Journal of Membrane Science 279 (2006) 336–346
Preparing highly porous chitosan/cellulose acetate blend hollow fibers as adsorptive membranes: Effect of polymer concentrations and coagulant compositions Chunxiu Liu a , Renbi Bai a,b,∗ a
Department of Chemical and Biomolecular Engineering, Faculty of Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260, Singapore b Division of Environmental Science & Engineering, Faculty of Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260, Singapore Received 18 August 2005; received in revised form 19 November 2005; accepted 10 December 2005 Available online 18 January 2006
Abstract In this work, we report the study on the effect of fabrication factors influencing the structures and morphologies of the chitosan and cellulose acetate blend hollow fibers as adsorptive membranes to achieve highly porous and macrovoids-free structures with different pore sizes. The factors investigated include chitosan (CS) and cellulose acetate (CA) concentrations in the spinning dope solutions, and the composition of the external and internal coagulants. For CA concentration at 12–18 wt% and CS concentration at up to 4 wt% in the spinning dope solutions, the blend hollow fibers were successfully prepared with outer surface pore sizes, specific surface areas and porosities in the range of 0.54–0.049 m, 10.4–14.5 m2 /g and 80.6–70.4%, respectively, depending on the CA and CS amounts in the spinning dope solutions and the coagulants used. Water, a weaker coagulant, can be used as both the external and internal coagulants in the fabrication process and the resultant CS/CA blend hollow fibers showed spongy-like, macrovoids-free and relatively uniform porous structures which are desirable for adsorptive membranes. By increasing the alkalinity of the coagulants, the coagulation rate of the blend hollow fibers was increased and the hollow fibers were observed to form relatively denser surface layers and to have smaller surface pore sizes and slightly greater specific surface areas, due to the stronger coagulation effect. Particularly, when NaOH solutions (1–3 wt%) were examined as the internal coagulant, more and larger macrovoids were formed in the blend hollow fibers at the near lumen side with the increase of the NaOH concentrations at low CS concentrations (<3 wt%), indicating the importance of coagulant compositions and polymer concentrations in the fabrication of adsorptive CS/CA blend hollow fibers with uniform porous structures. © 2006 Elsevier B.V. All rights reserved. Keywords: Chitosan; Cellulose acetate; Blend hollow fiber; Morphologies; Porous structures; Fabrication factors
1. Introduction In recent years, affinity or adsorptive membranes have attracted considerable research interest in the biomedical or biochemical field for the removal of toxins from human plasma, or purification of proteins, and in the environmental field for the removal and recovery of heavy metal ions [1–11]. These membranes have reactive or functional groups on the surfaces that can be used for specific or selective separation of components from a gaseous or liquid medium, and are sometimes called as
∗
Corresponding author. Tel.: +65 65164532; fax: +65 67744202. E-mail address:
[email protected] (R. Bai).
0376-7388/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2005.12.019
membrane adsorbers or membrane adsorbents. In comparison with the conventional granular adsorption bed, membrane-based adsorption systems provide the advantages of short diffusion distances by the convective flow across the membrane thickness, which allows fast separation of the targeted components because the process efficiency is dominated by the binding kinetics rather than by the mass transfer resistance [2]. An ideal adsorptive membrane requires tunable surface properties (i.e., hydrophobic or hydrophilic), large specific surface area, high porosity with desirable pore size and high density of reactive groups, in addition to the common requirement of membrane for good chemical and mechanical resistance [12–14]. Since most commercially available membranes are synthesized of relatively inert materials, the frequently used method to prepare affinity
C. Liu, R. Bai / Journal of Membrane Science 279 (2006) 336–346
or adsorptive membranes is therefore through surface modification of the commercial membranes (such as polyethylene [15], nylon [16], polyvinylidene difluoride [17] and polysulfone [18]) via the attachment of functional groups, such as OH, NH2 , SO3 H, COOH, CONH2 or epoxy, through chemical reactions or surface grafting polymerization. In general, such systems to prepare adsorptive membranes are complicated, involving three steps including the preparation of the base membranes, activation of the base membranes and coupling of affinity ligands to the activated membranes [19]. Moreover, the activation of the base membranes is usually conducted under various harsh physical and chemical conditions which can cause undesirable and irreproducible changes in the membrane structures [20]. It is also often difficult for such systems to obtain porous adsorptive membranes with high porosities and various desired pore sizes. An alternative method to prepare adsorptive membranes is to fabricate the membranes from a polymer or polymer blends which have reactive or functional groups on the polymer backbones, such as chitosan [3,21], ethylene vinylalcohol (EVAL) [22] and cellulose acetate/polyethyleneimine blend [9], etc. By controlling and varying the fabrication conditions, it is also possible to produce adsorptive membranes with different pore sizes and structures. This alternative fabrication method clearly has the advantage of system simplicity (only one step as compared to the three steps mentioned earlier). In our laboratory, we have successfully prepared adsorptive hollow fiber membranes from chitosan/cellulose acetate blend, with cellulose acetate (CA) acting as a matrix polymer and chitosan (CS) as a functional polymer to provide the membrane with coupling or reactive sites for affinity-based separations [23]. The preliminary study has indicated that the blend hollow fiber membranes have high mechanical strength, tunable hydrophilicity/hydrophobicity and good binding or adsorption capabilities toward heavy metal ions or albumins, even at the presence of a small amount of CS in the blend hollow fiber membranes. Hence, a further study on the structures and morphologies of the CS/CA blend hollow fiber membranes is of research and practical interest since the membrane structures can have significant impact on the membrane performances, such as the adsorption capacity and adsorption rate, especially when substances of different sizes are to be separated. For example, in metal ion removal and protein separation, it may be desirable to have the membranes with different pore sizes to allow adequately free passage of the small or large ions or molecules to be separated in the membrane pores for adsorption or separation to fully take place. Moreover, macrovoids-free membranes are desirable for adsorptive membranes because they can provide large internal surface areas for adsorption and uniform flow across the membranes to ensure the process efficiency. In this work, we report the study on the effect of fabrication factors influencing the structures and morphologies of the CS/CA blend hollow fiber membranes. The factors investigated include the CA and CS concentrations in the spinning dope solutions, the composition of the external and internal coagulants. The objectives are to obtain highly porous, macrovoids-free adsorptive hollow fiber membranes with different pore sizes.
337
Fig. 1. Schematic diagram of the hollow fibers spinning setup. (1) Gas cylinder, (2) pressure meter, (3) dope tank, (4) filter, (5) syringe pump, (6) spinneret, (7) coagulation bath and (8) rinsing tank.
2. Experimental 2.1. Materials CS was purchased from Aldrich (labeled as low molecular weight) and used as received. The degree of deacetylation and the molecular weight of the CS were determined to be 73.5% and 75,000 g/mol, respectively. The reason in choosing CS with low molecule weight was to allow a greater amount of chitosan could be added into the CS/CA blend dope. CA was supplied by Fluka and the acetyl content and molecular weight of the CA was 40% and 37,000 g/mol, respectively. Formic acid (FA, 98–100%) from Fluka was used as the co-solvent for both CS and CA. 2.2. Fabrication of CS/CA blend hollow fiber membranes The blend hollow fiber membranes were fabricated through a wet spinning process, as schematically represented in Fig. 1, and the details can be found elsewhere [23]. In brief, the blend spinning dope solution was prepared by mechanically stirring CA and CS together in the co-solvent at 3.33 Hz (200 rpm) overnight. The resultant blend spinning dope solution was then degassed and finally filtered through a 15 m stainless filter to remove any insoluble particles under the force of high pressure N2 gas. The clear and homogeneous blend dope solution was then forced through a stainless steel spinneret comprising an annular ring (with o.d. and i.d. of 1.3 and 0.5 mm, respectively) and extruded into an external coagulation bath. A bore liquid coagulant was simultaneously delivered through the inner core of the spinneret by a high pressure syringe pump (ISCO 100DX). The hollow fiber membranes were collected by a drum from the external coagulation tank and were then rinsed with 10 wt% NaAc solution to leach out excess solvent, followed by the rinse with tap water. Finally, the hollow fibers were stored in DI water for further use. To get dried samples for analyses, the hollow fibers were subjected to treatment of multi-step solvent exchange with 1-propanol and 1-heptane to retain the original porous structures [24]. A number of experiments were conducted with different dope compositions and external and internal coagulants and the conditions are summarized in Table 1. The total weight of
338
C. Liu, R. Bai / Journal of Membrane Science 279 (2006) 336–346
Table 1 Fabrication conditions of the CS/CA blend hollow fibers in this study Experimental series
Dope composition by weight (CS/CA/FA)
Composition of external coagulant
Bore fluid composition
Effect of CA concentrations
2.0/12.0/86.0 2.0/14.0/84.0 2.0/16.0/82.0 2.0/18.0/80.0
Tap water
DI water
Effect of CS concentrations
2.0/12.0/86.0 3.0/12.0/85.0 4.0/12.0/84.0
Tap water
DI water
Effect of external coagulants
2.0/12.0/86.0
3 wt% NaOH 10 wt% NaAc 3.6 × 10−4 wt% FA
DI water
Effect of internal coagulants
2.0/12.0/86.0 3.0/12.0/85.0
Tap water Tap water
2 wt% NaOH; 3 wt% NaOH 2 wt% NaOH; 3 wt% NaOH
the three components in each dope, i.e., CS, CA and FA, was set at 100 g while the CA concentrations in the dopes varied from 12 to 18 wt% and the CS concentrations varied from 2 to 4 wt%. Water was first examined as both the external and internal coagulants. Then, the composition of the external coagulant was adjusted by adding an adequate amount of NaOH (3 wt%, 6.2 kg) or sodium acetate (10 wt%, 22 kg) or a small amount of FA (3.6 × 10−4 wt%, 60 mL) in the coagulation bath of 200 L water (200 kg) and that of the internal coagulant was adjusted by the addition of a different amount of NaOH in DI water. 2.3. Cloud point study Cloud points of the dope solutions were measured by titration method. CA solution, CS solution and CS/CA blend solutions were prepared by dissolving them in formic acid, respectively, with mechanical stirring. Then, nonsolvent or coagulant (i.e., DI water or NaOH solution in this case) was added slowly into each of the solutions through a syringe pump. The cloud point was observed visually from the sudden occurrence of the turbidity of the solutions (indicating the production of polymer solid particles due to phase separation/inversion).
temperature. Before the analysis, samples were fully degassed overnight with pure nitrogen gas. Specific surface areas were calculated from five-point adsorption data in the relative pressure range of p/p0 = 0.05–0.30. The porosities of the blend hollow fibers were measured by the dry–wet weighing method. The dried hollow fibers were equilibrated with DI water for 24 h. The porosity was then determined by dividing the amount of water adsorbed (mL) with the amount of the wet hollow fibers (mL). The experiment was done for five samples and the average porosity was used for each type of the blend hollow fibers. The mechanical property of the wet hollow fibers was evaluated through the measurement of the tensile strength and strain at break. Tests were conducted with Instron 3345 Material Tester at a temperature of 25 ◦ C and a relative humidity of 60%. The initial gauge length was set to be 25 mm and the draw speed was set at 10 mm/min. In each measurement, sample of each fiber was cut into 5 cm length, and attached onto the two clamps of the machine. For reliability, five readings were taken for each type of the sample, and the average value was used in this paper. 3. Results and discussion
2.4. Other analyses of the blend hollow fibers
3.1. Cloud point data
The structures and morphologies of the blend hollow fiber membranes were investigated through SEM (JEOL JSM-5600 SEM) and FESEM (JEOL JSM-6700F FESEM) analysis. The dried hollow fibers were snapped in liquid nitrogen to give a generally clean break of the cross-section. As the polymers were non-conductive, the hollow fibers were coated with platinum powder on the surface for 40 s at 4 × 103 Pa (40 mbar) vacuum. The average surface pore sizes of the hollow fibers were measured with the software supplied by the manufacturers of the SEM/FESEM. The specific surface areas of the blend hollow fibers were measured with the BET method using a Quantachrome Nova 3000 Multi-point Gas Adsorption Analyzer at the liquid nitrogen
The cloud point data provide useful thermodynamic information about the phase separation/inversion process of the polymer solutions. Table 2 shows the experimental results of the cloud points for a few types of spinning dope solutions containing CA/FA, CS/CA/FA and CS/FA, respectively. The results indicate that both the ternary CS/CA/FA blend solutions and the binary CA/FA solution had high tolerance with the addition of the nonsolvent (or coagulant), i.e., water or NaOH solutions in this case, because the nonsolvent added to obtain the cloud point was as high as 40–27% by weight. This suggests that FA was indeed a solvent with high solubility for both CS and CA polymers. The cloud point for the CS/FA binary solution was however not observed with the addition of water up to 100%
C. Liu, R. Bai / Journal of Membrane Science 279 (2006) 336–346
339
Table 2 Cloud point data of different dope solution compositions (at 25 ◦ C) CA (wt%)
12 12 12 0 a
CS (wt%)
0 2 3 2
FA (wt%)
88 86 85 98
Nonsolvent concentration at cloud point (wt%) Water as nonsolvent
NaOH (3 wt%) solution as nonsolvent
39.3 32.7 29.6 NOa
34.5 30.2 27.3 95.6
Not observed.
by weight, and was only observed when the addition of NaOH solution (3 wt%) reached about 95% by weight. Since CS can dissolve in solutions of pH < 4, the cloud point (or phase separation/inversion) of the CS/FA solution would only occur when the addition of water or NaOH solution raised the solution pH to a value of 4 or above. The results in Table 2 also suggest that, though both the tap water and the NaOH solution may be used as the coagulants for the fabrication of the CS/CA blend hollow fibers, higher pH coagulant (e.g., NaOH solution) would serve as a stronger coagulant since the amount of the NaOH solution to be added was less than that of tap water to obtain the cloud point. With the increase of the polymer concentrations (consequently a reduction of FA), the phase separation/inversion also appeared to be easier as the cloud point occurred at a less addition of the water or NaOH solution. The CS concentration seemed to have a significant impact on the cloud point since a slight increase of the CS concentration caused a large reduction in the need of adding water or NaOH solution to obtain the cloud point (e.g., the dope solution of 12% CA, 2% CS and 86% FA needed 39.3% water, and that of 12% CA, 3% CS and 85% FA needed only 29.6% water, see Table 2). This trend of change indicates that thermodynamically CS worked favorably in enhancing the
demixing of the polymers with FA in the CS/CA/FA spinning dope solutions. 3.2. Effect of cellulose acetate concentrations Spinning dope solutions with a constant CS (2 wt%) and different CA concentrations (i.e., 12, 14, 16 and 18 wt%) were used to spin the hollow fiber membranes and water was used as both the external and internal coagulants (nonsolvent). Some typical results showing the overall and the cross-sectional structures of the hollow fiber membranes are given in Fig. 2, and those showing the effect of CA concentrations on the specific surface areas, porosities and surface pore sizes of the hollow fibers are given in Fig. 3. It has been found that all the hollow fiber membranes had spongy-like and macrovoids-free structures, with the pores across the cross-section being highly interconnected and displaying open porous networks, which is the desirable structure for adsorptive membranes to achieve large specific surface areas and uniform fluid flow. As indicated in Fig. 3, the porosities and surface pore sizes of the hollow fiber membranes decreased but the specific surface areas of the hollow fiber membranes increased with the increase of the CA concentrations in the dope
Fig. 2. SEM images showing the overall (a and c) and cross-sectional (b and d) structures of the CS/CA blend hollow fibers prepared from spinning solutions containing 12 wt% CA and 2 wt% CS (a and b) and 18 wt% CA and 2 wt% CS (c and d) (water was used as both the external and internal coagulant).
340
C. Liu, R. Bai / Journal of Membrane Science 279 (2006) 336–346
Fig. 3. Effect of CA concentration in the spinning dope (with 2 wt% CS) on the structural characteristics of the CS/CA blend hollow fibers: (a) specific surface areas and porosities and (b) averaged pore size of the outer and inner surfaces (water was used as both the external and internal coagulant).
solutions. For example, for CA from 12 to 18 wt%, the porosity reduced from 80.6 to 70.4%, and the outer surface pore size from 0.54 to 0.09 m, but the specific surface area increased from 10.4 to 14.5 m2 /g. Since CA acted as the matrix polymer,
it is easy to understand that more CA in the spinning dope solution resulted in the formation of denser matrix networks, hence lower porosity, smaller pore sizes but greater specific surface areas. However, there was a limitation in the polymer concentrations that can be used to prepare the spinning dope solutions as the high viscosity can eventually cause the dope solution to be non-spinnable. To further illustrate the effect of the CA concentrations on the morphologies of the hollow fiber membranes, Figs. 4 and 5 show the SEM images of the outer and inner surfaces of the hollow fiber membranes fabricated at different CA concentrations. Extremely open porous outer surfaces with a latex structure were obtained in the CA concentrations ranging from 12 to 16 wt% (see Fig. 4a–c) while a much less porous surface was obtained at the CA concentration of 18 wt% (see Fig. 4d). Although water was used as both the external and internal coagulants, the inner surfaces appeared to be even more porous with much larger pore sizes than the corresponding outer surfaces, with beak-like structure at the CA concentration of 12 wt% (see Fig. 5a) or latex structure at the CA concentrations of 14–18 wt% (see Fig. 5b–d). The larger inner surface pore sizes are also clearly shown in Fig. 3b. As seen in both Figs. 5 and 3b, the average pore sizes of the inner surfaces also decreased with the increase of the CA concentrations. Although the external and internal coagulation processes occurred simultaneously in the wet spinning system, the coagulation behavior at the lumen side can be different from that at the shell side [25]. The internal coagulant or bore fluid (i.e., water here) was small in the amount and can soon become a mixture of water and the polymer solvent (i.e., FA), and, as a consequence, the coagulation rate of the polymers at the lumen side can be much slower due to the lowered pH than that at the shell side with a large quantity of external coagulant (i.e., water). Hence, more porous inner surfaces were formed and the delayed phase separation (and
Fig. 4. SEM images showing the outer surfaces of the CS/CA blend hollow fibers prepared at CS concentration of 2 wt% but CA concentration of 12 wt% (a), 14 wt% (b), 16 wt% (c) and 18 wt% (d) (water was used as both the external and internal coagulant).
C. Liu, R. Bai / Journal of Membrane Science 279 (2006) 336–346
341
Fig. 5. SEM images showing the inner surfaces of the CS/CA blend hollow fibers prepared at CS concentration of 2 wt% but CA concentration of 12 wt% (a), 14 wt% (b), 16 wt% (c) and 18 wt% (d) (water was used as both the external and internal coagulant).
possibly the erosion of the un-solidified polymers) at the lumen side resulted in the formation of the larger pores. 3.3. Effect of chitosan concentrations CS was added as a functional polymer to provide the CS/CA blend hollow fiber membranes with excellent adsorptive performance [23]. Although it was desirable to fabricate the CS/CA blend hollow fiber membranes with greater amounts of CS, but the addition of CS significantly increased the viscosity of the spinning dope solution and the spinning process became increasingly more difficult. It was found that when CS concentration exceeded 4 wt%, a non-flow behavior of the spinning dope solution occurred. Therefore, the effect of CS concentration on the morphologies and structures of the blend hollow fiber membranes was examined at the CS concentrations from 2 to 4 wt% with the CA concentration being set constant at 12 wt% in this case. Again, water was used as both the external and internal coagulants in the study. In general, all the hollow fiber membranes showed the spongy-like and macrovoids-free porous structures, similar to those discussed in Section 3.2. However, the most significant differences in this case are that a small increase in the CS concentration would largely decrease the pore sizes of the hollow fiber membranes, as shown by the typical SEM images in Fig. 6. The effect of CS concentrations on the surface pore sizes, porosities and specific surface areas of the hollow fibers are shown in Fig. 7. The average outer surface pore sizes were found to decrease from 0.54 to 0.22 and to 0.063 m when CS concentration was increased from 2 to 3 and 4 wt%, respectively (see Fig. 7a). The surface pore size of the hollow fiber membrane at the CS concentration of 4 wt% was only about one-ninth of that
of the hollow fiber membrane at the CS concentration of 2 wt%. This phenomenon may be attributed to the much higher viscosity of the dope solution at higher CS concentrations. As expected, the porosities of the hollow fibers were not significantly changed due to the addition of a small amount of CS in the blend (see Fig. 7b). The results support the advantages of entrapping CS in the CA matrix to increase the reactivity of the blend hollow fiber membranes for improved or enhanced adsorptive performance [23]. Also, the addition of CS may be used as an effective way to change the pore sizes of the blend hollow fiber membranes when hollow fibers of different pore sizes are desirable for applications involving the separation of substances with different sizes. 3.4. Effect of coagulant composition The hollow fiber membranes mentioned in Sections 3.2 and 3.3 were fabricated by using water as both the external and internal coagulants. Generally, most polymeric hollow fiber membranes in literature are prepared through the wet or dry-jet wet spinning process of a polymer dope solution, with water being frequently used as the coagulant, largely due to the fact that water is a good nonsolvent for many polymers, has high mutual affinity with many polymer solvents, and is inexpensive as a large quantity of the external coagulant is usually needed. For the CS/CA blend spinning dope solutions, the results in Sections 3.2 and 3.3 also clearly illustrate that water can be used as the coagulant to make highly porous and macrovoids-free hollow fiber membranes. The relatively uniform porous structures of the hollow fibers suggest that delayed demixing of the hollow fibers during phase separation/inversion took place and water was a relatively weak coagulant for the CS/CA/FA ternary spinning dope solutions.
342
C. Liu, R. Bai / Journal of Membrane Science 279 (2006) 336–346
Fig. 7. Effect of CS concentrations in the spinning solution (with 12 wt% CA) on the structural characteristics of the CS/CA blend hollow fibers: (a) averaged pore size of outer surfaces and (b) specific surface areas and porosities (water was used as both the external and internal coagulant).
Fig. 6. SEM images showing the outer surfaces of the hollow fibers prepared at CA concentration of 12 wt% but CS concentration of 2.0 wt% (a), 3.0 wt% (b) and 4.0 wt% (c). Water was used as both the external and internal coagulants.
In literature, the now well established wet phase inversion method for fabrication of hollow fiber membranes usually uses aprotic polymer solvents and water as the coagulant, resulting in the production of asymmetric membranes, typically with a thin dense top layer (or skin) supported on a porous layer with macrovoids. The skin is formed by the instantaneous demixing of the membranes at the surface with the strong polymer nonsolvent (which also has high mutual affinity with the polymer solvent). One of the major difference in this work from many others in literature lies in that a protic polymer solvent, instead of aprotic polymer solvent, was used. Until now, there have only been a few reports on the fabrication of polyamide membranes employing protic polymer solvents [26–28]. In addition, Strathmann et al. have also demonstrated that one type of polymer dope solution may produce a range of membrane structures (spongylike to finger-like voids), depending on the choice of solvent and nonsolvent [29,30]. In the following, we present the results
for the CS/CA blend hollow fiber membranes fabricated with some alkali or acid added into the water coagulant to change the coagulant compositions. The interest in examining this was also arising from the cloud point study in Section 3.1 where the results show that using water or NaOH solution as coagulant affected the cloud point or phase separation of the CS/CA blend solutions. 3.4.1. Effect of external coagulant compositions Instead of using water as the external coagulant, NaOH solution (3 wt%), NaAc solution (10 wt%) and FA solution (3.6 × 10−4 wt%) were examined as the external coagulant to spin CS/CA blend hollow fibers with the spinning dope solutions containing 2 wt% CS and 12 wt% CA (see Table 1). In the case of using the NaOH solution, the hollow fibers were observed to show a faster phase separation/inversion than using water in the coagulation bath, supporting that the NaOH solution is a relatively stronger coagulant than water, and agreeing with the results from the cloud point experiments. The resultant hollow fibers were observed to have some sparkly distributed “tear-drop” shaped macrovoids which appeared near the outer surface of the fiber wall (see Fig. 8). With the use of the NaAc and FA solutions, the resultant hollow fibers did not show any such macrovoids across the cross-sections of the hollow fibers (results not shown), indicating that the NaAc and FA solutions were relatively weaker coagulants than the NaOH solution for the CS/CA blend hollow fibers.
C. Liu, R. Bai / Journal of Membrane Science 279 (2006) 336–346
343
Fig. 8. Overall view of the CS/CA blend hollow fibers prepared with 3.0 wt% NaOH solution as the external coagulant (CS/CA in the dope solution was 2.0/12.0 (g/g), and water was used as the internal coagulant).
The SEM images showing the outer surfaces of these hollow fibers from the analysis are given in Fig. 9 and the corresponding surface pore sizes, specific surface areas and porosities are also given in Table 3. It is clear that the hollow fibers generally had smaller surface pore sizes (surface pores from 0.47 to 0.087 to 0.049 m) and consequently greater specific surface areas (from 10.4 to 11.2 to 11.7 m2 /g) when the external coagulant was changed from the FA to NaAc to NaOH solutions (i.e., the solution pH increased). This result may be attributed to the relatively more rapid coagulation rate of the hollow fibers in a more basic coagulation solution. With basic (or alkali) coagulants, the porosity of the membranes also appeared to be slightly reduced (see Table 3). 3.4.2. Effect of internal coagulant (bore fluid) compositions In Section 3.2, it has been implied that the structures of the CS/CA blend hollow fiber membranes were sensitive to the internal coagulant because of its small flow rate and the strong polymer solvent (i.e., FA) used. In this part of the study, only NaOH solutions (i.e., strong nonsolvent) were examined. The spinning dope solutions were prepared with 12 wt% CA, plus 2 or 3 wt% CS. Water was used as the external coagulant, but NaOH solutions at 1–3 wt% were used as the internal coagulant, respectively. It was found that for the hollow fibers prepared with 2 wt% CS, large macrovoids were formed near the lumen side in the hollow fibers and more and larger macrovoids appeared for the NaOH solution of a higher concentration (see Fig. 10). In contrast, the hollow fibers prepared with 3 wt% CS did not show apparent macrovoids in the cross-sections (results not shown) for the NaOH solutions studied, possibly due to the high viscosity of the spinning dope solution. It is interesting
Fig. 9. Effect of external coagulant composition on the outer surface morphology: (a) 3 wt% NaOH solution, (b) 10 wt% NaAc solution and (c) FA solution (3.6 × 10−4 wt%) with pH of 3.21 (CS/CA in the dope was 2.0/12.0 (g/g), and water was used as the internal coagulant).
to note that the shape of these macrovoids was different from the tear-drop shape shown in Fig. 8 and also different from the typical finger-like shape usually reported in literature (the macrovoids in this case were very wide). The special shape of the macrovoids in this work may be attributed to the different phase separation behaviors of the CS/CA blend from many oth-
Table 3 Effect of external coagulant composition on the structural characteristics of the CS/CA blend hollow fibers (CS/CA in the dope was 2.0/12.0 (g/g), and water was used as the internal coagulant) External coagulant
Average pore size of outer surface (m)
Specific surface area (m2 /g)
Porosity (%)
3 wt% NaOH 10 wt% NaAc 3.6 × 10−4 wt% FA Water
0.049 0.087 0.47 0.54
11.7 11.2 10.4 10.4
79.0 79.2 80.4 80.6
344
C. Liu, R. Bai / Journal of Membrane Science 279 (2006) 336–346
Fig. 10. Effect of internal coagulant (bore fluid) composition on the cross-sectional structures of the CS/CA blend hollow fibers. NaOH concentration was (a) 2 wt% and (b) 3 wt% (CS/CA in the dope was 2.0/12.0 (g/g), and water was used as the external coagulant).
ers in literature using a single polymer in the spinning dope solution. The nucleation theory [31] has often been used to explain the formation of macrovoids. After a top surface layer is formed, the nonsolvent will penetrate through the top thin layer into the hollow fiber wall. During the period of growth of a droplet of nonsolvent in the sublayer, the influx of solvent from surrounding polymer solutions into the nonsolvent droplet makes the droplet a solvent and nonsolvent mixture. Hence, the surrounding polymer solution remains in the form of liquid and the solidification (or phase inversion) of the polymers could not be possible. Only when more and more solvent is extracted (at the same time the droplet becomes larger and larger in size), the surrounding polymers start to have phase separation. After the surrounding polymers are completely solidified, the spaces occupied by the large droplets of the solvent and nonsolvnet mixture thus form the macrovoids. The formation of macrovoids depends on the ratio of influx of the nonsolvent from the coagulant into the sublayer and the influx of solvent from the surrounding polymer solutions into the nonsolvent droplet. One of the factors dominating the influxes of nonsolvent and solvent is the kinetic hindrance from the polymer solutions which are dependent on the polymer concentrations and thus on the viscosities of the dope solution. In this work, the blend dope solution at 3 wt% CS showed much higher viscosity than that at 2 wt% CS. At such a high viscosity, the influx of the FA from the surrounding polymer solutions into the nonsolvent droplet was very slow. Under this condition, due to the high concentration nonsolvent in contacting with the surrounding polymers, the surrounding polymers solidified rapidly before a large
droplet of sovent and nonsolvent mixture formed, leading to the formation of microporous structures rather than macrovoids structures. Although hollow fibers prepared at 3 wt% CS did not show macrovoids in the cross-sections, a tendency of phase separation into polymer rich and polymer lean phases was also observed when NaOH solution was used as internal coagulant. The inner edges of the hollow fibers are shown in Fig. 11. It can be seen that a more porous region existed below the relatively denser inner surfaces. With increasing the NaOH concentration in the bore fluid, the pore size and porosity of this subsurface region obviously increased. Moreover, the starting point of this region was much closer to the inner surfaces with increasing NaOH concentration in the bore fluid. This might be caused by the different phase separation behaviors of CS and CA and will be examined further in the future. SEM images showing the inner surfaces of these hollow fibers are given in Fig. 12. Clearly, all these membranes had open porous inner surfaces, but the pore sizes were much smaller than those prepared with water as the internal coagulant (bore fluid) shown in Fig. 5. Also, a slight reduction in the surface pore sizes was observed with the increase of the NaOH concentrations in the bore fluid and with the increase of CS concentrations in the spinning dope solutions. The corresponding specific surface areas and porosities of these hollow fiber membranes are given in Table 4. At CS concentration of 2 wt%, the porosity increased at a high NaOH solution, which may be attributed to the presence of more and larger macrovoids as discussed earlier, though the changes in the specific surface areas were not significantly in this case. At CS concentration of 3 wt%, however, the changes
Fig. 11. Effect of internal coagulant (bore fluid) composition on the inner edges of the CS/CA blend hollow fibers. NaOH concentration was (a) 2 wt% and (b) 3 wt% (CS/CA in the dope was 3.0/12.0 (g/g), and water was used as the external coagulant).
C. Liu, R. Bai / Journal of Membrane Science 279 (2006) 336–346
345
Fig. 12. Effect of internal coagulant (bore fluid) composition on the inner surfaces of the CS/CA blend hollow fibers. (a) CS/CA = 2.0/12.0 (g/g), 2 wt% NaOH; (b) CS/CA = 2.0/12.0 (g/g), 3 wt% NaOH; (c) CS/CA = 3.0/12.0 (g/g), 2 wt% NaOH; (d) CS/CA = 3.0/12.0 (g/g), 3 wt% NaOH. Water was used as the external coagulant. Table 4 Effect of bore fluid composition and CS concentration on the structural characteristics of the CS/CA blend hollow fibers (water was used as the external coagulant) Dope composition by weight (CS/CA/FA)
Bore fluid composition
Specific surface area (m2 /g)
Porosity (%)
2.0/12.0/86.0
DI water 2 wt% NaOH 3 wt% NaOH
10.4 10.9 11.3
80.6 82.3 86.7
3.0/12.0/85.0
DI water 2 wt% NaOH 3 wt% NaOH
12.2 13.0 13.4
79.7 79.0 79.2
Table 5 Typical results of tensile stress and stain of some of the CS/CA blend hollow fibers Dope composition by weight (CS/CA/FA)
Composition of external coagulant
Bore fluid composition
Tensile stress (MPa)
Strain (%)
2.0/12.0/86.0 2.0/18.0/80.0 3.0/12.0/85.0 3.0/12.0/85.0
Tap water Tap water Tap water Tap water
DI water DI water DI water 3 wt% NaOH
7.8 18.2 7.6 8.3
22.1 27.6 23.7 25.3
of both the specific surface areas and the porosities did not appear to be significant. 3.5. Tensile stress and strain of the blend hollow fiber membranes Although for adsorptive membranes the mechanical strength is not as critical as the filtration membranes (because the process pressures do not have to be high), we examined the tensile stress and strain of the porous hollow fiber membranes developed in this study and some of the typical results are given in Table 5. It is found that the blend hollow fibers generally had sufficiently high tensile stress (7.8–8.3 MPa) and break elongations (22.1–25.3%) even for the highly porous membranes prepared
at low polymer concentrations (CS/CA = 2/12 or 3/12). With the increase of the CA concentration up to 18 wt%, the tensile stress was increased significantly from 7.8 to 18.2 MPa while the strain also increased moderately. These changes can be attributed to the increased cohesions among the CA molecules which formed the membrane matrix. The mechanical strength of the hollow fiber membranes also appears to be strengthened when NaOH solution instead of water was used as the internal coagulant. 4. Conclusions Highly porous CS/CA blend hollow fibers for adsorptive membranes have been successfully fabricated through a wet spinning process with CA in the concentration range of
346
C. Liu, R. Bai / Journal of Membrane Science 279 (2006) 336–346
12–18 wt% and CS concentration at up to 4 wt% in the spinning dope solutions. Depending on the coagulant compositions, the outer surface pore sizes, the specific surface areas and the porosities of the blend hollow fibers can change from 0.54 to 0.049 m, 10.4 to 14.5 m2 /g and 80.6 to 70.4%, respectively, with the increase of CA or CS amount in the spinning dope solutions for the polymer concentrations studied. Water can be used as both the external and internal coagulants in the fabrication process and the resultant hollow fibers showed spongy-like, macrovoids-free and relatively uniform porous structures which are desirable for adsorptive membranes, attributed to water being a weaker coagulant for CS and CA. The composition of the coagulants, especially the internal coagulant, also greatly affected the blend hollow fibers’ structures. By increasing the alkalinity of the coagulants, the coagulation rate of the blend hollow fibers was increased, resulting in the formation of relatively denser surface layers and smaller surfaces pore sizes. In particular, when NaOH solutions (1–3 wt%) were used as the internal coagulant, more and larger macrovoids were formed in the blend hollow fibers at the near lumen side, when the concentration of NaOH solution was increased (>1 wt%) and the CS concentration in the spinning dope solutions was low (<3 wt%). The present work demonstrates that the CS/CA blend hollow fibers can be made into highly porous adsorptive membranes with large specific surface areas and various desirable pore sizes, by properly controlling the CS and CA concentrations in the spinning dope solutions and by choosing the compositions of the external and internal coagulants. Acknowledgements The authors are thankful to Professor Neal Chung (Department of Chemical and Biomolecular Engineering) for using some analytical facilities in his laboratory. Thanks are also given to Associate Professor Toh Siew Lok and Mr. Zhang Yanzhong (Department of Mechanical Engineering) who helped in the mechanical strength test. The financial support of the Academic Research Funds from the National University of Singapore is also acknowledged. References [1] K. Kugel, A. Moseley, G.B. Harding, E. Klein, Microporous poly(caprolactam) hollow fibers for therapeutic affinity adsorption, J. Membr. Sci. 74 (1992) 115–129. [2] D. Keith Roper, N. Edwin, Lightfoot, Separation of biomolecules using adsorptive membranes, J. Chromatogr. A 702 (1995) 3–26. [3] X. Zeng, E. Ruckenstein, Supported chitosan-dye affinity membranes and their protein adsorption, J. Membr. Sci. 117 (1996) 271–278. [4] W. Kaminski, Z. Modrzejewska, Application of chitosan membranes in separation of heavy metal ions, Sep. Sci. Technol. 32 (1997) 2659–2668. ¨ Genc¸, C. Arpa, Bayramo˘glu, M.Y. Arica, S. Bektas¸, Selec[5] O. tive recovery of mercury by Procion Brown MX 5BR immobilized poly(hydroxyethylmethacrylate/chitosan) composite membranes, Hydrometallurgy 67 (2002) 53–62. [6] M.A. Teeters, S.E. Conrardy, B.L. Thomas, T.W. Root, E.N. Lightfoot, Adsorptive membrane chromatography for purification of plasmid DNA, J. Chromatogr. A 989 (2003) 165–173. [7] M.E. Avramescu, W.F.C. Sager, Z. Borneman, M. Wessling, Adsorptive membranes for bilirubin removal, J. Chromatogr. B 803 (2004) 215–223.
[8] M.M. Beppu, E.J. Arruda, R.S. Vieira, N.N. Santos, Adsorption of Cu(II) on porous chitosan membranes functionalized with histidine, J. Membr. Sci. 240 (2004) 227–235. [9] Z. Chen, M. Deng, Y. Chen, G. He, M. Wu, J. Wang, Preparation and performance of cellulose acetate/polyethyleneimine blend microfiltration membranes and their applications, J. Membr. Sci. 235 (2004) 73–86. [10] W. Shi, F. Zhang, G. Zhang, Adsorption of bilirubin with polylysine carrying chitosan-coated nylon affinity membranes, J. Chromatogr. B 819 (2005) 301–306. [11] C.X. Liu, R.B. Bai, Recovery of bovine serum albumin (BSA) by tripolyphosphate (TPP) crosslinked chitosan membrane, in: Presented at Recent Advances in Separation & Purification Techniques for Biological & Pharmaceutical Products Development, NTU, Singapore, 21–23 February, 2005. [12] B. Xia, G. Zhang, F. Zhang, Bilirubin removal by Cibacron Blue F3GA attached nylon-based hydrophilic affinity membrane, J. Membr. Sci. 226 (2003) 9–20. [13] S.-Y. Suen, Y.-C. Liu, C.-S. Chang, Exploiting immobilized metal affinity membranes for the isolation or purification of therapeutically relevant species, J. Chromatogr. B 797 (2003) 305–319. [14] V. Karakoc¸, H. Yavuz, A. Denizli, Affinity adsorption of recombinant human interferon-␣ on a porous dye-affinity adsorbent, Colloid Surf. A 240 (2004) 93–99. [15] H. lwata, K. Saito, S. Furusaki, Adsorption characteristics of an immobilized metal affinity membrane, Biotechnol. Progr. 7 (1991) 412–418. [16] K. Kugel, A. Moseley, G.B. Harding, E. Klein, Microporous poly(caprolactam) hollow fibers for therapeutic affinity adsorption, J. Membr. Sci. 74 (1992) 115–129. [17] A. Fabre, Hydrophilic PVDF semipermeable membrane, US Pat., 4,810,384. [18] K. Rodemann, E. Staude, Synthesis and characterizationof affinity membranes made from polysulfone, J. Membr. Sci. 88 (1994) 271–278. [19] E. Klein, Affinity membranes: a 10-year review, J. Membr. Sci. 179 (2000) 1–27. [20] H. Matsuyama, S. Berghmans, D.R. Lloyd, Formation of hydrophilic microporous membranes via thermally induced phase separation, J. Membr. Sci. 142 (1998) 213–224. [21] X. Zeng, E. Ruckenstein, Cross-linked macroporous chitosan anionexchange membranes for protein separations, J. Membr. Sci. 148 (1998) 195–205. [22] M.E. Avramescu, W.F.C. Sager, M.H.V. Mulder, M. Wessling, Preparation of ethylene vinylalcohol copolymer membranes suitable for ligand coupling in affinity separation, J. Membr. Sci. 210 (2002) 155–173. [23] C.X. Liu, R.B. Bai, Preparation of chitosan/cellulose acetate blend hollow fibers for adsorptive performance, J. Membr. Sci. 267 (2005) 68–77. [24] J.J. Shieh, T.S. Chung, Effect of liquid–liquid demixing on the membrane morphology, gas permeation, thermal and mechanical properties of cellulose acetate hollow fibers, J. Membr. Sci. 140 (1998) 67–79. [25] D.L. Wang, W.K. Teo, K. Li, Preparation and characterization of highflux polysulfone hollow fibre gas separation membranes, J. Membr. Sci. 204 (2002) 247. [26] R.M. Macdonogh, C.J.D. Fell, A.G. Fane, Characteristics of membranes formed by acid dissolution of polyamides, J. Membr. Sci. 31 (1987) 321. [27] C.W. Yao, R.P. Burford, A.G. Fane, C.J.D. Fell, R.M. Macdonogh, Effect of coagulation conditions on structure and properties of membranes from aliphatic polyamides, J. Membr. Sci. 38 (1988) 113–125. [28] Pall, Process for preparing hydrophilic polyamide membrane filter media and prodict, U.S. Patent, 4,340,479. [29] H. Strathmann, K. Kaock, P. Amar, R.W. Baker, The formation mechanism of asymmetric membranes, Desalination 16 (1975) 179–203. [30] H. Strathmann, Production of microporous media by phase inversion process, in: D.R. Lloyd (Ed.), Material Science of Synthetic Membranes, ACS symposium Series 269, American Chemical Society, Washington, DC, 1985, pp. 165–195. [31] C.A. Smolders, A.J. Reuvers, R.M. Boom, I.M. Wienk, Microstructures in phase-inversion membranes. Part 1. Formation of macrovoids, J. Membr. Sci. 73 (1992) 259–275.