The development of pervaporation hollow fiber membranes for isopropanol dehydration

CHAPTER ONE INTRODUCTION 1.1 Introduction of Membrane and Pervaporation A membrane is a layer of impermeable material which serves as a selective barrier between two phases to seperate particles, molecules, or substances when exposed to the action of a driving force [3]. Various membrane processes, such as reverse osmosis, ultrafiltration, microfiltration and dialysis, are widely applied in seawater desalination, ultra-pure water production, municipal and industrial waste stream treatment, purification of food and pharmaceutical products, fuel cells, controlled drug delivery and blood detoxification in hemodialysis and others applications. Because of the effectiveness, efficiency, energy and cost-saving of membrane process, many conventional separation processes have been replaced by large scale membrane processes. Membrane separation processes can be categorized into microfiltration, ultrafiltration, nanofiltration, reverse osmosis, dialysis, electrodialysis, gas separation, pervaporation, and membrane distillation, based on the driving force and the size of the molecules to be separated, as shown in Table 1.1. The driving force can be chemical potential gradient (i.e. concentration gradient or pressure gradient), or electrical potential 1 gradient across the membrane. The driving force across the membrane is differentiated by the mobility or concentration of each species in the membrane during selective transport of certain species across the membrane. Table 1.1 Industrial membrane separation processes [4-6] Membrane separation process Membrane type Driving force Method of separation Range of application Microfiltration Microporous membrane, 0.1 to 10 µm pore radius Pressure difference Sieving mechanism due to pore radius and absorption Sterile filtration clarification Ultrafiltration Microporous membrane, 0.1 to 1 µm pore radius Pressure difference Sieving mechanism Separation of macromolecular solutions Nanofiltration Microporous membrane, 0.01 to 0.1 µm pore Pressure difference Sieving mechanism radius Separation of macromolecular solutions Reverse Osmosis Dialysis Nonporous Pressure difference Solution diffusion mechanism Separation of salts and microsolutes from solutions Separation of salts Microporous membrane Concentration or Diffusion in and microsolutes 0.001 to 0.1 µm pore activity gradient convection free layer from macromolecular radius solutions Electrodialysis Cation and anion exchange membrane Nonporous or microporous Gas Separation Nonporous Pressure or concentration gradient Solution diffusion mechanism Separation of gas mixtures Nonporous Partial pressure gradient Solution diffusion mechanism Separation of close boiling point mixtures and azeotropic mixtures Pervaporation Electrical potential Electrical charge of gradient particle and size Desalting of ionic solutions 2 Among these membrane separation processes, pervaporation is attracting more and more attention due to its energy saving aspects and effectiveness [7] in separating azeotropic mixtures, close boiling point mixtures, isomers and heat-sensitive mixtures. Azeotropic mixtures separation requires special processes such as rectification with entrainer because the same composition at both liquid and vapor phases is not easy to separated by distillation, molecular sieve absorption or liquid-liquid extraction which are expensive and usually involve secondary treatment. Compare to traditional separation processes, pervaporation can effectively break the azeotropes by altering the liquid-vapor phase equilibrium with a selective dense membrane. The “pervaporation” is termed from “permselective evaporation” because of the unique phase change, i.e. the feed liquid changes to permeate vapor across the membrane [89]. Pervaporation is a membrane process that uses membrane as a barrier to separate solvent mixtures containing trace or minor amounts of the component to be removed. The membrane acts as a selective barrier between the two phases, the liquid phase feed and the vapor phase permeate through the membrane preferentially. It allows the desired componen of the liquid feed to transfer through it by evaporates as a lowpressure vapor at the other side of the membrane. Separation of components is based on a difference in transport rate of individual components through the membrane. The driving force for transport of different components is provided by a chemical potential 3 difference between the liquid feed/retentate and vapor permeate at each side of the membrane. Different from other membrane processes, pervaporation has the phase change across the membrane [10-12]. Figure 1.1 illustrates the schematic diagram of typical vacuum pervaporation. Feed Liquid Retentate Permeate Vapor Vacuum pump Figure 1.1 Schematic diagram of vacuum pervaporation 1.2 Recent research progress of pervaporation membranes The concept of “pervaporation” was initially introduced when the fast evaporation of water from aqueous solutions through a collodion bag reported by Kober in 1917 [13]. Farber made the earliest attempt to concentrate protein solution by pervaporation in 1935 [14]. Heisler et al. published a first quantitative study of pervaporation separation of aqueous ethanol mixture by a cellulose membrane in 1956 [15]. Many studies have reviewed the performance and characteristics of membrane materials in various pervaporation applications [11, 16-22]. Based on the recent research progress, three catagories including hydrophilic membranes, organophilic membranes and 4 organoselective membranes are classified according to various materials and applications. 1.2.1 Hydrophilic membranes The broadest industrial application of pervaporation is for the dehydration of organic solvents especially alcohols. Polymers which contain hydrophilic groups such as hydroxyl (-OH), carboxyl (-COOH), carbonyl (-CO) and amino (-NH2) groups are intensively studied for pervaporation dehydration process. • Highly hydrophilic materials Highly hydrophilic materials such as poly(vinyl alcohol) (PVA) and poly(acrylic acid) (PAA) have strong affinity to water but exhibit excessive swelling in aqueous solutions and this leads to drastic loss of selectivity. Cross-linked poly (vinyl alcohol) (PVA) is the most popular material for pervaporation dehydration. Its membranes have been successfully commercialized by GFT (now Sulzer Chemtech) after extensive researches to improve its permselectivity and stability. Crosslinking and grafting increase PVA membranes’ stability and selectivity, however decrease permeability. The crosslinking agents that have been studied are [23,24]: fumaric acid, glutaraldehyde (GA), HCl, citric acid, maleic acid, formic acid, amic acid, sulfursuccinic acid, and formaldehyde. 5 Blending is another economical and effective approach to suppress swelling and to enhance performance. Namboodiri and Vane studied blending of poly (allylamine hydrochloride) (PAAHCl) and PVA for ethanol and isopropanol dehydration, and found that both water flux and selectivity were increased with the addition of PAAHCl and the performance was tunable by varying blend composition and cure conditions [25, 26]. Incorporation of nanoparticles especially zeolite molecular sieves into the polymer matrix is also very promising to improve the physicochemical stability and enhance the separation performance. Despite some trade-offs between permeability and selectivity obtained by the early attempts of embedding zeolite 3A, 4A, 5A and 13X into PVA membranes [27], Guan et al. [24] successfully developed multilayer mixed matrix membranes with PVA and zeolite 3A as the selective layer crosslinked by fumaric acid. Both flux and separation factor for ethanol dehydration were enhanced significantly after the incorporation of zeolite particles. The key factors of making a successful mixed matrix membrane for gas separation are also applicable to the development of pervaporation membranes, i.e., the choices of appropriate polymer and filler, and the controlled interstitial defects between the polymer phase and the zeolite phase [28]. Wang et al. [29] fabricated composite PVA membranes containing delaminated microporous aluminophosphate and showed distinct improvement on flux and separation factor. Guo et al. [30] incorporated γ- glycidyloxypropyltrimethoxysilane (GPTMS) into PVA by an in situ sol-gel method 6 for ethylene glycol (EG) dehydration. The PVA-silica nanocomposite membranes effectively suppressed the swelling of PVA and exhibited desirable stability in aqueous EG solution. Adoor et al. [31] attempted to fabricate mixed matrix membranes (MMMs) containing soldium alginate (NaAlg), PVA and hydrophobic zeolite, i.e., silicalite-1. The incorporation of hydrophobic zeolite particles reduced swelling and led to increased selectivity but decreased permeability. Natural polymeric materials, such as chitosan, alginate and agarose, are abundant in nature, low cost, non-toxic and biodegradable. This group of materials is hydrophilic, but its swelling and instability in water are major problems for dehydration applications. Chitosan, produced from the N-deacetylation of chitin, has gain intensive attention for alcohol dehydration. Various modifications have been carried out to make the chitosan membranes more stable in water and to have better water permselectivity. Cross-linking with hexamethylene diisocyanate (HMDI) [32], glutaraldehyde [33], and sulfuric acid [34] have been investigated. Other modifications include blending with other polymers [35, 36] and incorporation of zeolite particles [37]. Prominent separation performance has been obtained from novel PBI hollow fiber pervaporation membranes [38]. Synthesized from aromatic bis-o-diamines and dicarboxylates, PBI has superior hydrophilic nature and excellent solvent-resistance with robust thermal stability (Tg of 420°C). The brittleness of PBI was successfully 7 overcome in a dual-layer composite form. The as-spun fibers without further crosslinking or heat treatment exhibit good separation performance for dehydration of tetrafluoropropanol (TFP) and isopropanol. • Aromatic polyimides Recently, the development of pervaporation dehydration membranes based on aromatic polyimide has achieved promising results. Aromatic polyimides possess very attractive properties such as superior thermal stability, chemical resistance and mechanical strength. Although polyimides may exhibit instability at high temperatures and high humidity due to the hydrolysis of the imide rings, most polyimides are suitable for the dehydration of organic solvents under moderate conditions [39]. Conventionally, polyimides are synthesized by two-stage polycondensation of aromatic dianhydrides with diamines to form a soluble poly(amic acid), followed by imidization via thermal treatment [40]. The interactions between water molecules and the functional groups of polyimides are through hydrogen bonding (Figure 1.2). The small free volume and rigid polymer backbone contribute to the high water selectivity of polyimide membranes. As a result, without strong hydrophilicity, a heat-treated P84 co-polyimide asymmetric membrane may exhibit very limited swelling even in high water content [41]. 8 H H O H H H O H N O O C H O C Figure 1.2 Interaction of water molecules with imide groups through hydrogen bonding [40] The separation performance of polyimide membranes varies with chemical composition and molecular structure of polymer chains, as well as preparation conditions, and operating conditions [11, 42-46]. Table 1.2 lists the pervaporation performance of recently developed polyimide membranes for pervaporation dehydration of alcohols. Although investigation of inherent membrane properties through dense membranes is essential, it is clear that recent research has moved from dense films to composite or asymmetric membranes because they have more commercial values. 9 10 11 To develop composite membranes with polyimide as the selective layer, various preparation methods have been attempted, such as chemical vapor deposition and polymerization (CVDP), dip coating, and cataphoretic electrodeposition. These methods have the advantages to produce a thin polyimide selective layer with the drawback of increasing the complexity of the membrane fabrication process. On the other hand, the as-fabricated or as- spun asymmetric polyimide membranes often show high flux but low separation factor due to defects in the selective skin layer [1, 41]. Post-treatments such as heat treatment and crosslinking have been conducted to reduce defects and enhance the separation property of polyimide membranes [41, 47, 48]. In general, heat treatment is easy to operate and effective for many materials such as polyimide, PBI, polysulfone and polyacrylonitrile (PAN) to reduce pore size and improve selectivity because heat treatment induces molecule relaxation and polymer chain repacking [49,50]. For example, Yanagishita et al. found that heat treatment of polyimide membranes at 300°C for 3hr increased mechanical strength and separation factor for ethanol dehydration [47]. Qiao et al. observed noticeably smoothed surface roughness, reduced d-space and densified skin layer structure of P84 co-polyimide flat asymmetric membranes at heat treatment temperatures above 200°C [41]. Liu et al. demonstrated the application of heat treatment to P84 hollow fibers and obtained much superior separation performance to those asymmetric flat sheet membranes [1]. 12 The significant performance enhancement at a heat treatment temperature lower than the polymer’s Tg, i.e., around 200°C, may be attributed to the local or segmental motions of polymer chains at β transition, as suggested by Zhou and Koros [51]. In addition, the segmental motions of polymer chains at a temperature above β transition enhance the formation of charge transfer complexes (CTCs) through their inherent electron donor (the diamine moiety) and electron acceptor (the dianhydride moiety) elements. The CTCs formation strongly depends upon heat-treatment temperature, i.e., the higher the heat treatment temperature, the more CTCs can be formed [52]. The intra- and inter-chain CTCs restrict the polymer chain mobility and act as crosslinking. The formation of CTCs can be characterized by both fluorescence and UV-vis spectrophotometer [53-56]. Adopted from gas separation membranes made from polyimide [56-58], chemical crosslinking has been proved as another economical and effective tool which can tune the pervaporation performance of polyimide membranes with or without the aid of thermal treatment. The modification of P84 polyimide asymmetric membranes with diamines for isopropanol dehydration was firstly investigated by Qiao and Chung [59]. With the introduction of amide groups after modification, P84 co-polyimide membranes exhibited higher hydrophilicity and apparently denser skin structure. There existed an optimum degree of crosslinking where separation factor achieved a maximum point then degraded; this was attributed to the increased hydrophilicity which caused excessive swelling. It was found that thermal treatment after 13 crosslinking also affected membranes’ property and performance. A low-temperature heat treatment facilitated the crosslinking reaction, while a high-temperature heat treatment caused the reaction reversed. The separation factor was further enhanced after heat treatment with a loss in flux. Jiang et al. [60] demonstrated that chemical crosslinking by 1,3-propane diamine (PDA) for Matrimid® hollow fibers apparently improved membrane selectivity in water/isopropanol separation. In addition, a thermal pretreatment followed by chemical crosslinking was found effective in revitalizing and enhancing the membrane performance regardless the initial status of the hollow fiber (e.g. defective or defective free). However, extensive experimental data have revealed that the effectiveness of diamine modification varies significantly with diamines chemistry and structure, polyimide moieties and chain structure, and the pre- or post-heat treatment conditions. Therefore, one must take these factors into consideration when conducting the modification. Other modification methods, e.g. blending with highly hydrophilic materials [48] , incorporation of zeolite molecular sieves [28] and inorganic nanoparticles [61], have also shown effectiveness in performance enhancement of polyimide membranes for the dehydration of organic solvents. Interestingly, the swell-up of polymer chains in the feed solution makes the adverse effect of interstitial defects between the polymer matrix and the inorganic particles much less significant compared to that in gas 14 separation membranes [59]. However, so far these mixed matrix modifications are only demonstrated in dense films; it will be more interesting and challenging if the currently developed knowledge can be extended to fabricate membranes with a composite or asymmetric structure. In addition, the high temperature required in the fabrication process of polyimide mixed matrix membranes probably can be reduced by the introduction of crosslinking agent and the modification of surface properties of inorganic particles, which may bring down the processing cost to achieve a good interaction between the filler particles and the polymer matrix. • Hydrophobic materials Hydrophobic materials have higher stability in aqueous solutions. If the hydrophobic nature of the material can be changed to hydrophilic and the degree of modification can be controlled, this type material can become a good candidate for pervaporation dehydration. For example, poly(ether ether ketone) (PEEK) had been modified by sulfonation reaction for pervaporation separation of water/isopropanol mixtures and the hydrophilicity-hydrophobicity balance was controlled by different degrees of sulfonation [62]. Tu et al. [63] developed hydrophilic surface-grafted poly(tetrafluoroethylene) (PTFE) membranes with good performance and wide applications in pervaporation dehydration processes. • Inorganic materials 15 Inorganic membranes are able to overcome the problems of instability and swelling of hydrophilic polymeric membranes. They show better structural stability and chemical resistance at harsh environments and high temperature operations [64-66]. Zeolite membranes have the advantages of high selectivity and high permeability due to their unique molecular sieving property and selective adsorption. The recently developed zeolite NaA, X and Y membranes exhibit impressive separation performance that is far superior to traditional polymeric membranes. The high separation factor is achieved because of the precise micropore structure of zeolite pores and the preferential sorption of water molecules. Microporous silica membranes are water selective and exhibit a much higher flux and less swelling but lower selectivity compared to polymeric membranes [67, 68]. Ceramic membranes are resistant to microbes; they can be easily sterilized by steam or autoclave. Ceramic membranes show high water permeation flux and relatively high separation factor for alcohol dehydration [65]. The major drawbacks of inorganic membranes are (1) the higher cost of fabrication process compared to that of polymeric membranes and (2) the brittleness. However, the superior stability and higher separation performance may level off the initial fabrication and installation cost of inorganic membranes. The performance of recently developed inorganic membranes is summarized in Table 1.3. It is obvious that the preparation procedure also plays an important role on membrane performance. By lowering the transport resistance of the support layer and minimize 16 the selective layer thickness, Sato and Nakane [69] developed NaA zeolite membranes with very high flux and comparable water/alcohol separation factor. 17 1.2.2 Organophilic membranes In organo-selective membranes for the separation of small amount of organics from water, the difference in solubility determines the membrane selectivity. This is because diffusivity always favors the smaller molecule, i.e., water. Membranes made from rubbery polymers such as poly(dimethyl siloxane) (PDMS) [70, 71], polyurethane [72], polybutadiene [73], polyamide-polyether block copolymers (PEBA) [16] and poly[1-(trimethylsilyl)-1-propyne] (PTMSP) [74, 75], and hydrophobic inorganic materials such as zeolite silicalite-1 and ZSM-5, have been intensively investigated for the separation of organics from aqueous streams. PDMS are currently the benchmark material for this application because of its high affinity and low transport resistance for organics, and stability in organic solutions [22]. It has been pointed out that the most important factor to advance organophilic pervaporation is to have breakthroughs in membrane materials and structure in addition to minimizing concentration polarization, optimizing the process, and improving energy efficiency [22, 76, 77]. The following approaches have been taken: (1) modification of currently available membranes by crosslinking, grafting or incorporation of adsorbent fillers, (2) development of novel membrane structures, and (3) development of new polymeric materials. For example, Uragami et al. [78] crosslinked PDMS membranes with divinyl compound and found both permeability and benzene permselectivity of the membranes were improved. A novel polymeric- 18 inorganic composite membrane made by coating cellulose acetate upon a tubular ceramic support was firstly developed by Song and Hong [79] for the dehydration of ethanol and isopropanol. Later this approach was adapted to coat a PDMS layer on top of a ceramic tubular support to extract ethanol from water [71, 80]. Using crosslinked PDMS as the selective layer and tubular non-symmetric ZrO2/Al2O3 membranes as the support layer, Xiangli et al. [80] developed composite membranes with remarkably high flux (i.e., flux of 12300g/m2hr and separation factor of 6 for a feed ethanol concentration of 4.3wt% and temperature at 40°C). This performance is superior to the performance of PDMS composite membranes with a polymeric support, owing to the significantly reduced transport resistance of the ceramic support layer. Recently, Nagase et al. [81] synthesized siloxane-grafted poly(amide-imide) and polyamide with a new reactive diamino-terminated PDMS macromonomer. The newly developed material exhibited durability and good permselectivity toward several organic solvents with high permeation rates and reasonable separation factors (i.e., flux of 37.4g/m2hr and separation factor of 9.78 for a feed ethanol concentration of 9.24wt% and temperature at 50°C). Table 1.4 summarizes the recent development of PDMS membranes. Except for benzene removal, the separation factors for alcohol removal are all below 100, which inhibit industrial scale applications. 19 20 1.2.3 Organoselective membranes Albeit of the great potential in chemical and petrochemical industries, the separation of organic/organic mixtures using pervaporation is the least developed area. There are wide streams of organic/organic mixtures and basically these mixtures can be categorized into four major groups: polar/non-polar, aromatic/aliphatic, aromatic/alicyclic and isomers. Smitha et al. [21] have given a good literature summary on membrane materials and their performance for the above four aspects, Villaluenga and Tabe-Mohammadi [18] gave a deeper insight on the membranes developed for benzene and cyclohexane separation. Membrane materials are selected based on the solubility differences of organic components in membrane. By improving the interaction between membrane material and one permeating component, the separation performance can be enhanced. Among the diversified applications in organic/organic separation by pervaporation, the separation of benzene/cyclohexane represents one of the most important but most difficult and complicated separation in petrochemical industry. The double bonds of benzene molecule have strong affinity to polar groups in a membrane; therefore hydrophilic membranes which possess polar groups such as PVA and benzoylchitosan show selectivity to benzene [82-83]. Benefit from the conjugated π bonds, graphite, carbon molecular sieve and carbon nanotube show preference to aromatics with effective π-π stacking interaction [84-85]. These inorganic materials have been used to 21 enhance PVA performance. Crystalline flake graphite was firstly incorporated into PVA or PVA/Chitosan blend membranes and resulted in significant increase of permeation flux and selectivity. Later carbon nanotubes with or without wrapped with chitosan were introduced to the PVA matrix [84-85]. The improvement in permeation flux and separation factor were attributed to the preferential affinity of carbon nanotubes towards benzene and the increased free volume by altering PVA polymer chain packing. Nam and Dorgan et al. [86] attempted modification of the solubility selectivity of glassy polymer polyvinylchloride by physical blend with crosslinked rubbery materials; and the resultant membranes showed permselectivity toward benzene. Table 1.5 summarizes the recently developed membranes for benzene/cyclohexane separation. 22 23 1.3 Industrial applications and commercial aspects The applications of pervaporation processes are mainly divided into three areas: (1) dehydration of alcohols or other aqueous organic mixtures; (2) removal of volatile organics from water; (3) organic/organic separation. Dehydration of organic solvents such as alcohols, esters, ethers, and acids has become the most important application of pervaporation due to the high demand in industries and the difficulties to obtain the anhydrous form of these chemicals by traditional distillation technology. Both diffusion and sorption selectivity of water over organic solvents can be simultaneously sought by hydrophilic pervaporation membranes because water has smaller molecular size and stronger affinity to hydrophilic materials than the organic solvents. The first commercial membrane which consisted of a dense cross-linked PVA as the selective layer, an ultrafiltration poly(acrylonitrile) and a fabric non-woven as the support layer was developed by Gesellschart für Trenntechnik (GFT, now belongs to Sulzer Chemtech) in 1980s for the dehydration of ethanol. Since then, 38 solvent dehydration plants for ethanol and isopropanol, 8 units for other solvents dehydration (i.e. ester) have been installed world widely [87]. There are also numbers of attempts to employ pervaporation for organics removal from water which aim on water purification, pollution control, solvents/aroma compounds recovery and biofuel production from fermentation broth. Applications in 24 this area include removal of trace amount of volatile organic compounds (VOCs) from aqueous streams. The emission of VOCs from industrial and municipal wastewater streams are of great concern due to the toxic and carcinogenic effects of VOCs. VOCs include solvents from petroleum industry, such as benzene, toluene and xylenes, and substances which contain chlorine, such as chloroform, 1,1,2-trichloroethane (TCA), trichloroethylene (TCE), perchloroethylene, and chlorobenzene. Due to the low solubilities of these compounds in water, the amount of these compounds dissolved in wastewater is very small; therefore treatment by distillation is not economically viable [19]. Traditionally, carbon adsorption and air stripping were employed as treatment processes; however, these treatments merely transfer the contaminant from water phase to another phase and further treatment is necessary. In addition, the regeneration of activate carbon is costly. Pervaporation is promising for VOCs removal or recovery by achieving the separation through preferential sorption of one component in the membrane without disruption of the process. If the concentration of the organic is sufficiently high in an aqueous stream, the recovery of organics is valuable. It has been demonstrated that a stream containing 2% ethyl acetate was concentrated to 96.7%, which was reused in the feed stream [88]. Combining with a fermentation process, pervaporation is applied to extract inhibitory products such as ethanol, butanol, and isopropanol from a fermentation broth in order to increase the conversion rate [6]. As crude oil price reaches new highs every year, pervaporation become promising for biofuel (e.g., bio-ethanol and bio-butanol) 25 recovery from fermentation broth [77, 89]. However, the flux and separation factor of membranes for separating organics from water by pervaporation are still lower than those for organic solvent dehydration, which seriously restrict the industrial-scale application of organophilic pervaporation [20]. Till 2002, only a few pervaporation systems for VOC removal were commercialized by MTR [87]. Organic/organic separation by pervaporation has large potential applications in chemical, petrochemical and pharmaceutical industries. The applications include the separation of polar/non-polar mixtures, e.g. methanol/methyl tert-butyl ether (MTBE) [90, 91]; aromatic/aliphatics, e.g. cyclohexane/benzene [18, 92]; aliphatic hydrocarbons, e.g. hexane/heptane [93]; isomers, e.g. C8 isomers (o-xylene, m-xylene, p-xylene and ethyl benzene) [94-96]; and enatioseparation, e.g. linalool racemic mixture separation [97]. Research in this area is extremely challenging; nevertheless, the application of pervaporation in organic/organic separation has not acquired industrial acceptance because of the lack of advanced performance and the instability in organic solvents of currently available membranes [21]. The first and only pervaporation plant using organoselective membranes was built by Air Products in 1991 for removal of methanol from MTBE. The commercialization of pervaporation was started in Europe by GFT (now Sulzer). Nowadays, main global pervaporation membrane manufacturers and suppliers are: Sulzer Chemtech (Swiss), CM-Celfa Membrantrenntechnik (Swiss), GKSS 26 (Germany), UBE Industries & Mitsui Engineering and Shipbuilding Co. Ltd (Japan), Daicel Chemical Industries, Ltd (Japan), Membrane Technology and Research, Inc. (USA), and MegaVision Membrane Technology and Engineering Co Ltd, (China). 1.4 Research Objectives and Organization of Dissertation It was found that previous research mainly focused on flat sheet dense or asymmetric membranes, however the research is lacking in the area of hollow fiber membranes especially as pervaporation dehydration membranes. The development of novel hollow fiber pervaporation membranes is therefore a main objective of this study. BTDA-TDI/MDI (P84) co-polyimide dense and asymmetric membranes have shown good permeability and selectivity for pervaporation dehydration of alcohols. This research study intends to extend previous research on BTDA-TDI/MDI (P84) copolyimide dense / asymmetric membranes to hollow fiber membranes. It comprises the understanding of pervaporation transport process, the development of novel hollow fiber pervaporation membranes based on BTDA-TDI/MDI (P84) co-polyimide, and the investigation of the effects of spinning conditions and modifications on membrane performance. This study selects IPA as a model solvent because of its high market value. The objectives of this study are: 1. To develop BTDA-TDI/MDI (P84) asymmetric hollow fiber membranes, and investigate the effects of various spinning conditions and post treatments including 27 silicone rubber coating and heat treatment on P84 hollow fiber membranes morphology and performance of pervaporation dehydration of IPA. 2. To develop BTDA-TDI/MDI (P84) / PES dual-layer hollow fiber membranes using a triple-orifice dual-layer spinneret via co-extrusion phase inversion process, and investigate the effects of various spinning conditions and post treatments including heat treatment and and the p-xylenediamine cross-linking modification on dual-layer hollow fiber membranes morphology and performance of pervaporation dehydration of IPA. This dissertation is organized and structured into six (6) chapters. Chapter 1 gives a general introduction to pervaporation membrane separation processes and current development. The research objectives and outline are also included in this chapter. Chapter 2 provides the literature review on theoretical background of the pervaporation transport mechanism, the formation mechanism of phase inversion membranes and the effects of important factors on membrane properties and pervaporation performance. Chapter 3 summarizes the experimental materials, methodologies and membrane preparation methods. Membrane characterization methods are also reported in this chapter. 28 Chapter 4 describes the development of BTDA-TDI/MDI (P84) asymmetric hollow fiber membranes, and investigates the effects of various spinning conditions and post treatments including silicone rubber coating and heat treatment on P84 hollow fiber membranes morphology and performance of pervaporation dehydration of IPA. Chapter 5 describes BTDA-TDI/MDI (P84) / PES dual-layer hollow fiber membranes using a triple-orifice dual-layer spinneret via co-extrusion phase inversion process, and investigates the effects of various spinning conditions and post treatments including heat treatment and and the p-xylenediamine cross-linking modification on dual-layer hollow fiber membranes morphology and performance of pervaporation dehydration of IPA. Chapter 6 summarizes the general conclusions drawn from this research works. 29 CHAPTER TWO THEORETICAL BACKGROUND 2.1 Fundamentals of pervaporation separation process The performance of a pervaporation membrane is dependent on the membrane materials, the structure of the membrane, and the interactions between permeantpermeant and permeant-membrane. The complex interactions between permeants and membrane make it difficult to present a comprehensive model to depict the transport process. Two major models, namely, solution-diffusion model and pore flow model, have been developed to illustrate transport mechanism in pervaporation process. The solution-diffusion model which has been widely adopted by most of pervaporation membrane researchers due to its good agreement between theory and experiments [11, 98] will be discussed in the following section. 2.1.1 Transport Mechanisms - Solution-diffusion model The mass transfer in pervaporation membrane is described as a three-step process: (i) the permeant is dissolved in the feed side of the membrane; (ii) the permeant diffuses though the membrane; and (iii) the permeant evaporates as vapor at the downstream side of the membrane. Figure 2.1 illustrates the solution-diffusion model. For a 30 pervaporation process, the transportation rates of molecules from the bulk feed liquid mixture to the membrane surface and the removal of vapors at the downstream side also play important roles to the overall mass transport. The former may affect concentration polarization, while the latter influences the productivity. Membrane Sorption Evaporation Diffusion Feed liquid Permeate vapor Figure 2.1 Schematic diagram of solution-diffusion model The solution-diffusion model is applicable to non-porous polymeric membranes in which the transport of permeating molecules relies on the thermally agitated motion of polymer chain segments. Based on the solution-diffusion model, the permeability coefficient P is given by the product of diffusivity D and solubility S: P=D S (2.1) For a binary pervaporation system consisting of components A and B, the diffusion selectivity (DA/DB) depends greatly on (1) the penetrant size and shape, (2) the 31 mobility of polymer chains, (3) the interstitial space between polymer chains, and (4) the interactions between penetrants and between penetrant and membrane material [3]. The sorption selectivity (SA/SB) prefers more condensable molecules or molecules which have special interaction with membrane materials [10]. The driving force for mass transport through a pervaporation membrane is the chemical potential gradient, i.e. partial pressure gradient (fugacity). The transport equation based on solution-diffusion mechanism for pervaporation can be derived as follows [3, 98]. The flux Ji of one component is proportional to its driving force, i.e. the chemical potential gradient across the membrane J i = − Li dµ i dx (2.2) Under isothermal conditions (constant T), the chemical potential in pressure and concentration driven processes is µ i = µ i0 + RT ln(ci ) + vi ( p − pi0 ) (2.3) The chemical potential gradient dµ i is given as dµ i = RT d ln(ci ) + vi d p (2.4) where vi is the molar volume. The pressure within the membrane is assumed to be constant in the solution-diffusion model; therefore combining Eqns. (2.2) and (2.4) gives Ji = − RTLi dci ci dx (2.5) 32 If RTLi/ci is replaced by diffusion coefficient Di, the integration of Eqn. (2.5) across the membrane provides Ji = Di (cif( m ) − cip( m ) ) l (2.6) where l is the membrane thickness, ci(m) represents the concentration of i component inside the membrane, and the superscripts f and p represent the feed and permeate side, respectively. By assuming the chemical potential equilibrium at the liquid mixture/membrane feed interface and a hypothetical vapor state in equilibrium with the liquid mixture, one can obtain cif( m ) = γ i f,G γ i f, L ⋅ pis p if = K i ⋅ p i f (2.7) where γ is the activity coefficient, p s is the saturation vapor pressure, and the subscripts G and L represent the gas and liquid phase, respectively. Similarly, the equilibrium at the permeate gas/permeate membrane surface gives cip( m) = γ ip γ ip ⋅ pis pip = S i ⋅ pip (2.8) where Si is defined as the sorption coefficient. Finally, Equation (2.6) can be rewritten as follows: Di S i ( p if − pip ) Pi ( pif − pip ) Ji = = l l (2.9) 33 where Pi, the membrane permeability as a product of diffusion coefficient (Di) and sorption coefficient (Si), can be obtained from the above equation. Conventionally, Di and Si are considered to be constant. In pervaporation separating liquid mixtures, the membrane may often be seriously swollen due to the much complicated and strong physicochemical interaction between the highly condensable permeant molecules and the membrane material. This would lead to the changes of sorption characteristics and diffusion properties in the membrane [99]. Different from air separation membranes where the selectivity obtained from mixed gases is not much different from the pure gas tests, the selectivity of pervaporation for a specific mixture is not only far lower than the permeability ratio of the pure components but also varies as a function of the feed composition. 2.1.2 Performance parameters: flux and separation factor The performance of a pervaporation membrane is typically characterized by flux (J) and separation factor (α), as defined by the following equations: J= Q At α 2 /1 = (2.10) y 2 / y1 x2 / x1 (2.11) where, Q is the total mass transferred over time t, A is the membrane area, x2 and y2 are the mole fractions of one component in the feed and permeate, respectively, and 34 x1 and y1 are the mole fractions of the other component in the feed and permeate, respectively. Flux is obtained from the amount of permeant collected from a laboratory setup at a certain time interval divided by membrane area, as defined in Eqn. (2.10); while separation factor is defined as the concentration ratio of two components in a binary system, as defined in Eqn. (2.11). For very dilute feed solutions, an enrichment factor is often used to represent membrane selectivity which is the ratio of concentrations of the preferentially permeating component in the permeate and in the feed, respectively [100]. Because of the existence of a trade-off relationship between flux and separation factor, that is, the flux and separation factor usually perform in the opposite way, Huang and his coworkers [101] introduced pervaporation separation index (PSI) to evaluate the overall performance of a membrane. PSI was originally defined as a product of permeation flux and separation factor: PSI = J t ⋅ α (2.12) Where Jt is the total permeation flux, and α is the separation factor. However, in this definition, the PSI can be large if the membrane has a high flux even when α is equal to 1. Therefore, the definition of PSI was later modified as a product of Jt and (α – 1) [102]. Flux and separation factor are direct performance data which are easy to use and compare. However, from the standpoint of investigating intrinsic membrane 35 properties (e.g. solution-diffusion model), these two parameters are not the pertinent guidelines for membrane materials comparison and membrane development. This is due to the fact that these two parameters combine both membrane properties and operating conditions into the calculation of pervaporation performance, thus one cannot discern the true and individual effects of operating conditions and membrane’s intrinsic properties on system performance. Therefore, this intermingled effect creates difficulties for membrane scientists to compare membrane materials for pervaporation because no intrinsic membrane properties can be derived from the data of flux and separation factor. 2.1.3 Performance parameters: permeance and selectivity Wijmans and Baker [103] were the pioneers proposing the use of permeance and selectivity instead of flux and separation factor to investigate pervaporation membrane performance and properties. Recently, Wijmans [104] re-emphasized the importance of using permeance and selectivity, while Guo et al. [105] and Qiao et al. [106] not only elaborated their differences but also gave detailed examples on the comparison of flux vs. permeance and separation factor vs. selectivity for performance interpretation. Equation (2.9) gives the relationship between flux, permeability, and driving force of vapor pressure. The partial vapor pressure (fugacity) of each component on the feed 36 side in this equation can be calculated based on its concentration in the feed liquid mixture as follows: pif = xi γ i pis (2.13) where the superscript s indicates the saturated state and γ is the activity coefficient. pis and γi can be obtained from the vapor-liquid equilibrium data with the aid of the HYSYS DISTIL software (Hyprotech Ltd, Canada) or the Antoine equation and Wilson equation, respectively. By substituting Equation (2.13) to Equation (2.9), one can obtain permeance ( Pi ) based on the following equation Pi = Pi / l = J i /( xi γ i pis − pip ) (2.14) Permeance is more convenient for an asymmetric/composite membrane where the dense selective layer thickness is not readily available, while permeability (P) is usually related to a dense membrane. Both are direct indicators of the intrinsic properties of a membrane, and can be determined directly from experiments with the help of the above equations. The membrane selectivity is defined as the ratio of permeability or permeance of two permeating components. Through investigating the dehydration of aqueous butanol mixtures through PVA membranes, Guo et al.[105] found that using permeance and selectivity could clarify and quantify the contribution by the nature of the membrane to the separation performance. For example, for the dehydration of aqueous butanol mixtures, water flux increased rapidly with increasing feed temperature; however, their corresponding 37 water permeance showed an opposite trend. Traditionally, the increased water flux at a higher temperature was explained by the increased thermal motion of polymer chains and the expansion of the free volume. However, the declining trend of permeance on temperature revealed that the driving force also played an important role. In addition, the negative temperature effect on sorption should also be accounted. Furthermore, Qiao et al. [106] investigated the dehydration of isopropanol and butanol. In contrast to the results obtained from water flux, a comparison of water permeances from different alcohol systems revealed that the mass transport of water inside the membrane was actually not strongly affected by the different alcohols. Similarly, based on permeance, the important physicochemical properties which influenced alcohol transport were mainly attributed to molecular linearity and their affinity to water and PVA as reflected by solubility and polarity parameters. In addition, the separation factor versus feed water content plots may mislead the analysis of water influence on membrane performance and exaggerate the plasticization phenomenon. The insight information would be very hard to be discovered if one only examines membrane performance by flux and separation factor. To give an example, the following equation describes the relation between separation factor (α) and selectivity (β) for water to ISOPROPANOL in the dehydration of an aqueous ISOPROPANOL system when the permeate side pressure p p is very small and negligible [106]: 38 α 2 /1 = J 2' / J 1' x 2 / x1 = x1 J 2 / 18 x P (γ x p s − p 2p ) γ 2 p 2s P 2 γ 2 p 2s = 3.3 1 2 2 2 s2 = 3 . 3 = 3 . 3 β 2 /1 s x 2 J 1 / 60 x 2 P1 (γ 1 x1 p1 − p1p ) γ 1 p1s P1 γ 1 p1 (2.15) where J is the mass flux and P is the permeance based on mass, and subscript 1 and 2 represent isopropanol and water respectively. 60/18 = 3.3 is the ratio of molecular weights of isopropanol to water. Here β is the mass-based membrane selectivity. If mole is the base for calculation, the above equation can be rewritten as follows: α 2 /1 = J 2' / J 1' x 2 / x1 = s p ' x1 P2 (γ 2 x 2 p 2 − y 2 p ) x1 P2'γ 2 x 2 p 2s P2'γ 2 p 2s γ ps = = = β 2' / 1 2 2s x 2 P1' (γ 1 x1 p1s − y1 p p ) x 2 P1'γ 1 x1 p1s γ 1 p1 P1'γ 1 p1s (2.16) where J ' is the molar flux and P ' is the permeance based on mole. x and y are the mole fractions at the feed and permeate side, respectively. β’, the mole-based membrane selectivity, is defined as β 2' / 1 = P2' / P1' (2.17) Obviously, the mole-based selectivity and separation factor are related by a term of γ 2 P2s / γ 1 P1s . Table 2.1 lists the calculation of the ratio γ 2 P2s / γ 1 P1s for a solution containing 85wt% (63mol %) isopropanol in water. The data in this table illustrates that the difference between membrane selectivity and separation factor increases with increasing temperature. Furthermore, since γ 2 / γ 1 is a concentration dependent term, 39 the membrane selectivity as defined in Eqn. (2.17) shows less dependence on feed water content compared to the separation factor. 2.2 Factors influencing pervaporation membrane performance It is self-evident that membrane separation performance is dependent on the membrane materials, the structure of the membrane, and the interactions between permeant-permeant and permeant-membrane. These factors are crucial for 40 understanding the pervaporation separation process and are critical for pervaporation material screening. 2.2.1 Interaction between permeant and membrane As aforementioned, membrane performance is mathematically determined by selective sorption and/or selective diffusion. However, these two theoretical parameters are actually not constant during the separation process and may vary significantly with physicochemical properties of membranes and penetrants. The most important factor which determines the selective sorption is the interaction between penetrant and membrane. This interaction can be qualitatively described and analyzed from three aspects; namely, solubility parameter approach, polarity parameter approach, and Flory-Huggins interaction parameter approach. Solubility parameter represents the nature and magnitude of the interaction force between different molecules. It is frequently used to qualitatively describe the affinity between solvent and polymer, the interactions between polymer and permeant, and permeant and permeant [107-109]. If a solvent has a solubility parameter close to that of a polymer, it is regarded as a good solvent for the polymer. The solubility parameter is calculated by Equation (2.18) δ sp2 = δ d2 + δ p2 + δ h2 (2.18) 41 where δd, δp, and δh represent dispersion force, polar force and hydrogen bonding component of the solubility parameter, respectively. The solubility parameters of solvents are available in literature [110], while the solubility parameters of polymers can be estimated by group contribution methods [108]. The Flory-Huggins interaction parameter describes the sorption of a penetrant in a polymer material or the thermodynamic interaction between a solvent and a polymer [111]. Although the assumption that the interaction parameter χ is a constant over the entire activity range remains questionable, it can be determined in a convenient way by a single sorption experiment in a pure liquid according to the following equation [112]: χ =− ln φ + (1 − φ ) (1 − φ ) 2 (2.19) where φ is the penetrant volume fraction in the membrane. A lower interaction parameter indicates a higher affinity between polymer and penetrant. Polarities of the membrane and penetrants also have strong relation to permeant transport. Thereby polarity parameters (ET) are used to describe polymer/liquid interactions as well. Yoshikawa et al. [47] investigated effects of polymer polarity on pervaporation separation performance for ethanol/water mixtures. When the polarity parameter of the membrane deviated from that of water, the selectivity of water decreased. The polarity parameter is particularly useful when a polymer structure is not available. Jonquieres et al. [113] showed a linear relationship between sorption 42 and liquid solubility parameters. Later they found a good correlation of flux and polarity parameter of alcohols for the pervaporation of alcohol/ETBE mixtures [114]. For a given solvent, the definition of polarity parameters ET(30) is given as the transition energy for the longest wavelength visible absorption band of the pyridinium-N-phenoxide betaine called Reichardt’s dye. The polarity parameters for more than 270 organic solvents are collected by Reichardt [115]. However, the real affinity and complicated interaction between the feed component and the membrane may be beyond the prediction by means of solubility parameter, polarity parameter and Flory-Huggins parameter because of the evolution of membrane structure with time during the tests. For example, highly hydrophilic materials such as polyvinyl alcohol (PVA), polyacrylic acid (PAA), agarose, alginate and chitosan may swell excessively and exhibit poor selectivity in aqueous organic mixtures due to the strong interaction between the feed water and the materials. Therefore, one needs to maintain a proper balance of hydrophilicity and hydrophobicity for the development of pervaporation dehydration membranes. Another example is that the physical structure of polyimide membranes may be influenced by operating conditions with the sorption of water and organic solvents, and thus the pervaporation performance is affected. Guo and Chung [46] demonstrated that the Matrimid® polyimide membranes had performance change after a thermal cycle process, due to the interactions between feed molecules and membrane, the plasticization of membrane and the non-equilibrium nature of the dense-selective skin 43 in an asymmetric membrane. It is therefore important to identify a proper conditioning procedure to maintain a better and stable performance. 2.2.2 Interactions between permeant and permeant Not only the interaction between permeant and membranes affects pervaporation performance, the interaction between permeant and permeant also influences their transport properties. This is so-called the “coupling effect” which has often been observed in the mass transfer through pervaporation membranes. Thermodynamically, the interactions between permeant molecules always exist. For example, the permeate molecules could form clusters in the hydrophilic environment in a membrane whereby induced thermodynamic coupling [62, 116]. On the other hand, the mutual drag between the diffusion of permeant molecules induces kinetic coupling. According to Drioli et al. [117], the dissolved permeant component may cause change in polymer free volume and membrane morphology, and thus result in facilitated or inhibited transport of other component through the membrane. Schaetzel et al. [118] investigated dehydration of aqueous ethanol solution through a PVA-PAA-comaleic acid membrane and suggested that coupling effect for the flux of ethanol (the slow permeant) was more significant than that for the flux of water (the fast permeant). Qiao et al. [119] observed a strong coupling between isopropanol and water through a Sulzer PVA membrane and illustrated that the degree of coupling not only depended on the membrane properties, such as the degree of cross-linking, affinity to water and 44 structure responses on temperature rise, but also relied on the physicochemical properties of penetrants, such as molecular linearity (or the aspect ratio) of penetrant molecules, and their solubility parameters and polarity parameters. 2.2.3 Kinetic aspect According to the solution-diffusion mechanism, high selectivity can be obtained if one component has strong affinity to the membrane and can diffuse faster in the membrane than other components. From the kinetic aspect, the movements of molecules from the upstream to the downstream side depend on the interstitial space created by the thermally agitated motion of polymer chains to allow molecule jump from one site to another site [11, 120]. As a result, the polymer chain stiffness and packing, which are determined by the monomer composition, chain configuration and functional groups, also affect the performance of a pervaporation membrane. The chain stiffness tends to hinder molecule diffusion and lowers permeability; while polymer chains consists of a larger interstitial space and a bigger free volume tends to have a higher permeation rate. Polyimide and polyamide are typical examples which show high selectivity but low permeability because of their rigid chain backbones and low free volumes. For this kind of materials, a high diffusion selectivity of water over other organic solvent can be achieved by tailoring the packing density of polymer chains. Peng et al.[121,122] demonstrated that the fractional free volume of PVA membrane can be 45 increased and the free volume cavity size can be tuned by the incorporation of graphite and carbon nanotube into the PVA matrix. The thermodynamic and kinetic influences can be potentially elucidated with the aid of molecular simulation and modeling. The molecular simulation may be able to predict the membrane performance of a designed material for a specific application and therefore reduce the tedious experimental testing during membrane development. The molecular dynamic simulation can be carried out by Materials Studio (Discover and Amorphous Cell modulus) or other software. For example, Peng et al. [84,85] employed Material Studio to predict the fractional free volume change of the PVA membranes after various modifications. They also illustrated that the simulation results provided good explanations for the observation from experiments. Adoor et al. [31] observed greater affinity between silicalite-1 particles and the sodium alginate (NaAlg) polymer matrix than that between the particles with the PVA polymer matrix. This observation was later confirmed by the greater interaction energy between NaAlg and zeolite calculated using a COMPASS (condensed-phase optimized molecular potentials for atomistic simulation studies) force field. A more rigorous and detailed thermodynamic model using the UNIFAQ-FV group contribution calculation was also proposed to study polymer blend formulations and their membrane performance [86]. However, there still lacks a universal model that can quantitatively predict the actual membrane performance for pervaporation because of the complicated thermodynamic interactions and sophisticated mass transports in pervaporation membranes. 46 2.3 Structure and formation of Pervaporation Membranes The membrane in a pervaporation process has to differentiate liquid molecules in angstrom scale (Å). For example, the difference in kinetic diameters between a water molecule (3.0Å) and an isopropanol molecule (4.7Å) is only 1.7Å. Although the difference is much bigger than the size difference between O2 and N2 molecules (kinetic diameter: 3.46 vs. 3.64Å) in gas separation, it is still necessary to use a dense membrane to achieve the separation. In general, in order to develop a suitable membrane for pervaporation, one should screen and choose appropriate materials to construct the crucial dense-selective layer which not only have superior separation performance (i.e., high flux and separation factor), but also possess good mechanical strength, chemical resistance, thermal stability, and minimal solvent-induced swelling [2, 21]. Koros et al. [123] illustrated the fundamental factors controlling material selection for polymeric membranes used for solution-diffusion based permeation separations. An entire hierarchy including four structure levels was summarized. That is, the chemical composition of the polymer which construct the selective layer, the steric relationship of polymer chains in the selective skin, the morphology of the selective skin and the structure integrity between skin and porous support. They also pointed out that because operation at high temperatures was desirable in pervaporation [77], the durability of the membrane material and the sealing material of modules presented great challenges. 47 Usually, dense membranes are prepared to obtain the intrinsic sorption and transport property of a membrane material. If the separation performance generated from dense membranes meets the requirements, asymmetric or composite membranes made of the same polymeric material will be developed. The development of asymmetric or composite membranes consisting of an ultra-thin dense selective layer represents the real interest in industrial applications. However, it is not an easy task because the resultant membranes may perform differently from its dense films as well as from each other if the process parameters are changed. A composite membrane consists of a dense selective layer and a substrate made from different materials. Its flux (or permeance), separation factor (or selectivity) and membrane mechanical strength may be a combination of different layers [10]. Composite membranes are typically prepared by first casting the microporous support, followed by deposition, coating or in situ polymerization of the selective dense layer on top of the support. Ideally, composite membranes are capable to combine the properties of different materials and offer a higher permeation rate than dense films due to the reduced thickness of the selective layer. However, the resistance and the separation properties of the sublayer may interfere with the overall membrane performance or even dominate the separation process [124]. Similar to gas separation membranes, a resistance model has been derived for pervaporation membranes to describe the relationship between the overall permeation resistance and the resistances 48 from skin and sublayer. A model analogous to an electric flow in a circuit is shown in Figure 2.2. The overall permeation resistance can thus be determined by the following equation (2.20) Rt = R1 + R2 R3 /( R2 + R3 ) where R1 represents the resistance of the skin layer, R2 and R3 are the resistances of the pores and the polymer matrix in the porous substrate, respectively. It has been illustrated by Huang and Feng that the selectivity achievable in an asymmetric pervaporation membrane was determined by the resistances in both skin layer and substrate [125]. In this work it is also showed that the performance enhancement of dual-layer P84 co-polyimide/polyethersulfone (PES) hollow fibers upon heat treatment was restricted by the drastically increased resistance of the PES inner layer. Skin layer R1 l1 l2 Substrate R2 R3 Pore R1 = resistance of the skin layer (=Rskin) R2 = resistance of the pores in the substrate R3 = resistance of the polymer matrix in the substrate Rsub = resistance of the substrate [=R2R3/(R2+R3)] Figure 2.2 Resistance model of mass transport through an asymmetric membrane [11,125] 49 The integrity of composite membranes is also of great concern when the top layer and the substrate exhibit different degrees of swelling and thus produce a big stress at the interface which consequently causes instability (e.g. delamination) of the membrane structure [22]. Traditionally, this problem was solved by crosslinking of the top layer, inserting another layer between the skin layer and the substrate, and in-situ polymerization. Dual-layer hollow fibers made by co-extrusion of two polymeric solutions provide the opportunity to form a good adhesion among inner and outer layers by dry-jet wet spinning [49,126,127]. Li et al. [127] demonstrated conceptually that delamination-free dual-layer hollow fibers could be produced by varying outerand inner-layer spinning dope composition, bore fluid composition as well as posttreatment process. In the air gap region, molecular diffusion may take place at the interface and create an interlocking structure at the interface [128]. The inter-layer molecular diffusion can be further enhanced with an indent spinneret design [129]. For pervaporation, In this work, we developed BTDA-TDI/MDI co-polyimide (P84) /PES dual-layer hollow fibers with good integrity for the dehydration of isopropanol/water mixtures [2], while Wang et al. [38] were the pioneer developing polybenzimidazole (PBI)/P84 dual-layer hollow fibers for the dehydration of tetrafluoropropanol (TFP). The interface can be further improved if the polymeric materials in the inner and outer layers have strong interactions. Of course, the traditional single-layer asymmetric membranes consisting of an integrally skinned selective layer and a porous support layer made from the same 50 material is still a promising structure due to the easy and simple fabrication procedure. These membranes are prepared by the phase inversion technique whereby a homogenous polymer solution is transformed into a three dimensional macromolecular network in flat sheet or hollow fiber geometry [130]. However, the reproducibility is very sensitive to the process conditions. Using the same polymer, significant differences in morphology and performance between asymmetric membranes under different preparation conditions have been observed [123]. Some key factors influence the formation of phase inversion flat sheet membranes include: polymer dope composition, the volatility of solvent and non-solvent additives, chemistry of the quench medium, cast thickness, evaporation time, drying process and post-treatment conditions [131,132]. The formation mechanism for hollow fibers is more complicated than that for flat sheet membranes and additional controlling factors are involved, such as air gap distance, fiber take-up rate, dope and bore fluid flow rates, bore fluid chemistry and spinneret dimension [1,133,134]. Table 2.2 briefly summarizes the effect of some controlling factors in hollow fiber fabrication process. Since the invention of phaseinversion technology by Leob and Sourirajan [135], it was generally admitted that the formation of a defect-free membrane with an ultra-thin skin layer is difficult due to the complexity in the phase-inversion process. As a result, various post treatments such as thermal treatment, crosslinking and coating are exploited to eliminate the defects and to improve the separation performance at a compensation of increasing processing 51 time, cost and reduction of flux. Recent advances in membrane fabrication proved that through the proper control and adjustment of the preparation conditions in the phase inversion process, defect-free asymmetric membranes may be obtainable without the aid of post treatments [134, 136-138]. Pinnau and Koros [136] fabricated defect-free asymmetric flat membranes by casting a dope contained solvent and non-solvent followed by forced convective evaporation of solvent though a dry/wet phase inversion process. Chung et al. [137] developed defect free 6FDA-durene hollow fibers from a modified Lewis acid: base complex dope. Clausi and Koros [138] produced defect-free Matrimid hollow fibers with a thin skin layer by spinning dopes comprising volatile and non-volatile solvents, and a polymer. Chung et al. [139] also proposed that the spinning dope should exhibit significant chain entanglement in order to produce hollow fibers with an ultra-thin skin layer and minimum defects. Peng and Chung [134] demonstrated that by varying spinneret dimension with a proper take-up rate and air gap distance, defect-free Torlon® hollow fibers with ultra-thin dense layers can be formed from a polymer/solvent binary system. Although defect-free membranes were mainly investigated for gas separation, the principles developed in these studies were principally applicable for the future development of pervaporation membranes. It is believed that one can obtain a defect-free pervaporation membrane without post-treatment through properly controlled parameters in the fabrication process. 52 2.3.1 Membrane materials selection In section of 1.2, several different materials including highly hydrophilic materials, aromatic polyimides, hydrophobic materials, and inorganic materials are categorized for hydrophilic membranes. Generally, polymers which contain hydrophilic groups in the structure are preferred for the dehydration of alcohol mixtures, it is because of the hydrophilic groups preferentially absorb water molecules, thus lead to high flux and separation factor. However, the hydrophilic groups often cause plasticizing effect that 53 leads to membrane swelling and results in low selectivity [140]. Therefore the hydrophilicity and hydrophobicity have to be balanced to produce membranes with high selectivity and reasonable permeability. Vice versa, hydrophobic materials are commonly chosen in studies of organophilic membranes. The intensively studies using natural materials i.e.chitosan, agarose, etc were reported for alcohol dehydration in last few decades. On the other hand, Inorganic materials i.e. zeolite, ceramic are getting more attentions in pervaporation memebrane research works because the advantages of high selectivity and high permeability oriented by unique molecular sieving property and selective adsorption. The drawbacks of inorganic membranes are high cost, more expensive fabrication process and the brittleness. The performances of recently developed inorganic membranes are summarized in Table 2.3. Table 2.3 Collection of pervaporation performance of some inorganic membranes in aqueous alcohol systems Membrane Material Zeolite NaA /Carbosep tube support Feed & concentration Operation T ºC Total flux (g/m2hr) SF Membrane structure 90wt% IPA 70 300 2000 Composite [141] Zeolite NaA Ceramic commercial membrane Pervap SMS® Silca membrane 70wt% EtOH 60 2100 2140 Composite [64] 95wt% IPA 70 2100 600 Tubular [65] 90wt% IPA 70 300 60 Porous [68] Reference 54 2.3.2 Membrane structures and configurations Dense non-porous membranes are homogeneous membranes which are usually used only for characterization of material intrinsic properties such as permeability, selectivity and sorption properties. The thick dense membranes are not suitable in real applications because of the very low permeation flux. Asymmetric membranes prepared by a phase inversion method were first developed by Loeb and Sourirajan [135]. In general, asymmetric membranes contain an integrally skinned selective layer and a porous support layer; therefore high permeation flux can be achieved. On the other hand, the defects existing in the skin layer may deteriorate the selectivity. A typical phase inversion process is to precipitate a homogenous polymer solution in a coagulation bath and followed by a suitable drying process. During the precipitation, a homogeneous polymer solution is transformed into two phases: a polymer-rich solid phase which forms the rigid membrane structure, and a polymer poor liquid phase which forms the voids. Figure 2.3 shows a typical structure of an asymmetric membrane. Permselective layer Microporous support Figure 2.3 Structure of an asymmetric membrane [10] 55 The structure of a composite membrane consists of a selective layer and a substrate made from different materials shows in Figure 2.4. Due to a combination of different materials, composite membranes are capable to combine the properties of permselectivity, mechanical stability and strength of different materials of composite membranes [10]. Composite membranes are typically prepared by casting, coating or in situ polymerization the selective dense layer on top of the microporous support. The commercial Sulzer polymeric membranes are typical composite membranes , it contains a non-woven fabric layer, a polyacrylonitrile (PAN) porous layer as the support, and a PVA dense layer as the selective layer. In this work, Silicon rubber coating technique is applied on the assymetric hollow fiber membranes in order to seal the defect and enhance the separation performance [1]. Permselective layer Microporous support Figure 2.4 Structure of a composite membrane [10] Hollow fiber membranes own the advantages of large surface to volume ratio, high packing density, and mechanical self-supporting. Figure 2.5 shows a schematic diagram of a hollow fiber membrane. Hollow fiber membranes are very popular in many other membrane configurations; however the development of hollow fiber 56 pervaporation membranes was far behind polymeric flat membranes. This deficiency may arise from (1) the permeate pressure builds up at the lumen side which lead to a significant decrease in the driving force; (2) the free-standing hollow fibers may exhibit more severe swelling than flat membranes cast on non-woven fabrics [1]. Microporous support Permselective layer Figure 2.5 Structure of a hollow fiber membrane [10] 2.3.3 • Membrane preparation methods Phase inversion Phase-inversion is the most popular polymeric membrane fabrication method [108]. In this process, a homogenous polymer solution experiences a phase change and separated into two phases during the solvent evaporation and precipitation steps. Thermodynamic factors determine the phase boundary line and kinetic factors includes the speed of solvent evaporation and the rate of solvent-nonsolvent exchange. Figure 2.6 shows a ternary phase diagram, including polymer, solvent, and nonsolvent, is used to describe the composition change during the membrane formation. 57 This diagram is important in determining the formed membrane structure [142]. The area I located between the solvent/polymer axis and the binodal line indicates the stable region; the area II located between the binodal and spinodal curve indicates the metastable region; and the area III located between the spinodal and polymer/nonsolvent axis indicates the unstable region [143]. The binodal and spinodal curves can be obtained theoretically from Flory-Huggins thermodynamics by Gibbs free energy of mixing [108, 144,145]; only binodal curve can be obtained experimentally by the determination of cloud point [146]. For a given composition located in the metastable and unstable region, the interception of tie line and binodal curve gives the compositions of polymer-rich phase and polymer-lean phase when equilibrium is reached. Polymer Dense layer Binodal curve II Tie line III Open-cell structure Spinodal decomposition I II Nucleation and growth Solvent Spinodal decomposition curve Porous structure Nonsolvent 58 Figure 2.6 Schematic representation of a ternary phase diagram under isothermal condition Kinetically, the rate of solvent evaporation, the solvent depletion and nonsolvent penetration in a coagulation bath affect the formation of membrane structures. Factors that have significant influences on membrane structure include: polymer concentration, choice of solvent and nonsolvent, composition of coagulation bath, temperature of the casting solution and coagulation bath, distance of liquid-liquid demixing, and evaporation time. There are two types of demixing process: instantaneous liquid-liquid demixing and delayed onset of liquid-liquid demixing. Instantaneous demixing occurs when the membrane is formed immediately after immersion in the nonsolvent bath, and the membrane obtained has a relatively porous skin layer which can be used as microfiltration/ultrafiltration membranes; on the contrary, delayed demixing onsets after a period of time and results in the formation of membrane with a dense skin layer which is suitable for gas separation/pervaporation [3]. Two type of phase inversion process are widely applied in membrane industries, are wet-phase inversion (immersion precipitation), and dry/wet phase inversion. Dry phase inversion process was invented by Leob and Sourirajan [135] and has become the most popular membrane formation process for commercial asymmetric membranes. The casting solution is immersed into a coagulation bath which contains nonsolvent. Phase separation is induced by solvent/non-solvent exchange in the coagulation bath. The formed membrane usually undergoes a solvent-exchange process before drying to maintain the microporous structure. 59 Dry/wet phase inversion is a combination process of evaporation and immersion precipitation. After a short evaporation period, the cast film is immersed into nonsolvent and then precipitates. The volatility of solvent, dope composition, evaporation time, cast thickness, chemistry of the quench medium influence the formation of the membrane structure [131, 132]. • Thermally-induced phase separation In thermally-induced phase separation process, a polymer solution contains a polymer and a latent solvent which is a solvent for the polymer at elevated temperatures and a non-solvent when the temperature is cooled down. Phase separation occurs during cooling down the temperature. This process can be utilized to produce isotropic microporous membranes [147, 148]. • Solvent evaporation A dense homogeneous membrane can be formed by casting, dip coating or spraying a polymer solution on a support followed by slowly evaporating solvent. Phase separation occurs when the solvent is slowly evaporating which then leads to the solidification of polymer matrix [149]. 2.3.4 Membrane post treatment and modifications 60 The purpose of post treatment and modification of membrane material is to alter the membrane physical and chemical properties in order to achieve the desired membrane properties and separation performance. The heat treatment, chemical treatment i.e. chemical cross-linking etc, blending with different polymers or blend with inorganic particles into the polymer matrix are widely applied to enhance the membrane separation capability. In this work, heat treatment and chemical cross-linking are used to enhance the hollow fiber membrane separation performance. • Heat treatment Heat treatment is easily operated and is capable to improve membrane stability and performance of pervaporation membranes [42]. Heat treatment usually induces increased density, reduced free volume, and thus leads to declined permeability and increased selectivity. Nagy et al. [151] modified cellulose hydrate membranes by means of dry and wet heat treatment. Tsai et al. [50] modified PAN hollow fibers at different temperatures for IPA/water dehydration and showed that the skin layer was densified and the separation performance was improved. As for polyimide membranes, the effect of heat treatment is also significant. Yanagishita et al. [42] found heat treatment of polyimide membranes at 300°C for 3hr could optimize the separation factor for ethanol dehydration. The increased mechanical strength and separation factor were attributed to cross-linking although no 61 additional chemical functional group was identified. On the other hand, the red shift in fluorescence indicated the charge transfer interaction was formed between imide and amine moieties. The thermally induced polymer chain relaxation and rearrangement and the formation of charge transfer complex (CTCs) have contributed to the improvement of gas transport properties and the suppression of the CO2 plasticization for polyimide gas separation membranes [53, 54]. The formation of CTCs can be characterized by both fluorescence and UV-vis spectrophotometer. Zhou and Koros [51] suggested that the considerable separation factor enhancement of Matrimid fibers at a heat treatment temperature lower than Tg could be attributed to the local or segmental motions of polymer chains at β transition, the segmental motions of polymer chains at temperature above β transition enhanced CTCs formation. • Chemical modification Cross-linking carried out by chemical reagents on membrane selective layer is used to control excessive swelling of highly hydrophilic materials, to enhance membrane stability and to change the polymer chemistry [34, 57, 58]. For polyimide membrane, the imide groups of polyimide can be modified by amines to form amide groups. Diamine or dendrimers with multi amine groups have been successfully used for cross-linking polyimide membranes and new gas separation membranes were obtained [55-58, 151-152]. The modification of polyimide 62 membranes by diamine has been extensively studied for gas separation, the studies of its effects on P84 co-polyimide membranes and the resulted pervaporation performance and properties are not well investigated yet. This work will conduct studies of p-xylenediamine cross-linking modification on dual-layer hollow fiber, the effects on membranes morphology and performance of pervaporation dehydration of IPA will be discussed. 63 CHAPTER THREE EXPERIMENTAL 3.1 Materials P84 co-polyimide and PES were purchased from HP Polymer GmbH and Amoco Polymers Inc., respectively. Figure 3.1 & 3.2 shows the chemical structure of P84 & PES. P84 has a glass transition temperature (Tg) of 315◦C, PES has a glass transition temperature (Tg) of 215◦C. All polymers were dried at 120 ◦C in a vacuum oven overnight before use. The cross-linking agent, i.e. p-xylenediamine was purchased from Sigma–Aldrich. The chemical structure and properties of p-xylenediamine is listed in Table 3.1. The solvent N-methyl-2-pyrrolidinone (NMP) and hexane were purchased from Merck, and methanol was purchased from Tedia. IPA was purchased from Fisher Scientific. All chemicals were used as received. Milli-Q ultrapure water was prepared in lab. O O O CH3 H N O N C O H 80 % n 20 % Figure 3.1 Chemical structure of P84 co-polyimide 64 Figure 3.2 Chemical structure of Polyethersulfone (PES) RADEL A-300 Table 3.1 The chemical and physical properties of p-xylenediamine Chemical Chemical structure p-xylenediamine H2N NH2 Molecular weight Molecular dimension 136 Boling point (ºC) Melting point (ºC) 230 61 a a: simulated by Cerius2 software. 3.2 Membrane preparation 3.2.1 P84 dope preparation and viscosity measurements Figure 3.3 shows the solution viscosity as a function of P84 concentration in NMP determined by an ARES Rheometric Scientific rheometer with a 25mm cone plate at room temperature (27°C) under a dynamic mode. 28 wt% P84 was chosen to prepare the spinning solution because a dope exhibiting significant chain entanglement may 65 produce hollow fibers with ultra-thin skin layer and minimum pore porosity [139]. The solution was prepared by the following procedures. P84 polymer powders were dispersed in a cold NMP solvent (0-3°C) with a Eurostar high speed stirrer. The low temperature slows down the dissolving rate and prevents polymer powder from agglomeration [139]. The solution was agitated until the polymer was fully dissolved, while the dope temperature slowly increased to room temperature. The relationship between shear stress τ (Nm-2) and shear rate γ&(s-1) for the 28 wt% P84/NMP spinning dope follows a power-law relation [153]: 0.93 τ = 57.722 γ& (3.1) With the known spinneret dimensions and dope flow rates, the shear rate, shear stress and their distribution can be calculated by commercially available FLUENT software [154]. Figure 3.3 Viscosity vs. concentration for P84/NMP system at 27 Deg.C 66 3.2.2 Spinning process and modules fabrication of asymmetric P84 hollow fiber membranes Figure 3.4 shows the schematic diagram of the single layer hollow fiber spinning system [139]. The polymer solution was degassed 24 hrs before loading into a syringe pump (ISCO 1000), with following 24 hrs degassing after loading. A mixture of 95/5 (w/w) NMP/water was employed as the bore fluid in order to make a porous substructure to minimize the substructure resistance. Tap water was used as the external coagulant. Two coagulation bath temperatures (6°C and 27°C) were studied. Both dope fluid and bore fluid were filtered through 15µm sintered metal filters before spinning. Polymer solution and bore fluid were extruded by two ISCO syringe pumps through a spinneret at room temperature with air gap distances varying from 0 to 15 mm. The spinneret has an outer diameter (o.d.) of 8 mm and an inner diameter (i.d.) of 5 mm. Table 3.2 lists the spinning conditions and Table 3.3 summarizes hollow fibers’ ID. and their preparation conditions. The nascent fibers entered freely into the coagulation bath without additional drawing. A constant ratio of the dope flow rate to the bore fluid rate was employed in order to reduce the complicated dragging and coupling effects of bore fluid flow rate on fiber formation [155]. The as-spun hollow fibers were stored in water for 3 days, then solvent exchanged by washing with methanol 3 times, each time 0.5 hr, followed by washing with hexane 3 times, each time 0.5 hr. The hollow fibers were dried in air naturally after solvent exchange. 67 Figure 3.4 The schematic diagram of the single layer hollow fiber spinning system Table 3.2 Spinning conditions of P84 asymmetric hollow fibers Table 3.3 List of P84 asymmetric hollow fibers’ ID & preparation conditions 68 Modules were prepared by loading 2~5 pieces of hollow fibers with an effective length of 20-22 cm into a stainless steel holder and sealing both ends with epoxy. Each module was cured 48 hrs at ambient temperature before evaluation. 3.2.3 Spinning process and modules fabrication of P84/PES dual-layer hollow fiber membranes The schematic diagram of the dual-layer hollow fiber spinning system is shown in Figure 3.5. The structure of the dual-layer spinneret is shown in Figure 3.6. [127]. The dimensions of three independent channels are 1.20 mm/0.97mm (o.d./i.d. of the outer- 69 layer flow channel), 0.81 mm/0.44mm (o.d./i.d. of the inner-layer flow channel) and 0.30mm (o.d. of the bore fluid channel), respectively. A 28 wt% P84/72 wt% NMP solution was used as the outer-layer dope to create a selective layer, whereas 32 wt% PES/68 wt% NMP solution was utilized as the inner-layer dope to create a porous support layer. The outer- and inner-layer dope solutions were degassed 24 h before loading into two syringe pumps (ISCO500 and ISCO1000), respectively, and followed by 24 h degassing after loading. A mixture of 95/5 (in wt%) NMP/water was employed as the bore fluid. Tap water was used as the external coagulant. The dope fluid and bore fluid were filtered through 15 micron sintered metal filters before spinning. Two polymeric dope solutions and the bore fluid were extruded simultaneously by three ISCO syringe pumps through the dual-layer spinneret at room temperature with air gap distances varying from 0 to 17 mm. The nascent fibers entered freely into the coagulation bath without additional drawing. Table 3.4 lists the spinning conditions. Table 3.5 summarizes untreated hollow fibers’ ID and their preparation conditions. The as-spun hollowfibers were stored in water for 3 days, then solvent exchanged by washing with methanol three times, each time 0.5 h, followed by washing with hexane three times, each time 0.5 h. The hollow fibers were dried in air naturally after solvent exchange. The dual-layer hollow fiber membrane modules were prepared by loading two pieces of hollow fibers with an effective length of 16.5 cm into a polypropylene module. Both ends were sealed with epoxy. Each module was cured 48 h at ambient temperature before evaluation. The heat treatment and chemical cross-linking 70 modifications of dual-layer hollowfibers were carried out before module fabrication. At least two modules were tested for each hollow fiber membrane sample. Table 3.4 Spinning conditions of P84/PES dual-layer hollow fibers Table 3.5 List of untreated P84/PES dual-layer hollow fibers’ ID & preparation conditions 71 72 Figure 3.5 The schematic diagram of dual layer hollow fiber spinning system A: inner layer dope fluid, B: outer layer dope fluid, C: bore fluid, D: syringe pump, E: 15 micron filter, F: dual layer spinneret, G: coagulation bath, H: take up unit, I: water spray, J: collection bath Figure 3.6 The structure of the dual-layer spinneret 73 3.2.4 Post treatments: silicon rubber coating, heat treatment and chemical modification of membranes • Silicon rubber coating The as-spun P84 hollow fiber samples were tested without coating, then dried at 60°C in a vacuum oven overnight. After that, the same samples were coated with 3 wt% poly-dimethylsiloxane (Sylgard® 184) in a hexane solution under vacuum for 0.5 hr followed by curing in air at room temperature for 24 hrs. • Heat treatment Three treatment temperatures are chosen; namely 200, 250 and 300°C. Heat treatment protocol 1 was carried out by increasing temperature at a rate of 10°C/10 min to 200°C, then holding there for 2 hrs followed by naturally cooling down in a vacuum oven. Heat treatment protocols 2 and 3 applied a heating rate of 10°C/10 min to 200°C, then reduced the rate to 12°C/20 min to 250°C and 300°C, respectively, held there for 1 hr followed by naturally cooling down in the vacuum oven, and held there for at least 6 hrs under vacuum. Different holding times were chosen in order to minimize the increase in dense layer thickness during heat treatment. Heat treatments of hollow fibers were conducted in a vacuum furnace before module fabrication. 74 • chemical modification The chemical cross-linking modifications of hollow fibers were carried out after solvent exchange. The hollow fibers were immersed in a 2.5% (w/v) pxylenediamine/methanol solution for 1, 2 and 3 h, respectively, followed by washing with fresh methanol three times to remove the residual cross-linking agents.. The hollow fibers were then dried in air naturally. 3.3 Pervaporation Experiments Figure 3.7 shows the pervaporation apparatus designed for the characterization of hollow fiber membranes. The system consists of 4 sets of hollow fiber modules. The aqueous IPA solution was delivered from a 2-liter stainless steel tank to membrane modules by a recirculation pump (Speck pumpen PY2071). The system pressure and flow rate were controlled by the pressure control valve and the return valve, and the recirculation flow rate was measured by a flow meter. The solution temperature was controlled by a thermocontroller (B. Braun) via a heating bath where the solution tank was immersed in. The feed temperature was monitored by a thermometer placed in the outlet of the hollow fiber modules. Permeate was collected individually by using cold traps immersed in liquid nitrogen, while vacuum was maintained by a vacuum pump (Rietschle VGC-6). A vacuum gauge (Vacuubrand DRV2) was installed at the permeate side to monitor the down stream vacuum pressure. 75 Figure 3.7 Flow diagram of pervaporation experimental setup 2 liters aqueous mixture containing 85 wt% IPA was used as the feed solution. The pressure at the permeate side was normally maintained at less than 0.5 kPa. The system was stabilized for 2 hrs before the collection of samples. The recirculation rate was set 76 at 36 liter/hr for each module. Two hollow fiber modules were tested for each sample and the data were averaged. The testing temperature was maintained at 60°C. Retentate and permeate samples were collected at certain time intervals (normally one hour) and their compositions were analyzed by a Hewlett-Packard GC 6890 with a HPINNOWAX column (packed with cross-linked polyethylene glycol) and a TCD detector. The permeate samples were weighted by a Mettler Toledo balance. Flux and separation factors were calculated by the following equations J= Q At α 2 /1 = (3.2) y 2 / y1 x 2 / x1 (3.3) where J is the flux, Q the total mass transferred over time t, A the membrane area, x2 and y2 are the mole fractions of water in the feed and permeate, respectively, and x1 and y1 are the mole fractions of alcohol in the feed and permeate, respectively. As shown in Chapter 2, the dependence of permeation flux on temperature follows the Arrheniustype relation [11]: J = J 0 exp( − E J / RT ) (3.4) where EJ is the apparent activation energy of permeation, J0 is the pre-exponential factor, and EJ can be obtained from the ln J vs. 1/T plot. 77 As depicted in Chapter 2, the driving force for pervaporation can be expressed as a vapor pressure difference. The basic pervaporation transport equations based on the solution diffusion model are [103] J1 = P1 f ( p1 − p1p ) = P1 ( p1f − p1p ) l (3.5) J2 = P2 ( p 2f − p 2p ) = P2 ( p 2f − p 2p ) l (3.6) where P1 and P2 are the membrane permeability for alcohol and water, respectively, which are the product of the solubility and diffusivity. Subscripts 1 and 2 correspond to alcohol and water, respectively, while superscripts f and p correspond to the feed and permeate. l is the membrane dense layer thickness. p denotes the partial vapor pressure, while P is the permeance. The membrane selectivity β is defined as P2 divided by P1 . The partial vapor pressure of each component on the feed side can be obtained based on its concentration in the feed liquid mixture p1f = x1γ 1 p1s (3.7) p 2f = x 2 γ 2 p 2s (3.8) where the superscript s indicates the saturated state and γ is the activity coefficient. In this study, p1s and p 2s were calculated by the Antoine equation [156] or obtained from HYSYS DISTIL software, while γ1 and γ2 were obtained from the HYSYS DISTIL software (version 5.0, provided by Hyprotech Ltd, Canada). By substituting Equations 78 3.7 and 3.8 to Equations 3.6 and 3.7, the permeance can be calculated by the following equations P1 = J1 /( x1γ 1 p1s − p1p ) (3.9) P2 = J 2 /( x2γ 2 p2s − p2p ) (3.10) 3.4 Membrane Characterizations The morphology of hollow fiber membranes was observed by using a JSM-6700F field emission scanning electron microscope (FESEM). Samples were prepared by fracturing membranes in liquid nitrogen then coating with platinum. Energy dispersion of X-ray (EDX) with a line scan mapping by using a JEOL JSM-5600LV scanning electron microscope (SEM) was applied to detect the distribution profile of silicon rubber coating layer in a coated hollow fiber’s cross-section. A Nanoscope III equipped with 1533D scanner (Digital Instruments, California, USA) was used to study the surface morphology of membranes by using a tapping mode atomic force microscope (AFM). A X-ray photoelectron spectrometer (Kratos XPS System-AXIS His-165 Ultra) was used to measure the surface chemical compositions of the original and cross-linked P84/PES dual-layer hollow fibers. A PC Processing software on the SUN SPARC station was used to quantify the spectra of N and O elements. 79 CHAPTER FOUR FABRICATION AND CHARACTERIZATION OF BTDA-TDI/MDI (P84) COPOLYIMIDE HOLLOW FIBER MEMBRANES FOR THE PERVAPORATION DEHYDRATION OF ISOPROPANOL 4.1 Introduction IPA is an important solvent use in pharmaceutical and electronic industries. Most of the applications require high purity, i.e. 99.5% of IPA content [157]. Because of IPA forms an azeotrope with water at around 88wt% at atmospheric pressure [158] and this makes it difficult to be separated from water by conventional processes. Compared to conventional processes such as extractive distillation or adsorption, pervaporation is very promising because of its interesting energy saving aspect [11] and effectiveness for azeotropic separation and solvent recovery. Asymmetric hollow fiber membranes have received great interests from both academia and industry. However, asymmetric flat membranes have always been the focus for the development of polymeric pervaporation membranes. Two decades ago, some researchers [8, 159] expressed concerns that asymmetric polymer membranes made by the phase inversion process might not be suitable for pervaporation separation, probably because of material limitations (i.e., severe swelling and poor inherent selectivity) and 80 difficulties of fabricating defect-free asymmetric pervaporation membranes with characteristics of high flux and high selectivity. The phase inversion process has the advantage of minimizing skin-layer thickness but increases the probability of membrane defects, which may become the roots of failures in the separation of water and organic mixtures in pervaporation processes. With the significant advance in membrane material research and membrane fabrication in the last two decades, nowadays the phase inversion process is a proven technique to produce defect-free asymmetric membranes through proper control and adjustment of membrane preparation conditions [137-138,160-161]. Hollow fiber configuration made by this process has been widely used in gas separation, reverse osmosis, ultrafiltration, microfiltration, and other membrane processes due to its large surface to volume ratio, mechanical self-support and high packing density. However, compared to other membrane processes, little attempts have been made to prepare defect-free integral asymmetric membranes for pervaporation, and very rare systematic studies can be found on the fabrication of defect-free hollow fiber membranes for pervaporation [11]. Tsuyumoto et al [162] studied the performance of a pervaporation plant using modified polyacrylonitrile hollow fiber membranes for the dehydration of ethanol (EtOH). They found that the pervaporation plant using hollow fibers not only has the economical superiority over the traditional distillation and the plant utilizing GFT flat membranes, but also saves membrane cost and space. Hollow fiber membranes made from others 81 materials such as Nafion 811 [163], cellulose acetate [164], polypropylene grafted with poly(acrylic acid) [165], grafted anion-exchange polyethylene [166], cellulose acetate / chitosan [167] have been studied for the pervaporation dehydration of aqueous EtOH mixtures or aqueous IPA mixtures. Table 4.1 summarizes their separation performance in terms of flux / permeance and separation factor / selectivity. Clearly, the performance of these hollow fiber membranes is not impressive and far behind those polymeric flat membranes [68] or membranes prepared from inorganic and organic / inorganic hybrid materials [64, 168-169]. This deficiency may arise from the following factors: 1) The difficulties of fabricating asymmetric hollow fiber membranes with a thin defect-free skin layer; 2) The feed-induced swelling of pervaporation membranes. Usually it may be more severe in free-standing hollow fibers than in flat membranes cast on nonwoven fabrics; 3) The effect of concentration polarization. It becomes more rigorous and lowers the separation performance if the feed is at the shell side. The first factor is an engineering issue because it is difficult to simulate the hollow-fiber spinning process using the fabrication conditions developed for asymmetric flat membranes. The former has two coagulations taking place simultaneously at the inner and external skins of hollow fibers, while the latter has only one major coagulation process occurring at the outer skin of asymmetric flat membranes. In addition, the chemistry of external and internal coagulants for hollow fiber spinning may be different, thus complicates the evolution of membrane formation. The second and third factors are material and system issues, respectively. Therefore, it is much more challenging to develop hollow fibers for pervaporation applications. One must select stable polymer materials with minimal 82 solvent-induced swelling and develop appropriate spinning technologies to fabricate hollow fiber membranes with an ultra-thin selective layer in order to produce useful membranes for pervaporation. Recently we have identified P84, a commercially available co-polyimide of 3,3’4,4’benzophenone tetracarboxylic dianhydride and 80% methylphenylene-diamine + 20% methylene diamine, as a promising material for the pervaporation dehydration of IPA and water mixtures. The asymmetric flat P84 membranes developed by Qiao displayed comparable or superior separation factor and flux to commercial flat sheet membranes 83 [41]. Therefore, the objectives of this study are 1) to explore if we can develop P84 hollow fiber membranes for pervaporation dehydration of IPA with equivalent or superior performance to asymmetric flat membranes, and 2) to identify key processing factors to produce defect-free P84 hollow fiber membranes. 4.2 Results and Discussion 4.2.1 Effects of air gap distance on membranes morphology Figure 4.1 displays the SEM images of cross-sections of hollow fibers coagulated at 6ºC with various air gap distances. Clearly, the wet-spun hollow fibers have the largest o.d.. This is mainly resulted from a rapid solidification of the outer skin because water is a powerful coagulant and because precipitation takes place much fast at low temperatures. In addition, wet spinning minimizes the effect of drawing created by gravity. As a result, the fiber o.d. is very close to the spinneret size (i.e., 0.715 vs. 0.8 mm). On the other hand, fibers spun with an air gap distance of 5 mm exhibit the largest wall thickness which is due to the die-swell effect. Highly viscous random-coil polymer solution tends to relax the shear-induced orientation by exhibiting expansion in the transverse direction (to the spinning direction) after passing through an annulus straight spinneret [170-171]. Figure 4.2 shows the surface morphology and roughness of as-spun hollow fibers examined by AFM. Clearly, the wet-spun hollow fibers have a skin with nodules of 0.05-0.15 micron [172]. Possibly because of die swell, vigorous quench and then stretch by gravity, fibers spun at an air gap of 5 mm have less nodule appearance but higher roughness. At the air gap distances of 10 and 15 mm, the skin 84 roughness and nodule size significantly decrease owing to the elongational stretch caused by gravity. In the same way, with an increase in air gap distance, the gravitational stretch is responsible for the decreases in both fiber o.d. and wall thickness. Figure 4.1 The SEM images of the cross-sections of hollow fiber spun at coagulant 6 ºC and a) wet spinning; b) 5 mm air gap; c) 10 mm air gap and d) 15 mm air gap Figure 4.2 AFM images of outer fiber surface spun at 6ºC and a) wet spun; b) 5 mm air gap; c) 10 mm air gap and d) 15 mm air gap (Ra: mean roughness and Rms: the root mean square of Z values) 85 As shown in Figure 4.1, the formation of finger-like voids appears to be also affected by the air gap distance, i.e. the macrovoids tend to disappear at a higher air gap distance. Similar air-gap dependent macrovoid structure has been reported for polyethersulfone hollow fiber membranes [139, 173]. Even though membrane scientists have debated and divided on the origins of macrovoid formation for the last three decades, we believe, for the case of hollow fibers, it most likely starts from local surface instability, material and stress imbalance, which result in weak points and induce intrusion of coagulants, as suggested by Strathmann et al [174] and recently summarized by Wang et al [175]. As illustrated in Figure 4.1, the wet-spun and dry-jet wet-spun P84 hollow fibers at an air gap of 5 mm have a regular array of finger-like macrovoids near the inner layer, followed by a microporous layer and finally a dense skin layer. At higher air gap distances of 10 and 15 mm, the numbers of macrovoids become much less. Even though the elongation induced orientation and a longer exposure of hollow fiber surface to moisture in the air gap region may cause viscosity increase and retard macrovoid formation [173], the most likely cause for the reduction of macrovoids may be the rapid shrinkage process of fiber diameter during the elongational stretch by gravity. It may induce radial outflow (i.e., negative normal stresses) which hinders the capillary intrusion or diffusion of coagulants, thus eliminate the chance of forming macrovoids [175-176]. 86 4.2.2 Effects of air gap distance on membrane’s pervaporation performance Figure 4.3 shows the pervaporation performance of the above hollow fibers spun with different air gap distances for IPA and water separation. Wet-spun hollow fibers have the highest separation factor / selectivity and relatively high flux / permeance because the instantaneous phase inversion of highly concentrated polymer solutions tend to form an ultra-thin and less defective skin layer [173]. Hollow fibers extruded at an air gap distance of 5 mm have the highest flux / permeance but relatively low separation factor / selectivity. This is consistent with AFM observations that the perfectibility of the outer skin may suffer from die swell and then vigorous precipitation. With a further increase in air gap distance, both the permeance and selectivity seem to deteriorate. This may arise from the fact that air gap distance may not only induce orientation but also create defects [173, 177]. The former reduces flux and permeance, while the later decreases separation factor and selectivity. These two factors compete with each other and determine the final membrane’s separation performance. Figure 4.3 Pervaporation performance of A-1~4 hollow fibers spun at 6 ºC, dope flow rate 0.4 ml/min and different air gaps. The dotted lines indicate permeance and selectivity, respectively 87 4.2.3 Effects of coagulation temperature and dope flow rate on membrane morphology and performance Two coagulation bath temperatures, namely, 6°C and 27°C were investigated in this study. The morphology of A series (spun at 6°C) and B series (spun at 27°C) is illustrated in Figure 4.4. Fibers spun at a higher temperature tend to have more as well as longer macrovoids in the cross section. This is attributed to the fact that the dope viscosity becomes higher at low coagulation temperatures and the low-temperature coagulation tends to tighten the nascent membrane structure. As a result, the resistance for coagulant intrusion increases and this retards the propagation of macrovoid formation. Even though hollow fibers precipitated at 6°C have a tighter skin layer than that precipitated at 27°C, the former seems to have a less perfect outer skin than the latter probably due to rapid quench, vigorous skin formation and low-temperature elongational stretch. Therefore, the former usually has lower flux / permeance and separation factor / selectivity than the latter, as shown in Figure 4.5. 88 Figure 4.4 Morphology of fibers spun at various air gap distances, dope flow rates and coagulant temperatures (Top A series: 6°C; Bottom B series: 27°C) Figure 4.5 Pervaporation performance of fibers spun at various coagulant temperatures and dope flow rates (A series: 6°C; B series: 27°C) The only difference between fibers A-1 to A-4 and A-5 to A-8 sets, B-1 to B-4 and B-5 to B-8 sets is their dope flow rates. Fibers A-1 to A-4, and B-1 to B-4 were spun at 0.4 ml/min, while the others were at 0.6 ml/min. Based on the FLUENT software, the calculated shear rate and shear stress for a dope flow rate of 0.4 ml/min are 104 s-1 and 89 4334 N/m2, respectively; and 156 s-1 and 6319 N/m2, respectively, for a dope flow rate of 0.6 ml/min. Figure 4.6 shows a comparison of their performance between each set. Interestingly, irrespective of coagulant temperature and air gap distance, higher separation factor / selectivity but lower flux / permeance are generally obtained for fibers spun at a higher shear rate (i.e., higher dope flow rate). This may be ascribed to enhanced molecular orientation of polymer chains at hollow fiber surface and increased wall thickness at high shear rates. Similar phenomena were observed for gas separation membranes and UF membranes [154, 170]. Figure 4.6 Effect of dope and bore fluid flow rate on membrane performance (A, B1~4: dope 0.4 ml/min and bore fluid 0.2 ml/min; A, B-5~8: dope 0.6 ml/min and bore fluid 0.3 ml/min). The dotted lines indicate permeance and selectivity, respectively 90 4.2.4 Effect of silicone rubber coating on membrane morphology and performance Silicone rubber coating was proven as an effective method to increase selectivity while maintaining permeability of gas separation membranes [178]. In pervaporation, silicone rubber is regarded as hydrophobic and organic selective [70,71]. However, a thin silicone rubber coating layer can enhance membrane’s selectivity in both pervaporative dehydration [179, 180] and alcohol removal [181]. Though the best performance of the as-spun P84 hollow fibers have a flux of 1160 g/m2hr and a separation factor of 139 which is comparable to the previous arts, silicone rubber coating is employed to enhance the membrane selectivity. The pervaporation performance of hollow fibers before and after silicone rubber coating is listed in Table 4.1 After coating, the flux / permeance drops about 18-60%, while the separation factor / selectivity increases at least 3 times. There are two possible reasons for these remarkable performance enhancements: 1) silicone rubber effectively seals the big defects; and 2) silicone rubber has high water permeability even though it has hydrophobic nature [182]. Based on the SEM-EDX silicone line scanning spectra shown in Figure 4.7, the thickness of the silicone rubber coating layer is about 2 µm. 91 Table 4.1 Pervaporation performance of P84 hollow fibers before and after silicone rubber coating Figure 4.7 The SEM-EDX line scanning spectra of silicon rubber coated A-8 hollow fiber cross section, carbon and silicon mapping 92 4.2.5 Effects of heat treatment on membrane property and separation performance Because B-5~8 hollow fibers exhibited impressive separation factor / selectivity and relatively high flux / permeance, they were selected for the post heat treatment study. Figure 4.8 compares the cross-section morphology of B-5 hollow fibers after different heat treatments. Macroscopically, the overall cross-sections do not differ from one another visibly. However, the enlarged outer-skin pictures show that the original B-5 hollow fiber has relatively loose packed polymer nodules underneath the outer skin; while the heat treated fibers apparently have more packed polymer nodules. Benefited from a less holding time, fibers heat treated under protocol 2 noticeably have a thinner dense sub-structure immediately underneath the outer skin than that under protocol 1. Figure 4.9 also shows the densification process of the outer skin morphology as a function of heat treatment conditions observed by FESEM. Figure 4.8 SEM images of B-5 hollow fibers and enlarged outer-skin cross-section morphology (a) original, (b) heat treatment protocol 1, (c) protocol 2 and (d) protocol 3 93 Figure 4.9 The SEM images of the outer skin surfaces of B-5 hollow fibers (a) original, (b) heat treatment protocol 1, (c) protocol 2 and (d) protocol 3 Table 4.2 compares the separation performance of hollow fibers before and after heat treatments at various temperatures. The decrease in flux / permeance with an increasing heat treatment temperature is compensated by a dramatic increase in separation factor / selectivity, because of the densification of dense-selective skin and the reduction of free volume and d-space [41, 46]. The drop in flux / permeance is about 30% to 45%; however, the increase in separation factor / selectivity is 20 to 100 times higher than the untreated fibers. This performance enhancement, particularly B-8 hollow fibers heattreated at 300°C, is very remarkable compared to the commercially available polymeric membranes, e.g. PERVAP 2510 membrane. The former has a flux of 883.5 g/m2hr and separation factor of 10585, whereas the latter has a flux of 857 g/m2hr and a separation factor of 1053 under the same operating conditions [106]. Table 4.1 also shows that the newly developed P84 hollow fiber membranes have much superior separation performance to other polymeric hollow fiber membranes. 94 Table 4.2 Pervaporation performance of P84 hollow fibers at various heat treatment temperatures A comparison of the calculated apparent dense layer thickness vs. the observed dense layer thickness for the original and heat treated B-5 hollow fibers is listed in Table 4.3. The apparent dense layer thickness is calculated by the ratio of water permeability of a 10 µm dense P84 membrane to the water permeance of an asymmetric hollow fiber membrane. The calculated apparent dense layer thickness varies from 0.43 to 0.57 µm. These values are in a reasonable range with the thicknesses observed from FESEM images. Table 4.3 Calculated apparent dense-layer thickness and observed dense-layer thickness by SEM for heat treated B-5 fibers and a dense P84 film 95 Moreover, comparing Table 4.2 with Table 4.3, one may find that the properly heat treated P84 hollow fibers may exhibit much higher selectivity than its homogeneous dense film. For example, the selectivity of heat treated B-8 hollow fibers at 300°C is 3616, whereas the 10 µm homogeneous P84 film heat treated at 250°C has a selectivity of 993 [41]. This difference may arise from the effects of elongational stretch during the hollow fiber fabrication process. The slightly stretched and oriented molecules have greater tendency to pack better during the annealing process. Interestingly, wet-spun hollow fibers exhibit much lower separation factor / selectivity than that of dry-jet wetspun hollow fibers after heat treatment. This again implies that elongation and shear stresses have different effects on chain orientation or molecular configuration during the hollow fiber spinning. The elongation stress is more powerful than shear stress to induce molecular orientation and alignment as pointed out by Cao et al [154]. 4.3 Conclusions P84 asymmetric hollow fiber pervaporation membranes were fabricated by phase inversion process. Effects of various spinning conditions and post treatments on P84 hollow fiber membranes morphology and performance were investigated. For dehydration of 85 wt% IPA in water mixture at 60°C, the best performance was obtained from hollow fibers spun from a 28/72 (w/w) P84 / NMP dope with water as a coagulant and 95/5 NMP / water as a bore fluid, at 15 mm air gap with coagulation at 27°C followed by a heat treatment at 300°C for 1 hr. This performance, i.e. flux 883.5 96 g/m2hr and separation factor 10585, far exceeds previous polymeric hollow fiber pervaporation membranes for IPA dehydration, indicating the potential for successful application of hollow fibers in pervaporation. The following conclusions can be drawn: 1) It was found that hollow fiber with a thin dense skin layer and macrovoid support layer is probably the best structure to attain both high permeability and selectivity. Through properly controlled spinning conditions and post treatments, i.e. air gap distance, dope / bore fluid flow rate, post treatment process and temperature, the high performance polymeric hollow fiber membranes were successfully developed by phase inversion process for pervaporation dehydration of IPA. 2) Wet-spun hollow fibers have the highest separation factor and relatively high flux among fibers without any post treatments. A die-swell phenomenon was observed at an air gap of 5 mm, and hollow fibers spun at an air gap of 5mm have a rough surface and apparently higher flux but lower separation factor. As air gap further increases, the flux decreases however the separation factor does not increase much. 3) Hollow fibers exhibit more finger-void structure when spinning at a high coagulation temperature (27°C); in contrast, fibers spun at a low coagulation temperature (6°C) have more microporous structure. The higher coagulation temperature results in fibers with a better performance (higher flux and higher separation factor). 4) Hollow fibers spun at a higher dope / bore fluid flow rate show higher separation 97 factor but lower flux due to the enhanced shear rate. 5) A thin layer silicone rubber coating increases separation factor about 3 times however the flux drops about 30% of its original value, indicating that a hydrophobic layer (silicon rubber coating) effectively seals big pores / defects however does not dominate the transport process. 6) Heat treatment can densify the P84 hollow fibers’ skin layer, remarkably increases selectivity especially for dry-jet wet-spun fibers. 98 CHAPTER FIVE FABRICATION AND CHARACTERIZATION OF BTDA-TDI/MDI (P84) /PES DUAL-LAYER HOLLOW FIBER MEMBRANES FOR THE PERVAPORATION DEHYDRATION OF ISOPROPANOL 5.1 Introduction Development of asymmetric hollow fibers for pervaporation applications is attractive because hollow fiber membranes have the following advantages: (1) a large surface area per volume ratio, (2) a self-supporting structure and (3) a self-contained vacuum channel where a feed can be supplied from the shell side while vacuum is applied on the lumen side. However, it is challenging to fabricate asymmetric hollow fiber membranes with an ultra-thin defect-free selective layer which have superior pervaporation performance, meanwhile possess the ability to overcome the solventinduced swelling. In order to successfully develop asymmetric hollow fibers for pervaporation, appropriate polymeric materials with superior performance and minimal solventinduced swelling must be employed to construct the crucial selectivity layer. However, the high material cost usually limits the real application of some promising but expensive materials [127]. One cost-saving method is to apply the expensive material with superior performance as a very thin selective layer, while choose the less 99 expensive material as a supporting layer to fabricate dual-layer hollow fiber membranes by means of simultaneous co-extrusion through a triple-orifice spinneret. For example, Li et al. [127] fabricated 6FDA-durene-mPDA/polyethersulfone duallayer hollow fibers for gas separation and reported 90% material cost saving. Studies of dual-layer hollow fibers started from late 1970s in the fields of hemodialysis [183] and water purification [184,185]. Intensive investigations have been carried out for gas separation membranes [126-128,176-177, 186-189] recently. Since two polymeric dopes made of different materials and different concentrations are the fabrication of dual-layer hollow fibers is more complicated than that of the single-layer ones. First of all, the morphology and structure of dual-layer hollow fibers become the focus of the concern of most researches. This is because the integrity of the fibers significantly affects the long-term stability during applications. Li et al. [177] investigated the morphology and structure of dual-layer hollow fibers by using Matrimid 5218 as the outer-layer and polyethersulfone (PES) or polyetherimide as the inner-layer. The higher shrinkage rate of the inner-layer than that of the outer-layer was identified as one of the key factors causing the interfacial delamination. They also showed that the formation of macrovoids in the outer-layer strongly depended on the viscosity of the outer-layer dope and the critical structuretransition thickness. They attempted to eliminate macrovoids in the inner-layer by means of additives and high elongational draw ratio [176,177]. Li et al. [187] fabricated polyethersulfone duallayer hollow fiber membranes with an ultra-thin dense-selective layer. No delamination was observed because the same material but different polymer concentration was used for 100 both inner and outer layers. The effects of fabrication conditions on separation performance of dual-layer hollow fibers are also of great interest. Jiang et al. [128] found the dope flow rates of both inner and outer layers affected gas separation performance of dual-layer hollow fiber membranes, especially the outer-layer dope flow rate. Recently, Chung and co-workers [188,189] extended their studies to duallayer mixed matrix hollow fibers using nanosize zeolite beta in order to further improve the gas separation performance. Previouslywehave successfully developed integrally skinned BTDA-TDI/MDI (P84) co-polyimide hollow fibers with superior performance for pervaporation dehydration of isopropanol (IPA) [127]. If this expensive material can be integrated as the outerlayer of a dual-layer hollow fiber membrane, the material cost can be significantly reduced by using a cheaper material as the inner-layer. As a consequence, the potential for industrial application can be appreciably enhanced. PES is purposely chosen in this study to construct the porous inner-layer because of its low cost, good mechanical properties and miscibility with aromatic polyimide [190,191]. However, due to the more complicated conditions in the coextrusion phase inversion process, one should choose proper spinning conditions and suitable post-treatments in order to achieve comparable performance with those of P84 single-layer hollow fibers. Therefore, the objectives of this paper are (1) to develop P84/PES dual-layer hollow fiber pervaporation membrane for IPA dehydration, (2) to investigate the effects of spinning conditions on P84/PES dual-layer hollow fiber membranes formation, morphology and pervaporation performance, and (3) to identify post-treatments such 101 as heat treatment and cross-linking modification and evaluate their effects on performance enhancement of P84/PES dual-layer hollow fiber membranes for pervaporation dehydration of IPA. 5.2 Results and Discussions 5.2.1 Effects of air gaps on membrane morphology and performance Figure 5.1 shows the SEM images of enlarged cross section of untreated hollow fibers (ID: A-1, A-3, A-4) spun at different air gap distances. No obvious delamination between outer and inner layers was observed. The delamination between outer and inner layers resulted from the different shrinkage rates during the phase inversion process is unfavorable because the delamination affects membrane’s integrity and structural stability [177]. This phenomenon can be properly controlled by the optimization of spinning parameters such as: (1) Material compatibility between the two dope fluids. PES has shown good interactions with most polyimides in molecular level [190,191]; therefore the mutual inter-layer diffusion may be helpful to construct a seamless interface between outer and inner layers [128,177]. (2) Proper dope concentrations of both inner and outer layers. A higher viscosity of dope fluid tends to form a more densified layer with a less shrinkage rate [10]; therefore, a high inner layer dope concentration, i.e. 32wt% PES in NMP, may be favorable to improve its adhesion with the outer layer. (3) An appropriate ratio of the outer to the inner layer dope flow 102 rates. In order to avoid mismatched shrinkage percentages between the inner and outer layers, one must optimize their dope flow ratio to balance the shrinkage strength in both layers [128,177]. Figure 5.1 The SEM images of untreated P84/PES dual layer hollow fibers spun at different air gap distances. (a) wet spun; (b) 10 mm air gap and (c) 15 mm air gap Figure 5.1 also illustrates that the outer diameters (OD) of dry-jet wet-spun dual-layer hollow fibers are much smaller than that of the wet spun one. The same phenomenon was observed in the fabrication of single layer P84 hollow fibers [1]. Moreover, the wet spun dual-layer hollow fibers have the thickest outer layer compared to the dry-jet wetspun fibers, and the outer layer thickness decreases when air gap distance increases. Similar to the single-layer P84 hollow fibers, the elongational stretch induced by 103 gravity in the air gap region during dry-jet wet-spun spinning process creates additional drawing effect which decreases both the outer layer thickness and the outer diameter of dual-layer hollow fibers. In the wet spinning process, this phenomenon is minimized because solidification occurs immediately after hollow fibers extruding into the coagulant [1,177,154,192]. Furthermore, as shown in Figure 5.1, macrovoids can be clearly observed in the outer layer of wet spun dual-layer hollow fibers. In contrast, no obvious macrovoids are detected in the outer layer of dry-jet wet-spun dual-layer hollow fibers. This phenomenon indicates that the macrovoids can be suppressed effectively by the elongational drawing created by gravity. The suppression of macrovoids in the outer layer of dry-jet wet-spun dual-layer fibers is probably due to (1) the elongational stretch induced radial outflow of spinning solvents from the nascent fiber [176]; (2) the elongational stretch induced more regular polymer chain packing which prevents the intrusion or diffusion of coagulants [154,192]; (3) the thickness of outer layer formed in dry-jet wet-spun spinning is less than the critical structure-transition thickness [177,193]. Figure 5.2 displays SEM images of enlarged cross sections of a wet-spun hollow fiber (ID: A-13), and a hollow fiber spun at a 15mm air gap (ID: A-12). Both of them were spun at a higher outer dope flow rate. It is clear that both the outer and inner layer structures are asymmetric even if the outer layer is very thin. Similar to the fibers spun 104 at a lower outer layer dope flow rate as shown in Figure 5.1, fibers spun at a higher outer layer dope flow rate have macrovoids in the outer layer of wet-spun fibers (ID: A13), whereas the macrovoids are suppressed in the fibers spun at 15mm air gap (ID: A12). The outward-pointed macrovoids in the outer layer of wet-spun fibers indicate that the fast precipitation with the aid of unbalanced localized stresses and non-solvent penetration from the coagulation bath are the major causes of the macrovoids [128,139, 173,187,194,195]. Furthermore, as shown in Figure 5.1(a) and Figure 5.2(a), the fingerlike voids in the inner layer of the wet-spun fibers have apparent openings towards the macrovoids in the outer layer, which probably suggests that the strong coagulant, water, may penetrate through the nascent outer layer into the inner layer, resulting in long and narrow finger-like passage ways throughout the inner layer. 105 Figure 5.2 The SEM images of cross-section (left) and enlarged outer selective layer (right) of P84/PES dual layer hollow fibers. (a) wet spun and (b) dry-jet wet-spun with 15mm air gap Figure 5.3 shows the pervaporation performance of the dual-layer hollow fibers spun at different air gap distances for the dehydration of 85% IPA/water solution. The flow rates of bore fluid, outer layer and inner layer dope were 0.2, 0.1 and 0.4 ml/min, respectively. It shows that the wet spun dual-layer hollow fibers (ID: A-1) have a higher separation factor/selectivity and higher flux/permeance than the dry-jet wet-spun ones (ID: A-3 ~ A-5). This phenomenon is very similar to the single layer asymmetric hollow fiber membranes [1]. Clearly, the instantaneous phase inversion of wet spinning 106 process facilitates the formation of an ultra-thin and less defective skin layer, while a long air gap distance tends to create defects [177]. Figure 5.3 Pervaporation performance of P84/PES dual layer hollow fibers spun at different air gap distances There are two competing factors induced by the elongational drawing in the air gap region: (1) a higher polymer chain orientation and packing may be formed which may enhance the membrane separation performance [154]. (2) The die swell and subsequently elongation-induced instability and phase separation may create defects on the selective layer [177]. The latter factor obviously dominates in the air gap distance from 10mm to 17mm. As a result, both permeance and selectivity decrease. 5.2.2 Effects of outer layer dope flow rate on membrane morphology and performance 107 Figure 5.4 shows SEM images of enlarged cross section of wet spun hollow fibers under different outer layer dope flow rates (ID: A-1, A-13). At a higher outer layer dope flow rate, a thicker outer layer is formed. An interesting phenomenon is that an almost uniform thickness of macrovoid-void free portion upon a layer full of finger-like voids was observed in the selective outer layers. This phenomenon may help to prove the existence of a critical structure-transition thickness under certain membrane formation conditions; if the membrane thickness exceeds the critical transition thickness, a macrovoid structure will form [177,193]. Interestingly, the thickness of the sponge-like structure portion of the outer layers remains about 12µm. It is very close to the critical thickness (i.e. 11±2µm) observed for P84 asymmetric flat membranes with a dope concentration of 20wt% [193]. Figure 5.4 The SEM images of enlarged cross section of wet spun P84/PES dual layer hollow fibers at bore fluid flow rate of 0.2 ml/min, and inner layer dope flow rate of 0.4 ml/min. (a) outer layer dope flow rate 0.1 ml/min; (b) outer layer dope flow rate 0.2 ml/min 108 Table 5.1 shows pervaporation performance of P84/PES dual-layer hollow fibers spun at different outer layer dope flow rates, while the bore fluid and inner layer dope flow rates were kept at the same. For the wet spun hollow fibers (ID: A-1, A-13), the flux decreases 40% and separation factor increases 44% when the outer layer dope flow rate was increased from 0.1 ml/min to 0.2 ml/min. The enhanced separation factor/selectivity may be attributed to shear induced better polymer chain packing at a higher outer layer dope flow rate. On the other hand, the flux/permeance is sacrificed due to a higher resistance at the selective layer. Interestingly, the increased outer layer dope flow rate does not show a beneficial effect on separation performance for the dry-jet wet-spun hollow fibers with an air gap of 15mm (ID: A-4, A-12). As shown in Table 4, the flux decreases 37% and the separation factor drops 51% when the outer layer dope flow rate increased from 0.1 ml/min to 0.2 ml/min. The higher outer layer dope flow rate incurs a higher resistance of the outer layer; meanwhile, it also creates greater gravitational drawing which is unfavorable due to increasing defects in the selective skin layer [177]. 109 110 5.2.3 Effects of inner layer dope flow rate on membrane morphology and performance Figure 5.5 shows the SEM images of P84/PES dual-layer hollow fiber spun at the outer-layer dope flow rate of 0.2 ml/min with different inner-layer dope flow rates. At a higher inner-layer dope flow rate, the outer layer thickness decreases, and the finger voids in the outer layer tends to disappear. This further demonstrates the macrovoids formation in the outer layer is thickness dependent and is influenced by the inner-layer dope flow rate; therefore the macrovoids in the outer layer of the P84/PES dual-layer hollow fibers can be suppressed under a higher inner layer dope flow rate. Figure 5.5 The SEM images of untreated P84/PES dual layer hollow fiber spun at outer layer dope 0.2 ml/min. (a) Wet spun, inner layer dope 0.8 ml/min; (b) 10 mm air gap, 111 inner layer dope 0.8 ml/min; (c) Wet spun, inner layer dope 0.4 ml/min; and (d) 10 mm air gap, inner layer dope 0.4 ml/min Figure 5.6 shows a pervaporation performance comparison between the dual-layer hollow fibers spun at different inner layer dope flow rates but with the same outer layer dope flow rate. For wet spun dual-layer hollow fibers, it appears that both flux and separation factor drop with an increase in inner layer dope flow rate. This may be arisen from a higher substructure (i.e., inner layer) resistance and more defects formed in the thinner outer selective layer because of the effects of a higher inner layer dope flow rate. The performance dependence on inner-layer dope flow rate seems to become less for fibers spun at an air gap distance of 15mm. As shown in Figure 8, the flux and separation factor do not vary much when the inner layer dope flow rate doubled. This indicates that other factors, such as the air gap distance and elongational stretch, and outer layer dope flow rate, may play more important roles than the inner layer dope flow rate on dual-layer membrane separation performance during the dry-jet wet-spun process. 112 Figure 5.6 Pervaporation performance of P84/PES dual layer hollow fibers spun at an outer layer dope flow rate of 0.2 ml/min and different inner layer dope flow rates 5.2.4 Comparison between P84/PES dual-layer membranes and P84 single layer membranes Table 5.1 also compares the pervaporation performance between as-spun P84 singlelayer hollow fibers [1] and P84/PES dual-layer hollow fibers spun at similar fabrication conditions. Impressively, the wet spun single-layer and dual-layer hollow fibers exhibit almost the same performance even though the latter has a slightly lower flux/permeance while slightly better separation factor/selectivity than the former. If we calculate the membrane material cost based on polymers and solvent consumption per unit membrane area (the unit price of P84, i.e. 110US$/kg, is about 6 times the unit price of PES), the material cost of the wet-spun P84/PES dual-layer hollow fibers is nearly half of the cost of P84 single-layer hollow fibers. This demonstrates that a significant 113 material cost saving can be achieved without sacrificing separation performance by choosing the dual-layer hollow fiber approach. However, when the air gap increases to 15mm, both flux/permeance and separation factor/selectivity of the single-layer fibers are superior to the dual-layer hollow fibers. Clearly, contradictory to the wet-spun single-layer and dual-layer hollow fibers where the outer skins may form vigorously and instantaneously because water is a powerful coagulant, it is harder to produce a defective-free outer selective skin in the dry-jet wetspun dual-layer hollow fiber. The complexity arises from the fact that air gap distance may have different effects on the inner and outer layers depending on their phase inversion rates as well as adverse effects on their separation performance. If the inner layer has a higher phase inversion rate than the outer layer, then the inner layer may bear more elongational stresses and have a higher elongation-induced orientation and chain packing (i.e., higher resistance), while the outer selective layer may be relaxed and less oriented. As a result, the combined pervaporation performance of the dry-jet wet-spun dual-layer hollow fibers is poorer than the single-layer ones. 5.2.5 Effects of heat treatment Since the as-spun fibers have very low separation factor/selectivity, heat treatment is employed to enhance their separation performance. For polyimide membranes, the heat treatment may induce chain relaxation and packing, and cross-linking reactions or the 114 formation of charge transfer complexes which are helpful to reduce free volume and defects, and restrict polymer chain mobility; therefore the membrane separation performance can be enhanced [41,53,196]. For P84 single-layer hollow fiber membranes, we have investigated their morphology changes after heat treatment, and reported a tremendous increase in separation factor/selectivity for IPA dehydration after a heat treatment at 300°C [1]. However, for P84/PES dual-layer hollow fibers, the effect of heat treatment is much more complicated because there are two materials involved. P84/PES dual-layer hollow fibers A-1 were chosen for heat treatment studies due to their relatively better separation performance. Figure 5.7 displays the evolution of densification in the P84 outer selective layer as a function of heat treatment temperature, whereas Figure 5.8 illustrates the morphology changes in the outer surface of PES inner layer with heat treatment temperature. Obviously, the pores in the outer surface of PES inner layer almost completely disappear when the heat treatment temperature is above 200°C. Figure 5.9 shows the pervaporation performances of heat-treated dual-layer hollow fiber membranes via different heat treatment protocols. After 200°C heat treatment, the flux/permeance decreases around 60% compared to the untreated ones. Meanwhile, the separation factor/selectivity increases 180% after heat treatment. Clearly, the densification of P84 skin layer and the significant reduction of pores in the PES inner layer contribute to the improved selectivity and decreased flux. However, when the heat treatment temperature is further increased, the membrane separation 115 performance does not show significant enhancement. Compared to the considerable selectivity improvements after heat treatment at 250◦C and 300°C for single-layer P84 hollow fiber membranes [1], the separation factor/selectivity enhancements for duallayer hollow fiber membranes under the same heat treatment protocols (250◦C and 300 ◦ C) are less impressive. Figure 5.7 The SEM images of outer layer cross section of P84/PES dual layer hollow fibers (ID: A-1) under different heat treatment protocols. (a) untreated; (b) 200˚ C; (c) 250˚ C; (d) 300˚ C Figure 5.8 The SEM images of inner-layer’s outer surface of P84/PES dual layer hollow fibers (ID: A-1) under different heat treatment protocols. (a) untreated; (b) 200° C; (c) 250° C; (d) 300° C 116 Figure 5.9 Pervaporation performance of P84/PES dual-layer hollow fibers (ID: A-1) under different heat treatment protocols As illustrated firstly by Pinnau and Koros [124] on gas separation membranes and then by Huang and Feng [125] on pervaporation, the selectivity of an asymmetric membrane depends not only on the resistance of the skin layer but also on the resistance of the substrate. When the substrate resistance becomes dominant, both permeance and selectivity can be considerably influenced. After P84/PES dual-layer fibers undergo the designated heat treatments, the resistance of PES sub-layer increases tremendously and dominates the separation process. This factor may offset the reduced defects in the P84 selective layer at a higher heat treatment temperature; therefore the overall separation performance enhancement is very limited. 5.2.6 Effects of p-xylenediamine cross-linking modification In order to circumvent the performance limitation caused by the densification of the inner layer during heat treatment, the cross-linking modification approach is adopted to 117 improve membrane separation performance as well as chemical and thermal stability [23,197-199]. It is believed that if an appropriate cross-linking agent can be found, which reacts only with the outer layer but is inert to the inner layer, then the limitation effects by the densification of inner layer on the overall separation performance can be minimized. P-xylenediamine was chosen to test our hypothesis because it has been applied as a cross-linking agent on 6FDA [57,58,200] and Matrimid [201] for gas separation and P84 single layer flat asymmetric membranes for pervaporation [59]. It was reported that this chemical cross-linking agent densified membrane skin layer, enhanced the membrane resistance to plasticization, increased the selectivity but decreased the permeability [34-38]. Since PES is inert to p-xylenediamine [202], the cross-linking reaction can be confined only in the P84 outer layer. Therefore, the chemical modification induced by p-xylenediamine for P84/PES dual-layer hollow fibers has the advantage of targeting the cross-linking reaction specifically in the outer layer without altering chemical composition and transport resistance in the inner layer. Figure 5.10 shows the pervaporation performances of p-xylenediamine cross-linked wet-spun P84/PES dual-layer hollow fibers with different immersion times. Clearly, the flux/permeance keeps decreasing with immersion time because of the resultant higher degree of cross-linking and densification in the selective layer. As illustrated in Figure 5.11, a thicker and more densified selective skin layer of the P84 outer layer can be 118 observed when the immersion time increases from 1 hour to 3 hours. Table 4 summarizes the XPS measurements of the N and O atomic concentrations, and N to O ratio upon the outer surface of cross-linked P84/PES dual-layer hollow fibers. Since the cross-linking reaction does not introduce oxygen or destroy oxygen from the polymer backbone, the oxygen content shall maintain at the same level, while the ratio of N/O indicates the degree of cross-linking [59]. As shown in Table 5.2, the N/O ratio increases with increasing immersion time, suggesting the degree of cross-linking increases. Figure 5.10 Pervaporation performance of p-xylenediamine cross-linked P84 / PES dual-layer hollow fibers (ID: A-13) with different immersion times 119 Figure 5.11 The SEM images of cross section of p-xylenediamine cross-linked P84/PES dual-layer hollow fibers (ID: A-13) with different immersion times. (a) 1 hour; (b) 2 hours; (c) 3 hours Table 5.2 Surface element concentrations of N and O of p-xylenediamine cross- linked P84/PES dual-layer hollow fibers and their concentration ratio An interesting phenomenon shown in Figure 5.10 is that the separation factor/selectivity increases significantly with cross-linking time within two hours, and then decreases with a prolonged immersion time. The initial increase in separation 120 factor is mainly attributed to the reduced defects by cross-linking and the densification of skin layer. The selectivity deterioration upon prolonged immersion time may be attributed to the enhanced substructure resistance within the outer layer because of high degrees of cross-linking reactions or due to the lesser rigid backbone of amide ring introduced by a higher degree of amidation which exhibits more polymer chain mobility during pervaporation process [40]. This reveals that cross-linking with pxylenediamine at a certain degree is helpful for the performance enhancement; however, too much cross-linking may not be favored. This phenomenon is consistent with the recent results of the cross-linked P84 flat asymmetric membranes [59]. Furthermore, the separation factor/selectivity of the 2-hr cross-linked P84/PES dual-layer hollow fibers increases 15 times while the flux decreases about 43% compared to the untreated ones, indicating a much better performance improvement than heat treatment for the pervaporation dehydration of IPA/water mixtures. 5.3 Conclusions BTDA-TDI/MDI co-polyimide (P84)/polyethersulfone (PES) dual-layer hollow fiber membranes were fabricated by the co-extrusion phase inversion process for the dehydration of IPA. Effects of spinning conditions and post treatments, including heat treatment and cross-linking modifications, on membranes morphology and pervaporation performance were investigated. The material cost saving of P84/PES 121 dual-layer hollow fiber was nearly half compared with P84 single-layer hollow fiber membranes. The following conclusions can be drawn: 1. Marcrovoids appear in the outer layer of wet-spun dual-layer hollow fibers, but are suppressed in the dry-jet wet-spun fibers. The finger-like voids in the inner layer of wet-spun fibers have apparent openings towards the outer layer, suggesting that the non-solvent penetration from outer layer may be a major cause of the macrovoids formation in the inner layer. Wet-spun fibers exhibit slightly better flux and higher separation factor than the dry-jet wet-spun fibers. 2. A higher outer layer dope flow rate enhances separation factor/selectivity however decreases flux/permeance of wet-spun fibers; while the increased outer layer dope flow rate has no beneficial effect for the dry-jet wet-spun fibers. 3. For wet-spun fibers, the increased inner layer dope flow rate creates a higher resistance of the inner layer and more defects in the outer layer. This results in the drop in both separation factor/selectivity and flux/permeance. For dry-jet wet-spun fibers, the inner layer dope flow rate does not apparently affect the pervaporation performance. 4. Heat treatment densifies the skin of P84 outer layer and eliminates the pores in the skin of PES inner layer. As a result, the flux/permeance is reduced and the separation factor/selectivity is enhanced. Further increased heat treatment temperature does not produce better separation performance because the drastically enhanced resistance of the PES sub-layer plays a dominant role in the separation process. 5. By reacting only with the P84 outer layer, p-xylenediamine cross-linking significantly increases the separation factor/selectivity of dual-layer hollow fibers. An 122 optimum crosslinking time of 2hr is observed. The p-xylenediamine cross-linking modification applied on P84/PES dual-layer hollow fiber shows a better performance improvement than heat treatment. 123 CHAPTER SIX CONCLUSIONS 6.1 Conclusions The development of hollow fiber membranes for IPA dehydration by pervaporation process were carried out in this study. It consisted of two parts: 1. Fabrication and characterization of BTDA-TDI/MDI (P84) co-polyimide hollow fiber membranes for the pervaporation dehydration of IPA. 2. Fabrication and characterization of BTDA-TDI/MDI (P84)/PES dual-layer hollow fiber membranes for the pervaporation dehydration of IPA. The following conclusions were derived from this study. 6.1.1 Fabrication and characterization of BTDA-TDI/MDI (P84) co-polyimide hollow fiber membranes for the pervaporation dehydration of IPA In Chapter 4, we studied P84 asymmetric hollow fiber pervaporation membranes fabricated by phase inversion process. Effects of various spinning conditions and post treatments on P84 hollow fiber membranes morphology and performance were investigated. The following conclusions can be drawn: Wet-spun hollow fibers have the highest separation factor and relatively high flux among fibers without any post treatments. A die-swell phenomenon was observed at an 124 air gap of 5 mm, and hollow fibers have a rough surface and apparently higher flux but lower separation factor. As air gap further increases, the flux decreases however the separation factor does not increase much. A thin layer silicone rubber coating increases separation factor about 3 times however the flux drops about 30% of its original value, indicating that a hydrophobic layer (silicon rubber coating) effectively seals big pores / defects however does not dominate the transport process. Heat treatment can densify the P84 hollow fibers’ skin layer, remarkably increases selectivity especially for dry-jet wet-spun fibers. With properly controlled spinning conditions and post treatments, high performance polymeric hollow fiber membranes were successfully developed by phase inversion process for pervaporation dehydration of IPA. For dehydration of 85 wt% IPA in water mixture at 60°C, the best performance, i.e. flux 883.5 g/m2hr and separation factor 10585, far exceeds previous polymeric hollow fiber pervaporation membranes for IPA dehydration. 6.1.2 Fabrication and characterization of BTDA-TDI/MDI (P84)/PES dual-layer hollow fiber membranes for the pervaporation dehydration of IPA In Chapter 5, we studied BTDA-TDI/MDI co-polyimide (P84)/polyethersulfone (PES) dual-layer hollow fiber membranes fabricated by the co-extrusion phase inversion process for the dehydration of IPA. Effects of spinning conditions and post treatments, including heat treatment and cross-linking modifications, on membranes morphology 125 and pervaporation performance were investigated. The material cost saving of P84/PES dual-layer hollow fiber was nearly half compared with P84 single-layer hollow fiber membranes. The following conclusions can be drawn: Marcrovoids appear in the outer layer of wet-spun dual-layer hollow fibers, but are suppressed in the dry-jet wet-spun fibers. The finger-like voids in the inner layer of wet-spun fibers have apparent openings towards the outer layer, suggesting that the non-solvent penetration from outer layer may be a major cause of the macrovoids formation in the inner layer. Wet-spun fibers exhibit slightly better flux and higher separation factor than the dry-jet wet-spun fibers. A higher outer layer dope flow rate enhances separation factor/selectivity however decreases flux/permeance of wet-spun fibers; while the increased outer layer dope flow rate has no beneficial effect for the dry-jet wet-spun fibers. For wet-spun fibers, the increased inner layer dope flow rate creates a higher resistance of the inner layer and more defects in the outer layer. This results in the drop in both separation factor/selectivity and flux/permeance. For dry-jet wet-spun fibers, the inner layer dope flow rate does not apparently affect the pervaporation performance. Heat treatment densifies the skin of P84 outer layer and eliminates the pores in the skin of PES inner layer. As a result, the flux/permeance is reduced and the separation factor/selectivity is enhanced. Further increased heat treatment temperature does not produce better separation performance because the drastically enhanced resistance of the PES sub-layer plays a dominant role in the separation process. By reacting only with the P84 outer layer, p-xylenediamine crosslinking significantly increases the separation factor/selectivity of dual-layer hollow 126 fibers. An optimum crosslinking time of 2hr is observed. The p-xylenediamine crosslinking modification applied on P84/PES dual-layer hollow fiber shows a better performance improvement than heat treatment. 127 REFERENCES [1] R.X. Liu, X.Y. Qiao, T.S. Chung, The development of high performance P84 copolyimide hollow fibers for pervaporation dehydration of isopropanol, Chemical Engineering Science 60 (2005) 6674. [2] R.X. Liu, X.Y. Qiao, T.S. Chung, Dual-layer P84/polyethersulfone hollow fibers for pervaporation dehydration of isopropanol, Journal of Membrane Science 294 (2007) 103. [3] Mulder, Marcel (1996). Basic principles of membrane technology (2 ed.). Kluwer Academic: Springer. ISBN 0-7923-4248-8. [4] Strathmann, H. Membrane separation processes. Journal of Membrane Science 9, (1981) 121. [5] Strathmann, H. Membrane separation processes: Current relevance and future opportunities. AIChE Journal 47, (2001) 1077. [6] Strathmann, H.; Gudernatsch, W. Pervaporation in biotechnology in R. Y. M. Huang edited Pervaporation Membrane Separation Processes. Elsevier Science Publisher B.V., New York (1991). 128 [7] Bravo, J. L.; Fair, J. R.; Humphrey, J. L.; Martin, C. L.; Seibert, A. F.; Joshi, S. Fluid Mixture Separation Technologies for Cost Reduction and Process Improvement. Noyes Publications, Park Ridge (1986). [8] Neel, J. Introduction to pervaporation, in R.Y.M. Huang edited Pervaporation Membrane Separation Processes, Elsevier Science Publisher B.V., New York.(1991). [9] Fleming, H. L., Slater, C. S. Pervaporation, in W.S.W. Ho and K.K. Sirkar edited, Membrane Handbook. New York: Van Nostrand Reinhold.(1992). [10] Crespo, J. G., Böddeker, K. W. Membrane Process in Separation and Purification. Kluwer Academic Publishers.(1995). [11] Feng, X., Huang, R. Y. M. Liquid separation by membrane pervaporation: a review. Industrial and Engineering Chemistry Research 36, (1997) 1048. [12] Moon, G. Y., Pal, R., Huang, R. Y. M. Novel two-ply composite membranes of chitosan and sodium alginate for the pervaporation dehydration of isopropanol and ethanol. Journal of Membrane Science 156, (1999) 17. 129 [13] Kober, P. A. Pervaporation, perstillation and percrystallization. Journal of American Chemistry Society 39, (1917) 940. [14] Farber, L. Applications of pervaporation. Science 82, (1935) 158. [15] Heisler, E. G., Hunter, A. S., Siciliano, J., Treadway, R. H. Solute and temperature effects in the pervpaoration of aqueous alcoholic solutions. Science 124, (1956) 77. [16] Ji, W., Sikdar, S. K., Hwang, S. T. (1994). Modeling of multicomponent pervaporation for removal of volatile organic compounds from water. Journal of Membrane Science 93, 1. [17] Lipnizki, F.; Field, R. W.; Ten, P. K. Pervaporation-based hybrid process: a review of process design, applications and economics. Journal of Membrane Science 153 (1999) 183. [18] Villaluenga, J. P. G., Tabe-Mohammadi, A. A review on the separation of benzene/cyclohexane mixtures by pervaporation processes. Journal of Membrane Science 169, (2000) 159. [19] Peng, M., Vane, L. M., Liu, S. X. Recent advances in VOCs removal from water by pervaporation. Journal of Hazardous Materials B98, (2003) 69. 130 [20] Bowen, T. C., Noble, R. D., Falconer, J. L. Fundamentals and applications of pervaporation through zeolite membranes. Journal of Membrane Science 245, (2004) 1. [21] Smitha, B.; Suhanya, D.; Sridhar, S.; Ramakrishna, M. Separation of organicorganic mixtures by pervaporation-a review. Journal of Membrane Science 241 (2004) 1. [22] Shao, P.; Huang, R. Y. M. Polymeric membrane pervaporation. Journal of Membrane Science 287 (2007) 162. [23] Yu, J., Lee, C. H., Hong, W. H. Performances of crosslinked asymmetric poly(vinyl alcohol) membranes for isopropanol dehydration by pervaporation. Chemical Engineering and Processing 41,(2002) 693. [24] Guan, H. M., Chung, T. S., Huang, Z., Chng, M. L., Kulprathipanja, S. Poly(vinyl alcohol) multilayer mixed matrix membranes for the dehydration of ethanol-water mixture. Journal of Membrane Science 268, (2006)113. [25] Namboodiri, V. V.; Ponangi, R.; Vane, L. M. A novel hydrophilic polymer membrane for the dehydration of organic solvents. Europ. Polym. J. 42 (2006) 3390. 131 [26] Namboodiri, V. V.; Vane, L. M. High permeability membranes for the dehydration of low water content ethanol by pervaporation. Journal of Membrane Science 306 (2007) 209. [27] Gao, Z., Yue, Y., Li, W. Application of zeolite filled pervaporation membrane. Zeolite 19, (1996) 70. [28] Qiao, X. Y.; Chung, T. S.; Rajagopalan, R. Zeolite filled P84 co-polyimide membranes for dehydration of isopropanol through pervaporation process. Chem. Eng. Sci. 61 (2006) 6816. [29] Wang, C.; Hua, W. M.; Yue, Y. H.; Gao, Z. Delaminated microporous aluminophosphate-filled polyvinyl alcohol membrane for pervaporation of aqueous alcohol solutions. Micropor. Mesopor. Mater. 105 (2007) 149. [30] Guo, R. L.; Hu, C.; Pan, F. S.; Wu, H.; Jiang, Z. Y. PVA-GPTMS/TEOS hybrid pervaporation membrane for dehydration of ethylene glycol aqueous solution. Journal of Membrane Science 281 (2006) 454. 132 [31] Adoor, S. G.; Prathab, B.; Manjeshwar, L. S.; Aminabhavi, T. M. Mixed matrix membranes of sodium alginate and poly(vinyl alcohol) for pervaporation dehydration of isopropanol at different temperatures. Polymer 48 (2007) 5417 [32] Nawawi, M. G. M., Huang, R. Y. M. Pervaporation dehydration of isopropanol with chitosan membranes. Journal of Membrane Science 124, (1997) 53. [33] Huang, R. Y. M., Pal, R., Moon, G. Y. Characteristics of sodium alginate membranes for the pervaporation dehydration of ethanol-water and isopropanol-water mixtures. Journal of Membrane Science 160, (1999) 101. [34] Lee, C. H., Hong, W. H. Influence of different degrees of hydrolysis of poly(vinyl alcohol) membrane on transport properties in pervaporation of IPA/water mixture. Journal of Membrane Science 135, (1997) 187. [35] Chanachai, A., Jiraratananon, R., Uttapap, D., Moon, G. Y., Anderson, W. A., Huang, R. Y. M. Pervaporation with chitosan/hydroxyethyl cellulose (CS/HEC) blended membranes. Journal of Membrane Science 166, (2000) 271. [36] Sridhar, S., Ganga, D., Smitha, B., Ramakrishna, M. Dehydration of 2-butanol by pervaporation through blend membranes of chitosan and hydroxy ethyl cellulose. Separation Science and Technology 40, (2005) 2889. 133 [37] Ahmad, A. L., Nawawi, M. G. M., So, L. K. Characterization and performance evaluations of sodium zeolite-Y filled chitosan polymeric membrane: Effect of sodium zeolite-Y concentration. Journal of Applied Polymer Science 99, (2006) 1740. [38] K.Y. Wang, T.S. Chung, R. Rajagopalan, Dehydration of tetrafluoropropanol (TFP) by pervaporation via novel PBI/BTDA-TDI/MDI co-polyimide (P84) dual-layer hollow fiber membranes, Journal of Membrane Science 287 (2007) 60. [39] Tanaka, K.; Okamoto, K. I. Structure and transport properties of polyimides as materials for gas and vapor membrane separation in Y. Yampolskii, I. Pinnau and B. D. Freeman edited Materials Science of Membranes for Gas and Vapor Separation. John Wiley & Sons Ltd, Chichester (2007). [40] Ohya, H., Kudryavtsev, V. V., Semenova, S. I. Polyimide membranesapplications, fabrications, and properties, Gordon and Breach Publishers.(1996). [41] Qiao, X. Y., Chung, T. S., Pramoda, K. P. Fabrication and characterization of BTDA-TDI/MDI (P84) co-polyimide membranes for the pervaporation dehydration of isopropanol. Journal of Membrane Science 264, (2005) 176. 134 [42] Yanagishita, H., Maejima, C., Kitamoto, D., Nakane, T. Preparation of asymmetric polyimide membrane for water/ethanol separation in pervaporation by phase inversion process. Journal of Membrane Science 86, (1994) 231. [43] Yanagishita, H., Kitamoto, D., Haraya, K., Nakane, T., Tsuchiya, T., Koura, N. Preparation and pervaporation performance of polyimide composite membrane by vapor deposition and polymerization (VDP). Journal of Membrane Science 136, (1997) 121. [44] Yanagishita, H., Kitamoto, D., Haraya, K., Nakane, T., Okada, T., Mastyda, H., Idemoto, Y., Koura, N. Separation performance of polyimide composite membrane prepared by dip coating process. Journal of Membrane Science 188, (2001) 165. [45] Fan, S. C. A study of polyamide and polyimide membranes for pervaporation and vapor permeation, PhD thesis. Chun Yuan University, Taiwan. (2002). [46] Guo, W. F., Chung, T. S. Study and characterization of the hysteresis behavior of polyimide membranes in the thermal cycle process of pervaporation separation. Journal of Membrane Science 253, (2005) 13. 135 [47] Yoshikawa, M., Yokoi, H., Sanui, K., Ogata, N., Shimidzu, T. . Polymer membrane as a reaction field. II. Effect of membrane environment on permselectivity for water/ethanil binary mixtures. Polymer Journal 16, (1984) 653. [48] Chung, T. S., Guo, W. F., Liu, Y. Enhanced Matrimid membranes for pervaporation by homogeneous blends with polybenzimidazole (PBI). Journal of Membrane Science 271, (2006) 221. [49] Chung, T. S. A review of microporous composite polymeric membrane technology for air-separation. Polymers and Polymer Composites 4, (1996) 269. [50] Tsai, H. A., Ciou, Y. S., Hu, C. C., Lee, K. R., Yu, D. G., Lai, J. Y. Heattreatment effect on the morphology and pervaporation performances of asymmetric PAN hollow fiber membranes. Journal of Membrane Science 255, (2005) 33. [51] Zhou, F., Koros, W. J. Study of thermal annealing on Matrimid fiber performance in pervaporation of acetic acid and water mixtures. Polymer 47, (2006) 280. [52] Hasegawa, M., Horie, K. Photophysics, photochemistry, and optical properties of polyimides. Progress in Polymer Science 26, (2001) 259. [53] Kawakami, H., Mikawa, M., Nagaoka, S. Gas transport properties in thermally cured aromatic polyimide membranes. Journal of Membrane Science 118, (1996) 223. 136 [54] Bos, A., Punt, I. G. M., Wessling, M., Strathmann, H. Plasticization-resistant glassy polyimide membranes for CO2/CH4 separations. Separation and Purification Technology 14, (1998) 27. [55] Shao, L., Chung, T. S., Goh, S. H., Pramoda, K. P. The effects of 1,3cyclohexanebis(methylamine) modification on gas transport and plasticization resistance of polyimide membranes. Journal of Membrane Science 267, (2005) 78. [56] Shao, L., Chung, T. S., Goh, S. H., Pramoda, K. P. Polyimide modification by a linear aliphatic diamine to enhance transport performance and plasticization resistance. Journal of Membrane Science 256, (2005) 46. [57] Liu, Y., Wang, R., Chung, T. S. Chemical cross-linking modification of polyimide membranes for gas separation. Journal of Membrane Science 189, (2001) 231. [58] Cao, C., Chung, T. S., Liu, Y., Wang, R., Pramoda, K. P. Chemical cross-linking modification of 6FDA-2,6-DAT hollow fiber membranes for natural gas separation. Journal of Membrane Science 216, (2003) 257. 137 [59] X.Y. Qiao, T.S. Chung, Diamine modification of P84 polyimide membranes for pervaporation dehydration of isopropanol, AIChE J. 52 (2006) 3462. [60] Jiang, L. Y.; Chung, T. S.; Rajagopalan, R. Dehydration of alcohols by pervaporation through polyimide Matrimid® asymmetric hollow fibers with various modifications. Chem. Eng. Sci. 63 (2008) 204. [61] Jiang, L. Y.; Chung, T. S.; Rajagopalan, R. Matrimid®/MgO mixed matrix membranes for pervaporation. AIChE J. 53 (2007) 1745. [62] Huang, R. Y. M., Shao, P., Feng, X., Burns, C. M. Pervaporation separation of water/isopropanol mixture using sulfonated poly(ether ether ketone) (SPEEK) membranes: transport mechanism and separation performance. Journal of Membrane Science 192, (2001) 115. [63] Tu, C. Y.; Liu, Y. L.; Lee, K. R.; Lai, J. Y. Hydrophilic surface-grafted poly(tetrafluoroethylene) membranes using in pervaporation dehydration processes. Journal of Membrane Science 274 (2006) 47. [64] Shah, D., Kissich, K., Ghorpade, A., Hannah, R., Bhattacharyya, D. Pervaporation of alcohol-water and dimethylformamide-water mixtures using 138 hydrophilic zeolite NaA membranes: mechanisms and experimental results. Journal of Membrane Science 179, (2000) 185. [65] Verkerk, A. W., Male, P. v., Vorstman, M. A. G., Keurentjes, J. T. F. Properties of high flux ceramic pervaporation membranes for dehydration of alcohol/water mixtures. Separation and Purification Technology 22, (2001) 689. [66] Bowen, T. C., Li, S., Noble, R. D., Falconer, J. L. Driving force for pervaporation through zeolite membranes. Journal of Membrane Science 225, (2003) 165. [67] Gallego-Lizon, T., Edwards, E., Lobiundo, G., Santos, L. F. d. Dehydration of water/t-butanol mixtures by pervaporation: comparative study of commercially available polymeric, microporous silica and zeolite membranes. Journal of Membrane Science 197, (2002) 309. [68] Gallego-Lizon, T., Ho, Y. S., Santos, L. F. d. Comparative study of commercially available polymeric and microporous silica membranes for the dehydration of IPA/water mixtures by pervaporation/vapour permeation. Desalination 149,(2002) 3. 139 [69] Sato, K.; Nakane, T. A high reproducible fabrication method for industrial production of high flux NaA zeolite membrane. Journal of Membrane Science 301 (2007) 151. [70] Feng, X., Huang, R. Y. M. Separation of isopropanol from water by pervaporation using silicone-based membranes. Journal of Membrane Science 74, (1992) 171. [71] Hong, Y. K., Hong, W. H. Influence of ceramic support on pervaporation characteristics of IPA/water mixtures using PDMS/ceramic composite membrane. Journal of Membrane Science 159, (1999) 29. [72] Gupta, T., Pradhan, N. C., Adhikari, B. Separation of phenol from aqueous solution by pervaporation using HTPB-based polyurethaneurea membrane. Journal of Membrane Science 217, (2003) 43. [73] Yoshikawa, M., Wano, T., Kitao, T. Specialty polymeric membranes. 1. Modified polybutadiene membranes for alcohol separation. Journal of Membrane Science 76, (1993) 255. [74] Fadeev, A. G.; Selinskaya, Y. A.; Kelley, S. S.; Meagher, M. M.; Litvinova, E. G.; Khotimsky, V. S.; Volkov, V. V. Extraction of butanol from aqueous solutions by 140 pervaporation through poly(1-trimethylsilyl-1-propyne). Journal of Membrane Science 186 (2001) 205. [75] Gonzalez-Velasco, J. R.; Gonzalez-Marcos, J. A.; Lopez-Dehesa, C. Pervaporation of ethanol-water mixtures through poly(1-trimethylsilyl-1-propyne) (PTMSP) membranes. Desalination 149 (2002) 61. [76] Lipnizki, F.; Hausmanns, S.; Ten, P. K.; Field, R. W.; Laufenberg, G. Organophilic pervaporation: prospects and performance. Chem. Eng. J. 73 (1999) 113. [77] Vane, L. M. A review of pervaporation for product recovery from biomass fermentation processes. J. Chem. Technol. Biotechnol. 80 (2005) 603. [78] Uragami, T.; Ohshima, T.; Miyata, T. Removal of benzene from an aqueous solution of dilute benzene by various cross-linked poly(dimethylsiloxane) membranes during pervaporation. Macromolecules 36 (2003) 9430. [79] Song, K. M.; Hong, W. H. Dehydration of ethanol and isopropanol using tubular type cellulose acetate membrane with ceramic support in pervaporation process. Journal of Membrane Science 123 (1997) 27. 141 [80] Xiangli, F. J.; Chen, Y. W.; Jin, W. Q.; Xu, N. P. Polydimethylsiloxane (PDMS)/ceramic composite membrane with high flux for pervaporation of ethanolwater mixtures. Ind. Eng. Chem. Res. 46 (2007) 2224. [81] Nagase, Y.; Ando, T.; Yun, C. M. Syntheses of siloxane-grafted aromatic polymers and the application to pervaporation membrane. Reactive Funct. Polym. 67 (2007) 1252. [82] Yamasaki, A.; Shinbo, T.; Mizoguchi, K. Pervaporation of benzene/cyclohexane and benzene/n-hexane mixtures through PVA membrane. J. Appl. Polym. Sci. 64 (1997) 1061. [83] Inui, K.; Tsukamoto, K.; Miyata, T. Permeation and separation of a benzene/cyclohexane mixture through benzoylchitosan membranes. J. Membr. Sci. 138 (1998) 67. [84] Peng, F. B.; Pan, F. S.; Sun, H. L.; Lu, L. Y.; Jiang, Z. Y. Novel nanocomposite pervaporation membranes composed of poly(vinyl alcohol) and chitosan-wrapped carbon nanotube. J. Membr. Sci. 300 (2007) 13. 142 [85] Peng, F. B.; Hu, C. L.; Jiang, Z. Y. Novel poly(vinyl alcohol)/carbon nanotube hybrid membranes for pervaporation separation of benzene/cyclohexane mixtures. J. Membr. Sci. 297 (2007) 236. [86] Nam, S. Y.; Dorgan, J. R. Non-equilibrium nanoblends via forced assembly for pervaporation separation of benzene from cyclohexane: UNIFAQ-FV group contribution calculations. J. Membr. Sci. 306 (2007) 186. [87] Jonquieres, A., Clement, R., Lochon, P., Neel, J., Dresch, M., Chretien, B. Industrial state-of-the-art of pervaporation and vapour permeation in the western countries. Journal of Membrane Science 206, (2002) 87. [88] Blume, I., Wijmans, J. G., Baker, R. W. The Separation of Dissolved Organics from Water by Pervaporation. Journal of Membrane Science 49, (1990) 253. [89] O'Brien, D. J.; Roth, L. H.; McAloon, A. J. Ethanol production by continuous fermentation-pervaporation: a preliminary economic analysis. J. Membr. Sci. 166 (2000) 105. [90] Nam, S. Y., Lee, Y. M. Pervaporation separation of methanol/methyl t-butyl ether through chitosan composite membrane modifed with surfactants. Journal of Membrane Science 157, (1999) 63. 143 [91] Yoshikawa, M., Yoshioka, T., Fujime, J., Murakami, A. Pervaporation separation of MeOH/MTBE through agarose membranes. Journal of Membrane Science 178, (2000) 75. [92] Wang, Y. C., Li, C. L., Huang, J., Lin, C., Lee, K. R., Liaw, D. J., Lai, J. Y. Pervaporation of benzene/cyclohexane mixtures through aromatic polyamide membranes Journal of Membrane Science 185, (2001) 193. [93] Knight, K. F., Duggal, A., Shelden, R. A., Thompson, E. V. Dependence of diffusive permeation rates and selectivities on upstream and downstream pressures : V. Experimental results for the hexane/heptane (ideal) and toluene/ethanol (nonideal) systems. Journal of Membrane Science 26, (1986) 31. [94] Wessling, M., Werner, U., Hwang, S. T. Pervaporation of aromatic C8-isomers Journal of Membrane Science 57, (1991) 257. [95] Miyata, T.; Iwanoto, T.; Uragami, T. Characteristics of permeation and separation for propanol isomers through poly(vinyl alcohol) membranes containing cyclodextrin. J. Appl. Polym. Sci. 51 (1994) 2007. 144 [96] Kusumocahyo, S. P., Kanamori, T., Sumaru, K., Iwatsubo, T., Shinbo, T. Pervaporation of xylene isomer mixture through cyclodextrins containing polyacrylic acid membranes Journal of Membrane Science 231, (2004) 127. [97] Paris, J., Molina-Jouve, C., Nuel, D., Moulin, P., Charbit, F. Enantioenrichment by pervaporation. Journal of Membrane Science 237, (2004) 9. [98] Wijmans, J. G., Baker, R. W. The solution-diffusion model: a review. Journal of Membrane Science 107,(1995) 1. [99] Yeom, C. K., Lee, S. H., Lee, J. M. Pervaporative permeations of homologous series of alcohol aqueous mixtures through a hydrophilic membrane. Journal of Applied Polymer Science 79, (2001) 703. [100] Carretier, E., Moulin, P., Beaujean, M., Charbit, F. Purification and dehydration of methylal by pervaporation. Journal of Membrane Science 217 (2003) 159. [101] Huang, R. Y. M., Yeom, C. K. Pervaporation Separation of aqueous mixtures using crosslinked poly(vinyl alcohol) (PVA). II. Permeation of ethanol-water mixtures. Journal of Membrane Science 51, (1990) 273. 145 [102] Huang, R. Y. M., Feng, X. Dehydration of isopropanol by pervaporation using aromatic polyetherimide membranes. Separation Science and Technology 28, (1993) 2035. [103] Wijmans, J. G., Baker, R. W. A simple predictive treatment of the permeation process in pervaporation. Journal of Membrane Science 79, (1993) 101. [104] Wijmans, J. G. Process performance = membrane properties +operating conditions. Journal of Membrane Science 220, (2003) 1. [105] Guo, W. F., Chung, T. S., Matsuura, T. Pervaporation study on the dehydration of aqueous butanol solutions: a comparison of flux vs. permeance, separation factor vs. selectivity. Journal of Membrane Science 245, (2004) 199. [106] Qiao, X. Y., Chung, T. S., Guo, W. F., Matsuura, T., Teoh, M. M. Dehydration of isopropanol and its comparison with dehydration of butanol isomers from thermodynamic and molecular aspects. Journal of Membrane Science 252, (2005) 37. [107] Mulder, M. H. V. Thermodynamic principles of pervaporation, in R. Y. M. Huang edited, Pervaporation Membrane Separation Processes. Amsterdam, Elsevier Science Publishers. (1991). 146 [108] Matsuura, T. Synthetic Membranes and Membrane Separation Processes, CRC Press, Boca Raton, Florida. (1993). [109] Yeom, C. K.; Lee, S. H.; Lee, J. M. Pervaporative permeations of homologous series of alcohol aqueous mixtures through a hydrophilic membrane. J. Appl. Polym. Sci. 79 (2001) 703. [110] Barton, A. F. M. Handbook of Solubility Parameters and Other Cohesion Parameters. Florida, USA, CRC Press: Boca Raton.(1990). [111] Flory, P. J. Principles of Polymer Chemistry. Ithaca, NY, Cornell University Press. (1953). [112] Jonquieres, A., Perrin, L., Durand, A., Arnold, S., Lochon, P. Modelling of vapour sorption in polar materials: Comparison of Flory-Huggins and related models with the ENSIC mechanistic approach. Journal of Membrane Science 147, (1998) 59. [113] Jonquieres, A., Roizard, D., Lochon, P. Use of empirical polarity parameters to describe polymer-liquid interactions: correlation of polymer swelling with solvent polarity in binary and ternary systems. Journal of Applied Polymer Science 54, (1994) 1673. 147 [114] Jonquieres, A., Roizard, D., Cuny, J., Lochon, P. Solubility and polarity parameters for assessing pervaporation and sorption properties. A critical comparison for ternary systems alcohol/ether/polyurethaneimide. Journal of Membrane Science 121, (1996) 117. [115] Reichardt, C. Solvents and Solvent Effects in Organic Chemistry, 2nd ed. Weinheim, VCH. (1998). [116] Bode, E., Hoempler, C. Transport resistances during pervaporation through a composite membrane: experiments and model calculations. Journal of Membrane Science 113, (1996) 43. [117] Drioli, E., Zhang, S., Basile, A. On the coupling effect in pervaporation. Journal of Membrane Science 81, (1993) 43. [118] Schaetzel, P., Vauclair, C., Luo, G., Nguyen, Q. T. The solution-diffusion model order of magnitude calculation of coupling between the fluxes in pervaporation. Journal of Membrane Science 191, (2001) 103. [119] Qiao, X. Y., Chung, T. S. Fundamental characteristics of sorption, swelling, and permeation of P84 co-polyimide membranes for pervaporation dehydration of alcohols. Industrial & Engineering Chemistry Research 44, (2005) 8938. 148 [120] Smit, E.; Mulder, M. H. V.; Smolders, C. A.; Karrenbeld, H.; Eerden, J. van; Feil, D. Modeling of the diffusion of carbon dioxide in polyimide matrices by computer simulation. J. Membr. Sci. 73 (1992) 247. [121] Peng, F. B.; Lu, L. Y.; Hu, C. L.; Wu, H.; Jiang, Z. Y. Significant increase of permeation flux and selectivity of poly(vinyl alcohol) membranes by incorporation of crystalline flake graphite. J. Membr. Sci. 259 (2005) 65. [122] Peng, F. B.; Lu, L. Y.; Sun, H. L.; Wang, Y. Q.; Liu, J. Q.; Jiang, Z. Y. Hybrid organic-inorganic membranes: solving the trade-off between permeability and selectivity. Chem. Mater. 17 (2005) 6790. [123] Koros, W. J., Fleming, G. K., Jordan, S. M., Kim, T. H., Hoehn, H. H. Polymeric membrane materials for solution-diffusion based permeation separations. Progress in Polymer Science 13, (1988) 339. [124] Pinnau, I., Koros, W. J. Relationship between substructure resistance and gas separation properties of defect-free integrally-skinned asymmetric membranes. Industrial & Engineering Chemistry Research 20, (1991) 1837. 149 [125] R.Y.M. Huang, X. Feng, Resistance model approach to asymmetric polyetherimide membranes for pervaporation of isopropanol/water mixtures, J. Membr. Sci. 34 (1993) 15. [126] Ekiner, O. M.; Hayes, R. A.; Manos, P. Novel multicomponent fluid separation membranes. U.S. patent (1992). [127] Li, D. F.; Chung, T. S.; Wang, R.; Liu, Y. Fabrication of fluoropolyimide / polyethersulfone (PES) dual-layer asymmetric hollow fiber membranes for gas Separation. J. Membr. Sci. 198 (2002) 211. [128] L.Y. Jiang, T.S. Chung, D.F. Li, C. Cao, S. Kulprathipanja, Fabrication of Matrimid/polyethersulfone dual-layer hollow fiber membranes for gas separation, J. Membr. Sci. 240 (2004) 91. [129] Widjojo, N.; Zhang, S. D.; Chung, T. S.; Liu, Y. Enhanced gas separation performance of dual-layer hollow fiber membranes via substructure resistance reduction using mixed matrix materials. J. Membr. Sci. 306 (2007) 147-158. [130] Kesting, R. E. Synthetic Polymetic Membranes: A Structure Perspective. New York, John Wiley & Sons, Inc.(1985). 150 [131] Pinnau, I., Koros, W. J. Influence of quench medium on the structures and gas permeation properties of polysulfone membranes made by wet and dry/wet phase inversion. Journal of Membrane Science 71, (1992) 81. [132] Paulsen, F. G., Shojaie, S. S., Krantz, W. B. Effect of evaporation step on macrovoid formation in wet-cast polymeric membranes. Journal of Membrane Science 91, (1994) 265. [133] Niwa, M.; Kawakami, H.; Nagaoka, S.; Kanamori, T.; Shinbo, T. Fabrication of an asymmetric polyimide hollow fiber with a defect-free surface skin layer. J. Membr. Sci. 171 (2000) 253. [134] Peng, N.; Chung, T. S. The effects of spinneret dimension and hollow fiber dimension on gas separation performance of ultra-thin defect-free Torlon® hollow fiber membranes. J. Membr. Sci. 310 (2008) 455. [135] Leob, S., Sourirajan, S. Sea water demineralization by means of an osmotic membrane, in R. F. Gould edited. Saline water conversation 2, ACS Symposium Series 38, American Chemical Society, Washington, D.C., U.S.A. pg 117.(1963). [136] Pinnau, I.; Koros, W. J. Defect-free ultrahigh flux asymmetric membranes. U.S. Patent 4,902,422, (1990). 151 [137] Chung, T. S.; Kafchinski, E. R.; Vora, R. Development of a defect-free 6FDAdurene asymmetric hollow fiber and its composite hollow fibers. J. Membr. Sci. 88 (1994) 21. [138] Clausi, D. T.; Koros, W. J. Formation of defect-free polyimide hollow fiber membranes for gas separations. J. Membr. Sci. 167 (2000) 79. [139] Chung, T. S., Teoh, S. K., Hu, X. Formation of ultrathin high performance polyethersulfone hollow-fiber membranes. Journal of Membrane Science 133, (1997) 161. [140] Maeda, Y., Kai, M. Recent progress in pervaporation membranes for water/ethanol separation, in R. Y. M. Huang edited, Pervaporation Membrane Separation Processes. Elsevier Science Publishers B. V. (1991). [141] Jafar, J. J.; Budd, P. M. Separation of alcohol/water mixtures by pervaporation through zeolite A membranes. Microporous Mater. 12 (1997) 305. [142] Reuvers, A. J. Membrane Formation: diffusion induced demixing processes in ternary systems, PhD Thesis. The Netherlands, Twente University of Technology.(1987). 152 [143] Pinnau, I. Skin formation of integral-asymmetric gas separation membranes made by dry/wet phase inversion. Texas, University of Texas at Austin. PhD Thesis.(1991). [144] Tompa, H. Polymer Solutions. London, Butterworths.(1956). [145] Kamide, K. Thermodynamics of Polymer Solutions. Amsterdan, The Netherlands, Elsevier Science Publishers B. V. (1990). [146] Lai, J. Y., Lin, S. F., Lin, F. C., Wang, D. M. Construction of ternary phase diagrams in nonsolvent/solvent/PMMA systems. Journal of Polymer Science: Part B: Polymer Physics 36, (1998) 607. [147] Mehta, R. H., Madsen, D. A., Kalika, D. S. Microporous membranes based on poly(ether ether ketone) via thermally-induced phase separation. Journal of Membrane Science, 107, (1995) 93. [148] Matsuyama, H., Berghmans, S., Lloyd, D. R. Formation of hydrophilic microporous membranes via thermally induced phase separation. Journal of Membrane Science 142 (1998) 213. 153 [149] Shojaie, S. S., Krantz, W. B., Greenberg, A. R. Dense polymer film and membrane formation via the dry-cast process part I. Model development. Journal of Membrane Science 94 (1994) 255. [150] Nagy, E., Borlai, O., Stelmaszek, J. Pervaporation of alcohol-water mixtures on cellulose hydrate membranes. Journal of Membrane Science 16, (1983) 79. [151] Chung, T. S., Chng, M. L., Pramoda, K. P., Xiao, Y.C.PAMAM dendrimerinduced cross-linking modification of polyimide membranes. Langmuir 20, (2004) 2966. [152] Xiao, Y., Shao, L., Chung, T. S., Schiraldi, D. A. Effects of thermal treatments and dendrimers chemical structures on the properties of highly surface cross-linked polyimide films. Industrial and Engineering Chemistry Research 44, (2005) 3059. [153] S.J. Shilton, Flow profile induced in spinneret during hollow fiber membrane spinning, J. Appl. Polym. Sci. 65 (1997) 1359. [154] C. Cao, T.S. Chung, S.B. Chen, Z.J. Dong, The study of elongation and shear rates in spinning process and its effect on gas separation performance of poly(ether sulfone) (PES) hollow fiber membranes, Chem. Eng. Sci. 59 (2004) 1053. 154 [155] K.Y. Wang, T. Matsuura, T.S. Chung, W.F. Guo, The effects of flow angle and shear rate within the spinneret on the separation performance of poly(ethersulfone) (PES) ultrafiltration hollow fiber membrane, J. Membr. Sci. 240 (2004) 67. [156] Hirata, M., Ohe, S., Nagahama, K. Computer aided data book of vapor-liquid equilibria,. Tokyo, Kodansha Limited and Elsevier Scientific Publishing Company.(1975). [157] Kondo, M., Yamamura, T., Yukitake, T., Matsuo, Y., Kita, H., Okamoto, K. IPA purification for lens cleaning by vapor permeation using zeolite membrane. Separation and Purification Technology 32, (2003) 191. [158] Gmehling, J., Onken, U., Arlt, W. Vapor-liquid equilibrium data collection, Vol. 1, Part 1. Dechema, Frankfurt. (1991). [159] W. Heinzelmann, Fabrication methods for pervaporation membranes, in R. Bakish (Ed.), Proceedings of 5th International Conference on Pervaporation Processes in the Chemical Industry, Bakish Materials Corporation, 1991. [160] G.H. Koops, J. Nolten, M.H.V. Mulder, C.A. Smolders, Integrally skinned polysulfone hollow-fiber membranes for pervaporation, J. Appl. Polym. Sci. 54 (1994) 385. 155 [161] A.F. Ismail, L.P. Yean, Review on the development of defect-free and ultrathinskinned asymmetric membranes for gas separation through manipulation of phase inversion and rheological factors, J. Appl. Polym. Sci. 88 (2003) 442. [162] M. Tsuyumoto, A. Teramoto, P. Meares, Dehydration of ethanol on a pilot-plant scale, using a new type of hollow-fiber membrane, J. Membr. Sci. 133 (1997) 83. [163] I. Cabasso, Z.Z. Liu, T. Makenzie, The permselectivity of ion-exchange membranes non-electrolyte liquid mixtures II the effect counterions (separation of alcohol/water mixtures with nafion membranes), J. Membr. Sci. 28 (1986) 109. [164] D.R. Seok, S.G. Kang, S.T. Hwang, Use of pervaporation for separation azeotropic mixtures using two different hollow fiber membranes, J. Membr. Sci. 33 (1987) 71. [165] Z.K. Xu, Q.W. Dai, Z.M. Liu, R.Q. Kou, Y.Y. Xu, Microporous polypropylene hollow fiber membranes Part II Pervaporation separation of water/ethanol mixtures by the poly(acrylic acid) grafted membranes, J. Membr. Sci. 214 (2003) 71. [166] W. Kujawski, S.R. Krajewski, Sweeping gas pervaporation with hollow-fiber ion-exchange membranes, Desalination 162 (2004) 129. 156 [167] H.A. Tsai, H.C. Chen, W.L. Chou, K.R. Lee, M.C. Yang, J.Y. Lai, Pervaporation of water/alcohol mixtures through chitosan/cellulose acetate composite hollow-fiber membranes, J. Appl. Polym. Sci. 94 (2004) 1562. [168] M.Y. Kariduraganavar, A.A. Kittur, S.S. Kulkarni, K. Rameshb, Development of novel pervaporation membranes for the separation of water–isopropanol mixtures using sodium alginate and NaY zeolite, J. Membr. Sci. 238 (2004) 165. [169] A. Huang, Y.S. Lin, W. Yang, Synthesis and properties of A-type zeolite membranes by secondary growth method with vacuum seeding, J. Membr. Sci. 245 (2004) 41. [170] J.J. Qin, R. Wang, T.S. Chung, Investigation of shear stress effect within a spinneret on flux, separation and thermomechanical properties of hollow fiber ultrafiltration membranes, J. Membr. Sci. 175 (2000) 197. [171] J.J. Qin, T.S. Chung, Effects of orientation relaxation and bore fluid chemistry on morphology and performance of polyethersulfone hollow fibers for gas separation, J. Membr. Sci. 229 (2004) 1. 157 [172] T.S. Chung, J.J. Qin, A. Huan, K.C. Toh, Study for the effect of dope shear rate on the outer surface morphology of hollow fiber ultrafiltration membranes by the atomic force microscope (AFM), J. Membr. Sci. 196 (2002) 251. [173] T.S. Chung, X.D. Hu, Effect of air-gap distance on the morphology and thermal properties of polyethersulfone hollow fibers, J. Appl. Polym. Sci. 66 (1997) 1067. [174] H. Strathmann, K. Kock, P. Amar, R.W. Baker, The formation mechanism of asymmetric membranes, Desalination 16 (1975) 179. [175] K.Y. Wang, D.F. Li, T.S. Chung, S.B. Chen, The observation of elongation dependent macrovoid evolution in single- and dual-layer asymmetric hollow fiber membranes, Chem. Eng. Sci. 59 (2004) 4657. [176] D.F. Li, T.S. Chung, R.Wang, Morphological aspects and structure control of dual-layer asymmetric hollow fiber membranes formed by a simultaneous coextrusion approach, J. Membr. Sci. 243 (2004) 155. [177] T.S. Chung, The limitations of using Flory-Huggins equation for the states of solutions during asymmetric hollow fiber formation, J. Membr. Sci. 126 (1997) 19. 158 [178] J.M.S. Henis, M.K. Tripodi, Multicomponent membrane for gas separations, US Patent 4,230,463 (1980). [179] J. Bai, A.E. Fouda, T. Matsuura, J.D. Hazlett, A study on the preparation and performance of polydimethylsiloxane-coated polyetherimide membranes in pervaporation, J. Appl. Polym. Sci. 48 (1993) 999. [180] C.A. Page, A.E. Fouda, R. Tyagi, T. Matsuura, Pervaporation performance of polyetherimide membranes Spin- and dip- coated with polydimethylsiloxane, J. Appl. Polym. Sci. 54 (1994) 975. [181] H. Matsuda, H. Yanagishita, H. Negishi, D. Kitamoto, T. Ikegami, K. Haraya, T. Nakane, Y. Idemoto, N. Koura, T. Sano, Improvement of ethanol selectivity of silicalite membrane in pervaporation by silicone rubber coating, J. Membr. Sci. 210 (2002) 433. [182] J. Comyn, F. de Buyl, Mobility of water and alcohols in a silica reinforced siloxane network, Euro. Polym. J. 37 (2001) 2385. [183] W. Henne, G. Dunweg, W. Schmitz, R. Pohle, F. Lawitzki, Method of producing dialyzing membrane, US Patent 4164437 (1979). 159 [184] T. Yanagimoto, Manufacture of ultrafiltration membranes, Japanese Patent 62019205 (1987). [185] H. Komada, K. Omuro, Hollow-fiber membranes for ultrapure water purification, Japanese Patent 62264280 (1987). [186] T. Yoshinaga, H. Shimazaki, Y. Kusuki, Polyimide bilayer hollow-fiber gas separation membranes and their manufacture, Japanese Patent 02251232 (1990). [187] Y. Li, C. Cao, T. S. Chung, K. Pramoda, Fabrication of dual-layer polyethersulfone (PES) hollow fiber membranes with an ultrathin dense-selective layer for gas separation, J. Membr. Sci. 245 (2004) 53. [188] L.Y. Jiang, T.S. Chung, D.F. Li, C. Cao, S. Kulprathipanja, Fundamental understanding of nano-sized zeolite distribution in the formation of the mixed matrix single- and dual-layer asymmetric hollow fiber membranes, J. Membr. Sci. 252 (2005) 89. [189] Y. Li, T.S. Chung, Z. Huang, S. Kulprathipanja, Dual-layer polyethersulfone (PES)/BTDA-TDI/MDI co-polyimide (P84) hollow fiber membranes with a submicron PES-zeolite beta mixed matrix dense-selective layer for gas separation, J. Membr. Sci. 277 (2006) 28. 160 [190] G.C. Kapantaidakis, G.H. Koops, M. Wessling, Preparation and characterization of gas separation hollow fiber membranes based on polyethersulfone-polyimide miscible blends, Desalination 145 (2002) 353. [191] K. Liang, J. Grebowicz, E. Valles, F.E. Karasz, W.J. MacKnight, Thermal and rheological properties of miscible poly(ether sulfone)/polyimide blends, J. Appl. Polym. Sci. 30 (1992) 465. [192] T.S. Chung, S.K. Teoh, W.Y. Lau, M.P. Srinivasan, Effect of shear stress within the spinneret on hollow fiber membrane morphology and separation performance, Ind. Eng. Chem. Res. 37 (1998) 3930. [193] D.F. Li, T.S. Chung, J.Z. Ren, R. Wang, Thickness dependence of macrovoid evolution in wet phase-inversion asymmetric membranes, Ind. Eng. Chem. Res. 43 (2004) 1553. [194] T.S. Chung, E.R. Kafchinski, The effects of spinning conditions on asymmetric 6FDA/6FDAM polyimide hollow fibers for air separation, J. Appl. Polym. Sci. 65 (1997) 1555. 161 [195] N. Widjojo, T.S. Chung, Thickness and air gap dependence of macrovoid evolution in phase-inversion asymmetric hollow fiber membranes, Ind. Eng. Chem. Res. 45 (2006) 7618 [196] C.C. Hu, C.S. Chang, R.C. Ruaan, J.Y. Lai, Effect of free volume and sorption on membrane gas transport, J. Membr. Sci. 226 (2003) 51. [197] R.Y.M. Huang, R. Pal, G.Y. Moon, Crosslinked chitosan composite membrane for the pervaporation dehydration of alcohol mixtures and enhancement of structural stability of chitosan/polysulfone composite membranes, J. Membr. Sci. 160 (1999) 17. [198] Y.M. Sun, T.L. Huang, Pervaporation of ethanol-water mixtures through temperature-sensitive poly(vinyl alcohol-g-N-isopropyacrylamide) membranes, J. Membr. Sci. 110 (1996) 211. [199] J.H. Kim, K.H. Lee, S.Y. Kim, Pervaporation separation of water from ethanol through polyimide composite membranes, J. Membr. Sci. 169 (2000) 81. [200] J.Z. Ren, R. Wang, T.S. Chung, D.F. Li, Y. Liu, The effects of chemical modifications on morphology and performance of 6FDA-ODA/NDA hollow fiber membranes for CO2/CH4 separation, J. Membr. Sci. 222 (2003) 133. 162 [201] P.S. Tin, T.S. Chung, S. Kawi, M. Guiver, Novel approaches to fabricate carbon molecular sieve membranes based on chemical modified and solvent treated polyimides, Micropor. Mesopor. Mater. 73 (2004) 151. [202] Y. Liu, T.S. Chung, R. Wong, D. Li, M.L. Chng, Chemical cross-linking modification of polyimide/poly(ether sulfone) dual-layer hollow-fiber membranes for gas separation, Ind. Eng. Chem. Res. 42 (2003) 1190. 163 PUBLICATIONS Journal papers: 1. R.X. Liu, X.Y. Qiao, T.S. Chung, The development of high performance P84 copolyimide hollow fibers for pervaporation dehydration of isopropanol, Chemical Engineering Science 60 (2005) 6674. 2. R.X. Liu, X.Y. Qiao, T.S. Chung, Dual-layer P84/polyethersulfone hollow fibers for pervaporation dehydration of isopropanol, Journal of Membrane Science 294 (2007) 103. Conference papers: 1. R.X. Liu, X.Y. Qiao, T.S. Chung, The pervaporation dehydration of isopropanol by BTDA-TDI/MDI (P84) co-polyimide membranes, AIChE 2005 annual meeting, Cincinnati, U.S.A. 2. X.Y. Qiao, R.X. Liu, T.S. Chung, The sorption characteristic and pervaporation performance of BTDA-TDI/MDI (P84) co-polyimide membranes for alcohol dehydration, NAMS 2006, Chicago, U.S.A. Book chapter: 1. X.Y. Qiao, L.Y. Jiang, T.S. Chung, R.X. Liu, Pervaporation Membranes for Organic Separation, in Tongwen Xu Edited, Advances In Membrane Science, Nova Science Publishers, Inc., New York. 2009 Patent: 1. T.S. Chung, X.Y. Qiao, R.X. Liu, A method of treating a permeable membrane, U.S. Patent PCT/SG2005/00430. 164 [...]... Organization of Dissertation It was found that previous research mainly focused on flat sheet dense or asymmetric membranes, however the research is lacking in the area of hollow fiber membranes especially as pervaporation dehydration membranes The development of novel hollow fiber pervaporation membranes is therefore a main objective of this study BTDA-TDI/MDI (P84) co-polyimide dense and asymmetric membranes. .. selectivity for pervaporation dehydration of alcohols This research study intends to extend previous research on BTDA-TDI/MDI (P84) copolyimide dense / asymmetric membranes to hollow fiber membranes It comprises the understanding of pervaporation transport process, the development of novel hollow fiber pervaporation membranes based on BTDA-TDI/MDI (P84) co-polyimide, and the investigation of the effects of. .. on dual-layer hollow fiber membranes morphology and performance of pervaporation dehydration of IPA Chapter 6 summarizes the general conclusions drawn from this research works 29 CHAPTER TWO THEORETICAL BACKGROUND 2.1 Fundamentals of pervaporation separation process The performance of a pervaporation membrane is dependent on the membrane materials, the structure of the membrane, and the interactions... nevertheless, the application of pervaporation in organic/organic separation has not acquired industrial acceptance because of the lack of advanced performance and the instability in organic solvents of currently available membranes [21] The first and only pervaporation plant using organoselective membranes was built by Air Products in 1991 for removal of methanol from MTBE The commercialization of pervaporation. .. also shown effectiveness in performance enhancement of polyimide membranes for the dehydration of organic solvents Interestingly, the swell-up of polymer chains in the feed solution makes the adverse effect of interstitial defects between the polymer matrix and the inorganic particles much less significant compared to that in gas 14 separation membranes [59] However, so far these mixed matrix modifications... performance This study selects IPA as a model solvent because of its high market value The objectives of this study are: 1 To develop BTDA-TDI/MDI (P84) asymmetric hollow fiber membranes, and investigate the effects of various spinning conditions and post treatments including 27 silicone rubber coating and heat treatment on P84 hollow fiber membranes morphology and performance of pervaporation dehydration. .. summarizes the recent development of PDMS membranes Except for benzene removal, the separation factors for alcohol removal are all below 100, which inhibit industrial scale applications 19 20 1.2.3 Organoselective membranes Albeit of the great potential in chemical and petrochemical industries, the separation of organic/organic mixtures using pervaporation is the least developed area There are wide streams of. .. ZrO2/Al2O3 membranes as the support layer, Xiangli et al [80] developed composite membranes with remarkably high flux (i.e., flux of 12300g/m2hr and separation factor of 6 for a feed ethanol concentration of 4.3wt% and temperature at 40°C) This performance is superior to the performance of PDMS composite membranes with a polymeric support, owing to the significantly reduced transport resistance of the ceramic... the segmental motions of polymer chains at a temperature above β transition enhance the formation of charge transfer complexes (CTCs) through their inherent electron donor (the diamine moiety) and electron acceptor (the dianhydride moiety) elements The CTCs formation strongly depends upon heat-treatment temperature, i.e., the higher the heat treatment temperature, the more CTCs can be formed [52] The. .. aspects The applications of pervaporation processes are mainly divided into three areas: (1) dehydration of alcohols or other aqueous organic mixtures; (2) removal of volatile organics from water; (3) organic/organic separation Dehydration of organic solvents such as alcohols, esters, ethers, and acids has become the most important application of pervaporation due to the high demand in industries and the ... dense / asymmetric membranes to hollow fiber membranes It comprises the understanding of pervaporation transport process, the development of novel hollow fiber pervaporation membranes based on... hollow fiber membranes especially as pervaporation dehydration membranes The development of novel hollow fiber pervaporation membranes is therefore a main objective of this study BTDA-TDI/MDI (P84)... 1980s for the dehydration of ethanol Since then, 38 solvent dehydration plants for ethanol and isopropanol, units for other solvents dehydration (i.e ester) have been installed world widely [87] There