Separation and Purification Technology 109 (2013) 98–106

Contents lists available at SciVerse ScienceDi rect

Sepa ration and Purification Techn ology

journal homepage: www.elsevier .com/locate /seppur

Characterization of PVDF hollow fiber membrane for CO 2 stripping by atomic force microscopy analysis

R. Naim a,b,c, K.C. Khulbe d, A.F. Ismail a,b,⇑, T. Matsuura d

a Advanced Membrane Technology Research Centre (AMTEC), Universiti Teknologi Malaysia, 81310 Skudai, Johor, Malaysia b Department of Gas Engineering, Faculty of Petroleum and Renewable Energy Engineering, Universiti Teknologi Malaysia, 81310 Skudai, Johor, Malaysia c Faculty of Chemical Engineering and Natural Resources, Universiti Malaysia Pahang, Lebuhraya Tun Razak, 26300 Kuantan, Pahang, Malaysia d Industrial Membrane Research Laboratory, Department of Chemical and Biological Engineering, University of Ottawa, 161 Louis Pasteur St., Ottawa, ON, Canada K1N 6N5

a r t i c l e i n f o a b s t r a c t

Article history: Received 7 December 2012 Received in revised form 28 February 2013 Accepted 28 February 2013 Available online 14 March 2013

Keywords:PVDF hollow fiber membrane Atomic force microscope Membrane contactor

1383-5866/$ - see front matter � 2013 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.seppur.2013.02.036

⇑ Corresponding author at: Advanced Membrane (AMTEC), Universiti Teknologi Malaysia, 81310 Skuda5535592; fax: +60 7 5535925.

E-mail addresses: afauzi@utm.my, fauzi.ismail@gm

Microporou s polyvinyliden e fluoride (PVDF) membranes with various concentrations of lithium chloride additives were prepared for CO 2 stripping test. Physical membrane characterizations such as atomic force microscopy analysis, gas permeation, contact angle measurement and liquid entry pressure were also conducted. Correlations among the membrane properties, i.e. contact angle, gas permeation, mean pore size, nodule aggrega tes and surface roughness obtained from experimen tal analysis were discussed. The overall trend showed that increasing lithium chloride concentration has decreased the mean pore size, mean nodule aggregates and mean surface roughness of the membrane. On the contrary, the membrane liquid entry pressure has been significantly improved . It was found that the mean pore size determined by tapping mode atomic force mic roscopy (TM-AFM) is 2.3 to 2.7 times larger than that of obtained from gas permeation test. The decreases in nodule size, mean pore size and low surface roughness have con- tributed to the enhancement of CO 2 stripping performance in membrane contactor system. Increasing LiCl concentration has increased the CO 2 stripping flux and membrane mass transfer coefficient. How- ever, the concentration of LiCl showed minimal effect on the liquid side mass transfer coefficient.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Membrane contactor has been extensive ly applied for commer- cial purposes such as beverage processes [1], extraction process [2],membrane distillation [3] and gas separation [4]. The key to success in such systems lies in the reliability of the membrane layer which acts as a separator for gas–liquid, liquid–liquid or liquid–solid phase in their respective applicati ons. The main criterion to have such membran es for the diverse applications is the ability of the membrane to remain non-wetted for prolonged hours when con- tacted with liquid absorbent, regardless of its temperature and con- centration. It was reported that some of the commercially available membranes have suffered from severe wetting problems due to the increase in the membran e porosity [5] and membran e degradation due to the reaction between the membrane and the absorbent li- quid itself [6]. Therefore, membran e material is preferably to have hydrophobic attributes as the surface contact angle would be much higher than that of hydrophilic membrane. In addition, absorbent liquid with high surface tension is also favoured in order to avoid

ll rights reserved.

Technology Research Centre i, Johor, Malaysia. Tel.: +60 7

ail.com (A.F. Ismail).

the intrusion of liquid into the membrane pores. Surprisingly , sev- eral studies have shown that hydrophilic polymer such as polyethe -rimide (PEI) polymers [7] that have been diversely used in ultrafiltration [8,9] and gas separation [10,11] can also be poten- tially applied in membrane contactor due to its high wetting pres- sure and high absorption flux performance [12].

Similar to absorption process, striping process is a reverse pro- cess where absorbed gases in the absorbent are stripped by means of temperature and pressure difference. To date, many studies have been focusing on the absorption/stri pping flux performanc e in membran e contactor but unfortunatel y not many of them high- lighted on the correlations of membrane performance with the membran e surface characteristics such as surface roughness, pore size and pore size distribut ion. The detail information about these paramete rs can be easily obtained by performing atomic force microscop y (AFM) analysis. AFM analysis has been extensive ly used in characterizi ng the properties of membranes for ultrafiltration[13], microfiltration [14], reverse osmosis [15] and gas separation [16]. It is a reliable tool in giving useful information regarding sur- face morphology, pore size, pore size distribution , surface rough- ness and pore density of a material [17]. It was introduced by Binning et al. [18] and consequentl y applied in polymer membrane surface by Albrecht and Quate [19] in ultrafiltration and reverse osmosis process. It serves as an effective means of characteri zing

Nomenc lature

LiCl lithium chloride AFM atomic force microscop yKo interceptPo slopeJG gas permeance (mol m�2 s�1 Pa�1)rp pore radius (lm)Lp effective pore length (m)e surface porosity (%)R gas constant (J mol �1 K�1)l gas viscosity (kg m�1 s�1)M gas molecula r weight (kg mol �1)T gas temperat ure (K)

P mean pressure (Pa)e/Lp effective surface porosity (m�1)v median ranks (50%)i number of measure d pore size by AFM n total number of the measure d pores by AFM dp pore size (nm)rp geometric standard deviation (nm)lp mean pore size of the membrane (nm)JCO2

CO2 stripping flux (mol m�2 s�1)Ql liquid flow rate (m3 s�1)Ai inner surface of the hollow fiber membr anes (m2)

Table 1Spinning condition param eters.

R. Naim et al. / Separation and Purification Technology 109 (2013) 98–106 99

membrane surface in most of membrane areas nowadays. Khulbe et al. [8] used AFM analysis to study the effect of air gap on the mor- phology of the surfaces of polyetherim ide hollow fibers membran efor ultrafiltration. They deduced a correlation between the rough- ness paramete r and pure water flux and concluded that with in- crease in air-gap, the outer surface roughness decreased while increasing the pure water flux. A similar study was also carried out using AFM analysis to characterize the inner and outer surface of polyethe rsulfone–polyimide hollow fiber membrane for gas sep- aration applicati on [20]. In addition, Khayet et al. [21] also per- formed AFM analysis to study the polyviny lidene fluoride (PVDF)membrane surface propertie s (e.g. mean pore size, pore size distri- bution, nodule size, pore density, surface porosity and roughness parameters) for membrane distillation applicati on. A study on PVDF hollow fiber membrane for ultrafiltration reported that addition of ethylene glycol as non-solvent additive has increased the pore size of the membrane as determined by AFM, gas permeation test and solute transport experime nt [22]. It was also noted that the pore size of the inner surface was larger than that of on the outer surface. Recently, Barbosa and Silva [23] studied on the surface propertie s of synthetic polymeric porous membran e (nitrocellulose, nylon, PTFE, and PVDF) by using AFM analysis (intermittent contact mode). They investigated the surface changes of the membran es before and after filtration test using protein solution. It was reported that after fil-tration test the roughness values (Ra and Rq) decreased for all membranes which can be associated to the surface deposition or incrustation with no significant statistical variation among the membranes . However , no direct comparison was made with the va- lue describe in the literature since the roughness parameter is di- rectly dependent on scanning scale and condition s.

To the best of our knowledge, AFM analysis has not yet been re- ported in the open literature for membrane contactor, or more spe- cifically for CO 2 stripping process. Despite all the above works, little is known about the effect of surface properties such as rough- ness, nodule aggregates and pore size on the CO 2 stripping perfor- mance. Therefore, the goal of this report is to elucidate the relationship between the surface properties and CO 2 stripping per- formance in membrane contactor system. For this purpose, the sur- face properties are evaluated based on varying the concentratio n of non-solvent additives; LiCl in the PVDF solution. The liquid entry pressure, gas permeation , mean pore size and effective surface porosity were also determined.

Bore flow rate (ml/min) 2.0 Dope extrusion rate (ml/min) 4.0 Bore fluid composition (wt.%) DMAc/H 2O: 80/20 Coagulation medium Tap water Spinneret OD/ID (mm/mm) 1.0/0.5 Air gap distance (cm) 0Spinning dope temperature (�C) 25 External coagulation temperature (�C) 25

2. Experimental

2.1. Materials

PVDF polymer (Kynar 740) was purchased from Arkema Inc. Philadelphia , USA in the form of pellets. Dimethyl acetamide

(DMAc, >99.5%) was used as the solvent without further purifica-tion. Lithium chloride (P99%) (Sigma–Aldrich) was used as the non-solv ent additive in the polymer solution. Methanol (GR grade, 99.9%) and n-hexane (99%) were used for post-treatmen t of the prepared membranes. Tap water was used as the coagulation bath medium in the hollow fiber spinning process. Diethanolam ine (DEA) (>99%) from Merck was dissolved in water to prepare aque- ous solutions as the liquid absorbent in the membrane contactor test. Pure CO 2 and N2 gas were employed as loaded gas and sweep gas, respectively.

2.2. Fabricatio n of hollow fiber membranes

The polymer was dried in a vacuum oven over a 24 h period at 60 �C in order to remove moisture content. The spinning dopes of 17 wt.% of PVDF polymer with non-solvent additive concentrations of 0, 3 and 5 wt.% were prepared by stirring polymer in solvent (DMAc) homogenously at 60–70 �C for several hours. The dopes were degassed for more than one hour prior to spinning. Details of the spinning process can be found elsewhere [24]. The spun fi-bers were immersed in water for three days to ensure the complete removal of solvents. The water in the hollow fibers was exchanged with methanol and then n-hexane to minimize fiber shrinkage and pore collapsin g prior to drying at room temperature. Spinning con- ditions and parameters applied in this work are given in Table 1.

2.3. Gas permeation measurement

Gas permeation testing is used to determine mean pore size and effective surface porosity on the outer skin layer of the asymmetric membran es [22]. Based on the common gas permeation method by Wang et al. [25], gas permeance JG for a porous membrane can be expresse d as:

JG ¼2rpe

3RTLp

8RTpM

� �0:5

þ rp2e8lRTLp

P or JG ¼ Ko þ PoP ð1Þ

where JG is gas permean ce (mol m�2 s�1 Pa�1), rp and Lp are pore radius and effective pore length (m), e is surface porosity, R is gas

100 R. Naim et al. / Separation and Purification Technology 109 (2013) 98–106

constant 8.314 (J mol �1 K�1), l is gas viscosity (kg m�1 s�1), M is gas molecul ar weight (kg mol �1), T is gas temperatur e (K) and P is mean pressure (Pa). Pure N2 was used as a standard gas for permeatio nexperime nts and permeatio n flux was measured at various pressure intervals from 50,000 Pa to 200,000 Pa. The test modul e was housed with one hollow fiber membrane of 10 cm in length. The gas perme- ation rate was measured at room temperat ure using a soap bubble flow meter. N2 permeance was then calculate d considerin g the out- er diamete r of the hollow fiber. By plotting the gas permeance vs. mean pressure, JG vs. P, the intercept and slope are given as (Ko)and (Po). The mean pore size and the effective surface porosity can be calcul ated from the following equations :

rp ¼ 5:333Po

Ko

� �8RTpM

� �0:5

l ð2Þ

eLp¼ 8lRTPo

r2p

ð3Þ

2.4. Liquid entry pressure (LEPw) and contact angle

A test module consisting of one or two fibers was used where deionized water was fed into the lumen side of the fiber using adiaphragm pump. The pressure was slowly increased at 50,000 Pa intervals and was kept at constant pressure at each pres- sure interval for about 20 min to examine for the appearance of water droplets at the outer surface of the hollow fiber. The pres- sure recorded when the first water droplets appear on the outlet of dry fiber skin is regarded as liquid entry pressure [26]. For water contact angle measure ment, five samples of hollow fibers were dried in a vacuum oven at 60 �C for 12 h. A technique called sessile drop using a goniometer (Model G1, Krauss GmbH, Germany ) was used to measure the contact angle of the outer surface of the fibers.An average of 15 contact angle values was measured at various positions of the sample.

2.5. Atomic force microscopy (AFM) anlysis

The technique for the AFM (tapping mode) for hollow fiber is described elsewhere [26,27]. AFM studies were conducted in tap- ping mode at the outer and inner of the hollow fibers using a Nano- scope III equipped with a 1553D scanner (Digital Instruments, CA).A crystal silicon probe cantilever was with a spring constant of 20–100 Nm �1, length of 125 lm and a nominal tip radius of curvature of 5–10 nm was used. Each hollow fiber was cut into small piece of 1 cm length which was fixed to a metal disk (a puck, 1.5 cm of out- er diameters ) by a double-si ded tape and the AFM study was done on the exterior surface. The laser beam of the AFM was focused on the pre-selected spot of the surface prior to the engagem ent of the cantilever. For the inner surfaces, the hollow fiber was cut at an in-

Fig. 1. The cutting technique used for AFM studies of the internal surface of hollow fiber membranes [28].

clined angle by a sharp razor under a microscope as shown in Fig. 1.

The roughness parameters were determined by the AFM soft- ware and the pore size and nodule size were determined by visual inspection of the line profile of different pores and nodules from various AFM images [28]. The roughness parameter depends on the curvature and the size of the AFM tip, as well as on the treat- ment of the captured image (plane fitting, flattering, filteringetc.). Therefore, the roughness parameters should not be consid- ered as absolute roughness values. In the present study, all AFM images were taken by the same cantilever and all AFM images were treated in a similar way. The mean diameter of a pore or anodule is based on at least 30 measure ments on the same scan size. For roughness parameters, at least six samples of each hollow fiberwere measured and the average of all is given.

In order to obtain the pore size distribution range, several paramete rs such as median ranks (x) and mean pore size and geo- metric standard deviation (rp) should be determined. The mea- sured pore sizes were arranged in ascending order and the correspond ing median ranks (50%) were calculated using the fol- lowing equation [16,22]:

x ¼ i� 0:3nþ 0:4

� �100 ð4Þ

where i is the order number of the measured pore size arranged in ascendin g order and n is the total number of the measured pores. Median ranks values were plotted against pore sizes in an ascending order and this will yield a straight line on the log normal probab ility paper. The mean pore sizes and the correspondi ng standard devia- tion can be calculated from the log-normal plot if this condition is satisfied. The mean value is correspondi ng to the 50% of the cumu- lative number of pores and the geometric standard can be deter- mined by the ratio of 84.13% of the cumulative number of pores to that of 50%. Based on the mean values and the geome tric stan- dard deviation s, the pores size distribution which can be described by the probability density function can be obtained as follows [16,22]:

df ðdpÞdðdpÞ

¼ 1

dp lnrpð2pÞ1=2 exp �ðln dp � ln lpÞ

2

2ðln rpÞ2

!ð5Þ

where dp is the pore size, rp the geometri c standard deviation and lp is the mean pore size of the membrane .

2.6. CO 2 stripping test

A stainless steel membran e contactor module was used to determine the CO 2 stripping performanc e of the prepared mem- branes. 10 hollow fibers were installed in a stainless steel module with detailed specifications as listed in Table 2. An aqueous DEA solution of (10 wt%) was preloaded with pure CO 2 in a gas–liquidmembran e contactor and used as liquid feed stream in the CO 2stripping module. Pure nitrogen as a sweep gas was simulta- neously flowed through the module shell side while preloaded DEA solution was similarly made to flowed inside the lumen side

Table 2Specification of gas–liquid membrane contactor system.

Module length (mm) 220 Module inner dia. (mm) 10 Fiber outer dia. (lm) 795 ± 2.6 Fiber inner dia. (lm) 545 ± 1.4 Effective fiber length

(mm)160

Number of fibers 10

R. Naim et al. / Separation and Purification Technology 109 (2013) 98–106 101

of the hollow fibers. A counter-current flow mode was applied for the gas and liquid phases. The pressure and the flow rates of the gas and liquid phases were controlled by the control valves. The pressure of the liquid stream was 20,000 Pa higher than that of gas side to avoid the formation of bubbles on the liquid side. Afixed operating temperature (80 �C) was applied during the exper- iment. The CO 2 concentration in the inlet and outlet of the liquid stream was determined by the double chemical titration method [30]. The system was in operation for 30 min to achieve a steady state condition before taking samples. The CO 2 stripping flux of the module was calculated as:

JCO2¼ ðCli � CloÞ � Q l

Aið6Þ

where JCO2is the CO 2 stripping flux (mol m�2 s�1); Ql is the liquid

flow rate (m3 s�1); and Ai is the inner surface of the hollow fibermembrane s (m2). The overall mass transfer coefficient of the PVDF membrane s was calculate d and discussed in details as reported elsewhere [31].

3. Results and discussion

3.1. Characterizati on result of PVDF membranes

The characterization results of PVDF membranes upon the addi- tion of LiCl additives in the polymer dope solution are presented in Table 3. It is observed that an increase in LiCl concentratio n has re- duced the gas permeation , contact angle and mean pore size of the membrane. Shi et al. [32] reported that addition of LiCl and glycerol (up to 4 wt%) into PVDF-HFP/N MP system has significantly reduced the macro-void formatio n thus reducing the membrane pore size. These results are directly related to the decrease in gas permeabil- ity for all PVDF membran es. The decreasing trend observed in con- tact angle measureme nt with the increasing LiCl might be probably due to the alteration of surface roughness. This phenomenon will be further discussed in the AFM analysis section. A reverse effect can be noticed for liquid entry pressure and effective surface porosity of the membrane as the concentration of LiCl increased. This is associate d to the decrease in the pore size at the membran einner surface which enhanced the liquid entry pressure according to the Laplace equation. Since the formatio n of macrovoid in the membrane was suppressed by the addition of LiCl additive, it is ex- pected that the effective surface porosity of the membrane will be increased.

3.2. Results on morphological characterization of membranes by AFM

The AFM images of the outer and inner surface of PVDF hollow fiber membran es are illustrate d in Figs. 2 and 3, respectively . Fig. 2shows the 2D and 3D images of the outer surface at 1 lm scan. The nodule aggregates can be regarded as bright high peaks and the darker spots as the pores. From Fig. 2, it is observed that the nodule aggregate size decreases with the addition of LiCl (from S1 to S2).The large nodule aggregate have broken into smaller sizes. This trend is further continued as the LiCl concentratio n increased from

Table 3Characteristics of PVDF hollow fiber membrane.

Membrane N2 permeance at 100 kPa (10�3 cm3/cm 2.s.cm.Hg)

Mean pore size (nm)

Effec(m�1

Plain PVDF 13.10 ± 0.6 430 ± 1.3 3PVDF + 3%

LiCl 8.25 ± 1.0 20 ± 0.4 1209

PVDF + 5% LiCl

1.00 ± 0.7 14 ± 0.8 2305

S2 to S3. It is also noticed that the rows of the nodule aggregat es are aligned in the plain PVDF hollow fiber, maintain ing some dis- tance from each other. With the increasing concentr ation of LiCl, the distance between the nodule aggregates was reduced, leading to the reduced formation of the dark spots.

Fig. 3 shows the 2D and 3D images of the inner hollow fiber sur- face at 2 lm and 1 lm scan, respectively . The same trend is ob- served as at the outer surface, i.e. the nodule size decrease d with the increase in LiCl concentration. Table 4 gives more quantitative data of the nodule sizes. According to the data in Table 4, the size of the nodule aggregat e at the inner surface of plain PVDF membrane (240 nm) is much larger than that of outer surface (76 nm). This huge difference is probably due to the difference between the com- positions of the bore fluid, in which the inner surface is in contact that has resulted in the larger nodule size at the outer surface. These results are probably attributable to the increase in polymer solution viscosity in the presence of LiCl as shown in Table 3. It is observed that high solution viscosity of polymer solution has in- duced the delay in polymer coagulation.

Table 5 shows the pore size of PVDF membranes obtained from the AFM analysis and gas permeation test. The size of membrane pores is often related directly with the size of nodules [17] inwhich the pore size decreases with the decrease in the nodule size. Therefore, as expected from the nodule sizes given in Table 4, the membran e pore size, both at the inner and outer membrane sur- face decrease d as the concentration of LiCl increased. The reason for the pore size reduction with the increase in LiCl concentr ation is probably the same as that for the reduction of the nodule size, i.e. an increase in LiCl concentr ation increased the spinning solution viscosity , hence led to the delayed demixing.

Comparing the pore sizes of the inner and the outer surface, the mean pore size of the inner surface is much greater than that of outer surface with respect to the plain PVDF membrane. The pore size of the inner surface also greater than that of outer surface with the LiCl additive but only minor differenc e was observed. The same tendency was observed for the nodule aggregate size (see Table 4).Comparing the mean pore sizes obtained from AFM and those obtained from gas permeation tests (see the last column in Table 5),the trend observed for the inner surface (PlainPVDF > PVDF + 3%LiCl > PVDF + 5%LiCl) agrees with that of ob- tained from the gas permeation test. However, the gas permeation test has given larger mean pore size than AFM for the plain PVDF while the trend was reversed for the membrane spun with LiCl in the dope. In both methods (AFM and gas permeation test), the obtained pore sizes should be considered not as absolute values but as relative values for comparison purposes.

Fig. 4 shows the probabili ty density function vs. pore size of the PVDF membran es for both inner and outer surfaces. It can be no- ticed that the pore size distribution curves shifted to the left when the LiCl concentr ation was increased in the PVDF polymer solution and the pore size distribut ion become narrower around the mean pore size.

The data of the mean surface roughness (Ra) of the inner and outer are summarized in Table 6. It is well accepted that the in- crease in pore size and nodule size normally enhances the surface

tive surface porosity, e/Lp

)LEP W(x105 Pa)

Contact angle (�C)

Viscosity (cP at 25 �C)

.20 3.0 ± 0.2 85 ± 0.3 510 ± 0.5

.05 6.0 ± 0.4 80 ± 0.6 1267 ± 0.5

.54 7.0 ± 0.1 75 ± 0.4 2381 ± 0.5

Fig. 2. AFM images for outer surface of different PVDF hollow fiber membrane (S1: plain, S2: 3% LiCl, S3: 5% LiCl) prepared at: (S1a–S3a) 1 lm scan (S1b–S3b) 1 lm scan (3Dimages).

102 R. Naim et al. / Separation and Purification Technology 109 (2013) 98–106

roughness parameter [33]. The surface roughness decrease d with increasing LiCl concentratio n in the spinning dope. Comparing the inner surface roughness with the outer surface, the former is higher than the latter. This is in accordance with the trend ob- served in the pore sizes measure d by AFM. In fact, when the sur- face consists less dark spots that characterize pores and low peaks which correspond to the formation of nodules, low surface roughness parameters are expected. It is worth mentioning that the introduct ion of LiCl in the membrane dope solution indirectly

reduced the peak nodule aggregates of the PVDF membrane thus forming smaller nodules with less dark spots and smoother surface (as shown in Figs. 1 and 2). Similar correlation between surface roughness and pore size has been also reported by Singh et al. [29].

The relationship between surface roughness and contact angle has been widely studied [34,35]. It is claimed that a rough or tex- tured surface has varying effects on the contact angles of wetting liquids which can be represented in two proposed models: Wen- zel’s model and Cassie–Baxter model. Based on the proposed

Fig. 3. AFM images for inner surface of PVDF hollow fiber membrane (S1 – plain, S2 – 3% LiCl, S3 – 5% LiCl) prepared at; (S1a–S3a) 2 lm scan and (S1b–S3b) 1 lm scan (3Dimages).

Table 4Mean, maximum and minimum sizes of the nodular aggregates on the inner and outer surface of PVDF hollow fiber membranes.

Type of membranes Nodule aggregates size (nm)

Inner surface Outer surface

Mean Maximum Minimum Mean Maximum Minimum

Plain PVDF 239.8 ± 3.5 310.4 ± 3.7 142.8 ± 2.4 76.2 ± 2.5 104.6 ± 1.3 40.9 ± 1.2 PVDF + 3Li 30.6 ± 2.6 59.65 ± 2.7 19.6 ± 1.4 71.5 ± 2.4 90.1 ± 1.1 38.2 ± 0.7 PVDF + 5Li 27.8 ± 1.8 58.1 ± 0.6 19.6 ± 0.8 45.5 ± 0.6 51.2 ± 0.6 30.1 ± 2.0

R. Naim et al. / Separation and Purification Technology 109 (2013) 98–106 103

Table 5Pore sizes of PVDF hollow fiber membranes.

Type of membranes Pore size (nm) From gas permeation

Inner surface Outer surface

Mean Maximum Minimum Mean Maximum Minimum

17PVDF 142.8 ± 3.2 310.9 ± 2.1 85.8 ± 1.4 58.5 ± 1.2 78.1 ± 2.3 39.0 ± 1.2 430 ± 2.5 17PVDF + 3Li 46.7 ± 0.3 68.6 ± 1.2 31.4 ± 0.7 45.4 ± 2.1 75.6 ± 1.8 35.0 ± 2.1 20 ± 1.6 17PVDF + 5Li 45.4 ± 0.9 47.0 ± 0.4 20.5 ± 1.7 38.6 ± 0.5 48.5 ± 1.3 22.1 ± 1.5 14 ± 0.4

Fig. 5. Results of CO 2 stripping flux for PVDF membranes.

Table 7Calculated individual mass transfer coefficient for PVDF membranes.

kL (ms�1) km (ms�1) Overall mass transfer coefficient(ms�1)

Plain PVDF 9.60 � 10�5 1.37 � 10�3 3.16 � 10�4

PVDF + 3% LiCl

9.77 � 10�5 2.24 � 10�3 4.51 � 10�4

PVDF + 5% LiCl

1.04 � 10�4 2.35 � 10�3 4.86 � 10�4

(MDEA = 1 M, VL = 0.40 m s�1, T = 80 �C, Henry’s constant of 1 M DEA = 0.767).

Fig. 4. Probability density function curves for (a) inner surface (b) outer surface; generated from the pore sizes measured from the AFM images of the PVDF membranes.

Table 6Mean surface roughness of PVDF hollow fiber membranes.

Type of membranes Mean surface roughness (nm)

Inner surface Outer surface

Plain PVDF 19.923 ± 1.6 12.583 ± 1.3 PVDF + 3% LiCl 10.05 ± 1.4 7.445 ± 1.5 PVDF + 5% LiCl 6.575 ± 1.7 6.471 ± 1.2

104 R. Naim et al. / Separation and Purification Technology 109 (2013) 98–106

models, surface with different textures can be classified into two; homogeneous and heteroge neous, where the regime of liquid fillsis very much different with each other. From Tables 3 and Table 5,the contact angle decreased as the roughness of the outer surface, where the contact angle decreased. Since LiCl, as the additive has been completely removed from the membrane during the water immersion for 3 days, while the membran e was being immersed in water for 3 days, the change in contact angle can be considered solely as the effect of the surface morphology. Most likely, this is due to the decrease of the effective solid surface area as a result

of the surface smoothing, which decrease d the interfacial energy between the solid and liquid thus leading to the decrease in con- tact angle.

3.3. Results of CO 2 stripping on membrane contactor

As can be seen in Fig. 5, it was observed that the CO 2 strippingflux of PVDF membranes increased with the increasing LiCl con- centration in the spinning dope. As well, the flux increased almost linearly with the absorbent liquid velocity. Interestingl y, the trend in the CO 2 stripping flux is almost opposite to the N2 gas flux. Upon addition of LiCl additive in the polymer spinning solution, the membran e pore size and nodule aggregates became smaller thus enhanced the effective surface contact for the gas and liquid mass transfer process to occur. As the membrane surface roughness de- creased with the additive concentration, the CO 2 stripping perfor- mance was improved in a significant manner. It is postulated that the decrease in membrane surface has reduced the membrane phase transfer resistance and with the increasing concentr ation of LiCl, the membrane surface became smoother thus the mem- brane mass transfer resistance became insignificant compared to liquid mass transfer resistance [36]. In addition, it can be assumed that since the variation of surface roughness on the inner surface of membran e with LiCl content of 3 wt% and 5 wt% is very small, the

R. Naim et al. / Separation and Purification Technology 109 (2013) 98–106 105

CO2 stripping results for both membranes have insignificant differ- ence in term of flux performanc e. In a similar work reported by Bakeri et al. [37], they reported on the AFM analysis on the outer surface of polyetherimide hollow fiber membrane upon addition of surface modified macromolecul es (SMM). It was claimed that the surface modified membran e was rougher than the pristine membrane and the CO 2 absorption test with distilled water for modified membrane exhibited higher absorption flux than the plain membran e. The surface roughness may affect the contact an- gle of the membrane but it is not confirmed to what extent this surface characteristic could influence the absorption/ stripping per- formance of the membrane. Based on the experimental results, it can be concluded the addition of LiCl in PVDF hollow fiber mem- branes has smoothen the membrane surface structure thus en- hanced the CO 2 stripping flux.

As for overall mass transfer analysis, the gas mass transfer resis- tance is assumed negligible and has insignificant effect on the overall mass transfer coefficient [36]. Therefore, only the liquid and membrane mass transfer coefficient will be further discussed in this study. As observed in Table 7, addition of LiCl has significanteffect on the membrane mass transfer coefficient where it can be seen that the membrane mass transfer coefficient was improved as the LiCl concentration increased from 0 to 3 wt% and 5 wt%. These results can be associate d to the decreasing values in the in- ner surface roughness upon increasing the additive concentr ation. It is assumed that the incorporation of LiCl additives in PVDF mem- brane has contributed to smoothin g effects on the inner and outer membrane surface thus reducing the interference for the mass transfer between the gas and liquid phase. However as for liquid side, only minimal increase in liquid mass transfer coefficient val- ues was achieved even with increasing LiCl concentratio n. Liquid flow rate and contact area at the liquid side has more prominent effect on the improvement in liquid mass transfer compare d to the increasing additive concentr ation. Therefore, it can be con- cluded that the presence of LiCl additive in the PVDF membrane polymer dopes has contributed to significant increase in the mem- brane mass transfer coefficient but only minimal effect in the li- quid mass transfer coefficient. In general, the greater the value of the mass transfer coefficient, the lower the resistance for mass transfer between phases to occurs [38]. However , the liquid mass transfer coefficient still remains as the dominant contributor to the overall mass transfer resistance of PVDF membran es.

4. Conclusions

PVDF hollow fibers were spun from the hollow fiber spinning dope with and without the addition of LiCl. The surface morphology of the membranes was then investiga ted by the gas permeation test and AFM analysis. The membranes were further analyzed for CO 2stripping in a membrane contactor system. From the present study, it can be noticed that both gas permeation test and AFM analysis showed that the pore size decreased with an increase in LiCl concen- tration in the membrane. Increasing LiCl concentration in the PVDF membrane dopes has a decreasing effect on the contact angle value, membrane surface roughness and nodules aggregates. Based on the experimental test, CO 2 stripping flux was precisely in a reverse order of gas permeation flux in which the flux increased with the reducing membrane pore size. Smoothing effect due to increasing concentra- tion of LiCl additive has eventually enhanced the CO 2 stripping fluxand the membran e mass transfer coefficient but has negligible effect on the liquid side mass transfer coefficient.

Acknowled gements

The authors gratefully acknowled ge the financial support from the Ministry of Higher Education (MOHE), Malaysia and Universiti

Malaysia Pahang. We also thank Dr. C.Y. Feng from the University of Toronto for his assistant in analysing the pore size distribution data.

References

[1] J. Mackey, J. Mojonnier, CO 2 injection using membrane technology, in: 8th International Conference on the Operation of Technologiec Advanced Beverage Plants and Warehouses, 1995, p. 1–22.

[2] L. Sciubba, D. Di Gioia, F. Fava, C. Gostoli, Membrane-based solvent extraction of vanillin in hollow fiber contactors, Desalination 241 (2009) 357–364.

[3] P. Wang, T.-S. Chung, Design and fabrication of lotus-root-like multi-bore hollow fiber membrane for direct contact membrane distillation, J. Membr. Sci. 421–422 (2012) 361–374.

[4] P.T. Nguyen, E. Lasseuguette, Y. Medina-Gonzalez, J.C. Remigy, D. Roizard, E. Favre, A dense membrane contactor for intensified CO 2 gas/liquid absorption in post-combustion capture, J. Membr. Sci. 377 (2011) 261–272.

[5] S. Khaisri, D. DeMontigny, P. Tontiwachwuthikul, R. Jiraratananon, CO 2stripping from monoethanolamine using a membrane contactor, J. Membr. Sci. 376 (2011) 110–118.

[6] R. Wang, D.F. Li, C. Zhou, M. Liu, D.T. Liang, Impact of DEA solutions with and without CO 2 loading on porous polypropylene membranes intended for use as contactors, J. Membr. Sci. 229 (2004) 147–157.

[7] G. Bakeri, A.F. Ismail, M. Shariaty-Niassar, T. Matsuura, Effect of polymer concentration on the structure and performance of polyetherimide hollow fiber membranes, J. Membr. Sci. 363 (2010) 103–111.

[8] K.C. Khulbe, C.Y. Feng, F. Hamad, T. Matsuura, M. Khayet, Structural and performance study of micro porous polyetherimide hollow fiber membranes prepared at different air-gap, J. Membr. Sci. 245 (2004) 191–198.

[9] Z.-K. Xu, L.-Q. Shen, Q. Yang, F. Liu, S.-Y. Wang, Y.-Y. Xu, Ultrafiltration hollow fiber membranes from poly(ether imide): preparation, morphologies and properties, J. Membr. Sci. 223 (2003) 105–118.

[10] K. Kneifel, K.-V. Peinemann, Preparation of hollow fiber membranes from polyetherimide for gas separation, J. Membr. Sci. 65 (1992) 295–307.

[11] D. Wang, K. Li, W.K. Teo, Preparation and characterization of polyetherimide asymmetric hollow fiber membranes for gas separation, J. Membr. Sci. 138 (1998) 193–201.

[12] G. Bakeri, T. Matsuura, A.F. Ismail, The effect of phase inversion promoters on the structure and performance of polyetherimide hollow fiber membrane using in gas–liquid contacting process, J. Membr. Sci. 383 (2011) 159–169.

[13] M. Khayet, C.Y. Feng, K.C. Khulbe, T. Matsuura, Study on the effect of a non- solvent additive on the morphology and performance of ultrafiltration hollow- fiber membranes, Desalination 148 (2002) 321–327.

[14] N. Dizge, G. Soydemir, A. Karagunduz, B. Keskinler, Influence of type and pore size of membranes on cross flow microfiltration of biological suspension, J. Membr. Sci. 366 (2011) 278–285.

[15] D. Wu, X. Liu, S. Yu, M. Liu, C. Gao, Modification of aromatic polyamide thin- film composite reverse osmosis membranes by surface coating of thermo- responsive copolymers P(NIPAM-co-Am). I: Preparation and characterization, J. Membr. Sci. 352 (2010) 76–85.

[16] N. Peng, T.-S. Chung, J.-Y. Lai, The rheology of Torlon solutions and its role in the formation of ultra-thin defect-free Torlon hollow fiber membranes for gas separation, J. Membr. Sci. 326 (2009) 608–617.

[17] K.C. Khulbe, C.Y. Feng, T. Matsuura, Synthetic polymeric membranes: characterization by atomic force microscopy, Springer, Germany, 2008.

[18] G. Binning, C.F. Quate, C. Gerber, Atomic force microscope, Phys. Rev. Lett. 56 (1986) 930–933.

[19] T.R. Albrecht, C.F. Quate, Atomic resolution with the atomic force microscope on conductors and non-conductors, J. Vac. Sci. Technol. A 6 (1988) 271–275.

[20] K.C. Khulbe, C. Feng, T. Matsuura, G.C. Kapantaidakis, M. Wessling, G.H. Koops, Characterization of polyethersulfone–polyimide hollow fiber membranes by atomic force microscopy and contact angle goniometery, J. Membr. Sci. 226 (2003) 63–73.

[21] M. Khayet, K.C. Khulbe, T. Matsuura, Characterization of membranes for membrane distillation by atomic force microscopy and estimation of their water vapor transfer coefficients in vacuum membrane distillation process, J. Membr. Sci. 238 (2004) 199–211.

[22] M. Khayet, C.Y. Feng, K.C. Khulbe, T. Matsuura, Preparation and characterization of polyvinylidene fluoride hollow fiber membranes for ultrafiltration, Polymer 43 (2002) 3879–3890.

[23] E.F. Barbosa, L.P. Silva, Nanoscale characterization of synthetic polymeric porous membranes: scrutinizing their stiffness, roughness, and chemical composition, J. Membr. Sci. 407–408 (2012) 128–135.

[24] A.F. Ismail, S.N. Kumari, Potential effect of potting resin on the performance of hollow fibre membrane modules in a CO 2/CH4 gas separation system, J. Membr. Sci. 236 (2004) 183–191.

[25] D. Wang, K. Li, W.K. Teo, Effects of temperature and pressure on gas permselection properties in asymmetric membranes, J. Membr. Sci. 105 (1995) 89–101.

[26] K. Smolders, A.C.M. Franken, Terminology for membrane distillation, Desalination 72 (1989) 249–262.

[27] C.Y. Feng, K.C. Khulbe, G. Chowdhury, T. Matsuura, V.C. Sapkal, Structural and performance study of microporous polyetherimide hollow fiber membranes made by solvent-spinning method, J. Membr. Sci. 189 (2001) 193–203.

106 R. Naim et al. / Separation and Purification Technology 109 (2013) 98–106

[28] M. Rafat, D. De, K.C. Khulbe, T. Nguyen, T. Matsuura, Surface characterization of hollow fiber membranes used in artificial kidney, J. Appl. Polym. Sci. 101 (2006) 4386–4400.

[29] S. Singh, K.C. Khulbe, T. Matsuura, P. Ramamurthy, Membrane characterization by solute transport and atomic force microscopy, J. Membr. Sci. 142 (1998)111–127.

[30] M.-H. Li, B.-C. Chang, Solubilities of carbon dioxide in water + monoethanolamine + 2-amino-2-methyl-1-propanol, J. Chem. Eng. Data 39 (1994) 448–452.

[31] R. Naim, A.F. Ismail, A. Mansourizadeh, Effect of non-solvent additives on the structure and performance of PVDF hollow fiber membrane contactor for CO 2stripping, J. Membr. Sci. 423–424 (2012) 503–513.

[32] L. Shi, R. Wang, Y. Cao, D.T. Liang, J.H. Tay, Effect of additives on the fabrication of poly(vinylidene fluoride-co-hexafluropropylene) (PVDF-HFP) asymmetric microporous hollow fiber membranes, J. Membr. Sci. 315 (2008) 195–204.

[33] W.R. Bowen, N. Hilal, R.W. Lovitt, C.J. Wright, Atomic force microscope studies of membrane surfaces, in: T.S. Sorensen (Ed.), Surface Chemistry and Electrochemistry of Membranes, Dekker, New York, 1999, p. 1.

[34] Y. Tamai, K. Aratani, Experimental study of the relation between contact angle and surface roughness, J. Phys. Chem. 76 (1972) 3267–3271.

[35] X.B. Zhou, J.T.M. De Hosson, Influence of surface roughness on the wetting angle, J. Mater. Res. 10 (1995) 1984–1992.

[36] R. Naim, A.F. Ismail, A. Mansourizadeh, Preparation of microporous PVDF hollow fiber membrane contactors for CO 2 stripping from diethanolamine solution, J. Membr. Sci. 392–393 (2012) 29–37.

[37] G. Bakeri, A.F. Ismail, D. Rana, T. Matsuura, Development of high performance surface modified polyetherimide hollow fiber membrane for gas–liquidcontacting processes, Chemi. Eng. J. 198–199 (2012) 327–337.

[38] P. Luis, T. Van Gerven, B. Van der Bruggen, Recent developments in membrane- based technologies for CO 2 capture, Prog. Energy Combust. 38 (2012) 419–448.