GAS SEPARATION MEMBRANES

Abstract

A composite membrane suitable for separating a gas from a gas mixture comprising a selective layer coated on a support, wherein said selective layer comprises: a) a polymeric matrix comprising an amine polymer; b) a graphene oxide nanofiller; and c) a mobile carrier selected from an ionic liquid or an amino acid salt.

Claims

1. A composite membrane suitable for separating a gas from a gas mixture comprising a selective layer coated on a support, wherein said selective layer comprises: a) a polymeric matrix comprising an amine polymer; b) a graphene oxide nanofiller; and c) a mobile carrier selected from an ionic liquid or an amino acid salt.

2. A composite membrane as claimed in any preceding claim wherein the polymeric matrix comprises a polymer comprising a repeating unit of formula (I) ##STR00009## wherein R.sub.1 and R.sub.2 are independently selected from hydrogen or a C.sub.1-C.sub.10 hydrocabyl group and the integer m is 0-6.

3. A composite membrane as claimed in any preceding claim wherein the polymeric matrix comprises a polyallylamine having a repeating unit of formula (II): ##STR00010## wherein R is a C.sub.1-C.sub.10 hydrocabyl group, preferably C.sub.1-C.sub.6 hydrocarbyl group, preferably C.sub.1-C.sub.6 alkyl group such as methyl, ethyl, n-propyl, isopropyl, n-butyl, sec-butyl, isobutyl or tert-butyl, more preferably R is isopropyl or sec-butyl.

4. A composite membrane as claimed in any preceding claim wherein the graphene oxide is physically or chemically modified, e.g. wherein the graphene oxide is porous.

5. A composite membrane as claimed in any preceding claim wherein the graphene oxide is porous and/or comprises a polymer grafted thereto, e.g. an oxygen- and/or nitrogen-containing polymer grafted thereto, preferably an oxygen-containing polymer (e.g. PEG) grafted thereto

6. A composite membrane as claimed in any preceding claim wherein the graphene oxide nanofiller has an average lateral dimension of 1000 nm or less, e.g. in the range 10-1000 nm, preferably 100-1000 nm, more preferably 300-900 nm, more preferably 400-800 nm.

7. A composite membrane as claimed in any preceding claim wherein the support is a flat sheet or in the form of one or more hollow fibers, especially in the form of one or more hollow fibers.

8. A composite membrane as claimed in any preceding claim wherein the nanofiller is present in the selective layer in an amount of less than 5 wt %, preferably less than 1 wt %, more preferably less than 0.5 wt %, preferably in the range 0.05 wt %-5 wt %, preferably 0.1-1 wt %, preferably 0.1-0.5 wt %, preferably 0.1-0.3 wt %.

10. A composite membrane as claimed in any preceding claim wherein the amount of mobile carrier in the membrane is in the range 1.0-40 wt %, preferably 2.0-30 wt %, more preferably 5.0-25 wt %.

11. A composite membrane as claimed in any preceding claim wherein the mobile carrier comprises an ionic liquid in which the cation is selected from 1-alkyl-3-methylimidazolium, 1-alkylpyridinium, fluorosulfonyl-trifluoromethanesulfonylimide (FTFSI) N-methyl-N-alkylpyrrolidinium or comprises the salt of a naturally occurring amino acid.

12. A composite membrane as claimed in any preceding claim wherein the selective layer has a thickness of 20 nm to 100 μm, preferably 50 nm to 10 μm, preferably 100 nm to 5 μm, more preferably 100 nm to 1 μm, more preferably 100 nm to 500 nm.

13. A composite membrane as claimed in any preceding claim wherein the support is made of polyethersulfone (PES), polytetrafluoroethylene (PTFE), polypropylene, sulphonated polysulfone, polyvinylidene fluoride (PVDF), polyacrylonitrile (PAN) and related block copolymers, cellulosics such as cellulose acetate (CA), polyimide, polyether imide (PEI), aliphatic polyamides, polyetheretherketone (PEEK), polyphenylene oxide (PPO) and polysulfone (PSf).

14. A composite membrane as claimed in any preceding claim wherein the support is porous.

15. A process for the formation of a composite membrane as clamed in any of claims 1 to 14, comprising the steps of: (I) forming an aqueous solution comprising: a) a polymeric matrix comprising an amine polymer; b) a graphene oxide nanofiller, and c) a mobile carrier selected from an ionic liquid or an amino acid salt; (II) casting said aqueous solution onto a support, such as a hollow fiber support.

16. A process as claimed in claim 15 wherein said support is a flat sheet support and the casting process uses a bar roller to apply the selective layer; or wherein the support is a hollow fiber and the casting process involves dip coating.

17. A process as claimed in any of claims 15 to 16 wherein the support is treated with a pore filling agent prior to casting the aqueous solution in step (II).

18. A process for separating a gas from a gas mixture, comprising a step of contacting the gas mixture with a membrane as claimed in claims 1 to 14.

19. Use of a membrane as claimed in claims 1-14 in the separation of a gas from a gas mixture, e.g. in separating carbon dioxide from a mixture containing the same, e.g. flue gas, biogas (e.g. biogas upgrading), natural gas (e.g. natural gas upgrading), or syngas

20. A composite membrane suitable for separating a gas from a gas mixture comprising a selective layer coated on a hollow fiber or flat sheet support, wherein said selective layer comprises: a) a polymeric matrix comprising an amine polymer; b) a porous graphene oxide nanofiller or PEG-modified graphene oxide nanofiller, and optionally c) a mobile carrier selected from an ionic liquid or an amino acid salt.

21. A composite membrane as claimed in claim 20 comprising a selective layer coated on a hollow fiber support, wherein said selective layer comprises: a) a polymeric matrix comprising an amine polymer and a polyvinyl alcohol; b) a porous graphene oxide nanofiller, and optionally c) a mobile carrier selected from an ionic liquid or an amino acid salt.

22. A composite membrane suitable for separating a gas from a gas mixture comprising a selective layer coated on a support, e.g. a hollow fiber or flat sheet support, wherein said selective layer comprises: a) a polymeric matrix comprising an amine polymer; b) a porous graphene oxide nanofiller or chemically-modified graphene oxide nanofiller, optionally wherein the chemically-modified graphene oxide nanofiller is graphene oxide with an organic unit grafted thereon, preferably wherein said organic unit is selected from a nitrogen and/or oxygen-containing organic unit, a polymer, or a nitrogen- and/or oxygen-containing polymer, preferably said chemically-modified graphene oxide nanofiller is PEG-modified graphene oxide nanofiller, and optionally c) a mobile carrier selected from an ionic liquid or an amino acid salt.

23. A composite membrane as claimed in any of claims 20-22, wherein the composite membrane, support, selective layer, polymeric matrix, graphene oxide nanofiller, and/or mobile carrier are as defined in any of claims 2-14.

Description

BRIEF DESCRIPTION OF FIGURES

[0179] FIG. 1 is a process flow chart for the gas permeation test. Flue gas from cement kiln is used as the feed gas in the permeation test. A membrane pump (KNF Model N036-ST-11-E) is used to suck the flue gas from the chimney. The basic operation conditions are feed flow 10 L/min; feed pressure 1.7 bar; temperature 60° C.; no sweep gas. Synthetic flue gas (CO.sub.2/O.sub.2/N.sub.2 mixture, 12.6/14/73.4% vol) is used as the make-up gas to adjust the water content in the feed gas by using a water evaporator (IAS Model HOVACAL).

[0180] FIG. 2 is a cross-section SEM image of (A) porous PVDF support (B) composite membrane containing SHPAA/PVA with 0.2 wt % pGO

[0181] FIG. 3 is a schematic illustration of alignment of GO-based fillers in composite membranes fabricated using bar coating technique.

[0182] FIG. 4 is an SEM imaging of neat SHPAA/PVA composite membrane freeze-fractured hollow fiber (A) and cross-section (B).

[0183] FIG. 5 shows mixed gas permeation performance of various SHPAA/PVA-based selective layers measured at 35° C. for flat composite membranes.

[0184] FIG. 6 shows a mixed gas permeation performance of SHPAA/PVA-composite membranes with GO-based fillers as a function of filler loading measured at 35° C. for flat composite membranes.

[0185] FIG. 7 shows a CO.sub.2/N.sub.2 mixed gas permeation performance of various (A) 0.2 wt % (B) 0.5 wt % loaded GO-based membranes measured at 35° C. for hollow fiber composite membranes.

[0186] FIG. 8 shows a CO.sub.2/N.sub.2 mixed gas permeation performance of various (A) 0.2 wt % (B) 0.5 wt % loaded pGO-based membranes measured at 35° C. for hollow fiber composite membranes.

[0187] FIG. 9 shows mixed gas permeation performance of various composite membranes measured at 35° C. for (A) CO.sub.2/N.sub.2 gas pair at 1.7 bar and (B) CO.sub.2/CH.sub.4 gas pair at 2 for hollow fiber composite membranes.

[0188] FIG. 10 shows CO.sub.2/CH.sub.4 mixed gas permeation performance of various 0.2 wt % membranes measured at 35° C. for hollow fiber composite membranes.

[0189] FIG. 11 shows module permeation performance of SHBPAA/PVA+0.2 wt % pGO membrane (tested at 60° C.; Permeate side: 0.3mbar vacuum; Feed: Stack gas @ 10 L min.sup.−1

[0190] FIG. 12 shows module permeation performance of SHBPAA/PVA+0.2 wt % pGO membrane with and without mobile carriers (tested at 60° C.; Permeate side: 0.3 mbar vacuum; Feed: Stack gas @ 10 L min.sup.−1).

[0191] FIG. 13 shows the effect of SO.sub.x and NO.sub.x on composite membranes with mobile carriers—Effect of (A) NO and (B) SO.sub.2 on CO.sub.2 purity of Module 2; Effect of (C) NO and (D) SO.sub.2 on CO.sub.2 purity of Module 3.

MATERIALS

[0192] Poly(allylamine hydrochloride) (Mw=120,000-200,000) was purchased from Thermo Fisher Scientific, Sweden, and was purified and modified into sterically hindered polyallylamine.

[0193] For hollow fiber studies, Graphene Oxide powder (2.5 wt % in water) was supplied by Graphene-XT, Italy and used as diluted dispersions.

[0194] Polyvinyl alcohol (Mw=89,000-98,000, 89% hydrolyzed), 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide, N-hydroxysulfosuccinimide, 8-arm-poly (ethylene glycol)-NH.sub.2 (hexaglycerol core, Mn=10,000), were used as received from Sigma-Aldrich. Polyvinylidene fluoride (PVDF) ultrafiltration membrane (50 k MW) with polypropylene (PP) substrate was obtained from Synder Filtration, USA.

[0195] 3M™ Fluorinert™ Electronic Liquid FC-72 was used as received from Kemi-Intressen, Sweden.

[0196] L-proline Reagentplus® (≥99 wt %), 1-(2-Aminoethyl) piperazine (99 wt %), 1-Ethyl-3-methylimidazolium acetate (97 wt %) and sarcosine (N-Methylglycine) (98 wt %) were purchased from Sigma-Aldrich.

[0197] Poly (p-phenylene oxide) (PPO) hollow fibers used for hollow fiber supports with inside diameter of 350 μm and outside diameter of 540 μm was obtained from Parker A/S Norway.

[0198] CO.sub.2/N.sub.2 mixture (10 vol. % CO.sub.2 in N.sub.2), CO.sub.2/CH.sub.4 mixture (40 vol. % CO.sub.2 in CH.sub.4), N.sub.2 and CH.sub.4 (99.95%), used for permeation tests, were supplied by AGA, Norway. Hydrogen peroxide (H.sub.2O.sub.2, 30% in water) used in the modification of GO was supplied by Sigma Aldrich, Norway.

Characterization Methods

[0199] Chemical changes to nanofillers were monitored by Fourier-transform infrared (FTIR) spectroscopy using Thermo Nicolet Nexus spectrometer equipped with smart endurance reflection cell in attenuated total reflectance mode with a diamond crystal. An average of 16 scans with a resolution of 4 cm.sup.−1 was used in the range of 4000 cm.sup.−1 and 800 cm.sup.−1 to build the spectra.

[0200] The surface chemical composition of synthesized GO was analysed with an X-ray photoelectron spectroscopy (XPS, XPS-theta probe, Thermo Fisher Scientific Co., USA) equipped with a monochromatic Al Kα source with C-correction of 284.5 eV.

[0201] Membrane morphologies were analysed by Field Emission SEM APREO (FEI, Thermo Fisher Scientific, USA) equipped with an in-lens detector under immersion mode. Before analysis, the samples were sputter-coated with 8 nm Pd/Pt alloy.

Polymers

Polymer Example 1—Synthesis of Sterically Hindered Polyallylamine (SHPAA)—For Flatsheet Membranes

[0202] Sterically hindered polyallylamine was obtained by modification of purified polyallylamine with 2-bromopropane. Poly allylamine reacts with 2-bromopropane in the presence of stoichiometric amounts of KOH at 50° C. under reflux conditions in methanol yields poly-N-isopropyl allylamine, as shown in reaction Scheme 1.

##STR00007##

[0203] The poly-N-isopropyl allylamine prepared herein has an estimated Mw of 120 to 250K.

Polymer Example 2—Synthesis of Sterically Hindered Polyallylamine (SHPAA)—For Hollow Fibers

[0204] Polyallyl amine hydrochloride was purified by reacting with equivalent amounts of KOH in MeOH precipitating KCl. Subsequently, purified PAA was modified in to poly-N-isobutyl allyl amine by reaction with equivalent amount of 2-bromobutane and KOH in MeOH at 50° C. (Scheme 2). The resulting polymer was purified by separating precipitated KCl crystals followed by drying in N.sub.2 atmosphere at 60° C.

##STR00008##

[0205] The polymer prepared has an estimated Mw of 120 to 250 K

Mobile Carrier

Synthesis of Mobile Carriers

[0206] Equivalent amounts of i-proline and KOH were dissolved in DI water to form a solution of 10 wt % total solids. The solution was then stirred at high speed in room temperature overnight to form potassium L-prolinate (ProK).

[0207] Similarly, equivalent amounts of 1-(2-Aminoethyl) piperazine and sarcosine were stirred in calculated quantities of DI water at room temperature to obtain 37.7 wt % of 2-(1-piperazinyl) ethylamine sarcosinate (PZEA-SARC).

[0208] 1-Ethyl-3-methylimidazolium acetate ([Emim][OAc]) was dissolved in DI water to form a 10 wt % solution and stirred overnight at room temperature.

Nanofillers

[0209] Nanofillers for the flat sheet membrane examples were prepared as follows:

GO Example 1 Synthesis of Graphene Oxide

[0210] Graphene oxide used in the flat sheet membranes was synthesized through modified Hummer's method. 10 g of graphite powder was mixed with 450 ml of Sulphuric acid under stirring at 5° C. for 1 h. 30 g of potassium permanganate was then added and stirred for 30 minutes, resulting in a colour change from black to dark green. The solution was further heated up to 40° C. for 1 h. 450 ml of deionized water was added dropwise and carefully to avoid a rapid increase in temperature. The solution is characterized by a colour change to dark brown at this point. The temperature was then maintained at 95° C. for 30 minutes followed by adding 300 ml of 10% hydrogen peroxide solution and then stirred for 15 minutes. The colour change to light brown marked the successful synthesis of graphene oxide. The GO was then purified multiple times with about 5 L of 10% hydrochloric acid through a Whatman glass microfiber filter followed by washing in 3 L of acetone. The filtered GO cakes were then dried at 40° C. for two days under vacuum to give graphene oxide flakes named GO herein.

[0211] A 2 mg ml.sup.−1 solution of GO was prepared, and tip sonicated for up to 3 h followed by bath sonication for 30 mins. AFM analysis revealed the presence of flakes in this solution with lateral dimensions of 1 μm or more.

GO Example 2—Physical Modification of Graphene Oxide

[0212] In order to physically modify GO flakes for better diffusion of penetrants, random pores were introduced by hydrothermal treatment of GO example 1 using hydrogen peroxide. 1M NaOH was used to adjust pH of 75 ml of 1 mg ml.sup.−1 GO solution (diluted from the 2 mg ml.sup.−1 solution prepared earlier) was taken and the pH was adjusted to 10 using 1M NaOH solution. The mixture was stirred for 5 mins at high speed followed by bath sonication for 10 mins. 10 ml of 3% dilute hydrogen peroxide solution was then added to the mixture and the solution is stirred for 10 mins at high speed followed by bath sonication for 10 mins. The resulting mixture is then treated at 180° C. in a Teflon autoclave for 6 h and then cooled down to room temperature.

[0213] The resulting pGO (porous graphene oxide) dispersion in the water had a concentration of about 1 mg ml.sup.−1 (pGO). The flakes are expected to inherit the same lateral dimensions as GO from example 1.

GO Example 3—Chemical Modification of Graphene Oxide

[0214] PEG groups were grafted onto the GO surface using the EDC coupling reaction. Synthesized GO dispersion in water is acidic. However, in order to activate multiple sites for PEG grafting for amide bond formation, further carboxylic groups on GO surface were introduced by treating 20 ml of 4 mg ml.sup.−1 GO dispersion from Example 1 with an equal volume of 3M NaOH followed by bath sonication for 1 h at 25° C. This reaction enabled the conversion of esters in GO to be hydrolyzed into carboxylic groups. Dilute HCl was then added to neutralize the solution followed by dilution to 1 mg ml.sup.−1, obtaining a dispersion of carboxylated GO in water. 100 mg of NHS and 150 mg of EDC was then added to the GO—COOH dispersion, followed by bath sonication in ice for 30 mins to activate the catalysts. 200 mg of 8-arm PEG was then added to the mixture, and the solution was stirred at room temperature for 24 h. The solution was then centrifuged at 7000 rpm to remove aggregates, and the dispersion was then dialyzed in water using Dialysis membrane Spectra/Por® 3 to remove the catalysts, salts and other unreacted components. The residual dispersion had a GO—PEG concentration of about 1 mg ml.sup.−1.

[0215] The flakes are expected to inherit the same lateral dimensions as GO from example 1.

Nanofillers for the hollow fiber composite membrane examples are prepared as follows:

GO Example 4—GO/pGO Nanosheets—for Hollow Fiber Composite Membranes

[0216] An important parameter of GO flakes that influence the gas permeation performance is the flake size (lateral dimensions). Different suppliers of GO provide dispersions of different flake size. For hollow fiber composite membrane experiments, we used GO from commercial supplier Graphene-XT.

[0217] The GO dispersion as received was first diluted to 1 mg g.sup.−1 solution followed by pH adjustment to 10 using 1M NaOH. The diluted solution was sonicated in a bath sonicator for 30 minutes at 25° C. The dispersion is then subject to ultrasonic disintegrator (Vibra-Cell™ Ultrasonic Liquid Processor) at an amplitude of 60% in an ice bath with a 3 second pulse followed by 2 second break. This procedure was carried out to simultaneously exfoliate and control the size of GO flakes by varying the time of operation.

[0218] The sonication was carried out for 3, 6 or 9 h and the resulting GO flakes was referred to as GO3, GO6, and GO9, respectively.

[0219] Sonication-assisted exfoliation procedure was employed to obtain monolayers of GO in water dispersion. In order to ensure the reproducibility of the methodology, the concentration of GO dispersion was kept constant at 2 mg mL.sup.−1 and the sample volume was maintained at 300 mL for all procedures. Sonication process imparts random fragmentation of 2D nanosheets induced by mechanical failure of defective sp.sup.3 regions. These random lacerations are followed by propagation of cracks leading to reduced flake sizes.

[0220] AFM analysis revealed the presence of large flakes with lateral dimensions more than 1 μm for GO3. Subsequent sonication resulted in smaller flakes in the range of 400-800 nm and down to less than 500 nm for GO6 and GO9 respectively. All samples were then subject to hydrothermal treatment for introduction of random pores.

[0221] The size controlled GO dispersions were also hydrothermally treated to introduce random non-selective pores. The GO dispersion was mixed with 3 wt % H.sub.2O.sub.2 solution and the mixture was stirred vigorously for 10 mins followed by bath sonication for 10 mins. Thereupon, the mixture is treated in a Teflon autoclave for 6 h at 180° C. The resulting pGO dispersions derived from GO3, GO6 and GO9 samples were named as pGO3, pGO6 and pGO9, respectively.

[0222] The successful introduction of non-selective pores in GO nanosheets through the hydrothermal treatment is confirmed by representative S(T)EM imaging of GO3 and pGO3.

[0223] Representative imaging of pGO flakes show further reduction in flake size after the hydrothermal treatment. This size reduction is confirmed with relative increase in presence of carbonyl groups (observed from FTIR) that are exposed along the edges of the pGO when compared to GO.

[0224] Chemical changes in the GO nanoplatelets during the hydrothermal treatment process was studied using FTIR spectroscopy. The sonication procedures barely affected the chemical structure of the GO and pGO nanoplatelets.

[0225] However, discernible peak changes appeared between the GO and pGO.

Support

Flat Sheet

[0226] The flat sheet support was PVDF as a flat sheet. It had a MWCO of 50,000. The PVDF support was first washed in tap water at 45° C. for 1 h followed by DI water for 30 mins to remove the pore protective agent. The support was dried at room temperature overnight prior to coating with a casting solution using a bar coating machine (See FIG. 3). Pore filler FC-72 was used to fill the pores to avoid pore penetration of the casting solution.

Hollow Fiber Support

[0227] PPO was used as the hollow fiber support. It has a MWCO of 30,000 to 50,000. To fabricate hollow fiber supports, the PPO supports prepared by conventional hollow fiber spinning techniques were hung vertically with the ends sealed using paper clips which also create tension and avoids slackening of fibers. DI water was used to wash the fibers two times to remove possible dust particles sticking to the surface followed by drying in room temperature.

Composite Membrane Formation

Flatsheet Membranes

[0228] 4 wt % PVA solution in water was prepared by dissolving PVA pellets in deionised water at 80° C. for 4 h under reflux conditions. The SHPAA solution in methanol post modification (polymer example 1) was dried at 60° C. under vacuum overnight to remove residual solvent. The resulting pristine polymer was then dissolved in water for 24 h in room temperature to obtain a 6 wt % solution.

[0229] In the case of flat sheet supports, a cast solution concentration of about 1 wt % “solids” was used. The blend polymer solution of SHPAA/PVA consisted of 90 wt % SHPAA and 10 wt % PVA based on the total polymeric “solids” present in the solution. The % amount of nanofillers (from Ex GO1 to GO3), was measured with respect to the total amount of polymer and nanofillers present in the cast solution. For example, a 0.5 wt % GO in SHPAA/PVA blend denotes that the amount of GO is 0.5% of the total “solids” content, i.e. of the polymer and nanofillers in the solution.

[0230] The casting solution comprises 1 wt % of a [0.5 wt % and 99.5 wt %] blend of nanofillers and SHPAA and PVA polymer blend was applied to the PVDF support to prepare a selective layer of thickness lower than 200 nm.

[0231] In this example, the selective layer is applied by the bar-coating method, as schematically represented in FIG. 3.

Hollow Fiber Composite Membranes

[0232] Purified and dried SHPAA post-modification (polymer example 2) was dissolved in DI water to obtain a 6 wt % solution and the polymer solution is stirred for at least 2 days at room temperature to obtain a clear polymer solution. In the case of PVA, a 4 wt % solution was prepared by dissolving PVA pellets in DI water at 80° C. for 4 h under reflux conditions.

[0233] To prepare the casting solutions, calculated quantities of polymer solutions were added to DI water and diluted to a casting solution concentration of 0.15 wt % total “solids”. The amounts of mobile carriers were measured as ratio of polymer phase while the amounts of nanofillers (Ex GO4) were measured in terms of total solid content as described in Equations 1 and 2 respectively—

[00001]$\begin{array}{cc}{w}_{\mathrm{mc}}=w_{\mathrm{pol}}\times \left[\frac{1}{\left(1-c_{mc}\right)/100}-1\right]& \left(1\right)\end{array}$$\begin{array}{cc}{w}_{\mathrm{nf}}=w_{\mathrm{pol}}\times \left(c_{\mathrm{nf}}/100\right)& \left(2\right)\end{array}$

[0234] where w.sub.mc is the weight of mobile carrier (g), w.sub.pol is the total weight of dry polymer (g), w.sub.nf is the weight of nanofiller (g), c.sub.mc is the concentration of mobile carrier (wt %) and c.sub.nf is the concentration of nanofiller (wt %).

[0235] Coating of thin selective layer is achieved by dip coating the fibers using the casting solution in both directions at a constant low speed (in the range of 6-8 cm s.sup.−1) with a time interval of 30 mins between successive coating procedures. Coating in opposite directions ensures defect-free selective layer. Additionally, the lean viscosity of casting solution owing to the low solid content leads to uniformity of selective layer thickness independent of coating speed and filler loading. The hollow fibers are then dried in room temperature followed by drying at 60° C. under vacuum for 2 hrs to remove residual solvent components. The resulting hollow fibers spectacle shiny appearance due to the presence of ultrathin selective layer coating. The thickness of the selective layer is about 200 nm.

[0236] In order to assemble the coated hollow fiber composite membranes into a module, few fibers (in the range of 2-5) were inserted carefully into a pre-assembled stainless-steel hollow fiber module designed using ¼ inch ⅜-inch Swagelok™ fittings. The ends are then sealed using epoxy adhesive. The bore side of the fibers are open by knocking of cured adhesive on an extension mould.

Composite Membrane Morphology

Flatsheet Composite Membranes

[0237] Stable dispersions of GO-based fillers with both PVA and SHPAA/PVA blend matrices were obtained at the entire range of GO filler loading from 0.2 wt % to 1 wt %. The blend polymer solution of SHPAA/PVA consisted of 90 wt % SHPAA and 10 wt % PVA based on the total polymeric “solids” present in the solution.

[0238] The concentration of casting solutions was maintained at 1 wt % of solids. Representative cross-sectional SEM imaging of 0.2 wt % pGO loaded selective layer reveals the presence of an ultrathin selective layer with a thickness lower than 200 nm on the PVDF porous support. The surface images of the selective layers show visible difference between the neat polymer and those loaded with nanofillers.

[0239] While neat polymer layers show a relatively smooth surface, dark patches are observed in the composite membrane samples, which could be associated with the aligned GO-based filler flakes. No evident protrusion or aggregation of the nanofillers were observed from the smooth surface, which ascertains the alignment of GO nanoflakes parallel to the coating surface along their larger two dimensions. In-plane alignment of GO is attributed to the lowering of surface free energies of GO-based fillers and the mechanically forced-alignment of the thin 2D flakes into the tangent to the cylindrical surface of the bar at the contact point. FIG. 2 shows cross-section SEM images of (A) porous PVDF support (B) composite membrane containing SHPAA/PVA with 0.2 wt % pGO.

Lab-Scale Gas Permeation Performance

Flatsheet Membranes with Facilitated Transport SHPAA Polymer Matrix

[0240] Facilitated transport membranes transport CO.sub.2 through a reactive pathway in addition to the solution-diffusion mechanism. The facilitated transport effect is brought in by the amine groups attached to the main chain of the backbone in the SHPAA polymer matrix which reversibly react with CO.sub.2 in the presence of water.

Gas Permeation Performance

[0241] The composite membranes were evaluated for gas permeation performance using humid mixed gas permeation test rigs. The feed composed of 90/10 v/v CO.sub.2/N.sub.2 mixture or 40/60 v/v CO.sub.2/CH.sub.4 mixture. The flow rate of the feed was 300 ml min.sup.−1 for the CO.sub.2/N.sub.2 tests and 400-600 ml min.sup.−1 for the CO.sub.2/CH.sub.4 tests. The difference in feed flow rates was mainly to recoup differences in membrane areas and targeting very low stage cut of below 0.5%. The sweep gas for CO.sub.2/N.sub.2 tests was CH.sub.4 and for the CO.sub.2/CH.sub.4 tests, N.sub.2 was used. In both cases, feed and the sweep gas streams were humidified in a bubble tank before the membrane module. The shell side of the membranes was used for the feed gas and the bore side of the fibers was used as the permeate/sweep side. The pressure on the feed side was maintained constant at 1.7 bar for the CO.sub.2/N.sub.2 tests and varied between 2-20 bar for the CO.sub.2/CH.sub.4 tests. Sweep side pressure was held at 1.02 bar. The temperature of operation was maintained at 35° C. for all tests. The exit gas compositions in both feed and sweep side were monitored continuously using a pre-calibrated gas chromatograph (490 Micro GC, Agilent for CO.sub.2/N.sub.2 tests and MGS, SRI Instruments Inc. for CO.sub.2/CH.sub.4 tests). The permeance of component ‘i’ was obtained using the following equation

[00002]$\begin{array}{cc}{P}_i=\frac{V_p\left(1-{\gamma }_{H_{2}O}\right)y_i}{\left(\mathrm{avg}\left(p_{i,f}-p_{i,r}\right)-p_{i,p}\right)A}& \left(1\right)\end{array}$

Where the total permeate flow V.sub.p is in ml s.sup.−1 measured at the exit using a bubble flow meter at steady state conditions. y.sub.H.sub.2.sub.O and y.sub.i denote the molar fraction of the water and permeating species in the permeate flow respectively. Partial pressures p.sub.i,f, p.sub.i,r and p.sub.i,p of the species ‘i’ in the feed, retentate and permeate, respectively, are in cm Hg.sup.−1. Permeance of components are represented in GPU, where 1 GPU=10.sup.−6 cm.sup.3(STP) cm.sup.−2 s.sup.−1 cmHg.sup.−1=3.35×10.sup.−10 mol m.sup.−2 s.sup.−1 Pa.sup.−1. The separation factor is calculated using permeance of each component according to the equation

[00003]$\begin{array}{cc}{\alpha }_{i/j}=\frac{y_i/x_i}{y_j/x_j}& \left(2\right)\end{array}$

[0242] The separation performance of neat SHPAA/PVA blend membrane is CO.sub.2 permeance of 383 GPU and a CO.sub.2/N.sub.2 separation factor of 55, as shown in FIG. 5.

[0243] The separation performance of SHPAA/PVA flat sheet composite membranes with GO-based fillers are summarised in FIG. 5. At a low loading of 0.2 wt %, both GO and pGO effectively disrupt polymer chain packing while simultaneously increasing CO.sub.2 sorption and reorienting water distribution in the matrix. Hence the CO.sub.2 permeance sharply increases to ˜455 GPU in the case of 0.2 wt % GO and to 610 GPU in the case of 0.2 wt % pGO. The selectivity of the membranes containing both GO and pGO at 0.2 wt % loading rapidly decreases to ˜34.

[0244] A reduced CO.sub.2 permeance of ˜250 GPU with a CO.sub.2/N.sub.2 separation factor of ˜37 was observed in the membranes with low loading of GO-PEG. This reduction of performance can be explained with the effect of filler loading. Membranes with GO-PEG feature as a classic case of the rigidified interface between nanofiller and polymer matrix. Due to the strong interactions between —OH groups and the amine-containing facilitated transport matrix, there exists a compactly packed polymer interface between the GO surface the adjacent polymeric matrix. The tightly packed chains that form a rigidified interface volume in conjugation with the GO barrier property leads to reduced permeance of CO.sub.2 while increasing the CO.sub.2/N.sub.2 selectivity markedly at higher loadings. Thus, the CO.sub.2/N.sub.2 separation factor sharply rose to ˜90 while reducing the CO.sub.2 permeance down to 205 GPU at a high loading of 1 wt % filler.

[0245] A similar effect was seen with higher loading of GO nanoplatelets, where the tortuous pathways for N.sub.2 permeation due to the multi-layer alignment of GO leads to increased CO.sub.2/N.sub.2 separation factor of about 65 at 1 wt % loading. Thus, the optimal loading of GO-based fillers for enhanced permeation was observed at 0.2 wt %, above which the effect of barrier property of GO and hence the tortuosity to gas permeation is simultaneously manifested. At this loading, the enhanced permeation of CO.sub.2 due to the polymer chain disruption and increased sorption brought about by the high aspect ratio nanoplatelets counteracts with the resistance caused by the additional tortuosity of the impermeable platelets.

[0246] In the case of pGO, the presence of non-selective pores reduces the tortuosity, but the presence of bulk distribution of small-sized impermeable platelets is still significant. Hence, there is a decline in CO.sub.2 permeance with increasing filler loading, although not as steep as in the case of GO or GO-PEG.

[0247] FIG. 6 shows the effect of nanofiller loading on permeance and selectivity.

[0248] So the skilled person is able to tailor membrane properties to favour permeance or selectivity by varying the nanofiller content.

Lab-Scale Gas Permeation Performance

Hollow Fiber Membranes

[0249] Both GO and pGO nanofillers (GO example 4) were dispersed in SHPAA/PVA solution at two filler loadings of 0.2 wt % and 0.5 wt % in the selective layer. The total solid content in the casting solution (i.e. polymer+GO total) was maintained low at 0.15 wt % (polymer-based) that resulted in an ultrathin selective layer thickness of ˜200 nm on the PPO hollow fibers. Mobile carriers are also added in amounts shown below. FIG. 4 shows a composite membrane of the invention with neat polymerin the selective layer. The facile dip coating procedure also ensures in-plane alignment of GO due to shear alignment.

[0250] Both GO and pGO-based hollow fiber membranes were developed where the size of the GO was varied with sonication time resulting in GO3/6/9 and corresponding pGO3/6/9. Testing followed the protocol explained above for the flat sheet membranes. The neat polymeric membranes with SHBPAA/PVA exhibited a CO.sub.2 permeance of 407 GPU and a CO.sub.2/N.sub.2 separation factor of 32.2.

[0251] A small addition of pGO6 (optimized from FIGS. 7 and 8) at 0.2 wt % doubled the permeance up to 790 GPU with the selectivity remaining at 31. FIGS. 7 and 8 demonstrate that there is both an optimum loading of nanofiller and an optimum size of nanofiller.

Addition of Mobile Carriers

Hollow Fiber Membranes with Mobile Carriers

[0252] In order to increase the amount of reactive sites for the CO.sub.2 to interact in the selective layer, low molecular weight CO.sub.2-philic components were added. These are referred to as mobile carriers as they diffuse through the membrane matrix and enhance permeation.

[0253] 0.2 wt % of pGO was dispersed in polymer matrix containing 10 wt % [Emim][OAc] or 20 wt % ProK. The compositions were chosen according to the optimal composition detected in the experiments and reported in the previous sections. These resulting composite membranes with mobile carriers had an increased performance for CO.sub.2/N.sub.2 separation for ProK containing membranes, increasing the CO.sub.2 permeance up to 810 GPU as seen in FIG. 9A. Even better results are demonstrated with CO.sub.2/CH.sub.4 separation performances as seen in FIG. 9B. For these tests, the feed gas consisting of 40/60 v/v CO.sub.2/CH.sub.4 mixture, mimicking typical biogas composition was used. Both mobile carriers containing membranes were characterized with a marked increase in CO.sub.2 permeance while the CO.sub.2/CH.sub.4 separation factor remained constant around 20. The composite membrane containing 0.2 wt % pGO6 with 20% ProK peaked at a CO.sub.2 permeance of 825 GPU with a CO.sub.2/CH.sub.4 separation factor of 20, while the neat polymer had a CO.sub.2 permeance of 497 GPU with a CO.sub.2/CH.sub.4 separation factor of 21 at a feed pressure of 2 bar. The corresponding composite membrane containing 0.2 wt % pGO6 without mobile carriers was limited to 727 GPU with CO.sub.2/CH.sub.4 feed gas mixture. A similar increase in CO.sub.2 permeance of 10% [Emim][OAc] containing membrane up to 782 GPU was observed. The mobile carriers containing membranes exhibited an increased CO.sub.2 permeance at the total upstream pressure of 2 bar with the CO.sub.2/CH.sub.4 feed mixtures due to the increased partial pressure CO.sub.2 and lower stage cut (higher feed flow).

[0254] The effect of pressure on composite membrane performance was also investigated. The upstream pressure was increased from 2 bar to a maximum of 20 bar. Increasing feed pressure led to further distinguishable separation performance. Composite membranes of the invention are characterized with carrier saturation phenomenon at a high partial pressure of CO.sub.2 in the feed. Since the availability of fixed CO.sub.2 carriers (amine groups) in the polymer matrix is limited, increasing CO.sub.2 partial pressure in the feed gas leads to carrier saturation, hence decreases the CO.sub.2 permeance. Consequently, in all the systems discussed in this work, increasing pressure in the feed side reflects a drop in permeance of CO.sub.2 as seen in FIG. 10.

[0255] Interestingly, the composite membrane with loading 2D fillers, both GO6 and pGO6, exhibited increased resistance to carrier saturation phenomenon especially in the pressures of 5 bar and 10 bar. Thus, the corresponding CO.sub.2 permeances remained at 340 GPU and 450 GPU when compared with neat polymer at 300 GPU at 5 bar, which is typical operational pressure for biogas upgrading.

[0256] The composite membrane that contained mobile carriers showcased further resistance to carrier saturation phenomenon even further due to the increase of available CO.sub.2 carriers as expected. The effect remains evident across the entire pressure range of testing for both 10 wt % [Emim][OAc] and 20 wt % Pro-K loaded membranes. These membranes showcased a CO.sub.2 permeance of 463 GPU and 468 GPU and CO.sub.2/CH.sub.4 separation factors of 24 and 25, respectively, at 5 bar feed pressure.

[0257] Selected membranes were scaled up and tested on-site

Industrial Testing of Composite Membranes with Mobile Carriers

[0258] Hollow fiber composite membranes are prepared using the same PPO support as previously described. The selective layer was applied by dip coating. The coating solution concentration was about 0.15 wt % total solids and the amount of pGO was always maintained at an optimized concentration of 0.2 wt % with respect to the polymer content in the solution. The amount of mobile carriers was 10 wt % for [Emim}[OAc] and 20 wt % for ProK.

[0259] 3 scaled up modules with membrane area ranging from 130 cm.sup.2 to 200 cm.sup.2 each were assembled and tested. The material configuration in the modules and the mechanical aspects are summarised in Table 1.

TABLE-US-00001 TABLE 1 Summary of second prototype modules with 3.sup.rd generation materials Module Polymer Nano- Mobile No. of Length Perme- # matrix filler carrier fibers of fibers ating area 1 SHBPAA/ 0.2 wt % none 70 20 ~200 cm.sup.2 PVA pGO 2 SHBPAA/ 0.2 wt % 10% 55 15 ~150 cm.sup.2 PVA pGO [Emim][OAc] 3 SHBPAA/ 0.2 wt % 20% ProK 55 20 ~175 cm.sup.2 PVA pGO

[0260] The gas test was carried out using flue gas from the emission stack (height: 105 m.) placed close to the 5-stage cyclone pre-heater of the grey clinker production line at the Colacem Cement Plant located in Gubbio (PG) Italy.

[0261] The sampling point was located 30 meters below the top of the stack. A hole was made on the sidewall of the chimney and the vacuum pump was used to suck the flue gas from the chimney to the membrane module. A 2 μm ceramic filter, part of the gas sample probe (M&C Model SP180H), was used to remove the suspended particulate matter in the flue gas. The temperature of the flue gas from the stack, during the plant tests, was about 115° C. The composition of the dry flue gas, during the plant tests, is summarized in Table 2.

TABLE-US-00002 TABLE 2 Composition of the flue gas from the grey clinker production line 2.sup.nd pre-pilot field test Cement kiln Production parameters Raw Meal feeding ~192 t/h (t/h) Fuel feeding (t/h)  ~9 t/h Clinker ~117 t/h Production (t/h) Stack emission Concentration in components the dry flue gas CO.sub.2 (% v/v) 10.5~12.0 O.sub.2 (% v/v) 14.0~15.5 N.sub.2 (% v/v) 73~76 CO (ppmv)  50~100 NOx (ppmv) 100~120 SOx (ppmv) 0~3 NH.sub.3 (ppmv) 20~40 HCl (ppmv) 0.5~2   Suspended P.M. 1~3 (mg/Nm.sup.3) (before removal)

[0262] The flow chart of the in-field membrane permeation test is shown in FIG. 1. Except for the removal of suspended particulate matter from the flue gas by the filter, no further pretreatment was carried out for the feed gas. All the gas transport pipes were covered by electric heating tubing bundle to control the desired temperature for the test mainly to avoid humidity condensation. One needle valve was placed at the retentate side to control the feed pressure. The gas composition of the feed gas was measured by the Multicomponent Analysis System ACF-NT of ABB S.p.A., while that of the retentate and permeate streams was measured by the HORIBA PG 350 SRM and with TESTO Model 350XL-350S gas analyzers. The flow rates of the three streams were measured by TSI Model 4143 and confirmed by some floating element flow meters selected in accordance with the flow rate to be measured. If not specified, all the tests were carried out without vacuum or sweep gas. In the case that sweep gas was used, the composition of the IP grade gas was: 20.93% Oxygen and 79.07% Nitrogen.

Module 1: Effect of Vacuum and Stability Tests

[0263] Module 1 showcased a significant increase in CO.sub.2 flux with the use of vacuum on the permeate side of the membrane. The recorded flux was double when compared to usage of sweep. The tests were also prolonged over a period of two weeks to simultaneous estimate the stability of the membrane at the maximum performance (FIG. 11). Given that there is no pretreatment of stack gas which still contains traces of SO.sub.x and NOx, the obtained long-term performances translate to significant material stability. Additionally, the drop in fluxes can be attributed to the water condensation in the pores and fluctuations in feed water content in real time.

Module 2 and 3: Effect of Water Content

[0264] Water plays an important role in facilitated transport membranes. Since three modules 1, 2 and 3 have three different chemical configuration with respect to amine chemistry, the modules were subject to changing water content in the feed. An external evaporator system (HOVACAL) was used to force additional water into the feed stream. The results of permeation performance is shown in FIG. 12.

[0265] In general, all modules exhibited increasing flux and purity of CO.sub.2 with increasing total water content in the feed gas. This proves that water as a carrier plays a major role in activation of the amine groups present in the backbone of the facilitated transport polymer chain. Additionally, high fluxes in both ProK and [Emim][OAc] containing modules were evident due to increase in total amount of CO.sub.2-reactive groups present in the selective layer. [Emim][OAc] also acts as a physical solvent for CO.sub.2 and hence the purity of the module containing [Emim][OAc] is lower than the other two modules due to physical sorption of both CO.sub.2 and N.sub.2 in the mobile phase.

[0266] Modules 2 and 3 were also tested for stability in the presence of SO.sub.x and NO.sub.x. Composite membranes containing mobile carriers (ProK and [Emim][OAc]) were exposed to SO.sub.x and NO.sub.x in simulated flue gas and virtually no change in purity of CO2 in permeate was observed. Results are presented in FIG. 13.

Comparison to Existing Membranes Tested at Pilot Scale

[0267] Performance of membrane modules at larger scale when compared to the lab scale tests have been quantified in literature with flux and selectivity/purity of CO.sub.2 in the permeate. However, there are challenges involved in estimation of fluxes in terms of GPU (1 GPU=10.sup.−6 cm.sup.3(STP) cm.sup.−2 s.sup.−1 cmHg.sup.−1=3.35×10.sup.−10 mol m.sup.−2 s.sup.−2 Pa.sup.−1). Permeance (flux) in GPU can be calculated only with assumptions on steady state driving force across the membrane. With these assumptions, the fabricated membranes have the following permeances in GPU are shown in Table 3.
These estimations are with the following assumptions— [0268] Partial pressure of CO.sub.2 is negligible in permeate side due to smaller module size and use of continuous vacuum [0269] Flat CO.sub.2 concentration profile in the feed side, this is justified by the low stage cut (<5%) and the shorter length of the module

TABLE-US-00003 TABLE 3 Estimation of CO.sub.2 permeances in pre-pilot modules (variations with water content) CO.sub.2 Permeance CO.sub.2 purity CO.sub.2 flux Module (GPU) (single stage) (NL m.sup.−2 h.sup.−1) Module 1:  860-1133 52-55 335.6-418.9 0.2 wt % pGO Module 2: 1093-1376 47-51 529.0-603.8 0.2 wt % pGO with 10% [Emim][OAc] Module 3: 1130-1419 49-53 635.9-752.6 0.2 wt % pGO with 20% ProK

[0270] A surprising aspect of the present invention is therefore that the membranes comprising the mobile carriers have increased CO.sub.2 permeance and increased CO.sub.2 flux compared to the same modules without the mobile carriers, at industrially relevant conditions.

[0271] Benchmarking performances of the fabrication membranes with other membranes tested in industrial pilots can be done using the Table 5 obtained from Int. J. Greenhouse Gas Control, 86 (2019), pp. 191-200 as shown below. (Reference to each membrane can be found in the article).

TABLE-US-00004 TABLE 4 Summary of membrane pilot scale CO.sub.2 capture test. CO.sub.2 Membrane Membrane Feed Temp permeance Selectivity/ materials type Pressure (° C.) (GPU) CO.sub.2 flux CO.sub.2 flux purity in permeate PES HF 6~8 bar —  60 — — 40 Prism HF 1.32 bar RT 400~500 ~142 ~142 3~6 (polysulfone) NLm.sup.−2h.sup.−1 NLm.sup.−2h.sup.−1 PVAm (FSC) HF 2.5 bar 35 740 ~500 ~500 135 NLm.sup.−2h.sup.−1 NLm.sup.−2h.sup.−1 PVAm (FSC) HF 1~6 bar 23-45 — ~20 ~20 ~65% CO.sub.2in NLm.sup.−2h.sup.−1 NLm.sup.−2h.sup.−1 permeate side Module 1 HF 1.7 60  860-1133 335.6-418.9 385.4-468.09 52-55 NLm.sup.−2h.sup.−1 NLm.sup.−2h.sup.−1 Module 2 HF 1.7 60 1093-1376 529.0-603.8 591.63 47-51 NLm.sup.−2h.sup.−1 NLm.sup.−2h.sup.−1 Module 3 HF 1.7 60 1130-1419 635.9-752.6 635.94 49-53 NLm.sup.−2h.sup.−1 NLm.sup.−2h.sup.−1 (HF = hollow fibre)

Conclusions

[0272] Composite membranes containing GO-based fillers in ultrathin selective layers were fabricated and tested. GO-based fillers were found to benefit composite membranes to increase the CO.sub.2 separation properties depending on their lateral dimensions and loading. pGO fillers derived from size-optimized GO nanosheets at a loading of 0.2 wt % form continuous CO.sub.2 permeation pathways along the CO.sub.2-philic pGO surface with reoriented water channels surrounding 2D structure in the matrix. These composite membranes were characterized with a high CO.sub.2 permeance of 780 GPU and a corresponding CO.sub.2/N.sub.2 separation factor of 30. Composite membranes with mobile carriers that reversibly react with CO.sub.2 were also developed as hollow fibers.

[0273] It was found that the mobile carriers ProK and [Emim][OAc] were especially able to improve the separation performance of fixed site SHPAA/PVA membrane due to their high mobility and reversible interaction with CO.sub.2 to form bicarbonate/carbonate species and carbene-CO.sub.2 adducts respectively. As a new concept, composite membranes were combined with mobile carriers that resulted in membranes with CO.sub.2 permeance of 825 GPU. These membranes were evaluated for both CO.sub.2/N.sub.2 and CO.sub.2/CH.sub.4 gas pairs and resulted in a separation factor of 31 for CO.sub.2/N.sub.2 and 20 for CO.sub.2/CH.sub.4. Due to the relative increase in the content of CO.sub.2-philic species and reinforcement with addition of pGO, the composite membranes with mobile carriers were stable for feed pressures up to 20 bar and exhibited increased resistance to carrier saturation phenomena. This high stability and gas separation performance when combined with easily scalable hollow fiber configuration establishes the commercial viability of the fabricated membranes for CO.sub.2 separation applications.

[0274] In particular, a surprising aspect of the present invention is that the membranes comprising the mobile carriers have increased CO.sub.2 permeance and increased CO.sub.2 flux compared to the same modules without the mobile carriers, at industrially relevant conditions.