cellulose acetate blend hollow fibers for adsorptive performance

Journal of Membrane Science 267 (2005) 68–77

Preparation of chitosan/cellulose acetate blend hollow fibers for adsorptive performance Chunxiu Liu a , Renbi Bai b,∗ b

a Department of Chemical and Biomolecular Engineering, Faculty of Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260, Singapore Division of Environmental Science and Engineering, Faculty of Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260, Singapore

Received 9 March 2005; received in revised form 31 May 2005; accepted 7 June 2005 Available online 18 July 2005

Abstract Novel chitosan/cellulose acetate blend hollow fibers were prepared by the wet spinning method to obtain adsorptive membranes. In the hollow fiber membranes, cellulose acetate (CA) acted as a matrix polymer and chitosan (CS) as a functional polymer to provide the membrane with coupling or reactive sites for affinity-based separations. Formic acid was used as the co-solvent for both CA and CS to prepare the dope solution and NaOH solution was used as the external and internal coagulants in the wet spinning fabrication process. The mixability and possible interactions between CA and CS in the blend hollow fibers were investigated through FTIR and XRD analyses. FTIR spectroscopy revealed that the blending of CS with CA involved some chemical interactions. XRD results showed a single peak for CS and CA blend hollow fibers, similar to the one for CA, suggesting that good mixability was achieved between the two polymers. The properties of the blend hollow fibers were characterized through water flux measurements, surface and cross-section examinations and adsorption performances to copper ions and bovine serum albumin (BSA) on the surfaces, and were compared with those for CA hollow fibers. SEM image clearly showed the sponge-like and macrovoids-free porous structures of the hollow fibers. The blend hollow fibers displayed good tensile stress even though the tensile stress reduced with the increase of the CS content in the blend. The blend hollow fibers achieved significantly better adsorption performance as compared to that of CA hollow fibers, indicating the benefit in adding CS into CA to make novel blend hollow fibers in improving the performance of the traditional CA hollow fibers, especially for the affinity-based separation applications. © 2005 Elsevier B.V. All rights reserved. Keywords: Chitosan; Cellulose acetate; Blend hollow fiber; Fabrication; Adsorption; Copper ion; BSA

1. Introduction Microfiltration membranes with surface functional groups that can be used as coupling sites or adsorptive sites for separation are of great interest in biomedical and many other industrial and environmental applications [1]. With such membrane-based separation systems, the short diffusion distances caused by the convective flow across the membrane thickness can allow fast purification of biomolecules [2]. Many commercial polymeric membrane materials, such as polysulfone (PS), polyethersulfone (PES), polyvinylidene ∗

Corresponding author. Tel.: +65 68744532; fax: +65 67791936. E-mail address: [email protected] (R. Bai).

0376-7388/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2005.06.001

difluoride (PVDF), polypropylene (PP) and nylon, etc., have good chemical, thermal and mechanical stability, but they are usually lack of reactive functional groups on the polymer backbones. Hence, membranes fabricated from these materials often have to be modified to eliminate the non-specific type of adsorption and to enhance the separation efficiency through improved adsorptive surfaces, especially for biomolecules such as proteins. However, membrane modification through various harsh physical and chemical post-treatment can result in undesirable and irreproducible inhomogeneities in the membranes [3]. Compared with the above-mentioned materials, cellulose and its derivatives are hydrophilic and have reactive hydroxyl groups, which can be modified with other reactive

C. Liu, R. Bai / Journal of Membrane Science 267 (2005) 68–77

functional groups to obtain adsorptive membranes. Among the various reactive functional groups such as hydroxyl group, amine group and sulfhydryl group, the amine groups or sulfhydryl groups are however much more reactive than the hydroxyl groups [4] and can be used directly as affinity adsorption sites or can be much more easily attacked by other modifying agents under mild conditions. Therefore, one of the choices to prepare adsorptive membranes can be to introduce some amine groups into cellulose or its derivatives (e.g., cellulose acetate) as the membrane materials. In recent years, chitosan (CS), a biopolymer widely available from seafood processing waste, has been increasingly studied as an adsorptive material for various applications [5–8], due to its abundance in the free amine groups. Most of these studies however used chitosan in the form of powders, flakes, or gel beads. There has been considerable research interest to prepare chitosan membranes (flat or preferably in hollow fiber form) for adsorptive separation purpose [9–11], but the scope of preparing pure chitosan membranes has been largely limited due to the poor mechanical strength and chemical stability of chitosan. To improve the mechanical strength, chitosan has often been coated on supports such as flat PES membranes [12] and cellulose membranes [13,14] to make composite chitosan membranes. The coating method has however encountered some problems, including easy detachment of the coated chitosan film or incomplete and non-uniform coverage of the support membranes. More recently, blending chitosan with other polymers has been found to be an effective way to overcome the shortcomings of chitosan [15–21], because blending at the microscopic level due to chemical interactions may form additional chemical bonds. Hasegawa et al. [17], and Li and Bai [21] have found that the anti-acid stability of chitosan was enhanced when chitosan was blended with cellulose. Therefore, blending chitosan with other high strength polymers, e.g., cellulose or its derivatives, is an attractive way to prepare adsorptive membranes for practical applications. In this study, CS was blended with cellulose acetate (CA) to spin blend hollow fiber membranes. CA was used as the polymer matrix and CS as the functional polymer to provide amine groups for the hollow fiber membrane in order to make it adsorptive. The choice of CA as the polymer matrix is due to its long-known good mechanical strength in fabricating hollow fibers. Two types of CS/CA blend hollow fibers (i.e., Blend I and Blend II) and pure CA hollow fibers were fabricated through a wet spinning method. Formic acid was used as the co-solvent for CS and CA to prepare the dope solution and NaOH solution was used as both the external and internal coagulants. The mixability and possible interaction of the two polymers in the blend hollow fibers were investigated through FTIR and XRD analyses. The morphologies, water permeate flux, contact angle and mechanical strength of the blend hollow fibers were also examined. The adsorption performance of the hollow fibers

69

was evaluated through the adsorption of copper(II) ions and bovine serum albumin (BSA). To our knowledge, CS/CA blend hollow fibers have so far not been reported in the literature.

2. Experimental 2.1. Materials Chitosan (CS), practical grade from crab shells, was supplied by Aldrich and used as received. The viscosityaveraged molecular weight of CS, as determined from the Mark–Houwink equation [22] in the study, was 319,000 g/mol. The N-deacetylation degree (DDA) of CS was measured with the titration method [23] and was found to be 79%. Cellulose acetate (CA) with an acetyl content of 40% was purchased from Fluka. Formic acid (98–100%) from Fluka was used as co-solvent for both CS and CA. Copper sulphate (CuSO4 ·5H2 O) and bovine serum albumin (BSA) from Sigma were used in the adsorption experiments. 2.2. Fabrication of hollow fiber membranes CA and CS were first dried at 65 ◦ C to remove extra moisture. The spinning dope solution was prepared by mechanically stirring CA and CS together in formic acid solution at 209.44 × 10−1 rad/s (200 rpm) overnight. The resultant dope solution was then degassed by leaving it in the dope tank overnight without stirring to free the air bubbles entrapped in the dope solution, and it was finally filtered through a 15 ␮m stainless steel filter under high pressure N2 gas to remove any insoluble particles. The hollow fibers were spun through a wet spinning process. The clear dope solution was forced through a stainless steel spinneret comprising an annular ring (with o.d. and i.d. of 1.3 and 0.5 mm, respectively) under high pressure N2 gas, and was protruded into the coagulation bath. A core liquid coagulant was delivered simultaneously through the inner core by a high-pressure syringe pump (ISCO 100DX). NaOH solution (3%, w/w) was used as both the external coagulant (in the coagulation tank) and the internal coagulant (the core flow). The hollow fiber was formed and solidified in the coagulation bath and the collection rate of the hollow fiber was carefully controlled so that there was no excess drag force imposed on the fiber. As soon as the fiber was drawn out of the coagulation bath, it was rinsed with tap water to leach out excess solvents and external/internal coagulation solution. The clean hollow fiber was then stored in DI water for further use. To get dry product, hollow fiber was subjected to 1-propanol and 1-heptane multi-step solvent exchange and then dried at room temperature (22–23 ◦ C) under tension. The purpose of multi-step solvent exchange process was to prevent the pore collapse resulting from the capillary force during drying [24].

C. Liu, R. Bai / Journal of Membrane Science 267 (2005) 68–77

70

2.3. Characterization of hollow fiber membranes FTIR studies were conducted with a Shimadzu H8400 spectrometer in the wave number range of 500–4000 cm−1 . Dried fiber was ground into powder and approximately 1–2 mg was mixed with 100 mg of KBr to prepare the sample. FTIR analyses were done on CA, CS and CA/CS blends. X-ray diffraction (XRD) measurements for the hollow fibers were carried out on a diffractometer (Lab XRD-6000) ˚ with the Cu K␣ ray source (λ = 1.54 × 10−10 m (1.54 A)). Scanning diffraction angle range was set at 2–40◦ and scanning rate was 2◦ /min (2θ). The spectra were recorded at 40 kV, 25 mA. The structure and morphology of the blend hollow fibers were investigated through FESEM (JEOL JSM-6700F FESEM) and SEM (JEOL JSM-5600 SEM). The dried hollow fibers were snapped under 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 40 mbar vacuum. For FESEM/SEM analyses, the electrical voltage was controlled at below 10 kV to prevent possible collapse of the membrane surface caused by electron beam scanning. The dimensions of the hollow fibers were measured by the software supplied by the manufacturers of the equipment. Pure water flux (PWF) measurements of the hollow fiber membranes were carried out in a dead end filtration set-up at room temperature. Ten fibers (length 20 cm) were assembled into the test module with glue. Pressure drop across the hollow fiber membranes was controlled by compressed N2 gas. In all runs, the pressure difference was maintained at 5.8 bars. Pure water was loaded into the lumen side and permeated out from the outer side. After the flux through the membrane stabilized, the permeate was collected and measured at desired time intervals. Each type of hollow fibers was tested three times and the average flux was calculated. The PWF was determined from the following expression: Jw =

Q (A T p)

(1)

where Jw is the water flux (L/m2 h bar), Q the quantity of water permeated (L), T the sampling time (h), A the membrane outer surface area (m2 ), and p the pressure drop across the hollow fiber membranes (bar). The water contact angle of the hollow fiber’s surfaces was measured in an automated contact angle goniometer (Model 100-0-230, Rame-Hart Inc.) equipped with digital imaging software. The sessile drop method was used. After the dried fiber was mounted in the telescope, 0.5 ␮L distilled water was dropped on the surface to measure the contact angle. The relative humidity and temperature in the test were 60% and 22 ◦ C, respectively. For each contact angle value reported, five readings from different parts of the fiber surface were measured and averaged.

The mechanical property of the wet hollow fiber was evaluated through the measurement of tensile strength, elongation ratio, and Young’s modulus at break. Tests were conducted with Instron 5542 Material Testing Instrument 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. Values of breaking force and elongation at break were then recorded. For reliability, five readings were taken for each sample, and the average value was used in this paper. The internal surface areas of the CS/CA blend hollow fibers were determined by N2 -sorption employing the BET method and using a Quanta Chrome Nova 3000 series instrument. The amounts of the amine end-groups in the hollow fibers were analyzed with the ninhydrin method [25]. Adsorption experiments with CA or CS/CA blend hollow fibers were conducted for the adsorption of copper ions and bovine serum albumin (BSA) at room temperature. The dry hollow fibers were cut into about 0.5 cm length. For copper adsorption experiments, approximately 2 g of the hollow fiber pieces were added into 50 mL of the copper solution with initial concentration of 50 mg/L. The initial pH of the copper solution was adjusted to 5 or 6, respectively by adding a small amount of 0.1 M HCl or 0.1 M NaOH. The choice of pH 5–6 was to minimize the addition of acid but maintain copper in its ion form for better adsorption performance [21]. For the isotherm study, the initial concentrations of copper were varied from 10 to 300 mg/L and the initial solution pH was set at 6. The concentrations of copper ions in the samples were measured using an inductive coupled plasma (ICP) spectrometer. For the BSA adsorption experiment, approximately 1.8 g of the fiber pieces were added into 20 mL BSA solution with an initial concentration of 0.5 g/L and an initial solution pH of 6.3 (The choice of pH 6.3 was that the BSA solution prepared from dissolving the BSA solids in the DI water gave a solution pH of about 6.3 and hence no acid or alkali was added, which avoided any possible deformation and agglomeration of BSA molecules in the solution by adding acid or alkali [14].). The amount of BSA adsorbed on the hollow fibers was determined from the samples before and after adsorption with a UV–vis spectrometer at the wavelength of 278 nm. For isotherm study, the concentrations of BSA were varied from 0.2 to 1.3 g/L. All the adsorption experiments were conducted on an orbit shaker operated at 157.08 × 10−1 rad/s (150 rpm) and room temperature.

3. Results and discussion 3.1. Hollow fiber membranes Two blend CS/CA hollow fibers (Blend I and Blend II) and pure CA hollow fibers were prepared and the detailed

C. Liu, R. Bai / Journal of Membrane Science 267 (2005) 68–77

71

Table 1 Dope compositions and other information for the different types of hollow fibers (bore fluid and external coagulant: 3%, w/w NaOH solution) Type of hollow fibers

CA/CS/formic acid (g)

DER (g/min)

BFFR (mL/min)

o.d. (␮m)

i.d. (␮m)

Wall thickness (␮m)

Pure CA Blend I Blend II

27.0/0.0/73 26.5/0.5/73 26.0/1.0/73

0.488 0.402 0.411

0.167 0.133 0.133

611 671 644

350 429 362

131 121 141

DER: dope extrusion rate; BFR: bore fluid flow rate; o.d.: outer diameter; i.d.: inner diameter.

information is summarized in Table 1. The dope composition plays an important role in the mechanical properties of the hollow fibers. As the supporting matrix material, CA content in the dope determines the strength of the blend hollow fibers. In industrial fabrication process, CA content of 23–29% by weight in the dope was normally used to fabricate CA hollow fibers. It is well known that an increase in the polymer content in the dope leads to an increase in the mechanical strength, but too concentrated dope can make the spinning of hollow fibers impossible due to the high viscosity. For the present study, we set the total weight percentage of polymers in the dope at 27%. Due to the objective of adding CS into CA being to provide reactive amine groups for the hollow fibers and due to the high viscosity of CS polymer, the weight ratio between the two polymers of CS and CA were set at 0:27, 0.5:26.5, and 1:26, respectively, in the study (the attempt to spin hollow fibers at a higher CS content was difficult due to the non-flow behavior of the resulting dope). The resultant hollow fibers had a size around 600–700 ␮m (see Table 1). 3.2. FTIR analysis of the hollow fibers The nature of mixing between the two polymers is of interest to the study. FTIR has often been used as a useful tool in determining specific functional groups or chemical

bonds that exist in a material. The presence of a peak at a specific wavenumber would indicate the presence of a specific chemical bond. For CS and CA blending, if specific interactions took place between the two polymers, the most obvious and significant difference would be the appearance of new peaks or shift of existing peaks. Fig. 1 shows the FTIR spectra of CS powder, CA and CS/CA blend hollow fibers. The FTIR spectrum of CS in Fig. 1a shows peaks assigned to the saccharide structure at 899 and 1153 cm−1 , the amine group peak at around 1601 cm−1 , N-acetylated chitin at 1655 cm−1 and the OH and NH peaks centered at 3418 cm−1 . The FTIR spectrum of CA in Fig. 1b shows peaks for the C O functional groups at 1755 cm−1 , the OH functional groups at 3528 cm−1 , the CH3 groups at 1384 and 1249 cm−1 , and the ether C−O−C functional groups at 1060 cm−1 . The FTIR spectra for Blend I and Blend II hollow fibers in Fig. 1c–d are very similar to that of CA, probably due to the small percentage of CS in the blend hollow fibers. A most obvious difference in the spectra for CA, Blend I and Blend II is observed to be the shift of the broad peak for OH and NH groups from 3528 to 3483 and to 3479 cm−1 with the increase of CS content in the hollow fibers, indicating that an increased amount of amine groups or nitrogen atoms were incorporated into the CA matrix and interacted with the OH groups in the hollow fibers.

Fig. 1. FTIR spectra of (a) CS; (b) CA hollow fibers; (c) Blend I hollow fibers; and (d) Blend II hollow fibers.

C. Liu, R. Bai / Journal of Membrane Science 267 (2005) 68–77

72

Fig. 2. X-ray diffractograms of (a) CA hollow fiber; (b) Blend I hollow fibers; (c) Blend II hollow fibers; and (d) CS.

3.3. XRD analysis of hollow fibers The XRD spectra of CS, CA and Blend I and Blend II hollow fibers are shown in Fig. 2. As observed, CS showed two peaks at the two diffraction angles of 2θ = 10.26 and 19.86◦ , in agreement with the finding of Puttipipatkhachorn et al. [26], and CA displayed a single crystalline peak at 2θ = 13.06◦ . The diffractograms of Blend I and Blend II hollow fibers however showed nearly no obvious difference from that of CA. Since no other new peaks were formed, the results indicate that CA structure was not significantly affected by the addition of a small amount of CS and CS did not form its own crystalline region in the blend. The XRD results therefore confirm the good compatibility and mixability between CS and CA in the blends, because, if CS and CA had low compatibility in the blend hollow fibers, each component would show its own crystal region in the blend hollow fibers and the X-ray diffraction patterns should show a simply mixed pattern of CS and CA. The crystallinity and peak diffraction angles of CS, CA and the two blends are listed in Table 2. It can be seen that the crystalline was suppressed when CS content in the blends was increased. The diffraction angles also shifted slightly to a lower value from CA to Blend I and Blend II hollow fibers with the increase of CS content in the blends. The XRD results thus provide supporting evidence to the FTIR result that some specific chemical interaction between CA and CS existed in the blend. 3.4. Surface morphology The CA hollow fibers were transparent and colorless and the CS/CA blend hollow fibers were translucent and had a milk-white color. Both CA and CS/CA blend hollow fibers Table 2 Crystallinity and peak diffraction angles of CS, CA and Blend I and II hollow fibers

Crystallinity (%) Peak diffraction angle (◦ )

CS

Pure CA

Blend I

Blend II

7.28 10.26/19.86

6.65 13.24

6.42 13.06

6.34 13.00

displayed a sponge-like and macrovoids-free cross-section, which may be desirable for adsorptive membranes. Fig. 3 shows the typical cross-section morphologies of CA and CS/CA Blend II hollow fibers. The cross-sections are full of highly interconnected pores, which would provide high internal surface areas for the membranes. Heterogeneous structure with top skin supported by finger-like macrovoid substrate was often observed in many other types of hollow fibers fabricated through the wet phase inversion method, due to the fast solvent and non-solvent exchange rate. Macrovoids need to be avoided for adsorptive membranes because the macrovoids not only decrease internal surface area of membranes but also lead to non-homogeneous flow behavior of solutions through the membranes. Many studies have been devoted to the methods that can depress macrovoids, or reduce macrovoids to yield a sponge-like porous structure in adsorptive hollow fiber fabrication, including the use of high polymer concentrations [27], the addition of high viscosity component [28], and the induction of delayed demixing of dope [29]. The macrovoidsfree structure of the hollow fibers prepared in this study may be attributed to the high polymer concentration and viscosity of the dopes, which decreased the solvent–non-solvent exchange rate due to the high diffusion resistance from the polymer aggregates and therefore suppressed the macrovoid formation. Experiments showed that the CS/CA blend hollow fibers became much more porous and had larger pores than the CA hollow fibers. This is clearly shown in Fig. 3 where the crosssection of CS/CA Blend II hollow fibers is observed to be much more porous than that of CA. The outer and inner surfaces of CA and CS/CA Blend II hollow fibers are shown in Fig. 4. It can be found that the blend hollow fibers exhibited a porous outer surface while the CA hollow fibers showed a relatively dense surface. The formation of a more porous surface of two blends may be due to the fact that the addition of CS affected the phase inversion step of the blend during coagulation. The inner surfaces of CA and CS/CA blend hollow fibers showed even more significant difference in the pore size and porosity. Analyses indicated that the pore size and the porosity of the inner surfaces of the hollow fibers increased in the order of CA, followed by Blend I and Blend II hollow fibers, with average pore size of 50 nm for CA and 72 nm and 80 nm for Blend I and Blend II hollow fibers, respectively. In comparison with the outer surfaces, the inner surfaces of both CA and CS/CA blend hollow fibers had larger pore sizes. Unlike in the fabrication of conventional flat membrane where only one coagulation process is involved, hollow fiber fabrication involves two coagulation processes at the same time. One is through the continuous flow of bore fluid in the lumen of the fiber, while the other is through the outside contact with the coagulant as the hollow fibers enter the coagulation bath. Although both external and internal coagulant used in this study was 3% (w/w) NaOH solution, the coagulation behavior in the lumen side was different from that at outside. The NaOH solution in the bore fluid, which was of very small quantity can be quickly neutralized by the formic acid dif-

C. Liu, R. Bai / Journal of Membrane Science 267 (2005) 68–77

73

Fig. 3. Cross-section morphologies of CA ((a) ×700 and (b) ×30,000) and CS/CA Blend II hollow fibers ((c) ×1000 and (b) ×30,000).

fused from the polymer dope. Therefore, the coagulation rate of the fiber at the lumen side was slower than that at outside, leading to a delayed phase separation and therefore more porous inner surfaces. This result also indicates that a more

dilute NaOH solution (<3%, w/w) should be used as outside coagulant in order to have the outer surfaces to be as porous as the inner surfaces or in order to make the outer surface more porous.

Fig. 4. Surface morphologies of CA and CS/CA Blend II hollow fibers: (a) CA-outer surface; (b) CA-inner surface; (c) CS/CA Blend II-outer surface; (d) CS/CA Blend II-inner surface (×50,000).

C. Liu, R. Bai / Journal of Membrane Science 267 (2005) 68–77

74

Table 3 Pure water fluxes and water contact angles of CS, CA and CS/CA blend hollow fibers Hollow fibers

Pure water flux (×10−2 L/m2 h bar)

Water contact angle (◦ )

Chitosan 1.5%a Pure CA Blend I Blend II

– 9.06 9.22 9.14

63.0 36.5 40.0 47.5

a Chitosan film prepared with formic acid as solvent and NaOH as coagulant.

3.5. Pure water fluxes, contact angles and mechanical strength Table 3 presents the water flux results and the contact angles of the different types of hollow fibers. It can be seen that the water fluxes of the CS/CA blend hollow fibers were only slightly higher than that of CA hollow fibers, even though the results in Figs. 3 and 4 show the CS/CA blend hollow fibers being much more porous than the CA hollow fibers. This phenomenon may be explained by the change in the hydrophilic nature of the hollow fibers upon the addition of chitosan. As can be seen from Table 3, CA is relatively more hydrophilic than CS, which has a higher contact angle value than CA. The blending of CS with CA therefore resulted in an increase of the contact angle values of the blend hollow fibers, as compared to that of CA. Thus, the blend hollow fibers became slightly more hydrophobic than the CA hollow fibers, which can affect the water flux of the blend hollow fibers. The tensile stress, break elongation and Young’s modulus of the three types of hollow fibers are summarized in Table 4. Stress is defined as the force per unit area, normal to the direction of the applied force, and break elongation as the extension per gauge length at break. It is found that the blend hollow fibers displayed good tensile stress, break elongation and Young’s modulus values although these values tended to decrease with the increase of CS in the blend hollow fibers. The decrease of the mechanical strength of the blended hollow fibers may be attributed to the decreased cohesions between the molecules of the CA polymer matrix. The tensile strength values of the resultant CA and CS/CA blend hollow fibers however appear to be slightly greater than that of many other hollow fibers and the values of elongation ratio and Young’s modulus are generally comparable with that of other hollow fibers.

Table 4 Mechanical test results of the different types of hollow fibers Hollow fibers

Tensile stress (MPa)

Elongation ratio at break (%)

Young’s modulus (GPa)

Pure CA Blend I Blend II

38.80 26.16 22.10

33.90 27.97 24.36

0.1145 0.0935 0.0900

Table 5 Internal surface areas and amounts of amino end-groups of CA and CS/CA blend hollow fibers Hollow fibers

Internal surface areas (m2 /g)

Amount of amino end-groups (mmol/g)

Pure CA Blend I Blend II

18.6 23.5 28.6

0.001 0.091 0.179

3.6. Adsorption properties To evaluate the adsorptive performance, the three types of hollow fibers, i.e., CA, CS/CA Blend I and CS/CA Blend II, were used as adsorbents to adsorb copper ions and BSA from aqueous solutions. The internal surface areas and the amounts of the amino end-groups of these hollow fibers adsorbents are given in Table 5. It can be seen that the blending of CS with CA significantly increased the internal surface areas (from 18.6 for CA to 23.5 and 28.6 m2 /g dry membranes for CS/CA Blend I and Blend II hollow fibers, respectively) and the densities of the amino end-groups in the hollow fibers (from negligible for CA to 0.091 and 0.179 mmol/g dry membranes for CS/CA Blend I and Blend II hollow fibers, respectively). 3.6.1. Adsorption of copper ions The adsorption amount of copper ions on the three types of hollow fibers at pH 5 and 6 are presented in Table 6. The uptake of copper ions by CA hollow fibers was almost negligible. The CS/CA blend hollow fibers showed greatly enhanced adsorption of copper ions even though the amount of CS in the blend hollow fibers was very small, indicating that CS is indeed effective in chelating copper ions from solutions and thus improving the adsorption performance of CA hollow fibers. The solution pH also influenced the adsorption performance and all the different types of hollow fibers showed an increase in the adsorption uptake for pH change from 5 to 6. This may be an indication of the effect of electrostatic

Table 6 Experimental adsorption amounts of copper ions on CA and CS/CA blend hollow fibers at different initial solution pH values (C0 = 50 mg/L) Hollow fibers

Pure CA Blend I Blend II

Copper ions adsorption amounts (mg) per g of hollow fibers

Calculated copper adsorption amounts (mg) per g of chitosan

pH 5

pH 6

pH 5

pH 6

0.0017 0.375 0.7415

0.0367 0.5476 1.093

– 20.250 20.020

– 29.570 29.511

C. Liu, R. Bai / Journal of Membrane Science 267 (2005) 68–77

75

Fig. 5. Adsorption isotherm of copper ions on CS/CA Blend II hollow fibers. (a) Adsorption capacity (qe ) vs. equilibrium concentration (Ce ); (b) 1/qe vs. 1/Ce (marks represent experimental results and line represents the fitted results from the Langmuir model).

interaction on adsorption since at a higher pH the fibers were more electrically negative. The adsorption isotherms of copper ions on Blend II hollow fibers were investigated at initial pH 6 at room temperature. The experimental results are fitted to the Langmuir isotherm model given below: 1 1 1 = + qe qm bqm Ce

Table 7 Reuse of Blend II hollow fibers for copper ions adsorption Reuse cycle

Adsorption amount (mg/g fiber)

1 2 3

1.093 0.823 0.814

(2)

where qe (mg Cu/g dry fibers) and Ce (mg/L) represent equilibrium adsorption uptake and equilibrium concentration of copper ions in the bulk solution, respectively, and qm (mg Cu/g dry fibers) represents the maximum adsorption amount, b (L/mg) the adsorption equilibrium constant. It is found from Fig. 5 that the Langmuir model can be fitted to the experimental results reasonably well, giving the r2 value of 0.994. The values of qm and b are found to be 4.146 mg copper per g of dry fibers and 5.85 × 10−2 L/mg, respectively. When the amount of adsorption were calculated for chitosan, it is found that the adsorption capacity becomes about 111.9 mg Cu2+ /g CS, i.e., CS is indeed very effective in enhancing the adsorption performance of the hollow fibers. The regeneration and reuse of the blend hollow fibers were also investigated. The Blend II hollow fibers were preadsorbed with copper ions at the condition of initial pH 6 and initial copper concentration of 50 mmol/mol (50 ppm)

and then regenerated by desorption in 50 mL of 0.05 M HCl. The adsorption–desorption process was repeated for three cycles and the adsorption amounts are given in Table 7. It can be seen that except a reduction of 24.7% in the first cycle, the hollow fibers showed almost no further reduction in the adsorption performance in the subsequent cycles, indicating that the blend hollow fibers can be effectively regenerated and reused. 3.6.2. Adsorption of BSA The adsorption of BSA on CA and CA/CS blend hollow fibers were also investigated. The equilibrium adsorption uptakes of BSA with the three types of hollow fibers at pH 6.3 are given in Table 8. Similar to the results in copper ion adsorption, blend hollow fibers showed significantly enhanced adsorption performance for BSA than the CA hollow fibers. The equilibrium adsorption amount, for example, for Blend II hollow fibers is about 3.4 mg/g fibers, in comparison with 0.292 mg/g for CA hollow fibers.

Fig. 6. Adsorption isotherm of BSA on CS/CA Blend II hollow fibers. (a) Adsorption capacity (qe ) vs. equilibrium concentration (Ce ); (b) 1/qe vs. 1/Ce (marks represent experimental results and line represents the fitted results from the Langmuir model).

76

C. Liu, R. Bai / Journal of Membrane Science 267 (2005) 68–77

Table 8 Experimental adsorption amounts of BSA on CA and CS/CA blend hollow fibers (initial solution pH 6.3, C0 = 0.5 g/L) Hollow fibers

BSA adsorption amounts (mg) on per g of hollow fibers

Calculated BSA adsorption amounts (mg) by per g of chitosan

Pure CA Blend I Blend II

0.292 1.117 3.411

– 60.318 92.097

Table 9 Reuse of Blend II hollow fibers for BSA adsorption

References

Reuse cycle

Adsorption amount (mg/g fiber)

1 2 3

3.411 3.120 3.106

[1] 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. [2] D.K. Roper, N.E. Lightfoot, Separation of biomolecules using adsorptive membranes, J. Chromatogr. A 702 (1995) 3– 26. [3] H. Matsuyama, S. Berghmans, D.R. Lloyd, Formation of hydrophilic microporous membranes via thermally induced phase separation, J. Membr. Sci. 142 (1998) 213–224. [4] 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. [5] H. Yoshida, T. Kataoka, Adsorption of BSA on cross-linked chitosan: the equilibrium isotherm, Chem. Eng. J. 41 (1989) B11– B15. [6] X. Zhang, R.B. Bai, Mechanisms and kinetics of humic acid adsorption onto chitosan-coated granules, J. Colloid Interface Sci. 264 (2003) 30–38. [7] E. Guibal, Interactions of metal ions with chitosan-based sorbents, Sep. Purif. Technol. 38 (2004) 43–74. [8] W.L. Yan, R.B. Bai, Adsorption of lead and humic acid on chitosan hydrogel beads, Water Res. 39 (2005) 688–698. [9] B. Krajewska, Diffusion of metal ions through gel chitosan membranes, React. Funct. Polym. 47 (2001) 37–47. [10] Z. Modrrzejewska, W. Eckstein, Chitosan hollow fiber membranes, Biopolymers 73 (2004) 61–68. [11] X. Zeng, E. Ruckenstein, Supported chitosan-dye affinity membranes and their protein adsorption, J. Membr. Sci. 117 (1996) 271– 278. [12] X. Zeng, E. Ruckenstein, Cross-linked macroporous chitosan anionexchange membranes for protein separations, J. Membr. Sci. 148 (1998) 195–205. [13] L. Yang, W.W. Hsiao, P. Chen, Chitosan–cellulose composite membrane for affinity purification of biopolymers and immunoadsorption, J. Membr. Sci. 197 (2002) 185–197. [14] C.X. Liu, R.B. Bai, Recovery of Bovine Serum Albumin (BSA) by Tripolyphosphate (TPP) Crosslinked Chitosan Membrane, presented at Recent Advances in Separation & Purification Techniques for Biological & Pharmaceutical Products Development, NTU, Singapore, 21–23 February 2005. [15] A. Dufresne, J.Y. Cavaill´e, D. Dupeyre, M. Garcia-Ramirez, J. Romero, Morphology, phase continuity and mechanical behaviour of polyamide 6/chitosan blends, Polymer 40 (1999) 1657– 1666. [16] A. Isogai, R.H. Atalla, Preparation of cellulose-chitosan polymer blends, Carbohydr. Polym. 19 (1992) 25–28. [17] M. Hasegawa, A. Isogai, S. Kuga, F. Onabe, Preparation of cellulose–chitosan blend film using chloral/dimethylformamide, Polymer 35 (1994) 983–987. [18] S.Z. Rogovina, T.A. Akopova, G.A. Vikhoreva, I.N. Gorbacheva, Solid state production of cellulose–chitosan blends and their modification with the diglycidyl ether of oligo(ethylene oxide), Polym. Degrad. Stab. 73 (2001) 557–560.

The adsorption isotherm of BSA on the Blend II hollow fibers is illustrated in Fig. 6. Again, the Langmuir isotherm model can be fitted to the experimental results, giving the r2 value of 0.988. The values of qm and b are found to be 13.85 mg/g fibers (corresponding to 373.95 mg/g CS) and 1.69 × 10−3 L/mg, respectively. Again, the regeneration and reuse of the hollow fibers for BSA were examined. BSA adsorbed on the fibers was desorbed in 6 mL of 50 mM KSCN solution. Table 9 shows the adsorption results from three cycles. Similar to that in copper adsorption, there was a reduction in the adsorption amount after the first cycle, but the adsorption performance stabilized in subsequent cycles.

4. Conclusions In this study, we have successfully fabricated novel CS/CA blend hollow fibers through the wet spinning method with formic acid as co-solvent of CS and CA and NaOH as coagulant in the hollow fiber fabrication. CA was chosen to be the polymer matrix providing high mechanical strength and chitosan to be reactive polymer contributing to enhanced adsorption performance for the separation of biomolecules and metal ions. Two typical blend hollow fibers, i.e., Blend I and Blend II, were spun from dopes containing CS and CA at ratios of 1.8% and 3.8%. FTIR and XRD analyses revealed that the two polymers were well mixable. The blend hollow fibers showed porous and macrovoids-free structures and highly interconnected pores, which can provide great surface area for adsorption. The blending of CS into CA, even though at a small CS amount, significantly improved the adsorption performance of the hollow fibers for copper ions and BSA from aqueous solutions. The blend hollow fibers can also be effectively regenerated and reused.

Acknowledgement The financial support of the Academic Research Funds, National University of Singapore, is acknowledged.

C. Liu, R. Bai / Journal of Membrane Science 267 (2005) 68–77 [19] Y.K. Twu, H.I. Huang, S.Y. Chang, S.L. Wang, Preparation and sorption activity of chitosan/cellulose blend beads, Carbohydr. Polym. 54 (2003) 425–430. [20] L. Jin, R.B. Bai, Mechanisms of lead adsorption on chitosan/PVA hydrogel beads, Langmuir 18 (2002) 9765–9770. [21] N. Li, R.B. Bai, Copper adsorption on chitosan–cellulose hydrogel beads: behaviors and mechanisms, Sep. Purif. Technol. 3 (2005) 237–247. [22] W. Wang, S. Bo, S. Li, W. Qin, Determination of the Mark–Houwink equation for chitosans with different degrees of deacetylation, Int. J. Biol. Macromol. 13 (1991) 281–285. [23] X. Jiang, L. Chen, W. Zhong, A new linear potentiometric titration method for the determination of deacetylation degree of chitosan, Carbohydr. Polym. 54 (2003) 457–463. [24] A.Z. Gollan, Anisotropic membranes for gas separation, U.S. Patent 4, 681,605 (1987).

77

[25] 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. [26] S. Puttipipatkhachorn, J. Nunthanid, K. Yamamoto, G.E. Peck, Drug physical state and drug–polymer interaction on drug release from chitosan matrix films, J. Controlled Release 75 (2001) 143– 153. [27] R.E. Kesting, A.K. Fritzsche, Polymeric Gas Separation Membranes, Wiley, New York, 1993. [28] Y. Liu, G.H. Koops, H. Strathmann, Characterization of morphology controlled polyethersulfone hollow fiber membranes by the addition of polyethylene glycol to the dope and bore liquid solution, J. Membr. Sci. 223 (2003) 187–199. [29] J.H. Kim, B.R. Min, J. Won, H.C. Park, Y.S. Kang, Phase behavior and mechanism of membrane formation for polyimide/DMSO/water system, J. Membr. Sci. 187 (2001) 47–55.