Selective Swelling and Functionalization of Integral Asymmetric Isoporous Block Copolymer Membranes

SNIPS stands for a membrane fabrication technique that combines the evaporation induced self-assembly of the block copolymers and the classical nonsolvent induced phase separation. It is a one-step readily scalable technique to fabricate integral asymmetric isoporous membranes. The prominent developments in the last decade have carved out a niche for SNIPS as a potential technique to fabricate next generation isoporous membranes. In the last decade, a rich polymer library and variety of membrane postmodiﬁcation routes have been successfully implemented to fabricate SNIPS membranes having the desired pore functionality. Some of these membranes form soft nanochannels in hydrated state due to swelling of the pore wall, i.e., the pore forming block of the block copolymer. These membranes having soft nanochannels have demonstrated the potential to perform several challenging separation tasks in ultraﬁltration and nanoﬁltration. This paper highlights the currently accessible pore functionality, the strategies to tune the swelling of the soft nanochannels, the potential applications, and future perspectives of these membranes.


Introduction
The fabrication of integrally skinned asymmetric membrane by Loeb and Sourirajan was a major breakthrough in the development of membrane technology of the 20th century. [1,2] The technique involves controlled precipitation of a polymer solution by phase inversion in an immersion bath, resulting in a porous solid polymer membrane, commonly referred to as nonsolvent induced phase separation (NIPS). Although Loeb and Sourirajan reported this technique to prepare reverse osmosis membranes using cellulose acetate, [1,2] later this technique became popular for the preparation of microfiltration, ultrafiltration, and nanofiltration membranes as well. The key task of porous membrane development is to obtain the desired morphology for a separation DOI: 10.1002/marc.202100235 using a suitable material. An inherent advantage of the membrane morphology prepared by NIPS is the gradually increasing pore size from the surface toward the bottom of the membrane. The asymmetric porous morphology leads to high surface selectivity and prevents the transport resistance far from the surface. The depth selective morphology of symmetric membranes is one of the features which lead to nonremovable or irreversible fouling of the membrane. Highly porous robust membranes can be fabricated by NIPS using a large number of polymers, which are solid at the temperature range of membrane preparation and application, e.g., polysulfone, polyvinylidene fluoride, polyacrylonitrile, etc. For commercial production of porous polymer membranes the popularity of NIPS has surpassed all other techniques, such as uniaxial stretching, track etching, sintering, thermally induced phase inversion, etc. However, this class of membrane ends up having a rather broad distribution of the pore size at the surface. The surface pore size is often correlated with the ability of a membrane to completely retain substances of specific molecular weight referred to as "molecular weight cut-off." The broad surface pore size distribution of the NIPS membranes has been accused of being responsible for a rather diffuse molecular weight cut-off (Figure 1). It has resulted in a research impetus toward the development of techniques to prepare membranes having a narrow surface pore size distribution often referred to as isoporous membranes. The track etching method is widely used to prepare isoporous polymeric membranes. In this technique a high energy ion beam is bombared on a dense polymer (e.g., polycarbonate) film to obtain isoporous cylindrical channels. [3] Besides being expensive, a disadvantage of this top-down method is a rather low pore areal density, which is necessary to avoid the formation of bigger pores by merging channels in the random bombardment. Therefore, it was desirable to find another bottom-up approach, which allows for the formation of similar sized pores with a large number density. Thus, the ability of the block copolymers to self-assemble in well-ordered microphase separated domains has been an intriguing feature to prepare isoporous membranes for decades. [4,5] Initial trials were based on fabrication of a dense block copolymer thin film containing vertically aligned cylindrical domains on a nonporous primary support followed by etching out the cylinder forming block to obtain an isoporous template. [6][7][8] Instead of etching out the cylinder forming block, the pores of such templates can also be generated by washing out a homopolymer blend partner of the cylinder forming block, [9] by removing a sacrificial moiety of the cylinder forming block via chemical postmodification [10] and selective swelling. [11] The isoporous layer must be transferred on top of a porous support from the primary substrate in order to ensure the mechanical robustness of the membrane. [11] Effort has been made to eliminate this step by coating the thin block copolymer layer directly on top of a macroporous support followed by cold zone annealing and etching out the cylinder forming block. [12] In 2007, a technique to prepare isoporous membranes using a combination of the evaporation induced self-assembly of block copolymer and the nonsolvent induced phase separation (SNIPS) was reported. [13][14][15][16] Neither the etching out of the pore forming block nor the transferring of the isoporous layer from one substrate to another are involved in SNIPS. A vertically aligned isoporous layer on top of a highly porous spongy substructure is obtained in one step via SNIPS. Due to the ease of scalability and the potential to use it commercially for fabrication of isoporous membranes, SNIPS has been extensively explored [15][16][17][18][19][20][21] ever since it was first reported. [13]

Integral Asymmetric Isoporous Membranes via SNIPS
Synthesis of a block copolymer of well-defined composition and molecular weight having narrow molecular weight distribution is the prerequisite of a SNIPS membrane preparation. So far living anionic polymerization has been mostly employed for the synthesis of block copolymers being used for SNIPS. Yet advances of controlled radical polymerizations are paving the way to increase the choice of possible monomers and to scale up the polymer synthesis by economically feasible methods. While controlled radical polymerzation techniques offer the synthetic accessibility of a wide variety of monomers, the major drawback has been the limitation to obtain a molecular weight high enough to ensure the mechanical integrity of a SNIPS membrane. Recent progress of the reversible addition-fragmentation chain transfer (RAFT) polymerization is particularly auspicious for the next big leap in this regard. [22][23][24][25] Polymers and block copolymers with sufficiently large degrees of polymerization and low dispersity can be synthesized by RAFT polymerization, [26] even rather eco-friendly methods using photoRAFT have been developed. [27] An asymmetric diblock copolymer is the most common choice for the fabrication of a SNIPS membrane. Successful fabrication of the SNIPS membranes have been reported using polystyreneblock-poly(4-vinylpyridine) (PS-b-P4VP), [13] polystyrene-block-poly(2-vinylpyridine) (PS-b-P2VP), [28] poly(tert-butylstyrene)block-poly(4-vinylpyridine) (PtBS-b-P4VP), [29] poly(4-trimethyl silylstyrene)-block-poly(4-vinylpyridine) (PtTMS-b-P4VP), [29] poly (4-methylstyrene)-block-poly(4-vinylpyridine) (P4MS-b-P4VP), [30] poly( -methylstyrene)-block-poly(4-vinylpyridine) (P MS-b-P4VP), [30] polystyrene-block-poly(ethylene oxide) (PS-b-PEO), [31] polystyrene-block-poly(2-hydroxyethyl methacrylate) (PS-b-PHEMA), [32] polystyrene-block-poly(glyceryl methacrylate) (PS-b-PGMA), [33] polystyrene-block-poly(N-isopropylacrylamide) (PS-b-PNIPAM), [34] polystyrene-block-poly(acrylic acid) (PS-b-PAA) [35] etc. (Figure 2). The major block of the diblock copolymer acts as the matrix forming block, while the minor block acts as the pore forming block of the isoporous layer. In a typical SNIPS protocol, the membranes are prepared by first dissolving the block copolymer in a mixture of solvents having different volatility and distinct affinity for the matrix forming and the pore forming block, respectively. The combination of tetrahydofuran and dimethylformamide is one of the most popular binary solvent mixtures while using a ternary solvent mixture (e.g., mixture of tetrahydofuran, 1,4-dioxane, and dimethylformamide) is a common practice as well. After casting the solution on top of a substrate (glass plate or nonwoven porous support), at least one of the solvents of the mixture must evaporate rapidly in order to drive the self-assembly of the block copolymer perpendicular to the surface. A high boiling solvent must sufficiently partition itself between the two blocks to ensure that the pore forming block adopts a stretched conformation, i.e., remains strongly swelled. Eventually, in immediate vicinity of the surface the matrix forming block acquires a rather collapsed conformation around the highly stretched pore forming block during the solvent evaporation (Figure 3). Meanwhile along the cross section of the solution layer a polymer concentration gradient develops which is caused by the evaporation taking place from the surface. At this nonequilibrium stage quenching in a precipitation bath leads to formation and fixation of the desired integral asymmetric isoporous membrane.
Several triblock terpolymers have also been used successfully for SNIPS membrane preparation. The third block can serve either as a matrix forming block or a pore forming block. For polyisoprene-block-polystyrene-block-poly(4-vinylpyridine) (PI-b-PS-b-P4VP) [36] and polyisoprene-block-polystyrene-blockpoly(N,N-dimethylacrylamide) (PI-b-PS-b-PDMA) [22] both the PI end block and PS middle block serve as the matrix forming blocks ( Figure 2). Compared to the SNIPS membranes prepared by diblock copolymer where only PS act as the matrix forming block the triblock terpolymer membranes having soft PI and hard PS are expected to have better mechanical www.advancedsciencenews.com www.mrc-journal.de robustness. In polystyrene-block-poly(2-vinylpyridine)-blockpoly(ethylene oxide) (PS-b-P2VP-b-PEO), [37] poly(styrene)-blockpoly(4-vinylpyridine)-block-poly(propylene sulfide) (PS-b-P4VPb-PPS), [38] and polystyrene-block-poly(4-vinylpyridine)-blockpoly(solketal methacrylate) (PS-b-P4VP-b-PSMA) [39] membranes only the PS end block acts as the matrix forming block ( Figure 2). The PEO block, the PPS block, and the PSMA block form the pore wall lining of these membranes, respectively. In this way, the triblock terpolymers bring about the opportunity to design membranes having the desired functional moieties along the pore wall of the isoporous layer. PS-b-P4VP-b-PPS having a terminal sulfhydryl group (i.e., thiol group) can be synthesized via sequential anionic polymerization by simply terminating the polymerization with alcohol where PPS is the last block of the sequence. The pore walls of a SNIPS membrane prepared from this polymer is decorated with thiol functional groups which can serve as the covalent binding sites via thiol-ene click chemistry. [38] Recently, a strategy to design pores having two www.advancedsciencenews.com www.mrc-journal.de distinct functional groups using miscible end blocks of a triblock terpolymer was demonstrated. On the example of a membrane having hydroxyl and pyridine moieties along the pore wall poly(4-(2-hydroxyethyl-thio)-2-methyl butene-random-4-(2-hydroxyethylthio)-3-methyl butene-random-isoprene)-block-polystyrene-blockpoly(4-vinylpyridine) (P(HTMB-r-I)-b-PS-b-P4VP) membrane was fabricated. [40] Using a segment having random distribution of two monomers as the pore forming block also leads to isoporous channels having two distinct functionalities. Partial esterification of the hydroxyl moieties of PS-b-PHEMA diblock copolymer using succinic anhydride and glycine leads to PS-b-P(HEMA-r-SEMA) [41] and PS-b-P(HEMA-r-HEMA-Gly-NH 2 ), [42] respectively ( Figure 2). The pores of a PS-b-P(HEMA-r-SEMA) membrane are composed of a mixture of hydroxyl and carboxylic groups, [41] while those of a PS-b-P(HEMA-r-HEMA-Gly-NH 2 ) membrane are decorated with hydroxyl and amine groups. [42] In the last decade, the fabrication of SNIPS membranes both in flat sheet [43][44][45][46][47] and hollow fiber [25,[48][49][50][51][52][53][54] geometry have been extensively explored. Empirical optimization of solvent mixture composition, polymer concentration, evaporation time, concentration of additives (if required), etc., are inevitable tasks of SNIPS membrane casting. Thus systematic study of the casting parameters of a series of polymers having different molecular weight and composition is one of the forefronts of SNIPS membrane research. [46,55] Some of the initial studies [56] invoked arguments [15] and inspired new research directions to find out the underlying structure formation mechanism of the integral asymmetric isoporous structure morphology of the SNIPS membranes using state of the art experimental techniques, e.g., cryo microscopy, in situ synchroton small angle X-ray scattering, time resolved grazing incedence small angle X-ray scattering, etc. [57][58][59][60][61][62][63] In recent years along with the empirical investigation computational modeling have also been employed to shed light on the complex nonequilibrium structure formation mechanism of these kinetically trapped membranes. [64] The study of correlations between casting parameters and membrane morphology has fostered deep fundamental understanding of the behavior of the block copolymers in the casting solution and the self-assembly mechanism during evaporation of the solvents. [57][58][59][60][61][62][63][64] In spite of the tremendous progress in this regard there remains concern of significant contribution of the porous substructure on retention of macromolecules by SNIPS membranes due to presence of some bottleneck pores. [65] The content of these bottleneck pores of the porous sublayer determines if the membrane is surface or depth selective. Although rather uncommon, formation of a dense layer at the interface between the integral asymmetric isoporous layer and the nonwoven porous substrate has been encountered as well. [66] Using in situ formed TiO 2 nanoparticles in the casting solution prepared by hydrolysis of titanium (IV) tetraisopropoxide has been pointed out as a possible strategy to counter such problems. So far this technique has been utilized only for PI-b-PS-b-P4VP [67] and polystyrene-block-poly(4-(2-hydroxyethyl-thio)-2-methyl butene-random-4-(2-hydroxyethyl-thio)-3-methyl butene-randomisoprene) (PS-b-P(HTMB-r-I)) [66] (Figure 2) membranes. In both cases, the block copolymer-TiO 2 coassembly led to organicinorganic hybrid membranes having a high open porous substructure. Typically, the scanning electron microscope (SEM) images of surface and cross section of the membranes serve as the basis of optimization of the casting parameters for SNIPS membranes. SEM images provide an overview of the porous morphology of the top isoporous layer and the spongy substructure. However, it is not possible to detect whether in the spongy porous layer there exist bottleneck pores which are smaller than the pores of the top layer. Although 3D reconstruction of the isoporous layer and the spongy substructure of the SNIPS membranes is possible from a series of microscopic images, [56] it is not practical to use the time consuming procedure for optimization of the casting parameters. Till now only a few studies have attempted to find out the molecular weight cut-off by retention of well-defined macromolecules, e.g., poly(ethylene glycol) (PEG) [22,25,65,67,68] along with optimization of casting parameters. Therefore, it has not yet been established whether the isoporous membranes have a sharper molecular weight cut-off compared to the nonisoporous membranes as expected ( Figure 1).

Strategies to Tailor the Soft Nanochannels of SNIPS Membranes
Unlike the conventional NIPS membranes the surface pore size of the SNIPS membranes can be tuned by changing the molecular weight and composition of the block copolymers [46] www.advancedsciencenews.com www.mrc-journal.de as well as blending two block copolymers of different molecular weight and compositions. [69] Several studies have demonstrated that controlled swelling of the pore forming block is an effective way to tune the separation performance of the SNIPS membranes. PS-b-P4VP has so far been the most popular choice for the fabrication of SNIPS membrane where the P4VP block acts as the pore forming block. P4VP is well known for its stimuli responsiveness due to protonation and deprotonation of the nitrogen moieties with the change of pH. At a pH below 5.5, the nitrogen moieties acquire a positive charge due to protonation. Under such condition, the protonated P4VP blocks swell in water. As the hydrophobic matrix forming PS blocks of the membrane are fixed in space, the protonated P4VP blocks are compelled to stretch toward the center of the pore which creates soft nanochannels. Typical SEM images represent the pore size of the membrane in a dry state where the P4VP blocks exist in a collapsed conformation. The effective pore size of the PS-b-P4VP membrane in a hydrated state at a pH below 5.5 becomes significantly lower than the dry state pore size of the membrane. Thus, the size of the open pores of PS-b-P4VP membrane can be controlled by adjusting the pH. Obviously this feature is also present in SNIPS membranes prepared from other polymers where pyridine moeities are present in the pore forming blocks, e.g., Small organic molecules [70][71][72] having affinity toward the pore forming block are popular additives to tune the dry state pore size of the SNIPS membranes. A combination of these two techniques, i.e., addition of small organic molecules in the casting solution to control the dry state pore size followed by swelling of the pore forming block using an external stimuli avail a broader regime to tune the size of the soft nanochannels. [72] Additional stimuli responsiveness and functionality can be introduced by chemical post modification of the membrane. The mussel inspired dopamine coating avails the cite to perform surface chemistry via both "grafting to" [73,74] and "grafting from" [75] methods. By linking amino terminated poly(N-isopropyl acrylamide) on a dopamine coated PS-b-P4VP membrane double stimuli (i.e., pH and temperature) responsive nature has been engendered. [73] The prerequisite to perform the chemical postmodification of a SNIPS membrane is to find a heterogeneous reaction route, which does not destroy the integral asymmetric isoporous morphology. The initial trials were mostly focused on aqueous phase reactions. In recent years, successful implementation of reactions using gas phase reactants [40,66,76,77] and fluorinated nonsolvents [78,79] have opened up new possibilities for chemical postmodifications. The P4VP blocks of PS-b-P4VP membranes have been quaternized by several reaction routes, such as ring opening coupling of 1,3-propane sultone, [77] reaction with chloroacetamide, [80] reaction with alkyl halides, [76,77] oxidation using a mixture of ethanoic acid, and hydrogen peroxide, [68] etc. Unlike P4VP, the oxidized P4VP block acquires a stretched conformation at a pH above 5 due to its polyelectrolyte nature. The 4VP moieties of the pore forming block are converted into zwitterionic moieties due to oxidation. As the pH goes below 5 the soft nanochannels shrinks in case of the oxidized PS-b-P4VP membrane. Consequently, while with the decrease of pH from 5 to 1 the water permeance through the PS-b-P4VP membrane decreases dramatically, water permeance increases for the oxidized PS-b-P4VP membrane. [68] Quaterniza-tion of the nitrogen moieties of the P4VP blocks using alkyl halide is the most popular PS-b-P4VP membrane postmodification technique. [40,76,77] Similar like the protonated P4VP block, the quaternized P4VP block also swells in the hydrated state. Thus, the size of the soft nanochannels can be tuned by controlling the degree of quaternization. The P4VP block acquires a fixed positive charge after quartenization with alkyl halide. With the increase of the degree of quaternization the available protonation cites decrease. A P4VP segment completely quaternized by alkylhalides does not respond to pH, rather it exists in a swollen state irrespective of pH. The hydrophilicity of the quaternized P4VP is also dictated by changing the size of the alkyl group used for quaternization. Such as for equal degree of quaternization, the hydrophilicity of the pore forming block gradually increases when 1-propyl iodide, ethyl iodide, and methyl iodide are used as the quaternization agent, respectively (Figure 4). [76] In recent years, SNIPS membranes having carboxyl and hydroxyl functional groups in the pore forming blocks have become popular. Vertically aligned hexagonally packed pores of narrow pore size distribution on a spongy substructure has been obtained from PS-b-PAA, PS-b-PHEMA, PS-b-PGMA, PS-b-P(HEMA-r-SEMA), PS-b-P(HTMB-r-I), etc. Moreover, PI-b-PS-b-PDMA is used to obtain the integral asymmetric isoporous morphology followed by hydrolysis of the PDMA block to obtain PI-b-PS-b-PAA membrane having carboxylic functional groups along the pore wall. [22][23][24][25] Similarly, hydrolysis of a PS-b-P4VP-b-PSMA membrane leads to formation of a PS-b-P4VP-b-PGMA having hydrophilic pores composed of hydroxyl functional groups. [39] Unlike PS-b-P4VP, the hydrophilic pore forming blocks of the PS-b-PHEMA membranes do not adopt a collapsed conformation in a hydrated state irrespective of the pH. Inclusion of hydrophobic groups along with the hydroxyl moieties is a possible way to alter the size of the hydrated soft nanochannels of the membrane. An example hereof is the postmodification of the hydroxyl moieties of PHEMA using ethyl isocyanate (i.e., urethane chemistry [81] ) in a perfluoro(methyl cyclohexane) nonsolvent bath (Figure 4). [78,79] Instead of using a membrane postmodification route it is also possible to synthesize a polymer having a random distribution of hydrophilic and hydrophobic moeities in the pore forming block, e.g., PS-b-P(HTMB-r-I) for membrane fabrication. However, the scope of this approach is limited due to the complications to find out the optimum casting parameters to obtain the desired SNIPS membrane. Particularly for PS-b-P(HTMB-r-I) it has been reported that the content of hydroxyl groups containing hydrophilic HTMB moieties must be above a threshold to ensure the formation of the isoporous channels via SNIPS. If the content of HTMB moieties is below that threshold, the solvents of casting solution do not sufficiently partition between the pore forming block and matrix forming block during the solvent evaporation stage of SNIPS which is mandatory for the formation of the isoporous channels. [66] Thus, it is more convenient to control the content of hydrophobic and hydrophilic moieties of the pore forming block simply by varying the reaction time of a suitable postmodification route as demonstrated by postmodification of PS-b-PHEMA membrane via catalyst free urethane chemistry. [79] On the other hand, hydrophilicity and negative charge density of the pore walls can be enhanced by replacing hydroxyl and carboxylic moieties with sulfonic acid moities. In this regard, the hydroxyl groups of a PS-b-P(HTMB-r-I) membranes were postmodified via a gas phase ring opening coupling of 1,3-propane sultone (Figure 4). [66] Sulfonic acid moieties have also been attached to the pore walls of a PI-b-PS-b-PAA membrane by a carbodiimide coupling reaction in aqueous phase to obtain PI-b-PS-b-PADSA membrane ( Figure 4). [25] The carbodiimide coupling chemistry has also enabled decorating the pore walls of PI-b-PS-b-PAA membrane with bioinspired heavy metal binding ligands glutathione and cysteamine. [24]

SNIPS Membranes for Fractionation of Proteins
Fractionation of proteins is the most widely addressed potential application of SNIPS membranes. In the last two decades, both membrane materials and membrane-based process design were extensively developed for fractionation of proteins. Although the membrane based technology to separate proteins having large difference in molecular weight is well developed at present, fractionation of proteins having identical molecular weight is still confronted with low selectivity of the current commercially available ultrafiltration membranes. In the quest of membranes having higher selectivity compared to the current commercially available membranes, the permeation and retention behavior of the SNIPS membranes have been investigated using several model proteins, such as bovine serum albumin, [41,65,79,80,[82][83][84][85] hemoglobin, [41,79,80,85] catalase, [41,79,86] -globulin, [82,83] lysosome, [86] myoglobin, [86] and ferritin, [86] etc. Fractionation of proteins is not only influenced by the pore size and pore size distribution of the membranes. There are strong contributions of pH and ionic strength of the protein solution on the electrostatic interaction between the proteins and membrane. As the conventional NIPS membranes have rather poor size selectivity owing to the broad pore size distribution, the tuning of electrostatic interactions between the membrane and the proteins by changing the pH and ionic strength of the solution has been widely exploited to enhance the selectivity of the protein pairs having small molecular size difference. [87][88][89][90][91][92][93] If the proteins have a strong affinity for the membrane material a cake layer forms at the surface of the membrane due to adsorption which eventually controls the permeation and rejection behavior of the membrane rather than the pores of the membrane itself. This phenomenon is currently used in some protein separation application by conventional NIPS membranes, such as separation of casein from milk for cheese manufacturing. Obviously formation of such cake layer on top of a SNIPS membrane is highly undesirable as it will belittle the advantage of the narrow surface pore size distribution of the membrane. The most widely used model protein bovine serum albumin has a tendency to adsorb at the surface of the PS-b-P4VP [65] and quaternized PS-b-P4VP [77] membranes at neutral pH. The isoelectric point of bovine serum albumin is around 4.7 which means at neutral pH it acquires a strong negative charge. As the bovine serum albumin is neutral at the isoelectric point it can permeate through the PS-b-P4VP membranes having sufficiently large pore size. At pH 4.7 the PS-b-P4VP membrane can be used to separate bovine serum albumin from hemoglobin which has identical molecular weight. [80] The isoelectric point of hemoglobin is at ≈pH 7.2. Hence, both hemoglobin and PS-b-P4VP membrane acquires a positive charge at pH 4.7, while the bovine serum albumin is neutral. Due to the electrostatic repulsion the PSb-P4VP membrane rejects the hemoglobin, while the neutral bovine serum albumin permeates through the membrane. The www.advancedsciencenews.com www.mrc-journal.de 2-choloracetamide quaternized PS-b-P4VP membrane has a positive charge and a smaller effective pore size at the hydrated state. A complete exculsion of one protein while allowing the permeation of others has been reported by optimizing the degree of quaternization of the PS-b-P4VP membrane. [80] This technique to separate bovine serum albumin and hemoglobin by taking advantage of the electroneutrality of one of the proteins was proposed more than two decades ago. A separation factor of 70 was achieved using a polysulfone membrane by performing the separation at the isoelectric point of hemoglobin. [94] It unequivocally is a potent technique to separate proteins of identical size. But a slight deviation of pH from the isoelectric point leads to a sharp drop of membrane selectivity as the success of the technique relies on the electroneutrality of one protein. In order to establish protein fractionation by SNIPS membrane as a realistic platform it is essential to demonstrate the better size selectivity of the isoporous membranes compared to the conventional nonisoporous membrane rather than relying on the electroneutrality of one protein for separation. By retention of single proteins from aqueous solution it has been demonstrated that the size selectivity of PS-b-PHEMA membranes can be improved by tuning the size of the soft nanochannels by two techniques-i) thermal annealing and ii) chemical postmodification of the pore forming block via urethane chemistry. [79] A membrane prepared by a phase inversion method tends minimize the surface tension and reach a more energetically favorable state compared to the kinetically trapped state. Thermal annealing of the PS-b-PHEMA membrane allows systematic reduction of the pore size due to controlled relaxation of the polymer segments. During chemical postmodification the reduction of the pore size occurs as a result of filling up of the empty space of the pores. The dry state pore size of the PS-b-PHEMA membranes were systemically reduced from 34 to 9 nm by controlling the time of thermal annealing and chemical postmodification. [78,79] But the effective pore sizes of the membranes in hydrated state are largely dictated by the hydrophilicity of the pore forming block. Due to the swelling of the pore forming block in hydrated stated the effective pore size of the membranes are significantly different from the dry state pore size. The controlled postmodification of the PS-b-PHEMA membranes resulted in higher ideal selectivity of the protein pairs. [79] The ideal selectivity of hemoglobin over catalase was increased ninefold compared to the unmodified PS-b-PHEMA membrane at pH 7.4. The increase of ideal selectivity was not due to formation of cake layer at the surface of the membranes as bovine serum albumin, hemoglobin and catalase do not adsorb at the surface of these membranes at pH 7.4. These three proteins and the PS-b-PHEMA membranes acquire a negative charge at this pH. Thus the increase of ideal selectivity was merely due to the increase of size selectivity by controlling the size of the soft nanochannels. However, the ideal selectivity of bovine serum albumin over hemoglobin (Figure 5) of these chemically postmodified and thermally treated PS-b-PHEMA membranes were not improved. [79] A threefold higher ideal selectivity compared to the unmodified PS-b-PHEMA membrane was reported by a PS-b-P(HEMA-r-SEMA) membrane. [41] The dimensions of hemoglobin (7 × 5.5 × 5.5 nm) resembles a sphere, while those of bovine serum albumin (14 × 4 × 4 nm) resembles a prolate ellipsoid. The possibility of the bovine serum albumin to orient itself in a way that the long axis is perpendicular to the surface of the membrane facilitate easier permeation through the pores of the membrane compared to that of spherical hemoglobin ( Figure 5). The PS-b-P(HEMA-r-SEMA) membrane takes advantage of the shape difference of the two proteins rather than relying on electrostatic repulsion, [41] but it remains unclear whether the combination of hydroxyl and carboxylic group along the pore wall provides any specific advantage for the permeation of bovine serum albumin through the pores. The ideal selectivities of the PS-b-P(HEMA-r-SEMA), the PS-b-PHEMA, the chemically postmodified PS-b-PHEMA, and the thermally treated PS-b-PHEMA membranes are higher than several commercial nonisoporous membranes. [41,79] It must be taken into account that the real selectivity of a protein mixture might deviate considerably from the ideal selectivities. The ideal selectivity is calculated from the ration of the transmission of the proteins through the membrane when a single protein aqueous solution is used as the feed solution. For the determination of the real selectivity of a protein pair, it is necessary to use an aqueous solution of the mixture of proteins as feed solution and determine the content of both of the protein in the permeate solution. The ideal selectivity values are not influenced by several factors www.advancedsciencenews.com www.mrc-journal.de which are unavoidable in a protein mixture, e.g., protein-protein interaction, composition of the protein mixture. Nevertheless these promising results demonstrate the potential of using SNIPS membranes for fractionation of proteins and tailoring the soft nanochannels to improve the membrane selectivity.

Nanofiltration -An Emerging Application of SNIPS Membranes
Nanofiltration bridges the separation zone between ultrafiltration and reverse osmosis in terms of molecular weight cut off. The nanofiltration membranes are reputed for the ability to separate monovalent ions from the divalent ions of water as the first nanofiltration membranes were utilized for softening of water. Apart from potable water softening, the nanofiltration membranes have already found applications in removal of heavy metals from waste water, desalting of dyes, removal of pesticides from agricultural waste water, etc. Conventional polymeric nanofiltration membranes have either integrally skinned asymmetric or thin-film composite structure. The integrally skinned asymmetric membranes prepared via NIPS consist of a rather dense top layer and a porous sublayer of the same material. [95,96] For the thin film composite membranes, a microporous support layer is prepared first by NIPS and then a thin selective layer is prepared on top of it by a different technique, such as interfacial polymerization, layer by layer assembly of polymer electrolytes, UV photo grafting, etc. [97] The lack of molecular level design limits the potential of these conventional nanofiltration membranes for many applications. The common motivation to design the next generation nanofiltration membranes is to overcome this limitation of the conventional membranes. Separation of molecules having dimensions in the range of 0.5-5 nm from each other are extremely challenging due to the difficulty associated with the control of the membrane pore size relative to the size of these molecules. Moreover, the separation of these small molecules requires specific functionalities along the pore wall having selective interaction toward the molecule of interest often referred to as molecular recognition feature of the membrane. [98][99][100] Strategies to control the pore size and introduction of functional group along the pore wall of the polycarbonate track-etched membranes are widely explored hereof. [99,100] Reducing the pore size of the commercial track-etched membranes by electroless gold deposition followed by chemisorption of functionalized thiols has become a common practice. [101][102][103] Electroless gold deposition has also been applied on PS-b-P4VP SNIPS membrane to reduce the pore size below 10 nm. [84] Preparation of isoporous template having sub 10 nm pore size by etching out the cylinder forming block of a self-assembled block copolymer pattern is challenging. It requires rather low molecular weight block copolymers having very high incompatibility between the blocks. [104] Significant progress has been made in recent years regarding preparation of block copolymer pattern having sub 10 nm domains. [105,106] A template having of 5 ± 1 nm pores was prepared from poly(cyclohexylethylene)-blockpoly(lactide) (PCHE-b-PLA) by selective etching of the cylindrical PLA domains. [107] However, a template prepared from a low molecular weight block copolymer may not have sufficient mechanical strength to withstand the transmembrane pressure during filtration. Similarly for the fabrication of a SNIPS membrane the molecular weight of the block copolymer should not be too low. The decrease of molecular weight of the block copolymer does not only lead to smaller pore size, the number of pores of the membranes simultaneously increase as well. [46] However, it is not practical to aim for a simultaneous increase of the number of pores and the reduction of pores size in the nanofiltration range as it will not be possible to ensure the mechanical robustness of the membrane. Directed assembly of liquid crystalline (LC) monomers into columnar LC mesophase followed by crosslinking of the monomers is an alternative techniques to obtain freestanding isoporous nanofilltration membranes having vertically aligned channels. [108][109][110] These membranes have very high pore density which is not possible to obtain by self-assembly of block copolymers. However, a freestanding membrane needs a minimum thickness to withstand the transmembrane pressure. So the advantage of high pore density will surpass the performance of asymmetric membrane only when the free standing membrane is thin enough to provide lower resistance compared to the ultrathin selective layer of an integral asymmetric isoporous membrane. Therefore, there are ongoing efforts to prepare composite membrane by mounting a thin crosslinked isoporous layer on a microporous support. [111] An effective strategy to obtain nanofiltration membrane via SNIPS is to use a blend of interacting AB/AC diblock copolymer of suitable molecular weight and compositions. A PS-b-P4VP/PS-b-PAA membrane capable of rejecting protoporphyrin IX (diameter 1.47 nm) from aqueous solution while allowing the permeation of lysine (diameter 0.9 nm) has been reported in this regard. [112] Controlled swelling of the pore forming block of SNIPS membrane is an excellent tool to tune the separation of such small molecules from each other. As mentioned earlier the P4VP block acquires a fixed positive charge after quaternization which makes the SNIPS membranes having quaternized P4VP as pore forming block ideal for removal of positively charged molecules from water [76] or separation of positively charged and neutral molecule in an aqueous media. [40] The retentions of the cationic methylene blue (lateral dimension 1.1 nm) from aqueous solutions have been tuned between 6% and 98% by controlling the swelling of the pore forming block via changing the degree of quaternization of PS-b-P4VP membranes. [76] Separation of cationic methylene blue and neutral reboflavin (lateral dimension 1.0 nm) is demonstrated using a quaternized P(HTMB-r-I)-b-PS-b-P4VP membrane. The positively charged soft nanochannels of this membrane strongly retained methylene blue and allowed permeation of riboflavin which led to a selectivity of 28.3 for a aqueous feed solution containing equimolar methylene blue and riboflavin. [40] Its worth mentioning here that the negatively charged molecules tend to adsorb on the quaternized P4VP block that neutralizes the positive charge and the pore forming block loses its ability to swell in the hydrated state. [40] The separation of two negatively charged small molecule can be tuned by controlling the size and charge density of a SNIPS membrane having negatively charged soft nanochannels which has been demonstrated using anionic model dye molecules. [40,66] The soft nanochannels of the PI-b-PS-b-PAA membrane becomes smaller when it is converted into PI-b-PS-b-PADSA membrane by attaching bulky disulfonic acid side groups along the pore wall. [25] Moreover, being composed of stronger polyelectrolyte pore forming the soft nanochannels of a PI-b-PS-b-PADSA membrane have lower degree of ionic strength response compared to those on PI-b-PS-b-PAA. As a result the molecular weight cut off by retention of neutral poly(ethylene glycol) molecules remains unchanged for a PI-b-PS-b-PADSA even when the feed solution containing NaCl or MgCl 2 have ionic strength upto 3M. For the PI-b-PS-b-PAA membrane the divalent magnesium ions alter the conformation of the hydrated pore forming which leads rather broad molecular weight cut off of the membrane. [25] Similarly, the sulfonation of hydroxyl moieties of PS-b-P(HTMB-r-I) [66] and P(HTMB-r-I)-b-PS-b-P4VP [40] membranes increase the negative charge density and swelling of the soft nanopores. Thus the rejection of the trivalent naphthol green B (lateral dimension 1.8 nm) and the hexavalent reactive green 19 (lateral dimension 1.9 nm) of sulfonated membranes are significantly higher compared to that of monovalent orange II (lateral dimension 1.3 nm). For an aqueous feed solution containing equimolar orange II and naphthol green B a selectivity of 44.6 was reported in case of sulfonated P(HTMB-r-I)-b-PS-b-P4VP membrane. [40] The successful separation of these model molecules proves the potential of using SNIPS membrane for separation of such small chemical, pharmaceutical, and biological molecules where the currently available commercial nanofiltration membranes are not applicable.

Challenges and Perspectives
The dimension, shape, and functionality of the surface pores play crucial roles in determining the selective permeation that in turn has a great impact on the separation performance of a membrane. The recent developments of integral asymmetric isoporous block copolymer membranes via SNIPS having hydrated soft nanochannels are summarized in this paper. The success of designing such membranes relies largely on controlled polymer synthesis and membrane postmodification routes while finding out the optimum membrane casting condition is an inevitable task as well. A rich polymer library to prepare SNIPS membranes and several membrane postmodification routes have been reported. Concurrently, the potential of using these fast developing functional membranes for several challenging separations in the ultrafiltration and nanofiltration regime has been demonstrated. Despite the progress achieved in the last decade several inadequacies remain to be solved. To ensure the progress of rational design of these membranes the emphasis of future research should be adopted toward several scientific and practical problems. Most of the block copolymers used preparation of SNIPS membranes so far contain polystyrene or derivatives of polystyrene as the major block ( Figure 2). As a result, the mechanical strength of the SNIPS membranes are lower compared to the conventional NIPS membranes prepared from polysulfones, polyetherimide, polyacrylonitrile, etc. The currently available SNIPS membranes possess sufficient mechanical strength for ultrafiltration. However, the lack of compaction resistance at around 10 bar transmembrane pressure remains a concern. Owing to the recent developments regarding the controlled swelling of the pore forming blocks, the SNIPS membranes are becoming attractive candidates also for the separations in nanofiltration regime. The lack of compaction resistance may turn out to be a critical limiting factor for nanofiltration applications. Although several facile strategies to control the swelling of the pore form-ing block of the SNIPS membranes to tailor the pore size of the soft nanochannels have been reported, it remains a challenge to determine the size of these channels in the hydrated state. Thus, permeation behavior of the membranes is often correlated with other parameters, such as postmodification time, degree of functionalization etc. Efforts have been made to determine the diameter of the swelled pore from water flux measurement using the Hagen-Poiseuille equation. This method leads to unrealistic values of the pore diameter as it ignores the hydrophilicity of the swelled isoporous channels and the resistance of the spongy substructure of the membrane. Atomic force microscope images of the membranes submerged in a liquid can demonstrate the swelling of the pores, but it is questionable whether this method can deliver quantitatively precise results of size of the swelled pores. Hence, the classical molecular weight cut-off experiments of neutral macromolecules or the retention of well-defined spherical particles are probably the most practical ways to estimate the size of the hydrated soft nanochannels. To gain insight into the separation mechanism of the SNIPS membranes it is important to correlate the permeation and retention behavior with the surface charge density and dimension of the soft nanochannels in hydrated state. Several membrane postmodification routes enhance the surface charge density of the SNIPS membranes which in turn improves the membrane selectivity. However, a chemical reaction occurs in the complete integral asymmetric isoporous layer of the membrane using the postmodifications strategies that have been reported so far. Therefore, the pore of the substructure also swells to some extent and the overall membrane resistance increases. The ideal chemical postmodification strategy would be to perform a reaction only at the isoporous layer of the membrane which is the selective layer of an ideal SNIPS membranes. As the isoporous layer and the substructure of a SNIPS membrane are composed of the same block copolymer, it will be challenging to design a scalable method which can selectively postmodify the isoporous layer. There are potentials to further adapt the already reported chemical postmodification techniques in this regard. Mostly aqueous solutions of model proteins and model dyes are used to demonstrate the separation performance of SNIPS membrane in the ultrafiltration and nanofiltration regime, respectively. While the retention behavior of aqueous feed solutions having single model component was widely studied in the last decade there have not been enough studies using aqueous solutions having mixtures of proteins and dyes. Moreover, the potential of using SNIPS membranes to perform separation in nonaqueous media remains unexplored. It is important to figure out the real life separation problems where the enhanced size selectivity of isoporous membranes can replace the nonisoporous membranes and mimic such separations in laboratory experiments. Being a one-step scalable method of integral asymmetric isoporous membrane fabrication technique, SNIPS has received a lot of attention by researchers ever since it was discovered. There has been tremendous progress in this membrane fabrication technique within a rather short span of time. The controlled swelling of the pore forming block of the isoporous layer of a SNIPS membrane has opened up the potential of using these membranes in a wide size regime, while the variety of pore functionality provides the option to explore the ideal membrane material for a desired separation. The prospect of using integral asymmetric isoporous membranes having www.advancedsciencenews.com www.mrc-journal.de hydrated soft nanopores to solve the real separation problems will become better if future research will address the above mentioned challenges.