Development of antifouling properties and performance of nanofiltration membranes modified by interfacial polymerisation

Desalination 273 (2011) 36–47

Contents lists available at ScienceDirect

Desalination j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / d e s a l

Development of antifouling properties and performance of nanofiltration membranes modified by interfacial polymerisation M.N. Abu Seman a,b, M. Khayet c, Nidal Hilal a,⁎ a b c

Centre for Water Advanced Technologies and Environmental Research (C WATER), School Engineering, Swansea University, Swansea, SA2 8PP, UK Faculty of Chemical Engineering and Natural Resources, Universiti Malaysia Pahang, Lebuhraya Tun Razak, 26300 Kuantan, Pahang, Malaysia Department of Applied Physics I, Faculty of Physics, University Complutense of Madrid, Av. Complutense s/n, 28040, Madrid, Spain

a r t i c l e

i n f o

Article history: Received 20 July 2010 Received in revised form 21 September 2010 Accepted 22 September 2010 Available online 13 October 2010 Keywords: Polyethersulfone Interfacial polymerisation Humic acid Membranes AFM

a b s t r a c t Two types of bisphenol monomers, Bisphenol A (BPA) and Tetramethyl Bisphenol A (TMBPA), with different concentrations of bisphenol aqueous solution (0.5% to 2.%w/v) and various interfacial polymerisation times (10 s, 30 s and 60 s) in the fixed 0.15%w/v organic solution of trimesoyl chloride (TMC)-hexane were studied. Irreversible fouling of both unmodified polyethersulfone NFPES10 and modified polyester thin-film composite polyethersulfone membranes were studied using humic acid model solutions at two different pH values, pH 7 and pH 3. It was observed that polyester thin-film composite membranes prepared by BPA exhibited fewer tendencies for irreversible fouling by humic acid molecules at neutral environment compared to unmodified NFPES10 and TMBPA-polyester series. This is most probably due to high electrostatic repulsion force between negatively charged of BPA-polyester layer and highly negative charged of humic acid at pH7. However, some modified membranes with rougher surfaces were severely fouled by humic acid molecules at acidic environment, pH 3. Under this acidic environment, carboxylic acid groups of humic acid lost their charge and the macromolecules of humic acid have smaller macromolecular configuration due to the increased hydrophobicity and reduced inter-chain electrostatic repulsion. Thus the molecules of humic acid may be preferentially accumulated at the valleys of the rougher membrane surface blocking them and resulting in a more severe fouling. In addition, the modification also affected membrane pore size and pore size distribution as shown by AFM images. It was also observed that the smaller pore size generated after modification does not have significant effect on humic acid removal due to the larger size of humic acid molecules. All the modified membranes posses smaller pore size than the unmodified NFPES10 (1.47 nm) in the range of 0.8–1.34 nm. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Interfacial polymerisation (IP) is a technique used for preparation of composite nanofiltration (NF) membranes. In general, the produced membranes exhibit high water permeation flux and high salt rejection [1]. The interfacial polymerisation does not only change the physical morphology (i.e. roughness, pore size, porosity etc.) of the membrane surface, it also significantly affects the chemistry properties (i.e. hydrophilicity, surface charge, etc.) of the membrane. Both physical and chemical properties indirectly affect the membrane performance (permeability and rejection) and certain degree of fouling. One of the common monomer used for production of nanofiltration membrane is M-phenylenediamine (MPD) [2,3]. Other amine monomers were also used for the production of polyamide membranes [4,5]. The polyamide membranes are produced from reaction of amine monomer with acyl chloride in the organic phase. ⁎ Corresponding author. Tel.: + 44 1792 606675; fax: + 44 1792 295676. E-mail addresses: [email protected] (M.N. Abu Seman), [email protected] (M. Khayet), [email protected] (N. Hilal). 0011-9164/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.desal.2010.09.038

The prepared membranes exhibit high rejection and permeate flux with controlled fouling properties [4,6,7]. For example, Ahmad et al. (2003) [4] prepared and characterised polyamide membranes incorporating highly hydrophilic aromatic primary amide (3,5-diaminobenzoic acid, BA). They observed an increase in pure water permeability up to 23% higher than the unmodified polyamide membrane. However high content of BA produced loose and defects in skin layer; their membranes has lower NaCl and Na2SO4 rejections due to defects caused by the BA addition. Chu et al. (2005) [6] modified ceramic-supported polyethersulfone membrane surface using interfacial polymerisation in order to reduce fouling, through increasing hydrophilicity, during the treatment of oil-water microemulsions. Their results showed that the hydrophilic surface consequently reduced the membrane fouling effectively. Kang et al. (2007) [7] produced RO polyamide-PEG grafted membrane using interfacial polymerisation. At the final stage of interfacial polymerisation (before drying), the unreacted acyl chloride groups were reacted with Aminopolyethylene glycol monomethylether (MPEG-NH2) solution to produce surface modified membrane. The fouling experiments showed that the modified membrane had anti-fouling property. These studies revealed that interfacial polymerisation technique changes the

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47

membrane surface properties (hydrophilicity, membrane charge and roughness) and therefore affects membrane performance (including permeability and rejection correspond to pore size reduction and porosity) and fouling as well. Compared to thin-film polyamide membranes prepared by interfacial polymerisation, only few research studies have been carried out using this technique to prepare other polymeric thin-film polyester and polyesteramide membranes [8–12]. Moreover, it appears that of the few studies on Bisphenol modified polyester desalination membranes have been conducted and did not focus on humic acid fouling but the on salt permeability. The main objective of this paper is to study the performance of polyester thin-film composite membranes with improved NF antifouling properties by means of interfacial polymerisation using different molecular structure of phenol (Bisphenol A, BPA, and tetramethyl Bisphenol A, TMBPA) reacted with trimesoyl chloride (TMC) as monomers. Different humic acid aqueous solutions of different pH values have been tested. The membrane performance (permeability, rejection), roughness (by AFM analysis), pore size (by solute transport method) and irreversible fouling of both modified and unmodified membranes have been compared.

2. Experimental 2.1. Materials The monomers Bisphenol A (BPA), tetramethyl Bisphenol A (TMBPA) and trimesoyl chloride (TMC) were purchased from SigmaAldrich Co. Their chemical structures are shown in Fig. 1. The solvent hexane (reagent grade) was purchased from Fisher Scientific, UK. Humic acid in powder form (Sigma-Aldrich Co.) was chosen as a model organic foulant. A stock aqueous solution of 1 g/L humic acid was prepared using deionised water (Milli-Q Plus, 18.2 M Ω cm) and adjusting the pH to a value slightly higher than 9 to make sure that all humic acid powder was completely dissolved. To adjust the pH of the feed humic acid solutions to the required values, hydrochloric acid (HCl, Sigma-Aldrich Co.) and sodium hydroxide (NaOH, Acros Organics) were used. For membrane pore size determination, polyethylene glycol (PEG, from Sigma-Aldrich Co.) with different molecular weight in the range of 200 g/mol to 3350 g/mol were employed.

37

The asymmetric commercial polyethersulfone nanofiltration membrane, NFPES10 was purchased from Hoechst Company (Germany) with 75 μm thickness (based on SEM image without backing material), was used as base support for interfacial polymerisation. In all experiments, deionised water from Milli-Q Plus with resistance of 18.2 MΩ cm was used. 2.2. Preparation of thin-film composite polyester membrane using interfacial polymerisation technique NFPES10 membrane was first kept overnight in deionised water and then followed by cleaning again using ultrasonic bath to make sure the preservative agent totally removed. NFPES10 nanofiltration membrane was first cleaned with deionised water and subsequently left in ultrasonic deionised water bath for 1 min. These two steps were repeated three times in order to remove any preservative agent (i.e. glycerine) used by the manufacturer. The cleaned membrane was immersed in a TMBPA aqueous solution for 15 min. The monomer TMBPA was dissolved in an aqueous solution of NaOH at pH of about 11 since it has very low solubility in water. The TMBPA monomer concentration in the aqueous solution was varied between 0.1% w/v and 2% w/v. The pre-soaked membrane was taken out from the aqueous solution and positioned vertically for 2 min to drain the excess monomer on the membrane surface. Subsequently, this membrane was dipped in the organic solution prepared by mixing the monomer TMC in hexane (0.15% w/v) and left in the organic solution for a predetermined time (10 s, 30 s or 60 s) for interfacial polymerisation. The same procedure was employed for the monomer BPA and all experiments were performed at ambient temperature. 2.3. Membrane characterisation 2.3.1. Atomic Force Microscopy (AFM) A multimode AFM (Veeco Instruments USA) was used to characterise the surface of both the unmodified and the modified membranes. Comprehensive reviews on membrane characterization by AFM are available in the literature [13,14]. The images were obtained over different areas of each membrane sample. In this study, tapping mode was used and the same tip was employed to scan the surface of all membranes. All captured images were treated in the same way (same temperature, air medium, same scale). Root mean square roughness (RMS) was determined over an area of 5 μm × 5 μm. At least five different samples were characterised to obtain an average value of RMS. 2.3.2. Membrane performance test Nanofiltration experiments were carried out using a membrane module with an effective area of 12.6 × 10− 4 m2, which can be operated under the trans-membrane pressure in the range of 1 × 105 Pa up to 9 × 105 Pa. The feed solution was circulated through the membrane module by means of a pressure pump (Tuthill Pump Co. of California). In nanofiltration experiments the feed and retentate flow rates were maintained at 0.4 L/min. The permeate flux (J) of each membrane sample was determined by weighing the obtained permeate during a predetermined time using an electronic balance (Precisa, Model XB3200C) connected to a computer and the following equation. J=

Fig. 1. Chemical structure of (a) Bisphenol A (BPA), (b) tetramethyl Bisphenol A (TMBPA) and (c) trimesoyl chloride (TMC).

W AΔt

ð1Þ

where W is the weight of the obtained permeate during a predetermined NF operation time Δt and A is the membrane area. Shimadzu UVmini-1240 spectrophotometer was used to determine the concentration of humic acid in the feed, retentate and

38

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47

permeate aqueous solutions. The humic acid rejection factor (α) was calculated as follows: α=

1−

Cp Cf

!

z

× 100

ð2Þ

where Cp and Cf are the humic acid concentration in the permeate and in the feed solutions, respectively. Before all NF experiments, each membrane was first immersed in deionised water and then placed in the filtration cell and pressurised at 7 × 105 Pa for at least 2 h using deionised water. Subsequently, the pure water experiments were conducted at different trans-membrane pressures, ΔP (4 × 105 Pa, 5 × 105 Pa, 6 × 105 Pa and 7 × 105 Pa) in order to determine the pure water permeation flux (Jw) using Eq. (1). The membrane permeability, Pm, was determined from the slope of the straight line which is obtained by plotting the permeate flux (Jw) against ΔP following the following equation. Pm =

Jw ΔP

ð3Þ

Humic acid aqueous solutions were then employed and the flux permeate rate (Jt) as well as the rejection factor (α) were determined as a function of time applying a trans-membrane pressure of 6 × 105 Pa. The initial flux of humic solution is represented by Jo. A stock aqueous solution of 1 g/L humic acid was prepared using deionised water and adjusting the pH to a value slightly higher than 9 to make sure that all humic acid powder was completely dissolved. In this study, the humic acid feed solution of 15 mg/L and two different pH values, 3 and 7, were used. The pH was adjusted using 0.1 M NaOH or 0.1 M HCl. For both unmodified and modified membranes, after the NF experiments conducted using humic acid solution, the system was washed with deionised water and pure water permeation flux (Jwf) was measured again in order to evaluate the irreversible fouling in terms of pure water flux reduction, called hereafter, irreversible fouling factor (FRw) and determined as follows [15,16]. FRW =

The correlation between the solute separation and the solute diameter is expressed as [19–21]:

Jw0 −Jwf 100 Jw0

ð4Þ

2.3.3. Pore size and pore size distribution characterization The pore size and pore size distribution experiments were conducted in aqueous solutions containing 200 ppm of each PEG solute and applying an operating pressure of 6 × 105 Pa. The used PEG concentration was selected taking into consideration that this value is the minimum solute concentration necessary to produce reliable separation data [17,18]. The feed solution temperature was maintained constant at room temperature. Each membrane was initially subjected to pure water experiments to determine the pure water permeation flux (PWP). Then PEG aqueous solutions were circulated through the membrane modules for about 1.5 h. The product permeation rate (PR), in presence of solute as feed was then determined. The solute concentration in the feed, concentrate and retentate were measured using the total organic carbon (TOC) analyzer (Model TOC-VCPH, Shimadzu) and the solute separation was calculated using Eq. (2). Stokes radius was used to characterise the size of the solute. The Stokes diameter of PEG is determined from its molecular weight using Eq. (5) [19–21]: −10

ad = 33:46×10

M

0:557

ð5Þ

where ad is Stokes diameter (cm), and M is the molecular weight of PEG (g/mol).

2 1 −u α = erf ðzÞ = pffiffiffiffiffiffi ∫ e 2 du 2π −∞

ð6Þ

where z=

lnds − lnμ s lnσ s

ð7Þ

and ds is the solute diameter, μ s is the mean size (i.e. geometric mean diameter of solute at a separation factor α = 50%), σs is the geometric standard deviation. Combining Eqs. (6) and (7), a straight line between α and ds can be obtained as: FðαÞ = mðlnds Þ + c

ð8Þ

where m and c are the slope and the intercept on the log-normal probability plot. The mean pore size μ p and geometric standard deviation σp can be determined from m and c values. The mean pore size corresponds to 50% of the solute separation while the geometric standard deviation is the ratio between the Stokes diameter corresponding to 84.13% of the solute separation and the mean pore size (μ p). In this paper, the following parameters were considered to be equal (μ p = μ s, σp = σs) since the dependence of solute separation on the steric and hydrodynamic interaction between solute and pore sizes can be ignored [19–25]. From the obtained mean pore size (μ p) and geometric standard deviation (σp), the pore size distribution can be expressed by the probability density function described by the following equation [19– 21]. " # ðlndp − ln μ p Þ2 df ðdp Þ 1 pffiffiffiffiffiffi exp − = dðdp Þ 2ðlnσ p Þ2 dp lnσ p 2π

ð9Þ

3. Results and discussions 3.1. Effect of TMBPA monomer concentration and reaction time on pure water permeability The obtained average pure water permeabilities, Pm, of both the unmodified NFPES10 and the TMBPA modified membranes are summarised in Table 1. It can be observed that all modified membranes showed lower water permeability than the original NFPES10 membrane Table 1 Average pure water permeability, Pm, of the original unmodified NFPES10 and TMBPA modified membranes. Membrane

Pm (L/m2 h bar)

0 (unmodified NFPES10) 0.1a/10b 0.1/30 0.1/60 0.5/10 0.5/30 0.5/60 1.0/10 1.0/30 1.0/60 2.0/10 2.0/30 2.0/60

15.55 6.66 4.61 2.99 5.18 4.40 2.83 1.38 1.21 0.86 1.13 0.77 0.58

a b

TMBPA monomer concentration (%w/v). Reaction time in seconds (s).

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47

with a water permeability as low as 0.58 L/m2 h bar. The modified membranes can be classified in the NF range as described by Bowen and Mohammad (1998) [25] except the membranes modified with a concentration of 1.0% w/v TMBPA at 30 s and 60 s reaction time, which have characteristics near those of reverse osmosis membranes (RO). Moreover, the membranes modified with the highest concentration of TMBPA (2.0%w/v) also exhibit permeabilities similar to those of RO membranes. The effect of the TMBPA monomer concentration as well as the reaction time on the water permeability of the modified NFPES10 were investigated. Fig. 2 shows the water permeability of both unmodified and modified membranes prepared using different TMBPA concentrations as a function of the reaction time. It can be seen from Fig. 2 that the water permeability of the modified membranes decreases with increasing both monomer concentration and reaction time. In contrast to the results observed for BPA polyester membranes reported elsewhere [26]. The reaction time has significant impact on water permeability even at a low concentration of 0.1%w/v TMBPA monomer. For example, with 10 s reaction time the reduction of TMBPA modified membrane permeability was almost 50% compared to that of the unmodified NFPES10 membrane, whereas BPA modified membrane maintained the water permeability in the range of the unmodified membrane [26]. At higher concentration (1.0 and 2.0%w/v), increasing the reaction time did not have a significant effect on the water permeability. This result can also be observed clearly in Fig. 3. The reduction of modified membrane permeability might be due to the formed denser and/or thicker layer of polyester on the NFPES10 membrane surface. This explanation was already discussed by Mohammad et al. (2003) [9] and Ji and Mehta (2001) [27]. Although the effect of TMBPA concentration and reaction time showed almost similar trends to those of BPA modified membrane, the membrane modified with TMBPA has lower permeability. Fig. 4 shows a comparison of water permeability of BPA and TMBPA modified membranes at the same modification conditions (i.e. monomer concentration and reaction time). Kwak et al. (1997) [8] investigated RO polyester membranes fabricated using both BPA and TMBPA monomers interfacially polymerised on polysulfone (PS) support with 1%w/w monomer concentration at 120 s of reaction time. It was claimed that TMBPA modified membrane had higher permeate flux than BPA membrane with permeabilities of 2.3 L/atm m2 h and 2.1 L/atm m2 h, respectively when the membranes were operated at 29.4 × 105 Pa. However, the difference between both permeabilities is not significant enough even at a very high applied pressure.

Fig. 2. Effect of the reaction time on water permeability of TMBPA modified membranes.

39

Fig. 3. Effect of the TMBPA concentration on water permeability.

The chemical properties of the newly formed polyester layer could be one of the answers for this difference in permeabilities. According to Veríssimo et al. (2006) [28], the hydrophobic/hydrophilic character of the polymer obtained by interfacial polymerisation is influenced by the chemical properties of the monomers involved in the process (either monomer in organic phase or monomer in aqueous phase). All the membranes were prepared, in this study, using the same acid chloride (TMC) at the same concentration (0.15% w/v), therefore only the hydrophobic/hydrophilic character of the bisphenol type can affect the hydrophobic/hydrophilic character of the resulting polymer. The lower permeability of TMBPA-polyester membrane than that of BPA-polyester membrane can be attributed to the increase in the hydrophobicity of TMBPA-polyester membranes compared to BPA-polyester membranes due to the additional four methyl group (–CH3) present in each repeat unit of TMBPA-polyester layer as shown in Fig. 1. In fact, methyl group is classified as a hydrophobic alkyl functional group. Introduction or attachment of these groups as substituent into a compound increases its lipophilicity (i.e. ability of a chemical compound to dissolve in oils, fats, lipids and non-polar solvents such as hexane or toluene) and reduces its solubility in water [29]. Ruiz-Trevino and Paul (1997) [30] modified PS for gas separation membranes by adding a series of bisphenol group (TMBPA, tetrabromobisphenol A, TBBPA and tetra t-butyl bisphenol F, TBuBPF) and other additives. The incorporation of bisphenol-based additives containing small groups symmetrically substituted on the ring, for example TMBPA, increases the selectivity at the expense of drastic reductions in permeability relative to the unmodified PS. Replacing the methyl groups (–CH3) in TMBPA by more 5 polar or by more bulky units like TBBPA or TBuBPF leading to more permeable but much less selective membranes relative to the unsubstituted additive.

Fig. 4. Water permeability of BPA and TMBPA modified membranes under the same modification conditions (*monomer concentration in %w/v, **reaction time in seconds).

40

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47

Despite the increase of hydrophobicity of the methyl group, it may be also affected by physical properties (e.g. rigidity) of TMBPA. Jho (1990) [31] found that polycarbonate (PC) homopolymer synthesised using TMBPA (TMBPA-PC) was more rigid than BPA-polycarbonate (BPA-PC). Research carried out by Kim and Liu (2001) [32] on molecular modelling of BPA-polycarbonate and TMBPA-polycarbonate confirmed the properties of both homopolymers studied by Jho (1990) [31]. Kim and Liu concluded that the radial distribution function shows that BPA-PC has higher chain mobility and flexibility than TMBPA-PC. Their results also showed that the atomic motions of the TMBPA-PC chains were restricted to small scale because of the methylene groups substituted on phenylene (Fig. 1). Moreover, it was found that the molecular motions of BPA-PC have higher diffusion constants than TMBPA-PC and BPA-PC therefore it is easier to have various conformations due to low restrictions in molecular motion. In other studies, Chng et al. (2007) [33] investigated the effects of chemical structure on gas transport properties of poly (aryl ether ketone) (PAEK) copolymer and observed that the glass transition temperature (Tg) of TMBPA modified PAEK was higher than that of BPA-PAEK. Compared to the structure of BPA-PAEK, TMBPA-PAEK copolyimides have four methyl substitutions on the phenyl ring, which disrupt the chain packing and hinder the phenyl ring rotation resulting in an increase in rigidity and/or stiffness of the polymer backbone. In addition, reactivity of monomer may influence the degree of cross-linking. It seems that TMBPA is more reactive than BPA. This observation is clear when referring to the results reported in Fig. 4. At very low concentration of monomer (0.1%w/v), although both TMBPA and BPA reacted during the same reaction time (10 s, 30 s and 60 s), the obtained permeabilities are different (i.e. the permeability of TMBPA-polyester membranes are almost 2 times lower than that of BPA-polyester membranes). This result is also supported by Lee et al. (2008) [2] where it is stated that an increase in cross-linking reaction time caused a decrease in membrane permeate flux. Lee et al. explained that at a higher cross-linking reaction, a denser structure of polymer might be produced reducing the hydrophilicity of the polymeric membrane. It can be concluded from the above discussion that apart from chemical properties (e.g. hydrophobicity), the physical properties such as the rigidity character of TMBPA can also affect the formed polyester structure, which is a denser, more compact and tighter layer with lower permeability compared to BPA-polyester membrane. 3.2. Permeate flux decline of TMBPA membranes Fouling of both modified and unmodified NF membranes was studied using humic acid aqueous solutions of different pH values 3 and 7. The effect of membrane modification by interfacial polymerisation on fouling tendency was investigated. In this case, four modified membranes modified with different TMBPA monomer concentrations (0.1% w/v, 0.5% w/v) and different reaction times (10 s, 30 s) were considered. The modified membranes (0.1TMBPA/ 10 s, 0.1TMBPA/30 s, 0.5TMBPA/10 s and 0.5TMBPA/30 s) were named hereafter 1TMBPA-1, 1TMBPA-3, 5TMBPA-1 and 5TMBPA-3, respectively. The selection was based on water permeability. The membrane modified with the TMBPA monomer concentrations 1%w/v and 2%w/v were not considered in this case since the corresponding membranes exhibited very low permeability. Figs. 5 and 6 show the normalised permeate flux (Jt/J0) of both the unmodified and the modified membranes. As can be seen in Fig. 5, at pH 7 no permeate flux decline was detected for all modified membranes during 240 min (4 h) NF experiment with humic acid solution. It is worth quoting that the membrane 1TMBPA-1 exhibits an enhancement in permeate flux while for the other modified membranes the permeate flux remains unchanged. The increase in permeate flux has been explained in previous work [26]. On the contrary, at pH 3 all membranes

Fig. 5. Permeate flux decline for unmodified NFPES10 and TMBPA modified membranes tested with 15 mg/L HA at pH 7 (transmembrane pressure 6 × 105 Pa and feed flow rate 0.4 L/min).

exhibited a faster permeate flux decline compared to the results at pH 7. This is because at acidic pH of 3, molecules of humic acid become coiled, spherical in shape and compact leading to high permeate flux decline compared to the obtained results at neutral environment (pH = 7) [15,34].

3.3. Irreversible fouling of TMBPA membranes The irreversible fouling factor (FRw) was determined as explained previously using Eq. (4). The results are plotted in Figs. 7 and 8 for the pH values 7 and 3, respectively. All modified membranes exhibit some fouling tendency at neutral pH. However, this is not significant compared to the unmodified membrane. This means that the pure water permeation flux of all modified membranes was recovered after cleaning with pure water. The highest irreversible fouling factor for the modified membranes is approximately 5% corresponding to the membrane 1TMBPA-3 compared to 80% irreversible fouling factor for the unmodified membrane, whereas the lowest irreversible fouling factor was observed for the membrane 5TMBPA-3 membrane (i.e. practically fouled membrane). At low pH value (pH 3 in Fig. 8) only the modified membranes with the lowest TMBPA monomer concentration (1TMBPA-1, 1TMBPA-3)

Fig. 6. Permeate flux decline for unmodified NFPES10 and TMBPA modified membranes tested with 15 mg/L HA at pH 3 (transmembrane pressure 6 × 105 Pa and feed flow rate 0.4 L/min).

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47

41

Fig. 9. Irreversible fouling of the unmodified NFPES10 and both the BPA and TMBPA modified membranes used for the treatment of 15 mg/L HA at pH 7 (transmembrane pressure 6 × 105 Pa and feed flow rate 0.4 L/min).

Fig. 7. Irreversible fouling of unmodified NFPES10 and TMBPA modified membranes used for the treatment of 15 mg/L HA at pH 7 (transmembrane pressure 6 × 105 Pa and feed flow rate 0.4 L/min).

exhibit lower irreversible fouling factors than the unmodified membrane. The other modified membranes are more susceptible to be fouled by humic acid under acidic environment than the unmodified membrane.

3.4. Comparison of irreversible fouling for BPA and TMBPA modified membranes The irreversible fouling factor of the unmodified membrane together with the BPA and TMBPA modified membranes at pH 7 and pH 3 are shown in Figs. 9 and 10, respectively. Within the BPA modified membranes, the lowest FRw value can be observed for the membrane 20BPA at both pH values (refer to our previous paper [26]). For the TMBPA modified membranes, the lowest FRw value was obtained for the membrane 5TMBPA-3 at neutral pH and for the membrane 1TMBPA-1 at more acidic environment (pH 3). Therefore, very high BPA concentration is required for BPA modified membranes with low fouling tendency. In fact, 20BPA membrane refers to 2.0%w/v of BPA concentration. On the contrary, for TMBPA membranes, very low concentration of TMBPA monomer is enough to produce a good

Fig. 8. Irreversible fouling of unmodified NFPES10 and TMBPA modified membranes used for the treatment of 15 mg/L HA at pH 3 (transmembrane pressure 6 × 105 Pa and feed flow rate 0.4 L/min).

membrane with better fouling tendency and higher performance. 1TMBPA refers to 0.1%w/v of TMBPA. It can be seen in Fig. 9 that some modified BPA-polyester membranes have negative irreversible fouling factors indicating an increase of the water permeate flux after humic acid experiment. This result can be attributed to the hydrophilic fraction of humic acid itself. Moreover, all TMBPA modified membranes have higher fouling tendency than the BPA membranes. For example, the membrane 5BPA shows higher fouling resistance than the membrane 5TMBPA-1 modified using the same monomer concentration and the same reaction time (i.e. the membrane 5TMBPA-1 has FRw value five times higher than that of 5BPA membrane). This may be attributed to the fact that the highly charged membrane surface could repeal the humic acid molecules reducing fouling. In addition, the more hydrophilic BPA active layer compared to TMBPA can also reduce interactions between humic acid molecules and membrane surface. Opposite trend was found when the modified membranes were tested in acidic feed solution (i.e. pH 3, Fig. 10). The BPA modified membranes, except the membrane 20BPA modified using the highest BPA concentration (2% w/v), were severely fouled even more than the unmodified membrane. In contrast, the membranes prepared with a very low concentration of TMBPA (0.1%w/v) have lower irreversible fouling factors. At higher concentration of TMBPA (0.5%w/v), the irreversible fouling factor is almost similar to the 1BPA, 5BPA and 10BPA membranes which modified with the BPA monomer concentrations of 0.1, 0.5 and 1.0%w/v, respectively. At low pH values, electrostatic repulsion between humic acid molecules and the negatively charged membrane surface may not be the dominant factor since both humic acid molecules and membrane surface lost their charges. The humic acid character (hydrophobicity, shape, size) and the physical properties of membrane surface such as roughness may influence the membrane performance [35,36]. If the hydrophobicity/

Fig. 10. Irreversible fouling of the unmodified NFPES10 and both the BPA and TMBPA modified membranes used for the treatment of 15 mg/L HA at pH 3 (transmembrane pressure 6 × 105 Pa and feed flow rate 0.4 L/min).

42

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47

hydrophilicity is considered as the only factor affecting fouling, all the BPA membranes should have the lower fouling than the unmodified NFPES10 but in reality, this is not true. This means that there are other factors may contribute to fouling behaviour. One of the common issues is membrane surface roughness. Higher degree of membrane fouling is normally related to rougher membrane surfaces even though some contradictions were revealed in some studies [37–40]. The effect of membrane surface roughness on fouling behaviour is discussed in the next section. 3.5. AFM analysis The obtained AFM images of the NFPES10 membrane and the modified membranes are presented in Fig. 11. The AFM images of each membrane sample were made in air at room temperature and at different locations on the membrane surface. Changes of modified membrane surface morphology (i.e. roughness) have been reported in several previous studies [41,42]. It can be seen in Fig. 11 that surface topography of the modified membranes is quite different from that of the unmodified membrane. Some modified membranes seem to have laterally inhomogeneous nodular structure and finely dispersed grainy structure. It is impossible to evaluate the membrane images quantitatively by direct observation. Therefore, the membrane morphology was studied in terms of surface roughness parameters which is determined by AFM technique. As the membrane roughness is one of the factors contributing to certain degree on fouling, the effect of surface roughness on irreversible fouling by humic acid molecules was investigated. The RMS roughness of the unmodified and the BPA modified membranes are presented in Fig. 12. Surface roughness may influence the degree of fouling when the size of molecules and the relative scale of the roughness are similar [43]. It is to be noted that the size of the humic acid molecule may be in the range between 1.7 nm and 3.5 nm, depending on their molecular weight 1 kDa and 10 kDa, respectively [44]. The molecular weight of the used humic acid is taken as 4.1 kDa [45], therefore the molecule size can be between 1.7 nm and 3.5 nm. However recent studies by Lee et al. (2008) [46] shows that Aldrich humic acid contains three main group of different molecular weight G1 (3 up to N 20 kDa), G2 (0.5 to 3 kDa) and G3 (b0.4 to 0.5 kDa). Based on this criteria, the humic acid size used in this study can be in the range between 1.7 nm and 3.5 nm. It must be mentioned that at different pH environments, the configuration of humic acid changes. It can be coiled, spherical in shape and compact at pH 3. In this case, the effect of roughness on fouling is more significant at this pH than at neutral environment (pH 7). For example, as can be seen in Fig. 12, there is a correlation between BPA-polyester membrane roughness and the irreversible fouling factor FRw. The marked increment in irreversible fouling FRw was found to be in the order NFPES10b20BPAb1BPAb5BPAb10BPA. It was speculated that the non-uniformity of the formed polyester layer might influence the degree of fouling. The AFM technique has already clarified this speculation. Fig. 12 proved that the higher degree of irreversible fouling found for the BPA membranes are affected by their higher roughness parameter. It is to point out that the membrane 20BPA having smoother surface than the other BPA modified membranes but slightly rougher surface than the unmodified NFPES10 still has the lowest FRw value. Therefore, as it was previously explained, there are other factors affecting humic acid fouling of the studied membranes such as the hydrophilic character of the new polyester layer that prevents adsorption of humic acid molecules on membrane surface reducing adhesion force between them. 3.6. Pore size and pore size distribution Pore size and pore size distribution of NF membranes can be characterised using non-ionic solute transport experiments [47].

From the obtained solute transport data, the pore size of NF membranes can be determined. Several common transport models including pore model (PM), steric hindrance pore model (SHPM), hybrid model (HM) and Donnan-steric partitioning model (DSPM) have been considered. In order to evaluate pore size, Wang and Chung (2005) [48] characterised NFPES10 using different molecular weight (MW) of neutral solutes covering glycerol, glucose, saccharose, raffinose and PEG1000. From the SHPM model, the pore radius of the NFPES10 membrane was calculated by correlating it to the individual neutral solutes in the range 0.69–1.45 nm with an average radius of 1.33 nm. The lowest value of 0.69 nm was obtained using PEG1000 solute transport data. Besides the neutral solutes, charged solutes especially salt are also useful to estimate the pore size of NF membranes. Bowen and Mukhtar (1996) [49] performed the rejection of single electrolytes (Na2SO4 and NaCl) using two commercial membranes NFPES5 and NFPES10 (Hoechst Separation Product, Germany). Such experimental data has been interpreted using a model based on the extended Nernst–Planck equation. It was found that the best fit value obtained for NFPES5 membrane was a pore radius of 0.72 nm. Unfortunately no information on NFPES10 pore size was reported. Ernst et al. (2000) [50], based on the rejection data of 10− 3 M NaCl of the membrane NFPES10 previously investigated by Bowen and Mukhtar (1996) [49], speculated that the pore radius of NFPES10 should be in the range between 0.7 nm and 1 nm. In other words, as the real rejection of NFPES10 is lower than NFPES5, its pore size should be greater than that of the membrane NFPES5. In other work, Bowen et al. (1997) [51] used charged (KCl, NaCl and LiCl) and neutral solutes (Vitamin B12, raffinose, sucrose, glucose and glycerine) together with AFM technique to characterise NFPES5 membrane. The result revealed that the average pore size obtained from neutral solutes is smaller than the result obtained by fitting the obtained data using salt tests. It is worth quoting that it was found that mean pore radius determined by AFM technique is smaller than that obtained from the analysis of both the uncharged and salt rejection data. This is most probably due to the AFM tip convolution effect. In this present study, instead of using different neutral solutes with different molecular weights, the membranes were characterised by a series of the same neutral solute polyethylene glycol (PEG) but with different molecular weights (200, 600, 1000 and 3350 g/mol). This technique has been applied previously to determine the molecular weight cut off (MWCO), mean pore size and pore size distribution of ultrafiltration (UF) membranes [17,20,52]. This technique has also been used to characterise NF membranes [17,53,54]. In the present study, the solute separation of both the unmodified and the modified membranes was plotted in Figs. 13–15 as a function of the PEG Einstein–Stokes diameters (i.e. membrane pore diameter) on a lognormal probability. Straight lines could be fitted with reasonably high correlation coefficients. The mean pore size (μp) and the geometric standard deviation (σp) were determined from the obtained straight lines. From the results plotted in Fig. 13 corresponding to the unmodified NFPES10 membrane, the obtained mean pore size is 1.47 nm and the geometric standard deviation is 1.73. The determined pore size of the NFPES10 membrane is not far from the value estimated by Bowen and Mohammad (1998) [25] for the same membrane (1.5 nm). This is also consistent with the value reported by Wang and Chung (2005) [48] determined from the SHP model using PEG1000 data (1.38 nm). This indicates that the pore diameter estimated by PEG solute transport method followed in this study is adequate. It is worth quoting that Al-Zoubi (2007) [55] characterised the same membrane NFPES10 by AFM analysis and found a smaller mean pore diameter of approximately 0.4 nm. This issue has been pointed out by Bowen et al. (1997) [51] as follows. AFM can only give information of the surface pore dimensions as the tip cannot probe into the depth of pore. For small pores, the dimensions provided by

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47

AFM should be treated with caution due to the convolution between the tip and pore shapes and one may expect in certain case, AFM might underestimate the pore dimensions. As can be seen in Fig. 16, most of the pores of the membrane NFPES10 are concentrated in the range 1 nm–1.5 nm and even larger pore sizes (3 nm–5 nm) exist but their numbers are too small.

43

Tables 2 and 3 summarise the values of μp and σp of the BPA and TMBPA modified membranes determined from the data presented in Figs. 14 and 15, respectively. The data show that all modified membranes posses smaller pore size than the unmodified one and the smallest pore size, 0.8 nm (more than 45% reduction of unmodified NFPES10 membrane), corresponds to the membranes

Fig. 11. 3D AFM images of the unmodified NFPES10 and the BPA-polyester membrane: (a) unmodified, (b) 1BPA , (c) 5BPA, (d) 10BPA and (e) 20 BPA.

44

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47

Fig. 11 (continued).

1TMBPA-1 and 5TMBPA-3. This reduction in pore size can possibly be due to the rigid and dense polyester layer formed by TMBPA reaction as discussed in detail in Section 3.1. No clear trends were detected between the pore sizes and the monomers concentrations. It seems for BPA modified membranes that the pore size decreases with the increase of the monomer concentration. As the monomer increases, the kinetic reaction rate become faster and generated thicker poly-

ester layer on the pore wall thus reduce the size of the pore. Previous studies carried out by Mohammad et al. (2003) [9] showed a reduction of the membrane pore size determined by AFM with the increase of the monomer concentration. It can be seen from Figs. 16 and 17 that the pore size distribution curves are shifted to the left for both BPA and TMBPA modified membranes showing reduction of the number of larger pores. This might be one of the reasons of the

Fig. 12. Membrane roughness and irreversible fouling factor of humic acid (at pH 3) for the unmodified NFPES10 and the BPA-polyester membranes.

Fig. 13. PEG solute separation vs. pore size of the unmodified NFPES10 membrane.

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47

45

Table 2 Mean pore size, (μ p) and geometric standard deviation (σp) of the BPA modified membranes. Membrane

μ p (nm)

σp

1BPA 5BPA 10BPA 20BPA

1.34 1.08 1.16 1.00

1.73 1.85 1.80 2.14

Table 3 Mean pore size, (μ p) and geometric standard deviation (σp) of the TMBPA modified membranes.

Fig. 14. PEG solute separation vs. pore size of the BPA modified membranes.

Membrane

μ p (nm)

σp

1TMBPA-1 1TMBPA-3 5TMBPA-1 5TMBPA-3

0.8 1.32 1.00 0.8

1.55 2.32 2.09 1.74

Fig. 15. PEG solute separation vs. pore size of the TMBPA modified membranes.

observed decrease in permeability suggesting that the formed polyester layers are constricting, blocking pores or shifting the pore size to smaller sizes.

Fig. 17. Probability density function curves of the unmodified NFPES10 and the TMBPA modified membranes.

Although the modification changes the membrane pore size and its distribution with pore sizes smaller compared to the unmodified NFPES10 membrane in magnitude of almost up to 50% reduction, these changes have no significant affect on humic acid removal, which was maintained higher than 97% rejection at both pH values. This is attributed to the larger mean humic acid molecular size (1.7–3.5 nm) compared to the pore size of the membranes (0.8 nm–1.32 nm).

4. Conclusions

Fig. 16. Probability density function curves of the unmodified NFPES10 and the BPA modified membranes.

Thin-film composite polyester nanofiltration (NF) membranes based on two different bisphenol monomers (BPA and TMBPA) were successfully prepared by interfacial polymerisation technique. At neutral environment, the BPA modified membranes exhibited practically no irreversible fouling compared to TMBPA-polyester and the unmodified membranes. For some modified BPA-polyester membranes, the irreversible fouling factor was found to be negative indicating an increase of the water permeate flux after humic acid experiment. This result was attributed to the hydrophilic fraction of humic acid itself.

46

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47

Most of the membranes were found to be more susceptible to irreversible fouling by humic acid components at more acidic environment (pH 3) and under this condition the membrane modified with high BPA monomer concentration (2% w/v) during 10 s interfacial polymerisation time exhibited the lowest fouling factor. For TMBPA series, under pH 7 of humic acid solution, the membrane modified with high TMBPA monomer concentration (0.5% w/v) and longer reaction time (30 s) has very low degree of fouling while lower concentration (0.1%w/v) and shorter reaction time (10 s) is enough for applications in acidic environment of humic acid solution. When considering chemical properties, the membrane 20BPA was found to be the best candidate since it has a very low fouling tendency in both neutral (pH 7) and acidic (pH 3) humic acid solutions. The AFM analysis shows that membrane modification by interfacial polymerisation significantly changes the membrane surface morphology reducing the pore size of the membrane NFPES10. The effect of roughness is found to be significant at pH 3 since in this case the membrane charge is not a dominant factor and the humic acid structure and its configuration change due to intermolecular interaction affecting the degree of fouling. At this pH, the smoother membrane surface is required in order to reduce the irreversible fouling. It was observed that the changes of pore size did not affect considerably the humic acid removal. Acknowledgement We would like to thank the Ministry of Higher Education, Malaysia for providing Mazrul Abu Seman with a scholarship to carry out this work. References [1] A.L. Ahmad, B.S. Ooi, A. Wahab Mohammad, J.P. Choudhury, Development of a highly hydrophilic nanofiltration membrane for desalination and water treatment, Desalination 168 (2004) 215–221. [2] H.S. Lee, S.J. Im, J.H. Kim, H.J. Kim, J.P. Kim, B.R. Min, Polyamide thin-film nanofiltration membranes containing TiO2 nanoparticles, Desalination 219 (2008) 48–56. [3] Y. Zhou, S. Yu, M. Liu, H. Chen, C. Gao, Effect of mixed crosslinking agents on performance of thin-film-composite membranes, Desalination 192 (2006) 182–189. [4] A.L. Ahmad, B.S. Ooi, J.P. Choudhury, Preparation and characterization of copolyamide thin film composite membrane from piperazine and 3, 5-diaminobenzoic acid, Desalination 158 (2003) 101–108. [5] S. Veríssimo, K.-V. Peinemann, J. Bardado, New composite hollow fiber membrane for nanofiltration, Desalination 184 (2005) 1–11. [6] L.Y. Chu, S. Wang, W.M. Chen, Surface modification of ceramic-supported polyethersulfone membranes by interfacial polymerization for reduced membrane fouling, Macromol. Chem. Phys. 207 (2005) 1934–1940. [7] G. Kang, M. Liu, B. Lin, Y. Cao, Q.A. Yuan, Novel method of surface modification on thin film composite reverse osmosis membrane by grafting poly(ethylene glycol), Polymer 48 (2007) 1165–1170. [8] S.Y. Kwak, M.O. Yeom, I.J. Roh, D.Y. Kim, J.J. Kim, Correlations of chemical structure, atomic force microscopy (AFM) morphology and reverse osmosis (RO) characteristics in aromatic polyester high-flux RO membranes, J. Membr. Sci. 132 (1997) 183–191. [9] A.W. Mohammad, N. Hilal, M.N. Abu Seman, A study on producing composite nanofiltration membranes with optimized properties, Desalination 158 (2003) 73–78. [10] B. Tang, Z. Huo, P. Wu, Study on a novel polyester composite nanofiltration membrane by interfacial polymerization of triethanolamine (TEOA) and trimesoyl chloride (TMC) I. Preparation, characterization and nanofiltration properties test of membrane, J. Membr. Sci. 320 (2008) 198–205. [11] U. Razdan, S.S. Kulkarni, Nanofiltration thin-film-composite polyesteramide membranes based on bulky diols, Desalination 161 (2004) 25–32. [12] M.M. Jayarani, S.S. Kulkarni, Thin-film composite poly(esteramide)-based membranes, Desalination 130 (2000) 17–30. [13] N. Hilal, W.R. Bowen, L. Alkhatib, O. Ogunbiyi, A review of atomic microscopy applied to cell interactions with membrane, Trans. IChemE A Chem. Eng. Res. Des. 84 (A4) (2006) 282–292. [14] W.R. Bowen, N. Hilal, R.W. Lovitt, C.J. Wright, Characterisation of membrane surfaces: direct measurement of biological adhesion using an atomic force microscope, J. Membr. Sci. 154 (1999) 205–212. [15] M. Mänttäri, L. Puro, J. Nuortila-Jokinen, M. Nyström, Fouling effects of polysaccharides and humic acid in nanofiltration, J. Membr. Sci. 165 (2000) 1–17. [16] M. Mänttäri, M. Nystrom, Critical flux in NF of high molar mass polysaccharides and effluents from the paper industry, J. Membr. Sci. 170 (2000) 257–273.

[17] C.T. Cleveland, T.F. Seacord, A.K. Zander, Standardized membrane pore size characterization by polyethylene glycol rejection, J. Environ. Eng. 128 (2002) 399–407. [18] T.A. Tweddle, A. Baxter, O. Kutowy, ‘Equipment and procedures for testing small coupon samples of RO and UF membranes, Special National Research Council of Canada Rep. No.C-1144-875, Wikimedia Foundation, Inc., Ottawa, 1987. [19] S. Singh, K.C. Khulbe, T. Matsuura, P. Ramamurthy, Membrane characterization by solute transport and atomic force microscopy, J. Membr. Sci. 142 (1998) 117–127. [20] M. Khayet, T. Matsuura, Determination of surface and bulk pore sizes of flat-sheet and hollow-fiber membranes by atomic force microscopy, gas permeation and solute transport methods, Desalination 158 (2003) 57–64. [21] M. Khayet, C.Y. Feng, K.C. Khulbe, T. Matsuura, Preparation & characterization of polyvinylidene fluoride hollow fiber membranes for ultrafiltration, Polymer 43 (2002) 3879–3890. [22] M. Ishiguro, T. Matsuura, C.A. Detellier, Study on the solute separation and the pore size distribution of a montmorillonite, Membr. Sep. Sci. Technol. 31 (1996) 545–556. [23] A.S. Michaels, Analysis and prediction of sieving curves for ultrafiltration membranes: a universal correlation? Sep. Sci. Technol. 15 (6) (1980) 1305–1322. [24] A.R. Cooper, D.S. Van Derveer, Characterization of ultrafiltration membranes by polymer transport measurements, Sep. Sci. Tech. 14 (1979) 551–556. [25] W.R. Bowen, A.W. Mohammad, Characterization and prediction of nanofiltration membrane performance – a general assessment, IChemE 76 (1998) 885–893 Part A. [26] M.N. Abu Seman, M. Khayet, N. Hilal, Nanofiltration thin-film composite polyester polyethersulfone-based membranes prepared by interfacial polymerization, J. Membr. Sci. 348 (2010) 109–116. [27] J. Ji, M. Mehta, Mathematical model for the formation of thin-film composite hollow fiber and tubular membranes by interfacial polymerization, J. Membr. Sci. 192 (2001) 41–54. [28] S. Veríssimo, K.V. Peinemann, J. Bordado, Influence of the diamine structure on the nanofiltration performance, surface morphology and surface charge of the composite polyamide membranes, J. Membr. Sci. 279 (2006) 266–275. [29] http://en.wikipedia.org/wiki/Methyl_group. [30] F.A. Ruiz-Trevino, D.R. Paul, Modification of polysulfone gas separation membranes by additives, J. Appl. Polym. Sci. 66 (1997) 1925–1941. [31] J.Y. Jho, Cooperative Molecular motion on Bisphenol-A Polycarbonate (Molecular Motion). (1990) PhD Thesis. The University of Michigan. [32] S. Kim, J. Liu, Molecular modelling of bisphenol-a polycarbonate and tetramethyl bisphenol-a polycarbonate, Korea Polym. J. 9 (2001) 129–142. [33] M.L. Chng, Y. Xiao, T.S. Chung, M. Toriida, S. Tamai, The effects of chemical structure on gas transport properties of poly(aryl ether ketone) random copolymers, Polymer 48 (2007) 311–317. [34] S. Hong, M. Elimelech, Chemical and physical aspects of natural organic matter (NOM) fouling of nanofiltration membranes, J. Membr. Sci. 132 (1997) 159–181. [35] M. Khayet, T. Matsuura, Surface modification of membranes for the separation of volatile organic compounds from water by pervaporation, Desalination 148 (2002) 31–37. [36] M. Khayet, A. Velázquez, J.I. Mengual, Direct contact membrane distillation of humic acid solutions, J. Membr. Sci. 240 (2004) 123–128. [37] C.Y. Tang, Q.S. Fu, C.S. Criddle, J.O. Leckie, Effect of flux transmembrane pressure and membrane properties on fouling and rejection of reverse osmosis and nanofiltration membranes treating perfluorooctane sulfonate containing wastewater, Environ. Sci. Technol. 41 (2007) 2008–2014. [38] S. Kang, E.M.V. Hoek, H. Choi, H. Shin Effect, Membrane surface properties during the fast evaluation of cell attachment, Sep. Sci. Tech. 41 (2006) 1475–1487. [39] E.M.V. Hoek, S. Bhattacharjee, M. Elimelech, Effect of membrane surface roughness on colloid-membrane DLVO interactions, Langmuir 19 (2003) 4836–4847. [40] W.R. Bowen, T.A. Doneva, H.-B. Yin, Separation of humic acid from a model surface water with PSU/SPEEK blend UF/NF membranes, J. Membr. Sci. 206 (2002) 417–429. [41] M. Khayet, Membrane surface modification and characterization by X-ray photoelectron spectroscopy, atomic force microscopy and contact angle measurements, Appl. Surf. Sci. 238 (1–4) (2004) 269–272. [42] N. Hilal, L. Al-Khatib, H. Al-Zoubi, R. Nigmatullin Atomic, Force microscopy study of membranes modified by surface grafting of cationic polyelectrolyte, Desalination 184 (2005) 45–55. [43] J.E. Kilduff, S. Mattaraj, M. Zhou, G. Belfort, Kinetics of membrane flux decline: the role of natural colloids and mitigation via membrane surface modification, J. Nanopart. Res. 7 (2005) 525–544. [44] A.I. Schäfer, A.G. Fane, T.D. Waite Nanofiltration, of natural organic matter: removal, fouling and influence of multivalent ions, Desalination 118 (1998) 109–122. [45] Y.P. Chin, G. Aiken, E. O'Loughlin, Molecular weight, polydispersity and spectroscopic properties of aquatic humic substances, Environ. Sci. Technol. 28 (1994) 1853–1858. [46] H.-H. Lee, Y.-H. Weng, K.C. Li, Electro-ultrafiltration study on Aldrich humic substances with different molecular weights, Sep. Pur. Tech. 63 (2008) 23–29. [47] S. Nakao, Determination of pore size and pore size distribution: 3. Filtration membranes, J. Membr. Sci. 96 (1994) 131–165. [48] K.Y. Wang, T.S. Chung, The characterization of flat composite nanofiltration membranes and their applications in the separation of Cephalexin, J. Membr. Sci. 247 (2005) 37–50. [49] W.R. Bowen, H. Mukhtar, Characterisation and prediction of separation performance of nanofiltration membranes, J. Membr. Sci. 112 (1996) 263–274. [50] M. Ernst, A. Bismarck, J. Springer, M. Jekel, Zeta-potential and rejection rates of a polyethersulfone nanofiltration membrane in single salt solutions, J. Membr. Sci. 165 (2000) 251–259.

M.N. Abu Seman et al. / Desalination 273 (2011) 36–47 [51] W.R. Bowen, A.W. Mohammad, N. Hilal, Characterisation of nanofiltration membrane for predictive purposes – use of salts, uncharged solutes and atomic force microscopy, J. Membr. Sci. 126 (1997) 91–105. [52] N.A. Ochoa, P. Pradanos, L. Palacia, C. Pagliero, J. Marchese, A. Hernandez, Pore size distributions based on AFM imaging and retention of multidisperse polymer solutes: charactersation of polyethersulfone UF membranes with dopes containing different PVP, J. Membr. Sci. 187 (2001) 227–237. [53] N. Ali, N.S. Suhaimi, Performance evaluation of locally fabricated asymmetric nanofiltration membrane for batik industry effluent, World Appl. Sci. J. 5 (2009) 46–52 (Special Issue for Environment).

47

[54] A.R. Hassan, A.F. Ismail, Characterization of nanofiltration membranes by the solute transport method: Some practical aspects in determining of mean pore size and pore size distributions. Regional Symposium on Membrane Science and Technology 2004, 21–25 April 2004, Puteri Pan Pacific Hotel, Johor Baru, Johor, Malaysia. (2004). [55] H. Al Zoubi, Development of novel approach to the prediction of nanofiltration membrane performance using advanced atomic force microscopy, Ph.D Dissertation, University of Nottingham, UK, 2006.