The present invention relates to a thin layer composite membrane incorporating a thin layer comprised of a microporous material of high intrinsic microporosity for the selective permeation and/or separation of a particular component of a fluid or solid/fluid mixture, methods for fabricating such a membrane and applications for such a membrane.

Throughout the present application the term 'microporous material' is intended to encompass a material which may also be described as a 'nanoporous material'.

Applications for membrane separations are found in the fields of fluid (i.e. gas, vapour or liquid) separation and separation of low molecular weight solids from fluids (nanofiltration). Separation processes of industrial relevance include: (i) separation of H2 from hydrocarbons, N2 or CO in, for example, the refinery industry; (ii) separation of CO2, H2O and H2S from natural gas (dewpointing, upgrading); (iii) air separation either to obtain enriched O2 or N2 for various applications; (iv) separation of volatile organic compounds (VOC) or lower hydrocarbons from air and other gases; (v) pervaporative separation from trace organic compounds found in aqueous streams; (vi) separation of low molecular weight compounds and oligomers from fluids especially organic solvents (nanofiltration).

For industrial application, permeable membranes must be capable of providing an acceptable level of selectivity of one component in a mixture over other components in the mixture whilst operating at an acceptable permeation rate. Since the permeation rate of molecules through the membrane generally depends linearly on the thickness of the membrane, membranes are usually designed to be as thin as possible to generate the highest possible permeation rates for gas separation. Unfortunately, the drive towards higher permeation rates leads to membranes which are so thin as to lack the necessary mechanical strength to retain their structural integrity during use and function efficiently. Composite membranes have therefore been developed which incorporate at least one relatively thin selective membrane layer, typically a thin film, and a thicker porous supporting layer.

In aqueous based pervaporation processes, to remove minor organic components only the flux of the water depends linearly on the membrane thickness. The flux of the organic component is relatively independent of the membrane thickness. In this case, therefore, thin membranes show decreasing selectivity with decreasing membrane thickness and very thin membranes are not favoured.

While selectivity and permeability are the principal factors determining the industrial applicability of composite membranes, additional factors which must also be considered include membrane area, pressure difference across the membrane and compressor size.

It is generally agreed that polymeric selective membrane materials exhibiting the highest possible permeability and at least reasonably good selectivity for the desired separation are favoured. However, it is also recognized that in certain applications the reverse conditions may be more suitable, i.e. it may be more important to employ a material exhibiting the highest possible selectivity and only reasonably high permeability. It is therefore important to develop methods for tailoring the properties of selective membranes to suit a particular application.

It has been found that for polymeric selective membrane materials, as the intrinsic permeability of the material (i.e. the permeability of the material without a supporting layer) increases, its selectivity for a particular gas in a pair of gases diminishes. A relationship (referred to as the 'upper bound', see Figure 1) for many gas pairs has been identified1'2'3'4 and it was concluded that an enhancement in sorption selectivity beyond that identified for current materials3'4 would be required to move above the 'upper bound' line identified previously1'5.

Table 1 lists a selection of known polymers exhibiting unusually high oxygen permeability. Unfortunately, as shown in Table 1, these materials exhibit relatively low selectivities for oxygen in preference to nitrogen. The permeability / selectivity relationship for the polymers listed in Table 1 and some additionally highly permeable polyimides6 are included in the 'upper bound' plot shown in Figure 1 (see reference 6 for a description of Figure 1).

Polymer Class Polymer O2- Selectivity Reference Permeability / P(O2W2) / Barrer polyacetylene PTMSP 10000 1.4 2 perfluoroethylene Teflon AF 1100 2.1 7 copolymer 2400 silicone rubber PDMS 780 2.2 2 perfluoroethylene Hyflon AD 190 2.5 8 copolymer 8OX Polyimide TMPA-6FDA 120 3.4 US 5,591,250

Table 1. Known polymers exhibiting unusually high oxygen permeability.

The polymers listed in Table 1 are distinguished by a very high accessible free volume.9 This high, accessible free volume may be described as porosity on a molecular level. It is anticipated that other materials, in particular polymers, of high intrinsic microporosity should exhibit similarly high oxygen permeabilities.

A need therefore exists for new materials which would lie above the 'upper bound' and which are also suitable to be incorporated in to composite membranes to provide membrane systems exhibiting high selectivity and high permeability.

Rigid polymer networks having large surface areas have been disclosed in WO-A- 2002/002838 (The Victoria University of Manchester) which describes "organic-based" microporous materials comprised of a 3 -dimensional network of planar porphyrinic macrocycles covalently interconnected by linkers which impose a particular relative orientation on the macrocycle rings they interconnect. Each such linker may connect two or more of the macrocyclic rings together and in the overall network the substantial majority (but not necessarily all) of the macrocyclic rings is associated with at least three, and ideally four, linkers each of which in turn links that macrocycle to at least one adjacent macrocycle so as to build up the overall 3-dimensional network. Due to their flat, plate-like (or cross-like) shape, the non-coplanar orientation of the planes of adjacent macrocycles ensures a microporous structure. The rigid linkers maintain the non- coplanarity of the planes of adjacent macrocycles that would otherwise allow the coalescence of macrocycles and thus collapse the desired void space within the material. While these materials represent an important advance in this area of technology and should find application in a number of different fields it is by no means evident from WO-A-2002/002838 whether the materials disclosed therein would be suitable for use in composite membranes.

A development of WO-A-2002/002838 is described in UK patent application no. 0317557.7 (The Victoria University of Manchester) and is based on the unexpected realisation that porphyrinic macrocycles are not essential for the formation of rigid 3- dimensional networks exhibiting intrinsic microporosity as described in WO-A- 2002/002838, such materials can in fact be produced using any suitable generally planar species connected by rigid linkers having a point of contortion.

In addition to identifying selective membrane materials which possess high intrinsic microporosity and which may therefore exhibit the appropriate selectivity/permeability properties for a given application, the properties of the support will also markedly influence the performance of a composite membrane. For example, it has been stated that, for polymeric membrane materials, the coil diameter of the solved polymer chains should be greater than the pore dimensions of the support so as to not fill and plug the pores which would reduce the overall permeability of the membrane.13 Furthermore, the permeability of the support should be significantly higher than the selective membrane material so that the permeability of the support does not reduce the overall permeability of the composite membrane. A resistance model approach was defined14 and improved,15'16 from which it can be concluded that if the expected permeability of the selective membrane layer is below 10 % of the permeability of the support, no reasonable resistance of the support has to be taken into account and hence the selectivity of the selective membrane layer is not appreciably influenced. It is therefore desirable that the support should possess high porosity, a small mean pore size, a permeability which is significantly higher than that of the selective membrane layer and, of course, the support should be stable to the conditions used to fabricate the composite membrane and the conditions of the separation process, e.g. solvents, temperatures, feed pressures etc.

Methods for fabricating composite membranes have been developed which involve either coating the composite membrane or coating the support prior to application of the selective membrane material. For example, US 3,980,456, describes a method for sealing pinholes and covering imperfections of ultrathin multi-layer composite membranes by applying an ultrathin membrane of a very flexible, permeable polymer (e.g. organopolysiloxane-polycarbonate copolymer) over the outer surface of the composite membrane. This material should be less selective than the basic ultrathin membrane and must exhibit at least about 3.5 times the permeability of the basic membrane.

US 4,230,463 describes coating an integrally asymmetric membrane with a thin layer of highly permeable silicone rubber and found a selectivity well above the intrinsic selectivity of silicone rubber. US 5,391,219 discloses a variety of elastomers, e.g. poly(4- methylpentene-1) or butadiene-styrene copolymer, to coat an integrally asymmetric polyimide membrane and found the same effect of increased separation factor with respect to the elastomer.

US 6,540,813 describes the application of a highly permeable perfluoropolymer, Teflon AF, in a perfluorinated solvent to a porous hollow fibre support to form a permselective perfluoropolymer layer on the surface of the porous support. Prior to coating, the support is impregnated with a fluid, e.g. water, which is immiscible with the perfluorinated solvent so as to reduce the tendency for the coating solution to enter pores on the surface of the porous support which would otherwise increase the effective thickness of the permselective layer and reduce the performance of the final membrane. It was observed that the residence time in the oven of the impregnated support prior to coating influences the quality of the final composite membrane. It was concluded that the water content of the support results in different thickness of the composite and affects anchoring of the permselective layer into the support.

US 4,840,819 and US 4,806,189 also disclose processes for producing composite membranes where the pores of the support are filled by a liquid so as to improve the homogeneity of the selective membrane layer when applied to the surface of the porous support. US 5,320,754 describes the use of perfluoroethers to wet the surface of a porous support prior to coating with a selective membrane material. The presence of the perfluoroether in the pores of the support is believed to prevent the selective membrane material from penetrating appreciably in to the support and thereby ensures the formation of a thin homogenous selective membrane layer on the surface of the support.

There are therefore a large number of factors which must be considered when designing a composite membrane for a specific purpose. Not only must the intrinsic microporosity, selectivity and permeability of the selective membrane layer be appropriate for the particular application, but the intrinsic porosity and mean pore size of the supporting layer must also be suitable. Furthermore, for the final composite membrane to exhibit the desired performance the interrelationship between the properties of the selective membrane layer and the supporting layer must be considered and carefully controlled.

An object of the present invention is therefore to provide composite membranes exhibiting improved performance over current composite membranes.

According to a first aspect of the present invention there is provided a thin layer composite membrane incorporating a selective layer comprising a microporous material and a porous supporting layer, wherein said microporous material comprises organic macromolecules comprised of first generally planar species connected by rigid linkers having a point of contortion such that two adjacent first planar species connected by the linker are held in non-coplanar orientation. The first aspect of the present invention is based on the realisation that composite membranes can be provided incorporating microporous materials which exhibit high intrinsic microporosity by virtue of the presence of suitable generally planar species connected by rigid linkers having a point of contortion.

A preferred embodiment of this aspect of the present invention is based on the unexpected observation that porphyrinic macrocycles are not essential for the formation of rigid 3 -dimensional networks exhibiting intrinsic microporosity. Thus, the selective layer may comprise a microporous material in which said first generally planar species are other than porphyrinic macrocycles. A selection of such microporous materials which may be used is described in more detail later in the specification. In a particularly preferred embodiment, the inventive membranes may incorporate a selective layer comprising polydioxane A (1) which possesses a repeating unit of formula:

Polydioxane A (1)

Furthermore, since it has been recognised that a highly cross-linked polymer network is not essential in order to ensure microporosity, a membrane according to the first aspect of the present invention may also be provided incorporating a selective layer comprised of a microporous material comprising organic macromolecules comprised of first generally planar species connected by rigid linkers predominantly to a maximum of two other said first species, said rigid linkers having a point of contortion such that two adjacent first planar species connected by the linker are held in non-coplanar orientation. In a further preferred embodiment, a composite membrane in accordance with the first aspect of the invention may be produced incorporating a selective layer comprising a microporous material of the kind disclosed in WO-A-2002/002838 and described in more detail later in the specification. In this instance, the first generally planar species are planar porphyrinic macrocycles and the linkers connect pyrrole residues of adjacent macrocycles so that the linkers restrain the adjacent macrocycles such that their porphyrinic planes are in a non-co-planar orientation. In this way, the microporous material is comprised of a rigid 3 -dimensional 'network' of planar porphyrinic macrocycles which exhibits high intrinsic surface area and may be represented by the following formula (II):

In the above formula, L represents a linker fused to, and connecting, pyrrole residues of adjacent porphyrinic macrocycles, M represents a metal ion or 2H+ (for a metal free macrocycle) and R is carbon or nitrogen.

Composite membranes in accordance with the first aspect of the present invention may be fabricated to incorporate a selective layer as defined in the first aspect of the invention, of any desirable thickness. It is preferred that the selective layer has a thickness which is less than or equal to 10 μm and more preferably, less than or equal to 1 μm. It is envisaged that certain applications may require even thinner selective layers, consequently, the selective layer may have a thickness of approximately 0.5 μm, or possibly in the range 0.1 μm - 0.4 μm. Yet more preferably the selective layer may have a thickness of less than approximately 0.1 μm, most preferably approximately 0.05 μm. Since the properties of the supporting layer may influence the performance of the composite membrane, it is preferred that the selective layer has a permeability which is less than the permeability of the supporting layer, i.e. the permeability of the supporting layer should be higher than the selective layer so that the permeability of the supporting layer does not reduce the overall permeability of the composite membrane. Preferably the selective layer has a permeability which is less than 50 % of the permeability of the supporting layer and most preferably the selective layer has a permeability which is less than 10 % of the permeability of the supporting layer. In preferred embodiments of this aspect of the invention, the supporting layer exhibits an intrinsic nitrogen permeability in the range 100 - 200 m3/m2 h bar, more preferably approximately 150 m3/m2 h bar.

Furthermore, it is desirable that the support should possess high porosity and a small mean pore size, the supporting layer preferably has a mean pore size of less than 25 ran, more preferably a mean pore size in the range 10 - 20 nm, and yet more preferably a mean pore size of approximately 15 nm. Additionally, it is desirable that the pore size of the supporting layer is highly uniform such that a relatively homogenous supporting layer surface is provided upon which the selective layer is to be applied.

In order to ensure that the supporting layer is structurally and chemically compatible with the selective layer and the separation process conditions the supporting layer preferably comprises an inorganic material, e.g. a ceramic material, or an organic polymeric material. The supporting layer may comprise an organic polymeric material selected from the group consisting of a polyimide, a polyamideimide, a polyethersulfone, a polyacrylate, a polyphenylenesulfide and polyacrylonitrile.

A preferred embodiment of the invention provides a composite membrane in accordance with the first aspect of the present invention in which a 46 μm thick film of the microporous material exhibits an intrinsic oxygen permeability of approximately 380 Barrer at 30 0C at an oxygen feed pressure in the range 200 - 300 mbar. A further preferred embodiment of the invention provides a composite membrane according to the first aspect of the invention in which a 46 μm thick film of the microporous material exhibits an intrinsic oxygen/nitrogen selectivity of approximately 4 at 30 0C at an oxygen feed pressure in the range 200 - 300 mbar.

Yet further preferred embodiments of the invention afford a membrane according to the first aspect of the invention in which the membrane has an oxygen permeability of up to 5 m3/m2 h bar at 25 0C and/or an intrinsic oxygen/nitrogen selectivity of up to 5 at 25 0C.

It is envisaged that composite membranes in accordance with the present invention may be formed into any desirable membrane configuration, such as flat sheets or hollow fibres.

While WO-A-2002/002838 describes a family of porphyrin-based microporous materials the unexpected determination that porphyrinic macrocycles are not a pre-requisite for such materials to possess intrinsic microporosity, the selective layer in the membrane according to the first aspect of the present invention may comprise a microporous material in which said first generally planar species are other than porphyrinic macrocycles.

Furthermore, a membrane according to the first aspect of the present invention may also be provided incorporating a selective layer comprised of a microporous material comprising organic macromolecules comprised of first generally planar species connected by rigid linkers predominantly to a maximum of two other said first species, said rigid linkers having a point of contortion such that two adjacent first planar species connected by the linker are held in non-coplanar orientation.

By predominantly we mean that at least 50 % by mole of the first planar species in the organic macromolecule are connected (by the rigid linkers) to a maximum of two other such first planar species. Preferably the macromolecules are such that at least 70 %, more preferably at least 80 %, and most preferably at least 90 % by mole of the first planar species are connected (by the rigid linkers) to a maximum of two other planar species. Particularly preferred microporous materials comprise a plurality of contorted polymer chains in which adjacent chains are prevented from packing together efficiently by virtue of their rigid contorted structure which results in such materials possessing intrinsic microporosity. Thus, preferred microporous materials possess intrinsic microporosity extending in three dimensions but may be considered as 'non-network' polymer materials since they do not have a cross-linked covalently bonded 3 -dimensional structure such as that possessed by the porphyrin-based materials disclosed in WO-A-2002/002838.

It is desirable to be able to provide a membrane incorporating a selective layer of relatively high intrinsic surface area (i.e. the surface area of the unsupported material) but which may also be tailored to suit a particular application. Thus it is preferred that the intrinsic surface area of the microporous material, as measured by nitrogen adsorption or a related technique, may be at least 300 m2 g"1 or at least 500 m2 g"1. Furthermore, the surface area may be in the range 600 to 900 m2 g"1 or 700 to 1500 m2 g"1.

Further properties which it is desirable to control in order to provide useful composite membranes are the pore diameter and number average mass of the microporous material, for example, as compared to polystyrene standards and measured by gel permeation chromatography. The microporous material comprised in the selective layer of the inventive membrane preferably has an average pore diameter of less than 100 nm and, more preferably, a pore diameter in the range 0.3 to 20 nm. Moreover, the inventive membrane preferably incorporates a microporous material having a number average mass (compared to polystyrene standards) in the range 1 x 103 to 1000 x 103 amu, more preferably in the range 15 x 103 to 500 x 103 amu. Additionally, the inventive membrane may incorporate a selective layer comprising a microporous material which has a number average mass in the range 2O x IO3 to 200 x 103 amu, or 10 x 103 to 100 x 103 amu.

The organic nature of the microporous materials enables the pore structure of membranes constructed from such materials to be functionalised to a high degree of specificity for a particular species over similar species. For example, chiral inner surfaces can be produced which may be useful in separation processes involving chiral molecules, such as amino acids.

In certain applications it will be of use to construct membranes which comprise additional entities designed to perform a specific task. Thus, membranes in accordance with the first aspect of the invention may include an additional entity selected from a catalyst species, an organometallic species, an inorganic species, at least one type of metal ion; and at least one type of metal particle. The additional entity may for example be a metal-containing organic catalyst such as a phthalocyanine or porphyrin. A preferred example of an inorganic species is a zeolite.

It is envisaged that a range of different methods may be used to produce thin layer composite membranes incorporating a selective layer comprising a microporous material of the kind hereinbefore described in relation to the first aspect of the invention. Thus, a second aspect of the present invention provides a method for producing a thin layer composite membrane in accordance with the first aspect of the present invention incorporating a selective layer comprising a microporous material and a porous supporting layer, the method comprising the steps of:

i) dissolving the microporous material in a solvent to form a solution of the microporous material; ii) contacting the supporting layer with said solution; and iii) evaporating the solvent to provide said selective layer comprising the microporous material on the supporting layer.

The intrinsic microporosity of the microporous materials incorporated in to membranes in accordance with the present invention is distinct from the microporosity induced within conventional glassy polymers in that it is not eliminated by aging (i.e. physical relaxation), annealing at elevated temperatures (e.g. between 100 0C and 300 0C), or the slow removal of solvent during film or membrane fabrication. Accordingly, the performance of composite membranes in accordance with the invention should be more reliable, predictable and controllable than known membranes. The high intrinsic microporosity of the microporous materials incorporated into composite membranes in accordance with the invention also provides the opportunity to produce membranes exhibiting significantly improved performance (gas permeability, selectivity, etc) than conventional membranes.

While any convenient method may be used to produce the inventive membranes, it is preferred that the supporting layer is contacted with said solution of the microporous material by dip-coating. This process may be carried out at any convenient temperature, although it is preferred that the dip-coating process is carried out at room temperature and final drying of the thin film above room temperature.

Most of the microporous materials of high intrinsic microporosity to be incorporated into membranes in accordance with the present invention are soluble in solvents of medium polarity. Thus it is preferred that the solvent is of medium polarity. The solvent is preferably an organic solvent, which is preferably tetrahydrofuran or a halogenated hydrocarbon. In a preferred embodiment of this aspect of the invention the halogenated hydrocarbon is chloroform.

Modifying the selective layer coating solution by addition of alternative or additional solvents with different solvent strengths to a preferred solvent for a particular selective layer material (e.g. THF may be considered a preferred solvent for PIM-I coating solutions) during or after drying is a known method for controlling the quality of the final selective layer. Solvents may be characterised by their 'polarity'.19 A more useful concept has been developed which introduced three-dimensional solubility parameters.20 The Hansen parameters characterising solubility can be used to derive fractional parameters, from which a triangular graph can be plotted to facilitate convenient selection of suitable alternative or additional solvents.

Adding nonsolvents to a casting solution for preparation of asymmetric membranes is a well known technique21 to improve the quality of the final selective layer. In order to investigate the effect of nonsolvents in the formation of composite membranes in accordance with the present invention, fractional solubility parameters were plotted in a triangular graph to facilitate selection of appropriate additives (i.e. nonsolvents or swelling agents) close in solubility properties to the solvents THF and CHCl3. The additives chosen for investigation were the hydrocarbons cyclohexanone, ethyl acetate and dioxane, although many other suitable additives could be used, such as aromatic or aliphatic hydrocarbons, alcohols, ethers, ketones or esters. Accordingly, the solvent in which the microporous material is dissolved in step i) of the second aspect of the invention may contain an organic additive. It is preferred that the additive possesses a boiling point below 120 °C.

The additive may be selected from the group consisting of an aromatic hydrocarbon, an aliphatic hydrocarbon, an alcohol, an ether, a ketone, an ester and halogenated derivatives thereof. Alternatively, the additive may be selected from the group consisting of allyl acetate, butyl acetate, propyl acetate, ethyl acetate. As a further alternative, the additive may be selected from the group consisting of 1,4-dioxane and 1,3-dioxolane.

Further preferred additives may be selected from the group consisting of diethyl ether, methyl ethyl ether, dibutyl ether and methyl-t-butyl ether, the group consisting of acetone, diethyl ketone dipropyl ketone, methyl butyl ketone, di-isobutyl ketone, methyl ethyl ketone, ethyl butyl ketone and cyclohexanone, or the group consisting of methanol, ethanol, n-propanol, iso-propanol and t-butyl alcohol.

Yet further preferred additives are 1,1-dimethoxy ethane or additives selected from the group consisting of 1,1,1-rrichloroethane and 1,1,2-trichloroethane.

An appropriate concentration of the additive in the solvent may be chosen to suit a particular solvent/additive system and will also depend upon the nature of the selective layer, supporting layer and the final composite membrane to be produced. The concentration of the additive in the solvent therefore may be less than 50 %, less than 20 %, or in the range 1 % to 15 %. The concentration of the additive in the solvent may be relatively low, e.g. approximately 5 %.

Cross-linking of the microporous material is likely to modify the properties of the selective layer and the final composite membrane. Thus, for certain applications it is preferred that the microporous material is at least partly cross-linked by treatment with a cross-linking agent before, during or after contacting the supporting layer with the solution of the microporous material. The cross-linking agent may be palladium dichϊoride.

Such cross-linking may render the membrane insoluble in organic solvents, which may be desirable in certain applications. While many microporous materials suitable for incorporation in to membranes in accordance with the present invention are soluble in common organic solvents, some are not. Such insoluble materials may be formed into membranes using a range of conventional techniques such as powder pressing. Alternatively, insoluble microporous materials may be mixed with a soluble microporous material so that the mixture can then be processed using any of the known techniques employed for fabricating membranes from soluble materials e.g. solvent casting or dip- coating.

Any desirable concentration of the microporous material in the solution of the microporous material may be used to suit a particular application, however, it is preferred that the microporous material is a relatively minor component of the solution. Accordingly, it is preferred that the concentration of the microporous material in the solution of the microporous material is less than 5 %, more preferably in the range 0.1 % to 2 %, or 0.5 % to 1.0 %. Most preferably the concentration of the microporous material in the solution of the microporous material is approximately 0.75 %.

Previous work has indicated that properties such as gas permeance and selectivity of composite membranes may be modified by the application of a treatment solution to the supporting layer prior to formation of the selective membrane layer. Thus, the supporting layer may be treated with a first treatment solution before the supporting layer is contacted with the solution of the microporous material. It is preferred that the first treatment solution is substantially miscible with the solution of the microporous material.

While any number of different appropriate first treatment solutions may be utilised, the first treatment solution is preferably water or a hydrocarbon based liquid. The first treatment solution may be an alcohol, such as isopropanol. Alternatively, the first treatment solution may be tetrahydrofuran. In certain applications it may be advantageous to ensure that the additive is at least partly removed from the support before the supporting layer is contacted with the solution of the microporous material.

Optimisation of properties of the final composite membrane may be obtained by employing a method for producing the membrane in accordance with the second aspect of the invention in which the first treatment solution comprises a gas permeable polymer, e.g polydimethylsiloxane. The concentration of polydimethylsiloxane in the first treatment solution is less than 1 %, more preferably less than 0.5 %. As a further preferred alternative, the first treatment solution may be a halogenated hydrocarbon based liquid, such as a perfluoroether.

Properties of the final thin layer composite membrane may be modified by the application of a layer of a gas permeable polymer, such as polydimethylsiloxane, to the supporting layer before the supporting layer is contacted with the solution of the microporous material.

To provide yet further control over the properties of a composite membrane according to the present invention, the composite membrane once formed may be treated with a second treatment solution, which may comprise a gas permeable polymer, such as polydimethylsiloxane or a fluorinated hydrocarbon. The concentration of the gas permeable polymer in the second treatment solution is preferably low, i.e. less than 1 %, more preferably less than 0.5 %. The second treatment solution may comprise a hydrocarbon based solvent, e.g. isooctane or a perfluoroether. Separation processes for which composite membranes in accordance with the present invention may be particularly suitable include: separation of H2 from hydrocarbons, N2 or CO; separation of CO2, H2O and/or H2S from natural gas (dewpointing, upgrading); enrichment of O2 or N2 in air; separation of volatile organic compounds (VOC) or lower hydrocarbons from air and other gases; pervaporative separation from trace organic compounds found in aqueous streams; and separation of low molecular weight compounds and oligomers from fluids especially organic solvents (nanofiltration).

According to a third aspect of the present invention there is provided a method for modifying the composition of a feed mixture comprising first and second species, the method comprising the steps of: i) applying the feed mixture to a feed side of a membrane in accordance with the first aspect of the present invention; and ii) collecting a retentate possessing a different composition to the composition of the feed mixture from the feed side of the membrane and/or collecting a permeate possessing a different composition to the composition of the feed mixture from an opposite side of the membrane.

The third aspect of the present invention provides a method employing an inventive thin layer composite membrane which is eminently suitable for use in nanofiltration processes in which it is desired to separate low molecular weight compounds and oligomers from fluids, in particular, organic solvents. Thus, the first species may be a low molecular weight compound and the second species may be an organic solvent. In this case, step ii) would preferably comprise collecting a retentate possessing a higher concentration of the low molecular weight compound compared to the composition of the feed mixture from the feed side of the membrane. When an inventive membrane is employed in a nanofiltration process in accordance with the third aspect of the present invention it is preferred that the percentage of the first species in the retentate compared to the percentage of the first species in the feed mixture is as high as possible, i.e. above 50 %, more preferably above 75 % and yet more preferably above 95 %. It is most preferred that the percentage of the first species in the retentate compared to the percentage of the first species in the feed mixture is approximately 99 %.

According to a fourth aspect of the present invention there is provided a method for separating a first species from a mixture of said first species and a second species, the method comprising the steps of: i) applying the mixture to a feed side of a membrane in accordance with the first aspect of the invention; ii) causing the first species to pass through the membrane; and iii) collecting the first species from an opposite side of the membrane and/or the second species from the feed side of the membrane.

The term 'separating' in the method forming this aspect of the invention is intended to encompass both 'separation' and 'removal' of a first species from a mixture of said first species and a second species.

The inventive membranes may also be used to enrich a particular component in a mixture of that component with at least one further component. Accordingly, an fifth aspect of the present invention provides a method for enriching a first species in a first mixture of said first species and a second species, the method comprising the steps of: i) applying the first mixture to a feed side of a membrane in accordance with the first aspect of the invention; ii) causing the first mixture to pass through the membrane; and iii) collecting a second mixture of the first and second species, which is enriched in respect of the first species compared to the first mixture, from an opposite side of the membrane and/or collecting a third mixture of the first and second species, which is enriched in respect of the second species compared to the first mixture, from the feed side of the membrane.

The mixture which is applied to the feed side of the membrane in step i) of the third, fourth or fifth aspects of the present invention may be in the gas or vapour phase, or alternatively, in the liquid phase.

It will be clear to the skilled person that the method in accordance with the third aspect of the invention is eminently suitable for modifying the composition of a mixture comprising a first species and at least two further species. Moreover, the fourth aspect of the invention is eminently suitable for the separation (or removal) of a first species from a mixture of said first species and at least two further species. Similarly, the method forming the fifth aspect of the invention is eminently suitable for the enrichment of a first species in a mixture of said first species and at least two further species.

Regarding the third, fourth and fifth aspects of the invention, preferably at least one of the first and second species is an organic compound. In one preferred embodiment the first species is an organic compound and the second species is water. The organic compound may be an alcohol or a halogenated hydrocarbon compound. In a further preferred embodiment the first and second species are organic compounds and the first species is an isomer of the second species. In this case, at least one of said organic compounds may be an alcohol or a halogenated hydrocarbon compound. Additionally, at least one of the first and second species may be a metal-containing compound. Furthermore, the first species may be oxygen and the second species may be nitrogen.

Membrane separation is based primarily on the relative rates of mass transfer of different species across a membrane. A driving force, typically a pressure or a concentration difference, is applied across the membrane so that selected species preferentially pass across the membrane. The inventive membranes may be used for purification, separation or adsorption of a particular species in the liquid or gas phase. The inventive membranes may, for example, be used for the separation of proteins or other thermally unstable compounds, e.g. in the pharmaceutical and biotechnology industries. The membranes may also be used in fermentors and bioreactors to transport gases into the reaction vessel and transfer cell culture medium out of the vessel. Additionally, the membranes may be used for the removal of microorganisms from air or water streams (nanofiltration), water purification, ethanol production in a continuous fermentation/membrane pervaporation system, and detection or removal of trace compounds or metal salts in air or water streams. The inventive membranes may be used in gas/vapour separation processes in chemical, petrochemical, pharmaceutical and allied industries for removing organic vapours from gas streams, e.g. in off-gas treatment for recovery of volatile organic compounds to meet clean air regulations, or within process streams in production plants so that valuable compounds (e.g., vinylchloride monomer, propylene) may be recovered. Further examples of gas/vapour separation processes in which the inventive materials may be used are hydrocarbon vapour separation from H2 in oil and gas refineries, for hydrocarbon dewpointing of natural gas (i.e. to decrease the hydrocarbon dewpoint to below the lowest possible export pipeline temperature so that liquid hydrocarbons do not separate in the pipeline), for control of methane number in fuel gas for gas engines and gas turbines, and for gasoline recovery. It is possible for the membranes of the invention to incorporate a species that adsorb strongly to certain gases (e.g. cobalt porphyrins or phthalocyanines for O2 or silver(I) for ethane) to facilitate their transport across the membrane.

The inventive membranes may also be used in the separation of liquid mixtures by pervaporation, such as in the removal of organic compounds (e.g., alcohols, phenols, chlorinated hydrocarbons, pyridines, ketones) from water such as aqueous effluents or process fluids. A membrane in accordance with invention which is ethanol-selective would be useful for increasing the ethanol concentration in relatively dilute ethanol solutions (5 - 10 % ethanol) obtained by fermentation processes. Further liquid phase examples include the separation of one organic component from another organic component, e.g. to separate isomers of organic compounds. Mixtures of organic compounds which may be separated using an inventive membrane include: ethylacetate- ethanol, diethylether-ethanol, acetic acid-ethanol, benzene-ethanol, chloroform-ethanol, chloroform-methanol, acetone-ispropylether, allylalcohol-allylether, allylacohol- cyclohexane, butanol-butylacetate, butanol-1-butylether, ethanol-ethylbutylether, propylacetate-propanol, isopropylether-isopropanol, methanol-ethanol-isopropanol, and ethylacetate-ethanol-acetic acid, as suggested by Tusel and Bruschke.17 The inventive composite membranes are eminently suitable for gas separation. Examples of such separation include separation of an organic gas from a 'permanent' gas (i.e. a small inorganic gas such as nitrogen or oxygen), and the separation of organic gases from each other.

The inventive membranes may be used for separation of organics from water (e.g. ethanol and/or phenol from water by pervaporation) and removal of metal and organic compounds, low molecular weight compounds and/or oligomers from liquids such as water or organic solvents (nanofiltration).

An additional application for the inventive membranes is in chemical reactors to enhance the yield of equilibrium-limited reactions by selective removal of a specific product in an analogous fashion to the use of hydrophilic membranes to enhance esterification yield by the removal of water.18

As hereinbefore described, microporous materials to be employed in membranes in accordance with the present invention comprise first generally planar species connected by rigid linkers having a point of contortion such that that two adjacent first planar species connected by the linker are held in non-coplanar orientation. Preferably the point of contortion is a spiro group, a bridged ring moiety, or a sterically congested single covalent bond around which there is restricted rotation. The point of contortion may be provided by a substituted or unsubstituted spiro-indane, bicyclo-octane, biphenyl or binaphthyl moiety. Each of the first planar species preferably comprises at least one aromatic ring. In a further preferred embodiment each of the first planar species comprises a substituted or unsubstituted moiety of the formula:

where X is O, S or NH.

Moreover, the inventive material may comprise repeating units of formula:

which may be substituted or unsubstituted.

Furthermore, the inventive material may comprise repeating units of formula:

which may be substituted or unsubstituted.

The microporous materials for incorporation in to membranes of the present invention can be prepared with the type of chemistry used for the preparation of high performance polymers from a large variety of suitable monomers without compromising the intrinsic microporosity. The robust chemical nature of the five and six-membered rings which dominate the structure of the microporous materials are similar to those chosen to construct high performance engineering materials and therefore promise improved chemical and physical stability as compared to PTMSP.

In a preferred embodiment, the inventive membranes incorporate a selective layer comprising a class of microporous polymers, exemplified by polydioxane A (1), which offer similar (perhaps greater) microporosity than PTMSP due to their rigid and contorted molecular structure.

Polydioxane A (1)

The microporosity of polydioxane A (which represents a preferred embodiment of the present invention) is demonstrated by its high surface area (approximately 680 - 850 m2 g"1) determined using nitrogen adsorption measurements (BET calculation). The presence of the cyano and methyl groups is optional, they may be omitted or replaced with other simple substituents. Each phenyl group may contain one or more substituents. Additionally, the nature and arrangement of substituents on the spiro-indane moiety may be chosen to provide any desirable configuration around the carbon atom common to both 5-membered rings.

Polydioxane A may be prepared in good yield from the aromatic nucleophilic substitution reaction between a bis(catechol) (2) and tetrafluoroterephthalonitrile (3) as shown below in reaction scheme A: (2) (3) (1) Correct formula (1) Reaction scheme A: synthesis of polydioxane A (1); i) K2CO3, DMF, 5O0C

The resulting polymer, polydioxane A (1), is freely soluble in THF and DMF, partially soluble in chloroform and insoluble in acetone, methanol and water. By gel permeation chromatography, it has a number average mass of 170 x 103 amu as compared to polystyrene standards. In solution, as a powder and as a solvent cast membrane it is highly fluorescent (yellow). Most importantly, it displays a surface area in the range 680 - 850 m2 g"1 (measured from 10 different samples). Simple molecular modelling shows that polydioxane A (1) is forced to adopt a contorted configuration due to the presence of the spiro-indane centres, each of which acts as a point of contortion'. In addition, the fused ring structure ensures that the randomly contorted structure of each polymer molecule is locked so that the molecules cannot pack efficiently resulting in microporosity.

Further polymers with similarly rigid and contorted structures to that of polydioxane A (1) can be produced and incorporated in to membranes in accordance with the present invention. In essence, any combination of monomers, which a) leads to a very rigid polymer and b) gives a polymer within which there are sufficient structural features to induce a contorted structure leads to microporosity.

Within the class of microporous polymers exemplified by polydioxane A (1) are five preferred sub-classes of microporous materials, each of which is characterised by the monomer units from which the final materials are produced. The first sub-class of materials includes polydioxane A (1). Materials falling in to this sub-class are produced from the reaction of a Nu2-R-Nu2 monomer and an X2-R^-X2 monomer. In each case, R and RΛ represent organic-based moieties linking the Nu2 or X2 groups. R and RΛ may or may not be the same moiety. Provided at least one of R and R" contains at least one point of contortion, the resulting polymer will possess intrinsic microporosity. R and/or R' may contain one or more point of contortion, hi essence Nu represents a nucleophile and X represents a good leaving group for nucleophilic substitution, particularly aromatic nucleophilic substitution in cases where X is bonded to an aromatic group. Thus the reaction of Nu2-R-Nu2 with X2-R^-X2 involves a nucleophilic substitution reaction in which each Nu group bonds to a carbon atom bearing an X group and displaces X as exemplified below in idealised reaction scheme B.

Reaction scheme B

Scheme B illustrates a preferred embodiment of the microporous materials to be incorporated in to membranes in accordance with the present invention in which the four nucleophilic groups (Nu) in each monomer compound are arranged in two pairs in which the members of each pair are bonded to adjacent carbon atoms. Moreover the four leaving groups (X) in each monomer compound are likewise arranged in two pairs in which the members of each pair are bonded to adjacent carbon atoms. Arranging the four Nu and X groups in this way results in a pair of covalent bonds being formed to give a six-membered ring of atoms between each pair of monomers containing two carbon atoms from each monomer. Consequently, rotation of one monomer unit, once incorporated into the polymer, relative to its neighbour is restricted which results in a polymer structure in which adjacent monomers units are essentially 'locked' together. Removing rotational freedom in this way provides greater control over the microporosity of materials formed using such an arrangement of Nu and X groups since the predetermined points of contortion built into the structure are the predominant factor ensuring the contorted shape of the polymer chains and thereby the microporosity of the resulting material.

Each Nu and X may or may not be the same type of functional group. Moreover, each B and X may be any particular functional group provided it satisfies the above requirements. In preferred embodiments of the present invention Nu is an OH, NH2 or SH group and X is a halogen, OPh, OTs or trifϊate group.

Alternative embodiments are envisaged in which one monomer comprises suitably arranged nucleophilic groups which can react with suitably arranged leaving groups on another monomer so as to form a desired polymer. It will be clear to the skilled person that a point of contortion may be present in the R-group of the nucleophile-containing monomer and not in the R'-group of the leaving group-containing monomer or vice versa. Additionally, a point of contortion may be present in both types of monomers.

It is preferred that the point of contortion is present in the monomer bearing the nucleophilic groups and that the generally planar species carry the leaving groups. Examples of Nu2-R-Nu2 compounds which include points of contortion are: (2) (4) (5) (6) (7) (8) (9) (10) (H) (12) (13) (14) (15)

Structures 2 and 9 may be used as the racemate or R or S enantiomers. In structure 5, R' and R' can be 2H (i.e. an etheno bridge) or a fused benzo unit (i.e. a triptycene).

A preferred point of contortion is provided by the spiro-indane moiety (substituted or unsubstituted). A further preferred point of contortion comprises moieties linked by a sterically congested single covalent bond linking adjacent hydrocarbon moieties. By virtue of the steric congestion there is significant restriction of relative movement of the linked moieties around the single covalent bond. Examples of X2-RZ-X2 monomers which represent the generally planar species are:

(19) (20) (21)

(27) (28) (29) Alternatively, the point of contortion may be present in the monomer containing the leaving groups and the generally planer species may be the nucleophile-containing monomer. Examples OfX2-RZ-X2 compounds which comprise points of contortion are: (31) (32) (33) Examples OfNu2-R-Nu2 compounds representing the generally planar species are:

(38) (39) (40) (41) (42) In certain applications it may be desirable to convert a first nucleophile-containing monomer to a monomer containing leaving groups prior to reaction of the (newly formed) monomer containing leaving groups with a second nucleophile-containing monomer. One way in which this can be achieved is shown in reaction scheme C.

(45)

Reaction scheme C: i) oxidation; ii) acetic acid, IQO0C.

It will be evident that the process shown in scheme C is generally applicable to any of the Nu2-R-Nu2 compounds which include points of contortion (2, 4-15). The second sub-class of materials are produced by reacting a (H2N)2-R-(NH2)2 monomer with a (keto)2-R'-(keto)2 or (keto)(hydroxy)-R'-(hydroxy)(keto) monomer. The point of contortion may be present in the (keto)2-R'-(keto)2 or (keto)(hydroxy)-R'- (hydroxy)(keto) monomer alone, in which case the (H2N)2-R-(NH2)2 monomer represents the planar species (exemplified by reaction scheme D).

(43) (46) (47)

Reaction scheme D

Alternatively, the point of contortion may be present in the (H2N)2-R-(NH2)2 monomer alone, in which case the (keto)2-R'-(keto)2 or (keto)(hydroxy)-R'-(hydroxy)(keto) monomer represents the planar species (exemplified by reaction scheme E).

(48) (49) (50)

Reaction scheme E Further examples of planar (keto)2-R'-(keto)2 monomers are presented below:

(51) (52)

Further examples of planar and contorted (H2N)2-R-(NH2)2 monomers are presented below:

(56) (57) (58)

(59) (60)

(61)

In addition to reaction schemes D and E above, a point of contortion may be present in both the (keto)2-R-(keto)2 / (keto)(hydroxy)-R-(hydroxy)(keto) monomer and the (H2N)2- R~(NH2)2 monomer. Compounds falling in to the third sub-class of materials are produced by the reaction of a nucleophilic monomer comprising an (H2N)2-R-(NH2)2 compound with a bis-anhydride or bis-dicarboxylic acid monomer. As before, the point of contortion may form part of the (H2N)2-R-(NH2)2 monomer, bis-anhydride or bis-dicarboxylic monomer, or both types of monomers. Reaction scheme F below exemplifies the situation where the point of contortion, in this case a spiro-indane moiety, is contained in an acidic bis-anhydride monomer and the planar species is represented by the basic compound 1,2,4,5- tetraaminobenzene (46):

(63)

Reaction scheme F

It will be apparent that 1,2,4,5-tetraaminobenzene (46) may be substituted in reaction scheme F with any of the (H2N)2-R-(NHa)2 monomers described above (48, 53-61). Examples of carboxylic anhydride or acid monomers which represent the planar species are shown below:

(64) (65) (66)

The overriding principle concerning compounds falling in to the second and third sub¬ classes of materials is that one of the monomers comprises a plurality of nucleophilic groups (e.g. NH2) and the other monomer comprises a plurality of electropositive carbon atoms bonded to electronegative atoms, such as oxygen atoms forming part of keto-, hydroxyl-, anhydrido- or carboxylic acid groups. The fourth sub-class of materials comprises compounds containing orthocarbonate groups. In this case, any one of the Nu2-R-Nu2 compounds (2, 4-15, 34-42) is first converted to the corresponding bis-orthocarbonate compound, as exemplified in reaction scheme G:

(2) (67) Reaction scheme G

The bis-orthocarbonate produced (67) is then halogenated thus:

(67) (68) Reaction scheme H

The halogenated bis-orthocarbonate (68) may then be reacted with one of the planar Nu2- R-Nu2 compounds (34-42) or one of the Nu2-R-Nu2 compounds containing a point of contortion (2, 4-15), provided that at least one of the halogenated bis-orthocarbonate or the Nu2-R-Nu2 compounds contains a point of contortion. For example, since the halogenated bis-orthocarbonate (68) produced in reaction scheme H contains a point of contortion (the spiro-indane moiety) it can be reacted with either a planar Nu2-R-Nu2 compound (34-42) or a Nu2-R-Nu2 compound containing a further point of contortion (2, 4-15), as shown below in reaction scheme I, to produce a microporous material possessing intrinsic microporosity. Whereas, if the halogenated bis-orthocarbonate did not contain a point of contortion it would have to be reacted with a Nu2-R-Nu2 compound which did contain a point of contortion, e.g. any of compounds 2, 4-15.

(2) (68) (69)

Reaction scheme I

Materials falling into sub-classes one and four are clearly related by the fact that each comprise polymers which have been produced by the substitution of halogen atoms by nucleophiles such as hydroxyl groups.

The fifth sub-class of materials are formed by the reaction of a Nu2-R-Nu2 monomer (2, 4-15, 34-42) with a compound containing a metal ion (such as Ti, Sn, Al, B, Ni, Cr, Co, Cd), or phosphorus or silicon (generally designated M in formula 70 below) as exemplified by the reaction of compound 2 below:

(1) (70)

Reaction scheme J

It will be clear to the skilled person that depending upon the nature of M, at least one additional M-coordinating ligand or counter ion may be required. Furthermore, depending upon the conformational constraints imposed on the structure of the repeating unit by the arrangement of ligands around the M-centre (for example, planar (Ni), tetrahedral (Si), pentahedral (P), e.g. trigonal bipyramidal or square based pyramidal, or octahedral (Cr)) the Nu2-R-Nu2 monomer may or may not contain a point of contortion as long as the repeating unit contains at least one point of contortion.

It is envisaged that polymers may be made which contain any number of types of nucleophilic monomer bonded to any number of types of monomer comprising suitable leaving groups. Furthermore, any number of types of sites of contortion may be combined with any number of types of rigid linker to produce a microporous organic material possessing the desired characteristics to suit a particular application.

In addition, it should be noted that many of the microporous materials can be modified using simple reactions of the functional group(s) that they contain, e.g. hydrolysis of nitrile substituents (exemplified by reaction scheme K) and quaternarisation of amine functionality (exemplified by reaction scheme L). Such reactions may also be used to cross-link the inventive materials to render them insoluble, which may be desirable for certain applications.

(1) (71) Reaction scheme K (73) Reaction scheme L The first aspect of the present invention provides thin layer composite membranes incorporating a selective layer of a microporous material which comprises organic macromolecules comprised of first generally planar species connected by rigid linkers having a point of contortion such that two adjacent first planar species connected by the linker are held in non-coplanar orientation.

As hereinbefore described, in a preferred embodiment of the present invention the first generally planar species are planar porphyrinic macrocycles and the linkers connect pyrrole residues of adjacent macrocycles so that the linkers restrain the adjacent macrocycles such that their porphyrinic planes are in a non-co-planar orientation. Such microporous materials may be represented by the following formula (II):

In the above formula, L represents a linker fused to, and connecting, pyrrole residues of adjacent porphyrinic macrocycles, M represents a metal ion or 2H+ (for a metal free macrocycle) and R is carbon or nitrogen.

Methods for producing thin films comprising porphyrin-based microporous materials (i.e. 'Network' polymers) are described in WO-A-2002/002838. It will be evident to the skilled person that such porphyrin-containing films may be incorporated into composite membranes in accordance with the present invention using similar methods to those described herein in relation to the production of composite membranes incorporating 'Non-network' polymer films, e.g. films comprising PIM-I (1) or PIM-N2 (74). In one preferred embodiment of the invention, a composite membrane is provided which incorporates a selective layer comprising a microporous material of the general formula:

in which M represents a metal ion or 2H+ (for a metal free macrocycle) and Li is a linker group.

It is preferred that the porphyrin-based microporous material has an intrinsic surface area of at least 300 m2g"1, more preferably at least 400 m2g"1, and most preferably of 700 -

The inventive membrane may incorporate a selective layer in which said porphyrin-based microporous material is a phthalocyanine network possessing a basic repeating phthalocyanine repeating unit of formula:

(VI) The linkers may be such that the porphyrinic macrocycles they interconnect are orthogonal to each other. However orthogonality is not essential and it is possible, for example, for the porphyrinic planes of macrocycles connected by a linker to lie at angles of 60° to 90° to each other. It is also possible for the adjacent macrocycles connected by a linker to lie in parallel planes. In preferred embodiments the porphyrinic plane of a macrocycle connected by the linker does not intersect any portion of another macrocycle connected by that linker.

The linkers which connect adjacent porphyrin macrocycles preferably comprise planar fused ring systems connected by an at least one "orientating moiety" which provides for orientation of these rings systems such that the porphyrinic plane of one macrocycle is not co-planar with that another macrocycle to which it is connected by the linker. The fused ring systems of the linker each preferably comprise at least three fused rings and the fused rings of the linker are preferably six-membered rings. The terminal fused ring systems may be of the formula:

where "a" represents the side fused to the pyrrole residue and "b" represents the bonding to the orientating moiety.

Many different chemical groups may be employed to act as the "orientating moiety". For example, the "orientating moiety" may be a substituted or unsubstituted spiro-indane moiety of formula:

where the sides "c" are those sides that are fused to planar fused ring systems of the linker. Alternatively, the "orientating moiety" may be a bridged ring entity to the sides of which are fused the terminal planar ring systems of the linker. In certain applications it would be advantageous if the bridged ring system is a bicyclo[2,2,0]octane ring. A particularly preferred embodiment utilises a linker of the formula:

where "a" represents the sides fused to pyrrole residues of adjacent porphyrinic macrocycles.

In a yet further alternative linker arrangement the orientation within the linker is provided by steric effects. For example, the linker may be of the formula:

wherein "e" represents sides of the linker fused to pyrrole residues of adjacent porphyrinic macrocycles. The present invention is illustrated with reference to the following non-limiting Examples and accompanying figures, in which:

Figure 1 is an 'upper bound' plot of permeability verses selectivity for a range of polymers of high permeability;

Figure 2 is a graphical representation of the temperature dependence of gas permeation for a 91 μm thick film of Polydioxane A (1);

Figure 3 is a schematic representation of a pervaporation rig employed in Example 2;

Figure 4 is a N2 adsorption/desorption isotherm for a 60 μm thick film of Polydioxane A (i);

Figure 5 is a graphical representation of the variation in phenol concentration in a permeate solution as a function of phenol concentration in a feed solution for a series of phenol/water mixtures subjected to a pervaporation process using a 60 μm thick film of Polydioxane A (1);

Figure 6 is a graphical representation of separation factor and flux data for a series of phenol/water mixtures subjected to a pervaporation process using a 60 μm thick film of Polydioxane A (1);

Figure 7 is a graphical representation of the variation in ethanol concentration in a permeate solution as a function of ethanol concentration in a feed solution for a series of ethanol/water mixtures subjected to a pervaporation process using a 60 μm thick film of Polydioxane A (1) and the zirconia supported membrane of Example 1; and

Figure 8 is a graphical representation of separation factor and flux data for a series of ethanol/water mixtures subjected to a pervaporation process using a 60 μm thick film of Polydioxane A (1) and the zirconia supported membrane of Example 1. A series of experiments have been conducted to investigate firstly the intrinsic properties of microporous materials to be used in the selective layer of composite membranes in accordance with the present invention. Thin layer composite membranes according to the invention were then tested for their permeance and selectivity to a range of gases. Treatment of the supporting layer prior to application of the selective layer, modification of the solvent system used during application of the selective layer and treatment of the composite membrane following application of the selective layer were investigated. Finally, preliminary experiments were carried out to assess the suitability of the inventive membrane to nanofiltration processes. Intrinsic Properties of Microporous Materials

Polydioxane A (1), also referred to herein as PIM-I3 was prepared as described below and

formed in to films of thickness 46 μm and 91 μm in accordance with the method set out

below. Gas permeation data were then measured for a range of gases, which are listed in

Tables 2 and 3.

GAS Diffusivity, Solubility, (cm3/cm3 Permeability, V(XIN2) (cnvVs x E"8) cmHg x E"3) Barrer O2 82 46 380 4.0 N2 23 42 94 1.0 He 2300 3.0 670 7.1 Ar 40 50 200 2.1 Xe 0.45 1200 55 0.59 H2 1700 7.8 1360 14 CO2 26 910 2370 25 CH4 6.9 180 126 1.3 Propene 0.44 3800 160 1.7 Propane 0.040 3100 12 0.13

Table 2. PIM-I (1) film thickness 46 μm.

GAS Diffusivity, Solubility, (cmVcm3 Permeability, P(XTN2) (cm7s x E-8) cmHg x E-3) Barrer O2 150 36 535 3.5 N2 45 35 155 1.0 He 5000 1.5 740 4.8 H2 3000 5.3 1600 10 CO2 55 670 3700 24 CH4 17 140 240 1.5

Table 3. PIM-I (1) film thickness 91 μm.

The measured gas permeation data were used to calculate permeability / selectivity values

for the two films which were then plotted on an 'upper bound' plot as shown in Figure 1

(PM-I = 46 μm film and PIM-1_2 = 91 μm film). The results indicated that for most

gases, including oxygen and nitrogen as shown in Figure 1, PIM-I (1) exhibited a permeability / selectivity relationship above the 'upper bound' which represents a significant advance over prior art materials.

The gases decreased in permeability according to: CO2 > H2 > He > O2 > Ar > CH4, Propene > Xe > Propane. For the gases CO2, H2, He, O2, N2, CH4, and Xe an increase of permeability with decreasing feed pressure was observed. For propene and propane a decrease of permeability with decreasing feed pressure was found. E.g., P(O2) at 320 mbar feed pressure was 367 Barrer and increased to 383 Barrer at 76 mbar feed pressure. Propene decreased from 202 Barrer at 690 mbar feed pressure to 126 Barrer at 190mbar. This behaviour is well known for polymers with extremely high free volume (Stern 1994). The higher permeability and slightly lower selectivity for the thicker sample is explained by different evaporation conditions thus forming a less tightly packed polymer. These results indicate that PIM-I (1) would be especially well suited for separation of O2 or N2 from air, of CO2 from N2, CH4 and/or Xe, of H2 from N2 and/or CH4.

The stability of the PIM-I (1) polymer films to temperature and to thermal aging was tested by varying the temperature for gas permeation experiments from 25 to 55 °C and back to 30 0C. Within the 8 days of measurement no change in permeability was detected thus demonstrating the thermal stability of the polymer. The results of these experiments for the 91 μm PDVI-I (1) polymer film are shown in Figure 2. The slope and intercept of the linear regression are reported in Table 4. For example, with increasing temperature the oxygen permeability increased from 511 Barrer (at 20 °C) to 600 Barrer (at 60 °C) accompanied by a decrease in selectivity (P(O2/N2)) from 4.0 to 2.8.

Gas Regression O2 y = 7.58 -0.394x N2 y = 9.22 -1.28x He y= 8.77 -0.647x H2 y = 8.66 -0.382x CO2 y = 6.43 + 0.543x CH4 y = 9.74 -1.3χ

Table 4. Temperature dependence of permeability of 91 μm PDVI-I (1) polymer film. For comparative purposes, an alternative microporous polymer, PIM-N2 (74) was

synthesised as described below and formed in to a thin film of thickness 28 μm in

accordance with the method set out below. Gas permeation data were then measured for 8

of the 10 gases for which tests were conducted using PIM-I. The results of the tests

carried out using PIM-N2 are shown in Table 5.

GAS Diffusivity, Solubility, (cnvYcm3 Permeability, P(X/N2) (cmVs x E-8) cmHg x E 3) Barrer O2 62 30 190 4.5 N2 16 26 42 1.0 He 760 6 450 10.8 Ar 30 34 100 2.4 Xe 0.37 710 26 0.62 H2 940 9.6 866 21 CO2 21 530 1100 26 CH4 5.1 120 62 1.5

Table 5. PM-N2 film thickness 28 μm.

In comparison to PBVI 1, PIM-N2 exhibited a higher selectivity but a lower permeability.

Even though PM-N2 exhibited a permeability for each gas which was approximately

half that of PIM-I, it should be noted that the permeabilities for PIM-N2 are still

acceptably high. Thus, these results indicate that PIM-N2 would also be suitable for the

separation of O2 or N2 from air, of CO2 from N2, CH4 and/or Xe, of H2 from N2 and/or

CH4. Theoretical Prediction of Properties of Composite Membranes Incorporating PIM-I And PIM-N2

The permeability of permanent gases (e.g. oxygen, nitrogen etc) across thick polymer films depends linearly on the thickness of the film. For extremely thin films close to monomolecular layers this dependence is no longer valid. During the transition from molecular layers to thicker films the bulk polymeric properties are approached at some point, which may vary with the surface properties of the support. Disregarding this complication, a gas permeability can be calculated for polymer films of thicknesses not to close to the monomolecular level. The common polymer silicone rubber (PDMS) was selected and gas permeances of films of varied thickness were calculated for comparison to values calculated for the 91 μm PIM-I film and 28 μm PIM-N2 film. The results of these calculations are shown in Tables 6 and 7.

PIM-I: PIM N2: PO2=500 Barrer, P(O2/N2)=3.5 PO2=ISN) Barrer, P(O2/N2)=4.5 Thickness O2 permeance Thickness O2 permeance / μm / m3/m2 h bar / μm / nrVni2 h bar 1.00 1.35 1.00 0.51 0.50 2.70 0.50 1.00 0.25 5.40 0.25 2.05 Table 6. Calculated oxygen permeance for PIM-I and PIM-N2 films.

PDMS: PO2=600 Barrer, (O2/N2)=2.1 Thickness O2 permeance / μm / m3/m2 h bar 1.00 1.62 0.50 3.24 0.25 6M Table 7. Calculated oxygen permeance for a PDMS film.

The results presented in Tables 6 and 7 illustrate that PIM-I should exhibit more than 80 % of the permeability of PDMS and exhibit a selectivity approximately 170 % higher than PDMS. PIM-N2 should exhibit an even higher selectivity (around 210 % of the PDMS value) but at a lower permeance of around 30 % of that of PDMS. Example 1 - Zirconia Supported Composite Membrane Incorporating PIM-I

A solution of PIM-I (1) was prepared in accordance with the method set out for the preparation of a 60 μm film and was then deposited on a porous zirconia ceramic support reinforced with metal mesh and left to dry under a stream of nitrogen.

Example 2 - PAN Supported Composite Membrane Incorporating PIM-I

A support made from polyacrylonitrile (PAN) with a N2-permeance of 150 m3/m2 h bar (±50) was selected, which should exceed the expected gas permeance of the selective layer by at least 10 times. The reasoning underlying this selection is as follows. Assuming a selective layer thickness of 0.25 μm, a maximum permeance of about 40 mVm2 h bar was calculated for composite membranes of PIM-I with the most permeable gas, CO2, at 100 % CO2. However, in a real separation mixture CO2 will be a minor component (i.e. below 50 %) and it is reasonable to calculate permeances of up to 15 m3/m2 h bar. If the separation of O2 from air is targeted, even lower permeances of below 5 ni3/m2 h bar should be considered. Therefore, it is reasonable to select a porous support with a gas permeance of about 150 m3/m2 h bar. Since the pore size of the support is also a predominant factor in determining the quality of the thin selective layer a low mean pore size of below 17 ran was selected (as detected by a capillary flow porometer).

A PIM-I / THF coating solution was prepared with low PIM-I concentrations in the range 2.00 - 0.60 %. A thin layer of each solution was applied by dip-coating to a PAN support using a coating machine to provide composite membranes having an area of 34 cm2. Example 3 - Separation By Pervaporation Of A Phenol / Water Mixture

The 60 μm thick film of PIM-I (1) was tested using a standard batch pervaporation rig (Figure 3). The circular flat membrane was clamped into a sealed glass test cell above a porous support with an elastomeric O-ring and a silicone rubber compound (RS), forming a leak free seal, giving a pervaporation area of 33.17 cm2. The cell was filled with the feed solution (mixture of phenol in water; 400 mL) and a mechanical stirrer placed in the cell. Stirring was necessary to reduce the effects of concentration polarization arising from the high selectivity of the membrane and the low concentration of phenol in the feed. Three samples were tested using feed solutions containing phenol in water at concentrations of 1, 3 and 5 wt.% phenol. The cell temperature was controlled and measured with a thermocouple and electronic temperature control system. The cell temperature was controlled at temperatures in the range 50-80 °C. A vacuum pump on the downstream side maintained a low pressure. The pressure was measured between the cell and the cold trap. The permeate was condensed and frozen within the cold trap, which was cooled with liquid nitrogen. The concentration of phenol in the permeate was measured by UV spectroscopy using the phenol absorption band at 270 nm.

The percentage of phenol in the permeate versus that in the feed for the three samples at 70 °C is shown in Figure 5, the dashed line would be followed if no separation occurred. The results presented in Figure 5 illustrate that the concentration of phenol in the permeate has increased compared to that in the feed. These results demonstrate that effective separation of phenol from the phenol/water mixture is taking place across the inventive membrane.

The separation factors for each sample are given in Figure 6 and the values of flux for phenol and water are shown in Figure 6. Separation factor is defined as

Separation factor = (YiIY2) I

where (IVr2) is the weight ratio of component 1 (i.e., the organic compound) to component 2 (i.e., water) in the permeate and (X\IXi) is the weight ratio of component 1 to component 2 in the feed. Example 4 - Separation By Pervaporation Of Ethanol / Water Mixtures

A 60 μm thick film of PIM-I (1) and a zirconia supported composite membrane were prepared as described above and tested using a standard batch pervaporation rig (Figure 3) with a stainless steel test cell. The circular flat membrane was clamped into the steel test cell above a porous support with an elastomeric O-ring and a silicone rubber compound (RS), forming a leak free seal, giving a pervaporation area of 24.6 cm2. The cell was filled with the feed solution (mixture of ethanol and water, 400 mL). The solution was stirred to reduce the effects of concentration polarization. Samples were tested using feed solutions containing ethanol in water at concentrations in the range 10- 70 wt.%. The cell temperature was controlled at 30 °C. A vacuum pump on the downstream side maintained a low pressure. The permeate was condensed and frozen within the cold trap, which was cooled with liquid nitrogen. The concentration of ethanol in the permeate was measured using a calibrated refractometer.

The percentage of ethanol in the permeate versus that in the feed is shown in Figure 7 for the 60 μm thick film of PIM-I (1) and for the supported membrane. The dashed line would be followed if no separation occurred. The results presented in Figure 7 illustrate that the concentration of ethanol in the permeate has increased compared to that in the feed, with the supported membrane giving slightly higher concentrations of ethanol in the permeate. These results demonstrate that effective separation of ethanol from the ethanol/water mixture is taking place across the inventive membrane.

The separation factors and total fluxes achieved for each membrane at various feed compositions are shown in Figure 8. Example 5 - Effect of PIM-I Concentration On Properties of PAN Supported Composite Membrane Incorporating PIM-I

A composite membrane incorporating a support made from polyacrylonitrile (PAN) and a selective membrane layer comprising PIM-I was prepared in accordance with Example 2.

Four solutions of PIM-I of varying concentration in tetrahydrofuran (THF) were applied to the PAN support. The permeances for oxygen and nitrogen were measured at room temperature with membrane samples of 34 cm2 area. The values for oxygen permeance for each PIM-I solution represent an average of 5-10 samples. The experimental results, together with the calculated effective thickness of each composite membrane are given in Table 8.

Entry Solvent Cone. PIM-I P(O2) P(O2/N2) CaIc. Thickness / % / m3/m2 h bar / μm 5.1 THF 2.00 0.28 3.22 3.9 5.2 THF 1.00 0.47 3.85 2.3 5.3 THF 0.75 4.70 2.82 0.2 5.4 THF 0.60 4.16 2.84 0.3

Table 8. Results for a PAN supported composite membrane incorporating PDVI-I .

The results presented in Table 8 illustrate that with decreasing polymer concentration from 2.00 % to 0.75 % the oxygen permeances increased at some loss of selectivity. The highest permeance (4.7 mVm2 h bar) was observed for the composite membrane formed using a 0.75 % PEVI-I solution. A thickness of 200 to 300 nm was calculated for the membranes with highest permeance. Example 6 - Gas Permeances of a PAN / PIM-I Composite Membrane

The permeance and selectivity of the inventive PAN / PIM-I composite membrane

produced in accordance with Example 2 (treated with a 0.75 % PIM-I solution in THF)

was tested for a range of gases.

Gas permeation was measured at 27 °C in the order N2, O2, CO2, CH4, N2, propene,

propane, N2, and O2. The feed pressure applied was from 400 to about 100 mbar. The

data are reported in Table 9.

Gas Permeance / nr(N)/m h bar Selectivity / P(x/N2) N2 0.98 1.00 O2 2.78 2.83 CO2 15.85 16.14 CH4 1.79 1.83 N2 0.98 1.00 propene 12.24 12.47 propane 2.58 2.62 N2 2.00 O2 4.68 2.34

Table 9. Gas permeation results for a PAN / PIM-I composite membrane.

After measurement of permeance of the hydrocarbons, a higher permeance was detected

for some time accompanied by lower selectivity. Propene and, to a less extent, propane

showed a clear dependence of permeability to the feed pressure (see Table 10).

Feed pressure P(propene) P(propane) / mbar / m3(N) / m2 h bar / m3(N) / m2 h bar 300 8.8 2.4 200 10.8 2.9 150 11.9 3.1 100 13.2

Table 10. Feed pressure dependence of permeability for propene and propane.

The results included in Table 9 indicate that the oxygen selectivity (P(O2/N2)) was 81 %

and the carbon dioxide selectivity (P(CO2/N>)) was 65 % of the intrinsic selectivity of the 91 μm thick PM-I film (see Table 3). Propene, however, permeates about 10 times faster compared to a thick polymer film which may be due to some interaction of the condensable gas with the very thin polymer film.. Furthermore, the behaviour in response to changes in feed pressure was reversed. For the 46 μm thick PIM-I film a decrease in permeability was observed with decreasing feed pressure (see discussion following Tables 2 and 3). However, for the PAN / PEVI-I composite membrane propene permeance increased from 8.8 mVm2 h bar (at 300 mbar) to 13.2 mVm2 h bar at 100 mbar feed pressure. Example 7 - Modification of Composite Membrane Properties by Pre-Coating Treatment of Support

In order to investigate how properties such as gas permeance and selectivity of composite membranes may be modified by the application of a treatment solution to the support prior to formation of the selective membrane layer and/or by modification of composition of the PIM-I coating solution the following three experiments (Examples 7.1 - 7.3) were carried out. The basic membrane used was a PAN / PIM-I membrane produced in accordance with Example 2 and as used in the preceding experiments employing a PAN / PEVl-I membrane.

Example 7.1 - Pretreatment Of Support With A Solvent Miscible With PIM-I Solvent

The PAN / PIM-I membrane exhibiting the highest permeability in Table 8 (Entry 5.3) was selected. Prior to application of the selective membrane the support contacted by isopropanol (i-C3) or THF and the surface dried. The results are reported in Table 11.

Entry Treatment Solvent Cone. PIM-I P(O2) P(O2/N2) CaIc. Thickness / % / m3/m2 h bar / μm 6.1 i-C3 THF 0.75 4.27 2.66 0.3 6.2 THF THF 0.75 ~1 6.3 THF THF 0.50 0.66 3.76 1.5

Table 11. Pretreatment of support with a solvent miscible with PIM-I solvent.

The results in Table 11 illustrate that no improvement in the properties of the selective layer compared to the untreated selective layer (Table 8, Entry 5.3) was obtained. Example 7.2 - Pretreatment Of Support With PDMS

The oxygen permeance and selectivity was determined, and in one instance (Entry 7.2) the effective separation layer thickness calculated, for four different PAN and/or PIM-I membranes. The results obtained are presented in Table 12.

A first membrane (Entry 7.1) consisted solely of the PAN support treated with a 0.25 % solution of cross-linkable PDMS in hexane. A second membrane (Entry 7.2) consisted of the PDMS-treated support from Entry 7.1 coated with a 0.5 % solution of PIM-I in THF. Third and fourth membranes were investigate for comparative purposes to determine what effect, if any, a thicker defect free PDMS layer would have on the performance of the PAN support and PAN / PIM-I composite membrane. The third membrane (Entry 7.3) therefore consisted of a PAN support with a dense defect-free PDMS layer of - lμm thickness, and the fourth membrane (Entry 7.4) consisted of the PDMS coated PAN support from Entry 7.3 coated with a PIM-I selective layer (0.75 % PIM-I solution in THF).

Entry Treatment Solvent Cone. PIM-I P(O2), P(O2/N2) CaLc. Thickness / % / m3/m2 h bar / μm 7.1 0.25% PDMS hexane no -100 7.2 0.25% PDMS THF 0.50 0.93 3.46 1.1 7.3 PDMS - lμm no 4.90 2.00 7.4 PDMS ~ lμm THF 0.75 1.03 3.02

Table 12. Pretreatment Of Support With PDMS.

The PDMS-treated support, prior to application of the selective membrane layer, exhibited a permeance to oxygen of around 100 m3/m2 h bar (Entry 7.1), which is around two-thirds of that of the untreated support (approximately 150 m3/m2 h bar). Coating of this surface by a 0.5 % solution of PIM-I in THF (Entry 7.2) yielded a composite membrane with a higher P(O2/N2) selectivity of around 3.5 (c.f. P(O2/N2) of untreated membrane = 2.82, Table 8, Entry 5.3) but a lower oxygen permeance of close to 1 mVm2 h bar (c.f. P(O2) of untreated membrane = 4.70, Table 8, Entry 5.3). The selectivity observed for the membrane incorporating the PDMS-treated support (P(O2/N2) ~ 3.5) is close to that of the membrane incorporating an untreated support to which has been applied a 2.00 % (P(O2ZN2) ~ 3.2) or a 1.00 % (P(O2ZN2) ~ 3.9) solution of PIM-I in THF. This teaches that precoating the porous structure of the supporting layer with PDMS enables more diluted PIM-I solutions to be used to provide highly selective membrane layers which exhibit acceptably high oxygen permeances. An effective thickness of the selective membrane layer of about 1 μm was calculated for the membrane incorporating the PDMS-treated support (Entry 7.2).

For comparison, a support with a dense defect-free PDMS layer of ~ lμm thickness (Entry 7.3) was investigated. The supporting layer with a ~ lμm PDMS layer prior to coating with a PBVI-I solution exhibited an oxygen permeance of 4.90 m3Zm2 h bar and a selectivity of 2.00. Following application of a PEVI-I selective layer (0.75 % PEVI-I solution in THF, Entry 7.4) the permeance of the composite membrane (1.03 m3Zm2 h bar) was similar to that of the composite membrane incorporating a support treated with a 0.25 % PDMS solution (0.93 mVm2 h bar) but the selectivity of the composite membrane incorporating the thick PDMS layer was lower (3.02) than that of the membrane incorporating the support treated with the dilute PDMS solution (3.46). The thin film of the PEVI-I polymer showed an unusual, marked adhesive strength on the PDMS layer.

It is clear from the above experiments involving pre-treatment with PDMS that a thinner PDMS layer (such as that provided by pre-treatment with a 0.25 % PDMS solution) can provide highly selective membrane layers at acceptably high gas permeances and that thicker defect-free PDMS layers are not necessarily required. Example 7.3 - Modulation of Solvent and Pretreatment of Support With a Perfluoroether

It is recognised that the quality of the final selective layer may be controlled by modifying the selective layer coating solution by addition of alternative or additional solvents with different solvent strengths to a preferred solvent for a particular selective layer material during or after drying. Suitable alternative or additional solvents were selected by use of a triangular graph plotted from fractional parameters derived from the Hansen parameters characterising solubility. The hydrocarbons cyclohexanone, ethyl acetate and dioxane were chosen in order to investigate their effect on the formation of composite membranes in accordance with the present invention.

The oxygen permeance and selectivity was determined for six different PAN and/or PIM- 1 membranes. In all possible cases (Entries 8.2 - 8.6) the effective separation layer thickness calculated. The results obtained are presented in Table 13.

Penetration of the selective layer solution into the pores of the support is not possible with water because of the fast and unlimited miscibility of water and the PIM-I solvent, THF. Perfluorinated ethers, however, are not miscible with most organic solvents and can be used for this purpose. FC-75 (3M™) with a boiling point of 102 0C was selected for use in the following investigation.

A first membrane (Entry 8.1) consisted of an untreated PAN support coated with a 1.0 % solution of PEVI-I in cyclohexanone. A second membrane (Entry 8.2) consisted of an untreated PAN support coated with a 0.65 % solution of PIM-I in a solvent mixture consisting of 13 % cyclohexanone in THF. A third membrane (Entry 8.3) consisted of a PAN support pretreated with a liquid perfluoroether (FC-75) coated with a 0.65 % solution of PEvI-I in a solvent mixture consisting of 13 % cyclohexanone in THF. A fourth membrane (Entry 8.4) consisted of an untreated PAN support coated with a 0.69 % solution of PEvI-I in a solvent mixture consisting of 4.3 % ethyl acetate in THF. A fifth membrane (Entry 8.5) consisted of a PAN support pretreated with a liquid perfluoroether (FC-75) coated with a 0.69 % solution of PIM-I in a solvent mixture consisting of 4.3 %

ethyl acetate in THF. A sixth membrane (Entry 8.6) consisted of a PAN support

pretreated with a liquid perfluoroether (FC-75) coated with a 0.43 % solution of PIM-I in

a solvent mixture consisting of 9 % dioxane in THF.

Entry Treatment Solvent Cone. P(O2) P(O2/N2) CaIc. PIM-I / m3/m2 h bar Thickness / % / μm 8.1 No cyclohexanone 1.00 ~1 8.2 No THF 0.65 0.09 2.40 10.0 (13%cyclohexanone) 8.3 FC-75 THF 0.65 0.08 3.49 14.0 (13 %cyclohexanone) 8.4 No THF 0.69 0.65 4.10 1.6 (4.3%ethyl acetate) 8.5 FC-75 THF 0.69 0.50 4.60 2.0 (4.3%ethyl acetate) 8.6 FC-75 THF 0.43 1.39 2.24 0.8 (9%dioxane)

Table 13. Modulation of Solvent Strength and

Pretreatment of Support With a Perfluoroether

When pure cyclohexanone was used as the solvent (Entry 8.1) a clear solution was

observed whilst the solution was at a temperature above room temperature which turned

turbid upon cooling to room temperature. No dense layer was formed on the support.

The coating solution consisting of a 0.65 % PIM-I solution in 13 % cyclohexanone / THF

(Entry 8.2) provided a composite membrane which exhibited a low O2-ρermeance (0.09

m3/m2 h bar) and low selectivity (2.4). Pretreatment of the PAN support with FC-75 prior

to coating increased the selectivity to 3.49 at a low O2-permeance (0.08 m3/m2 h bar).

Using the low O2-permeance values measured for Entries 8.2 and 8.3 allowed the

calculation of effective selective layer thickness of 10.0 μm and 14.0 μm respectively.

From the low PIM-I concentration applied it is concluded that the high boiling additive,

cyclohexanone (b.p. = 155°C), causes a more densely packed selective layer to be formed

on the surface of the support which, as would be expected, exhibits a low permeability. It can be assumed that the real selective layer thickness will be much lower than the calculated values.

Addition of a small amount of ethyl acetate to a 0.69 % PM-I / THF solution enabled a composite membrane to be formed (Entry 8.4) which exhibited a slightly higher selectivity (4.10) than the intrinsic selectivity (4.0) of the unsupported 46 μm PEVI-I film (Table 2) and significantly higher than the selectivity (2.82) exhibited by a similar PAN / PM-I membrane (Table 8, Entry 5.3) formed using a 0.75 % PIM-I / THF coating solution. The oxygen permeance of the untreated composite membrane formed using the 4.3 % ethyl acetate / THF solution (Entry 8.4) was seven times higher than that of the untreated membrane formed using the 13 % cyclohexanone / THF solution (Entry 8.2) and 9 times higher than that of the unsupported 46 μm PM-I film (Table 2). Pretreatment of the PAN support with FC-75 prior to coating further increased the selectivity to 4.60 at an O2-permeance of 0.50 m3/m2 h bar. The O2-ρermeance values were used to calculate effective selective layer thickness of 1.6 μm and 2.0 μm for the composite membranes incorporating the non-pretreated and pretreated supports respectively. It will be evident to the skilled person that on the basis of these results composite membranes exhibiting even higher oxygen permeances may be obtained by using coating solutions comprising ethyl acetate as an additive and lower concentrations of the selective layer polymer, e.g. PM-I.

Dioxane as an additive at low PM-I concentration (0.43 %) provided a composite membrane which exhibited a reasonably high O2-permeance of 1.4 mVm2 h bar but at a selectivity below the intrinsic selectivity (4.0) of the unsupported 46 μm PM-I film (Table 2) and the selectivities of the other four composite membranes produced using a method employing THF as the predominant solvent (Entries 8.2 - 8.5).

The composite membranes formed following pre-treatment of the support with the perfluoroether FC-75 (Entries 8.3 and 8.5) exhibited only slightly lower oxygen permeances but significantly improved selectivities compared to the membranes incorporating the non-pretreated support (Entries 8.2 and 8.4). It therefore may be concluded that one way in which higher selectivities can be achieved is to pretreat the supporting layer with a perfluoroether, such as FC-75, prior to coating of the selective membrane layer. Example 8 - Modification of Composite Membrane Properties By Treatment of Composite Membrane Following Coating; of Selective Layer on Supporting Layer

In addition to pre-treatment of the supporting layer, it has also been determined that treating the composite membrane following its formation, i.e. after the selective membrane layer has been formed on the supporting layer, may also modify membrane performance such that membrane properties could be controlled to suit a particular application.

Example 8.1 - Treatment of Composite Membrane with PDMS

A composite membrane in accordance with the present invention consisting of an untreated PAN support coated with a 0.69 % solution of PIM-I in a solvent mixture consisting of 4.3 % ethyl acetate in THF was produced (corresponding to the membrane denoted Entry 8.4 in Table 13). The composite membrane was immersed for 20 minutes in n-hexane, dried for 30 minutes in ventilated air at room temperature and finally coated with a 0.35 % solution of PDMS in isooctane.

The gas permeance and selectivity for a range of gases were determined and the results are reported in Table 14. Treatment Gas Pressure Flux Permeance P(X/N2) / bar / ml/min / m3/m2 h bar N2 5.95 31 0.08 1.0 O2 5.96 150 0.40 4.9 n-hexane immersion N2 1.05 879 13.23 1.0 O2 1.08 1520 22.24 1.7 He 1.11 1370 19.50 1.5 H2 1.10 3400 48.84 3.7 CH4 1.10 1840 26.43 2.0 N2 1.08 930 13.61 1.0 measurement after 1 h N2 1.08 850 12.44 1.0 O2 1.08 1450 21.21 1.7 PDMS coating N2 1.98 116 0.93 1.0 O2 0.94 154 2.59 2.8 He 0.95 179 2.98 3.2 H2 0.95 350 5.82 6.3 CH4 0.95 143 2.38 2.6 CO2 0.37 367 15.67 16.9

Table 14. Treatment of Composite Membrane with PDMS

The O2-permeance of the composite membrane before immersion in n-hexane was 0.40

m3/m2 h bar. Following immersion in n-hexane the O2-permeance increased initially to

22.24 m3/m2 h bar and fell slightly to 21.21 m3/m2 h bar one hour later. Following coating

with a dilute, crosslinkable PDMS solution (0.35 % PDMS in isooctane) the O2-

permeance fell to 2.6 m3/m2 h bar. The selectivity of the untreated composite membrane

was 4.9 and fell to 2.8 following treatment although it is worth noting that the selectivity

of the treated membrane is above the intrinsic selectivity of the PDMS thin film given in

Table 7. Example 8.2 - Treatment of Composite Membrane with Teflon AF 2400

A composite membrane incorporating a support made from PAN and a selective membrane layer comprising PIM-I was prepared in accordance with Example 2 using a PIM-I concentration of 0.60 % in THF (corresponding to the membrane denoted Entry 5.4 in Table 8). The membrane was coated with a 0.35 wt% solution of Teflon AF 2400 in FC-75. The permeances for a range of gases were measured at room temperature and used to calculate selectivities relative to nitrogen. The results are presented in Table 15.

Entry P(N2), P(O2), P(He), P(H2), P(CH4), P(CO2), nrVni2 h bar m3/m2 h bar m3/m2 h bar m3/m2 h bar nrVm2 h bar nvVm2 h bar 11.1 0.130 0.56 1.71 2.66 0.22 3.85 11.2 0.077 0.35 1.27 1.53 0.12 2.20 11.3 0.105 0.50 1.61 2.34 0.18 3.14 Av. 0.104 0.47 1.53 2.18 0.17 3.06

Entry P(O2/N2) P/(He/N2) P/(H2/N2) P(CH4ZN2) P(CO2/N2)

11.1 4.3 13 21 1.7 30 11.2 4.5 17 20 1.6 29 11.3 4.7 15 22 1.7 30 Av. 4.5 15 21 1.7 29

Table 15. Treatment of Composite Membrane with Teflon AF 2400

Relative to the untreated PAN/PM-1 membrane (Table 8, Entry 5.4) an increase of P(O2/N2)-selectivity from 2.84 to 4.5 was observed and was accompanied by a decrease in oxygen permeance from 4.16 to 0.47 m3/m2 h bar.

The above experiment was repeated using more dilute solutions of Teflon AF 2400 (0.2 and 0.1 wt %) and provided composite membranes which exhibited higher permeances than the membrane coated with the 0.35 wt% Teflon AF 2400 solution and which still exhibited acceptable selectivities for a range of potential applications.

All of the composite membranes coated with a Teflon AF 2400 solution exhibited a higher P(XZN2) selectivity relative to the intrinsic selectivity of the unsupported 91 μm PM-I film. For example, a comparison of the P(X/N2) selectivity results for the 91 μm

PIM-I film to those presented in Table 15 for a selection of gases are shown in Table 16.

Selectivity / P(X/N2) Gas 91 μm PIM-I film Teflon AF 2400 coated Percentage membrane Increase O2 3.5 4.5 -130 He 4.8 15 -300 H2 10 21 -200 CO2 24 29 -120

Table 16. Comparison of P(X/N2) selectivity.

The marked increase of permeability is explained by a swelling of the unsupported thin

film and a widening of the microporous polymer network. This can be influenced by

directed crosslinking of the polymer chains as described above in relation to the

microporous materials suitable for incorporation into the inventive membranes. It will be

evident to the skilled person that partial crosslinking of the polymer chains will obstruct

swelling and hence result in a less pronounced increase in permeance. Example 9 - Application of Inventive Composite Membranes to Nanofiltration

Preliminary nanofiltration testing was carried out using a composite membrane incorporating a support made from PAN and a selective membrane layer comprising PM-I prepared in accordance with Example 2 using a PDVI-I concentration of 0.60 % in THF (corresponding to the membrane denoted Entry 5.4 in Table 8). As reported in Table 7, the composite membrane permeates O2 at 4.16 m3/m2 h bar and an effective layer thickness of 0.3 μm was calculated for the separation layer.

1 wt% solutions of polyethylene glycol (PEG) with a molecular weight of 900 g/mol in acetone and toluene were applied to the inventive PAN / PIM-I membrane and the retention of PEG determined. The results are reported in Table 17.

Solvent Pressure / bar Flux / 1/m2 h bar Retention / % acetone 2 29 38 toluene 2 31 29 toluene 4 29 29

Table 17. Preliminary nanofiltration results for an inventive membrane.

A very high flux of about 30 1/m2 h bar was observed in all cases. Retentions of 38 % (acetone) and 29% (toluene) were measured. It will be evident to the skilled person that crosslinking of the selective layer will result in much higher retention, potentially up to 99 %, accompanied by a reduced flux. Preparation of PIM-I (1)

(2) (3) (D A mixture of anhydrous potassium carbonate, 3,3,3',3'-tetramethyl-l,l"-spirobisindane- 5,5',6,6'-tetrol (2; 10.25 g, 30.1 mmol) and 2,3,5,6-tetrafluoroterephthalonitrile (3; 6.02 g, 30.1 mmol) in dry DMF (200 mL) was stirred at 65 °C for 72 h. On cooling, the mixture was added to water (300 mL) and the crude product collected by filtration. Repeated reprecipitations from methanol gave 13.15 g (95%, yield) of fluorescent yellow polymer 1. (Found C, 74.85; H, 4.23; N, 6.04% C2PH2ON2O4 requires C, 75.64; H, 4.38; N, 6.08%);% (300 MHz, d6-CDCl3): 1.20-1.5 (12H, br m), 2.15-2.5 (4H, br m), 6.5 (2H, br s), 6.8 (2H, br s); Mn = 170 x 103 amu; Mw = 290 x 103 amu (using GPC vs. polystyrene standards); Surface area (powder, N2 adsorption, BET calculation) = 811 m2 g"1.

A N2 adsorption/desorption isotherm for PIM-I (1) is shown in Figure 4. Preparation of PIM-N2 (74)

A mixture of anhydrous potassium carbonate, 3,3,3',3'-tetramethyl-l,r'-spirobisindane- 5,5',6,6'-tetrol (2; 0.58 g, 1.7 mmol) and monomer 45; 1.05 g, 1.7 mmol) in dry DMF (70 mL) was stirred at 120 0C for 72 h. On cooling, the mixture was added to water (300 mL) and the crude product collected by filtration. Repeated reprecipitations from methanol gave 0.98 g (71%, yield) of a yellow polymer 74. (Found C, 77.43; H, 5.35; N, 6.57; Cl, 1.32% C54H44N4O4 requires C, 79.78; H, 5.46; N, 6.89; Cl5 0.00%);<5H (300 MHz, O6- CDCl3): 1.40 (12H, br m), 1.66 (12H, br m), 2.15-2.5 (4H, br m), 2.70 (4H, br m), 6.5 (2H, br s), 6.9 (2H, br s), 7.3-7.7 (6H, br m), 7.95 (2H, br s); Mn = 18 x 103 amu; Mw = 50 x 103 amu (using GPC vs. polystyrene standards); Surface area (powder, N2 adsorption, BET calculation) = 695 m2 g"1; Surface area (film, N2 adsorption, BET calculation) = 640 mV- Preparation of Films

PM-I (1) Preparation of 46 μm and 91 μm Films

PIM-I (1) prepared as set out above was dissolved in THF (2-5 wt-%) and cast on a glass plate. After evaporation of the solvent and drying dense, clear, yellow films were obtained. Gas permeation data were measured at 30°C with pure gases using a pressure increase time-lag apparatus operated at low feed pressure (typically at 200-300 mbar). Permeation (P) was calculated from the slope in the steady state region, apparent diffusion (Dapp) coefficients from the time-lag and sorption by division of ΫfDapp.

Preparation of a 60 urn Film

0.5g of PM-I (1) prepared as set out above was added to a conical flask. Solvent (THF; 25 ml) was poured into the conical flask, a magnetic stirrer bar added, then the flask was closed with a cap. The flask was placed on a magnetic stirrer until the polymer completely dissolved. The dissolved polymer was poured into a glass petri dish of diameter 12 cm. The petri dish was placed in a dessicator under a slow flow of nitrogen and left until the polymer was dry (about 4 days). The thickness of the membrane was measured as 60 μm (average of 7 measurements) with a micrometer screw gauge. The membrane was yellow in colour and exhibited luminescence under ultraviolet light. It had a surface area of 622 m2 g"1 by nitrogen adsorption.

PM-N2 (74) Preparation of a 28 urn Film PM-N2 (74) was prepared as set out above and a 28 μm film formed from CHCl3 solution. REFERENCES

1. Robeson, L.M., J. Membrane Sci. 62 (1991) 165-185. 2. Robeson, L.M., Curr. Opin. Solid St. M. 4 (1999) 549-552. 3. Freeman, B.D., Macromolecules 32 (1999) 375-380. 4. Alentiev, A. Y., Yampolskii, Y., J. Membrane Sci. 165 (2000) 201-216. 5. Robeson, L.M., Burgoyne, W.F., Langsam, M., Savoca, A.C. Polymer 35 (1994) 4970-4978. 6. Al-Masri, M., Kricheldorf, H.R., Fritsch, D., Macromolecules. 32 (1999), 7853- 7858. 7. Alentiev, A.Y.,Yampolskii, Y.P., Shantorovich, V.P., Nemser, S.M., Plate, N.A., J. Membrane Sci. 126 (1997) 123-132. 8. Arcella, V., Colaianna, P., Maccone, P., Sanguineti, A., Gordano, A., Clarizia, G., Drioli, E., J. Membrane Sci.163 (1999) 203-209. 9. Hofmann, D., Entrialgo-Castano, M., Lerbret, A., Heuchel, M., Yampolskii, Y., Macromolecules 36 (2003) 8528-8538. 10. Ilinich, O.M., Lapkin, A.A., Polymer 43 (2002) 3209. l l. Masuda, T., Isobe, E., Higashimura, T., Takada, K., J. Am. Chem. Soc. 105 (1983) 7473. 12. Nagai, K., Masuda, T., Nakagawa, T., Freeman, B.D.,Pinnau, L, Progr. Polym. Sci. 26 (2001) 7221. 13. Rezac, M.E., Koros, WJ., J. Appl. Polym. Sci. 46 (1992) 1927-1938. 14. Henis, J.M.S., Tripodi, M.K. J. Membrane Sci., 8 (1981) 233. 15. Lopez, J.L., Matson, S.L., Marchese, J., Quinn, J.A., J. Membrane Sci., 27 (1986) 301-325. 16. Pinnau, I., Wijmans, J.G., Blume, L, Kuroda, T., Peinemann, K.V., J. Membrane Sci. 37 (1988) 81-88. 17. Tusel G.F. and Bruschke H.E.A. Desalination 53 (1985) 327. 18. Jafar, JJ., Budd, P.M., Hughes, RJ. Memb. Sci. 199 (2002) 117. 19. Abboud et al, Pure Appl. Chem. 71 (1999) 645. 20. Hansen CM., Hansen Solubility Parameters, CRC Press, 1999. 21. Kinzer, Lloys, D.R., Wightman, J-P., McGrath, J.E., Desalination, 46 (1983) 327- 334.