Wastewater treatment is becoming more and more important to humans because of public health, water shortage and environmental issues. The U.S. Geological Survey published in 2009 suggested that although two-thirds of the earth is covered in water, 97% of it is of no use to humans and animals (except marine animals) due to the amount of salt within the water. The remaining 3% of freshwater is mainly captured in glacier and less than 0.01% is readily available for use. The distribution of water on earth is illustrated in Figure 1. In addition, the water consumed is becoming polluted due to human activities such as improper water use and lack of water treatment. The shortage of water is further aggravated by recent climatic changes. The World Health Organization reported that water shortage affects 40% of the world's population and over 25% of the world's population suffer from water related problems that affect, amongst other things, hygiene and health. Thus, the recycling of wastewater plays an increasingly important role in the provision of water. Due to the nature of wastewater, recycled or treated wastewater has to reach high-level qualities before usage.
Figure 1. Distribution of water on Earth (U.S. Geological Survey, 2009).
Many attempts through physical, chemical and biological treatments have been carried out to treat wastewater (Figure 2). An example is fungal wastewater treatment (Sankaran et al., 2010). However, many of the conventional methods have drawbacks especially those pertaining to residual chemical disinfectants and consequently cannot reach the high standard of quality needed for irrigation, consumption or industrial use. In addition, industrial wastewater nowadays produces large amounts of heavy metals and new synthetic organic compounds that require new treatment technologies to remove them effectively (United Nations ESCWA, 2003).
Figure 2. List of general wastewater treatment processes (United Nations ESCWA, 2003).
Membrane technology in this area has recently gained considerable attention due to its excellence in water filtration with no chemical and/or disinfection drawbacks. This technology, which used to be limited to desalination activities, has been improved and is now suitable for use in high quality water treatment. However, there are certain operational problems and limitations caused by system clogging and membrane fouling. These problems, which lead to reduced flux and inefficiency of the membrane, are listed in Figure 3 (Curcio & Drioli, 2009). Frequent backwashing and chemical cleansing are required in order that the system can run smoothly. Therefore, membrane material selection and pre-treaments are crucial to minimize the problems (Mulder, 1996).
Figure 3. Common problems in membrane processes and technology (Curcio & Drioli, 2009).
2. 1 Definition of membrane
Mulder (1996) defined membrane as a permselective barrier between two homogeneous phases. As there is a broad variety of membranes, their classification is based on their chemical and physical properties, such as membrane material, structure and preparation.
There are two types of membrane materials: biological and synthetic. Synthetic membranes can be subcategorised into organic or inorganic membranes. Membranes can also be distinguished by their morphologies (Figure 4), i.e. symmetric (homogeneous) and asymmetric (heterogeneous) structures, and porous (microfiltration, ultrafiltration) and nonporous (gas separation, pervaporation, dialysis) structures. Membrane preparations come in the form of flat sheet, spiral wound or tubular form. Such membrane forms are commonly used in plate-and frame or in hollow fibre, capillary and tubular modules respectively. (Mulder, 1996 and Curcio & Drioli, 2009).
Figure 4. Morphological membrane classification (Curcio & Drioli, 2009)
2. 2 Membrane material and PVDF
Many polymers are used as membrane materials, the most commonly used polymers are listed in Table 1. Since the research is based on PVDF, this review will focus on PVDF.
Table 1. Common polymeric materials for membrane preparation (Curcio & Drioli, 2009).
The polymeric membrane material poly(vinylidene fluoride) (PVDF) is known for its good chemical and physical properties. PVDF is advantageous due to its hydrophobic and chemical resistance towards corrosives, i.e. acids, bases, oxidants and halogens (Lovinger et al., 1982)
PVDF has good thermal stability, mechanical strength and chemical resistance. Such properties make PVDF a better membrane material than polysulfone (PS), poly(acrylonitrile) (PAN) and poly(ethersulfone) (PESf). PVDF has a semi-crystalline structure, comprising both crystalline phases and amorphous and/or rubbery regions (Busch et al., 2007), which gives the polymer membrane thermal stability and flexibility. In comparison with other hydrophobic, crystalline materials such as polytetrafluoroethylene (PTFE) and polypropylene (PP), PVDF is easier to process because it is more soluble in organic solvents such as dimethylformamide (DMF), dimethylacetamide (DMAc) and N-methyl-pyrrolidinone (NMP). Different types of membranes have been made from PVDF, e.g. flat sheet and tubular membrane modules. PVDF hollow fibre membranes are of particular interest recently because they have a high surface area per volume and they do not require support for the membrane. PVDF hollow fibres with favourable asymmetric structure can be prepared via dry-wet phase inversion process. On the other hand, PTFE and PP membranes are prepared via thermal or stretching processes that result in symmetric membranes.
PVDF's advantageous properties make it a good membrane material for industrial treatment such as oil emulsion (Jou & Huang, 2002), organic/water separations (Yuan & Li, 2008) gas absorption and stripping (Lau & Ismail, 2009) and membrane distillation (Khayet et al., 2002). However, its hydrophobic nature can lead to membrane fouling and reduced permeability, limiting its application in wastewater treatment. Thus, many studies have been done to improve its hydrophilicity through surface modifications (Mansourizadeh et al., 2010) as discussed further in this review.
3 Membrane Fouling
Membrane fouling is the accumulation of particles and molecules deposited or retained on the membrane surface, pore wall or inside pores. It is caused by the physical interaction and chemical degradation between membrane and foulants such as organic and inorganic substances (Mulder, 1996). During membrane separation, the membrane performance varies with time as illustrated in Figure 5. Du et al. (2009) explained that the flux through the membrane decreases over time due to fouling. Fouling resistance is generally measured using the bovine serum albumin (BSA) model. The phenomenon causes severe flux decline mainly for microfilitration, reverse osmosis and ultrafiltration, with flux being 5% less than that of pure water flux. Gas separation and pervaporation are less affected by the problem. Fouling can either temporarily or permanently decrease membrane flux. Initial flux can be restored through washing, but not once the membrane is permanently fouled. It is widely accepted that higher membrane hydrophilicity provides greater fouling resistance due to the hydrophobic nature of proteins and foulants (Rana & Matsuura, 2010).
Time (Minutes)
Flux (L m-2h-1)
Figure 5. Flux behaviour as a function of time.
4. 1 Membrane Fabrication
There are different techniques for membrane fabrication with a given membrane material. The technique to employ depends on the material used and the structure wanted, e.g. porous versus non-porous membranes (Mulder, 1996). Different kinds of synthetic materials are used for membrane preparation. There are a number of techniques used for preparing synthetic membranes. The more common and important ones are listed below.
4. 2 Sintering
Sintering is a technique that involves the compression of particles with a particular size and then sintering at high temperatures, during which the contacting interfaces between the particles vanish (Figure 6). The temperature used is dependent on the chosen material (Mulder, 1996).
Figure 6. A schematic of the sintering process (Mulder, 1996)
Sintering is widely used for production of inorganic membranes and symmetric membranes such as polypropylene (PP) and polytetrafluoroethylene (PTFE) due to the insolubility of the polymer in solvents. Polymeric membranes made via sintering tend to have low porosity (10%), except in the case of metal membranes (80%) (Mulder, 1996).
Regarding PVDF fabrication, Glasrock Products Inc. reported preparing PVDF membranes using the sintering method. PVDF powder was put in methyl-isobutyl- ketone and dispersed into droplets. The droplets were then sintered at certain temperatures to form porous PVDF membranes (Dickey & Mcdaniel, 1975).
4. 3 Stretching
A crystalline polymeric material such as PTFE, PP or PE is extruded and stretched into a film or foil perpendicular to the direction of extrusion. Mechanical stress is then applied, causing small ruptures that result in a porous structure. The porosity of membranes prepared by stretching is higher than sintering and reaches up to 90%, with pore sizes ranging from 0.1µm to 3 µm (Mulder, 1996).
The fabrication of polyurethane-poly(vinylidene fluoride)-poly(ethylene glycol) (PU-PVDF-PEG) and PU-PVDF hollow fibre membranes by melt spinning and stretching processes have been reported by Hu et al., 2010 and Hu et al. ,2011. The hollow fibres were spun, stretched and underwent thermal treatment and further stretching to obtain the final hollow fibre membranes.
4. 4 Track etching
A membrane film or foil is exposed to high energy particle radiation applied perpendicularly to the film. The particles create tracks as they damage the polymer matrix. The film is then submerged in acid or alkaline and etching is performed along the tracks to form parallel and symmetric cylindrical pores. The technique yields a membrane with low porosity ( ~10% maximum), with pore sizes ranging from 0.02 to 10µm. The porosity is dependent on radiation duration and the pore size is dependent on the etching time and temperature used (Mulder, 1996).
Track etching is sometimes used in PVDF membrane preparation. Komaki (1979) reported the formation of fine holes in PVDF film when it was subjected to oxygen fission fragments and etched in sodium hydroxide solutions. Grasselli and Betz (2005) prepared a PVDF membrane with a few hundred nanometers in pore size using an etching method. PVDF film was irradiated with tin ion beams and then subjected to etching conditions.
4. 5 Phase inversion
Phase inversion is the most industrially used technique for preparing polymeric membranes. Phase inversion is the controlled transformation of a polymer from a liquid state to a solid state. Solidification is initiated by the conversion of one liquid state into two liquids, a process known as liquid-liquid demixing. The initial phase of demixing is crucial to membrane morphology, where either porous or nonporous membranes can be obtained. There are a number of techniques involving phase inversion; they are controlled evaporation, precipitation from the vapour phase, thermally induced phase separation (TIPS) and immersion precipitation (IP). Among the listed techniques, IP is the most commonly used method to prepare polymeric membranes such as PVDF (Mulder, 1996).
Immersion precipitation
Asymmetric membranes such as PVDF are prepared by IP (Kong & Li, 2000) as it is easy for PVDF to dissolve in general organic solvents. The polymer solution is projected on a support and then immersed in a coagulation bath with a nonsolvent. The exchange of solvent from the polymer solution and non-solvent from the coagulation bath, known as liquid-liquid demixing results in precipitation and the formation of a membrane. The simplicity of the IP process makes it popular for industrial purposes. A variety of membranes with different morphologies and properties can be prepared with different formation conditions, as discussed later (Mulder, 1996).
The Principles of immersion precipitation
The principles of immersion precipitation, i.e. the liquid-liquid demixing process can best be described using a phase diagram (Figure 7). The process involves three main components namely: solvent, non-solvent (coagulant) and polymer. Phase diagrams are based on theoretical calculations adopted by Flory-Huggins and the analysis of the coexisting phases (Altena & Smolders, 1982 and Yilmaz and McHugh, 1986). The binodal curve within the diagram is an equilibrium curve that separates the triangle into two different regions (Aroon et al., 2010):
(i) One phase region: The compositions of solvent, non-solvent and polymer lead to the formation of a homogeneous one-phase solution. This is where the dope composition should be.
(ii) Two phase region: The compositions of solvent, non-solvent and polymer divide into polymer rich and lean phases.
Figure 7. Ternary phase diagram (Aroon et al., 2010)
The spinodal curve illustrates all of the variations that can result in phase separation. The region between the bimodal and spinodal curves is a region with metastable compositions. Phase separation in this region occurs via nucleation and growth. The intersection of the two curves is known as the critical point (C) (Machado et al., 1999).
The various mechanisms of membrane formation during phase inversion is represented in the Gibbs phase triangle (Figure 8) (Bazarjani et al., 2009).
Path 1: When the casting solution has a higher polymer concentration and a high solvent outflow to non-solvent inflow ratio, the top layer composition of the casting solution can follow path 1 (Barth et al., 2000). The path leads to the one-phase region and towards the vitrification region where the polymer molecules are solidified by gelation and/or crystallized into a compact, dense structure due to strong film shrinkage. Such structure is impermeable to water and so is not ideal for wastewater treatment (Bazarjani et al., 2009).
Path 2 & Path 3: When the solvent outflow to non-solvent inflow ratio is relatively low, the top layer composition of the casting solution can follow path 2. The path crosses the binodal boundary curve into the phase two region where phase separation, nucleation and polymer growth begins at the polymer-lean phase (Barth et al., 2000). Nucleation in path 2 is slow but faster in path 3. The polymer solution is in a metastable condition. The nuclei form membrane pores during growth in phase inversion. Membranes prepared via these pathways form small pores and have a small resistance to water flux (Bazarjani et al., 2009).
Figure 8. Precipitation paths for phase inversion (Bazarjani et al., 2009).
Path 4: The demixing path crosses the critical point, going into the unstable region where spinodal decomposition predominates (Barth et al., 2000). The initial stages of phase inversion involve the formation of a co-continuous structure from which the terminal pores are created by penetration into the miscibility gap. The polymer solution becomes unstable in this region and small changes in concentration can induce phase inversion. Membranes produced via this pathway tend to have a larger water flux (Bazarjani et al., 2009).
Path 5: When polymer concentration is low, the composition pathway crosses below the critical point where nucleation and growth of polymer-rich phase occurs. Phase inversion therefore occurs close to the critical point, forming a bi-continuous structure. This path leads to a compacted membrane with partially sintered beads. The water permeation through the packed membrane is therefore high (Bazarjani et al., 2009).
5. 1 Factors affecting membrane morphology and properties
The finishing membrane structure and properties are dependent on certain experimental parameters, such as polymer dope composition (e.g. concentration, additives, solvent), the material used (e.g. glass, polymer, metal), viscosity of the polymer solution, the temperature of the polymer solution, and duration of phase inversion process. The relationship of these parameters has an impact on the diffusion and demixing process, thereby affecting membrane morphology. Membrane morphology depends on the type of formation mechanism, e.g. instantaneous or delayed demixing. This results in either porous or non-porous membranes (Mulder, 1996).
5. 2 Effect of polymer concentration and composition
The polymer dope composition and more specifically its concentration affects the membrane morphology and permeability. Increasing PVDF concentration reduces finger-like macrovoids. This is due to the decreasing exchange rates of solvent and non-solvent with increasing dope viscosity, leading to a higher diffusion resistance from polymer aggregation. Lower PVDF concentration results in greater pore size distribution and mean pore size on membrane surfaces (Ren et al., 2006). García-Payo et al. (2010) showed that increasing the co-polymer, poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP) concentration decreased precipitation rate, resulting in a sponge-like membrane structure with reduced pore size.
5.3 Effect of solvent
The solvent is the one of the primary variables affecting membrane performance and properties. It is therefore crucial to choose the appropriate solvent. According to Klein & Smith (1972), one of the difficulties in producing a successful membrane is the lack of a method to systematically predict and select an appropriate solvent.
Yang et al. (1996) showed that membrane morphology is governed by the miscibility between solvent and nonsolvent. In order words, controlling the process of liquid-liquid demixing can manipulate membrane structure. A dense membrane with better mechanical strength is formed with delayed demixing whereas a porous membrane with more weak points, i.e. macrovoids is formed with instantaneous demixing (Shieh & Chung 1998).
Various studies have been conducted on the effects of different solvents on PVDF membrane structure and performance. In general, strong polar and high boiling point solvents for PVDF include: N,N-dimethylacetamide (DMAc), N,N-dimethylformamide (DMF), dimethylsulphoxide(DMSO), N-methyl-2-pyrrolidone (NMP), trimethyl phosphate (TMP), triethyl phosphate (TEP) and hexamethylphosphoramide (HMPA). Low boiling point weak solvents include: acetone or tetrahydrofuran (THF). According to Bottino et al. (1991) and Yeow et al. (2003), the solvent strength to PVDF polymer is as follows: HMPA>DMAc>NMP>DMF>TEP>TMP.
Wu et al., 2010 reported that greater pure water flux and lower BSA retention is observed when DMAc is used as solvent. On the other hand, the usage of DMSO as solvent resulted in a membrane with lower pure water flux which is difficult for protein macromolecules to pass through.
Figure 9. SEM images of PVDF hollow fibre membranes prepared with solvent A)DMAc, B) NMP, C)DMP, and D) DMSO (Wu et al., 2010).
Membranes prepared with DMAc have a higher pure water flux and lower BSA retention compared to membranes prepared with NMP and DMF. The resulting membrane morphology from solvent DMAc consists of finger-like pores in the inner membrane layer and a thinner inner and outer skin layer (Figure 9A). In comparison, solvents NMP and DMF produce a thicker outer layer (Figure 9B, C). Finger-like pores with less volume and thicker inner and outer walls are generated by DMSO solvent (Figure 9D). The results shown are consistent with Yeow et al. (2004).
Li et al. (2009) showed the resulting membrane morphologies and characteristics for four different mixed solvents, TMP-DMAc, TEP-DMAc, tricresyl phosphate (TCP)-DMAc and tri-n-butyl phosphate (TBP)-DMAc. Mixed solvents TMP-DMAc and TEP-DMAc resulted in a stronger solvent to PVDF. The two mixed solvents also showed faster precipitation rate and a reduction in membrane shrinkage, leading to higher flux. TCP-TCP-DMAc and TBP-DMAc resulted in greater membrane shrinkage and shortened macrovoids. Thus, a dense membrane structure is observed. The membrane porosity and flux was significantly reduced. Out of all four solvents, TBP-DMAc exhibited the thinnest membrane.
5. 4 Effect of additives
Non-solvent additives are often added during membrane fabrication to improve structure and performance. They can be categorized into high molecular weight additives and low molecular weight additives.
High molecular weight additives
According to studies, polymeric additives such as polyethylene glycol (PEG) and polyvinylpyrrolidone (PVP) have been shown to induce macrovoid formation (Deshmukh & Li, 1998, & Bottino et al., 1988). Deshmukh & Li (1998) reported that the addition of PVP resulted in macrovoid formation and larger cavities due to a higher precipitation rate during membrane casting. In addition, PVP is a pore-forming agent. A slight reduction in pore size is observed despite the increase in membrane surface porosity and hydrophilicity. In other words, as the concentration of PVP increases, so does the water flux, but the retention ability decreases. A disadvantage of PVP is that it is a water-soluble polymer that could be washed away in later process (Wang et al., 1999).
Another additive, PEG has been reported to increase water flux and decrease solute rejection. PEG also functions as a pore former and thus increases membrane porosity. However, PEG reduces mechanical strength due to the formation of macrovoids with increasing PEG molecular weight (Zuo et al., 2008).
Low molecular weight additives
Studies have been conducted on the effect of low molecular inorganic salts such as lithium chloride (LiCl) and lithium perchlorate (LiClO4) (Fontananova et al., 2006 and Yeow et al., 2005). Tomaszewska (1996) reported structural changes in PVDF membranes with the addition of LiCl in the cast solution. The presence of LiCl resulted in larger cavities and a more porous membrane. Tomaszewska (1996) also observed better permeate flux and reduction in mechanical strength. On the other hand, Wang et al. (2000) observed that the addition of LiCl improved the mechanical properties of membranes. According to Fontananova et al. (2006), low levels of LiCl enhanced permeate flux due to the thermodynamics effect on liquid-liquid demixing. At high levels of LiCl, macrovoid formation was reduced due to the increase in solution viscosity, which led to better mechanical strength. The addition of LiClO4 with DMAc solvent in PVDF membrane preparation has been reported to enhance membrane pore size (Yeow et al., 2005).
5. 5 Effect of coagulation bath composition
The semi-crystalline structure of PVDF means preparation by phase inversion is governed by liquid-liquid demixing and crystallization, which is dependent on the medium of the coagulation bath. Water is known to be a strong non-solvent, therefore using water in the coagulation bath leads to instantaneous liquid-liquid demixing. This results in the formation of finger-like voids in asymmetric membranes (Mulder, 1996). Deshmukh and Li (1998) observed that the addition of ethanol into the coagulation bath resulted in sponge-like membranes. Such phenomena were also observed by Khayet et al. (2002) which led them to suggest that the addition of ethanol delayed the coagulation process, resulting in a sponge-like membrane structure.
5. 6 Effect of temperature
The temperature of solvent and non-solvent mixing and precipitation are factors that control membrane morphology. Cheng (1999) observed that the increase in coagulation bath temperature resulted in the change in membrane morphology from a symmetrical structure with spherical crystallites to a dense, asymmetric membrane with cellular structure and spherical particles. Yeow et al. (2004) reported that the increase in dope and coagulant temperature, especially with the presence of additives, repressed crystallization and affected membrane morphology. It was believed that at higher temperatures, liquid-liquid demixing was favoured and gelation was suppressed. This resulted in cellular membrane morphology. At low temperatures, gelation was induced by crystallization and the slow liquid-liquid demixing prevented the growth of crystals, leading to irregular macrovoid formation (Wang et al., 2008 and Yeow et al., 2004).
5. 7 Effect of air gap
Air gap is the distance between the spinneret and the coagulation bath during spinning of the hollow fibre. A greater air gap means the polymer is exposed to air for a longer period of time before it goes into the coagulation bath (Mulder, 1996).
Wang et al. (1999) reported a thinner skin layer was formed with a shorter air gap which led to greater permeation flux. When an amphiphilic copolymer additive was incorporated into the spinning dope, the effect of the air gap was more important. A short air gap resulted in better surface segregation.
Widjojo & Chung (2006) reported the effects of air gap distance in macrovoid formation. The number of inward-pointed macrovoids increased with air gap distance while the number of outward-pointed macrovoids decreased.
6. 1 Surface and blending modification of membranes
Synthetic polymer membranes such as PVDF membranes tend to be hydrophobic due to the nature of materials used. This results in fouling, affecting the quality of permeate and operating cost. A hydrophilic membrane has high surface tension and allows hydrogen bonding with water, forming a water layer between the membrane and solution. This layer decreases the adhesion of foulants, as an increase in energy is required for hydrophobic molecules such as proteins to disrupt the structured water boundary (Brant and Childress, 2004). Thus, studies have been carried out to reduce fouling and exclusion of foulants from membranes. Yamamura et al. (2008) suggested that certain membrane properties such as hydrophilicity and charge improved fouling resistance. Hydrophilic membranes prevented the build up of particles at the solution-membrane boundary (Kabsch-Korbutowicz et al., 1999), allowing better flux and reduced fouling. Therefore, many studies have been done to modify and improve membrane surface hydrophilicity via surface coating, blending and grafting etc. to minimize irreversible membrane fouling (Hester et al., 1999 and Taniguchi et al., 2003).
6. 2 Surface Coating
One method of obtaining a more hydrophilic membrane is by coating a hydrophilic layer on the membrane surface. This can be achieved by physical adsorption of water-soluble polymers (Reddy et al., 2003) or surfactants (Jönsson & Jönsson, 1991), cross-linking (Du et al., 2009) and sulfonation (Baroña et al., 2007).
Studies on hydrophobic membrane coating have been done with hydrophilic polymers such as polyvinyl alcohol (PVA) by the dip-coating method followed by interfacial cross-linking. The coating improved membrane smoothness and hydrophilicity which resulted in better water permeation and anti-fouling when compared with unmodified membranes (Du et al., 2009, Mansouri & Fane, 1999). Recently, Chanachai et al. (2010) reported that coating PVDF membrane surface with chitosan via cross-linking provided better wetting-out protection and higher water flux.
In addition, the charge of foulants has a great impact on fouling (Brink & Romijin, 1990). Membrane foulants such as microbial cells and proteins are anionic (Wilbert et al., 1998). Therefore increasing the negativity of membrane surface would raise the repulsive electrostatic forces between the membrane surface and foulants, resulting in less fouling by such compounds (Mockel et al., 1999). Negative charges such as hydroxyl groups (-OH) and sulfonyl groups (-SO3) are often used for such purposes (Baroña et al., 2007).
Despite its advantages, surface coating can block membrane pores thereby reducing water flux. Further, the instability of the coated surface due to the weak physical interaction between PVDF membrane and the layer may result in the coating being washed away during operational cleaning processes. Chemical treatments such as cross-linking and sulfonation can be carried out to secure the layer on the membrane. On the other hand, surface grafting on the membrane surface eliminates the problem by immobilizing the functional chains via covalent bonding.
6. 3 Surface Grafting
A promising method to enhance long-term chemical stability on membrane surface is by surface grafting. Surface grafting is achieved by covalently attaching macro-monomers or polymers to membrane surfaces. This can be done via several means, e.g. free-radical, photochemical, radiation, redox and plasma-induced grafting, atom transfer radical polymerization (ATRP) (Belfer et al., 2004, Belfer et al., 1998, Hilal et al., 2003, Taniguchi et al., 2003 & Bryjak et al., 2004). Similar to surface coating, surface grafting can block membrane pore surfaces leading to a reduction in permeability. Table 2 summarizes the advantages and disadvantages of the different grafting methods. The mechanism of grafting involves the formation of free radicals on the membrane surface by exposure to UV irradiation, low-temperature plasmas and electron beam radiation etc, allowing the graft polymer to covalently attach to the membrane (Zhang et al., 2009, Yamagishi et al., 1995 and Chen et al., 2009 ).
Table 2. The advantages and disadvantages of various surface grafting methods.
Grafting Method
Advantages
Disadvantages
UV
Simple and relatively cheap (Zhang et al., 2009 )
Grafting on PVDF by UV irradiation is difficult because PVDF is very resistant (Botelho et al., 2008). Strong initiation irradiation source, e.g. X-ray or ozone required (Chan 1994). Pre-treatment of surface required for further modification.
Plasma
Majority of membrane properties not affected. Versatility for both porous and non-porous membranes (Chan et al., 1996).
Expensive as vacuum system required (Chan et al., 1996).
Redox
Simple and can be used in aqueous medium at room temperature without external activation (Sarac, 1999).
High monomer concentrations required to achieve a substantial degree of grafting. (Bernstein et al., 2010)
Electron Beam
No expensive vacuum system required. Can operate at room temperature. Contamination free as no catalyst or additives are required (Liu et al., 2007).
Long linear graft chains while grafting at high degree can lead to pore plugging (Liu et al., 2007).
Atom Transfer radical polymerization (ATRP)
Easy to set up. Inexpensive catalyst.Allow molecular weight and functionality to be controlled. Have a wide range of choice for alkyl halide initiators and transition metal catalysts. Allow free radical concentration to be controlled and remain low throughout the process. (Mueller et al., 2009)
The transition metal complex has to be removed after ATRP for aesthetic, environmental and stability reasons. (Mueller et al., 2009)
Atom Transfer Radical Polymerization
The general mechanism of ATRP is illustrated in Figure 10. ATRP establishes a dynamic equilibrium between a low oxidation-state transition metal complex (MtnLm) and its higher oxidation-state complex (Mtn+1Lm). Alkyl halides (R-X) such as alkyl chloride or bromide are initiators for ATRP. A reversible reaction initiates between the low oxidation-state complex and an alkyl halide via an electron redox process, generating a radical (Râ-) and a high oxidation-state complex. The radical may then react with a vinyl monomer during propagation or it can terminate by disproportionation and coupling, or become reversibly deactivated by the high oxidation state complex (Mueller et al., 2009).
Figure 10. Schematic of ATRP mechanism (Matyjaszewski, 1994)
An example of ATRP is the surface segregation of amphiphilic copolymer on PVDF (Hester et al., 2002), which is one of the objectives of this research. Amphiphilic graft copolymers were prepared with PVDF functioning as the macroinitiator. Hydrophilic poly(oxyethylene methacrylate) (POEM) side chain was grafted onto PVDF backbone via ATRP, forming PVDF-g-POEM (Figure 11). The grafting was performed at 90°C, involving ligand 4,4′-dimethyl-2,2′-dipyridyl (DMDP), co-initiator copper (I) chloride (CuCl) and solvent NMP.
Figure 11. Mechanism of PVDF-g-POEM grafting by ATRP, where R is -(CH2-CH2-O)9-CH3 (Inceoglu et al., 2004).
The resulting amphiphilic copolymer (PVDF-g-POEM) could be used as an additive in the PVDF casting solution to enhance membrane hydrophilicity. Since the hydrophibic POEM side chains were grafted onto the hydrophobic backbone of PVDF, this backbone allowed the additive to mix well with the PVDF solution. This resulted in a stable, long lasting additive. In addition, the hydrophilic chains could organise themselves on the membrane surface as water was used for membrane precipitation. This process is known as surface segregation. Enhanced resistance to biofouling and better water flux were observed with such grafted membranes. Although chemical grafting can significantly improve hydrophilicity of membrane, a drawback is that it can alter the PVDF main chain and affect the membrane performance (Xu et al., 2005).
7 Macrovoids
Macrovoids are large (10-50µm) pores that occur in asymmetric membranes fabricated by phase-inversion. Asymmetric membranes are made up of a thin top layer and a supportive porous sublayer. Macrovoids are often found in the sublayer (Figure 12). In general, formation of macrovoids in membranes are undesirable as they cause weak mechanic points that can result in membrane failure under high pressure and continuous backwashing. Although PVDF has arguably good mechanical strength and properties, macrovoids can cause substantial compression and fracture of the permselective layer. This can reduce flux through the membrane and result in loss of permselectivity, making the system less efficient. Thus, macrovoid-free hollow fibre membranes are highly desirable and are of great interest to the academic and industrial world.
Figure 12. Structures of macrovoids in membranes.
Many studies have been done on macrovoid formation and growth. Currently, there are five reported types of macrovoids (Figure 13). They are elliptical, micelle-like, teardrop, inward- and outward-pointed finger-like macrovoids (Widjojo & Chung, 2006 and Teoh & Chung 2009). Pekny et al. (2002) and Shojaie et al. (1994) suggested the formation of elliptical and teardrop macrovoids is through solutocapillary convection or osmotic pressure (McKelvey and Koros, 1996). Finger-like macrovoids have been proposed to originate from solvent intrusion or local surface instability. (Peng et al., 2008)
Figure 13. The five proposed types of macrovoids. 1) inward-pointed; 2)outward-pointed; 3) elliptical; 4) teardrop; and 5) micelle-like macrovids (Widjojo & Chung, 2006 and Teoh & Chung 2009).
A number of different mechanisms of macrovoid formation have been proposed. They include instantaneous demixing and diffusional growth (Reuvers, 1987 and Smolders et al., 1992), the solutocapillary hypothesis (Pekny et al., 2002), shrinkage stress in skin layer induced by solvent syneresis (Strathmann et al., 1975 & 1985), instability in concentration gradient during solvent casting (Ray et al., 1985). Not one of these mechanisms in itself alone is sufficient to explain the presence of macrovoids for all conditions because some proposed mechanisms are developed based on wet-casting while others are based on dry-casting. However, these studies imply the involvement of several different mechanisms to explain macrovoid formation depending on the casting conditions.
Understandably, different attempts have been made to reduce macrovoids, such as by the induction of delayed demixing and gelation, by the use of high concentration of polymer solution, by high spinning shear rates, and by the addition of high-viscosity components and addition of surfactants (Wang et al., 1998, Fontananova et al., 2006 and Widjojo & Chung, 2006). Most of the studies done on macrovoid-free membranes are focused on glassy polymers such as polyamide, cellulose acetate etc. For glassy polymers, the precipitation during phase inversion is driven primarily by liquid-liquid demixing. For semi-crystalline polymers, both liquid-liquid demixing and solid-liquid demixing govern the precipitation. The liquid-liquid demixing leads to a porous structure created by the polymer-rich phase forming a matrix and surrounding the polymer-lean phase. Conversely, solid-liquid demixing leads to an interlinked semi-crystalline structure which makes it very difficult for PVDF to achieve a macrovoid-free membrane morphology. Recently, Bonyadi and Chung (2009) proposed the use of a two-phase flow with the solvent-dope solutions in air gap and a dual layer spinneret for immersion in external coagulant to prepare porous, macrovoid-free PVDF hollow fibre membranes.
The rheological behaviour of the dope solution during spinning can be investigated by altering the spinning parameters. Ren et al. (2006) and Chung et al. (2000) concluded that membrane morphology was also dependent on the spinneret shear rate where a higher rate resulted in a larger mean pore size and wider distribution. In addition, the elongation rate of fibres in the air gap also contributed to membrane morphology and mechanical properties. The elongation viscosity of the polymer solution hardened with increasing elongation rates, resulting in a denser membrane morphology with less macrovoids and therefore better mechanical strength (Ekiner & Vassilatos, 2001). These studies highlight the importance of rheological behaviour in macrovoid formation, particularly in preparing macrovoid-free PVDF membranes.
8 Summary
Worldwide water shortage has resulted in the need of high quality water for population and industrial growth. With the large amount of wastewater produced every day , wastewater treatment is necessary to retrieve water for consumption or for industrial purposes. There are many technologies and projects for wastewater treatment, in which membrane technology is one of the most recent ones. It has many advantages since the physical approach of membrane filtration eliminates risks of any residual chemicals in the permeate water.
PVDF has excellent chemical and thermal stabilities with acceptable mechanical strength which makes it a desirable membrane material. PVDF hollow fibre membranes are of particular interest as they have a high surface area per volume and they do not require support for the membrane.
Although PVDF has many advantages over other materials, it also has limitations due to its hydrophobic nature. Many studies have been conducted to minimize fouling due to the hydrophobicity of PVDF. Nonetheless, studies on macrovoid-free PVDF hollow fibre are scarce. Since there is room for the further exploration and establishment of methods and technical know how for fabricating macrovoid-free PVDF membranes for both specific and general use, it would be interesting and useful to study the circumstances affecting their development, fabrication, their membrane structure and their mechanical strengths and properties.
9 The research objectives and milestones
Objective 1: To study ATRP and prepare PVDF-g-POEM amphiphilic copolymer.
Objective 2: To prepare hollow fibre membranes from PVDF/PVDF-g-POEM solutions via phase inversion
Objective 3: To investigate the effect of adding the PVDF-g-POEM copolymer in the spinning dope on PVDF hollow fibre morphology and properties, such as membrane thickness, mechanical strength, flux, porosity and formation of macrovoids.
Objective 4: To investigate the effect of various spinning parameters on the PVDF hollow fibre morphology and properties.
Objective 5: To produce PVDF hollow fibres with reduced macrovoids
Objective 6: To investigate and characterize the performance and strength of PVDF hollow fibres with reduced macrovoids
Table 3. Timetable and milestones.
Feb
Mar
Apr
May
June
July
Aug
Sept
Objective 1
Objective 2
Objective 3
Exams
Objective 4
Objective 5
Objective 6
Write up thesis
