https://orcid.org/0009-0006-3261-3209
Figures
Abstract
Polymer membranes offer an extremely attractive solution for achieving sustainable carbon dioxide capture and have the potential for large-scale application. The authors prepared almost amorphous polyethylene oxide separation membranes through cross-linking methods, with a CO2/H2 separation coefficient of 7.9. However, there is still the drawback of low CO2 permeability. To improve the performance of the cross-linked membranes, this paper utilized the swelling property of the cross-linked membranes and immersed them with different end groups of PEG aqueous solutions. The effects of different end groups on the gas separation performance of the cross-linked membranes were investigated. The results showed that after being impregnated with different end group PEG, the gas permeabilities of the cross-linked membranes increased. When using a polyethylene glycol with a molecular weight of 250 g/mol for impregnation, a 47.0wt.% increase in weight was observed, and the gas permeability and CO2/H2 separation coefficient of the cross-linked membranes increased significantly. The CO2 permeability increased from 170 Barrer of the original membrane to 1457 Barrer, and the separation coefficient of H2 increased from 7.9 to 13.
Citation: Ji S, Wang T, Guan L, Zhao C, Quan S, Dong X, et al. (2026) Effect of impregnation with polyethylene glycols (PEGs) of different end groups on gas separation performance of cross-linked polyethylene oxide (PEO) membranes. PLoS One 21(4): e0346667. https://doi.org/10.1371/journal.pone.0346667
Editor: Anshu Sharma, Central University of Haryana, Mahendergarh, INDIA
Received: November 27, 2025; Accepted: March 20, 2026; Published: April 10, 2026
Copyright: © 2026 Ji et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: This study is supported by the Shandong Provincial Natural Science Foundation (Grant No. ZR2022MB100), Shandong Province University Youth Innovation Team Development Plan (Grant No.2024KJH159). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
1 Introduction
The massive emission of carbon dioxide (CO2) is one of the main causes of global climate change, seriously threatening the human living environment [1,2] Therefore, it is necessary to effectively capture and separate CO2. Compared with traditional CO2 separation methods, membrane technology has the advantages of good stability, high efficiency, low cost, simple operation and less environmental pollution [3–5]. Polyethylene oxide (PEO) materials are considered as one of the best materials for preparing CO2 capture gas separation membranes because the ether oxygen groups contained in them have a strong interaction with CO2, facilitating the mass transfer of CO2 in the membrane [6,7]. The solubility coefficient of gas in polymers is mainly affected by the compressibility of the gas itself and the interaction between the gas and the polymer matrix [8]. The ether oxygen functional groups in the PEO structure can form a strong tetrahedral dipole distance interaction with CO2, thus PEO materials have a high CO2/light gas solubility selectivity (CO2 is much greater than H2), and there is often a trade-off relationship between the solubility selectivity and diffusion selectivity in the CO2/H2 separation process (SCO2/H2 is high, DCO2/H2 < 1) [9–11]. During the separation process, solubility selectivity plays a dominant role, with CO2 being the fast gas and H2 being the slow gas, thereby achieving the reverse selection of CO2. However, traditional PEO materials have a high crystallinity due to their regular chain segments, which significantly reduces the gas permeability of the crystalline region [12–14]. Sun et al.[15] prepared high-performance PEO membranes (named BPM) via UV-crosslinking of bisphenol A ethoxylate diacrylate (BPA), poly(ethylene glycol) methyl ether acrylate (PEGMEA) and low molecular weight PEGDME without additional solvents. When PEGDME content reached 50 wt%, the membrane showed an unprecedented CO₂ permeability of 4883 Barrer with a high CO₂/N₂ ideal selectivity of 43, surpassing the 2019 CO₂/N₂ upper bound. This aligns with the broader goal of optimizing CO₂/N₂ separation performance, a key focus in carbon capture research [16]. Fan et al.[17] proposed a MOF-in-COF strategy for CO₂/H₂ separation: confining MOF unit cells in 2D COFs’ 1D channels to build hierarchical pores. The membranes showed ultrahigh H₂ permeance (>3000 GPU) and CO₂/H₂ selectivity exceeding Robeson upper bounds (via size sieving and fast transport), addressing pure COFs’ wide pores and traditional bilayer membranes’ underused sieving, offering a new high-performance gas separation paradigm. Therefore, preparing amorphous PEO membrane materials is very important. To inhibit the crystallization of PEO materials, scholars at home and abroad have adopted methods such as blending, copolymerization and cross-linking [18–20]. To increase the CO2 permeability, Castro-Munoz et al. [21] prepared Matrimid 5218/PEG200 blend membranes with different PEG contents, and conducted single gas permeation tests at 35°C and 10 bar feed pressure. The results showed that the CO2 permeability increased with the increase in additive content, from 7.7 Barrer of the pure membrane to 22.0 Barrer of the blend membrane with 20wt.% PEG200 content. The CO2/CH4 selectivity slightly increased from 35 of the pure membrane to 40 of the blend membrane with 5wt.% PEG200 content. When the content is higher, it decreases. In contrast, Vroulias et al.[8] observed that increasing IL loading (from 30 to 40 wt%) in PEO-based copolymer membranes consistently improved CO₂ permeability, with the 40 wt% IL-loaded membrane showing significantly higher permeability than the 30 wt% counterpart, which may be attributed to the stronger plasticization effect of higher IL content. In addition, ether groups, endowed with excellent CO₂ affinity, have also been incorporated into polynorbornene matrices for the fabrication of gas separation membrane materials. Alentiev et al. reported the synthesis and gas transport performance of two series of addition-type and metathesis polynorbornenes grafted with flexible ether moieties as pendant groups. The resulting polymers were found to be amorphous and exhibited enhanced solubility-controlled CO2 permeation behavior [22]. Additionally, Medentseva investigated the gas permeability of two isomeric vinyl-addition polymers synthesized from commercially available ester-functionalized norbornenes; these polymers displayed outstanding ideal and mixed-gas selectivity, along with favorable permeability for several industrially relevant gas pairs. Notably, acetoxy-functionalized polynorbornene (AcPNB) achieved a CO2 permeability of 270 Barrer, with a CO2/N2 separation selectivity exceeding 20.[23]
Despite this, completely amorphous PEO membranes still have problems of low gas permeability and low gas separation coefficient [24,25]. This paper represents a highly versatile strategy for markedly enhancing the gas transport performance of swellable PEO membranes. Leveraging the swelling behavior of cross-linked PEO membranes in organic solvents, we intentionally incorporated low-molecular-weight polyethylene glycol (PEG) into the PEO membrane matrix, and investigated the influence of different functional group PEG immersion on the gas permeability of the cross-linked membranes under similar molecular weight conditions. At the same time, the dissolution-diffusion theory was used for explanation.
2 Experiment
2.1 Experimental materials
O,O′-bis (2-aminopropyl) polypropylene glycol-block-polyethylene glycol-block-polypropylene glycol (PEO-600, molecular weight of 600 g/mol) and poly (ethylene glycol) diglycidyl ether (PEO-526, molecular weight of 526 g/mol) were purchased from Sigma-Aldrich and used as received. Poly(ethylene glycol) (PEG: molecular weight of 300 g/mol and 400g/mol, PEG300/PEG400), Methoxypolyethylene glycol (PEGME: molecular weight of 350 g/mol) and Poly(ethylene glycol) dimethyl ether (PEGDME: molecular weight of 250 g/mol) were purchased from Sigma-Aldrich and used as received.
2.2 Preparation of cross-linked PEO membranes and impregnated membranes
The preparation of cross-linked PEO membrane has been described in our previous work PEO-600 and PEO-526 were mixed directly according to the mole ratio of reactive groups with magnetic stirring under room temperature [26]. After continuously stirring more than 6 hours, the mixed homogeneous liquid was poured onto a Teflon dish. Subsequently, the membrane was pre-cured at 80°C for 3 hours. Then, the temperature was raised to 100°C with a heating rate 1°C/min and kept there for 2 hours. Finally, the temperature was raised to 120°C with a heating rate 0.5°C/min and maintained at 120°C for 0.5 hour. The obtained membranes (named CM) were ready for characterizations.
The known weight (m1) cross-linked PEO membranes were immersed in the PEG-water solution (70/30wt%). After 2h of ultrasonication to promote the incorporation of PEG molecule in the solution into cross-linked PEO membranes, the solution in which the membranes were immersed was standing under room temperature for 12 ~ 48h. Then, the Impregnated Membrane (IMs) were washed with distilled water immediately to remove PEG on the surface of IMs. The IMs were annealed at 60°C in vacuum for 48h and weighted as m2. As a result, a serial of IMs with different high loading rate ((m2-m1)/m1* 100%) were obtained. The IMs with different PEG loading rate were denotedas IM-x-y, where x represented PEG with different end group (EG represented PEG300 or PEG400, ME represented PEGME350 and DME represented PEGDME250), the y represented loading rate. The photographs of the membranes can be found as S1 Fig(Supporting information).
2.3 Measurements of pure gas transport properties
The solution-diffusion mechanism was employed to explain the gas transport properties through dense membranes. The permeability(P) with the unit of Barrer (1 Barrer = 10−10 cm3(STP) cm/cm2 sec cmHg) can be expressed as:
where D with the unit of cm2/s and S with the unit of cm3(STP)/cm3 polymer cmHg are the diffusivity and solubility coefficients, respectively. The ideal selectivity of a membrane for gas A over gas B can be defined as follows:
The pure gas permeability was obtained by using a constant volume method. The permeabilities of various gases were tested in the sequence of H2, N2 and CO2. The gas permeabilities were measured at 35°C. The H2 is tested under 3.5 atm for safety reason and other gases are tested under 10 atm. For comparisons, CO2 gas is tested under both 10 atm and 3.5 atm. The gas permeability was determined from the rate of downstream-pressure increase (dp/dt) obtained when permeation reached steady state according to the following equation:
where P is the permeability of a membrane to a gas and its unit is in Barrer. V is the volume of the downstream chamber (cm3), l is the membrane thickness (cm). A refers to the effective area of the membrane (cm2), T is the testing temperature (K) and p2 means the upstream operating pressure (psia). The diffusivity can be calculated by the time-lag method as Equation 4 expressed, and then the solubility can be simply deduced according to Equation 1.
where θ is the diffusion time lag extrapolated from the plot of pressure with time at steady state to the time axis.
Results and discussion
The structure of cross-linked membranes (CM) have been described in our previous works [26], which will not be discussed here. Fig 1 shows the gas permeabilities of original membranes and IMs with different end groups.
As can be seen from Fig 1, CO2 permeability of all membranes are much more hihger than that of N2 and H2, which due to the strong interaction between CO2 and EO groups in PEO membranes causing high solubility. Interestingly, H2 permeability is higher than N2 permeability, which can be attributed to the high diffusivity caused by its smaller kinetic diameter. Meanwhile, the gas permeability of IMs impregnated with PEG solution has significantly increased compared to the original membranes. For example, the CO2 (10 atm) permeability increased from 170 Barrer of the original membrane to 530 Barrer of the PEGOH-400 membrane and 710 Barrer of the PEGME-350 membrane. More delightfully, the gas permeability of the membrane impregnated with PEGDME-250 was further significantly enhanced, the highest CO2 permeability reached 1090 Barrer. Through the above analysis, it can be found that the different end groups have a significant impact on the gas permeability of the IMs. This may be because when the PEG end group is a hydroxyl group, the small molecules PEG impregnated into the membrane are more likely to form hydrogen bonds with the cross-linked membrane segments, hindering the diffusion process of gases. When using PEGME-350, the number of hydroxyl groups decreases, and the hindering effect of hydrogen bonds weakens; when using PEGDME-250, the hydrogen bond effect is further weakened, thereby significantly increasing the gas permeability. Moreover, it can also be seen that the permeabilityes of CO2 and N2 increase significantly, while the permeability of H2 increases only slightly. This phenomenon indicetaes that the impregnation treatment effectively improves the free volumes of IMs which is beneficial to the gas molecules diffusion process [27,28]. However, the improved free volumes is usefulness for H2 due to its smaller kinetic diameter. In order to nvestigate the effect of PEG with different terminal groups on the free volume of the membrane, The positron annihilation lifetime spectroscopy (PALS) was used to determine the size of free volume and the fractional free volume of the dense membranes. A detailed description of the PALS test method can be found in the supporting information. Table 1 shows the o-Ps lifetime τ3, intensity I3 and FFV results of original membranes and IMs with different end groups.
According to Table 1, it appears that the τ3 values, which represent the mean radii of the free volume cavities according to the Tao-Eldrup model in ESI, show an increasing trend as the chang of diffrent end groups. In addition, the FFV, associated with both τ3 and I3, shows a monotonically increment, causing the significant increase of gas permeability. This indicates that that different types of end groups exert a significant influence on the FFV of the membrane. A higher FFV enables gases to diffuse more easily inside the membrane, which is also the reason for the increase in gas permeability.
Fig 2 shows the gas selectivities of original membranes and IMs with different end groups. From the figure, it can be seen that the CO2/N2 selectivity changes slightly, remaining between 54 and 61; while the CO2/H2 selectivity has significantly increased, from 8.5 of IM-EG300 to 12.6 of IM-DME250, approximately increasing by 50%. This is mainly due to the fact that the increase in CO2 permeability is greater than that of H2. The molecular kinetic diameter of H2 (2.89 Å) is much smaller than that of CO2 (3.30 Å) and N2 (3.64 Å), so the free volume change and hydrogen bond effect of the impregnated membrane have a relatively small impact on the diffusion process of H2, while they have a greater impact on CO2 and N2 with larger molecular kinetic diameters. Therefore, when impregnated with PEGDME-250, the permeabilities of CO2 and N2 have significantly increased due to the improvement of the diffusion process, while the permeability of H2 has increased slightly.
Fig 3 shows the relative permeability increment of IM-PEGDME as a function of PEG loading compared with originan membrane. As can be seen from Fig 3, the gas permeability of IM-DME significantly increased with the increase of PEGDME250 loading rate. Moreover, the growth rates of the permeabilityes of the three test gases were in the order of N2 > CO2 > H2. The permeabilityes of H2, N2, and CO2 increased from 21.5 Barrer, 2.64 Barrer, and 170 Barrer of the original membrane to 100 Barrer, 24.1 Barrer, and 1457 Barrer with a loading rate 47.0% of IM. The growth rates of the permeabilityes reached 365%, 810%, and 750% respectively. The significant increase in gas permeability can be attributed to the free volume fraction of the membrane increased after impregnation, which is beneficial for the diffusion of gas molecules in the membrane; and after impregnation, due to the presence of PEGDME, the number of ether oxygen groups on the membrane surface increased, which can improve the dissolution process of gases on the membrane surface.Table 2 shows the o-Ps lifetime τ3, intensity I3 and FFV results of IM-PEGDME as a function of PEG loading.
According to Table 2, the τ3 values show an increasing trend as the PEGDME250 loading increases from 0 to 47.0%. This is ascribed to the increased interchain spacing arising from the incorporation of PEGDME molecules after the swelling immersion treatment. Additional free volume is created within the membrane by the embedded PEGDME molecules. In addition, the FFV shows a monotonically increment as the PEGDME loading increases, indicating that the incorporation of PEGDME molecules would create additional free volume cavities in the polymer matrix.
Fig 4 illustrates the solubility (a) and diffusion (b) of the IM-DME with different loading rate e in which there is no data of H2 because the time lag of H2 is extremely small. The CO2 solubility is much more higher than that of N2, due to the strong interaction between CO2 and PEG. Meanwhile, the CO2 solubility inceases with the inceasing PEGDME loading rate. On the contrary, the N2 diffusivity is slightly higher than CO2 diffusivity duing to the smaller kinetic diameter of N2. Furthermore, the diffusivities of N2 and CO2 increse with the incresing PEGDME loading rate, attributed to the improvment of free volumes of IM-DME.
Fig 5 shows the gas selectivity of IM-DME as a function of PEG loading rate. From the figure, it can be seen that the selectivity of CO2/N2 slightly decreases, from 64 of the original membrane to around 60. This can be attributed to the fact that after PEGDME impregnation, the increase in N2 permeability is greater than that of CO2. At the same time, the selectivity of CO2/H2 shows a significant increase, from 7.0 of the original membrane to 13 when the loading rate of PEGDME reaches 47.0wt.%, with an growth rate of 86%. It can be seen that the use of PEGDME-250 impregnated CM membrane has a very obvious effect on improving the gas permeability of the membrane and the separation coefficient of CO2/H2.
In addition, time-dependent permeation performance of IMs was also investigated. Fig 6 shows the CO2 permeability of IM-DME with a loading 47.0% as a function of test duration.
According Fig 6, the CO2 permeability of the membrane remained relatively stable and essentially unchanged with the extension of the test time. This indicates thatthe PEG molecules are firmly entrapped within the membrane due to the cross-linked structure inside the membrane, with no leaching observed.
Conclusions
Through this work, it was discovered that by immersing the prepared cross-linked PEO in a small-molecular-weight PEG solution, the gas permeability of the cross-linked membrane could be effectively improved. This was because the immersion treatment improved the intrinsic volume of the cross-linked membrane. Different end groups of PEG also played an important role in improving the performance of the cross-linked membrane. The results showed that the immersion treatment with dimethyl ether polyethylene glycol (PEGDME) was the most effective. Moreover, as the weight gain during immersion increased, the permeability of the cross-linked membrane also improved significantly. In this work, when the weight gain of PEGDME was 47.0 wt.%, the gas permeability of the cross-linked membrane was the best. The CO2 permeability increased from 170 Barrer of the original membrane to 1457 Barrer, and the separation coefficient for H2 increased from 7.9 to 13. This was mainly because the free volume fraction of the membrane increased after immersion, which was beneficial for the diffusion of gas molecules within the membrane; at the same time, due to the presence of PEGDME after immersion, the number of ether oxygen groups on the membrane surface increased, which improved the dissolution process of gas on the membrane surface. Meanwhile, the impregnated membranes exhibit excellent temporal stability.
Acknowledgments
I sincerely thank my supervisor for guiding this study, from design to analysis. Thanks to lab colleagues for experimental help and Sigma-Aldrich for materials. Gratitude to family and friends for support, and to reviewers and staff for refining the work and administrative aid.
References
- 1.
Pörtner HO, Roberts DC, Adger WN, et al. Climate Change 2023: Synthesis Report. Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva, Switzerland: Intergovernmental Panel on Climate Change (IPCC); 2023.
- 2. Kamio E, Yoshioka T, Matsuyama H. Recent advances in carbon dioxide separation membranes—a review. J Chem Eng Jpn. 2023;56(1):1–44.
- 3.
Baker RW. Membrane Technology and Applications. 4th ed. Hoboken, NJ, USA: John Wiley & Sons; 2021.
- 4.
Freeman BD, Yampolskii Y, Pinnaau I, eds. Materials Science of Membranes for Gas and Vapor Separation. Hoboken, NJ, USA: John Wiley & Sons; 2006.
- 5. Han Y, Ho WSW. Recent advances in polymeric membranes for CO₂ capture. J Membr Sci. 2022;641:119928.
- 6. Zhao YY, Zhang YQ, Li FR. Ultra-robust superwetting hierarchical membranes constructed by coordination complex networks for oily water treatment. J Membr Sci. 2021;627:119234.
- 7. Borsap S, Charernsuk M, Tangjituabun K, Tantirungrotechai J. Influence of Silane‐Type Electron Donors on Ziegler–Natta Catalysts in Ethylene Polymerization and Ethylene/α‐Olefin Copolymerization. ChemCatChem. 2025;17(12).
- 8. Vroulias D, Staurianou E, Ioannides T, Deimede V. Poly(ethylene oxide)-Based Copolymer-IL Composite Membranes for CO2 Separation. Membranes. 2022;13(1):26.
- 9. Park S, Morales-Collazo O, Freeman B, Brennecke JF. Ionic Liquid Stabilizes Olefin Facilitated Transport Membranes Against Reduction. Angew Chem Int Ed Engl. 2022;61(25):e202202895. pmid:35384196
- 10. Han GL, Lv J, Chen M. ZIF-67 incorporated sulfonated poly(aryl ether sulfone) mixed matrix membranes for pervaporation separation of methanol/methyl tert-butyl ether mixture. Membranes. 2023;13(4):389.
- 11. Sun Z, Wang S, Xiong H, Wu K, Shi J. Optimal nanocone geometry for water flow. AIChE Journal. 2021;68(3).
- 12. Sanders DF, Smith ZP, Guo R, et al. Energy-efficient polymeric gas separation membranes for a sustainable future: A review. Polymer. 2013;54(17):4729–61.
- 13. Tzirakis F, Papadopoulos AI, Seferlis P. CO₂ solubility in aqueous N-methylcyclohexylamine (MCA) and N-cyclohexyl-1,3-propanediamine (CHAP) solutions. AIChE J. 2023;69:e17982.
- 14. Xiao W, Wang H, Cheng A, Wang H, Yang Z, Wu X, et al. Design and Optimization of a Novel Hybrid Membrane-Electrochemical Hydrogen Pump Process for Recovering Helium from NRU off Gas. Membranes (Basel). 2023;13(7):689. pmid:37505055
- 15. Sun W-S, Yin M-J, Zhang W-H, Li S, Wang N, An Q-F. Green Techniques for Rapid Fabrication of Unprecedentedly High-Performance PEO Membranes for CO2 Capture. ACS Sustainable Chem Eng. 2021;9(30):10167–75.
- 16. Noorani N, Mehrdad A. Impregnation of amine functionalized deep eutectic solvents in NH₂-MIL-53(Al) MOF for CO₂/N₂ separation. Sci Rep. 2023;13(1):13012.
- 17. Fan H, Peng M, Strauss I, Mundstock A, Meng H, Caro J. MOF-in-COF molecular sieving membrane for selective hydrogen separation. Nat Commun. 2021;12(1):38. pmid:33397939
- 18. Dehane A, Merouani S, Hamdaoui O. Effect of carbon tetrachloride (CCl₄) sonochemistry on the size of active bubbles for the production of reactive oxygen and chlorine species in acoustic cavitation field. Chem Eng J. 2021;417:129214.
- 19. Jull EIL, Mandle RJ, Raistrick T, Zhang Z, Hine PJ, Gleeson HF. Toward In Silico Design of Highly Tunable Liquid Crystal Elastomers. Macromolecules. 2022;55(11):4320–30.
- 20. Park T, Choi GH, Lee D. Metal-phenolic network-coated hollow fiber catalytic membranes via solvent transfer induced phase separation (STRIPS) for Suzuki coupling reaction. J Membr Sci. 2021;634:119386.
- 21. Castro-Muñoz R, Fíla V, Martin-Gil V. Enhanced CO₂ permeability in Matrimid®5218 mixed matrix membranes for separating binary CO₂/CH₄ mixtures. Sep Purif Technol. 2018;209:844–53.
- 22. Alentiev DA, Nikiforov RYu, Rudakova MA, Zarezin DP, Topchiy MA, Asachenko AF, et al. Polynorbornenes bearing ether fragments in substituents: Promising membrane materials with enhanced CO2 permeability. Journal of Membrane Science. 2022;648:120340.
- 23. Medentseva EI, Khrychikova AP, Bermesheva EV, Borisov IL, Petukhov DI, Karpov GO, et al. CO2-separation performance of vinyl-addition polynorbornenes with ester functionalities. Journal of Membrane Science. 2024;705:122916.
- 24. Mehringer KD, Davis BJ, Kemp LK, Ma G, Hunt SB, Storey RF, et al. Synthesis and Morphology of High-Molecular-Weight Polyisobutylene-Polystyrene Block Copolymers Containing Dynamic Covalent Bonds. Macromol Rapid Commun. 2022;43(24):e2200487. pmid:35822234
- 25. Yi H, Gao J, Lin S, Ma J, Guo J. Photoresponsive α-cyanostilbene-containing fluorescent liquid crystal polymers based on ring-opening metathesis polymerization for information storage and encryption. Polymer. 2022;258:125289.
- 26. Shao L, Quan S, Cheng XQ, et al. Developing Cross-linked Poly(ethylene oxide) Membrane by the Novel Reaction System for H2 Purification. Int J Hydrogen Energ. 2013;18:5122–32.
- 27. Car A, Stropnik C, Yave W, Peinemann K-V. PEG modified poly(amide-b-ethylene oxide) membranes for CO2 separation. Journal of Membrane Science. 2008;307(1):88–95.
- 28. Car A, Stropnik C, Yave W, et al. Pebax®/polyethylene glycol blend thin film composite membranes for CO2 separation: performance with mixed gases. Sep Purif Technol. 2008;62:110–7.