WO2016162518A1 - Composite membrane - Google Patents

Abstract

This invention developed a high performance membrane by mimicking the natural enzyme and combining it with membrane separation. The addition of selected mimic enzymes (Zn-cyclen or Zn-cyclam) in a water swollen membrane significantly increased both the CO₂ permeance (by 5 times as compared to PVA membrane) and selectivity of CO₂ from other less reactive gases (doubled).

Description

Composite Membrane

This invention relates to a composite membrane. In particular, it relates to a composite membrane comprising a selective layer comprising a polymer matrix and a zinc complex wherein the zinc is coordinated to a multi-dentate nitrogen containing l igand with a molecular weight of less than 2000 g 'mol . The invention also relates to a process for selectively separating C0₂ from a mixed gaseous feed stream using said composite membrane. Background of the Invention

Combustion of fossil fuels has met the ever growing energy demand but it has resulted in unchecked levels of C0₂ emission. These carbon dioxide emissions are considered to be a major cul rit in global warming and cl imate change. Carbon capturing technologies vary significantly depending on point source due to the great diversity in pressure, temperature and composition of sour gas streams. Several technologies have been investigated for CO_? capture but high capital investment and operational costs are major hindrances in their large scale industrial application.

Currently, amine-ba.sed absorption is the leading technology for C0₂ separation. However, amine absorption is energy intensive and has associated pollution risks due to solvent emission. Membrane separation is well recognized to be environmentally friendly and less energy intensive but w ith a very smal l share in the market. This is at least in part because the cost of the separation using current commercial membranes is much higher than amine-based absorption as a result of limited membrane separation performance when using conventional membrane materials. Innovative membrane materials may be the solution to significantly reduce C0₂ separation costs. However, such membranes should ideally have both high selectivity and a C0₂ permeance that is above the so-called Robeson upper bound.

Facil itated transport membranes for C0₂ separation exhibit a potential to achieve both high selectivity and C0₂ permeance. In a C0₂ facil itated transport membrane, C0₂ transport occurs by means of a reversible reaction of C0₂ with complexing agents (carriers) in the membrane. Un-reactive gases such as N₂ and CM i permeate only by the solution-diffusion mechanism.

The first studied facil itated transport membranes for gas separation were supported liquid membranes (SLMs) with mobile carriers, in which carrier solutions were impregnated in the pores of a microporous support. SLMs have serious degradation problems. The loss of carrier solution due to evaporation or entrainment with the gas stream, or the deact ivation of the comple ing agent makes these types of membranes unstable and therefore not suitable for large scale applications.

Fixed-site-carrier (FSC) membranes were introduced to overcome the above limitations and contained carriers covalentiy bonded to the polymer backbones (e.g. amino groups). In FSC membranes the carriers have restricted mobility, hence the stabil ity is no longer an issue, but the gas diffusivity (and thus permeabil ity) in these membranes is 2-3 orders of magnitude lower than that in a mobile carrier membrane.

As a special type of facilitated transport, the use of immobilized carbonic anhydrase (CA) in water swollen membranes to mimic the mechanism of the mammalian respiratory system (a special type of C0₂ facilitated transport ) reportedly leads to very high C0₂ selectivity over other gases, even at low partial pressures.

Carbonic anhydrases are metalloproteins that reversibly catalyse the hydration and dehydration of C0₂ at ambient temperature and physiological pH. Carbonic anhydrase facilitates extremely fast CO_? hydration rates, which arc typically l imited only by the diffusion rate of C0₂ to the active site. It has the abil ity to catalyze the hydration of 600,000 molecules of C0₂ per molecule of CA per second, 4000 times faster than m o n o e t h a n o I a m i n e (MEA) in terms of catalytic activity and with an energy consumption of almost 7 times less.

Very high selectivity at very low C0₂ concentrations (1%-0.1%) was documented by Yang et al in 2006 for a membrane with immobilized CA.

Moreover, a liquid membrane system catalysed by CA has been developed by Carbozyme Inc. through a NETL project (US), and preliminary results show a potential for significant decreases in energy penalty and cost.

The immobilization of CA has been extensively investigated and studies have focused on aspects such as cross-linking the enzymes to form a gel. CA can also be covalcntly attached or encapsulated within polymeric membranes. However, the lack of long-term stability (life time around 6 months) and the permanent loss of enzyme activity due to slight changes in pH and temperature are major drawbacks associated w ith this type of membrane. In addition, the enzyme is very costly and hence not suitable for large scale applications.

A number of small molecules are know n to mimic the active site of carbonic anhydrase, including macrocyclic and tripodal complexes of zinc ( I I ) and other

2__|_

metals. The active site consists of Zn ^' coordinated by three histidine imidazole groups and either a w ater molecule or hydroxy! group, depending on pH. The generally accepted mechanism of C0₂ hydration by carbonic anhydra.se involves nucleophilic attack by the Zn-hydroxyl group on the dissolved CO_? molecule.

In US 8066800, tetrahedral ly coordinated zinc hydroxide moieties are used as synthetic analogues of the active site of carbonic anhydrase and are incorporated into a membrane comprising a polymer matrix for use in CO_? separation. The zinc moieties are chemical ly linked to the polymer to provide increased selectivity of the membrane for C0₂. However, this l inkage means that the membrane has a quite rigid structure and thus gas permeability is low. Moreover, these systems are primarily hydrophobic and thus are not compatible with aqueous conditions or the separat ion of CO_? from feed streams comprising water vapour. There therefore remains a need to develop and identify new membrane-based systems for use in carbon dioxide capture.

In view of the above-mentioned chal lenges, it is an object of the present invention to develop a new membrane for C0₂ separation. A membrane w hich offers high separation efficiency of C0₂ from the other components in the feed stream, such as N₂, H₂ and CH (, is sought. Ideal ly, the membrane should also have high C0₂ permeabil ity to help maximise efficiency. It would be advantageous if the membrane worked in an aqueous environment. Ultimately, it is desired if the membrane is suitable for commercial applications.

It has surprisingly been found that this may be achieved by utilising a hydrophilic, w ater swollen selective layer comprising a polymer matrix and a zinc complex w herein the zinc is coordinated to a multidentate nitrogen containing ligand with a molecular w eight of less than 2000 g/mol . These synthetic zinc complexes resemble the active site of CA enzymes and offer the advantages of being more stable to impurities and having high thermal stability (over 200°C) and good activity over a wide pH range compared to the enzyme itself. Unexpectedly, both CO:

selectivity and permeability are found to be high. The low molecular weight of the zinc complex leads to additional benefits over the use of the enzyme: there are more active sites per unit mass of compound, leading to faster reaction rates per unit weight. The present invention thus offers the potential advantages of a more cost- effective, durable and stable membrane substitute to those comprising CA enzymes.

Moreover, it is possible to tailor these facilitated transport membranes by adding one or more nano-fillers to the selective layer. This offers the possibility for a further improvement in separation performance. The nano-filier can open up the polymer chains in the membrane to improve the facilitated transport effect and gas diffusion, thereby increasing the CO_? permeation. Summary of the inv ention

Thus, viewed from one aspect, the invention provides a composite membrane comprising a water swollen selective layer for separating C0₂ from a mixed gaseous feed stream, wherein the selective layer comprises:

(i) a polymer matrix; and

(ii) a zinc complex, wherein zinc is coordinated to a multidentate

n i t rogen -con t a i n i n g ligand with a molecular weight of less than 2000 g/mol;

and a porous support on which said selective layer is carried.

In a particularly preferred embodiment, the selective layer further comprises at least one nano filler.

Viewed from a further aspect, the invention provides a process for separating

CO_? from a mixed gaseous feed stream containing CO_?, said process comprising contacting said gaseous feed stream with a composite membrane comprising a water swollen selective layer for separating CO_? from a mixed gaseous feed stream comprising:

(i) a polymer matrix: and (ii) a zinc complex, wherein zinc is coordinated to a multidentate n i trogen -con ta i n i n g ligand with a molecular weight of less than

and a porous support on which said selective layer is carried.

Viewed from a further aspect, the invention provides the use of a composite membrane as hereinbefore defined in a process for separating C0₂ from a mixed gaseous feed stream.

Detailed Description of the Invention

The present invention describes a hydrophilic, water swollen composite membrane which may be used in a process for separating C0₂ from a mixed gaseous feed stream. The composite membrane comprises a selective layer carried on a support. The selective layer comprises a polymer matrix and a zinc complex, wherein in the zinc complex the zinc is coordinated to a multidentate nitrogen- containing ligand with a molecular weight of less than 2000 g/mol. The selective layer is carried on a porous support.

Gaseous Feed Stream

The mixed gaseous feed stream used in the process of the invention may be any gas stream comprising a mixture of at least two gases, wherein one of these gases is C0₂. The use of flue gas is especially preferred.

In a preferred embodiment, the feed stream comprises (e.g. consists of) nitrogen (N₂) and C0₂. In an alternative preferred embodiment, the feed stream comprises (e.g. consists of) methane (CH₄) and C0₂. In a further alternative embodiment, the feed stream comprises (e.g. consists of) hydrogen (H₂) and C0₂.

The feed stream may comprise 1 to 90 vol%, preferably 2 to 85 vol%, more preferably 5-60 vol%, such as 10-50 vol% CO? relative to the total amount of gas present. In one particular preferred embodiment, the feed stream comprises 5 to 50 vol% C0₂ relative to the total amount of gas present (e.g. when the feed stream comprises natural gas, flue gas or biogas).

It will be appreciated that in addition to the gases mentioned above, the gaseous feed stream may comprise further gases. Examples of such further gases include hydrogen, methane, nitrogen, NOx, carbon monoxide, hydrogen sulfide, hydrogen chloride, hydrogen fluoride, sulfur dioxide, earbonyl sulphide, ammonia, oxygen and heavy hydrocarbons such as hexane, octane or decanc.

In a particularly preferred embodiment, the gaseous feed stream may comprise flue gas from powerplants or other inductrial sources, such as cement and steel manufacturers. It will be understood that by "flue gas" we mean a mixture comprising nitrogen, NOx and sul fur dioxide in addition to carbon dioxide and other optional gases such as oxygen.

In a particularly preferred embodiment, the gaseous feed stream may comprise syngas, most preferably pre-combustion syngas, i.e. syngas which has yet to be combusted for power production. It will be understood that by "syngas" we mean a mixture comprising hydrogen and carbon monoxide in addition to carbon dioxide and other optional gases such as hydrogen sulfide.

Alternatively, in a further preferred embodiment, the gaseous feed stream may comprise biogas or natural gas, i.e. a mixture of gases comprising methane and carbon dioxide in addition to other optional gases such as hydrogen sulphide and carbon monoxide.

When the gaseous feed stream comprises biogas or natural gas, the process of the invention is primarily used to separate C0₂ from CH₄. Natural gas is a combustible mixture formed primarily of methane, but it can also include sour gas carbon dioxide and hydrogen sulphide. The composition of natural gas can vary widely, but typically contains methane (70-90 vol%), ethane/butane (0-20 vol%), nitrogen (0-5 vol%) carbon dioxide (0- 1 2 vol%) and hydrogen sulphide (0-5 vol%) before it is refined. C0₂ in natural gas should ideally be removed (natural gas sweetening) to meet specifications in order to increase heating value (Wobbe index ) and reduce corrosion of pipelines. Ideal l y, the C0₂ content shoul d be reduced to < 2 vol%. Biogas is a mixture of gases generated from anaerobic microbial digestion from organic wastes such as manure, landfill or sewage. The composit ion of biogas varies depending on the source. Typically biogas contains 60-65 vol% CH₄, 35-40 vol% CO₂, small amounts of hydrogen sulfide (H₂S), water vapour and t races of other gases. Depending on the source, nitrogen (N₂) may be present. The removal of carbon dioxide (C0₂) from biogas to a level of methane (CH₄) > 90 vol%, termed "upgrading^{" "}, can not only effectively increase the Wobbe index, but also reduce corrosion caused by acid gas and therefore e tend the biogas utilization as a renewable energy resource. Upgraded biogas containing at least 98 vol% of CH₄ may be compressed and liquefied for vehicle fuel or injected into a public natural gas grid.

Composite Membrane The composite membranes of the invention arc selective barriers which have a retentate side and a permeate side. The "retentate" side comprises those components which have not passed though the membrane and the "permeate" side comprises those components which have passed through the membrane.

The membranes comprise a selective layer comprising a polymer matrix and a zinc complex as hereinbefore defined.

The selective layer is hydrophilic, i.e. having an affinity for water. In the present case, the selective layer is water swollen and is therefore hydrophilic. A hydrophilic selective layer allows for a high water swell ing degree, thus providing sufficient water to take part in the C0₂ hydration cycle.

The selective layer is water swollen. By "water swollen" we mean that the selective layer has been swelled in the presence of water, e.g. in the form of water vapour. The selective layers may thus be considered h yd rated, i.e. containing water molecules. The selective layer may be sw el led by 25-85wt% of water relative to the weight of the polymer in selective layer as a whole. In general, the higher the amount of water the better, subject of course, to the membrane maintaining good mechanical properties. The swel l ing degree varies depending on the feed gas humidity and the type of polymer employed. Ideally, the selective layer may operate in a humid atmosphere, e.g. at least 75% relative humidity.

The membrane of the invention may be in the form of a bundle of hollow fibres or a flat sheet.

The membranes of the invention may be described as a composite membrane with a non-porous or dense selective layer on a preferably asymmetric porous support.

The thickness of the selective layer is preferably in the range of less than 5 um, preferably less than 2 um, such as 1 um or less. The layer may be at least 0.10 um in thickness such as at least 0.25 um in thickness.

The selective layer of the invention is carried on a porous support. The combination of the selective layer and the porous support may be termed "composite membrane". Figure 1 shows a typical structure of the selective layer of the invention carried on a porous support.

Suitable porous supports are porous membranes known in the art and arc ones which are porous to the gas being transported. Typical supports are made of polymers including pol sulfone (PSf), poly-vinyl idinc fluoride (PVDF), polyether- imide, poiyethersulfone, po I y t e t ra fl u o ro ethylene, polypropylene, polyimide, polycthcrketone. polyphenylenc oxide (PPO), aliphatic polyamides,

polyacrylonitrile or a cel lulose acetate support. Polysulfone is particularly preferred.

The support may be in the form of a flat sheet membrane or hollow fibre membrane. In a flat sheet support, a non-woven fibre layer is commonly used to prov ide mechanical strength.

In al l embodiments, it is preferred if the porous support layer has a thickness of less than 500 um, preferably less than 300 um, more preferably 200 um or less, such as 50- 100 um.

Typically, the porous support will be asymmetric, i.e. the pores vary in size across the support, typically graduating from smal ler pores at the side of porous support closest to the selective layer to larger pores at the side of porous support furthest from the selective layer.

The molecular weight cut off (MWCO) of the porous support may be more than 20,000, preferably more than 25,000, more preferably more than 30,000, such as more than 50.000. MWCO is essentially a measure of the pore size of the support, with larger MWCO values representing higher pore sizes. By MWCO we mean the molecular weight of the components which are substantially (i.e. at least 90%) retained on the retentate side of the composite membrane and are prevented from passage through the porous support.

In addition to the porous support, polymer matrix and the zinc complex, the composite membrane may comprise further components. Additives such as polyamines and hydroxides of alkali, alkali earth metals may be added to increase the pH. This facilitates deprotonation of the iigand. Other additives include poly amides, pH buffers of alkali metals/ alkal i earth metals that can maintain a pH greater than 7.

In a particularly preferred embodiment, the composite membrane further comprises at least one nanofilier. The nanofilier(s) typically form part of the selective layer. By "nanofilier" we mean any fil ler material which is dispersed in the membranes (preferably the selective layer) of the invention and which has at least one dimension in the nanoscale. i.e. in the range 1 to 100 nm. The nanofilier(s) of the invention typically have at least one dimension in the range 1 to 50 nm, preferably 1 to 25 nm. The nanofil ier may be one-dimensional, two-dimensional or three-dimensional. By "one-dimensional" we mean that only a one dimension of the material is in the nanoscale, by "two-dimensional" we mean that two dimensions of the material are in the nanoscale, and by "three-dimensional" we mean that all three dimensions of the material are in the nanoscale. Dispersion of the nanofiiier(s) in the selective layer of the composite membrane is considered to increase the volume of the water swollen in the polymer matrix. As water plays an important role in the facil itated transport mechanism of the membranes, this improved water uptake is considered to enhance the CO_? transport through the membrane, and hence to improve the separation performance.

At least one nanofilier may be present. Thus, it is possible for a single nanofilier or a mixture or two or more nanofi l lers to be used. In a preferred embodiment, a single nanofil ier is employed.

The nanofilier may be selected from any suitable material know n in the art. Preferably, however, the nanofilier is selected from the group consisting of carbon nanotubes (CNTs), graphene oxide nanosheets, nanocellulose, zeolite frameworks, metal-organic frameworks (MOFs) and particles of TiO_? or silica. In a particularly preferred embodiment, the nanofil ler comprises, preferably consists of, carbon nanotubes, especially hydrophilic carbon nanotubes.

The at least one nanofil ler is typical ly present in an amount of up to 20 wt%, such as 0.1 to 10 wt%, preferably 0.2 to 5 wt%, more preferably 0.5 to 1.5 wt%, such as 1.0 wt% relative to the total weight of the membrane. A particularly preferred amount of the nanofiller is 1 .0 wt%. It will be understood that if more than one nanofiller is present these amounts correspond to the total amount of all nanofi l lers in the membrane.

The pH of the membrane is usual ly in the range 2 to 14, preferably 5 to 12, especially 12.

Particularly high C02 permeance together with good separation performance is found for a membrane at pH 12 comprising 1.0 wt% nanofiller, relative to the total weight of the membrane. This therefore represents a particularly preferred embodiment of the in vention .

The composite membrane of the invention may be prepared by methods known in the art. Typical methods involve casting a solution of the selective layer components onto the support or immersion of the support in such a solution. The method used may be dependent upon the form of the composite membrane. The selective layer is typically cast on to the support using a coating process. Such processes are well known in the art and can include processes such as solution casting, dip-coating and spray-coating. Alternatively it can be made by interfacial polymerization or in-situ polymerization, or any other phase inversion method.

It will be appreciated that the composite membrane when in use is water swollen. Water can be introduced into the composite membrane during manufacture (e.g. via a humid atmosphere) or after the membrane is synthesised. If the composite membrane has a high percentage swelling, the more h yd rated the membrane and thus the more water molecules are present. This enhanced water uptake helps to increase the speed of reaction with C0₂, since H₂0 is required for the reaction. The process of operation ideally involves the use of humidified gas to ensure water swells into membrane. Polymer matrix

The selective layer comprises a polymer matrix. By "polymer matrix" we mean a polymer material which forms the base material of the selective layer. The polymer matrix is water swollen so comprises any polymer that is hydrophiiic, i.e. having an affinity for water.

The weight average molecular weight of the polymer is preferably in the range 10,000 to 500,000, such as 40,000 to 200,000. Ideally, the Mw of the polymer matrix is higher than the molecular weight cut off (MWCO) of support substrate.

Exemplary polymers include Poly vinyl Alcohol ( P V A ) and Copolymers, Poly vinyl amine ( PVAm ), Poly al ly! amine ( PAAm ), Poly vinyl imidizol and Copolymers, poly vinyl pyridine (PVP) and Copolymers, polyethylene imine (PEI), Cellulose acetate (CA), Poly amidoamine (PA MAM ), Poly ethylene oxide (PEO), poly ethylene glycol (PEG) and Polyacrylonitrile (PAN ).

In addition to these polymers, other hydrophiiic commercial ly available polymers such as PEBAX® or Sul fonated PolyEther Ether Ketone (SPEEK) may be used.

The polymer may be a homopolymer or a copolymer, preferably a homopolymer. Where the polymer is a copolymer, typical comonomers include linear or branched alpha olefins such as CI -6 a!pha-oiefms.

A particularly preferred polymer is PVA, especial ly a PVA homopolymer.

In one embodiment, the polymer is crosslinked.

Zinc complex

The zinc complex comprises a zinc ion coordinated to a multidentate n i t rogen -con t a i n i n g l igand which has a molecular weight of less than 2000 g/mol. The zinc complex may be termed a "mimic enzyme" because it mimics the active site of carbonic anhydrase.

2

The zinc is preferably in the +2 oxidation state, i.e. Zn ^' . By "multidentate" we mean a ligand which is capable of donating two or more pairs of electrons to the zinc ion. Preferably, the ligand is bidentate, tridentate, tetradentate, pentadentate or hexadentate. Most preferably, the ligand is tridentate or tetradentate, i.e. is capable of donating three or four pairs of electrons, respectively.

By "nitrogen-containing" we mean that the ligand contains at least one nitrogen atom. In ail embodiments, it is preferable if the multidentate ligand comprises two or more nitrogen atoms, such as three or four nitrogen atoms.

Ligands with five or six nitrogen atoms are also env isaged. Ligands containing four nitrogen atoms are particularly preferred.

In the context of the invention it is env isaged that the nitrogen atoms in the l igand will act as the donor atoms which coordinate to the zinc metal ion. i.e. the atoms which donate a pair of electrons to the zinc. How ever, it is also possible for other atoms such as oxygen or sulfur to coordinate to the zinc. Preferably, the donor atoms of the l igands of the invention arc nitrogen atoms.

The multidentate ligand may be saturated or unsaturated and may comprise aliphatic and/or aromatic moieties.

The molecular weight of the multidentate l igand is less than 2000 g/moi. Preferably, the molecular weight is less than 1000 g/moi, more preferably less than 800 g/moi, such as less than 500 g mol, e.g. less than 400 g'mol.

In a preferable embodiment, the l igand is a n i t rogen-contai n i n g cyclic l igand

(i.e. a macrocycle). Where the ligand is a cyclic ligand, the l igand typically comprises at least two nitrogen atoms l inked by aliphatic and/or aromatic carbon chain linkers to form a cycl ic molecule.

In a preferable embodiment, the linker is selected from the group consisting of linear or branched C_1-20 alkyi groups, C_3-12 cycloalkyl groups, and C₆-20 ary! groups, each of which may be optionally substituted. One or more heteroatoms such as oxygen or sulfur may also be present in the linker.

Preferably, the linker is a l inear or branched Ci_₂o a!kyl group, such as a Ci_io alky! group. The term "alkyi" is intended to cover linear or branched alkyl groups such as ethyl, propyl, butyl, pentyl and hexyi. It will be understood that the "alkyl" group in the context of the linker is divalent and thus may also be referred to as "alkylene". Particularly preferable alkyl groups are ethyl and propyl . In all embodiments, the alkyl group is preferably l inear.

Examples of aryl groups include phenyl, benzyl, phenylalkyl or naphthyl. Optional siibstituents may be selected from the group consisting of aryl groups, alkyl groups having 1 to 8 carbon atoms, acyl groups, or a nitro group.

In a cycl ic multidentate ligand. the nitrogen atoms are typical ly bonded to two linkers as hereinbefore defined and a further moiety selected from the group consisting of hydrogen and C_1-10alkyl, preferably hydrogen or methyl.

In an alternative embodiment, the nitrogen atoms may be present as primary amine siibstituents on a C_3-12 cycloalkane, such as cyclohexane.

Examples of cycl ic l igands include:

Tetramethylcyclam H₃q CHi

[13]aneN4

1 ,4,7-triazacyclononane 1 1

1 ,4,7, 10,13, 16- Hcxaazacyclocctadccane

' J

Pentacyclen

In a more preferable embodiment, the l igand is an aza macrocycle and typical ly comprises at least three itrogen atoms, preferably three to six nitrogen atoms. Particularly preferred aza macrocycl ic l igands are tri-aza macrocyclic ligands or a tetra-aza macrocycl ic ligands. By "tetra-aza macrocyclic ligand" we mean a cyclic molecule containing four nitrogen donor atoms that can coordinate to a metal centre. Thus, the four nitrogen donor atoms in the ligands used in the present invention coordinate to the zinc metal ion.

In a particularly preferred embodiment, the l igands of the invention are tetra- aza macrocyclic ligands, especial ly cyclen and eye I am, which have the structures shown below:

Cyclen Cyclam

An illustration of the catalytic mechanism of CO_? hydration promoted by the zinc complex is shown in Fig 2 and reactions 1-3. A carbon dioxide molecule

9

attaches to the Zn^{" '} active site to form a meta-stablc complex. The complex is then attacked by a Lewis base (OH^") to produce bicarbonate (HCCh ). In this a two-step process, CO_? is converted to HCO ¾ and the active site in the mimic enzyme are left unreacted. The mechanism is further elaborated in equations 1-3 below.

(2)

(3)

In the presence of water, in addition to the multidentate l igand. the zinc metal will typically have a water molecule or hydroxyl group bound to it, depending on pH . As shown above, the presence of this water molecule is necessary in the mechanism of action of the mimic enzyme. The kinetic rate constant for the reaction, and thus the hydration of C0₂ can be estimated by correlation between rate constant and dissociation constant (pKa). The hyd ation of C0₂ by the zinc complex is a simple protonation reaction, so it is possible to predict kinetics of C0₂ hydration based on its pKa. Since H₂0 is involved in the CC HCC ^" reaction cycle, a highly hydrophil ic pol ymer is preferred for use as the membrane material in the selective layer to retain water, and the selective layer is designed to operate in the highly water swollen state to achieve the best separation.

The pKa of the Zn complex is preferably less than 10, more preferably in the range 7 to 10. The pKa is defined as the dissociation constant of the complex, which in this case is the pH above which 90% of the water molecule bound to the Zn is deprotonated to a hydroxyl group. This deprotonation facilitates the reaction. The process is therefore ideal ly performed at a pH which is higher than the pKa of the Zn complex.

The zinc complex is preferably not chemical ly linked to the polymer matrix, but rather is free to move around within polymer matrix.

The amount of Zn complex in the polymer matrix is typically less than 1 0 ⁴ mo I, preferably less than 10° mol in per gram of polymer matrix such as PVA in membrane. There is preferably a minimum of 10^~8 mol of Zn complex, such as at least 1 0^"7 mol of Zn complex per gram of polymer matrix. Suitable amounts include 0.001 to 2 mmol/g, preferably 0.002 to 1 .2 mmol/g of polymer matrix.

Process

The processes of the in vention are used to separate C0₂ from a mixed gaseous feed stream.

In all embodiments, it is preferred if the process of the invention results in the capture of at least 80%, more preferably at least 85%, even more preferably at least 90%, such as at least 95%, e.g. at least 99%> of C0₂ present in the original gaseous feed stream.

The gaseous feed stream, after contact with the membrane, typical ly comprises less than 10 vol%, preferably less than 5 vol%, more preferably less than 2 vol%, such as less than 1 vol.% of C0₂ relative to the total amount of gas present.

The invention will now be described with reference to the following non limiting examples and figures.

Description of Figures Figure 1. Composite membrane with a selective layer coated over an asymmetric microporous support

Figure 2. C0₂ hydration promoted by mimic enzyme

Figure 3. PVA films containing various concentrations of CNTs

Figure 4. SEM picture of the PVA membrane selective layer on PSf support

(MWCO50,000)

Figure 5. Cross-section SEM picture of a PVA-CNT membrane on a PSf support. Figure 6. Humidified membrane testing setup

Figure 7. Titration of Zn -eye I am (a) and Zn-cyclen (b) against aOH to determine dissociation constant (pKa)

Figure 8(a). Comparison of C0₂ selectivity in membrane with and without mimic enzyme at different pH Figure 8(b). Comparison of C0₂ permeance in membrane with and without mimic enzyme at different pH

Figure 9. Effect of mimic enzyme concentration of membrane ( 0.0025- I mM of mimic enzyme/gram of PVA) on CO permeance

Figure 10. Effect of mimic enzyme concentration of membrane (0.0025- 1 mM of mimic enzyme/gram of PVA ) on CO₂/N₂ selectivity

Figure 11. Swell ing behavior of PVA membranes w ith and without CNT loadings, at pH 5 and pH 12 under fully humid conditions at 25"C.

Figure 12. Effect of CNT loading in a PVA membrane (at pH 5 ) on the C0₂ permeance and CO₂/N₂ selectivity. Fully humidified feed gas containing 10% CO₂ in N₂ at 1 .2 bara and 25°C.

Figure 13. Effect of pH on the CO₂ permeance and CO: ^selectivity of a PVA membrane containing 1 .0 wt% CNTs. Fully humidified feed gas containing 10% C0₂ in N₂ at 1 .2bara and 25"C.

Figure 14. Effect of relative humidity on the CO₂ permeance (a) and CO₂ /N₂ selectivity (b) of a PVA-mimic enzyme membrane with and without 1 .0 wt% NTs at pH 12. Feed gas containing 10% C0₂ in N₂ at 1.2bara and 25°C.

Figure 15. Effect of pressure on the C0₂ permeance (a) and CO₂/N₂ selectivity (b) of a PVA membrane with and w ithout CNTs and mimic enzyme at pH 12. The humidified feed gas contains 10% CO₂ in N₂, at 25°C.

Examples Materials

Analytical grade Zinc perch lorate hexahydrate (in crystal line form ), 1 ,4,7, 10- Tet raazac yc I ododec an e (97% purity ), 1 ,4,8, 1 1-Tetraazacyclotetradecane (98% purity), absolute ethanol, D₂0 (99.9%), Polyv inyl alcohol Mwt 89000-99000 (99% hydrolysed ) and cellulose acetate bio fibers were purchased from Sigma Aldrich, NaOH and HCT were purchased from VWR, PSf ultrafiltration membrane ( MWCO 50,000) was suppl ied by Al fa Lava and the special ly treated hydrophilic multi- wal led CNTs (VGCF-X™, diameter 1 5nm, length 1-3 μηι ) were supplied by Showa Denko K.K. (SDK., Japan). Ail these chemicals were used without further purification. Synthesis and characterization of Zn cyclen and cyclam Synthesis

The synthesis of the catalysts involves the treatment of the aza-macrocyclic scaffold with the perch locate salt of the metal, and heating of the mixture at 50-60°C. All the raw materials for catalysts are soluble in ethanol but both catalysts are insoluble in ethanol . 5 ml solution of cyclen /cyclam in ethanol added with 1 0 ml of Zn (C lOjb solutions in ethanol over a period of 1 .5 hr. Both reactants were added in a molar ratio of 1 : 1. The reactions involved in the synthesis of catalysts are presented I equation 4-5.

At the end of reaction produced catalyst was separated from solution using vacuum filtration and washed several times by absolute ethanol. ¹ H NMR 400 and MS were used to verify the completion of reaction.

Characterization of mimic enzymes

ID ' [ { M R sepctra of mimic enzymes in D₂0 was recorded on a Brukcr Avance DPX 400 Hz NMR spectrometer at 25°C. Samples were prepared by dissolving mimic enzymes in D₂0 at room temperature. Solutions were then placed in NMR tunes and were subjected to ultrasonic mix ing to ensure that composition of sample is constant throughout the sample. Data recorded by NMR was tested using TopSpin™ software. Furthermore, Mass spectroscopy was conducted to determine the molecular composit ion of produced mimic enzymes and determine their molecular weight experimentally.

Dissociation experiment (calculation of pKa values) Potentiometric measurements were carried out in a Mettier Toledo G20 compact titrator equipped with a pH-electrodc DSC-115 (uncertainty ± 0.02 pH) and temperature sensor DT100 (uncertainty ±0.1°C). The temperature in the jacketed glass vessel ( 1 00 ml vol ume) used for titration was control led using a Julabo M4 heating circulator (temperature stability ± 0.1 C). Dow 10 cSt silicone oil was used as heat transfer medium. The cal ibration of the glass pH-eiectrode was performed at 25°C temperature using buffer solutions traceable to the National Bureau of Standards (pH 4.01 , 7.00, 9.2 1 and 12 from Mettier Toledo).

About 30 g of catalyst solution ( 10 mg/ 30 g-H₂0) were titrated with 0.1 M

NaOH solution. The optimum titrant addition rate was selected in order to keep the temperature in the titration vessel constant within ±0. 1 "C. All measurements were carried out in duplicate and average values are reported. The titration procedure was computer control led. All data were recorded using the LabX 3. 1 software provided by Mettier Toledo.

The pKa values were calculated in the software as pH at half equivalence. Uncertainties in the experimental data are mainly determined by the uncertainty of the electrode used, which is 0.02 pH.

Preparation and characterization of membranes

Membrane preparation

A thin dense layer of Poly vinyl alcohol (PVA ) containing mimic enzyme was casted over PSf support membrane using dip coating procedure. A calculated amount of PVA powder was added in deionized water to produce a 1 0 wt% mixture.

The mixture was kept at 90"C for 1.5 hour to get a transparent, clear solution, and then left for rol ling overnight at room temperature. This PVA solution was then diluted with mimic enzyme solution (in Dl water), and 0. 1 M NaOH solution. Ratio of these solutions was adjusted to produce a dilute solution with PVA concentration varying from 1 to 3 wt%, and pH up to 12, based on calculation for various membrane thicknesses. All membranes were produced by dip coating technique. A flat sheet Psf ultrafiltration membrane washed with DI water was masked on a glass slab to prevent solution from entering the back side of the membrane. This support membrane was then dipped into the said solution for 30 seconds. After lifting the membrane above the coating bath, it was held nearly vertical (5-10° against a wall) to allow extra solution flowing off evenly and form a homogeneous and continuous film. The coated membrane stayed dust-free to dry at ambient condition for 3 hours and then recoated following the same procedure. Every membrane was coated twice to ensure that the entire membrane surface is evenly coated and to eliminate possible defects. The twice coated membrane was left in the dust -free space overnight, and then to dry in a convection oven at 45°C for 3 hours.

Carbon nanotube ( CNT)-containing membrane preparation

PVA nanocomposite membranes containing mimic enzyme and well-dispersed CNTs in a thin, dense selective layer over a PSf porous support were prepared by dip coating. The thicknesses of the selective layers were determined by adjusting the concentrations of the PVA in the casting solutions. A concentrated PVA solution ( 1 0 wt%) was first prepared, and then this PVA solution was diluted with CNT dispersed water and sonicated for 4 hours at an intensity level of 40 and then for 1 hour at an intensity level of 20 by using a sonication device (Sonics Vibra-Celi™). PVA is a hydrophiiic material with a pH value in the casting solution at 5 (acidic). As it is necessary to maintain a pH value higher than the pKa of the chosen mimic enzyme to catalyze the hydration reaction, the pH value of the P V A-based casting solution was adjusted to be basic by adding 0.1M NaOH solution. The weight of the bottle containing this PVA-CNT solution was measured before and after sonication to determine the loss of water by evaporation. The deionized water was added to this solution to maintain the concentration of PVA at 2.0 % by weight. The loadings of the CNTs were calculated with respect to the PVA in the solution. The casting solution containing 2.0 wt% PVA and different loadings of CNTs (0.5, 1 and 1 .5 wt%) with respect to the PVA was filtered by using a 5iim PTFE porous filter to remove large CNT clusters and to ensure the continuous dispersion of the nano- fillers. Effect of CNT concentration on the casting solution can be observed visually. PVA films containing CNTs (0.5, 1 and 1.5 wt%) were casted. As presented in Figure 3, the degree of darkness in the color of the sample films increases with respect to CNT concentrations.

Defect free membranes containing different concentrations of CNTs were casted by dip coating on porous polysulfone (PSf) support. A flat sheet PSf support membrane washed with deionized water was masked on a glass slab and dipped into the casting solution for 30 seconds. The coated membrane was withdrawn and held nearly vertically to allow any extra solution to flow off evenly and form a homogeneous and continuous film. This procedure was repeated twice to ensure a defect-free and consistent coating. The twice-coated membrane was dried overnight at room temperature in a dust-free envi onment and then heat treated at 1 10"C for 1 hour to produce a physical iy cross-linked membrane.

Membrane morphology

The cross-section and surface of the membranes were examined using field emission scanning electron microscopy (FES EM, Zeiss Ultra 55 Limited Edition). Cross-sectional samples were prepared by fracturing the membranes in liquid nitrogen. Samples were coated with gold before observation. A SEM image of the cross-section of a PVA membrane coated for the study is shown in Figure 4.

For the nanotube membranes, the SEM images of the membrane surface indicate a uniform and defect-free membrane. Th e thickness of the membrane can be determined from the cross-section SEM image, as seen in Figure 5. A casting solution containing 2 wt% PVA was used to cast the membrane, as shown in Figure 3, with a thickness of approximately (). 3iim. The thicknesses of all the membranes were controlled by adjusting the concentration of the PVA in the casting solution at 2.0 wt%. Multiple samples were casted at this concentration and the selective layer thicknesses were measured varying between 0.81 and 0.85 iim.

Swelling test

The swelling behavior of the membrane materials under humid conditions at 20"C was measured. The samples (polymer films) were vacuum dried at 20"C for 3 hours prior to the tests, and then placed in a closed container saturated with water vapors. Samples were weighted after regular intervals to determine the uptake of water by membrane from the water vapour saturated air. Degree of swelling was calculated based on gravimetric analysis of samples. Equation (6) for calculation of swelling degree is presented below.

W - W

% swell =— * *- x 100 (6)

W_d

Where W_s and \\ \ are the masses of the swollen and dry membranes, respectively.

For the carbon nanotube containing membranes, the experiments in a preliminary study showed that the dispersion of more than 2.0 wt% CNTs in a PVA solution was not stable, and hence the CNT loadings in the PVA polymer solution were set to be less than 2.0 wt%. The low ratio of stable CNT dispersion could be due to the long and entangled chains of PVA and the high v iscosity of the PVA solutions. The CNT loading was thus varied at 0.5, 1 .0, and 1.5 wt % in PVA, and the effect of the CNT loadings was evaluated to find the optimal loading. PVA membranes with various CNTs loading and with and without pH adjustments were cast in a Teflon dish to produce self-supported films. The casting solution (5ml) was poured onto a Teflon dish and dried at 45"C overnight to produce samples of similar thickness. The sample films were peeled off from the Teflon dish and heated at 1 10°C for 1 hour, then cooled down to room temperature and dried in vacuum chamber at 25°C for at least 3 hours to remove the absorbed moisture. The sw elling behav ior was then measured as described above. Permeation test

Separation performances of membranes were tested in a specially designed gas permeation rig with humidifiers (see Figure 6). Feed gases were supplied from a pre mixed gas cylinder ( 1 0 vol . % of CO_? in CO2/N2 gas mixture, AG A AS ). A flat sheet type membrane module was mounted in a thermostatic cabinet with a temperature control system. A membrane (diameter of 50mm.) was sandwiched between the permeate chamber and the feed gas chamber, supported by a porous metal disk and sealed with rubber O-rings. Both feed gas and sweep gas were saturated with water vapour by bubbl ing through their respective humidifers. A bypass line with a precise valve paralleling with the upstream humidifer was attached to adjust the humidity of feed gas. F low rate and pressure were recorded and control led by a flow control ler, flow indicator and pressure transmitters (M S) respectively and logged directly into a computer (by Labview). Relative humidity of feed gas was measured online by a humidity analyser. The composition of the permeate gas was analyzed online by a gas chromatograph equipped with a thermal conductivity detector (MicroGC3000). Some operation conditions, such as feed gas flow rate, relative humidity, feed pressure, were varied to investigate their respective effects on the membrane performance. A sweep gas was used on permeate side (He or CM 4 indicated in Fig. 6) for better recording of fluxes and gas compositions. Permeance of a species i was defined as the flux divided by the partial pressure differences betw een the upstream and downstream of the membrane and reported in units of [ m (STP)/(m h bar)] and given the symbol Pj. The selectivity ( a ) was calculated from permeance of CO , P( 02 and N2, P 2 , as expressed in Eq. (7):

(7) CO2- 2 ( 10 vol.% C0₂) mixed gas was used as feed gas. Permeation experiments were carried out at 25°C with a feed pressure varying from 1 bar to 3 bar, and separation performance was recorded when the system had been stabilized.

Characterization of mimic enzyme

Mimic enzymes synthesized for this work were tested using UN MR and Mass spectroscopy to validate the completion of reaction. The results from NMR verify the existence of mimic enzymes. Elemental analysis in MS validated the structure and molecular weight of the two compounds. Molecular weight of both mimic enzymes calculated based on elemental analysis is presented in Table 1 .

Table 1. Molecular weight of mimic enzymes

Dissociation constant of Mimic enzymes

A mimic enzyme needs to be protonated to catalyze the hydration of C0₂ . In order to efficiently use the full potential of mimic enzyme, the dissociation constant

(pKa) should be known and the process should be designed to operate at a pH higher than pKa value of mimic enzyme. Titrations were conducted at 25"C to determine the pKa of the selected mimic enzymes (Zn-cyclen and Zn-cyclam). Experimental results of titration with NaOH are presented in figure 7 and the dissociation constant of mimic enzymes are listed in Table 2.

Table 2 Dissociation constant of mimic enzymes

The results from experimentation are in good agreement with literature. These values indicate both mimic enzymes protonate at a pH over 7 (neutral). It suggests that slightly basic environment must be obtained in the hydration process. Furthermore, compared to Zn-cyclam shown in figure 7, Zn-cyclen has a sharper increase in pH with addition of NaOH over p a value. Hence it can be concluded that Zn-cyclen has a lower pKa and higher rate of protonation, which means that Zn- cyclen is easier to be activated and more suitable to be used in the membrane.

Characterization of membrane

PVA is a hydrophil ic polymer with excel lent film forming abilities. However, an aqueous solution of PVA is acidic in nature (pH 5-6). Sodium hydroxide was added in membrane casting solution to achieve a basic pH (9-12). Experiments were to study the effect of NaOH in the casting solution on separation performance of PVA membrane.

Mimic enzyme is a catalyst which hydrates C0₂ and a small quantity of mimic enzyme can show significant improvement in hydration rate. PVA

membranes containing O.O l mmol of mimic enzyme per gram of PVA were prepared and the solution adjusted to be at pl l 5, 7, 9 and 1 2 and tested under humidified conditions. Experimental results presented in figure 8(a,b) compares the

performance of membranes containing mimic enzyme with that of their counterparts not containing mimic enzyme. CO_? permeance of the membrane containing mimic enzyme with the solution adjusted to pH 9 shows the highest value of 0.68 m (STP)/ (m₂ bar h) ( Figure 8 (a)), and C0₂ permeance of the two membranes containing mimic enzyme are appreciably higher than that of their counterparts w ithout mimic enzyme at the same pH value when the pH is more than 7. Similarly the two membranes containing m imic enzyme show appreciably higher CO₂/N₂ selectivity than that of their counterparts without mimic enzyme at the same pl l value when the pH is more than 7, i.e. CO₂/N₂ selectivity 105 and 1 13 at pH 9 and 12, respectively.

From these figures it is quite clear that mimic enzyme in a membrane promotes the CO₂ transport and improves both CO₂ permeance and selectivity of CO₂/N₂ at the same time in membranes with pH value larger than the pKa value of the mimic enzyme Zn-cyclen. It is important for the mimic enzyme to be in protonated state for the facilitation of hydration reaction. Experimentation with membranes at pH 7 and 5 shows no significant increa.se in CO₂ permeance or selectiv ity, as the pH level of casting solutions in these membranes is lesser than pKa of mimic enzyme Zn-cyclen (pKa =7.9 ), hence lack excessive amount of OH ions to protonate the mimic enzyme. The results also show that by adjusting the pH from 9 to 12, only slight increase in selectivity (from 1 07 to 1 13) was obtained with no noticeable change in CO₂ permeance. Based on this study it can be said that pH plays an important role in membrane separation performance, and non-protonated mimic enzym e (pH of casting solution lesser than the pKa of mimic enzyme) present in membrane does not catalyze the hydration reaction, or provide facil itated transport function.

Effect of m imic enzyme concentration on the performance of membrane was studied by adjusting its concentration in casting solution. Six different

concentrations of mimic enzyme (0.0025- 1 mM per gram PVA ) were tested. Figure 9 compares the performance of these membranes based on their CO₂ permeance. When the mimic enzyme concentration in the casting solution is more than

0.005m , a decline in performance of membrane with increasing the mimic enzyme concentration is observed.

Although by increasing the concentrat ion the number of catalyst sites should increase, the CC½ permeance in fact decreased. It is due to that mimic enzyme is a catalyst which deactivates in the presence of HCC ions. When relatively high concentration of mimic enzyme presented in membrane, firstly the excess OH^" ions resulting from pH adjustment may become not sufficient for the protonation of the active sites of the mimic enzyme, and secondly the excess production of HCO₃ ^" ions produced by hydration reaction deactivates the catalyst sites. Although the HCO₃ ^" ions produced by hydration reaction diffuses through the membrane and dissociates from the permeate side of the membrane, the equilibrium between C0₂, HCCh and water exists. Hydration reaction of C0₂ by mimic enzyme requires water and a small quantity of catalyst. Few mmoies of mimic enzyme is a smal l quantity of catalyst in casting solution, yet its

concentration rises appreciably as the coated membrane is dried. Depending upon the degree of swel l ing, concentration of mimic enzyme in selective layer can be several moles per gram of water in membrane. High concentration of mimic enzyme not only exhausts the supply of OH ions by pH adjustment but also leads to a higher p a value.

imic enzyme is an amorphous solid with high solubil ity in w ater.

How ever, its solubility might have a limitation. As in figure 9, permeance of membrane with ImM mimic enzyme in casting solution is less than PVA membrane without mim ic enzyme. This shows that the water filled amorphous domains of polymer membrane have become less available for the CO₂ transport. The mimic enzyme due to its high concentration in membrane has filled up these domains and resulted in low ering of permeance. CO₂/N₂ selectivity of this membrane is as low as that of the not treated PVA membrane (without pH adjustment or addition of mimic enzyme).

Characterization of CNT containing membranes

PVA membranes containing 0- 1 .5 wt% CNTs were prepared with the pH value in the casting solution adjusted to be 5. 9 and 1 2 to study their effect on the

performance of the membrane. The concentration of the mimic enzyme in the membrane casting solutions was O.OOSmmol /g PVA.

Effect of CNT loadings on the degree of water swelling

The effect of CNT loadings on the swelling degree (water uptake) of membrane was first studied and optimized. Four sets of sample films with and without CNTs at pH 5 and 12 were placed in a humidified environment at room temperature for several days to investigate their swell ing conditions during the process. A grav imetric analysis of these samples was conducted periodically to determine their degree of swelling based on the weight gain. The degree of swelling was calculated based on equation 6. The percentage of water swelling in the four samples is plotted with time in Figure 1 1.

As presented in Figure 1 1 . it can be seen that the degrees of water swelling of al l sample films increase with time. The PVA sample without pH adjustment seems reached its equilibrium state at the end of the test, i.e. 96 hrs, while the other three samples tend to approach their respective equilibrium degrees of swelling, but at different rates. The PVA sample shows the lowest water swelling capacity, while the sample of PVA with the addition of CNTs shows a much quicker increase in the degree of swel ling with time, and a higher degree of swelling at the end of the test, suggesting a higher equil ibrium water uptake. By the addition of 1 .0 wt% CNTs with respect to PVA, the degree of swelling increases by appro imately 30% after 96 hours. By adjusting the pH from 5 to 12, the degree of swelling of the

CNTs/PVA sample increases dramatically from 130% to 212%. These results clearly show that the addition of CNTs can significantly improve the water swelling capacity of PVA, and the effect of CNT addition on the degree of swelling is more prominent in the PVA samples made at pi 1 12 as compared to the samples made at pH 5. The increases in the degree of swelling implies that the addition of the hydrophil ic CNTs may have positive effect to water solubility in the PVA matrix, which results in an increase in water uptake of the membrane.

Effect of CNT loadings on the CO₂/N₂ separation performance

The separation performance of a PVA membrane containing CNTs (0.5-1.5%) was studied to determine the optimal CNT loading. Figure 1 2 shows the effect of CNTs on the separation performance of a membrane without pH adjustment (i.e., pH=5).

As can be seen, the CO; permeance increases with the increase of the CNT loading, and reaches the highest C0₂ permeance of 0.34 m (STP)/(m₂ bar h) at the CNT loading of 1 .0 wt%, which is nearly doubled of that of the PVA membrane without CNTs. A further increase in the CNT loading from 1 .0 wt% to 1 .5 wt%, however, results in a notable drop in the C0₂ permeance. Meanwhile, the C0₂/N₂ selectivity decreases monotonically with an increase in CNT loading, but the effect is not significant: only an approximately 10% decrease was recorded from a CNT loading of 0 to 1 .0 wt%. The increase in C0₂ permeance with the CNT loading suggests that the addition of the surface modified hydrophiiic CNTs has positive influence to the C0₂ transport in this membrane. This may be due to the enhancement of water uptake capacity in the CNT -PVA matrix. Nevertheless, the presence of CNTs and the enhanced water swel l ing capacity did not lead to the facilitated transport of C0₂, as the C0₂/N₂ selectiv ity did not increase.

When the CNT loading is increased from 1 .0% to 1.5 wt% with respect to PVA, an obvious decl ine in separation performance can be observed. The C0₂ permeance for the PVA membrane containing 1.5 wt% CNTs was found to be appreciably lower than that of membrane containing 1.0 wt % CNTs. The C0₂/N₂ selectivity of the PVA membrane containing 1 .5 wt% CNTs is also slightly lower than all other tested membranes. This is bel ieved to be due to the more occupied gas diffusion path by the dispersed CNTs in the PVA matrix at a higher CNT loading. The CNTs used in this work has a very low density (0.08g/cm³), which is 1 5 times low er as compared to the density of the PVA film (1.2- 1 .3g'cm ). An increase in CNT loading from 1.0 to 1 .5 in weight percentage with respect to PVA would result in a very large increase in terms of the volume percentage. The benefit again by the enhanced water swel ling from the increased CNTs is thus counterweighed.

Experiments confirm that the addition of w el l-dispersed CNTs in the PVA membrane is an effective method to enhance water uptake and hence the separation performance in this membrane, especially the C0₂ permeance. As a significant drop in membrane separation performance was obtained by increasing the CNT loading to over 1 .0 wt%, the optimal CNT loading w ith respect to PVA was determined to be 1 .0 wt% CNTs in this membrane.

Effect of pH on the CO₂/N₂ separation performance

The effect of pH on the separation performance of a PVA membrane containing the optimal loading of CNT (1.0 wt%) is presented in Figure 13. It can be seen that both the C0₂ permeance and the C0₂/N₂ selectivity increase with the increase of the pH value. By adjusting the casting solution from pH 5 to 12, the C0₂ permeance of this membrane increased from 0.34 to 0.44 m (STP) (m_? bar h ) (25% increase) and the selectivity increased from 52 to 60 (13% increase).

The above results suggest that the separation performance of the water swollen CNT/PVA membrane can be improved by adding CNTs and increasing the pi 1 value to be basic (by the addition of the NaOH solution). The improvements can be attributed to the increase of water swelling capacity in the membrane, which m ay- increase the chain flexibility and the free volume in the PVA matri and eventual ly results in an increase in the CO_? permeance. The results also show that the addition of CNTs and adjustment of the pH can only contribute to the increase in CO_? permeance without fundamental impact on CO₂/N₂ selectiv ity, suggesting that the CO₂ transport mechanism through water swollen PVA is still solution-diffusion: the CO₂/N₂ selectivity is not above the theoretic selectivity (approx. 60) calculated from the solubility and diffusivity values of CO₂ and N₂ in water at the operating conditions. Effect of relative humidity on the CO₂/N₂ separation performance

The water swollen in this membrane not only opens up the PVA polymer network and increases the gas/ion diffusivity, but most importantly, it also attaches to the mimic enzyme as a temporary ligand to complete the catalytic hydration cycle. By adding CNTs to the membrane, the water swelling capacity of the PVA selective layer increases (see Figure 1 1). The excess water favors the activation of the mimic enzyme to promote the CO_? facil itated transport. Relative humidity in the feed side of the membrane has previously been found to strongly influence the water swelling degree of the membrane. The effect of relative humidity in the feed gas (50- 1 00%) on the CO₂ permeance and CO₂/N₂ selectiv ity of the PVA-mimic enzyme membranes with and without CNTs was studied, which confirms the influence of CNT addition, as presented in Figure 14.

Figure 14 shows the effect of relative humidity on the CO_? permeance and CO_? /N₂ selectivity of a mimic enzyme promoted PVA membrane with and without the addition of 1 .0 wt% CNTs. As can be seen in Figure 14, the relative humidity in the feed gas has a very strong influence on the CO_? separation performance. Both PVA- mimic enzyme membranes w ith and without CNT loading show clear increase in C0₂ permeance as well as CO₂/N₂ selectivity with increasing relative humidity, which suggests that the mimic enzyme-based PVA membrane should operate under fully humidified conditions to take advantage of the mimic enzyme catalytic function in the CO_? hydration reaction to improve the CO₂ /N₂ separation

performance.

The plots in Figure 14 also show that the effect of the relative humidity on the performance in the membranes with and without CNT loading is different. The CO_? permeances of both membranes in Figure 14(a) are quite similar in the low relative humidity range, i.e., when relative humidity is lower than 80%, where the CO_? N_? sclectivities of both membranes arc also similar. When the relative humidity increases to be 80% and above, the difference in CO₂ permeance becomes significant in the membranes with and without CNT loading. As mimic enzyme requires a water l igand to function, the separation performance qualitatively reflects the water uptake in the mimic enzyme membranes. Based on the separation performance, it is reasonable to assume that the rate of water swelling in both membranes was similar in the range of low humidity (<80%), but at higher humidity level, more water was retained in the membrane containing CNTs, most l ikely due to the nano- spacer effect resulted from the nano-sizc structure of the CNTs. The Free volume increase mechanism and Solubility increase mechanism in

nanocomposite membranes can explain the nano-spacer effect.

A CO₂ permeance of 0.98 m (STP)/ (m₂ bar h) is obtained at the ma imum relative humidity in the PVA-mimic enzyme membrane with 1 .0 wt% CNT loading, whereas that of the counterpart membrane without CNT loading is only 0.65 m (STP)/ (m₂ bar h), which is more than 30% lower. Both membranes have similar CO_? N_? selectivity at the same relative humidity when relative humidity is lower than 80%, but the selectiv ity increases more in the PVA-mimic enzyme membrane with CNT loading, maybe due to the increased water uptake from nano-spacer effect and that more accessible water favors the function of water for the catalytic effect for CO_? facilitated transport. Effect of feed gas pressure on the CO₂/N₂ separation performance

The CO₂/N₂ separation performances of the PVA and the PVA-mimic enzyme membranes with and without CNTs were tested at various feed pressures ( 1 -4 bara) to study the effect of pressure on CO? permeance and CO₂/N₂ selectivity. The results are plotted with operating pressures as x -axis i Figure 15. Figure 1 5(a) shows the CO₂ permeances of the four membranes. It can be seen that the influence of the feed pressure on the CO₂ permeance is different in the membranes with and without mimic enzyme. Both PVA-mimic enzyme membranes exhibit a dramatic decline in CO₂ permeance with increasing feed pressures, while the PVA membranes without mimic enzyme show a relatively stable CC½ permeance. The CO₂/N₂ selectivity of these membranes have a similar trend, as shown in Figure 15(b), where the CO₂/N₂ selectivity of two PVA membranes with and without CNT loading are nearly the same at various pressures.

From Figure 15(a) it can be also seen that the effect of the addition of CNTs in PVA membranes with and without mimic enzyme is different. In the PVA membrane without mimic enzyme, the CO₂ permeance decreases slightly with an increase in feed pressure, which may be attributed to the compaction of the membrane under pressure. Polymeric membranes are usually subjected to compaction due to the pressure difference between feed and permeate gas streams. This effect becomes intensified in highly water swollen membranes. The addition of CNTs may reduce the compaction and thus maintain a higher water uptake in the membrane. The PVA membrane with CNT loading can therefore maintain a better separation performance even at higher feed gas pressures. As expected, the addition of CNTs increases the CO₂ permeance of the PVA membrane, which becomes more significant at higher pressures. Nevertheless, in these two membranes, the influence of neither CNT addition nor pressure change on the membrane separation performances is significant. This indicates that CO₂ separation in these two membranes is mainly based on the solution-diffusion mechanism, in which gas permeance is determined by the solubility and diff usivity of the gas in the membrane that theoretically do not change with the change of feed gas concentration or partial pressure.

However, both the CO_? permeance and CO_?/N_? selectivity in the two PVA membranes containing mimic enzyme clearly decrease with the increase in feed pressure: this trend is considered a characteristic feature of facilitated transport membranes resulted from the saturation of the CO_? carriers at high C0₂ partial pressures, which has been reported by many researchers working with facil itated transport membranes. This suggests that the transport mechanisms of the membranes with and without mimic enzyme arc different. In the mimic enzyme membranes it is bel ieved that the transport of C0₂ through the water swollen PVA film is facil itated by the mimic enzyme catalyzed C0₂ hydration reaction, as the C0₂ molecules absorbed the membrane are hyd ol sed in the presence of mimic enzyme and transport through water swollen domains as bicarbonate ions. At a higher feed pressure, the equilibrium concentration of C0₂ or HCO in the membrane is higher, which may have caused the deactivation of some catalytic sites in the mimic enzyme and a reduction in C0₂ separation performance, as mimic enzyme is a catalyst that is deactivated in a bicarbonate rich environment. The decreasing trend in C0₂ permeance with an increase in feed pressure indicates that the CNT enhanced PVA- mimic enzyme membrane is most suitable for the removal of C0₂ from gas streams with a low C0₂ partial pressure, as in the case of post-combustion C0₂ capture. The results show that the addition of 1.0 wt% of CNTs in both PVA and PVA- mimic enzyme membranes can improve the C0₂ separation performance. The improvement is more significant in PVA membranes containing mimic enzyme. An appro imately 30% increase occurs at 1 .2bara. The increase in C0₂ permeance at a higher pressure is lower (e.g.. an increase of 0.23 [m (STP)/ (m₂ bar h)] at 2 bara, and approximately 0. 14 [m (STP) (m₂ bar h )] at 3 bara); however, the ratio of the increase due to the addition of CNTs is quite similar, at around 30%. This suggests that in the mimic enzyme membrane the decrease in C0₂ permeance with the increasing pressure is not resulted from compaction, but mainly due to the reduced facilitated effect at higher feed pressures. The C0₂ permeance of a PVA-mimic enzyme membrane containing CNTs at 1.2bara was determined to be 0.98 [ m (STP)/(m²bar h)] with a C0₂/N₂ selectivity of around 120. By adding CNTs to a mimic enzyme membrane the separation performance of the PVA membrane increased appreciably.

The CNT-enhanced PVA-mimic enzyme membrane shows a C0₂/N₂ separation performance comparable to some high performance membranes reported in recent years, as presented in Table 3 from the comparison of the separation performance of this membrane with some state-of-the-art CO? facilitated transport membranes and nanocomposite membranes.

Table 3: Comparison with the literature data

Feed gas Selectivity Permeance Reference

Membrane m

Pressure

Composition a (STP)/m bar

(kPa)

hr

Deng et al. Int. J.

Greenh. Gas

PVAm/PVA-CNTs 10% C0₂ in CH₄ 200-1000 45 0.35-0.15

Control, 2014,

26(0), 127-134

Deng et al, J.

Membrane Sci.,

PVAm/PVA 10% CC½ in ₂ 200-1000 174-66 0.58-0.13

2009, 340(1-2), 154-

163.

Sandru et al., J.

Membrane Sci.

PVAm/PPO 10% CO, in N , 200-1000 120-60 0.998-0.10

2010, 346(1), 172-

186.

Yu et al., J.

DNMDAM- Membrane Sci.,

20% C0₂ in N₂ 200-1000 70-56 0.47-0.27

TMC/PS 2010, 362(1-2), 265-

278.

Yao et al., Chem

PVl-Zn complex 1 15% C0₂ in N₂ 200-1600 75-50 1.23-0.82 Commun., 2010,

48(12), 1766-1768.

Yao et al., Chem

PVI-Zn complex 2 15% C0₂ in N₂ 200-1600 65-40 0.95-0.41 Commun., 2010,

48(12), 1766-1768.

Saeed et al. J.

PVA/mimic

10% C0₂ in N₂ 100-300 107-79 0.69-0.49 Membrane Sci., enzyme

2015, 494, 196-204.

CNTs +PVA 10% C0₂ in N₂ 100-300 60-55 0.44-0.26 This work CIMTs+PV A/mimic

10% C0₂ in N₂ 100-300 120-70 0.98-0.49 This work enzyme

Claims

Claims

A composite membrane comprising a water swollen selective layer for separating C0₂ from a mixed gaseous feed stream, wherein the selective layer comprises:

(i) a polymer matri ; and

(ii) a zinc complex, wherein zinc is coordinated to a multidentate

nitrogen-containing l igand with a molecular weight of less than 2000 g/moi;

and a porous support on w hich said selectiv e layer is carried.

A membrane as claimed in claim 1 wherein the polymer matrix comprises

PVA.

A membrane as claimed in claim 1 or 2, further comprising at least one nanofil ler.

A membrane as claimed in claim 3 w herein the at least one nanofiller is present in an amount of up to 20 wt%, preferably 0. 1 to 1 0 wt%, more preferably 0.2 to 5 wt%, even more preferably 0.5 to 1 .5 wt%, e.g. 1 .0 wt%.

A membrane as claimed in claim 3 or 4 wherein the nanofiller is selected from the group consisting of carbon nanotubes, graphene oxide nanosheets, nanocellulose, zeol ite frameworks, metal-organic frameworks and particles of Ti0₂ or silica, or a mixture thereof

A membrane as claimed in any of claims 3 to 5 wherein the filler comprises, preferably consists of, carbon nanotubes, preferably hydrophilic carbon nanotubes

7. A membrane as claimed in any preceding claim w herein the polymer matrix has a Mw of 10,000 to 500,000.

8. A membrane as claimed in any preceding claim wherein the multidentate l igand has a Mvv of less than 1 000 g/mol.

9. A membrane as claimed in any preceding claim wherein the multidentate l igand has a Mw of less than 500 g/moi.

1 0. A membrane as claimed in any preceding claim w herein the multidentate ligand is an aza macrocyiic ligand.

1 1. A membrane as claimed in any preceding claim wherein the multidentate ligand is cyclen or eye I am.

12. A membrane as claimed in any preceding claim, w herein the zinc comple is not chem ically l inked to the polymer matrix.

13. A membrane as claimed in any preceding claim wherein the selective layer is less than 5 microns in thickness.

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

1 5. A membrane as claimed in any preceding claim wherein the porous support has an MWCO of 25,000 or more.

1 6. A membrane as claimed in any preceding claim wherein the selective layer comprises 25-85% w ater relative to the amount of polymer matrix.

1 7. A process for separating CO_? from a mi ed gaseous feed stream, said process comprising contacting said gaseous feed stream with a membrane as defined in any of claims 1 to 16.

18. Use of a membrane as defined in any of claims 1 to 16 in the separation of CO_? from a mixed gaseous feed stream.