Preparation and characterization of polyvinylidene fluoride (PVDF) hollow fiber membranes

Preparation and characterization of polyvinylidene fluoride (PVDF) hollow fiber membranes

Journal of Membrane Science 163 (1999) 211–220 Preparation and characterization of polyvinylidene fluoride (PVDF) hollow fiber membranes Dongliang Wa...

540KB Sizes 35 Downloads 168 Views

Journal of Membrane Science 163 (1999) 211–220

Preparation and characterization of polyvinylidene fluoride (PVDF) hollow fiber membranes Dongliang Wang, K. Li ∗ , W.K. Teo Department of Chemical and Environmental Engineering, National University of Singapore, 10 Kent Ridge Road, Singapore 119260, Singapore Received 28 December 1998; received in revised form 21 April 1999; accepted 30 April 1999

Abstract Polyvinylidene fluoride (PVDF) hollow fiber membranes were prepared by dry/wet and wet phase inversion methods. In spinning these PVDF hollow fibers, dimethylacetamide (DMAc) and polyvinyl pyrrolidone (PVP) were used as a solvent and an additive, respectively. Water was used as the external coagulant. Water or ethanol was used as the internal coagulants. The membranes were characterized in terms of water flux, molecular weight cut-off for the wet membranes. Gas permeation fluxes and effective surface porosity were determined by a gas permeation method for the dried membranes. The cross-sectional structures were examined by scanning electron microscopy. The effects of polymer concentration, air-gap, PVP molecular weight, PVP content in the polymer dope, and the internal coagulant on the permeation properties and membrane structures were examined. Highly permeable PVDF hollow fiber membranes could be prepared from a polymer dope containing low molecular weight PVP and using ethanol as the internal coagulant. ©1999 Elsevier Science B.V. All rights reserved. Keywords: Polyvinylidene fluoride; Hollow fiber; Membrane preparation

1. Introduction Polyvinylidene fluoride (PVDF) is a semi-crystalline polymer containing a crystalline phase and an amorphous and/or rubbery phase. The crystalline phase provides thermal stability and the amorphous phase has flexibility towards membranes. PVDF is stable while it is attacked by most of the corrosive chemicals and organic compounds including acids, alkaline, strong oxidants and halogens [1,2]. In addition, the hydrophobicity of this polymer provides a potential application in the membrane-based gas absorption and oil/water separation [3,4]. ∗ Corresponding author. Tel.: +65-8746388; fax: +65-7791936 E-mail address: [email protected] (K. Li)

Preparation and characterization of PVDF flat-sheet ultrafiltration and microfiltration membranes by dry/wet phase inversion processes were studied extensively by Uragami et al. [5–7] and Bottino et al. [8–11]. The membranes were prepared from binary solutions containing the PVDF polymer and a high boiling point solvent such as N-methyl-2-pyrrolidone (NMP), dimethylacetamide (DMAc), dimethylformamide (DMF) and dimethylsulfoxide (DMSO), and from ternary solutions containing an additional low boiling point co-solvent such as acetone, methyl ethyl ketone and tetrahydrofuran (THF). Composition effect of the coagulation bath on the membrane structure was also investigated [10]. The coagulation media studied include water, alcohols, a mixture of water and various alcohols, and a mixture of water

0376-7388/99/$ – see front matter ©1999 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 6 - 7 3 8 8 ( 9 9 ) 0 0 1 8 1 - 7

212

D. Wang et al. / Journal of Membrane Science 163 (1999) 211–220

Table 1 Composition and viscosity of spinning solutions Solution No.

PVDF (wt.%)

DMAc (wt.%)

PVP (wt.%)

η (cp) 25o C

I II III IV V VI

20 22 20 20 25 25

73 70 73 73 73 70

7 8 7 7 2 5

1620 2500 12000 1130 2000 3190

(24K) (24K) (360K) (10K) (10K) (10K)

and solvents [10]. In general, large voids were formed when using water as the coagulant, while the voids were reduced when using alcohol and the mixtures of alcohol and water or the mixture of water and solvent. Solvent effect on the membrane morphology was investigated by Bottino et al. [11] who employed eight different solvents including hexamethylphosphoramide (HMPA), tetramethylurea (TMU), NMP, DMAc, DMF, triethylphosphate (TEP) and DMSO. No relationship between the membrane structure and solvent strength was observed. The PVDF membranes prepared were not of practical interest because of their low permeate fluxes. A study on improving the PVDF membrane permeability was conducted by Uragami et al. [12] who reported that the PVDF membranes with improved fluxes could be prepared by adding polyethylene glycol (PEG) as a pore forming agent in the solution containing NMP and THF. However, the addition of PEG reduces the strength of the membranes. Bottino et al. [13] introduced an inorganic salt, LiCl, in their polymer dopes and observed that the addition of LiCl greatly increases the viscosity of the PVDF polymer solution. The membranes cast from the LiCl containing polymer dope possesses improved membrane porosity and water permeation flux and solute rejection. All the studies conducted above were focused on the flat-sheet PVDF membranes prepared on laboratory-scale. Preparation of porous PVDF hollow fiber membranes have mostly been reported in the Patents [14,15] and detailed studies regarding the preparation and characterization of porous PVDF hollow fibers prepared by phase inversion methods are still lacking. This study aims at preparing various porous PVDF hollow fiber membranes with different structures and

permeation properties. The effects of spinning conditions such as polymer concentration, polyvinyl pyrrolidone (PVP) molecular weight, PVP content, air-gap and internal coagulants on the membrane morphology and the permeation performance are the objectives of this study and are discussed in detail below.

2. Experimental 2.1. Preparation of hollow fiber The PVDF pellet, Kynar® -720, was purchased from Elf Autochem, USA. PVP and 99+% DMAc were used as additive and solvent, respectively. Three types of PVP with molecular weights of 10,000, 24,000 and 360,000 were purchased from Sigma Chemical and Fluka Chemie. The polymers were used as received. The water content of the polymers was determined by heating in a vacuum oven for 20 h at a temperature of 80–100◦ C. The water contents of PVDF and PVP were 0.1 and 8–10 wt.%, respectively. PVDF, DMAc, and PVP were mixed at 50–60◦ C under agitation for 2 or 3 days. Spinning solutions with different polymer concentrations (20–25 wt.%) and different PVP contents were formulated. All the polymer solutions prepared were clear and homogeneous at room temperature. Their compositions and viscosity at 25◦ C are listed in Table 1. Viscosity was measured using rotary viscometer (Rheology International). Tap water at a temperature of 25 ± 1◦ C was used as the external coagulant in preparing these hollow fibers. Water or ethanol (99.8 vol%) were used as an internal coagulant. The spinneret with an orifice diameter/inner diameter of the tube of 0.9/0.21 mm was used. In spinning these hollow fibers, the take-up velocity (6–8 m min−1 ) was nearly the same as the free falling velocity of the nascent hollow fiber. The air-gap was in the range of 0 to 20 cm. The details of the spinning equipment and procedure on hollow fiber spinning have been described elsewhere [16]. 2.2. Water permeation flux and molecular weight cut-off The fibers were kept in fresh water for at least 1 week. Testing modules with both ends open were

D. Wang et al. / Journal of Membrane Science 163 (1999) 211–220

made. The module contained about four to ten wet fibers with the length of 10–20 cm. Water permeation flux and solute rejection were determined using a membrane testing unit (Nitto Denko Corporation, Japan). An aqueous solution containing about 1000–1800 ppm dextran® (Sigma) with a molecular weight of 110,000 or 500,000 was used as a testing liquid. The water solution was fed into the shell side of the module. The operating pressure (1 bar, gauge) and solution re-circulation flow velocity (about 2.5 m min−1 ) through the module were controlled by the adjusting valve. The water permeation flux in the permeate side was measured at the temperature of 25 ± 1◦ C. The water permeation flux was determined based on the hollow fibers outer diameter (OD). The concentration of dextran on the permeate side and the feed side was determined using TOC analyzer (TOC 5000A, Shimadzu). 2.3. Pore size and surface porosity For a porous asymmetric membrane, the determination of pore size, particularly surface porosity over effective pore length, is very important in studying mass transfer in the membrane absorption and membrane extraction processes. However, the gas permeation method suggested by Yasuda et al. [17] was useful in determining the volume porosity of porous membranes. In our recent study [18], a modified gas permeation method was introduced to determine the mean pore size and the effective surface porosity over the effective pore length of the asymmetric membrane. For a considerable porous asymmetric membrane, the solution–diffusion contribution is negligible, and therefore, the total gas permeation rate through the asymmetric membrane is the combination of Poiseuille flow and Knudsen flow [19]: if we assume that the pores in the skin layer are approximated by cylindrical pores with the modal pore radius, r, and the effective length, Lp , Eq. (1) can be obtained [18]: Ji =

2 3



8RT πM

0.5

p¯ r 2 ε 1 rε + RT Lp 8µi RT Lp

or Ji = K0 + P0 p¯

(1)

213

where r is mean pore radius of the membrane (m); ε is surface porosity; Lp is effective pore length (m); η is the viscosity of gas (Pa s); R is the gas constant (8.3144 m3 Pa mol−1 K−1 ); p¯ is the mean pressure (Pa); M is molecular weight of the gas and T is absolute temperature (K). As the skin layer thickness of asymmetric membranes is usually an unknown value and cannot be determined using available methods, the gas permeation flux, Ji , (mol m−2 Pa−1 s−1 ) is determined by the following equation: Ji =

Nt,i At 1p

(2)

Where Nt,i is total gas permeation rate (mol s−1 ), At is total permeation area (m2 ) and 1p is the pressure difference across the membrane (Pa); and subscript i refers to the gas species concerned. Gas permeation flux, Ji , through the asymmetric hollow fiber membrane at different pressures was first measured. By plotting Ji with mean pressures according to Eq. (1), the mean pore size can be calculated from the intercept (K0 ) and slope (P0 ): r=

16 3



P0 K0



8RT πM

0.5 µ

(3)

The effective surface porosity over pore length, ε/Lp can also obtained from the slope as follows: 8µRT P0 ε = Lp r2

(4)

After drying hollow fibers under ambient condition (25 ± 1◦ C and RH = 60–65%), a test module containing one or two fibers with the length of about 5 cm was made. The module was then connected to the gas permeation apparatus described elsewhere [19]. Nitrogen was used as the test gas. The test apparatus used was based on the volume displacement method. The upstream pressure was in the range from 5 to 30 psi (gauge), which was measured by the pressure transducer (Basingstoke, England). The N2 permeation rate was measured at 25 ± 1◦ C in atmosphere using soap-bubble flow meter. The gas permeability was then calculated based on Eq. (2) according to outer diameter of the hollow fiber.

214

D. Wang et al. / Journal of Membrane Science 163 (1999) 211–220

Table 2 Permeation properties and pore structure parameters of the PVDF hollow fibers prepared from different polymer concentrations and at different air-gapsa Concentration (wt.%)

Air-gap (cm)

Fw (Lm−2 h−1 bar−1 )

R (%)

(P/L)N2 (cm3 cm−2 s−1 cm−1 Hg)

r (m)

ε/Lp , (m−1 )

22

15 10 5

81.9 89.8 110

64 57 61

6.56 × 10−3 7.4 × 10−3 9.0 × 10−3

3.6 × 10−8 6.53 × 10−8 3.26 × 10−8

3.98 × 102 4.81 × 102 7.48 × 102

20

10 5

149 152

12.3 16.7

2.56 × 10−2 2.69 × 10−2

3.52 × 10−8 3.47 × 10−8

1.93 × 103 1.99 × 103

a

Fw : water solution flux; R: rejection; (P/L): gas permeability obtained at the pressure difference of 1 bar and 25 ± 1◦ C.

2.4. Membrane morphology The membrane morphology was examined using a scanning electron microscope. A piece of the wet fiber was immersed in liquid nitrogen. The membrane was carefully fractured. The specimen were put on a metal support and dried under vacuum for at least 24 h. Then the specimen was coated by sputtering gold. The SEM graphs of the cross-section of the fibers were taken by scanning electronic microscopy (Hitachi S-2150).

3. Results and discussion 3.1. Effect of polymer concentration and air-gap The PVDF hollow fiber membranes were first prepared from the 20 and 22 wt.% polymer dopes (I and II) containing 24K PVP. The membranes were formed at different air-gaps using water as the internal and the external coagulant. The prepared hollow fibers exhibited an outer diameter of 700–800 ␮m and an inner diameter of 400–500 ␮m. Water permeation flux and solute rejection of the wet hollow fibers were measured using an aqueous solution containing 1800 ppm dextran (MW = 500,000). The results are shown in Table 2. It can be seen that the water permeation flux increases and that the solute rejection decreases as the polymer concentration is reduced. The water permeation flux tends to increase with decreasing air-gap for the PVDF concentration of 22 wt.%. Based on the above experimental results, it may be said in conclusion that the concentrated polymer solution decreases the pore volume in the skin layer and short air-gap tends to form a thinner skin layer, which are commonly observed when preparing asymmetric membranes by the dry/wet immersion method. The membranes prepared

exhibit large pores as indicated by their low rejection for the solute of dextran with a molecular weight of 500,000. The measured Nitrogen permeability and the resulting pore size, and the effective porosity (ε/Lp ) determined by the gas permeation method are also shown in Table 2. As can be seen, the membranes with high water permeation flux show high gas permeability and large ε/Lp . Fig. 1(a) and Fig. 1(b) show the cross-sectional structures of the hollow fibers prepared from both the polymer dopes (I and II) at the air-gap of 10 cm. The cross-sectional structures are similar for both the membranes. It can be seen that near the inner wall of the hollow fiber, big cavities are formed, while near the outer wall, smaller macrovoids are formed. Beneath the smaller macrovoids, sponge-like structures are present. The formation of the fiber structures shown can be attributed to mechanisms of the coagulation process. In the dry/wet spinning process, water is initially in contact with the inner polymer solution. Phase separation occurs quickly on the inner edge and the inner skin is formed. The solvent concentration in the internal coagulant increases with time due to the outflow of solvent from the solution. Under the skin layer, a small amount of water penetrates into the solution and liquid–liquid phase separation occurs. Water diffusion along the polymer-lean phase results in the growth of polymer-lean phase due to strong interaction between water and PVDF. On the other hand, the addition of hydrophilic PVP improves water diffusion. The long time gap between the initial phase separation to the final solidification along the fiber wall develops the micro-phase separation well. So those large cavities were formed under the inner skin. For the outer layer, when the nascent fiber is immersed in the water bath, a large amount of water makes the outer layer

D. Wang et al. / Journal of Membrane Science 163 (1999) 211–220

215

Fig. 1. SEM graphs of cross-sectional structures of PVDF hollow fibers prepared from different polymer concentrations and at an air-gap of 10 cm. (a) 22 wt.%; (b) 20 wt.%. Table 3 Permeation properties and pore structure parameters of the PVDF hollow fibers prepared with different molecular weight PVPa PVP (MW) 10000 360000 aF

w:

Air-gap (cm)

Fw (L m−2 h−1 bar−1 )

R (%)

(P/L)N2 (cm3 cm−2 s−1 cm−1 Hg)

R (m)

ε/Lp (m−1 )

15 10 5

101 179 6.7

50.3 39.5 32.2

5.62 × 1−2 7.80 × 10−2 3.84 × 10−7

9.73 × 10−8 6.43 × 10−3 3.47 × 10−7

1.51 × 103 3.18 × 103 6.2 × 102

water solution flux; R: rejection; (P/L): gas permeability obtained at the pressure difference of 1 bar and 25 ± 1◦ C.

undergo solidification in a shorter time-period. Limited polymer-lean phase growth results in the membrane with small macrovoids near the outer edge.

3.2. Effect of PVP molecular weight The PVDF hollow fiber membranes were prepared from the 20 wt.% polymer dopes containing 7 wt.% PVP with the other two different molecular weights of 10,000 and 360,000. The membranes were characterized by ultrafiltration, gas permeation and scanning electron microscopy. Table 3 gives the water solution flux and the solute rejection determined for the aqueous solution with the concentration of 1800 ppm dextran (MW = 500,000). Nitrogen permeability, pore size and effective porosity (ε/Lp ) are also shown in Table 3. A comparison of the experimental results in Tables 2 and 3 reveals that high water solution permeation flux and good solute rejection could be obtained for the membrane prepared using low molecular weight PVP (10,000). The hollow fibers prepared from high molecular weight PVP (360,000) exhibited very low permeation flux. The N2 permeation flux of

the membrane prepared from PVP (360,000) is about five order of magnitudes lower than that of the membranes prepared from low molecular weight PVP. Low molecular weight PVP tends to form small pores and easily leaches out from the membrane, while most of the high molecular weight PVP stays in the membrane and may block the void interconnection path. Low rejection and large pore size are caused by the fact that large molecular weight PVP tends to form thicker skin layer containing bigger pores. The cross-sectional structures of the hollow fiber membranes prepared from both kinds of PVP (MW = 10,000 and 360,000) were examined. Fig. 2(a) and Fig. 2(b) depict the structures of the membranes prepared from these two polymer dopes. A comparison of Fig. 2(a) and Fig. 1(a) reveals that the cross-sectional structures of the hollow fibers prepared from PVP (MW = 10,000) and PVP (MW = 24,000) exhibit similar pore morphology. Still, the two layers were observed on the fiber wall and the inner layer has big cavities. The inner layer cavities on the membrane prepared from PVP (MW = 360,000) were reduced and no clear boundary between two layers exists on the membrane wall.

216

D. Wang et al. / Journal of Membrane Science 163 (1999) 211–220

Fig. 2. SEM graphs of cross-sectional structures of PVDF hollow fibers prepared from 20 wt.% polymer dope containing different molecular weight PVP. (a) MW = 10,000; (b) MW = 360,000. Table 4 Permeation properties and pore structure parameters of the PVDF hollow fibers prepared with different PVP (MW = 10,000) contentsa PVP (%)

Air-gap (cm)

Fw (L m−2 h−1 bar−1 )

R (%)

(P/L)N2 (cm3 cm−2 s−1 cm−1 Hg)

r (m)

ε/Lp (m−1 )

2

19 10 10

8.8 11.9 30.2

94 93 90.5

1.96 × 10−2 8.44 × 10−3 2.61 × 10−2

4.03 × 10−8 9.12 × 10−8 6.94 × 10−8

1.29 × 102 1.72 × 102 8.44 × 102

5 a

Fw : water solution flux; R: rejection; (P/L): gas permeability obtained at the pressure difference of 1 bar and 25 ± 1◦ C.

3.3. Effect of PVP concentration It was shown that membranes prepared from low polymer concentration exhibit weak mechanical strength and big pores. The experimental results also indicated that the hollow fibers prepared using low molecular weight PVP (10K) show good performance. Subsequently, the effect of PVP concentration on the membrane performance was studied for the PVDF hollow fibers prepared from 25 wt.% polymer dope with the PVP concentration of 2 and 5 wt.%, respectively. The water solution fluxes for the 1300 ppm (Dextran-110,000) aqueous solution were measured and the solute rejection was calculated. The results are shown in Table 4. As can be seen, the decrease in PVP concentration results in the membranes with low water solution flux and high rejection. It is interesting to note that gas permeability and ε/Lp values are still very high. The membranes prepared from high polymer concentration with low PVP content exhibited good mechanical strength. When dried in the air, the membranes keep good pore structures and

interconnection. The use of a low concentration of smaller molecular weight of PVP is preferable for making porous asymmetric membranes as it is easier to wash it out from the membrane compared to high molecular weight PVP. The cross-sectional structures of the membranes prepared from both the polymer dopes (V and VI) with the air-gap of 10 cm are shown in Fig. 3(a) and Fig. 3(b). The structures of the inner layer and the outer layer are similar. Under both the inner and the outer skin layers, many finger-like cavities exist. The increase in PVP content from 2% to 5% has little influence on the membrane cross-sectional structures. 3.4. Effect of internal coagulant In some applications, the internal skin is not required and the existence of the skin may provide an additional resistance for permeation. For example, in membrane based gas-absorption process, inner skinless structures with considerable porous substructures can certainly reduce the gas transport resistance when

D. Wang et al. / Journal of Membrane Science 163 (1999) 211–220

217

Fig. 3. SEM graphs of cross-sectional structures of PVDF hollow fibers prepared from 25 wt.% polymer dopes containing different PVP (10,000) contents and at air-gap of 10 cm. (a) 2 wt.%; (b) 5 wt.%.

Table 5 Permeation properties and pore structure parameters of the PVDF hollow fibers prepared with different internal coagulantsa Internal coagulant

Dope

Fw (L m−2 h−1 bar−1 )

R (%)

(P/L)N2 (cm3 cm−2 s−1 cm−1 Hg)

r (m)

ε/Lp (m−1 )

Ethanol Water Ethanol Water Ethanol Water

II II V V VI VI

156 89 39.7 24.4 155 85.1

36.4 69 83.4 86.5 31 61

2.8 × 10−2 1.45 × 10−2 2.52 × 10−1 6.83 × 10−3 2.81 × 10−1 3.80 × 10−2

1.17 × 10−7 4.66 × 10−8 3.88 × 10−8 3.56 × 10−8 4.40 × 10−8 6.33 × 10−8

6.54 × 103 4.5 × 102 1.52 × 104 4.33 × 102 1.47 × 104 1.35 × 103

a

Fw : water solution flux; R: rejection; (P/L): gas permeability obtained at the pressure difference of 1 bar and 25 ± 1◦ C.

gas is fed into the lumen of the hollow fiber. Attempts were made in this study to remove the internal skin so as to improve the membrane permeation flux. For the hollow fiber spinning, one of the possible methods is to use a very weak internal coagulant. In this case, the coagulation process takes place dominantly through the outer surface, e.g. the production of a flat-sheet membrane. The previous study in our laboratory [20] revealed that the outer skin layer could not be formed when using an aqueous solution containing more than 50% ethanol as the external coagulant. However, the gas permeation flux decreases with increasing ethanol content to 50% and circular hollow fiber could not be spun when using high concentration of ethanol in the external coagulant [20]. For PVDF, the alcohols are very weak non-solvents. This means that the coagulation rate between the polymer solution and the alcohol is very slow. The use of ethanol as the internal coagulant was investigated in this study. It should

be noted that the wet spinning process has to be used for making fine hollow fiber due to the slow internal coagulation. On the other hand, the internal coagulant flux should be high enough to allow the hollow fiber to have circular lumen. The hollow fiber membranes were prepared from the three polymer dopes (II, V and VI) using ethanol or water as the internal coagulant and at the air-gap of 0. Water solution flux and solute rejection for the membranes prepared using the polymer dope II were determined using 500,000 dextran® in the concentration of about 1800 ppm. The other membranes were examined using 110,000 dextran® in the concentration of about 1200 ppm. The results are shown in Table 5. As can be seen, the water solution flux of the membrane prepared using ethanol as the internal coagulant is much higher than that of the membrane prepared using water as the internal coagulant. The water flux for the former is nearly twice that for the latter. It is clearly indicated that the inner

218

D. Wang et al. / Journal of Membrane Science 163 (1999) 211–220

Fig. 4. SEM graphs of cross-sectional structures of PVDF hollow fibers spun from a polymer dope consisting of 22 wt.% PVDF, 8 wt.% PVP (MW = 24,000) and 70 wt.% (DMAc) using different internal coagulants: (a) water; (b) ethanol.

Fig. 5. SEM graphs of cross-sectional structures of PVDF hollow fibers spun from a polymer dope consisting of 25 wt.% PVDF, 2 wt.% PVP (MW = 10,000) and 73 wt.% (DMAc) using different internal coagulants: (a) water; (b) ethanol; (c) inner edge.

D. Wang et al. / Journal of Membrane Science 163 (1999) 211–220

219

Fig. 6. SEM graphs of cross-sectional structures of PVDF hollow fibers spun from a polymer dope consisting of 25 wt.% PVDF, 5 wt.% PVP (MW = 10,000) and 70 wt.% (DMAc) using different internal coagulants: (a) water; (b) ethanol; (c) inner edge.

skin gives similar resistance to water permeation. The rejection is lower, particularly for the membranes prepared from the dope containing high content PVP. As expected, large ε/Lp and high gas permeation flux were obtained for the hollow fibers prepared using ethanol as the internal coagulant. The ε/Lp of the membranes prepared from ethanol as the internal coagulant is approximately one order of magnitude larger than that of the membranes prepared using water as the internal coagulant. In the membrane absorption process, large ε/Lp is beneficial for reducing the membrane resistance so as to increase the mass-transfer coefficient. The results also show that a high content of PVP can greatly enhance the water permeation flux for the wet membranes, but that it has little effect on gas permeability. The residence of PVP in the membrane may

block the pore during the drying process. The results show that higher PVP content in the spinning dope is not favorable when making PVDF porous hollow fibers for membrane absorption and membrane extraction. Examination of the membrane structures indicates that high permeation flux when using ethanol as the internal coagulant was mainly caused by the fact that the internal skin was eliminated by the use of ethanol as the internal coagulant as demonstrated in Fig. 4(b), Fig. 5(b) and Fig. 6(b). When water was used as the internal coagulant, internal skin and external skin were generated for the membranes prepared from the three polymer dopes (Fig. 4(a), Fig. 5(a) and Fig. 6(a)). Under the skin layer, there are many finger-like structures. Inner skin layer and big macrovoids were not

220

D. Wang et al. / Journal of Membrane Science 163 (1999) 211–220

present if ethanol was used as an internal coagulant. Upon further examination at high magnification (Fig. 5(c) and Fig. 6(c)), loose, porous interconnection network structures were observed.

4. Conclusions PVDF hollow fiber membranes with different pore size and porosity as well as morphology were prepared from the solution containing DMAc and PVP. High permeation flux and good solute rejection could be obtained for the membrane prepared using low molecular weight PVP (10,000). The use of PVP with very large molecular weight tends to form the membrane having very low permeability. The increase in polymer concentration and the decrease in PVP content in the polymer dope greatly decrease the water solution flux, but they had little influence on gas permeability and the ε/Lp value. The use of alcohol as an internal coagulant eliminated the inner skin layer and formed the loose porous network inner layer of the hollow fiber. The resultant membranes exhibited greatly enhanced permeation flux. The proper conditions for making high performance PVDF hollow fibers include the use of low molecular weight PVP (10,000), suitable PVP content (2–5 wt.%) and PVDF (20–25 wt.%) concentrations, and the use of ethanol as an internal coagulant.

Acknowledgements The authors gratefully acknowledge the research funding (GR6456) provided by the Wheelabrator Water Technology (S) Ltd., Singapore.

References [1] J.E. Dohany, L.E. Robb, Polyvinylidene fluoride, in: Kirk-Othmer Encyclopedia of Chemical Technology, vol. 11, 3rd ed., Wiley, New York, NY, 1980, pp. 64–74. [2] A.J. Lovinger, poly(vinylidene fluoride), in: D.C. Bassett (Ed.), Development in Crystalline Polymers, vol. 1, Applied Science, London, 1982, p. 195. [3] Q. Zhang, E.L. Cussler, Microporous hollow fibres for gas absorption I: Mass transfer in the liquid, J. Membr. Sci. 23 (1985) 321.

[4] N.P. Tirmizi, B. Raghuraman, J. Wiencek, Demulsification of water/oil/solid emulsions by hollow-fiber membranes, AIChE J. 42 (1996) 1263. [5] M. Sugihara, M. Fujimoto, T. Uragami, Effect of casting solvent on permeation characteristics of polyvinylidene fluoride membranes, Polym. Prepr. Am. Chem. Soc. Div. Polym. Chem. 20 (1979) 999. [6] T. Uragami, M. Fujimoto, M. Sugihara, Studies on syntheses and permeabilities of special polymer membranes 24. Permeation characteristics of poly(vinylene fluoride) membranes, Polymer 21 (1980) 21. [7] T. Uragami, M. Fujimoto, N. Sugihara, Studies on syntheses and permeabilities of special polymer membranes 27. Concentration of poly(styrene sulphonic acid) in various aqueous solutions using poly(vinylene fluoride) membranes, Polymer 22 (1981) 240. [8] A. Bottino, G. Capannelli, S. Munari, Permeation characteristics of poly(vinylene fluoride) membranes, Polymer 21 (1980) 21. [9] A. Bottino, G. Capannelli, S. Munari, Factors affecting the structure and properties of asymmetric polymeric membranes, in: E. Drioli, M. Nakagaki (Eds.), Membrane and Membrane Processes, Plenum Press, New York, 1986, p. 163. [10] A. Bottino, G. Capannelli, S. Munari, Effect of coagulation medium on properties of sulphonated polyvinylene fluoride membranes, J. Appl. Polym. Sci. 30 (1985) 3009. [11] A. Bottino, G. Camera-Roda, G. Capannelli, S. Munari, The formation of microporous polyvinylidene difluoride membranes by phase separation, J. Membr. Sci. 57 (1991) 1. [12] T. Uragiami, Y. Naito, M. Sugihara, Studies on syntheses and permeability of special polymer membranes 39. Permeation characteristics and structure of polymer blend membranes from polyvinylidene fluoride and polyethylene glycol, Polym. Bull. 4 (1981) 617. [13] A. Bottino, G. Capannelli, S. Munari, A. Turturro, High performance ultrafiltration membranes cast from LiCl doped solutions, Desalination 68 (1988) 167. [14] T. Nohmi, T. Yamada, Polyvinylidene fluoride type resin hollow filament microfilter and process for producing the same, US Patent 4,399,035, August 1983. [15] S. Masaharu, Polyvinylidene fluoride hollow fiber membranes and their manufacture for ultrafiltration, JP 08299770A2, November 1996. [16] D.L. Wang, K. Li, W.K. Teo, Polyethersulfone hollow fiber gas separation membranes prepared from NMP/alcohol solvent systems, J. Membr. Sci. 115 (1996) 85. [17] H. Yasuda, J.T. Tsai, Pore size of microporous polymer membranes, J. Appl. Polym. Sci. 18 (1974) 805. [18] K. Li, J.F. Kong, D. Wang, W.K. Teo, Tailor made asymmetric hollow fibers for soluble gas removal, AIChE J., in press. [19] D.L. Wang, K. Li, W.K. Teo, Effects of temperature and pressure on gas permselection properties in asymmetric membranes, J. Membr. Sci. 105 (1995) 89. [20] S.P. Deshmukh, K. Li, Effect of ethanol composition in water coagulation bath on morphology of PVDF hollow fiber membranes, J. Membr. Sci. 150 (1998) 75.