The use of fossil fuels produces vast amounts of flue gas, which mainly consists of CO2 and N2, and has become one of the most significant contributors to CO2 emissions. One important strategy to reduce CO2 emissions is to capture CO2 from the post-combustion flue gas. Polymeric membranes have emerged as one of the leading technologies for CO2 capture due to their high energy efficiency, facile scalability, operation and maintenance, and small footprint. [1][2][3] As the flue gas has a huge volume and low CO2 partial pressure, membrane materials should exhibit both high CO2 permeability and CO2/N2 selectivity to minimize the size of the membrane skid and maximize the purity of the product. 3 Various polymeric materials have been designed and engineered for high-performance CO2/N2 separation, such as thermally rearranged (TR) polymers, [4][5][6] polymers of intrinsic microporosity (PIMs), 7,8 polyamines, 9,10 polyimides, 11 and mixed-matrix materials (MMMs). 9,12 In particular, polymers containing poly(ethylene oxide) (PEO) have emerged as leading materials for this application because of their affinity towards CO2 (resulting in superior CO2/gas solubility selectivity) and flexible polymer chains (leading to high CO2 diffusivity). [13][14][15][16] For example, block copolymers (e.g., Pebax ® and PolyActive TM ), [17][18][19][20][21][22] cross-linked PEO, 15,[23][24][25][26] and MMMs [27][28][29][30] have been demonstrated with an excellent combination of high CO2 permeability and CO2/N2 selectivity.
Further increase of ether oxygen contents in polymers is expected to improve CO2/N2 solubility selectivity and thus permeability selectivity. Recently, we have synthesized polymers based on 1,3-dioxolane (DXL with an O/C molar ratio of 0.67), which has ether oxygen content higher than PEO (O/C ratio: 0.5). 31,32 As high molecular weight polymers based on DXL tend to crystallize (reducing gas permeability), hyperbranched polymers (PDXLAn) with short branches of DXL were synthesized from poly(1,3-dioxolane) acrylate (DXLAn with n = 12). The short branches suppress crystallization while achieving high content of ether oxygens (40 mass%). 25,32 For example, PDXLA12 was amorphous and had glass transition temperature (Tg) of -66 ℃, and it displayed CO2 permeability of 200 Barrer (1 Barrer = 10 -10 cm 3 (STP) cm cm -2 s -1 cmHg -1 ) and CO2/N2 selectivity of 60 at 35 ℃. Such performance is better than its PEO analog prepared from poly(ethylene glycol) acrylate (PEGA with n = 7), which showed CO2 permeability of 120 Barrer and CO2/N2 selectivity of 46 at 35 ℃. 25 However, there lacks a systematic study of the impact of the n values on physical and gas transport properties of the polymers, which is critical to understand the structure/property relationship and guide the design of high-performance membranes for CO2/N2 separations.
Herein we present an improved method to synthesize a systematic series of DXLAn with a well-controlled DXL chain length (n) ranging from 4 to 20, which were then photopolymerized to form hyperbranched PDXLAn films. The structural and gas transport characteristics of PDXLAn are thoroughly investigated to derive the relationship between structure and CO2/gas separation properties. Pure-and mixed-gas CO2/N2 separation properties were thoroughly evaluated. PDXLA8 displays the best combination properties with CO2 permeability of 220 Barrer and CO2/N2 selectivity of 56 at 35 ℃, surpassing Robeson's 2008 upper bound. When challenged with model flue gas containing 20% CO2 and 80% N2, PDXLA8 shows robust CO2/N2 separation performance at temperatures up to 60 ℃. Additionally, PDXLA8 exhibits good CO2/CH4 separation performance at 35 and 50 ℃, making it appealing for practical natural gas processing.
Synthesis and characterization of the DXLAn and PDXLAn. Figure 1a displays the synthesis of the macromonomers of DXLAn from 1,3-dioxolane (DXL) using cationic ringopening polymerization with triflic acid (TfOH) as the initiator and 2-hydroxyethyl acrylate (HEA) as the chain transfer reagent. 31,32 The TfOH:HEA molar ratio increases from 1:16 in the previous method to 1:9 in this work, which reduces the reaction time from 12 to 3 h while achieving a similar yield of DXLn and better control of the DXL chain length. Figure 1b shows the 1 H NMR spectrum of DXLA8 and the assignment of the chemical shifts for the protons. The protons (g and f) in the DXL unit have chemical shifts of 3.73 and 4.76 ppm, and α-acryloyl protons (a, b, and c) have chemical shifts of 5.82, 6.13, and 6.42 ppm. The n values are calculated by comparing the intensities of g and f with those of a, b, and c. In this study, five DXLAn macromonomers were synthesized with actual n values of 3.6, 6.5, 8.3, 9.8, and 11.9, and they are named DXLA4, DXLA6, DXLA8, DXLA10, and DXLA12, respectively. The NMR spectra of DXLA4, DXLA6, DXLA10, and DXLA12 are presented in Figure S1. The calculated n values are very similar to the theoretical values (i.e., the molar ratio of DXL to HEA). [32][33][34] The molecular weight and polydispersity of the DXLA8 were also analyzed using electrospray ionization (ESI)-mass spectroscopy (MS). The Mn of DXLA8 is 797 g/mol (Figure S2), which is close to that (731 g/mol) measured by NMR. Additionally, the DXLA8 shows the Mw of 836 g/mol and thus PDI of 1.05, indicating the narrow molecular weight distribution of DXLA8 and good control of chain length using our improved method. DXLAn (n ≤ 12) were then photopolymerized using 1-hydroxycyclohexyl phenyl ketone (HCPK) as an initiator to prepare PDXLAn films (Figure 1a). Because of its high molecular weight (998 g/mol), DXLA12 is a viscous liquid at ≈ 23 ℃ and needs precaution to avoid any bubbles when preparing the PDXLA12 films, and therefore, only PDXLAn (n ≤ 12) films were prepared and tested for gas transport properties. As shown in the FTIR spectra (Figure S3), the absorption peaks of double bonds in DXLA8 almost disappear after the photopolymerization, indicating the near-complete conversion of the acrylate groups. The PDXLAn films cannot be dissolved in typical solvents (e.g., water, methanol, and dimethylformamide), indicating the occurrence of cross-linking. Similar phenomena have been noted for the polymers derived from PEGA 25 and DXLA12, 32 and the cross-linking can be ascribed to the trace amount of diacrylates produced during the macromonomer synthesis or chain transfer during the photopolymerization. As such, PDXLAn cannot be directly used to prepare coating solutions to manufacture thin-film composite (TFC) membranes, and a controlled polymerization method should be developed to prepare high molecular weight, soluble polymers, which is beyond the scope of this study.
Figure 2a shows differential scanning calorimetry (DSC) thermograms for PDXLAn.
There are no crystallization or melting peaks, suggesting that the PDXLAn are completely amorphous. The X-ray diffraction (XRD) patterns of the PDXLAn films also confirm their amorphous structure at ≈ 23 ℃ (Figure S4a). These results demonstrate that the approach of short branches successfully prohibits the crystallization of polar ether groups. Figure 2b shows that the Tg of PDXLAn decreases with increasing DXL branch length before leveling off. For example, the Tg decreases from -46 ℃ for PDXLA4 to -60 ℃ for PDXLA8, PDXLA10, and PDXLA12, which can be ascribed to the change of the content of the hydroxyl (-OH) groups in the PDXLAn. The -OH groups can form hydrogen bonding with the oxygen atom, reducing the polymer chain flexibility and increasing the Tg. 25,35,36 As shown in Table S1 and Figure S4b, PDXLA4 has an -OH content of 4.4 mass%, almost double that of PDXLA8 (2.3 wt%), and further increase of the chain length from 8 to 12 only decreases the -OH content to 1.7 mass%. Consequently, PDXLA8, PDXLA10, and PDXLA12 show similar Tg values. The PDXLAn samples are also compared with a PEO analog prepared from poly(ethylene glycol) acrylate (PEGAn with n = 7 and Mn = 380 g/mol). As shown in Table S1, PEGA7 shows a Tg of -40 ℃ because of its high -OH content (4.5 mass%). The mechanical properties of the PDXLAn films were also tested (Figure S5). Increasing the chain length reduces the tensile strength and Young's modulus and improves the elongation at break, which can also be ascribed to the decreased -OH content and thus hydrogen bonding. As such, PDXLA4 with the higher -OH content shows greater tensile strength and Young's modulus and lower elongation at break than PDXLA8 and PDXLA12.
The density (ρp) of the PDXLAn films decreases with increasing the n values due to the decreased -OH content. The fractional free volume (FFV) is often estimated using the following equation:
where VW is the van der Waals volume estimated using the group contribution method. Figure 2b presents that the FFV of PDXLAn increases with increasing n values, consistent with the decreased -OH content.
Effect of the branch length on pure-gas transport properties of PDXLAn. Figure 3a displays that the pure-gas permeability of PDXLAn at 35 ℃ follows an order of CO2 > H2 > C2H6 > CH4 > N2, similar to the PEO-based polymers. 14,25 Increasing the n value increases the gas permeability before leveling off at the n value greater than 8. For example, CO2 permeability increases from 96 Barrer for PDXLA4 to 220 Barrer for PDXLA8 and 250 Barrer for PDXLA12 because PDXLA8 and PDXLA12 have lower Tg and higher FFV than PDXLA4. Figure 3b depicts the effect of the DXL chain length on CO2/gas selectivity. As the n value increases from 4 to 6, the CO2/N2 selectivity significantly increases from 46 to 57, and it becomes almost stable after n ≥ 6. The CO2/H2 selectivity slightly increases while CO2/C2H6 selectivity gradually decreases with increasing n values (Figure 3b). The chain length does not have a significant impact on the CO2/CH4 selectivity (ranging from 18 to 22), given its uncertainty of ≈ 10%. By contrast, PDXLA8
shows CO2 permeability (220 Barrer) and CO2/CH4 selectivity (22) higher than PEGA7 (with CO2 permeability of 120 Barrer and CO2/CH4 selectivity of 16), demonstrating the advantage of the DXL-based polymers over PEO-based polymers. (d) Modeling of gas permeability using the free volume model (i.e., Equation 2).
Figure 3c presents the effect of the feed pressure on the pure-gas permeability of PDXLA8 at 35 ℃. CO2 permeability increases from 220 Barrer at 2.0 bar to 240 Barrer at 5.1 bar because of the high CO2 sorption and thus plasticization of polymer chains, increasing the polymer chain flexibility and gas diffusivity. 37,38 On the other hand, the permeability of other gases (H2, N2, CH4, and C2H6) is independent of the feed pressure because of their low sorption in the polymer. 38 Gas diffusivity depends sensitively on the FFV of the polymer. If the polymer compositions have negligible impact on the gas solubility, gas permeability can be expressed using Equation 2: 37
where A is a front factor (Barrer), and B is a parameter increasing with the increasing kinetic diameter of the penetrants. Figure 3d exhibits that the permeabilities of all tested gases can be satisfactorily modeled using Equation 2. The constant B is 0.15 for H2 (with the kinetic diameter of 2.89 Å), 0.26 for CO2 (3.3 Å), 0.20 for N2 (3.64 Å), 0.25 for CH4 (3.8 Å), and 0.30 for C2H6 (4.44 Å), which follows the same increasing order of their kinetic diameter except for CO2.
To thoroughly elucidate the effect of the DXL chain length on gas transport properties, CO2 and C2H6 sorption isotherms of PDXLAn were determined at 35 ℃, and the results are displayed in Figures 4a,b and S4. N2 sorption is below the detection limit of our apparatus, and therefore, C2H6 is used as a marker for gas molecules without specific interactions with PDXLAn. 32 For all PDXLAn films, the CO2 and C2H6 sorption increase linearly with the pressure, and the sorption isotherms can be described by Henry's law because of their rubbery nature. 25,39 As presented in Figure 3c and Table S2, CO2 (with a critical temperature of 304 K) exhibits higher solubility than C2H6 (305 K) despite their similar critical temperature, confirming the affinity of PDXLAn with CO2. Figure 4c shows that increasing the n value decreases C2H6 solubility due to the decreased FFV but has a negligible effect on CO2 solubility caused by the opposite effect of decreased FFV and increased ether oxygen content (Table S1). As such, the CO2/C2H6 solubility selectivity slightly decreases with the increase of DXL chain length. The PDXLAn demonstrates CO2/C2H6 solubility selectivity higher than PEGA7 and PEGDA-co-PEGMEA (≈2.5), 25,40 further confirming that higher ether oxygen content leads to higher CO2/gas solubility selectivity. As gas solubility is independent of the n values in PDXLAn, the increase of gas permeability with increasing n value can be ascribed to the increase in gas diffusivity. Figure 4d confirms that both CO2 and C2H6 diffusivity increase significantly with increasing DXL chain length because of the increased FFV and lowered Tg. For example, CO2 diffusivity increases from 5.4×10 -7 cm 2 /s for PDXL4 to 1.2×10 -6 cm 2 /s for PDXL8 by 120%. Figure 4d presents that CO2 has much higher diffusivity than C2H6 because of the smaller kinetic diameter for CO2, and the CO2/C2H6 diffusivity selectivity is 2 -3, comparable with that of PEGA7 (cf. Table S2).
Potential of PDXLA8 for mixed-gas CO2/N2 separation. PDXLA8 exhibits both good mechanical properties and CO2/N2 separation performance at 35 ℃, and thus, it was thoroughly tested with CO2/N2 mixtures at various temperatures (Table S3). When tested with a mixed gas (CO2:N2=15:85) at 6.2 bar and 35 ℃, the sample shows CO2 permeability of 190 ± 4 Barrer, N2 permeability is 3.3 ± 0.2 Barrer, and CO2/N2 selectivity of 57 ± 3 (cf. Figure 5a). By contrast, the polymer exhibits pure-gas permeability of 240 Barrer for CO2 and 3.9 Barrer for N2 at 6.2 bar and 35 ℃ (cf. Table S3). This difference between pure-and mixed-gas permeability is often attributed to the opposite effects of gas compression (decreasing FFV) and CO2 plasticization (enhancing polymer chain flexibility). 28,41,42 The presence of non-plasticizing N2 in the gas mixture compresses the sample and mitigates the CO2 plasticization, reducing CO2 permeability, while the existence of CO2 neutralizes the compression and maintains N2 permeability. Figure 5b presents the mixed-gas CO2/N2 separation performance of PDXLA8 at 60 ℃, a typical temperature for the flue gas from coal-fired power plants. 3 ). Together with its good stability at high temperatures, PDXLA8 is promising for industrial applications. Although some facilitated transport membranes, such as polyvinyl amine (PVAm), show CO2/N2 separation performance above 2019 upper bound, 1 their performance is highly dependent on the water content. In practical applications, the water vapor contents in the feed and permeate need to be carefully controlled to retain the excellent CO2/N2 separation performance, which increases the complexity of the membrane system.
Potential of PDXLA8 for mixed-gas CO2/CH4 separation. PDXLA8 was also investigated for CO2/CH4 separation at various feed pressures, feed gas compositions, and temperatures due to its balanced CO2 permeability and CO2/CH4 selectivity. Figure 6a shows that both CO2 and CH4 permeability increase slightly with increasing CO2 partial pressure at 35 ℃ and 6.5 bar because of the plasticization effect. For example, CO2 permeability increases from 186 ± 3 (10% CO2 in the feed gas) to 204 ± 3 Barrer (80% CO2 in the feed gas). Similar to the mixedgas CO2/N2 tests, the mixed-gas CO2 permeability is lower than the pure-gas values due to the compression by less condensable CH4. In addition, the plasticization decreases the CO2/CH4 selectivity from 14 ± 1 (10% CO2 in the feed gas) to 12 ± 1 (80% CO2 in the feed gas), which is a typical phenomenon observed in polymers. 46,47 Consequently, the CO2/CH4 selectivity improves from 12 to 20.
Figure 6c shows that increasing the temperature enhances gas permeability and reduces CO2/CH4 selectivity. For example, increasing the temperature from 35 to 50 ℃ increases CO2 permeability from 240 to 320 Barrer and decreases CO2/CH4 selectivity from 20 to 13 when tested PDXLA8 exhibits CO2/CH4 selectivity similar to and CO2 permeability much higher than CA (indicating its potential for industrial applications), though its separation performance is below the upper bound. However, PDXLA8 exhibits higher permeability of C2+ hydrocarbons (C2H4, C2H6, C3H6, and C3H8) than CH4 (Table S4) and thus much lower selectivity of CO2/C2+ hydrocarbons than CA, leading to higher hydrocarbon loss. 47 Nevertheless, biogas and landfill gas contain mainly CH4 and CO2, and PDXLA8 does not suffer from the disadvantage of the high loss of C2+ hydrocarbons.
Figure 7 demonstrates the stability of PDXLA8 at different feed gas conditions and temperatures. PDXLA8 displays stable CO2 permeability and CO2/CH4 selectivity for ~100 h at 35 ℃ (Figure 7a) and 50 ℃ (Figure 7b). By contrast, the n value has essentially no effect on CO2 solubility because of the decreased FFV.
When tested with model flue gas containing 20% CO2 and 80% N2, PDXLA8 demonstrates robust CO2/N2 separation performance at temperatures up to 60 ℃, superior to 2008 upper bound.
PDXLA8 also shows mixed-gas CO2/CH4 separation performance superior to CA, a workhorse material for industrial gas separation. This work demonstrates that hyperbranched architectures are an effective tool to design amorphous polymers incorporating high content of highly polar groups with a great tendency to crystallize and that chain end groups have an important impact on structures and gas transport properties.
Anhydrous DXL (99.8%), triethylamine (TEA, ≥99%), and HCPK (99%) were obtained from Sigma-Aldrich Corporation (St. Louis, MO). 2-Hydroxyethyl acrylate (HEA, 97%)
and triflic acid (TfOH, 99%) were acquired from Acros Organics (Fair Lawn, NJ).
Dichloromethane (DCM, ≥99.5%) was procured from Fisher Scientific (Hampton, NH) and dried with 3A molecular sieves (Acros Organics) before use. CO2, H2, and N2 (99.9%) were supplied by Airgas, Inc. (Radnor, PA).
DXLAn was first synthesized following our previous work with minor modifications, 31,32 and the chain length (n value) of DXLAn is almost the same as the molar ratio of DXL to HEA. As an example of synthesizing DXLA8, a flask was first purged with argon for 10 min to remove O2 and H2O and then kept in an ice bath. Second, DCM (50 mL), HEA (7.34 mL, 62 mmol), and TfOH (0.607 mL, 6.89 mmol) were added into the flask, and the solution was stirred at 600 rpm for 10 min. Third, DXL (35 ml, 500 mmol) was added slowly to the solution using a syringe under a continuous argon purge. After 10 min, the flask was transferred to a water bath (21 ℃) for the reaction to continue for 3 h. Fourth, TEA (2.5 mL) was added to terminate the reaction, and the solution was stirred for another 10 min. Finally, the solution was extracted with 100 mL deionized (DI) water 3 times, and the DCM phase was rotary-evaporated under vacuum at 50 ℃ for 3 h to obtain the DXLAn product. The yield of DXLAn is about 75% (based on the mass of the added DXL and HEA).
To synthesize PDXLAn, the prepolymer solution was prepared by mixing DXLAn (3.0 g) and HCPK (3.0 mg). 28,32 The solution was photopolymerized using 254-nm UV light at 3.0 mW cm -2 for 5 min. The film thickness (~200 µm) was measured using Starrett 2900 micrometer (The L.S. Starrett Co., MA). All the PDXLAn films were used for gas permeation tests without further treatment.
Characterizations. Rigaku Ultima IV diffractometer with Cu Kα radiation (Rigaku, JP)
was used to obtain X-ray diffraction patterns. ATR-FTIR spectra were collected using a vertex 70
Burker spectrometer (Billerica, MA). NMR spectra were recorded on a Varian Inova-500 spectrometer. Mass spectra were obtained using a Fourier Transform Ion Cyclotron Resonance Mass Spectrometer (FT-ICR MS) (Solarix 12 T, Bruker) using electrospray ionization (ESI).
Thermal transitions of the polymers were determined using Q2000 DSC (TA Instruments, New Castle, DE). Samples were scanned from -90 ℃ to 80 °C in N2 atmosphere at 10 °C min -1 for three cycles, and the last heating cycle was used to obtain the properties. The mechanical properties of PDXLAn films were determined using static tensile loading with a dynamic mechanical analysis (DMA, Q800 TA Instrument). The tensile tests were conducted at temperatures of 25°C. Uniaxial tensile loading on the sample stripes (20 mm × 3-4 mm) was carried out with an initial strain of 0.01% at a constant strain rate of 1.0%/min until the sample fractured. The film density was measured using Mettler Toledo XS64 with a density measurement kit based on Archimedes' principle. 28 Pure-gas permeability (PA, Barrer) was determined using a constant-volume and variablepressure system at 35 ℃. 48 The leak rate of the system was measured before permeation tests and should be ten times lower than the steady-state gas permeation rate. Mixed-gas CO2 and N2 permeability were measured by a constant-pressure and variable-volume system with H2 as a sweep gas on the permeate side. 32,48 The gas permeability has an uncertainty of ≈10% estimated using an error propagation analysis unless specified. 49 A dual-volume and dual-transducer system was used to determine gas sorption isotherms at 35 ℃. 50 Gas solubility (SA, cm 3 (STP) cm -3 atm -1 ) can be calculated using Equation 3:
where CA is the gas sorption (cm 3 (STP) cm -3 ), and pA is the gas pressure at equilibrium (atm). The solubility has an uncertainty of <10% using the error propagation analysis. 49 Based on the solution and diffusion mechanism, gas diffusivity (DA, cm 2 s -1 ) can be obtained using Equation 4:
■ ASSOCIATED CONTENT
The Supporting Information is available free of charge at https://pubs.acs.org.
*E-mail: haiqingl@buffalo.edu