View all hidden authors and organizations

Edited by Howard A. Stone, Princeton University, Princeton, New Jersey, approved on February 23, 2021 (review received on December 1, 2020)

Xylene is an important raw material for manufacturing packaging materials, general chemicals, industrial solvents, etc. The purification of xylene isomers is one of the most important but energy-intensive separations of organic mixtures in the chemical industry. We use carbon molecular sieve (CMS) membranes derived from spirobifluorene-based polymers (PIM-SBF) to achieve the separation of xylene isomers, which may reduce energy consumption, carbon emissions and equipment footprint. CMS membrane is a solvent-resistant and high-temperature resistant material, and can withstand high transmembrane pressure when it is made into a hollow fiber form. The new CMS membrane produced here shows competitive performance with the most advanced zeolite under high xylene loading, and its development provides basic insights and guidance for the manipulation of the CMS pore structure.

Inherent microporous (PIM) polymers have been used as precursors for the preparation of porous carbon molecular sieve (CMS) membranes. PIM-1 is a typical PIM material, which uses a fused ring structure to increase the chain stiffness between the spirobisindene repeating units. These two factors inhibit effective chain accumulation, resulting in high free volume within the membrane. However, the reduction of pore size and porosity was observed after pyrolysis of PIM-1 into CMS membrane, which was attributed to the destruction of the screw core, which resulted in the "flattening" of the polymer main chain and the graphite-like carbon chain. accumulation. Here, a spirobifluorene-based intrinsically microporous polymer (PIM-SBF) was synthesized and used to make a CMS membrane, which showed a significant increase in p-xylene permeability (about four times), p-xylene/o-xylene There is almost no loss of xylene selectivity (13.4 vs. 14.7). Compared with PIM-1 derived CMS membrane, equimolar xylene vapor is separated. This work shows that it is feasible to manufacture such a highly microporous CMS membrane, whose performance exceeds the current state-of-the-art zeolite under high xylene loading.

Xylene is a chemical raw material widely used in the production of solvents and synthetic polymers. P-xylene (p-xylene), with an annual global production of 29 million tons (1), is an important raw material for materials such as polyester and polyethylene terephthalate. O-xylene (o-xylene) can be converted into phthalic anhydride, an important plasticizer precursor (2), while m-xylene (m-xylene) can be converted into isophthalic acid (3), A precursor of polyethylene terephthalate. Xylene is mainly produced by catalytic reforming, which converts the naphtha distillate into a liquid rich in octane (4), called reformate, which is an important resource for aromatic compounds. Although benzene, ethylbenzene, and toluene are easy to separate, xylene isomers are difficult to separate by conventional distillation because their atmospheric boiling points are similar: o-xylene is 144 °C, meta-xylene is 139 °C, and meta-xylene is 139 °C. Xylene is p-xylene at 138 °C. The number of theoretical plates required to separate xylene to commercial specifications exceeds 360, which is not feasible (5).

The most advanced separation technology for xylene isomers is fractional crystallization and adsorption. The former is separated according to the different freezing points of xylene isomers: o-xylene is -25 °C, meta-xylene is -48 °C, and p-xylene is 13 °C. Although fractional crystallization accounts for about 25% of the separation of p-xylene, it has two main disadvantages (6, 7): due to the eutectic point and economic considerations in the operation of the crystallizer, the recovery rate of crystalline p-xylene is about 60 % To 70%, and the required cooling makes the process energy intensive. Therefore, in most commercial projects, crystallization is only suitable for streams with a paraxylene concentration of more than 80% (5). The adsorption separation of xylene isomers has higher efficiency and lower energy loss than crystallization. It is carried out on an industrial scale on a simulated moving bed (SMB), which was developed by Universal Oil Products (UOP) in 1960. Technology developed in the 19's. A typical SMB unit uses faujasite (FAU) zeolite as the adsorbent and operates at around 180 °C. The recovery rate of p-xylene is 97% to 99%, and the purity is 99.7% to 99.9%. It should be noted that there is usually an additional distillation separation between p-xylene and p-diethylbenzene (the desorbent in the SMB process). To further improve these energy-intensive and expensive methods, membranes using materials such as polymers, silica gel, zeolite, and metal organic framework (MOF) have been explored for xylene separation.

Polycrystalline zeolite MFI membranes have been extensively studied, usually using pervaporation or vapor permeation separation methods (8, 9). From a basic point of view, these zeolites are unlikely to be surpassed in the separation of xylene isomers because their pore structure can be precisely adjusted to the task. For example, Lai and colleagues optimized the microstructure of the zeolite Socony Mobil-5 (ZSM-5) zeolite membrane for xylene pervaporation (9), providing oriented ZSM-5 with a para-xylene/o-xylene separation factor of up to 400 The permeability of the membrane and p-xylene is as high as 3 × 10−7 mol/m2-s-Pa. However, these materials generally cannot maintain an ideal structure in practice: the MFI framework will undergo structural deformation when adsorbing xylene molecules, especially near ambient temperature and high xylene loading. This deformation will cause the phase transition of the MFI crystal from the orthogonal phase (ORTHO) to the second orthogonal phase (PARA), making the structure unable to distinguish the xylene isomers and reducing the separation efficiency of the membrane (10, 11). This problem Coupled with the low xylene permeability under the high fractional occupancy rate of guest molecules, it indicates that the MFI zeolite membrane will be difficult to provide satisfactory xylene isomer separation under the full load conditions that may be required in industrial processes. In addition, these membranes require expensive supports and are difficult to produce on a large scale (12). Although these long-standing problems can be solved in principle, they hinder the practical application of MFI membranes in xylene separation, despite their outstanding performance in the laboratory.

Nanoporous carbon molecular sieve (CMS) materials are produced by pyrolyzing well-defined polymer precursors under controlled temperature and atmosphere (13⇓ ⇓ –16). CMS membrane has solvent resistance and temperature resistance, and can withstand high transmembrane pressure when it is made into a hollow fiber form (17⇓ –19). Hollow fibers can be made into "asymmetric membranes," which have a thin separation layer and can transition to more pore structures that provide mechanical support. Usually, this membrane asymmetry is produced in a single-step phase inversion process. The asymmetric structure is critical to achieving high product permeability while providing mechanical integrity to the membrane and bypassing the problem of creating a defect-free layer on the membrane support. We have recently shown that asymmetric CMS hollow fiber membranes derived from cross-linked polyvinylidene fluoride can separate xylene isomers in a manner called "Organic Solvent Reverse Osmosis" (OSRO) (20). Although CMS membranes have advantages in scalability and resistance to actual operating conditions, existing materials exhibit lower para-xylene/o-xylene selectivity and lower performance than zeolite. Therefore, we have turned our attention to the production of CMS films with improved properties by changing the polymer precursor (21) and processing its pyrolysis conditions.

Under high temperature inert atmosphere, polymer chains are activated by pyrolysis and rearrange into a stable, highly carbonized structure. Although the precise details of the molecular details of this process are unknown, the resulting structure may have short-range order in the form of well-defined micropore spaces, the production of which is driven by entropy (22). This local ordering can be designed to distinguish certain pairs of molecules. Polymers with high free volume or interconnected micropores [for example, polyimide (18, 21, 23, 24), PIM-1 (16, 19, 25), functionalized polyimide with inherent micropores Amines (PIM-PI) (26⇓ ⇓ ⇓ ⇓ ⇓ –32) etc.] tend to form highly porous CMS materials. As demonstrated by previous work (16, 19, 25), PIM-1 (Figure 1A) is a successful precursor for the production of highly porous CMS membranes that can be used for organic solvent separation. Its rigid, twisted random coil structure is composed of screw-core biaromatic monomers connected by fused ring connectors, which results in high free volume in the membrane by inhibiting effective chain accumulation (33⇓ ⇓ ⇓ ⇓ –38). However, it was found that the size of the ultra-micropores inside the PIM-1 derived CMS is very similar to N2 (3.64 Å), which severely limits the transmission rate of p-xylene (16). Previous work has shown that CMS derived from PIM-1 is pyrolyzed in a reducing environment (4% H2/Ar) to produce 5 to 7 Å ultramicropores that intersect the size of xylene molecules (5.8 to 6.8 Å) (25) . Nevertheless, compared with the polymer precursor, the pore size and porosity of the PIM-1-derived CMS were still observed to decrease after H2-containing pyrolysis (16, 25).

Reaction scheme and material characterization. (A) The reaction scheme for the synthesis of PIM-1. (B) The reaction scheme for the synthesis of PIM-SBF. (C) Digital photos of PIM-SBF polymer film and (D) PIM-SBF-derived CMS dense film. (E) SEM cross-sectional image of PIM-SBF-derived CMS dense film. (F) Pore size distribution measured by nitrogen physical adsorption at 77 K.

We believe that this undesirable collapse of the pores may be due to the destruction of the spiro carbon center, which will lead to the "flattening" of the carbonaceous chain produced by the pyrolysis reaction. Consistent with the previous suggestion that more rigid polymer chains provide better performance for polymer membranes (ie enhanced permeability and permeability selectivity between gas molecules) (39, 40), we believe that the spiral core structure The greater rigidity and thermal stability in the unit may prevent chain flattening and subsequent collapse of pores during pyrolysis, resulting in the formation of CMS membranes with better separation properties. We describe here the successful application of this principle.

With a rotatable CC bond in the indane unit, the spiral center of the PIM-1 spirobisindan monomer has a certain degree of conformational flexibility (41). McKeown and colleagues demonstrated that the use of spirobifluorene variants improves gas separation performance, presumably due to the replacement of flexible indan with a rigid aromatic part (42). Here, we show that this spirobifluorene-based intrinsically microporous polymer (PIM-SBF, Figure 1B) is an attractive polymer precursor for CMS membrane manufacturing. Two properties of polymers are believed to contribute to this result. First, we believe that compared with the sp3 center of PIM-1, the high aromaticity of the spirobifluorene structure facilitates the formation of twisted carbonized chains that cannot be stacked together efficiently, and the latter may tend to break or allow the chains to move. Secondly, compared with the 490 °C of PIM-1, the thermogravimetric analysis of PIM-SBF at 580 °C (SI Appendix, Figure S1) shows the onset of significant degradation, which may help prevent pyrolysis. The first derivative of the thermogravimetric analysis (DTGA) curve (SI appendix, Figure S1) highlights the slight degradation of PIM-SBF, with an onset temperature as low as 400 °C, indicating that PIM-SBF may be rearranged by "pre-pyrolysis" The stage is at a low pyrolysis temperature (for example, 500°C), but is significantly pyrolyzed at a higher temperature. The PIM-SBF-derived CMS membrane has higher performance than other CMS membranes in the separation of xylene isomers, and is competitive with the characteristics of zeolite, especially in the case of high xylene loading.

The intrinsic transmission characteristics of adsorption-diffusion membranes are described by two main parameters: "permeability", a measurement of intrinsic productivity, and "selectivity", a measurement of separation efficiency. For single-component permeation, the permeability (ℙA) is equal to the ratio of the thickness-normalized flux to the transmembrane fugacity: ΔfA is the difference in transmembrane fugacity. In the adsorption-diffusion transport mechanism, guest molecules are adsorbed to the upstream side of the membrane due to the presence of a chemical potential gradient, diffuse through the membrane, and desorb on the downstream side. Permeability can be expressed as the product of DA, transmission diffusion coefficient and SA, that is, solubility or adsorption coefficient: ℙA=DA×SA. [2] The adsorption coefficient SA is a thermodynamic factor, which is mainly controlled by the condensability of the gas permeant and the interaction of the membrane permeant. The diffusion coefficient DA is a kinetic characteristic, which is related to the ability of guest molecules to "jump" in the membrane; in small and microporous membranes, the diffusion rate is well described by the transition state theory (43).

The ideal permeability selectivity αAB of guest molecule A relative to B reflects the separation efficiency of the membrane, which is defined as the ratio of the permeability of the fast component to the slow component when a vacuum is applied downstream. The main factors of selectivity can be defined using the adsorption-diffusion model, which shows that the permeation selectivity is the product of the diffusion selectivity DA/DB and the adsorption selectivity SA/SB: αAB=ℙAℙB=(DADB)×(SASB). [3]

The pore structure of the PIM-SBF polymer precursor and the corresponding PIM-SBF-derived CMS membrane (Figure 1 CE) were characterized by nitrogen physical adsorption experiments at 77 K. As shown in Figure S2 of the SI Appendix, the final pyrolysis temperatures (ie 500, 800 and 1,100 °C) were studied at a fixed volume fraction of hydrogen in the 4 vol% H2 pyrolysis gas, and the two hydrogen volume fractions (ie 0 and 4 vol% H2) The final pyrolysis temperature was 500 °C under the following conditions. As shown in the SI appendix, Figure S2 A and C, the PIM-SBF precursor shows a sharp absorption in the low pressure range, followed by a linear increase due to adsorption-induced polymer expansion (44⇓ –46). The apparent lack of swelling in the PIM-SBF-derived CMS sample (SI Appendix, Figure S2 B and D) indicates that the CMS sample has a more rigid structure than the polymer precursor, as expected. As shown in Table S1 of the SI Appendix, the PIM-SBF-derived CMS membrane has a high surface area (366 to 855 m2/g) and a large pore volume (0.153 to 0.380 cm3/g). It is worth noting that some PIM-SBF-derived CMS membranes (ie CMS_PIM-SBF_500 °C_4% H2) have been observed to have increased surface area and pore volume relative to the precursor. These results are consistent with our hypothesis that a more rigid polymer precursor can prevent the polymer chains from flattening and pores from collapsing to a certain extent during pyrolysis. It is important that the Brunauer-Emmett-Teller (BET) surface area and pore volume of PIM-SBF-derived CMS are larger than the value of PIM-1 derived CMS manufactured under the same pyrolysis conditions. These higher pore volumes are advantageous because they contribute to the high permeability (ie flux or throughput) of the CMS membrane.

The pore size distribution of the polymer precursor and CMS membrane was derived from the nitrogen isotherm at 77 K using the two-dimensional non-local density functional theory (2D-NLDFT) method. The pore size distribution curve is illustrated using two different y-axis scales so that both the ultramicropores and micropores are visible. As shown in Figure 1F, PIM-SBF has 5 to 8 Å ultra-micropores. After pyrolysis (under pure argon or 4% H2/Ar), the "medium-sized" ultra-micropores (ie 5 to 7 Å) remain in the CMS membrane, which will result in effective interaction between organic matter of appropriate size Molecular separation. Solvent molecules (for example, xylene isomers). It is worth noting that the reasonable nitrogen physical adsorption isotherm of CMS_ PIM-1 _500 °C_0% H2 at 77 K cannot be obtained, which indicates that the ultrafine pore size in the CMS is very similar to that of N2 (3.64 Å), resulting in extremely Slow N2 diffusion (16). In contrast, CMS_PIM-SBF _500 °C_0% H2 showed ultra-micropores ranging from 5 to 7 Å, indicating that it is feasible to manufacture CMS membranes with "medium-sized" micropores without reducing the environment. Avoiding hydrogen species in the pyrolysis environment will make the pyrolysis process safer and reduce the complexity of the overall manufacturing process.

Table 1 summarizes the full width at half maximum (FWHM), average pore size, and micropore and ultramicropore volume of PIM-SBF-derived CMS. The distribution of ultra-micropores is narrower, and the average ultra-micropore size is smaller. The hydrogen content is reduced from 4 vol% to 0 vol%. By comparing the pore size distribution of PIM-SBF-derived CMS samples under pyrolysis at 500, 800 and 1100 °C, it can be concluded that the average size of ultra-micropores decreases with the increase of pyrolysis temperature, indicating that CMS tightens the matrix At a higher pyrolysis temperature. The micropore volume in the PIM-SBF-derived CMS membrane increases with the decrease of pyrolysis temperature or the increase of hydrogen content. As shown in Table 1, the pyrolysis of PIM-SBF at relatively low temperatures (≤800°C) will result in the preparation of CMS membranes with large ultra-micropore volumes (0.230 to 0.280 cm3/g). However, very high pyrolysis temperatures (for example, 1,100 °C) can cause a decrease in the volume of micropores and ultramicropores. It is worth noting that the ultra-micropore distribution of CMS_PIM-SBF_500 °C_4% H2 is narrow (FWHM is 1.30 Å vs. 2.69 Å), but compared to the average ultra-micropore size of CMS_PIM-1_500 °C_4% (7.1 Å vs. 5.6 Å) H2. These two effects compete with each other (that is, a larger pore size will result in a decrease in selectivity, but a tighter pore size distribution will increase selectivity), so it is difficult to a priori estimate changes in precursors based on changes in membrane performance. As will be shown later, while the permeability increases sharply, the selectivity remains basically unchanged.

Compared with CMS_PIM-1 _500 °C_4% H2, the full width at half maximum, average pore diameter, micropore and ultramicropore volume of PIM-SBF derived CMS formed under different conditions

X-ray photoelectron spectroscopy (XPS) was used to further characterize the CMS film to study the carbon bonding properties in the film. The C1s spectra of different CMS samples (SI Appendix, Figure S3) can be deconvolved into three Gaussian peaks. As shown by the reduced square root of χ2 being less than 3 and the coefficient of determination R2 being higher than 0.99 for all CMS samples, a good fit was obtained. The two most important peaks with a relative binding energy distance of about 1 eV correspond to two different hybridization states. The signal with higher binding energy is related to sp3 hybrid carbon, while the signal with an energy shift of about 1 eV is attributed to sp2 hybrid carbon (47⇓ –49). In addition, the peak observed near 289 eV indicates the presence of the carbon state of CO (50). The sp3 hybrid carbon is a three-dimensional structure that can destroy the carbonaceous plate accumulation and contribute to high guest molecule flux. The sp2 hybrid carbon (a two-dimensional graphite layered structure) enables the plates to be stacked to form a more compact microstructure with smaller ultra-micropore gaps in the plates (25). Our previous studies have shown that in PIM-1 derived CMS, a higher sp3/sp2 hybridized carbon ratio results in a more permeable but less selective structure (25). In this study, the sp2 and sp3 hybrid carbon content in each PIM-SBF-derived CMS sample can also be estimated by the peak area ratio of its XPS spectrum. As shown in Table 1, similar to the results of PIM-1-CMS, the sp3/sp2 carbon ratio in PIM-SBF-CMS also increases with the decrease of pyrolysis temperature or the increase of hydrogen concentration. It is worth noting that the sp3/sp2 carbon ratio of PIM-SBF-derived CMS is higher than that of PIM-1-derived CMS manufactured under the same pyrolysis conditions. The higher sp3/sp2 carbon ratio in PIM-SBF-CMS helps to give the CMS membrane a higher flux of guest molecules. The supplementary N1s XPS, Fourier Transform Infrared Spectroscopy (FTIR) and Raman spectroscopy of PIM-SBF and 500 °C pyrolysis CMS_PIM-SBF samples and their respective analyses can be found in the SI appendix, Figure 1 and Figure 2. S4-S6.

Collect the adsorption isotherms of CMS_PIM-SBF_500 °C_4% H2 and CMS_PIM-1_500 °C_4% H2 at 55 °C. As shown in Figure 2 A and B, all isotherms show that the adsorption level in the low saturation region increases sharply, and then stabilizes at the higher saturation value. As expected, for the same type of CMS membrane, the adsorption capacity of p-xylene under each pressure condition is very similar to the adsorption capacity of o-xylene (the difference in adsorption capacity is within 5 wt%) because they have similar chemistry And physical properties, which indicate that there is no adsorption selective separation mechanism in the CMS membrane. It is worth noting that the adsorption capacity of the CMS_PIM-SBF_500 °C_4% H2 membrane of p-xylene and o-xylene is a little higher (that is, about 1.2 times higher) than the adsorption capacity of CMS_PIM-1_500 °C_4% H2 at all relative pressures. Below, shows the porous structure in the case of spirobifluorene. This observation is very consistent with the measurement results of nitrogen physical adsorption (Table 1), which indicates that the pore volume of CMS_PIM-SBF_500 °C_4% H2 is higher (0.043 cm3/g and 0.034 cm3/g) compared to CMS_PIM-1_500 ° C_4% H2.

The adsorption and diffusion characteristics of CMS. The single-component adsorption isotherms of p-xylene and o-xylene in (A) CMS_PIM-1_500 °C_4% H2 and (B) CMS_PIM-SBF_500 °C_4% H2 were measured at 55 °C. (C) Transmission diffusion coefficients of xylene isomers in CMS_PIM-1_500 °C_4% H2 and CMS_PIM-SBF_500 °C_4% H2.

Although p-xylene and o-xylene have similar adsorption characteristics in PIM-SBF CMS, the “medium-sized” ultra-micropores (ie 5 to 7 Å) in the rigid CMS membrane can realize diffusion-based xylene isomer molecules Separation. The single-component kinetic absorption curve (SI appendix, Figure S7) was determined by the vapor adsorption analysis method and used to estimate the transmission diffusion coefficient D of xylene isomers in different CMS samples. The transmission diffusion coefficients of xylene isomers in CMS_PIM-1_500 °C_4% H2 and CMS_PIM-SBF_500 °C_4% H2 and the transmission and diffusion selectivity of these two CMS materials between p-xylene and o-xylene are shown in Figure 2C. Show. In the same pyrolysis environment, the transmission diffusion coefficients of p-xylene and o-xylene in the PIM-SBF-derived CMS were significantly increased compared to the PIM-1-derived CMS sample (4.0 × 10-9 vs. 1.0 × 10-9 Xylene is cm2/s, 2.3 × 10−10 for o-xylene is 4.0 × 10−11 cm2/s). Consistent with our observations on nitrogen physical adsorption measurement, the pore volume of CMS_PIM-SBF_500 °C_4% H2 is much higher than that of CMS_PIM-1_500 ° (0.161 cm3/g vs. 0.380 cm3/g, as shown in Table S1 in the SI Appendix) C_4 % H2, which will cause the guest molecules to diffuse faster through the PIM-SBF-derived CMS. It is important that CMS_PIM-SBF_500 °C_4% H2 exhibits a smaller diffusion selectivity (17.4 vs. 25.0) compared to CMS_PIM-1_500 °C_4% H2, which is mainly due to the larger average ultramicropore size (7.1 Å vs. 5.6 Å).

The Wicke-Kallenbach osmosis device was used to test the separation performance of the CMS membrane, in which the total pressure difference on both sides of the membrane was kept at zero. The feed is an equimolar p-xylene/o-xylene mixture vapor carried by nitrogen, which is flushed upstream, while a nitrogen purge brings the permeate to the gas chromatograph to determine the xylene flux through the membrane. The influence of polymer precursors on the permeability of CMS membrane to xylene isomers is shown in Figure 3A. As shown in the figure, the p-xylene permeability of the CMS membrane derived from PIM-SBF is four times higher than that of the membrane derived from PIM-1. This is consistent with our characterization results, that is, the pore volume of PIM-SBF-derived CMS is larger than that of PIM-1-derived CMS manufactured under the same pyrolysis conditions (CMS_PIM-SBF_500 °C_4% H2 is 0.380 cm3/ g and 0.161 cm3/g of CMS_PIM-1_500 °C_4% H2, as shown in Table S1 in the SI Appendix). The larger pore volume in the spirobifluorene-based CMS membrane will create more diffusion pathways for guest molecules and increase the diffusion rate, which is ultimately beneficial to permeability. Unlike the permeability, the permeation selectivity of p-xylene/o-xylene has a much smaller change. This may be because the selective permeability is mainly determined by the ultra-micropores in the CMS membrane. CMS_PIM-SBF_500 °C_4% H2 and CMS_PIM-1_500 °C_4% H2 The inner ultra-micropore size is about 5 to 7 angstroms (note that the xylene isomers to be separated, p-xylene and o-xylene, have kinetic diameters ) Are 5.8 Å and 6.8 Å respectively). The smaller full width at half maximum (CMS_PIM-SBF_500 °C_4% H2 is 1.30 Å, and CMS_PIM-1_500 °C_4% H2 is 2.69 Å) can achieve high p-xylene/o-xylene permeation selectivity.

Permeation and separation of xylene isomers. (A) CMS_PIM-SBF_500 °C_4% H2 (blue square mark), CMS_PIM-SBF_500 °C_0% H2 (orange diamond mark), CMS_PIM-SBF_800 experiment (equimolar xylene vapor mixture Wicke-Kallenbach tested at 55 °C ) Separation performance °C_4% H2 (green left triangle mark), CMS_PIM-SBF_1100 °C_4% H2 (pink right triangle mark) and CMS_PIM-1_500 °C_4% H2 (red upper triangle mark). The Maxwell-Stefan model predicts the performance of CMS_PIM-SBF_500 °C_4% H2, using (marked by blue solid circles) and not using (marked by blue hollow circles) to consider the friction coupling effect. (B) The p-xylene/o-xylene separation performance of PIM-SBF-derived CMS as a function of sp3/sp2 hybrid carbon ratio, based on an equimolar xylene vapor mixture Wicke-Kallenbach test at 55°C. Draw lines to guide the eyes. (C) The p-xylene/o-xylene separation performance of PIM-1 derived CMS as a function of sp3/sp2 hybrid carbon ratio, based on the Wicke-Kallenbach test of an equimolar xylene vapor mixture at 55 °C (25). Draw lines to guide the eyes. (D) Estimate the p-xylene vapor permeability as a function of p-xylene feed pressure through MFI zeolite (silicalite-1) membrane and CMS_PIM-SBF_500 °C_4% H2 (operating temperature = 55 °C). CMS_PIM-SBF_500 °C_4% H2, penetrating p-xylene pressure = 0.1 Pa (red line) or 100 Pa (pink line). MFI zeolite (silicalite-1) membrane penetration p-xylene pressure = 0.1 Pa (black line) or 100 Pa (blue line).

Single-component adsorption and diffusion data are used as input to the Maxwell-Stefan (MS) model to predict the permeability of mixtures with and without molecular friction coupling effects (51). Here, the frictional coupling effect between xylene isomers is estimated using the Venus-type correlation (52, 53). Figure 3A also shows the comparison of the experimental results of the equimolar p-xylene/o-xylene vapor mixture separated by the dense CMS_PIM-SBF_500 °C_4% H2 membrane measured at 55°C through the Wicke-Kallenbach test and the prediction of the MS model (Detailed modeling parameters can be found in Table S2 of the SI Appendix. The experimental para-xylene permeability is slightly higher (1.05 times) than the predicted value of the MS model using the friction coupling effect, and it is observed that the result is higher than that predicted by the MS model without the coupling effect 60.6% lower. In this work, the maximum loading of xylene isomers in the CMS material is achieved by using the total pore volume of the membrane (measured by nitrogen physical adsorption test at 77 K) and the moles of xylene isomers However, this assumption may overestimate the xylene isomer adsorption capacity tested by Wicke-Kallenbach. The main reason is that the micropore volume contributes to the adsorption of xylene molecules, while the ultramicropore volume contributes little. This overestimation will lead to an overestimation of the predicted permeability of the model. The experimental p-xylene/o-xylene selectivity is between the selectivity predicted by the MS model, with and without coupling effects. The results show that the friction coupling is considered In the case of the effect, the loss of selectivity in the film is not as severe as predicted by the MS mixture. This indicates that the Venus-type correlation used to estimate the frictional coupling effect here does not accurately capture the degree of frictional coupling. Or, maybe There is a gradient of friction coupling effect, which is not considered here (ie, a constant "cross-coupling" diffusion coefficient Ð12 is used in these estimates). Such a gradient indicates that the xylene isomers are highly coupled on the highly active side of the membrane, while The low activity side is relatively uncoupled, which reasonably explains the difference between the two models and experiments.

The effect of the final pyrolysis temperature and the hydrogen concentration in the pyrolysis environment on the separation performance of the PIM-SBF-derived CMS membrane is also shown in Figure 3A. The results show that when the hydrogen concentration in the pyrolysis environment is increased from 0 vol% to 4 vol%, a higher p-xylene permeability and similar permeation selectivity are observed. This result is consistent with the measurement of nitrogen physical adsorption, that is, hydrogen will help create a CMS membrane with a higher BET surface area and larger pore volume. In addition, as expected, when the final pyrolysis temperature was increased from 500 °C to 1,100 °C, lower paraxylene permeability and higher permeation selectivity were observed.

As mentioned earlier, the sp3/sp2 hybrid carbon ratio of PIM-SBF-derived CMS is higher than that of PIM-1-derived CMS manufactured under the same pyrolysis conditions. We believe that the higher sp3/sp2 hybrid carbon ratio in PIM-SBF-CMS contributes to the high flux of guest molecules to the CMS membrane. Figure 3B illustrates the separation performance of the PIM-SBF-CMS membrane as a function of the sp3/sp2 hybrid carbon ratio. As shown in Figure 3B, as the sp3/sp2 hybrid carbon ratio increased from 0.29 to 0.89, the permeability of p-xylene through the PIM-SBF-CMS membrane increased significantly from 4.3 × 10-14 to 2.4 × 10-13 mol -m/m2-s-Pa (>5×, an increase of 458%), while the permeation selectivity was only slightly reduced from 17.9 to 13.4 (a decrease of 25%). This observation indicates that by adjusting the sp3/sp2 hybrid carbon ratio of the CMS membrane, the separation performance of the CMS membrane can be improved. Figure 3C shows the separation performance of the PIM-1-CMS membrane as a function of the sp3/sp2 hybrid carbon ratio (25). The response of p-xylene permeability to sp3/sp2 hybrid carbon ratio (Figure 3B and C) shows the difference between the CMS structures derived from PIM-1 and PIM-SBF. For PIM-SBF-derived CMS, the penetration promoting effect of sp3-rich carbon chains increases as the concentration of sp3-rich carbon chains increases. However, for PIM-1-derived CMS, the penetration promoting effect of sp3-rich carbon chains decreases as the concentration of sp3-rich carbon chains increases. Further research is needed to understand the source of this different reaction. We hypothesize that further destruction of the formation of carbonaceous plates by adding sp3 hybrid carbon inclusions will continue to increase the porosity of the polymer, but will eventually destroy the formation of well-defined ultra-micropores.

Compared with the most advanced zeolite membrane, CMS_PIM-SBF_500 °C_4% H2 is expected to show better performance in the actual separation of concentrated xylene mixtures. Figure 3D compares the theoretical p-xylene vapor permeability through the perfect MFI zeolite (silicalite-1) membrane (20) and the CMS_PIM-SBF_500 °C_4% H2 membrane as a function of the p-xylene feed pressure (the detailed performance evaluation process can be Found in the SI appendix). When the xylene diffusivity is constant, increasing the feed pressure of p-xylene will result in a decrease in the adsorption coefficient of p-xylene (determined by the nature of the Langmuir isotherm), and ultimately reduce the passage of p-xylene through the CMS_PIM-SBF_500 °C_4% H2 membrane The penetration rate. As shown in Figure 3D, when the p-xylene osmotic pressure is 0.1 Pa, the p-xylene permeability through CMS_PIM-SBF_500 °C_4% H2 gradually decreases from 2.1×10-2 to 4.4×10-6 mol/m2-s -Pa (four orders of magnitude) increases with the p-xylene feed pressure from 0.11 Pa to 5475.7 Pa (the saturation pressure of p-xylene at 55 °C). Unlike CMS_PIM-SBF_500 °C_4% H2, it has been shown that silicalite-1 nanopores have a strong restriction on p-xylene molecules, which leads to a decrease in the diffusivity of p-xylene with increasing p-xylene loading (54). In this case, the p-xylene permeability through the silicalite 1 membrane decreased significantly as the feed pressure increased. Specifically, when the p-xylene osmotic pressure is 0.1 Pa, the permeability of p-xylene through the silicalite-1 membrane drops sharply from 4.7×10-2 to 8.4×10-7 mol/m2-s-Pa (5 Orders of magnitude) as the feed pressure of paraxylene increases from 0.11 Pa to 5475.7 Pa. Even under low load conditions, silicalite-1 membrane exhibits a higher p-xylene permeability than CMS_PIM-SBF_500 °C_4% H2, but the sharp drop in diffusivity leads to a significant decrease in the passage of p-xylene through silicalite under high load- 1 The permeability of the membrane. When the p-xylene osmotic pressure increases to 100 Pa, under the same feed pressure, the p-xylene permeability of siliconite-1 membrane is about three orders of magnitude lower than that of CMS_PIM-SBF_500 °C_4% H2. Although both CMS and MFI membranes show significant permeation loss with increasing xylene loading, CMS materials are more successful in maintaining permeation levels relative to MFI. The experimental xylene vapor separation performance of PIM-SBF-CMS, PIM-1-CMS and the most advanced MFI zeolite membrane is also compared, as shown in Figure S8 in the SI Appendix. Compared with PIM-1-CMS, the p-xylene permeability of the PIM-SBF-derived CMS membrane is increased by about five times, while the sacrifice of p/o-xylene selectivity is negligible. Considering the p-xylene permeability and p-/o-xylene selectivity, the PIM-SBF-derived CMS shows comparable performance to silicalite-1 membrane under similar operating conditions.

The CMS membrane made by PIM-SBF under standard and hydropyrolysis conditions is significantly better than the membrane made by PIM-1 in the separation of xylene isomers. Although it is found that the hydrogen concentration in the pyrolysis atmosphere and the final pyrolysis temperature have a significant effect on the pore structure of the resulting CMS film, it is found that the spirobifluorene-based CMS material has a high surface area and pore volume and does not require a reducing atmosphere, which will ultimately simplify Manufacturing process. The optimized PIM-SBF-derived CMS membrane has a diffusion selectivity between p-xylene and o-xylene of about 17.4, which has important practical value and enhanced parameters in the Wicke-Kallenbach permeation experiment. The high permeability of PIM-SBF-derived CMS membranes is believed to be due to their larger pore structure (characterized by nitrogen physical adsorption), BET surface area and pore volume. The adsorption isotherms of o-xylene and p-xylene were also measured, indicating that there is no adsorption selective separation between xylene isomers in the PIM-SBF-derived CMS membrane. In addition, it was found that the theoretical p-xylene vapor permeability through the CMS membrane derived from PIM-SBF as a function of the p-xylene feed pressure at 500 °C and 4% H2 is higher than that of the MFI zeolite membrane for higher xylene loading More robust, indicating that CMS materials can be used for high-throughput separations.

The SI appendix provides detailed information about reagents, monomer synthesis, material characterization methods, organic adsorption tests, and Wicke-Kallenbach permeation measurements.

PIM-1 is synthesized using a low-temperature polycondensation method, as shown in Figure 1A (35). The two purified monomers tetrafluoroterephthalonitrile (TFTPN) and 5,5',6,6'-tetrahydroxy-3,3,3',3'-tetramethyl-1,1'-spiro Bisindene (TTSBI) is added to anhydrous dimethylformamide (DMF) in a round bottom flask in an equimolar ratio. After the monomer is completely dissolved, anhydrous highly crushed K2CO3 (2.5 molar equivalent times relative to TFTPN) is added to the solution. Then, the polymerization reaction was continuously stirred at 65°C for 72 hours under a nitrogen atmosphere. After the reaction, after cooling, the reaction was quenched with deionized water and the PIM-1 polymer was precipitated. The crude product was then collected by filtration and washed with additional deionized water to remove salts and residual solvents. Repeated reprecipitation from chloroform further purified the polymer. Finally, the fluorescent yellow PIM-1 polymer was vacuum dried at 70°C for 12 hours. Compared with polystyrene standards, the molecular weight determined by gel permeation chromatography (GPC) in tetrahydrofuran (THF) is Mn = 46,500 and the polydispersity index (PDI) = 1.5.

PIM-SBF is synthesized using standard PIM to form an aromatic nucleophilic substitution polymerization reaction, as shown in Figure 1B (35, 42). Combine 2,2',3,3'-tetrahydroxy-9,9'-spirobifluorene (4.18 g, 11 mmol, 1 equiv.) and tetrafluoroterephthalonitrile (2.2 g, 11 mmol, 1 equivalent) ). Dry DMF (55 mL) was added via a syringe, and the reaction mixture was stirred at room temperature under nitrogen until the two monomers were completely dissolved. Potassium carbonate (12.1 g, 88 mmol, 8 equivalents) was added in one portion, and the reaction mixture was stirred at 65°C for 92 hours. After completion, the reaction mixture was poured into water, filtered, and washed with water, methanol and acetone. The crude material was redissolved in chloroform, precipitated in 2:1 methanol:acetone, filtered, and dried in a vacuum oven at 100°C to obtain the title compound as a bright yellow powder. The weight average molecular weight measured by GPC in THF relative to the polystyrene standard is Mw = 54,000 and PDI = 1.5. There are also a large number of higher molecular weight parts (as shown in Figure S9 in the SI Appendix). The Mw of this material observed by GPC is 721,701 kDa, which we think is unrealistic. Instead, we assume that this material is hyperbranched and is caused by the reaction of tetrafluoroterephthalonitrile with trace amounts of water or potassium carbonate (55). The molecular structures of all chemical substances involved in the synthesis of PIM-SBF are shown in Table S3 in the SI Appendix, and the synthesis details are provided in the SI Appendix.

The dried PIM-SBF or PIM-1 was dissolved in chloroform to form a 10 wt% polymer solution, and placed on a drum at room temperature for 6 hours to form a uniform polymer solution. The resulting solution was then used to prepare a polymer film by a solution casting method at room temperature. The glove bag (Glas-Col) containing the polymer solution vial, glass plate, spatula, and beaker containing excess chloroform was placed in a fume hood before the film was cast. After the glove bag was sealed, it was saturated with chloroform for 5 h and purged with nitrogen 3 times. Then, the polymer solution was transferred from the vial to a glass plate and cast into a uniform film. Subsequently, as the chloroform slowly evaporates in the glove bag for 3 days, the film solidifies and then vacuum-dried at 70°C for another 24 hours.

The CMS membrane is made by pyrolyzing a polymeric precursor (PIM-SBF or PIM-1) in a furnace located in a fume hood, as shown in the SI appendix, Figure S10. The polymer precursor is first placed on a stainless steel mesh plate, placed in a quartz tube, and then loaded into a three-zone tube furnace (OTF-1200X-III-S-UL, MTI Corporation). The sealing of the quartz tube is ensured by a pair of SS 304 vacuum flanges with double high temperature silicone O-rings. A hydrogen (4 vol%)/argon mixed gas cylinder is used to provide a pyrolysis environment containing hydrogen. Before pyrolysis, the entire system is purged with the required gas mixture for at least 12 hours, until the oxygen concentration in the tube furnace is less than 0.5 ppm, by the online oxygen analyzer (R1100-ZF Rapidox 1100ZF, CEA Instruments, Inc.) Measurement (25). For safety reasons, if the hydrogen concentration in the fume hood exceeds 8,000 ppm, the surface-mounted hydrogen detector will be triggered. The heating protocol used is described in Table S4 in the SI Appendix.

The Wicke-Kallenbach osmosis device was used to test the separation performance of the CMS membrane, in which the total pressure difference on both sides of the membrane was kept at zero. The feed is an equimolar p-xylene/o-xylene mixture vapor carried by nitrogen, which is flushed upstream, while a nitrogen purge brings the permeate to the gas chromatograph to determine the xylene flux through the membrane. The free-standing dense CMS membrane is fixed between rings of aluminum tape (0.003 inch thick, McMaster-Carr) with an outer diameter of 1 inch and an inner diameter of 3/8 inch, and is made of chemically resistant epoxy resin (MarineWeld 8272, JB welding) .

All data can be found in the text and SI appendix.

We thank ExxonMobil Research and Engineering for funding this research and Young Hee Yoon (Georgia Institute of Technology) for assisting with sample pyrolysis.

Author contributions: MGF and RPL design research; YM and NCB conducted research; NCB contributed new reagents/analysis tools; YM, NCB, FZ and RPL analysis data; YM, NCB, FZ, MGF and RPL wrote this paper.

The author declares no competing interests.

This article is directly contributed by PNAS.

This article contains online support information https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2022202118/-/DCSupplemental.

Released under PNAS license.

Thank you for your interest in advertising on PNAS.

Note: We only ask you to provide your email address so that the people you recommend the page to know that you want them to see it, and that it is not spam. We do not capture any email addresses.

Feedback privacy/legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID and Research4Life.