could be realized for zeolite materials by combining the host-guest interactions together with a suitable pore channel.

Methods

Chemicals and reactants

Colloidal silica (Ludox AS-40, 40 wt.% suspension in H2O, Sigma-Aldrich), tetraethylorthosilicate (TEOS, 98%, Sinopharm Chemical Reagent Co., Ltd.), fumed silica (Silica, Shanghai Macklin Biochemical Co., Ltd.), N,N-dimethylethylamine (C4H11N, 98%, TCI Shanghai), 1,6-dibromohexane (C6H12Br2, 97%, Shanghai Aladdin Biochemical Technology Co., Ltd.), tetrapropylammonium hydroxide aqueous solution (TPAOH, 40 wt.%, Sinopharm Chemical Reagent Co., Ltd.), tetrapropylammonium bromide (TPABr, 98%, Aladdin Chemistry Co., Ltd.), N,N,N-Trimethyl-1-adamantammonium hydroxide aqueous solution (TMAdaOH, 25 wt.%, Bide Pharmatech Co., Ltd.), sodium hydroxide (NaOH, 98%, Shanghai Titan Scientific Co., Ltd.), sodium aluminate (NaAlO2, AR, Shanghai Titan Scientific Co., Ltd.), sodium chloride (NaCl, 99.5%, Aladdin Chemistry Co., Ltd.), ammonium fluoride (NH4F, 96%, Sinopharm Chemical Reagent Co., Ltd.), ammomium fluosilicate ((NH4)2SiF6, 99.7%, Sinopharm Chemical Reagent Co., Ltd.), acetonitrile (CH3CN, 99%, Sinopharm Chemical Reagent Co., Ltd.), ethyl ether (C4H10O, 99.7%, Sinopharm Chemical Reagent Co., Ltd.) and deionized water obtained from pure-water equipment are used in this work.

4 A zeolite (LTA framework structure), ZSM-5 zeolite with abundant defects (MFI framework structure, ZSM-5-OH), Ferrierite zeolite (FER framework structure), ZSM-22 zeolite (TON framework structure), NaY zeolite (FAU framework structure) and Beta zeolite (*BEA framework structure) are purchased from Nankai University Catalyst Co., Ltd.

Preparation of zeolites

Synthesis of organic template 1,6-bis (N, N-dimethylethylammonium)-hexane hydroxide

In the typical synthesis of 1,6-bis (N, N-dimethylethylammonium)-hexane bromide [6BDMEA(Br)2], 74 g of N, N-dimethylethylamine and 244 g of 1,6-dibromohexane (molar ratio ~ 2.2) are dissolved in 800 mL acetonitrile and then refluxed overnight at 60 °C. Then, white powder are filtered and washed with ethyl ether twice to remove most of the solvent and the unreacted N, N-dimethylethylamine. After drying under vacuum with A-tube, pure 6BDMEA(Br)2 could be obtained successfully, which is confirmed by 1H and 13C NMR spectra in Supplementary Fig. 2. Subsequently, the 6BDMEA(Br)2 is converted into hydroxide form 6BDMEA(OH)2 using hydroxide exchange resin in water, and the obtained solution is titrated against 0.1 M HCl aqueous solution.

Synthesis of SSZ-74 zeolite with -SVR framework structure using 6BDMEA(OH)2

In a typical synthesis of SSZ-74 zeolite with -SVR framework structure using 6BDMEA(OH)2 as organic template, 2.5 g of Ludox, 0.04-0.132 g of NaAlO2, 0.067-0.134 g of NaOH, 7.066 g of 6BDMEA(OH)2 (25 wt.%) and 5.23 g of deionized water are mixed and stirred for 2 h. The molar compositions of the synthetic gel are 1.0 SiO2: 0.01-0.033 NaAlO2: 0.2 6BDMEA(OH)2: 0.05-0.1 NaOH: 22.5 H2O. After stirring, the gel is sealed in an autoclave and heated at 160 °C for 168 h under rotation conditions (50 rpm). The sample is obtained from filtrating and washing 2 times with deionized water. After drying at 80 °C overnight, the sample is calcined at 550 °C for 4 h to remove the organic template. The SSZ-74 zeolite obtained from the gel composition of Run 18 in Supplementary Table 1 is denoted as SSZ-74-100 and used for the most of characterizations. Moreover, synthesis with different molar compositions is investigated as well, the gel compositions are 1.0 SiO2: 0-0.067 NaAlO2: 0.05–0.6 6BDMEA(OH)2: 0.15 NaOH: 10-40 H2O. Similarly, the gels are heated at 160 °C for 168 h under rotation condition (50 rpm), filtrated and washed 2 times with deionized water, and dried at 80 °C overnight to obtain the final products. Details about the gel compositions and products are listed in Supplementary Table 1.

Synthesis of repaired SSZ-74 zeolite

In a typical synthesis of repaired SSZ-74 zeolite, 1 g of calcined SSZ-74-100 zeolite material is dispersed in 40 mL of 0.01 mol/L (NH4)2SiF6 aqueous solution, and then the repairing procedure is carried out at 80 °C for 72 h with stirring49. The product is filtrated and washed 2 times with deionized water. Thereafter, it is dried at 80 °C overnight and calcined at 550 °C for 5 h. The repaired sample is denoted as SSZ-74-Si.

Synthesis of ZSM-5 zeolite under fluoride condition

In a typical synthesis of ZSM-5 zeolite under fluoride condition50, 3 g of fumed silica, 0.059 g of NaAlO2, 0.08 g of TPABr, 1.85 g of NH4F and 18 g of deionized water are mixed and stirred for 2 h. The molar compositions of the synthetic gel are 1.0 SiO2: 0.01 NaAlO2: 0.08 TPABr: 1.0 NH4F: 20 H2O. After stirring, the mixture is sealed in an autoclave and heated at 180 °C for 48 h. The sample is obtained from filtrating and washing 2 times with deionized water. After drying at 80 °C overnight, the sample is calcined at 550 °C for 5 h to remove the organic template. The obtained ZSM-5 zeolite is denoted as ZSM-5-F.

Synthesis of high silica SSZ-13 zeolite

The high silica SSZ-13 zeolite is obtained under fluoride-free condition with the seed-assisted aging treatment51. 2.5 g of Ludox, 0.02 g of NaAlO2, 2.83 g of TMAdaOH (25 wt.%), 0.134 g of NaOH, 5.4 g of deionized water and 0.02 g SSZ-13 zeolite seed (2 wt.% to SiO2) are mixed and stirred for 12 h. The molar compositions of the synthetic gel are 1.0 SiO2: 0.01 NaAlO2: 0.2 TMAdaOH: 0.2 NaOH: 30 H2O. After stirring, the gel is sealed in an autoclave and heated at 170 °C for 48 h under rotation condition (40 rpm). The sample is obtained from filtrating and washing 2 times with deionized water. After drying at 80 °C overnight, the sample is calcined at 600 °C for 6 h to remove the organic template.

Ion exchange

The ZSM-5 zeolite with abundant defects (ZSM-5-OH) and high silica Y zeolite are ion exchanged with 1 M NaCl solution twice at 80 °C for 2 h and then calcined at 550 °C for 5 h.

Characterizations

Powder X-ray diffraction (PXRD) patterns are measured with Bruker D8 Advanced X-ray diffractometer equipped with a Pixel detector using Cu Kα radiation (λ = 1.5418 Å, 40 kV, 40 mA). The samples are studied in the 5–40° 2θ range with a scanning step of 0.02°. Scanning electron microscopy (SEM) images are taken on JEOL JSM-7800F Prime instrument. Transmission electron microscope (TEM) image and electron diffraction (ED) pattern are taken on JEOL JEM-F200 instrument with an acceleration voltage of 200 kV. High-resolution Cs-corrected scanning transmission electron microscope (STEM) images are acquired using the JEOL GrandARM 300 F equipped with double correctors operated at 300 kV. Aberrations are calculated and corrected using the Ronchigram from the amorphous carbon on the TEM grids, and 2D difference filter is applied to process images. The N2 sorption isotherms and vapor sorption isotherms are measured on BELSORP-MAXII instruments and all the samples are degassed at 400 °C under vacuum for at least 2 h. Fourier transform infrared (FTIR) analysis was performed on a Perkin-Elmer Spectrum TM GX instrument. Prior to the measurement, the self-supporting wafer (ca. 10 mg) of the catalysts was activated in an in situ cell under a vacuum (<10−3 Pa) at 400 °C for 60 min, and FTIR spectrums were collected after the samples were cooled to 150 °C. For benzene-FTIR experiments, the samples were pretreated in situ at 400 °C for 120 min under evacuation (<10−3 Pa), then cooled to room temperature where a certain amount of benzene vapor was injected into the in situ cell, the pressure of benzene vapor was gradually increased to the maximum pressure. Then, the physically adsorbed benzene was removed by evacuating for 30 min and the samples were heated in vacuum. The test temperature was gradually increased from 30 to 400 °C at a heating rate of 1 °C/min. All FTIR spectra were collected as an average of 64 scans at 4 cm−1 resolutions. Thermal gravimetric analysis (TGA) is obtained on PerkinElmer TGA 4000 instruments under a N2 flow rate of 20 mL/min with a ramp rate of 10 °C/min. 1D 1H Magic Angle Spinning (MAS) NMR measurements were conducted using a Bruker Avance III 400 WB spectrometer, operating at a 1H resonance frequency of 399.33 MHz. The 1H MAS NMR spectra were acquired with a 4 mm probe, employing a spinning rate of 12 kHz, a π/2 pulse length of 3.8 μs, and a recycle delay of 5 s. In the 1H MAS NMR studies, the samples underwent further dehydration in a vacuum at 400 °C for 10 hours. The solid-state 13C, 27Al and 29Si NMR spectra were recorded on a Bruker ADVANCE 400 MHz Solid NMR spectrometer, and the samples used for 29Si MAS NMR were degassed at 200 °C for 8 h. The solution NMR spectra are recorded with a Bruker Avance-500 MHz spectrometer. 2D 1H-1H double quantum-single quantum (DQ) MAS NMR experiments, DQ coherences were excited and reconverted with a POST-C7 pulse sequence, and the excitation and reconversion time were set to 0.48 ms. The increment interval in the indirect dimension was set to 80 μs, 72 t1 increments, and 180 scan accumulations for each t1 increment were used. The elemental analysis is determined by inductively coupled plasma atomic emission spectrometry (ICP-AES) on a Thermo IRIS Intrepid II XSP atomic emission spectrometer after dissolving the samples in HF solution.

For the structure analysis, three-dimensional electron diffraction (3D ED) datasets are collected on JEOL JEM-F200 (200 kV) equipped with a hybrid-pixel electron detector (Cheetah 1800, 512 × 512 pixels, pixel size 55 μm, Amsterdam Scientific Instruments) and the software Instamatic52 with modification. 3D ED data are processed with XDS software53 to get unit cell parameters and reflection intensity information. The ab initio structure solutions are obtained with SIR2014 software54 and the structure refinements are performed in Olex2 packages55 with SHELXL program56. The reconstruction of 3D reciprocal space is processed by the software EDT-Process. PXRD data for Rietveld refinement is acquired on SmartLab X-ray diffractometer with D/teX Ultra250 detector using Cu Kα1 radiation (λ = 1.5406 Å) in a transmission mode (capillary: 0.5 mm, angle range: 5–115°, step size: 0.01°, total counting time: 20 h). As made SSZ-74-100 sample is degassed at 200 °C for 6 h before sealed in capillary. The calcined SSZ-74-100 sample is degassed at 400 °C for 4 h, and then sealed in capillary after adsorbing benzene. The Rietveld refinements are performed with the commercial program TOPAS V6.0. The background of PXRD patterns is removed manually using the tool Lines (www.github.com/stefsmeets/lines). Structure refinement is initiated using the structural model of -SVR reported from the International Zeolite Association (www.iza-structure.org/databases).

Computational method

Stabilization energy, which is defined as the difference between the energy of the zeolite-template complex with the energy of templates and empty zeolite framework (Equation below), has been widely utilized to predict new templates for the synthesis of zeolites46,47,57,58. Therefore, the stabilization energy of several reported33,35/designed templates is calculated to evaluate their directing role and to find out possible templates for synthesizing SSZ-74 zeolite. The SSZ-74 zeolite model with the -SVR framework structure, is retrieved from the IZA database (www.iza-structure.org/databases) and treated as a neutral pure silica framework. The terminal oxygen atoms on the silanol are saturated with hydrogen atoms. The -SVR framework, organic templates, as well as their complexes are firstly optimized under the constant volume condition and then performed molecular dynamics (MD) simulation using the Dreiding forcefield59 taking advantage of the General Utility Lattice Program (GULP) package60. The reported energy values are taken as the average energy calculated from the last 5 ps of a 30 ps MD simulation with a timestep of 0.5 fs at 433 K (the synthesis temperature of SSZ-74 zeolites) in the NVT ensemble.

$${{\rm{Stabilization}}}\; {{\rm{energy}}}= [{{\rm{E}}}({{\rm{zeolite}}}\_{{\rm{n}}}\; {{\rm{template}}})-{{\rm{E}}}({{\rm{zeolite}}})\\ -{n}{{\rm{E}}}({{\rm{template}}})]/(n\times {NT})$$

where E(zeolite_n template) is the energy of the zeolite-template complex, n is the number of templates in -SVR framework and is equal to 4 for all the zeolite-template complexes, E(zeolite) and E(template) are the energy of empty -SVR framework and isolated template, respectively. NT is the number of Si atoms and is equal to 92 in -SVR framework.

In order to gain insight into the host-guest interaction between -SVR framework and the guest molecules (i.e., benzene and cyclohexane), the interaction energy is calculated based on periodic density functional theory (DFT) using the Vienna Ab-initio Simulation Package (VASP)61,62. The initial structures for optimization were taken from the ab initio molecular dynamics (AIMD) trajectories every 10 ps (i.e., configurations at 10 ps, 20 ps, and 30 ps), and the strongest interaction was selected to present. Planewaves are constructed with a kinetic energy cutoff of 400 eV using the projector augmented wave (PAW) method63,64. The 1 × 2 × 1 k-point mesh is used and centered at the Γ point. All calculations are performed using the Perdew-Burke-Ernzerhof (PBE) generalized gradient approximation (GGA)65 in conjunction with Grimme’s D3 correction66 to account for dispersion interaction. The self-consistent field (SCF) electronic energies and atomic forces are converged to 10−5 eV and to less than 0.02 eV/Å, respectively. Furthermore, to clarify the impact of silanol in SSZ-74 zeolite on separating benzene and cyclohexane molecules intuitively, the AIMD simulations are carried out by the CP2K package67,68,69 with Gaussian plane wave basis sets (GPW)70,71. The AIMD simulation is performed in the NVT ensemble for 30 ps with a timestep of 0.5 fs. The first 5 ps is used for equilibrium and the last 25 ps is applied for calculating the distribution of C…H distances. The temperature is held at 298 K and controlled by the Nosé-Hoover thermostat72,73 with a coupling time constant of 0.1 ps. The revPBE functional is chosen for its improved catalytic ability to describe energies compared with the commonly used PBE functional for solid-state calculations74. The D3 correction is also adopted, just like in DFT calculations. The double-zeta valence polarized (DZVP) basis set and the Goedecker-Teter-Hutter (GTH) pseudopotentials75 are employed in the AIMD simulations, and the plane wave cutoff energy and the relative cutoff is 400 Ry and 60 Ry, respectively. The trajectories are recorded at every step.

Breakthrough experiments

Breakthrough experiments are measured using BSD-MAB multi-component adsorption breakthrough curve analyzer. The sample is mixed with quartz wool and loaded into an 8 ml penetrating column. Samples are activated 200 °C for 4 h with He purging. The test is carried out with feed gases (Benzene/Cyclohexane/He, 2/2/96) at flow rate of 50 sccm. The outlet gas is analyzed by using a mass spectrometer.

The dynamic selectivity is calculated as the ratio of adsorbed vapors in the sorbents from the breakthrough experiments:

$${{\rm{Dynamic}}}\; {{\rm{selectivity}}}=\frac{{{\rm{Adsorbed}}}\; {{\rm{benzene}}}}{{{\rm{Adsorbed}}}\; {{\rm{cyclohexane}}}}$$

The relative uptake is calculated as the percentage of benzene in the adsorbed vapors in the sorbents from the breakthrough experiments:

$${{\rm{Relative}}}\; {{\rm{uptake}}}=\frac{{{\rm{Adsorbed}}}\; {{\rm{benzene}}}}{{{\rm{Adsorbed}}}\; {{\rm{benzene}}}\; {{\rm{and}}}\; {{\rm{cyclohexane}}}}\times 100\%$$

Mass-based cyclohexane productivity is defined as the breakthrough amount of cyclohexane (L) over an adsorption bed packed with 1 kg of adsorbent. The mass-based productivity is calculated by integrating the breakthrough curves in the period from t1 to t2, during which the cyclohexane purity is kept higher than a threshold value of 95% (breakthrough point) or 99.9999% (ultrapure, cyclohexane <1 ppm). The productivity of cyclohexane is calculated as follows:

$${{\rm{Mass}}}-{{\rm{based}}}\; {{\rm{productivity}}}\; {{\rm{of}}}\; {{\rm{benzene}}}({{\rm{L}}}/{{\rm{kg}}})=\frac{{\int }_{{t}_{1}}^{{t}_{2}}\,f\, \left(t\right)\, {dt}}{{M}_{{sorbent}}}$$

where f(t) is the outlet flow rate of cyclohexane calculated by multiplying the time-dependent C/C0 and the gas flow rate recorded using a mass flow meter.