16

Click here to load reader

Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

Embed Size (px)

Citation preview

Page 1: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

E L S E V I E R Desalination 173 (2005) 223-238

DESALINATION

www.elsevier.com/locate/desal

Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

D. Violleau a'b, H. Essis-Tome c, H. Habarou b'd, J.P. Croud b*, M. Pontid a'c* aUniversit~ d'Angers, UMR.MA Paysages et Biodiversit~,

Laboratoire des Sciences de l'Environneraent et de l'Ara~nagement, 2 Bid Lavoisier, 49045 Angers Cedex, France bEcole Sup~rieure d'Ing~nieurs de Poitiers, Laboratoire de Chimie de l'Eau et de l'Environnement,

UMR CNRS No. 6008, 40 Ave de Recteur Pimeau, 86022 Poitiers Cedex, France TeL +33 (0) 549453924; Fax +33 (0) 549453768; e-mail: [email protected]

CEcole Nationale Supdrieure de Chimie de Paris, Laboratoire d'Electrochimie et de Chimie Analytique, UMR CNRS No. 7575, A CI JC No. 4052 (2002), 11 rue Pierre et Marie Curie, F- 75231 Paris Cedex 05, France

Tel. +33 (1) 241735207; Fax +33 (1) 241735352; e-mail: [email protected] dVEOLIA WATER, Anjou-Recherche, Chemin de la Digue, 78603 Maisons-Laffitte, France

Received 20 February 2004; accepted 29 July 2004

Abstract

The present study compares the fouling properties of two selected and well-characterized hydrophobic (denoted HPOA) and hydrophilic (denoted TPIA) natural organic matters (NOM) fractions with a commercial polyamide (PA) nanofiltration (NF) membrane (denoted NF-55). Analytical tools such as elemental analyses using microanalyses, specific UV absorbance, solid-state cross-polarization magic angle spinning (CPMAS) 13C-NMR, high pressure size exclusion (HPSEC) chromatography, HPLC system and acid/base titration for characterizing NOM fractions and contact angle, hydraulic permeability, streaming potential (SP) and observed rejection of a NaC1 aqueous solution to evaluate the affinity of the PA material with both selected NOM fractions isolated from the Blavet river (Britany region, France). The PA material was found to be more sensi- tive to hydrophobic NOM adsorption leading to irreversible fouling with drastic modifications of the initial physico-chemical properties of the membrane: (a) increase of its hydrophobicity; (b) decrease of its hydraulic permeability associated with a decrease in its pore size and consequently (c) increase the observed rejection of salty solutions. The higher decrease in the performances of this PA NF membrane is observed for the more hy- drophobic foulant, HPOA. At the same time a displacement of the isolectric point (IEP) of the membrane material was observed from 4.5 for the clean membrane (KC1 10 -4 mol.L -1) to 3.4 after HPOA sorption. At a lower pH range than the IEP, the effects of cations and I-I + on the charge properties of the membranes increases near the

* Corresponding authors

0011-9164/05/$- See front matter © 2005 Elsevier B.V. All rights reserved doi: 10.1016/j .desal.2004.07.048

Page 2: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

224 D. Violleau et al. / Desalination 173 (2005) 223-238

shear plane, yielding more positive SP values. For the hydrophilic TPIA foulant no displacement of the IEP was observed. Then the results of SP experiments conducted through the membrane with a homemade apparatus has indicated that HPOA is more retained inside the pores as compared to the TPIA that was mainly sorbed at the surface of the membrane. Furthermore the membranes acidic-basis properties were amplified after foulants deposit in comparison to the cleaned membrane where a dominant specific sorption of monovalents and diva- lents ions occurred.

Keywords: Nanofiltration; Natural organic matter; Fouling ~3C-NMR; Streaming potential; Wettability

1. Introduction

Successful utilization of membrane techno- logy has been greatly limited by membrane fouling. Then fouling phenomenon increases operation and maintenance costs by deteri- orating membrane performances (flux decline vs. time, zeta potential changing during time, etc.) and ultimately shortening membrane life. Natural organic matter (NOM) plays an import- ant role in membrane fouling. Today's nanofil- tration (NF) industrial plants are just starting to grow to a large scale and long-term ageing studies haven't been conducted. The permeate flux decline, generally well observed in ultra- filtration (UF) and microfiltration (MF) of natural water resources, depends strongly on the adsorption of chemical substances dissolved un- der the filtration material. The membrane pro- cesses performances in terms of flux are strongly correlated to the raw water quality. Accumulation of material on or in the mem- brane skin leads to a reduction of the water flux through the membrane. In the literature this phenomenon is referred to as membrane foul- ing, [1-5]. For MF/UF membranes this fouling is partially reversible by applying a backwash that removes part of the accumulated matters. After being backwashed, the membrane regains partially its initial hydraulic permeability de- pending on the nature of the foulants. The fre- quency and length of the backwashes are dic- tated by the raw water quality and the nature of the membrane material and both need to be increased when the water quality deteriorates.

Unfortunately, in the NF process, pure water backwashing operations cannot be applied.

Natural organic matter (NOM) fouling in- creases operation and maintenance costs by de- teriorating membrane performance (flux decline vs. time) and ultimately shortening membrane life. Several studies have demonstrated that humics and non-humics, polysaccharides and proteins are the dominant foulants.

The goal of our work was to study the foul- ing properties of a humic (i.e. hydrophobic, denoted HPOA) and non-humic (i.e. hydro- philic, denoted TPIA) NOM fractions isolated from the Blavet river toward a commercial polyamide (PA) NF membrane (denoted NF-55). The comparison of the affinity of both selected NOM fractions with the PA NF mem- brane was conducted using contact angle, hy- draulic permeability, streaming potential (SP) and NaC1 aqueous solution permeation mea- surements. Our study is dedicated to the optimi- sation of NF process performances by limiting its rapid ageing due to NOM deposit in and/or on the membrane material.

2. Experiments

2.1. Membranes and modules

The main membrane under study is a thin film composite membrane built up of two layers, a thin PA film denoted active surface and a large mesoporous polysulfone denoted support layer, as illustrated in Fig. 1 and de- tailed elsewhere [6]. The membrane was re- ceived as a flat sheet. It was purchased from Filmtec (Dow, Denmark) and its commercial- ised name is NF-55. The chemical structures of the support and active layers materials are reported in Fig. 2. Mean pores radii is 0.65 nm,

Page 3: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

ultrathin barrier layer in polyamide

polysulfone

reinforcing fabric in polysulfone

D. Violleau et al. / Desalination 173 (2005) 223-238

0,3 -3 pm

40 IJm

120 Nrn

Fig. 1. Schematic diagram of the thin film NF-55 com- posite membrane, adapted from R.J. Petersen [6].

as previously detailed [7]. Before all experi- ments the membrane was cleaned by means of standard procedures to remove preservatives and rinsed with ultra-pure (UP) from a system denoted MilliQ water (Millipore, France) until the conductivity of the permeate remained be- low 1 ixS/cm. The effective membrane surface area was 138 cm 2. A UF membrane in polyether-

225

sulfone, purchased from Millipore with a mole- cular weight cut-off (MWCO) of 100 kDa, is also used in order to compare the SP measure- ments of UF and NF membrane vs. ionic strength variations.

2.2. Isolation and characterization of the NOM fractions

Fractions of NOM were isolated from a high DOC surface water, the Blavet river (DOC=12 mg/L) sampled in December 1995, 500 m downstream of the Keme Uhel reser- voir, near the water treatment plant of Lanrivin (Cfte d'Armor, Brittany region, France). The humic NOM also referred to as hydrophobic acid NOM fraction (i.e. HPOA NOM) and the non-humic NOM referred to as transphilic acid NOM fraction (i.e. TPIA NOM) were isolated using the XAD-8 and XAD-4 resins in series, following the approach ofCrou6 et al. [8].

CH 3 O - A -

o--1- n

/ \I NH N ~ N H - - C O ~ ~ / C O

\ / n

\

N H x ~ N H - - C O . c o ¢~ ~',~

COOH

/ l -n

- B -

Fig. 2. Chemical structures of the support (A) and active layers (B) of the NF-55 composite membrane.

Page 4: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

226 D. Violleau et al. / Desalination 173 (2005) 223-238

The NOM fractions were characterized using a large diversity of analytical tools. Among the different ways of characterizing NOM isolates elemental analysis is generally the first ap- proach that researchers use [9]. Elemental ana- lyses were carried out in the laboratory of the Centre National de la Recherch6 Scientifique (CNRS) in France. Microanalyses developed at the CNRS laboratory allowed the determination of the C, H, O, N, S contents (in %) with an ab- solute precision of 0.3% and a standard devi- ation of 0.2%. Specific UV absorbance (SUVA254 or simply SUVA) is defined as the UV absorbance of a given sample determined at 254 nm and divided by the DOC concentration of the solution. It is expressed in units of m -x L/mg C. UV absorbance at 254 nm was determined using a "Safas" spectrophotometer "Double energy system 190 DES" with 1 or 5 cm long quartz cells. DOC analyses were carried out using a "Dohrmann DC 80" carbon analyser. 13C-NMR analysis provides both NOM organic structure information and carbon-con- mining functional group information. Solid-state cross-polarization magic angle spinning (CPMAS) ~3C-NMR spectra were obtained from the United States Geological Survey labo- ratory at Arvada, Colorado, USA, using a 200 MHz Chemagnetics CMX spectrometer with a 7.5 mm diam probe. The spinning rate was 5000 Hz. The acquisition parameters for the freeze-dfied samples included a contact time of I ms, pulse delay of 1 s, and a pulse width of 4.5 bts for the 90 ° pulse. Molecular weight of the NOM fractions was determined using high pressure size exclusion (HPSEC) chromato- graphy using the same analytical procedure as the one described by Chin et al. [10]. Note that these authors showed that the results obtained with this method were in good agreement with the ones determined by vapor pressure osmo- metry (VPO, a colligative property measure- ment) and field flow fractionation. The HPLC system included a WATERS 717 plus auto-

sampler, a WATERS 600 pump and a WATERS 996 Diode Array detector. Solutions of NOM fractions (50 mg C/L) were prepared with the mobile phase and injected on a Pro- teinPak 125 column (ID: 7.8 mm; length: 30 cm) using a 20 o.L loop. The molecular weight calibration was operated usingpolystyrene sul- fonates (8, 5.4 and 1.8 kDa), terephtalic acid and acetone (linear regression R2= 0.998). Weight-average molecular weight (Mw) and number-average molecular weight (Mn) were determined using the Millenium Waters soft- ware (2010 version). Carboxylic and OH phen- olic functional group contents were determined using the acid/base titration protocol proposed by Thurman [ 11 ].

2.3. Nano f i l t ra t ion runs

NF runs were performed with 5 mg TOC/L NOM solutions of both fractions, isolated from the Blavet fiver, at pH 5.8-6 (the pH was ad- justed with a few drops of diluted solutions of sodium hydroxide) on rigs (Osmonics, USA) equipped with a gears pump and a planar mem- brane design with a tangential flow velocity of 1 m.s -~ (Re = 6200) and a constant flow rate of 10% minimizing the concentration polarization effects. The fouling experiments were conduc- ted during 6 h at a pressure of 8 bars. At the end the membranes were rinsed copiously with UP water until the conductivities of the permeate and concentrate solutions were stabilized.

The applied transmembrane pressures varied in the range of 0-9 MPa. Permeated solutions were recycled during the runs except for samples withdrawn for the calculation of observed rejection (Robs) according to:

Rot~. = 1 - Cp/Co (1)

where Cp and Co are the solute concentrations in the permeate and in the bulk solutions respectively. The low range of transmembrane pressure used permits us to assimilate the ob- served rejection with the real rejection due to

Page 5: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

D. Violleau et aL / Desalination 173 (2005) 223-238

the high flow rate (around 1 m/s). The ions were assayed with specific electrodes from Radio- meter-Analytical (France) calibrated for the chloride ions coupled with an Ag/AgC1 reference electrode.

2.4. SP measurements

SP measurements occurred using a home- made tool easy to handle, as previously des- cribed [12,13]. A pair of silver-chloride elec- trodes from Radiometer-Analytical (France) judiciously introduced in the SP device, was used to measure the potential difference between both sides of the membrane vs. the transmem- brane pressure. The electric contact between the electrode and the retentate solution was estab- lished via a glass tank full of electrolyte solu- tion isolated from the bulk solution by a glass vycor. The electrodes were connected to a high impedance voltmeter. Thus we have to detail that using our properly SP measurements across NF55 membranes induces some theoretical problems due to the evaluation of real zeta po- tential from SP measurements across the mem- brane, as detailed elsewhere [4,7]. But our SP design apparatus permits only the characteri- zation of the active layer of the membrane, as described elsewhere [14,15]. Furthermore we have to note that in the current study we were only interested in the qualification of the mem- brane charge and not its quantification. The main interest in our homemade apparatus is to be able to distinguish between pore walls block- age and/or surface modifications of the mem- brane.

2.5. Contact angle measurements

The contact angle measurements were car- ried out by the sessile drop technique: a liquid droplet of 1-2 I.tl in volume placed onto a fiat homogeneous surface. The contact angle of the droplet with the surface is measured with an apparatus purchased from KRUSS G1 (Deutsch- land) and the contact angle measurements are obtained under the goniometer. The contact

227

angle measurements were carried out on dried films of flat sheet membrane samples (5x5 cm2). The different membranes were primarily washed twice with deionised water for a period of 15 min and then dried at room temperature in a closed tank. After drying the films over silicagel in a dessicator, a droplet of UP water solution was deposited onto the mem- brane surface by means of a microsyringe. Before being deposited the pH of the droplets was adjusted with the addition of little amounts of HCI and NaOH. Reported values are the averages contact angle (right and left) of 10 droplets. During the short measurement time (less than 1 min), no change in contact angle was observed. Contact angles do not give abso- lute values but allow a comparison between each material. All the membranes tested have a sufficient low mean roughness (around 20 nm using A F M - Autoprobe CP100, Park Scientific Instruments, with a window study of 8x8 ~tm 2, as reported elsewhere [15]) to neglect its effect under contact angle measurements. A variation of 2 ° in the angle is needed to differentiate the different samples analyzed.

2.6. Chemicals

All the salts used (KCI, NaOH, HC1) were purchased from Aldrich (France) with an ana- lytical grade. All solutions were prepared from a MilliQ water design (Millipore system, USA) with a high purity water presenting a conduc- tivity lower than 1 p.S/cm.

3. Results and discussions

3.1. Characterization o f the Blavet River NOM fractions

Table 1 gives some structural characteristics of the NOM fractions isolated from the Blavet river. The elemental analysis indicates that the HPOA and TPIA fractions are almost pure NOM with a very low ash content. As observed with other source waters [16], the HPOA frac- tion shows higher C/O (more oxygenated func-

Page 6: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

228 D. Violleau et al. / Desalination 173 (2005) 223-238

Table 1 Structural characteristics of the Blavet River NOM isolates

Fraction Elemental analysis, % Acid base titration, meq/g Molecular weight, UV254/TOC, of TOC daltons m -~.L/mg C

C H N O Ashes COOH OH phenolic Mw Mn SUVA

HPOA 47.0 4.5 2.0 38.8 6.2 8.8 4.1 1570 900 4.3

TPIA 43.3 4.6 2.9 40.6 2.7 11.7 4.3 1040 700 2.6

tional groups), C/N (more proteinaceous-type structures), and C/H (more unsatured carbon bonds) ratios as compared to the TPIA fraction. The HPOA fraction also incorporates a larger proportion of high molecular weight (Mw) structures (i.e. higher Mw determined by HPSEC/UV254 analysis) and a larger proportion of UV absorbing moieties (i.e. aromatic moie- ties) as indicated by its higher SUVA value (i.e. Specific UV Absorbance, UV absorbance at 254 nm divided by DOC).

The ~3C-NMR spectra of the NOM isolates are presented in Fig. 3. The integrated areas under various portions of the ~3C-NMR spectra curves can be assigned to specific features. Table 2 gives the assignments for the Blavet river NOM fractions. The results confirmed that the HPOA fraction is more aromatic in nature than the TPIA fraction (i.e. higher relative pro- portion of aromatic carbon). The two small peaks on each side of the aromatic carbon band of the HPOA fraction (Fig. 3) are indicators of the presence of phenolic structures. As indi- cated by the results of potentiometric titration and in accordance with the elemental analysis (i.e. C/O ratio), the TPLA_ fraction is more enriched in COOH functional groups, in C-O bond (i.e. aliphatic alcohols, ethers and esters) and anomeric carbon (i.e. sugars, 90-110 ppm chemical shift) as compared to the HPOA frac- tion. Consequently the HPOA fraction contains a larger proportion of aliphatic C-C and C-H.

All these structural findings confirmed that the HPOA fraction has a stronger hydrophobic character than the TPIA fraction.

R

I • • • • I • • • • I "

200 I00 0 ppm

2

Fig. 3. ~3C-NMR spectra of the Blavet river NOM isolates, 1-HPOA 2-TPIA.

3.2. Characterization o f the initial NF-55 mem- brane

3.2.1. SP results

SP measurements were conducted on a clean NF-55 membrane after removal of the preserv- atives, as mentioned above. These experiments

Page 7: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

D. Violleau et aL / Desalination 173 (2005) 223-238 229

Table 2 Integrated areas of 13C-NMR spectra of the Blavet River NOM fractions

Fraction Chemical shift, ppm

0--60 60-90 90--110 110-160 160-190 190-220 C-C/C-H C-O O-C-O Aromatic C COOH C=O

Relative area, %

I-IPOA 39 13 4 28 13 3

TPIA 38 24 7 13 16 2

2 4 6 8

0

-6

-8

-10

10 I

Fig. 4. SP measurements, Ad~ vs. the transmembrane pressure, AP for the NF-55 membrane, KC1 = 10 -4 mol.L -1, pH = 6.8, adapted from M. Ponti6 [151.

were carried out in order to determine the iso- electric point (IEP) of the membrane, as detailed elsewhere [ 17]. As shown in the Fig. 4, a good linearity was observed between the membrane potential difference (Adp) and the transmem- brane pressure (AP), in good agreement with the following Helmotz-Smoluchowski equation:

A /AP =

where ~ is the zeta potential of the membrane material (V) and e the permitivity (CV -1 rn -1 ), !~

the viscosity (Pa.s) and Z the conductivity (Sm -~) of the electrolyte solution.

The use of this equation is restricted to a high ionic strength ( r -1 small) in large capil- laries (re large), as reported recently [14,18]. In the case of microporous membranes the slope obtained is sufficient to qualify the charge modifications of the pore walls before and after NOM deposition and to determine the isoelec- tfic point of the membrane material. It was found that the polyamide membrane pore walls exert a negative charge in presence of a KC1 electrolyte at a concentration of 0.0001 mol.L -1 and for a neutral pH.

However, as proposed in Fig. 5a and 5b, SP results indicated that A~p/AP is changing with pH and ionic strength. At the isoelectric point (IEP) SP values are equal to zero. However the IEP increased with the concentration of the electrolyte solution of KC1 from 4.4 to 5.8 with 10 -4 and 10 -3 mol.L -~, respectively. This ob- servation relates a specific adsorption of the KC1 ions on the PA material, as similarly reported elsewhere [7,14]. Furthermore at a specific ionic strength we have observed that the IEP depends on the nature of the electrolyte solution, as fully reported elsewhere [7,18-22]. Using a 10 -4 mol.L -~ ionic strength electrolyte solution, SP results reported show a shift of the IEP from 4.4 for KCI to 5.5 for NaC1 and 6.5 for CaCI2 (Fig. 5b). Indeed when a polymer is im- mersed in an electrolyte solution, some ions can be adsorbed at the interface between membrane

Page 8: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

230

(a) 1.5

1

0.5

-= 0 ot~

-0.5

-t.5

D. Violleau et al. i Desalination 173 (2005) 223-238

El 10 "~ mole/L , ~ ~ 0 10 4 mole/L

pH i=5.8

pH i=4.4

........... ~ ...... i " ' i ' '=' "i 'i

2 3.5 5 6.5 8 9.5 II

pH

(bi~ -~ 0 KCL 2 1 I"1 I NaCI

"L" 1.5

0.5

-0.5

-1 i i I i ! z |

-1.5 3,0 4.0 5.0 6.0 7.0 8.0 9.0 ] 0

pH

Fig. 5. SP variations versus pH for the determination of the IEP of the NF-55 membrane: (a) for two ionic strength, electrolyte KC1, and (b) for three different electrolyte solutions, KCI, NaCI and CaC12 at a fixed concentration = 10 -4 mol/L.

T 0

> - 2 e t l . - 4 -

-8

-I0 , , I

10 "~ 10 -4 0.001 0.0l O.l 1 [KCI] mol/L

Fig. 6. Comparison of SP measurements, explained in mV/bars, vs. ionic strength, explained in terms of KC! concentration solutions, for mesoporous, UF, and mi- croporous, NF, membranes.

and solution. This phenomenon leads to the mo- dification of the surface charge of the material. When adsorbing ions are those of the solvent it- self, as usually observed in UF membranes [13,14,21], the surface charge of the membrane is dependent on the pH of the solution and the IEP corresponds to a pKa value of the mem- brane material because the membrane mainly exerts acidic-basis properties. In Fig. 6 the dif- ferent behaviours of the mesoporous and micro- porous membranes are compared in contact with different concentrations of KCI solutions. For the microporous membrane, specific ions sorption is dominant due to the increase of the electrostatic forces in narrow pores. Further- more it appeared that the SP values are depend-

ant on the electrolyte concentration with a sign inversion around 0.01 mol.L -~.

On the contrary with the mesoporous mem- brane no specific sorption of the electrolyte ions has been observed. Furthermore at high ionic strength it can be assumed that electroneutral channels are formed in the pores which is in good agreement with the electrical double layer compaction theory [23]. As reported elsewhere [24-26], the PA material seems to have an ampholytic behaviour under various pH's, ionic strength and transmembrane pressure. Con- sequently, if the ampholytic polymer contains weak basis and/or weak acidic groups, the membrane charge is affected by pH. Therefore we can use the ampholytic properties of the membrane material to make it negatively or positively charged or neutral by adjusting the pH value of the solution to be treated in order to minimize electrostatic sorption. IEP of the NF-55 membrane is dependent on the ionic strength and the nature of the electrolyte solu- tions in contact with the material. Contrary to the observations of Lebrun et al. [24] and Elimelech et al. [26], the ampholytic properties of the NF-55 PA membrane are not clear due to specific sorption of the electrolytes ions that have replaced I-I + and OH-. Furthermore the present results have detailed that this chemical adsorption is reversible and that the initial

Page 9: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

D. Violleau et al. / Desalination 173 (2005) 223-238

properties of the membrane can be regenerated under sonication in UP water [7,27]. But in the case of Ca 2÷ electrolytes solutions, membrane initial charge recovery was more difficult to re- generate than for K + or Na + electrolytes solu- tions due to its higher affinity with the PA material. Selectivity tests were performed to decide the relative importance of charge valence of cations in terms of the surface charge of membrane. Then our results have corro- borated the results reported by Cho et al. [28] that divalent cations (Ca 2+, Mg 2+) increase more significantly the membrane surface potentials compared to monovalent ones (Na +, K ÷) be- cause divalent cations have a greater potential in approaching membrane surfaces (i.e., inside the Stem layer). Thus, divalent cations can provide a greater double layer compaction and, when near the shear plane (available for both the potential measurement methods), exist to a lesser extent than monovalent cations.

We show in this paper that membrane foul- ing by natural organic matter has an extensive influence on the salt-rejection by the NF-55 membrane. Nevertheless, in order to improve the understanding of phenomena at the mem- brane solution interface, further research is needed. Particularly, new investigations must be achieved to take into account interactions between NOM and ions in solution.

It is well known today that the speciation and mobility of ions in natural waters are strongly affected by the presence of NOM, due to the formation of dissolved complexes. Actually, NOM consists of a mixture of large molecules bearing various kinds of complexing functional groups that exhibit a wide range of affinities for ions. The major binding sites in humie organics are usually attributed to the carboxylic and phe- nolic groups and the total binding capacity for metal ions usually in the range of 200--600 Ixrnol/g [29]. These inter-actions are not only important for the ion distribution, but also for the solubility and mobility of NOM, and could have a dual role: it can reduce the free ion concentration in solution and con-

231

sequently reduce their influence on membrane surface potentials, but the complexation of NOM with cations can also lead to increase the aver- age molecular weight of organic matters and improve membrane fouling. So, it seems to be obvious that these interactions can have a large impact in membrane technologies, especially in regard to the salt rejection behavior and the fouling processes. Further effects are correlated to specific adsorption of Ca 2÷ or Mg 2+ playing the role of a bridge between membrane material and NOM. Therefore electrostatic interaction between colloids and NOM, competition be- tween colloids and NOM for Ca 2+, and structure (porosity, density and thickness) of combined cake layer, should be analyzed to elucidate these complex interactions, as recently reported [30].

3.2.2. Hydraulic permeability measurements vs. p H

The hydraulic permeability is determined by following the AFNOR protocol [31] and cor- responds to the slope of the linear curve: hy- draulic flux vs. the transmembrane pressure. We have found a linear evolution of both para- meters in accordance with Poiseuille's law (re- sults not reported). The results plotted in Fig. 7 show that the hydraulic permeability of the clean NF-55 membrane decreases vs. pH in all the ranges of pH studied. This result is different in comparison to Childress et al. [26] observa- tions, especially in the range of low pH. Childress et al. [26] have obtained an increase of the hydraulic permeability from 2 to 5, a maximum at pH 5 (around the isoelectric point of the membrane) and a decrease at higher pH's. We have to notify that our experiments were conducted using UP water whereas Childress et al. [26] used a 0.01 mol/L KC1 solution. Then as detailed by these authors, the hydraulic per- meabilities decreasing vs. pH can be attributed to: (a) increase pore size due to conformational changes of the cross-linked membrane polymer structure, (b) increase apparent water perme- ability due to decrease electroviscous effect and/or (c) increase net driving pressure due to

Page 10: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

232

~ , = - - ~ Clean NF-55 :0.22 -O-NF-55 fouled HPOA

=-..e-.-NF-55 fouled TPIA

0.18 .J V

0. .<1 0.16

0.14

0.12

0.1

D. Violleau et al./ Desalination 173 (2005) 223-238

increases the salt rejection of the electrolyte added to adjust the pH increases. Increased salt rejection, coupled with concentration polari- zation, results in the increase of the osmotic pressure. Because the operating pressure of the system was kept constant, the increased osmotic pressure results in a decreased net driving pres- sure which in turn leads to decreased water flux.

I

12

"-O.O__0.--O n n n I

2 4 6 8 10 pH

Fig. 7. Hydraulic permeabilities vs. pH for the NF-55: clean, fouled with HPOA and fouled with TPIA, pH = 6.7, UP water solutions, T= 20°C, membrane con- tact time with NOM = 6 h.

increase osmotic pressure at the membrane sur- face.

Some authors [24-26] have described that with carboxylic acid groups the pore size of the membrane was found to be significantly reduced at higher pH values because the charged (=CO0) groups adopt an extended chain conformation due to electrostatic repulsion between them. This expanded conformation reduces the pore size (or pore volume) of the membrane and thereby causes decreased flux and increased salt rejection. Expending this explanation to the NF-55 membrane, which has both carboxyl and amine functional groups [25], the pore size of the membrane would be reduced at both high and low pH. In both cases, the electrostatic re- pulsion between the charged groups would cause a reduction in pore size. In our case the electroviscous effects could be due to specific sorption of the added ions for adjusted pH.

Furthermore, changes in osmotic pressure at the membrane surface may also explain the de- crease of the hydraulic permeability. As the pH

3.2.3. Wettability characterization o f the clean NF-55 membrane

Concerning the wettability, the contact angle measurements occurred by using the sessile drop technique and have shown no observable pH effect under the water contact angle measured on the PA material surface (see Fig. 8). As reported elsewhere [7,13,14,18,21 ], other materials such as polythersulfone (PES) have shown ampholytic properties due to acidic- basic complexes between the functional groups at the surface of the solid and the aqueous solu- tions [32]. This approach is another way to check

55

g - V

50

co

~m 4 5

o

40

35

30

.,,. • i , o - - - O . •

o clean NF-55 • NF-55 fouled TPIA a NF-55 fouled HPOA

o o o . . . . . . q , 8

o I I I

2 4 6 8 pH

10

Fig. 8. Contact angle measurements vs. pH for the NF-55 membrane: clean, fouled-HPOA and fouled- TPIA, pH = 6.7, volume of the water droplet: 2 IxL.

Page 11: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

D. Violleau et al. / Desalination 173 (2005) 223-238

the absence of ampholytic properties of the NF-55 membrane surface.

3.2.4. KCI rejection o f the clean membrane

In Fig. 9, we have reported the observed rejections of KCI aqueous solutions vs. the transmembrane pressure plotted for three con- centrations ranging from 10 -3 to 10 -1 mol/L. The results show that the rejections are decreas- ing with the ionic strength. It is well known that the electrochemical double layer thickness can be decreased under ionic strength increase, effect opening membranes pore diam (the debye length changes from 9.7 nm to 0.31 nm for the electrolyte concentration from 0.001 to 0.1 mol/L, in comparison with the 0.65 nm of the mean pore radius). At low ionic strength overlapping of the double electrochemical layers occurred and the consequences are a decrease in the membrane pore size, as well described else- where [14,23]. The observed rejection is also increasing vs. the transmembrane pressure due to the growth of the convective part of the mass transfer under pores size increasement due to ionic strength variations. Then at high pressure water convection mechanism could be dominant under hydration-diffusion mass transfer, as fully reported [33-49]. As a consequence the ob-

lOO

'...- 80 gl O

~ ao T ~ 4 0

~ 20

o

~ KCI 0.1 M

• - " B - - KCI 0.01 M

- ,e, - KC10.001 a

I I I I 2 4 6 8 10

AP (bars)

Fig. 9. Observed rejections vs. the transmembrane pres- sure for different concentrations of KCI aqueous solu- tions, Ph = 6.8.

233

served rejection increases. But at the same time the concentration polarization effect is working to accumulate electrolytes on the membrane sur- face increasing the concentration at the mem- brane solution interface and increasing the gradient of concentration on each side of the membrane. Then the observed rejection of KC1 decreases and its accumulation under the mem- brane surface contributes to increasing the os- mosis pressure difference between each side of the membrane. Consequently the hydraulic per- meability was decreased. In order to limit this dramatic decreasement a few strategies are en- gaged like increasing the flow rate, but the energetic cost is prohibitive. A more recent solution concerns the possibility of a chemical modification of the polyamide NF membranes [25,50].

3.3. Membrane properties after NOM fouling

Membrane fouling by humic substances is influenced both by the characteristics of the humic substances and also by the membrane material used, according to fully previously studied [50-75]. The hydrodynamic conditions and the chemical composition of the feed water could also play a role that we have to better understand. All of these parameters are essen- tial for a better control of membrane fouling by humic substances.

In the present studies, the different tools for membrane autopsies were engaged, such as the hydraulic permeabilitiy, the contact angle, the SP and the NaC1 solutions rejection mea- surements, they were conducted under the ini- tial and the fouled NF-55 membranes. The foulants are two selected NOM (denoted HPOA and TPIA) extracted from a natural water, as reported in the experimental section. Table 1 reports the elemental analysis of both NOM. The main differences between both NOM is their hydrophobicity with a higher hydrophobic character for the HPOA. The NF-55 membrane was coated by both foulants by permeation of a 5 mg/L concentration solution of each NOM (under a pH value of 5.9) during 6 h. Further-

Page 12: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

234 D. Violleau et al . / Desalination 173 (2005) 223-238

more the level of ionic strength was kept appro- ximately constant among the NOM fractions used because it affects the size and charge of NOM and the charge and size of the pore of the membrane, thus influencing the rejection and fouling properties of each foulant during NF operation.

The hydraulic permeabilities were followed vs. pH, as reported in Fig. 7. We have observed for both substances a dramatic decrease of the hydraulic permeability. Then the fouled mem- branes exhibited different behaviour in flux decline between each foulants. More quantita- tively a decrease of 40% was obtained for the HPOA and only 25% for the TPIA. Then whereas described by the reported articles [55,57,62,66,74], the hydrophobic acids were observed to cause greater flux decline than the transphilic acids. This was related to the higher content of large molecules in the hydrophobic fraction, and the fact that the more hydrophobic acids adsorb on the hydrophobic membranes to a greater extent than the less hydrophobic acids [57]. This was attributed mainly to the electro- static repulsion between the negatively charged membranes and the negatively charged NOM fractions. Moreover, it was less likely that there was any significant hydrophobic sorption of the charged fraction on the hydrophobic mem- branes as this fraction was hydrophilic [71 ].

The pH effect under measured hydraulic permeabilities is not sensitive enough after and before fouling. But the evolution observed is completely different for both NOM. Before NOM deposits a regular decrease of the measure hydraulic permeability was observed with in- creasing pH. But both fouled membranes present their proper behaviour. For HPOA an optimum is observable at pH 5 that we have attributed to the pKa of the carboxylic func- tional groups. For TPIA after a significant in- crease of the measurements in the acidic range of pH, a stabilization is observed. These obser- vations are typical of the physico-chemical changes of the membrane surface properties after NOM fouling. Then for the TPIA a maxi-

mum in water flux permeation is observed near the hypothetic isoelectric point of the fouled membrane. Indeed, the electroviscous effect is least pronounced at the pore surface point of zero charge where double-layer effects are neglected. At low pore surface charge the per- meating solution appears to exhibit a reduced viscosity when low flow rate is compared with the flow at high pore surface charge. Accord- ingly the flux would be maximum when the capillary is uncharged, or in other words, at the membrane isolectric point, as previously des- cribed [26] and may predominantly contribute to the maximum flux, shown in Fig. 7 after TPIA sorption. In the case of HPOA we have observed a minimum in the water flux at a pH of 5.2 (the novel isoeleetric point of NF-55 membrane fouled by HPOA), as similarly ob- served for proteins fouling [1 ].

To conclude, it seems that the NF-55 coated membranes have well acquired the physico- chemical properties of both NOM deposited.

In order to complete this macroscopic ap- proach based under the hydraulic permeabilities measurements, we have also measured the modifications of the wettability of the PA mem- brane surface occurring by water contact angles measurements (Fig. 8). We have observed that the initial hydrophilic PA membrane became more hydrophobic after foulants sorption.

As reported elsewhere, typically the hydro- phobieity of humie substances increases with increasing molecular weight and decreasing acidity [66,76]. Furthermore, the different wet- tabilities of the fouled membranes follow the hydrophilicPaydrophobic properties of the foul- ants. These results underline the difference in behaviour of these two humic substances. Fur- thermore, if we observed in detail the contact angle results obtained we have observed a dra- marie change in the wettability of both fouled membranes. We can hypothesise that the NF membrane has acquired predominant acidic- basis properties as a consequence of NOM sorption. Then the physico-chemical properties of the initial material has changed dramatically

Page 13: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

D. Violleau et al. / Desalination 173 (2005) 223-238

and the fouled membranes have acquired those of the foulants. Hydrophobic interactions play a significant role with hydrophobic membranes causing a reduction of permeate flux and we have checked that hydrophilic interactions play a significant role with hydrophilic membranes also causing a reduction of permeate flux. Finally, some authors [52] claimed for 10 years that hydropilic NOM had little influence on per- meate flux but only on hydrophilic materials. Today, as recently observed by Makdissy [74,75], it depends on the resource contents and higher from seasonal variations [56]. Then hy- drophilic/hydrophobic properties of the initial membrane seem not very crucial.

As a consequence of membrane fouling the selectivity of NF membranes could be modified as fully observed [51,60,62,66,67]. In order to evaluate the consequences of such modifi- cations, we have determined the observed rejec- tion of a NaC1 aqueous solution vs. transmem- brane pressure in the range of 0-9 bars. The re- suits are reported in Fig. 10. The fouled mem- branes have shown a higher modification of their

100

90

70

Fig. 10. Observed rejection vs. transmembrane pressure for a KCI solution: membranes NF-55: clean, fouled- HPOA and fouled-TPIA, pH = 6.7, KCI concentration 0.001 mol.L -1.

235

selectivity for NaCI due to the reduction of their mean pore size. The modifications are more visible for the significant foulant. Then the accu- mulation of NOM within and/or on the mem- brane shows an increase of the observed rejec- tion. HPOA appeared the more limiting NF per- formances between both fractions and then in- creased a lot the selectivity of the NF-55 membrane.

3.3.1. Effects o f N O M on the membrane sur- face charge

Hydrophilic/hydrophobic effects are not the only forces engaged in membrane fouling phe- nomena. Electrostatic effects have to be taken into account. In order to distinguish between fouling inside and/or on the membrane, a home- made apparatus for SP measurements has been accounted. This eleetrokinetic method, as de- tailed elsewhere [12], is useful to estimate charges on the surface pores walls. Fig. 11 shows the IEP displacement of the NF-55 membrane before and after NOM's deposit. We observed no IEP displacement for the TPIA, as for the HPOA fraction the initial IEP was dis- placed from 4.4 to 3.5. Similar displacement was also observed by Cho et al. [27]. Then those pre- liminary experiments, in combination with the

m

A ~ NF-55 clean + NF-55 fouled TPIA

NF-55 fouled HPOA

.o 6

• 0

0 8 - 2

Ap (bar) 2.5 3.5 4.5 5.5 6.5

pH

Fig. 11. SP measurements vs. pH for the NF-55: clean, fouled-TPIA and fouled-HPOA.

Page 14: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

236

results detailed before, permit to differentiate an inside fouling to a pore blocking mechanism which can help the sorption behaviour of the HPOA and TPIA onto the PA membrane. This observation could not be explained by a size effect due to the fact that HPOA has a bigger size than TPIA (see Table 1). We can hypothes- ise that a chemical affinity might exist between the HPOA and the PA material that contributes a lot to its intrinsic fouling. On the contrary, the anionic properties of the TPIA could explain its electrostatic repulsion with the membrane nega- tive charge. We want to notify that diffusion of TPIA could happen but not during the short time (6 h) of the experiments because diffusion is a very low mass transfer. It can be hypothes- ised that NOM acids approach and accumulate near the shear plane on the membrane surface and can even form complexes with the function- al groups on the membrane, thus significantly influencing the surface charge (particularly the potential).

The major differences in properties between the hydrophobic and transphilic NOM are their molecular structures in terms of aromatieity. This difference results in a shift toward the high- er negative potential range of the hydrophobic HPOA vs. the hydrophilic transphilic TPIA, be- cause of the hydrophobic interactions between the PA membrane surface and the hydrophobic NOM (based on aromaticity). In the case of non-isolated NOM, however, the amounts of shifts were small as compared with the NOM acids, because it contains large amounts of hydrophilic NOM with non-charged molecules. Considering the potential of both foulants, the HPOA-coated membrane had a much higher positive potential than the TPIA-membrane, especially at low pH.

Furthermore, a large increase in the SP values was observed after HPOA deposit in high acidic pH range not for the TPIA. In that range of pH only NH3 + groups of the PA material can contribute to a positive charge of the membrane surface. Then we can hypothesise that it occurs when a specific sorption of the ions is added to

D. Violleau et al. / Desalination 173 (2005) 223-238

adjust the pH under the PA-HPOA membrane surface.

4. Conclusion

Natural organic matters play an important role in membrane fouling. Today's NF industrial plants, just starting on a large scale, need long- term studies that they do not have. The present study is presented as an alternative for the limit- ation of NOM fouling during NF operation. In the present paper we have compared the fouling properties of two selected hydrophilic and hydrophobic NOM fractions extracted from the Blavet river (France) in contact with a hydro- philic PA membrane.

The analytical tools accounted micro- analyses, specific UV absorbance, solid-state cross-polarization magic angle spinning (CPMAS) 13C-NMR, high pressure size exclu- sion (HPSEC) chromatography, HPLC system, acid/base titration, contact angle, hydraulic per- meability, SP and observed rejection of a NaC1 aqueous solution) have shown that HPOA NOM is higher limiting PA membrane perform- anees than the TPIA. But for each foulant the membrane initial properties has been drama- tically changed due to NOM deposit.

The modifications observed are dependant on which compounds are mainly adsorbed in, or on the membrane. We tried to identify the com- pounds which are the more important on the fouling phenomena observed. We have obtained that the NF-55 membrane modified with the selected foulants became two novel membranes with smaller pores size and with novel acidic- basis properties that they acquired after NOM sorption.

All the analytical tools developed during this work to qualify both NOM and membrane material physico-chemical properties are the best approach in order to determine the optimum membranes material to be selected for a done water feed.

Further investigations are being conducted now in order to elaborate synthetic solution by combining HPOA, TPIA, proteins and polysac-

Page 15: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

D. VioUeau et al. / Desalination 173 (2005) 223-238

charides, in order to simulate the contents o f a real water feed. The interest in the good know- ledge of the water resource contents can facilitate an antifouling strategy in order to opti- mise the chemical washing procedures and/or the pretreatments of the water feeds.

Acknowledgements

This study was supported by the Center Region (France) and by VEOLIA WATER. Furthermore the Action Concert~e Incitative Jeune Chercheur (ACI), No. 4052 (2002) from the National Funds for Science has permitted us to elaborate a new homemade apparatus for the SP measurements. Grateful thanks for fruitful discussions go to Richard W. B6wen, Professor at the University of Wales-Swansea (UK), Hilal Nidal, lecturer at Nottingham University (UK) and Gladys Makdissy, postdoctorate at the University of Massachusetts (USA).

References

[1] M. Cheryan, Ultrafiltration and Microfiltration Handbook, Teehnomic Pub. Co. Inc., Lancaster, 1998.

[2] M. Mulder, Basic Principles of Membrane Tech- nology, Hardcover, 1996.

[3] W.S. Winston Ho and K.K. Sirkar, (eds), Mem- brane Handbook, Chapman et Hall, 1992.

[4] M.M. Ponti~, M. Rumeau, R. Audinos, F. Bedioui, M. Cassir, R. C16ment, J. Devynek, C. Jolivalt, J.P. Labb6, P. Letellier, P. Mandin, M. Minier, S. Sarrade, L. Schrive and V. Week, Les Techniques S6paratives a Membranes: Th6orie, Applications et Perspectives, CEE-UIE Paris, 2001.

[5] P. Aimar, Separations par Membrane Dans les Proc6d~s de L'industrie Alimentaire, Tech. Doe., Paris, 1998.

[6] R.J. Petersen, J. Membr. Sci. 83 (1993) 81. [7] M. Ponti~, C.K. Diawara and M. Rumeau, Desali-

nation, 151 (2002) 267. [8] J.P. CrouP, B. Martin, P. Simon and B.B. Legube,

Wat. Supply, 11 (1993)51. [9] E.W.D. Huffman and H.A. Stuber, in: G.R. Aiken

et al., (eds), Analytical Methodology for Elemental Analysis of Humic Substances, Humie Substances in Soil, Sediments and Water - Geochemistry, Iso- lation and Characterization, Wiley-Intersciences, NY, 1985.

237

[10] Y. Chin, G. Aiken and E. O'Loughlin, Environ. Sci. Technol., 28 (1994) 1853.

[ll]E.M. Thurman, Developments in Biochemistry: Organic Geochemistry of Natural Waters, M. Nijhoff and Junk W. Publishers, Dordrecht, 1985.

[12] M. Ponti6, X. Chasseray, D. Lemordant and J.M. Lain6, J. Membr. Sei., 129 (1997) 125.

[13] M. Ponti6, D. Lemordant and J.M. Lain~, Sep. Purif. Tech., 14 (1998) 1.

[14] M. Ponti6, L. Durand-Bourlier, D. Lemordant and J.M. Lain~, J. Claim. Phys., 94 (1997) 1741.

[ 15] M. Ponti~, Electrokinetics phenomena and electro- lyres transfer in microporous membranes, PhD thesis, University of Tours, France, 1996.

[16] J.P. Crou6, D. Violleau and L. Labouyrie, Disin- fection by-products formation potentials of hydro- phobic and hydrophilic natural organic matter fractions: a comparison between a low and a high humic water, in: S.E. Barrett, S.W. Krasner and G.L. Amy (eds), Natural organic matter and dis- infection by-products: characterization and control in drinking water, ACS Symposium Series 761, American Chemical Soc., (2000) 139.

[17] M. Ponti~, J. Membr. Sci., 154 (1999) 213. [18] P. Fievet, M. Sbai, A. Szymczyk and A. Vidonnc,

J. Membr. Sci., 226 (2003) 227. [19] R. Zimmermann, S. Dnkhin and C. Werner, J.

Phys. Chem. B, 105 (2001) 8544. [20] M.D. Afonso, G. Hagmeyer and R. Gimbel, Sep.

Purif. Tech., 22(23) (2001) 529. [21] M.J. Ariza and J. Benavente, J. Membr. Sci., 190

(2001) 119. [22] A. Yaroshchuk and V. Ribitsch, Chem. Eng. J., 80

(2000) 203. [23] R.L Hunter, Introduction to Modem Colloid Sci.,

Oxford Sci. PubI., 1996. [24] Y. Xu and R.E. Lehrun, Desalination, 122 (1999)

95. [25] A. Akbari, J.C. Remigy and P. Aptel, Chem. Eng.

Process, 41 (2002) 601. [26] A.E. Childress and M. Elimelech, Environ. Sci.

Technol., 34 (2000) 3710. [27] M. Pontie and D. Lemordant, J. Membr. Sei., 141

(1998) 13. [28] Y. Shim, H. Lee, S. Lee, S. Moon and J. Cho, En-

viron. Sei. Teehnol., 36 (2002) 3864. [29] S. Lee, J. Cho and M. Elimelech, Simulation of

feed water recovery and concentration factor and their influence on colloid and NOM fouling of NF membranes, Membrane Teeh. Conf., American Water Works Assoc., 2003, pp. 1-12.

[30] D. Violleau, Int&~t du fractionnement et de l'ex- traction des mati~res organiques naturelles d'eaux

Page 16: Fouling studies of a polyamide nanofiltration membrane by selected natural organic matter: an analytical approach

238 D. Violleau et al. / Desalination 173 (2005) 223-238

de surface pour l'~tude de leurs propri~t~s struc- turales et de leur pouvoir complexant vis-/l-vis du cuivre, PhD thesis, University of Poitiers, France, 1999.

[31] AFNOR, Filtration des Liquides: Membranes Pore- uses; Norme NF )(45-101, 1996.

[32] K.J. HiRtinger, H. Wien and G. Krekel, J. Adhesion Sci. Technol., 6 (1992) 317.

[33] C. Menjeaud, M. Ponti6 and M. Rumeau, Entropie, 179 (1993) 13.

[34] M. Pontie, H. Buisson, C.K. Diawara and H. Essis- Tome, Desalination, 157 (2003)127.

[35] W.R. Bowen, A.W. Mohammad and N. Hilal, J. Membr. Sci., 126 (1997) 91.

[36] C. Combe, C. Guizard, P. Aimar and V. Sanchez, J. Membr. Sci., 129 (1997) 147.

[37] G.M. Rios, R. Joulie, S.J. Sarrade and M. Carl6s, AIChE J., 42 (1996) 2521.

[38] B. Chaufer, M. Rabiller-Baudry, L. Guilhard and G. Daufin, Desalination, 104 (1997) 37.

[39] M. Ponti6, C. Diawara, M. Rumeau, D. Aureau and P. Hemmery, Desalination, 158 (2003) 277.

[40] A. Lhassani, M. Rumeau, D. Benjelloun and M. Ponti6, Wat. Res., 35 (2001) 3260.

[41] C.K. Diawara, M. Ponti6, S.M. Lo, O. Sarr and M. Rumeau, J. Membr. Sci., 219 (2003) 103.

[42] P.Y. Pontalier, A. Ismail and M. Ghoul, Sep. Purif. Techn., 12 (1997) 175.

[43] Y. Garba, S. Taha, N. Gondrexon and G. Dorange, J. Membr. Sci., 160 (1999) 187.

[44] J. Schaep, C. Vandecasteele, M.A, Wahab and R.W. B6wen, Sep. Purif. Tech., 22(23) (2001) 169.

[45] S. Choi, Z. Yun, S. Hong and K. Ahn, Desalina- tion, 133 (2001) 53.

[46] J. Schaep, B. Van der Bruggen, C. Vandecasteele and D. Wilms, Sep. Purif. Tech., 14 (1998) 155.

[47] B. Van der Bruggen, L. Braeken and C. Vandeeas- teele, Desalination, 147 (2002) 281.

[48] Y. Xu and R.E. Lebran, J. Membr. Sci., 158 (2003) 93.

[49] V. Freger, J. Gilron and S. Belfer, J. Membr. Sci. 209 (2002) 283.

[50] A.L. Sch~fer, A.G. Fane and T.D. Waite, Desali- nation, 118 (1998) 109.

[51] A.L Sch[ifer, A.G. Fane and T.D. Waite, Desali- nation, 131 (2000) 215.

[52] G.F. Crozes, J.G. Jacangelo, C. Anselme and J.M. Lain6, J. Membr. Sci., 124 (1997) 63.

[53] C. Ventresque, V. Gisclon, G. Bablon and (3. Chagneau, Desalination, 131 (2002) 1.

[54] A.E. Childress and S. Deshmukh, Desalination, 118 (1998) 167.

[55] C. Jarusutthiriak, G. Amy and J.P. CrouP, Desali- nation, 145 (2002) 247.

[56] N. Her, G. Amy and C. Jarusutthirak, Desalina- tion, 132 (2000) 143.

~57] C. Jucker and M.M. Clark, J. Membr. Sci., 97 (1994) 37.

[58]W.R. Bowen, T.A. Doneva and H.B. Yin, J. Membr. Sci., 206 (2002) 417.

[59] J.A. Nilson and F. DiGiano, J. AWWA, 5 (1996) 53.

[60] A.Seidel and M. Elimelech, J. Membr. Sci., 203 (2002) 245.

[61] S. Hong and M. Elimelech, J. Membr. Sci., 132 (1997) 159.

[62] E.M. Gwon, M.J. Yu, H.K. Oh and Y.H. Yiee, Wat. Res., 37 (2003) 2989.

[63] J.S. Vrouwenvelder, J.W.N.M. Kappelhof, S.G.J. Heijman, J.C. Schippers and D. van der Kooij, De- salination, 157 (2003) 361.

[64] K. Kosutic and B. Kunst, Desalination, 150 (2002) 113.

[65] A. Combe, E. Molis, P. Lucas, R. Riley and M.M. Clark, J. Membr. Sci., 154 (1999) 73.

[66] J. Cho, G. Amy, J. Pellegrino and Y. Yoon, De- salination, 118 0998) 101.

[67] J. Cho, G. Amy, J. Pellegrino and Y. Yoon, Wat. Res., 33 (1999) 2517.

[68] M. Manttari, L. Puro, J. Nuorila-Jokinen and M. Nystrt}m, J. Membr. Sci., 165 (2000) I.

[69] T.L. Champlin, Desalination, 131 (2000) 105. [70] F. Linhua, J.L. Harris, F.A. Roddick and N.A.

Booker, Wat. Res., 35 (2001) 4455. [71] J. Meier-Haach, N.A. Booker and T. Caroll, Wat.

Res., 37 (2003) 585. [72] T. Caroll, N.A. Booker and J. Meier-Haach, J.

Membr. Sci., 203 (2002) 1. [73] A.R. Roudman and F.A. DiGiano, J. Membr. Sci.,

175 (2000) 61. [74] H. Habarou, G. Makdissy, J.P. Crou6, G. Amy, H.

Buisson and C. Machinal, Toward an understand- ing of NOM fouling of UF membranes, Proc. AWWA Conf. San Antonio, 2001, p.1.

[75] G. Makdissy, Etude des propfi6t6s colmatantes des mati~res organiques naturelles vis-/t-vis des mem- branes d'ultrafiltration, Phi) thesis, University of Poitiers, France, 2003.

[76]A.K. Zander, C. Cleveland and N.K. Curry, Measurement of membrane surface hydropho- boeity and charge as affected by solution charac- teristics, Proc. Membr. Technol. Conf. AWWA, Rcno, Nevada, 1995.