behaviours of natural organic matter in membrane filtration for surface water treatment

0011-9164/06/$– See front matter © 2006 Elsevier B.V. All rights reserved

Desalination 194 (2006) 211–231

Behaviours of natural organic matter in membrane filtration forsurface water treatment — a review

A.W. Zularisama,b, A.F. Ismaila,*, Razman Salimc

aMembrane Research Unit, Universiti Teknologi Malaysia, Skudai Johor, MalaysiaTel. +60 (7) 553-5592; Fax: +60 (7) 558-1463; email: [email protected]

bEnvironmental Engineering Focus Group, Faculty of Civil and Environmental Engineering, Kolej UniversitiKejuruteraan and Teknologi Malaysia, Gambang, Pahang, Malaysia

cFaculty of Civil Engineering, Universiti Teknologi Malaysia, Skudai Johor, Malaysia

Received 13 July 2005; accepted 13 October 2005

Abstract

Membrane application in surface water treatment provides many advantages over conventional treatment. However,this effort is hampered by the fouling issue, which restricts its widespread application due to increases in hydraulicresistances, operational and maintenance costs, deterioration of productivity and frequency of membrane regenerationproblems. This paper discusses natural organic matter (NOM) and its components as the major membrane foulants thatoccur during the water filtration process, possible fouling mechanisms relating to reversible and irreversible of NOMfouling, current techniques used to characterize fouling mechanisms and methods to control fouling. Feed properties,membrane characteristics, operational conditions and solution chemistry were also found to strongly influence thenature and extent of NOM fouling. Findings of such studies are highlighted. The understanding of the combined rolesof controlling factors and the methods used is very important in order to choose and optimize the best technique andconditions during surface water treatment. The future potential of membrane application for NOM removal is alsodiscussed.

Keywords: NOM; Fouling mechanism; Surface water treatment; Membrane filtration

1. Introduction

Increasing population and improper indus-trialization practices have led to detrimentalsurface and underground water pollution. Thisphenomenon has raised concern in the public for

*Corresponding author.

stringent environmental legislation and alterna-tive technology in water treatment. The presenceof free chlorine content that is used as a disin-fectant in conventional treatment is found to reactwith residual natural organic matter (NOM). Thisreaction process has been found to have a ten-dency to form disinfection by-products (DBPs)

doi:10.1016/j.desal.2005.10.030

A.W. Zularisam et al. / Desalination 194 (2006) 211–231212

such as trihalomethanes, haloacetics and otherhalogenated organics. DBPs are carcinogens, anddirect exposure can lead to cancers, miscarriagesand nervous system complications. Its small size,easier maintenance and superior water qualityproduced by membrane filtration has made thisadvanced technology possible to replace conven-tional treatment processes that consist of ozona-tion, precipitation, coagulation, flocculation,chlorination and gravel filtration [1]. In addition,the membrane filtration process offers the extraadvantages over conventional treatment such as asmall footprint, compact module, lower energyconsumption, environmental friendliness and thecapability of handling wide fluctuations in feedquality.

Membrane filtration processes involvingmicrofiltration (MF), ultrafiltration (UF) andnanofiltration (NF) in potable water productionhave increased rapidly over the past decade. MFand UF are employed to remove microparticlesand macromolecules, which generally includeinorganic particles, organic colloids (i.e., micro-organisms) and dissolved organic matter (DOM).However, the use of membrane processes doesnot directly eliminate the problem of DBPs; eventhough MF, UF, NF or reverse osmosis (RO)have excellent performance on the removal ofmicrobial particles, the disinfection process is stillnecessary.

DOM is ubiquitous in natural surface watersand is often claimed as an important factor forboth the reversible and irreversible fouling inwater filtration [2–6]. As for other membraneapplications, the preferred membrane materialsfor surface water filtration are mostly polymericwith a configuration ranging from flat sheet tohollow fibre (Table 1).

Fouling is the main obstacle in membranefiltration efficiency as it causes a reduction inproductivity. Productivity decline can be definedas a decrement in flux with time of operation dueto the increment of hydraulic resistance. Pro-ductivity decline may also be interpreted as a

need for additional energy supply to the filtrationsystem so as to keep the system performanceconstant. Therefore, a fundamental knowledge ofthe possible foulants and how they cause foulingis essential before any remediation work is car-ried out. In the present paper the characteristics ofNOM fouling are reviewed during surface waterfiltration that relate to both reversible and irrever-sible fouling mechanisms. Discussion is alsodevoted to the chemistry of NOM, current tech-niques used to characterize foulants, foulingmechanisms, membrane characteristics, foulingcontrol methods, feed properties, operationalconditions and solution chemistry that have beenreported to potentially influence the NOM foulingin surface water filtration.

2. Chemistry of NOM

It is vital to identify foulants and foulingmechanisms before membrane fouling can bealleviated. “Membrane fouling” refers to bothreversible and irreversible alteration in membraneproperties. Reversible fouling means depositionof retained solutes on the membrane surface thatgenerally exists as a gel cake layer. Irreversiblefouling refers to adsorption or pore plugging ofsolutes in and within the membrane pore matrix.Concentration polarizations are the accumulationof retained materials in the boundary layer abovethe membrane due to osmotic pressure and thehydraulic resistance effect. Increments and vari-ation of hydraulic resistances may come from avariety of organic substances, inorganic particles,colloids and microorganisms with different foul-ing behaviours.

Fouling behaviour is found to be significantlyinfluenced by various chemical and physical fac-tors of the foulants. The foulant can be charac-terized according to its molecular structure,surface charge, molecular size and functionalgroups. One of the most important identified foul-ants found in surface water filtration is NOM.

A.W. Zularisam et al. / Desalination 194 (2006) 211–231 213

Table 1Membrane characteristics for water treatment

Membrane materials Configuration Pore size Reference

Sulfonated PESPolyslphonePolysulfonePolypropylenePESRegenerated cellulosePVDFCellulose esterPANMetalComposite (PA+PS)PolypiperazinePESPolyethylenePANPolypropylenePolysulphoneSulfonated PESPVDFRegenerated cellulosePolyamide TFCPolypiperazinePolypropylenePolysulphonePolysulphonePESCellulosicAcrylicPolyethylene

Flat sheet (cross flow)Plate and frame (cross flow)Flat sheet (cross flow)Hollow fibre (dead end)Flat sheet (dead end)Flat sheet (dead end)Flat sheet (dead end)Flat sheet (dead end)Hollow fibre (cross flow)Plate and frame (submerged)Flat sheet (cross flow)Flat sheet (spiral wound)Flat sheet (spiral wound)Hollow fibre (submerged)Flat sheet (cross flow)Flat sheet (dead end)Hollow fibre (cross flow)Flat sheet (cross flow)Flat sheet (dead end)Flat sheet (cross flow)Flat sheet (cross flow)Flat sheet (cross flow)Hollow fibre (dead end)Capillary (cross flow)Hollow fibre (cross flow)Flat sheet (dead end)Hollow fibre (cross flow)Hollow fibre (cross flow)Hollow fibre (dead end)

0.7 nm10 kDa10 kDa0.2 µm100 kDa100 kDa0.22 µm0.22 µm50 kDa0.2 µm0.65 nm360 g/mol2000 g/mol0.1 µm110 kDa0.2 µm0.01 µm20 kDa0.22 µm3 kDa8 kDa290 Da0.2 µm40 kDa100 kDa0.16 µm100 kDa50 kDa0.1 µm

[5][7][7][8][9][9][9][9][10][11][12][13][13][14][15][16][17][18][19][20][20][21][22][23][24][25][26][26][27]

In water NOM is a complex mix of particulateand soluble components of both inorganic andorganic origin that vary from one source toanother [16]. NOM is a heterogeneous mixturewith wide ranges in molecular weight (MW) andfunctional groups (phenolic, hydroxyl, carbonylgroups and carboxylic acid) and is formed byallochthonous input such as terrestrial and vege-tative debris and autochthonous input such asalgae. Among these components, DOM is foundto have the most detrimental effect on membraneperformance as it can result in irreversible foul-

ing during surface water filtration. DOM is aubiquitous constituent in natural waters and gene-rally is comprised of humic substances; poly-saccharides; amino acids; proteins; fatty acids;phenols; carboxylic acids; quinines; lignins; car-bohydrates; alcohols; resins; and inorganic com-pounds such as silica, alumino-silicates, iron,aluminium, suspended solids and microorganisms(bacteria and fungus).

NOM that occur in natural brown water arepolyphenolic molecules with MW ranging from5,000 to 50,000 Dalton [23]. In particular, NOM


Fig. 1. Fraction of NOM in surface water based on DOC [29].

can be fractionated into three segments: hydro-phobic (humic substances), hydrophilic andtransphilic fractions. The hydrophobic fractionrepresents almost 50% of dissolved organiccarbon (Fig. 1) with larger MW. The hydrophilicfraction is composed of 25–40% of dissolvedorganic carbon (DOC) with lower MW (poly-saccharides, amino acids, protein, etc.) and isoperationally defined as a non-humic fraction.The transphilic fraction is comprised of approxi-mately 25% DOC in natural water but with MWin between hydrophobic and hydrophilic frac-tions. A major fraction of the NOM arises fromhumic substances which are reported to represent60–70% of TOC in soils and 60–90% of DOC inmost natural waters [28]. This statement is foundto be inconsistent with the information reportedby Thurman [29] in Fig. 1, but this conflictingfinding is not surprising since humic substanceconcentrations may vary with season and source[24].

Fan et al. [19] reported that the major fraction(over 50% of DOC) of NOM is composed ofhumic substances that are responsible for thecolour of natural water. A humic substance is thepredominant fraction of NOM and generally isdivided into three categories: humic acid (HA),fulvic acids (FA) and humin. HA and FA areanionic polyelectrolyte with negatively chargedcarboxylic acid (COOH-), methoxyl carbonyls(C=O) and phenolic (OH-) functional groups.

Figs. 2 and 3 show both models of HA and FAstructures. HA is soluble at higher pH — nor-mally 10 — while fulvic acid is soluble in waterat any pH [30]. Humin exists naturally as black incolour and is not soluble in water at any pH(Table 2). A humic fraction has been identified asthe major foulant in membrane water filtration,which controls the rate and extent of fouling [31].It causes more fouling than any other NOM com-ponent due to its adsorptive capacity on themembrane surface [32]. A study done by Malle-vialle et al. [33] showed an organic matrix formeda structure of fouling layer that served as a gluefor inorganic constituents. Similar results werereported by Kaiya et al. [27] in analyzing adeposited layer formed on a MF hollow fibreduring filtration of Lake Kasumigaura water.

NOM deposition has been found as the domi-nant factor causing flux decline along with man-ganese constituents. A study by Mo and Huang[14] on the purification of micro-polluted rawwater revealed that fouling on the exterior surfacewas a combined effect of microorganisms andinorganic matter, while on the inner surface, itwas mainly due to biofouling. These studiesfound that organic foulants were of low MW andthe primary inorganic substance was Ca2+. Theirinvestigations on membrane permeability recov-ery showed that alkaline cleaning was effective inremoving organic foulants while acidic cleaningwas more effective for inorganic scale. It has also


Fig. 2. Schematic of humic acid model structure [35].

Fig. 3. Schematic of fulvic acid model structure [36].

Table 2Physical and chemical characteristics of humic substances[35]

Fulvic acid Humic acid Humin

Lightyellow

Yellowbrown

Darkbrown

Greyblack

Black

Increase in degree of polymerization

2000 Increase in MW 300,000 and above45% Increase in carbon content 62%48% Decrease in oxygen content 30%1400% Decrease in exchange acidity 500%

Decrease in degree of solubility

been shown that hydrophobic fraction of NOMcauses much more fouling than hydrophilic frac-tions [34]. Nilson and DiGiano [34] performedNF of a hydrophilic membrane with aquaticNOM using DAX-8 to fractionate the NOMcomponents. Hydrophobic fraction (absorbable toDAX-8) was mainly responsible for the permeateflux decline. On the other hand, the hydrophiliccomponent which passed through the DAX-8showed a lower fouling effect. Humic macro-molecules with higher hydrophobicity were foundto favourably be adsorbed onto hydrophobicmembranes than a hydrophilic fraction [34].

Previous studies done by many researchersshowed humic substances caused irreversiblefouling of membranes [25,37]. Yuan and Zydney[25] studied HA fouling on a 0.16 µm hydrophilicMF and found that aggregate HA was responsiblefor the initial stage of fouling. Furthermore, thefouling mechanism was substantially due to con-vective deposition with little internal pore adsorp-tion. This finding is well supported by Schafer etal. [38] who observed that HA caused a 78%decline in flux compared to FA (15%). HA wasobserved to have a greater impact on membraneperformance (irreversible fouling) than FA andhydrophilic fraction (reversible fouling). Thisscenario might be due to its high aromaticityproperties, adsorptive behaviour, hydrophobia


and greater MW that led to the tendency to foul.A similar result was also reported by Lahoussine-Turcaud et al. [39] during UF observations withseveral organic and inorganic compounds fromthe Seine River. The flux decline observed wasprimarily due to HA deposition on the membranesurface.

The cellulose acetate membrane (hydrophilic)flux was twice as great as the hydrophobic poly-ethersulfone (PES) during UF of river water. Thehydrophilic components were thought to haveless of an impact on water quality than the humicfraction. However, recent studies done by Lin etal. [24] and Carroll et al. [22] have claimed thatthe non-humic fraction of NOM (hydrophilic andneutrals) materials was responsible in deter-mining the rate and extent of flux decline. Carrollet al. [22] performed MF of hydrophobic hollow-fibre membranes with a single water source andconcluded that the major cause of fouling wasdue to hydrophilic neutralily and not humic sub-stances. Fan et al. [19] reported the order offouling potential of NOM fraction as hydrophilicneutral > hydrophobic acids > transphilic acids >hydrophilic charged. They also found that hydro-phobic membranes had the greatest fouling effectthan hydrophilic membranes of similar size, sug-gesting that the fouling mechanism was governedby adsorption.

In addition, high MW components were iden-tified as having the largest impact on membranefouling compared to smaller DOM. This findingwas well supported by the study carried out byLee et al. [40] who found that polysaccharidesand protein that have larger MW and lower UV toHPSEC–DOC/UV response significantly fouledtheir low-pressure (MF/UF) membrane. Polysac-charides are aldehyde derivatives of high poly-hydric alcohols, which have neutral charactersthat can cause adsorption on a charged hydro-phobic UF membrane [41]. The authors suggestedthat the large neutral NOM fractions were theprime foulants rather than humic substances.Speth et al. [42] in their study also found

hydrophilic neutrals fouled more than hydro-phobic acids. This can be due to a bulky, macro-molecular shape and neutral character of thepolysaccharides that make it prone to foul andadsorb on membrane surface.

Lin et al. [24] studied the effect of fractionatedNOM onto a negatively charged UF membraneand observed that both the large scale of hydro-phobic and hydrophilic NOM components causedflux decline. However, the hydrophilic fractionwas found to induce the worst fouling. Jaru-sutthirak et al. [43] in their study on the effect ofeffluent organic matter for UF membranes alsofound that the high MW of hydrophilic com-ponents was the prime contributor of NOM foul-ing. It can be claimed that NOM fouling was aresult of low UV absorbing compounds and highMW hydrophilic components that occurredthrough adsorption mechanisms.

Inorganic particles can also affect the foulingbehaviours of organic substances. The presenceof inorganic particles such as clay minerals insurface water created a significant competitionbetween NOM and inorganic particles to adsorbonto the membrane surface or in the pores. Highsurface areas of inorganic particles enhance theadsorption of organic substances on clay mineralsand affect the fouling characteristic. This pheno-menon either resulted in enhancing particledeposition on the membrane or a decrease ofsorption of NOM onto the membrane and henceincrease membrane permeability.

2.1. Effect of solution chemistry on NOM fouling

The extent of NOM fouling was also found tobe strongly governed by ionic strength, pH anddivalent cations [20]. NOM particles are seen toagglomerate more at low pH and in the presenceof high multivalent cations concentrations. Con-trarily, the particles stretch to more linear chainsin low concentration, at neutral pH and low ionicstrengths [44]. Many previous studies have shownthat permeate flux is substantially reduced with


lower pH, high ionic strength and increasingdivalent cation concentration [45]. Most of thisbehaviour could be explained by changes of intra-and intermolecular electrostatic gradients of thefunctional groups (COO- and COOH). Increasingionic concentration changed rejection by buffer-ing or shielding the charge on the solute mole-cules. This condition encouraged coiling andaggregation of NOM that led to higher apparentMW and surface area.

Hydraulic resistance of humic acid was foundto increase at low pH, high ionic strength and inthe presence of calcium ion. Calcium ions reducethe humic acid solubility and increase its aggre-gation by canceling (protonation) the negativecharge effect of the functional group [20] or bybridging the negative membrane surface with thenegative charge functional groups (carboxylic,phenolics and methoxy carbonyls) of humicsubstances. Braghetta et al. [46] claimed low pHand high ionic strength caused compaction ofmembrane pores size that restricts the entrance ofsolute through the membrane. It was alsohypothesized that low pH and high ionic strengthcontribute to the reduced intermolecular electro-static repulsion of NOM. Subsequently, mole-cules densely accumulated on the membrane sur-face which reduced pure water permeability.

Incremental surface water pH was found toalter the physical and chemical properties [23].The authors ascribed the increase of UV254absorbance to an increase in pH as a result ofionization of the carboxyl groups process thatincreased intramolecular repulsion and solubility.A decrease of UV254 absorbance values at low pHconditions was because NOM carboxyl groupswere protonated and formed large complexes,which were less soluble and precipitated. Theprecipitation process removed most of the humicacid fraction and left only soluble fulvic acid inthe solution. This hypothesis was proven from theresult of the E4/E6 ratio (absorbance at 465 and656 nm) that was inversely proportional with thevalue of obtained UV254 absorbance.

In a study of chemical and physical impact ofNOM fouling in NF, Hong and Elimelech [3]found that fouling increased with increasing ionicstrength, decreasing pH and with addition of Ca2+.This finding was supported by Ghosh andSchnitzer [47] who found that it was due to lowerintramolecular repulsion. A variation of thissolution chemistry leads to changes in the netelectrostatic charge of the humic’s functionalgroups. Divalent cations specifically interact withhumic functional groups and thus greatly reducethe NOM interchain repulsion between humicmacromolecules. Reduction in the electrostaticrepulsion later results in the formation of adensely packed fouling layer. The authorsconcluded that the rate of fouling is mainly con-trolled by the interplay between drag and electro-static double layer repulsion in NF.

Jucker and Clark [48] explained the effect ofelectrolytes on humic substance aggregation inthe form of apparent molecular weight (AMW).He found that humic substance’s AMW distri-bution was shifting to a higher value at low ionicstrength compared to results obtained at highionic strength. This phenomena is probably due tothe fact that humic substances are more compactat high ionic strength as a result of a contractionof the molecule which is caused by diminution ofintramolecular electrostatic repulsion as thecharges are neutralized [49]. Calcium concen-tration adsorbed on the membrane is found to bein proportion to the amount of humic substancesadsorbed [50]. Therefore, calcium content in anysurface water should be taken into account asexcessive calcium presence definitely resulted inthe substantial adsorption of humic substances onthe membrane.

Hong and Elimelech [3] experienced a 50%flux decline after 70 h of filtration with a highpermeation rate of 3.1×10!5 m/s and less than10% with a low permeation rate (0.4×10!5 m/s).They found that a faster fouling rate happened athigh ionic strength and with the addition ofcalcium ion. They hypothesized that both pheno-


mena were due to a reduction in electrostaticrepulsion between the charged humic acids andthe membrane. The same finding was observedby Bob and Walker [51] where they hypothesizedthat increasing the ionic strength led to a con-tinous change in humic acid from uncoiledmacromolecules to a fully coiled state. Coiledhumic acids posed higher surface area and AMWthat made it easier to be retained by the mem-brane. Subsequently, this increased the humicacid rejection, hydraulic resistance and foulingrate as well.

3. Identification of foulantsResearchers have used a number of techniques

including inline attenuated total reflection (ATR)Fourier transform infrared (FTIR) spectrometry,UV254 absorbance, HPSEC–DOC–UV, non-ionicmacroporous ion exchange and pyrolysis–GC/MSto analyze and characterize membrane foulants.

3.1. ATR–FTIRATR–FTIR spectrometry provides insight into

the nature of foulants in the membrane texture. Itappears to be a valuable tool for foulant autopsies[16]. ATR–FTIR can also be used to determinethe functional groups of certain unknown foulantswhich correspond to their vibrational energy ofatomic bonds. Different functional groups absorbenergy at different specific wavelengths that latercan be translated into an intensity response.Frequent absorption bands seen are shown inTable 3 [52]. Nevertheless, this method may beinsignificant in identifying certain functionalgroups when the absorption reading gives broadover-lapping bands. This phenomenon occurreddue to the heterogeneity of natural water.

3.2. UV254 absorbanceResearchers [15,53,58] use UV254 absorbance

to identify the permeate and retentate of rejected

Table 3Common IR spectra for humic substances, poly-saccharides and proteins [52]

Bands (cm!1) Functional group

Humic substances:3400–3300 O-H stretching

N-H stretching2940–2900 Aliphatic C-H stretching1725–1720 Carboxylic acids1660–1630 C=O stretching of amide group1620–1600 Aromatic C=C1590–1517 COO–, N-H deformation1460–1450 Aliphatic C-H1400–1390 OH deformation, C–O stretching

of phenolic OH1280–1200 C-O stretching, OH deformation

of COOH1170–950 C-O stretching of polysaccharidePolysaccharidesgroup:3400 Alcohol (1,2,3, Ar)2940 Alkane1480 Alkane1370 1370 (starch)1170 Tertiary alcohol1120 Secondary alcohol1040 Aliphatic ether1000 Primary alcohol775 EthylProteins group3300 Alcohol1640 Alkene in aromatic1540 Mono substituted amide1100 Ether

compounds specifically for humic substances inunit of cm!1 or m!1. The presence of unsaturatedcompounds generally produced a distinct colourand can therefore be detected by UV–Vis [4].

UV254 absorbance is sensitive to aromaticcomponents and is an indicator for both HA andfulvic acid presence. Samples were first filteredthrough 0.2 µm to remove particulate matter andDI was used as a blank. The difference in readingof UV254 absorbance between feed and permeate


indicates the quantity of rejected humic substanceby the membrane. Sspecific ultraviolet absorb-ance (SUVA) is a ratio of UV at the wavelengthof 254 nm and DOC. High SUVA means higharomaticity or hydrophobicity of samples in thelimited DOC. The permeate from membrane thefiltration process with a high SUVA value con-forms to most of the rejected compounds that arenon-humic and that resulted in high values ofUV254 in the permeate. Previous studies reportedthat SUVA of NOM from natural water or groundwater was in the range of 2.4–4.3 L/mg-m to 4.4–5.7 L/mg-m, respectively [54].

3.3. HPSEC with online UV and DOC detection

The MW distribution of NOM was normallydetermined using high-performance liquid chro-matography (HPSEC) with online UV and DOCdetection [21]. The HPSEC contains a porous gelthat allows separation of molecules based on theirmass and MW. Smaller molecules access most ofthe pore volume while larger molecules that can-not pass the pores are eluted first. Subsequently,non-humic compounds with greater MW such ascarboxylic protein and polysaccharides exhibitsignificant DOC peaks but with a low area ofUV254 peak. On the other hand, humic fractionssuch as HA and fulvic acid exhibit high peakswith a molecular mass between 500 to 2000 Dal-tons with a high UV response.

3.4. Non-ionic macroporous ion exchange

The most common techniques for isolation ofNOM fractions are gel filtration, UF and adsorp-tion using non-ionic macroporous ion-exchangeresins DAX-8 and XAD-4 [8,55]. The surfacewater is fractionated into hydrophobic, which isDAX-8 adsorbable; transphilic, which is XAD-4adsorbable; and hydrophilic components, whichpass through the DAX-8 and XAD-4 resin with-out any adsorption (Fig. 4).

Fig. 4. Schematic diagram of HPO, TPI and HPI frac-tionation [8].

3.5. Pyrolysis–GC/MS

The application of pyrolysis and the GC/MStool as an analytical method to characterize com-plex organic matter was successfully used bySpeth et al. [42] and Jarusutthirak et al. [43]. Arecent study done by Speth et al. [42] usingpyrolysis with GC/MS showed that hydrophilicfractions of NOM were the major foulants forriver water filtration.

The pyrolysis–GC/MS method was developedby Bruchet et al. [56]. This tool is useful for char-acterizing NOM in terms of biopolymers such aspolysaccharides, polyhydroxyaromatics, aminosugar and protein when pyrolyzing refractorycompounds released volatile fragments, which areseparated and analyzed by GC/MS. Those frag-ments are then characterized by relative percen-tages according to their biopolymer. However,the pyrolysis–GC/MS technique is considered asa semi-quantitative technique due to variation of


fragments characteristics with their biopolymerstructures. Besides that, almost 50% of the totalpeaks are classified as miscellaneous, and thestandards used may also not effectively representthe classes of required compounds.

The ATR–FTIR and DOC fractionationmethods using non-ionic macroporous ion-exchange resins such as DAX-8 and XAD-4 havebeen the most popular techniques used by manyresearchers to characterize NOM. But ATR–FTIRmay generate unreliable IR spectra readings dueto overlapping bands. On the other hand, DOCfractionation through ion-exchange resins(DAX-8 and XAD-4) exhibited total DOC re-covery of less than 100%. This can be due to animproper elution procedure of DOC from theresins, or the commercial NOM did not representactual NOM in the natural environment as itvaries with season and origin.

4. Model interpretation

The extent of rejection of solutes by mem-branes is the most critical parameter in membranefiltration. For a clean membrane the extent ofrejection is largely influenced by pore size,whereas for a fouled membrane it is determinedby the electrostatic interactions between thesolute and membrane [6]. Fouling of membranesis likely to happen in many instances due to anumber of mechanisms such as pore blocking ofsolutes, cake deposition and precipitation of inor-ganic and organic particles at the membranesurface (Table 4).

Bowen et al. [57] elucidated the consecutivesteps of membrane blocking in flux declineduring MF as follows: (1) the smallest pores areblocked by all particles arriving to the membrane;(2) the inner surfaces of bigger pores are covered;(3) some particles arriving to the membrane coverother already arrived particles while othersdirectly block some of the pores; and (4) cakestarts to be built.

NOM fouling mechanisms on membrane pro-cesses are different and are dependent on mem-brane type. For MF, pore plugging, pore blockageand cake formation were found responsible forfouling that reduced pore size and increasedrejection. In the case of UF, internal poreadsorption reduced the internal pore diameter andenhanced rejection, while in NF the foulingmechanism was mostly governed by cake deposi-tion and concentration polarization.

The effect of adsorptive fouling on membranefiltration has been well explored and explained bymany researchers [32], but the causes of foulingby dissolved NOM are still not well understood[22]. This is because NOM interactions with amembrane are complex due to many influentialfactors that govern the process such as solutionchemistry (pH, ionic strength and water hard-ness), membrane characteristics (hydrophobicity,surface/pore charge, MWCO and morphology),operational conditions, inorganic presence andNOM characteristics (concentration, humic andnon-humic fractions, charge, MW distribution).The contribution of NOM to fouling dependsgreatly on the natural water properties, charac-teristics of NOM and the type of pretreatment.The fouling phenomenon is caused by particlessmaller than membrane pores being adsorbed intothe membrane pores, then followed by particlesof similar size to pore diameter before cakeformation by deposited particles [58,59].

NOM with a high content of colloidal mattercan expedite fouling by forming a cake on themembrane surface or by precipitating and adsorb-ing (dissolved material) within the membranepores. In the case of accumulation of retainedmaterials on the membrane surface, the hydraulicresistance is related to the permeability of surfacecake while for the adsorbing fouling of DOM, thefiltration resistance is mainly governed by thesize of membrane pores (effective membranepermeability). Altmann and Ripperger [60] statedthat it was more difficult for large particles to foulthe membrane surface than smaller particles. This


Table 4Constant pressure filtration laws [49]

Law Equation Description

Complete blocking(pore blocking)

Intermediate blocking(long-term adsorption)

Standard blocking(direct adsorption)

Cake filtration(boundary layer resistance)

Particles seal pores: dparticle . dpore .Particulates do notaccumulate on each other.

Particles accumulate on each other and seal membranepores: dparticle . dpore.

Particles deposit on the internal pore walls, decreasingthe pore diameter: dparticle < dpore.

Particles are retained due to sieving and form cake on thesurface; deposition occurs on other particles, membranearea is already blocked: dparticle > dpore.

phenomenon is supported by the Kozeny equationthat expresses the specific resistance (Rc) of anincompressible cake. Based on this equation, thecake-specific resistance increases when the dia-meter of deposited particles (dp) decreases in size.

where g is the porosity of gel and dp is the dia-meter of the particle deposited [32].

Membrane filtration performance is accessedthrough permeate flux and rejection parameters.These parameters are strongly correlated to thephysicochemical properties of the feed solution.Unfortunately, surface water contains a complexfeed solution of thousand heterogeneous parti-culate mixtures and soluble components that canlead to significant impact on membrane fouling.Among these components, DOM was found tohave the most detrimental effect on membraneperformance as it could result in irreversiblefouling during surface water filtration. DOM has

been reported as the main cause of irreversiblefouling in surface water filtration [33,61], and thisfouling tendency is governed by solution chemis-try, membrane characteristics and operationalconditions. Laine et al. [61] found DOM adsorp-tion as the most important parameter in mem-brane fouling in lake water, and this was true asmany other researchers also claimed that poly-saccharides, polyhydroxyaromatics, amino sugarsand proteins were the significant fractions foundin the cakes formed during membrane filtration.

Many researchers suggested that humic sub-stances play a vital role in irreversible fouling ofmembranes. Maartens et al. [23] claimed UFmembranes can remove NOM from natural brownwater up to 98%, but this process had an impacton flux decline in the permeate volume that wasdue to an irreversible fouling mechanism. Hydro-phobic interaction between the hydrophobicNOM fraction and a hydrophobic membrane maycause more flux decline than that of hydrophilicmembranes. NOM with a variety of organic frac-tions of different hydrophobicity, hydrophilicity,molecular weight, sizes and charge densities gave

(1)


different interactions in membrane filtration.Yuan and Zydney [25] found that NOM adsorbedboth inside the pores and on the membranesurface to form a cake layer. Cake layer forma-tion is generally known to occur during surfacewater filtration using tight UF, NF and RO whilepore blockage or direct adsorption usuallyhappened using MF.

Cake layer formation is caused by an electro-chemical interaction, and the degree of accumu-lation depends on a balance between convectivetransport of solutes towards the membrane andback-diffusion transport. The transport of largeparticles by drag force (convective force) isgoverned by an orthokinetic mechanism (inertiallift and shear induced diffusion). Inertial liftinduced by the wall effect tends to reduce largerparticles to the membrane especially at high crossflow velocity [32]. Furthermore, shear-induceddiffusion is found to increase back transport ofparticles.

Both inertial lift and shear-induced diffusioninvolving back transport are functions of particlesize. The larger the particle, the higher the possi-bility that it will be back transported [62]. On theother hand, back transport of small particles iscontrolled by Brownian diffusion, which has lesseffect compared to inertial lift and shear-induceddiffusion. Subsequently, large particles in cakestend to produce less resistance for the same massdeposited. Lahoussine-Turcaud et al. [39] de-scribe NOM fouling as primarily governed bypore adsorption and gel formation. In their study,they experienced a 25% flux reduction for thefirst 5 min of 1 nm UF hollow fibre of 10 mg/LHA filtration and a further decline of flux (55%)after 300 min. They concluded that the first 5 minof rapid decline were due to irreversible adsorp-tion of HA foulants. The continued flux declinewas caused by HA gel deposition (reversiblefouling) by convective transport. In anotherNOM–UF study carried out by Chang andBenjamin [63], it was shown that the cake depo-sition was the primary fouling mechanism to their

membrane flux decrement. This was proven by aSEM picture that showed most of the foulingoccurred on the upper membrane surface. In thatstudy they used a 10-nm UF hollow fibre with4 mg/L NOM that gave a flux decline up to 80%after 20 h of filtration.

Some researchers used the zeta potential as afunction of fulvic acid removal and an indicatorfor the degree of fouling mechanisms. Fu andDempsey [64] used the addition of divalent cation(Ca2+) to increase fulvic acid adsorption on amembrane surface. They experienced a decreasein zeta potential as more FA was rejected andincreased in flux decline. The effect of parti-culates and adsorptive fouling on membrane per-formance were well described. However, foulingdue to dissolved NOM–membrane interactionwas not well described and is not yet well under-stood, though there are many reports that it isgenerally controlled by charge interaction andadsorptive behaviour.

Researchers have used different kinds ofhydrophobic membranes and humic substances toprove that membrane fouling is due to adsorptivefouling. However, no NOM fouling was foundwhen the membranes were running at moderatefluxes [44,65]. Cho et al. [66] claimed the ratiobetween flux and mass transfer coefficient wasthe dominant parameter that determines fluxdecline. They found NOM hydrophobicity andmembrane materials did not significantly affectmembrane fouling. Moreover, Howell et al. [67]found that critical flux was useful in reducing thefouling rate, and they also observed that usingintermittent flux substantially helped to reducelong-term fouling.

All these findings support the theory that con-vection and diffusion dominate the extent and rateof flux declination rather than adsorptive orcharge interaction between NOM and mem-branes. Schafer et al. [6] claimed that the rejec-tion characteristics of membranes depend moreon the current fouling state of the membrane andthe nature of the foulant than on initial membrane


characteristics. In their study on NF membranesthey found cation rejection was influenced by thecharge of the precipitates on the membrane sur-face. Cations showed higher rejection comparedto negatively charged low MW acids when themembrane was covered with a positive chargeprecipitate compared to more neutral or negativeprecipitates. Therefore, interaction and combinedeffects of NOM with the membrane matrix,membrane pore blockage and plugging, adsorp-tion, precipitation, deposition and gel cake layerformation were the factors that determined thetypes of fouling mechanisms and the extent offlux reduction.

In summary, NOM adsorption or cake depo-sition on membrane surfaces is the result ofphysical, chemical and electrostatic interactionsbetween NOM molecules and membrane surfaceproperties. In physical adsorption it is caused byattractive dispersion forces that are relativelyweak in comparison, while for chemical adsorp-tion the electrons between NOM and membranesurface molecules are shared, which produce arelatively high strength bond. Electrostaticadsorption occurs between negatively chargedfunctional groups of NOM and the membranecharged polymer.

5. Membrane materials

Membrane properties such as MWCO, poresize, surface charge and hydrophilicity/hydro-phobicity play vital roles in determining therejection, types of retained solutes, rate of fluxreduction and types of membrane foulingmechanisms. A good knowledge of charac-terization of how membranes and foulants inter-act is foremost as this assists in avoiding mem-brane configurations that end up with irreversibleadsorption.

Previous researchers have claimed that ahydrophobic membrane tends to foul more than ahydrophilic membrane while the membrane sur-

face a with positive charge tends to adsorb tonegatively functional groups of NOM, thus in-creasing the potential of fouling. NOM adsorp-tion is greater for hydrophobic membranes suchas polysulphone and polyethersulphone than forhydrophilic membranes.

Laine et al. [61] observed better permeabilityoccurring for more hydrophilic membranes(cellulose) than hydrophobic ones (acrylic) whenrunning a static adsorption of surface water. Theinfluence of hydrophobicity on flux rate does notonly affect the NOM fouling patterns but also themembranes [34]. The study done by Nakatsuka etal. [68] showed that hydrophobic filtration mem-ranes tend to foul, causing a much greater fluxreduction than hydrophilic membranes. Theyfound that cellulose acetate membranes (hydro-philic) had twice as large a flux as polyether-sulfone during UF of river water.

Membranes with negative charge have beenreported to reduce flux decline when the solute isalso negatively charged [69]. Some researchersused chemicals to make the membrane surfacecovered with an anionic charge in order to reduceHA adsorption. Cho et al. [20] found that UFmembranes with a negatively charged surface hada greater NOM rejection than the expectednominal MWCO performance. Although the usedmembrane was intrinsically hydrophobic, theadsorbed NOM found was less in quantity withless fouling.

There are two controlling factors that deter-mine HA adsorption on the membrane surface:the electrostatic repulsion force due to a similarityin charge with the membrane, and adsorptiveproperty due to its high hydrophobicity. Humicsubstances possess significant negative chargeswhich prevent humic substances from beingadsorbed onto the negatively charged membranesurface. However, in many cases reported mem-brane foulants could be hydrophobic materials aswell [25,34,37]. This scenario occurred whenhydrophobic interactions managed to overcomethe electrostatic repulsion force between NOM


and the membrane. Adsorption of NOM on themembrane surface was found to alter membraneproperties (hydrophilicity and surface potentialcharge). Membrane surface potential becamemore negative due to rapid adsorption of NOM(humic fraction) on hydrophilic membranes [70].In addition, Wiesner and Aptel [32] reported thatthe membrane surface became more hydrophilicdue to adsorption of NOM. It is therefore criticalto choose suitable membrane materials that canresult in low hydraulic resistance and high pro-ductivity.

Thus, it can be concluded that fouling ishigher for negatively charged membranes whenthey are exposed to neutral (hydrophilic) mater-ials and protein (bases) compared to high nega-tively charged materials (humic substances)though they possessed high adsorptive behaviourdue to their high hydrophobicity.

6. Fouling controlMembrane efficiency in the filtration opera-

tion depends greatly on tedious management offouling. Membrane fouling is the prime bottle-neck that retards membrane effectiveness andwide application [14]. The usage of suitablefouling control techniques result in longer mem-brane life and lower operational costs. Foulingcontrol comprises physical and chemical proce-dures. Physical methods such as intermittentbackwashing, application of critical flux, criticalTMP, intermittent suction operation, low TMP,high cross flow velocity and the hydrodynamicshear stress scouring effect produce only tempo-rary recovery of membrane flux and require highenergy consumption. On the other hand, theapplication of effective chemical cleaning agentssuch as NaOCl, NaOH, HCl and HNO3 have beenproven to recover completely initial membranepermeability. However, these procedures areexpensive, can cause severe membrane damage,chemical contamination and may produce toxicby-product wastes [4,71].

In practical engineering, chemical cleaning isvery effective in removing the deposited foulantsand can be adopted as a long-term solution forinevitable fouling, but this procedure is out of thefocus of this paper. The backwash technique isdependent on the nature of the fouling mechanismand is only suitable for backflushing a weaklyadhered cake layer. In the case of pore pluggingand pore adsorption (irreversible fouling), theconsumption of chemical agents is more favour-able. Surface water pretreatment prior to mem-brane filtration can be done either by adjustingthe solubility of NOM or reducing the NOM con-centration using precoagulation [48].

Aluminium-based or iron-based coagulantshave long been used to remove NOM in the con-ventional method [26]. Subsequently, coagulationpretreatment prior to membrane filtration wasalso used to enhance permeate quality, as MF andUF alone are inadequate [10,14,22,72,73]. SinceMF/UF have their own limitations due to theirlarger MWCO to the relative molecular mass ofNOM, pretreatment processes such as coagulationand PAC would definitely help to improve theseweaknesses and are capable of meeting waterquality requirements for NOM removal. How-ever, Lahoussine-Turcaud et al. [39] stated thatcoagulation pretreatment could only reduce therate of reversible fouling but not the irreversiblefouling of low MW polysaccharide compounds.Carroll et al. [22] found that coagulation can beused as an efficient pretreatment to improveNOM removal and minimize fouling in MF ofsurface water. Coagulation of colloidal materialand NOM were found to reduce the rate of foul-ing by aggregating fine particles that result inimproving cake permeability, less dense, highlyporous flocs and precipitation or adsorption ofdissolved material into flocs [32]. Increase inparticle size by coagulation helped to reduce foul-ant penetration into pores and formed a higherpermeability cake on the membrane surface [58].In addition, coagulation can also be used toassemble microorganisms with coagulated matter


though it is not as effective as other disinfectantagents.

Maartens et al. [23] suggested alteration of pHand application of metal ions as pretreatmenttechniques of feed water to reduce fouling ofpolysulfone UF membranes caused by naturalbrown water (NBW). NBW with pH 7 was sus-tained at 69% of its original flux after 300 min offiltration, whereas NBW with pH 2 was only33%. In their experiment, using coagulants aspretreatment agents, they hypothesized that thepresence of Al3+ and Ca2+ in the NBW would helpto block the functional groups of NOM by form-ing large precipitated organic materials of metalions and hence influence the potential of adsorp-tive behaviour of NOM membrane-bindingactivities. However, the results of their studyindicated precoagulation with metal ions couldnot prevent membrane fouling, and, as a matter offact, resulted in an increase of NOM adsorptionand a much worse irreversible fouling mechan-ism. They explained that the increased foulingwas due to the greater adsorption of NOM ontothe PSF membrane (40 kD) caused by metal ioncomplexes. This finding was supported by ano-ther study, which also experienced an increase infouling after coagulation [39].

Another chemical addition such as powderedactivated carbon (PAC) has been found to reduceboth cake deposition and DOC adsorption byrapid adsorption of dissolved foulants. PAC helpsto improve hydraulic properties of cake by in-creasing its permeability and reducing com-pressibility. Most researchers [61,74] reportedthat the addition of PAC during membranefiltration helped decrease hydraulic resistance andirreversible fouling. A study done by Koniecznyet al. [75] using PAC and granular activatedcarbon in the hybrid processes and PES foundthat they managed to reduce 99% TOC on cera-mic membranes and 90% on PES membranes.However, Nilson and Digiano [48] and Lin et al.[31] found PAC adsorption to act adversely. Theyclaimed PAC adsorption resulted in large and

more hydrophobic residuals that were detrimentalto membranes. Lin et al. [31] reported the fluxdecline was due to small particles left by PACadsorption.

Operating modes are important factors in opti-mizing membrane filtration. Operating conditionssuch as operating flux, TMP and CFV are thecritical parameters that influence the efficiency ofmembrane filtration. In addition, there was adirect relationship between system performanceand operating flux. Flux decline of constant pres-sure operation or TMP increases of constant fluxoperation can be taken as surrogate parameters ofsystem performances. Many researchers [76,77]reported that running a membrane system withcritical flux resulted in maximum flux withoutsignificant deterioration of system performance.Field et al. [78] defined critical flux as the maxi-mum flux for which little or no fouling occurred.The operation of a membrane below or at thecritical flux helped to reduce the fouling rate [67].This statement was supported by Seidel and Eli-melech [79] who claimed that the high fluxesimposed (40–60 L/m2h), CFV and Ca2+ concen-tration have a significant impact on flux declineduring 50 h of filtration operation. Thorsen [69]suggested that there is a critical flux at which thefouling rate starts to change with conditions athigher fluxes.

The fouling phenomenon is a function of con-vection and diffusion of solutes away from themembrane and is found to be linked with thehydrodynamic conditions of the system. Thehydrodynamics of CFV play an important role ininfluencing NOM fouling. The hydrodynamics ofcrossflow membrane filtration, especially CFV,was observed to be capable of reducing NOMfouling at high CFV. CFV drags the back-transported particles off the membrane surface.The increment of CFV helps to increase thephysical scouring effect at the membrane surfaceand improve back-transport into the bulk solu-tion. Shear stress generated from this CFV helpsto reduce the rate of cake deposition and the


Fig. 5. Effect of CFV on hydraulic resistance [80].

resultant cake thickness. However, flux as a func-tion of CFV is dependent on the imposed CFVvalue and particle size factors. Thomas et al. [80]reported that the increase of CFV above 3 m/sresulted in increased fouling resistance (Fig. 5).This was likely due to a more compact foulinglayer and increased pore plugging caused byhigher pressure. Previous study showed that CFVincreased the small particle flux better than largerparticles [71,81]. This phenomenon was due toparticle classification at the membrane surface.

Increasing CFV with increment pump speedalso generates finer colloidal particles, which canform a denser cake layer on the membrane sur-face. For an immersed membrane reactor the CFVis replaced by bubble aeration. The bubble actsdirectly against the fibres by colliding with andscouring the foulants. The bubble creates hydro-dynamic effects that suppress the fouling layer.However, the efficiency of bubble aeration relieson the applied flow rate. A study done by Bou-habila et al. [82] showed that the air flow rate hada significant effect on hydraulic resistance, butthis effect only worked for higher filtrate fluxes(Fig. 6).

Fig. 6. Variation of hydraulic resistance with filtrate fluxfor various CFVs [82].

Hong and Elimelech [3] showed that the NOMfouling rate was substantially increased with anincreasing initial permeation rate, particle con-centration and TMP. This fouling condition wasworsened when the operation was carried out forlong hours.

7. Future direction of membrane filtration forsurface water treatment

Membrane technology for surface water treat-ment is considered to be the most promisingdevelopment in water treatment as it could pro-vide permeate quality far beyond the currentregulatory requirement for potable water con-sumption. This technology becomes even moreattractive when compact technology (small foot-print) and careful water management are requiredwith a critical water shortage problem to address.With the current technology, problems such aslow membrane life, expensive membranes thatare prone to fouling and questionable permeatequality are easily eased out. Currently, membranepotential has been specifically manipulated totheir best application (Table 5).

However, there are still many technical chal-lenges to optimize and make membrane tech-nology more competitive in the market, especiallyfor large-scale industries. There is a need for


Table 5Specific use of membrane based on types [83]

Membrane type Application

MFUFNFROMBRHybrid

Removes turbidity and microorganismsRemoves viruses and large macromolecules recovery in water recyclingRemoves colour, large ions and higher flux than RO for surface water treatmentPotable use and industrial reuse removal of small ionsTreat difficult waste possibly with specific organisms for special wasteUV–ozone–precoagulation or combined membrane systems becoming much effective than a single membrane

further research of advanced membrane materialsthat are resistant to both chemical and mechanicalattacks during surface water treatment as thiswould help in prolonging the membrane lifespanand induce long-term performance. The identifi-cation of the best practices in terms of design,treatment configuration and operating parameterswould help to project minimal capital for design,construction and operational costs. The develop-ment of clear fundamental knowledge on under-standing and minimization of membrane fouling(due to NOM and particulates in surface water) isalso vital and can be realized through usingaccurate process design and operating conditions.Furthermore, the proper selection of pretreat-ments, improvement of cleaning strategies andmembrane systems with low energy requirementsmay help to position this technology to gain theconfidence of the market.

8. ConclusionsHeterogeneity of NOM in surface water has

made the nature and extent of fouling moredifficult to control and predict. Membrane foulingis a phenomenon that cannot be prevented but canbe minimized. Fouling of NOM occurs due tomany factors and mechanisms. The factors affect-ing NOM and membrane interactions includeNOM characteristics, operating conditions, mem-brane characteristics and solution chemistry.

NOM fouling occurs when dissolved organic orinorganic solute is adsorbed or deposited on themembrane. The adsorption mechanism happensmore quickly compared to cake formation butdepends on the membrane properties, ionicstrength, pH and presence of divalent cations.

Solute deposition or gel formation occurs inparallel with the magnitude of convective fluxand the extent of concentration polarization.Hydrophobicity and electrostatic interactionsbetween solute and membrane are also reported tobe the dominant factors that affect the extent ofNOM fouling. The presence of electrolyte com-position, low pH and high ionic strength has beenfound to strongly enhance the degree and rate offouling.

MF and UF are drinking water treatment pro-cesses that are particularly suitable for the remo-val of suspended solids and colloidal materialssuch as bacteria, algae, protozoa and inorganicparticulates. However, these filtration types areless successful for the removal of dissolved con-taminants, especially NOM in surface water.Coagulation was introduced to address this weak-ness as it has proven to be effective for decreas-ing hydraulic resistance, increasing critical fluxand improving NOM removal.

There is still controversy over how NOMaffects membrane fouling mechanisms. Somestudies suggested that charge interaction andadsorptive behaviours are the responsible factors


that control NOM fouling, whereas others claimconvective and diffusive particle transport thatmainly dominate fouling in NOM filtration. As amatter of fact, many previous studies were doneusing various types of membranes and wereoperated at high fluxes in their experiments withregard to NOM fouling. These conditions con-tributed to physical accumulation due to con-vection, diffusion and adsorptive fouling. As aresult, it was difficult to distinguish the dominantfactor that was responsible for the fouling. Hencefurther study needs to be carried out in order toclarify this ambiguity and help in the properselection of membrane properties and configura-tions, pretreatment, filtration process, operatingconditions and cleaning conditions of surfacewater filtration so that the process can beoperated economically and successfully.

References[1] M. Clever, F. Jordt, R. Knauf, N. Rabiger, R.

Rudebusch and H. Scheibel, Process water pro-duction from river by ultrafiltration and reverseosmosis. Desalination, 131 (2000) 325–336.

[2] J. Mallevialle, C. Anselme and O. Marsigny, Effectof humic substances on membrane process. AquaticHumic Substances, Proc. 193th meeting, AmericanChemical Society, Denver, 1996, pp. 749–767.

[3] S. Hong, and M. Elimelech, Chemical and physicalaspects of natural organic matter (NOM) fouling ofnanofiltration membrane. J. Membrane Sci., 132(1997) 152–181.

[4] A. Maartens, P. Swart and E.P. Jacobs, Humicmembrane foulants in natural brown water: charac-terization and removal. Desalination, 153 (1998)215–227.

[5] M.L. Manttari, J. Puro, N. Jokinen and M. Nystrom,Fouling effect of polysaccharides and humic acid innanofiltration. J. Membr. Sci., 165 (2000) 1–17.

[6] A.I. Schafer, A.G. Fane and T.D. Waite, Foulingeffect on rejection in the membrane filtration ofnatural waters. Desalination, 131 (2000) 215–224.

[7] J.H. Kweon and D.F. Lawler, Fouling mechanisms inthe integrated system with softening and ultra-filtration. Water Res., 38 (2004) 4164–4172.

[8] S.R. Gray, C.B. Ritchie and B.A. Bolto, Effect offractionated NOM on low pressure membrane fluxdeclines. Water Sci. Technol., 4 (2004) 189–196.

[9] N. Lee, G. Amy, J.P. Croue and H. Buisson,Identification and understanding of fouling in low-pressure membrane (MF/UF) filtration by naturalorganic matter (NOM). Water Res., 38 (2004)4511–4523.

[10] S. Xia, J. Nan, R. Liu and G. Li, Study of drinkingwater treatment by ultrafiltration of surface water andits application to China. Desalination, 170 (2004)41–47.

[11] T. Leiknes, H. Odegaard and H. Myklebust, Removalof natural organic matter (NOM) in drinking watertreatment by coagulation–microfiltration using metalmembranes, J. Membr. Sci., 242 (2004) 47–55.

[12] D. Violleau, H.E. Tome, H. Habarou, J.P. Croue andM. Pontie, Fouling studies of a polyamide nano-filtration membrane by selected natural organicmatter: an analytical approach. Desalination, 173(2004) 223–238.

[13] F.H. Frimmel, F. Saravia and Gorenflo, NOMremoval from different raw waters by membranefiltration. Wat. Sci. Technol., 4 (2004) 165–174.

[14] L. Mo and X. Huang, Fouling characteristics andcleaning strategies in a coagulation–microfiltrationcombination process for water purification. Desali-nation, 159 (2003) 1–9.

[15] S. Mozia and M. Tomaszewska, Treatment of surfacewater using hybrid processes — adsorption on PACand ultrafiltration. Desalination, 162 (2004) 23–31.

[16] K.J. Howe, K.P. Ishida and M.M. Clark, Use ofATR/FTIR spectrometry to study fouling of micro-filtration membranes by naturals waters. Desali-nation, 147 (2002) 251–255.

[17] P. Park, C.H. Lee, S.J Choi, K.H. Choo, S.H. Kimand C.H. Yoon, Effect of the removal of DOMs onthe performance of a coagulation–UF membranesystem for drinking water production. Desalination,145 (2002) 237–245.

[18] H. Lee, G. Amy, J. Cho, Y. Yoon, S.H. Moon andS.I. Kim, Cleaning strategies for flux recovery of anultrafiltration membrane fouled by natural organicmatter. Water Res., 35 (2001) 3301–3308.

[19] L. Fan, J.L. Harris, F.A. Roddick and N.A. Booker,Influence of the characteristics of natural organicmatter on the fouling of microfiltration membranes.Water Res., 35 (18) (2001) 4455–4463.


[20] J. Cho, G. Amy and J. Pellegrino, Membrane fil-tration of natural organic matter: factors and mechan-isms affecting rejection and flux decline withcharged ultrafiltration (UF) membrane. J. Membr.Sci., 164 (2000) 89–110.

[21] N. Her, G. Amy and C. Jarusutthirak, Seasonalvariation of nanofiltration (NF) foulants: identifi-cation and control. Desalination, 132 (2000) 143–160.

[22] T. Carroll, S. King, S.R. Gray, B.A. Bolto and N.A.Booker, The fouling of microfiltration by NOM aftercoagulation treatment. Water Sci., 34 (1999) 2861–2868.

[23] A. Maartens, P. Swart and E.P. Jacobs, Feedwaterpretreatment methods to reduce membrane foulingby natural organic matter. J. Membr. Sci., 162 (1999)51–62.

[24] C. Lin, T. Lin and O.J. Hao, Effects of humicsubstance characteristics on UF performance. WaterRes., 34(4) (1999) 1097–1106.

[25] W. Yuan and A.L. Zydney, Humic acid foulingduring microfiltration. J. Membr. Sci., 157 (1999)1–12.

[26] G.F. Crozes, J.G. Jacangelo, C. Anselme and J.M.Laine, Impact of ultrafiltration operating conditionson membrane irreversible fouling. J. Membr. Sci.,124 (1997) 63–76.

[27] Y. Kaiya, Y. Itoh, K. Fujita and S. Takizawa, Studyon fouling materials in the membrane treatmentprocess for potable water. Desalination, 106 (1996)71–77.

[28] Y. Choi, Critical flux, resistance and removal ofcontaminants in ultrafiltration (UF) of naturalorganic materials. PhD Thesis, Pennsylvania StateUniversity, 2003.

[29] E.M. Thurman, Organic Geochemistry of NaturalWaters, Martinus Nijhoff/Dr W. Junjk, Dordrecht,1985.

[30] J.A. Leenheer, Chemistry of dissolved organic matterin rivers, lakes and reservoirs. Amer. Chem. Soc.,237 (1994) 196–221.

[31] C. Combe, E. Molis, P. Lucas, R. Riley and M.Clark, The effect of CA membrane properties onadsorptive fouling by humic acid. J. Membr. Sci.,154 (1999) 73–87.

[32] M.R. Wiesner and P. Aptel, Mass transport andpermeate flux and fouling in pressure driven process.

AWWA. Water Treatment: Membrane Processes.McGraw-Hill, New York, 1996.

[33] J. Mallevialle, C. Anselme and O. Marsigny, Effectof humic substances on membrane process, in: I.H.Suffet and P. MacCarthy, eds., Aquatic HumicSubstances, ACS, Washington, DC, 1989, pp. 749–767.

[34] J.A. Nilson and F.A. DiGiano, Influence of NOMcomposition on nanofiltration. J. AWWA., 88(5)(1996) 53–66.

[35] F.J. Stevenson, Humus Chemistry. Wiley, NewYork, 1982.

[36] J.A.E. Buffle, Les substances humiques et leursinteractions avec les ions mineraux. Proc. Commis-sion d’hydrologie appliquee de AGHTM l’Universited’Orsay, 1977, pp. 3–10.

[37] K.L. Jones and C.R. O’ Melia, Protein and humicacid adsorption onto hydrophilic membrane surfaces:effect of pH and ionic strength. J. Membr. Sci., 165(2000) 31–46.

[38] A.I. Schafer, U. Schwicker, M.M. Fisher, A.G. Faneand T.D. Waite, Microfiltration of colloids andnatural organic matter. J. Membr. Sci., 171 (2001)151–172.

[39] V. Lahoussine-Turcaud, M.R. Wiesner and J.-Y.Bottero, Fouling in tangential-flow ultrafiltration: theeffect of colloids size and coagulation pretreatment.J. Membr. Sci., 52 (1990) 173.

[40] J. Lee, S. Lee, M. Jo, P. Park, C. Lee and J. Kwak,Effect of coagulation conditions on membrane fil-tration characteristics in coagulation–micro-filtration.Environ. Sci. Technol., 34 (2002) 3780– 3788.

[41] J. Cho, G. Amy and J. Pellegrino, Membranefiltration of natural organic matter: comparison offlux decline, NOM rejection and foulant duringfiltration with three UF membranes. Desalination,127 (1999) 283–298.

[42] T. Speth, A.M. Guses and R.S. Summers, Evaluationof nanofiltration pretreatments for flux loss control.AWWA, Mem. Tech. Conf., Long Beach, CA.,Vol. 7(2), 2000.

[43] C. Jarusutthirak, G. Amy and J.P. Croue, Foulingcharateristics of wastewater effluent organic matter(EFOM) isolates on NF and UF membranes.Desalination, 145 (2002) 247–255.

[44] T. Thorsen, Concentration polarisation by naturalorganic matter (NOM) in NF and UF. J. Membr. Sci.,233(1–2) (2004) 79–91.


[45] A.H. Braghetta, The influence of solution chemistryand operating conditions on nanofiltration of chargedand uncharged organic macromolecules. PhD Thesis,University of North Carolina, Chapel Hill, N.C,1995.

[46] A. Braghetta, F.A. DiGiano and W.P. Ball, Nano-filtration of natural organic matters: pH and ionicstrength effects. J. Env. Eng., 123(7) (1997) 628–641.

[47] K. Ghosh and M. Schnitzer, Macromolecular struc-ture of humic substances. Soil Sci., 129 (1980) 266–276.

[48] C. Jucker and M.M. Clark, Adsorption of aquatichumic substances on hydrophobic ultrafiltrationmembranes. J. Membr. Sci., 97 (1994) 37–52.

[49] E. Aoustin, A.I. Schafer, A.G. Fane and T.D. Waite,Ultrafiltration of natural organic matter. Sep. Purif.Technol., 22–23 (2001) 63–78.

[50] S. Cohen, M.A. Fleer, G.J. Lyklema, J. Norde andJ.M.H.M. Scheutjens, Adsorption of ions, poly-electrolytes and proteins. Adv. Coll. Interface Sci.,34 (1991) 477–535.

[51] M. Bob and H.W. Walker, Enhanced adsorption ofnatural organic matter on calcium carbonate particlesthrough surface charge modification. Coll. Surface:Physicochemical Eng. Aspects, 191 (2001) 17–25.

[52] C. Jarrusutthirak, Fouling and flux decline of reverseosmosis (RO), nanofiltration (NF) and ultrafiltration(UF) membranes associated with effluent organicmatter (EFOM) during wastewater reclamation/reuse. PhD Thesis, University of Colorado at Boul-der, 2002.

[53] G.V. Korshin, C.W. Li and M.M. Benjamin,Monitoring the properties of natural organic matterthrough UV spectroscopy: a consistent theory. WaterRes., 31(7) (1997) 1787–1795.

[54] S.W.Krasner, J.P. Croue, J. Buffle and E.M. Purdue,Three approaches for characterizing NOM. J.AWWA, 88 (1996) 66–79.

[55] G.R. Aiken, D.M. McKnight, K.A. Thorn and E.M.Thurman, Isolation of hydrophilic organic acids fromwater using non-ionic macroporous resin. Org.Geochem., 18(4) (1992) 567–573.

[56] A. Bruchet, C. Rosseau and J. Mallevialle, Pyro-lysis–GC–MS for investigating high-molecularweight THM precursors and other refractoryorganics. J. AWWA., 82(9) (1990) 66–74.

[57] W.R. Bowen, J.I. Calvo and A. Hernandez, Steps ofmembrane blocking in flux decline during proteinmicrofiltration. J. Membr. Sci., 101 (1995) 153–165.

[58] D. Sakol and K. Konieczny, Application of coagu-lation and conventional filtration in raw waterpretreatment before microfiltration membrane.Desalination, 162 (2003) 61–73.

[59] J. Jacob, P. Pradanos, A. Calvo, A. Hernadez andG. Jonsson, Fouling kinetics and associateddynamics of structural modifications. Colloids Surf.A., 138 (1998) 173–183.

[60] J. Altmann and S. Ripperger, Particle deposition andlayer formation at crossflow microfiltration. J.Membr. Sci., 124 (1997) 119–128.

[61] J.M. Laine, J.P. Hagstrom, M.M. Clark and J.Mallevialle, Effect of ultrafiltration membrane com-position. J. AWWA, 81(11) (1989) 61–67.

[62] S. Chellam and M.R. Wiesner, Particle back-transport and permeate flux behaviour in crossflowmembrane filters. Envi. Sci. Technol., 31 (1997)819–824.

[63] Y. Chang and M.M. Benjamin, Iron oxide adsorptionand UF to remove NOM and control fouling. J.AWWA, 90 (1996) 74–88.

[64] L.F. Fu and B.A. Dempsey, Effects of charge andcoagulant dose on NOM removal and membranefouling mechanisms. Proc. Membrane TechnologyConference, AWWA, LA, 1997, pp. 1043–1058.

[65] M.M. Dal-Cin, F. McFellan, C.N. Striez, C.M. Tam,T.A. Tweddle and A. Kumar, Membrane perfor-mance with pulp mill effluent: relative contributionof fouling mechanisms. J. Membr. Sci., 120 (1996)273.

[66] J. Cho, J. Sohn, H. Choi and G. Amy, Effects ofmolecular weight cutoff f/k ratio (a hydrodynamiccondition) and hydrophobic interactions on naturalmatter rejection and fouling in membranes. Aqua.,551(2) (2002) 109.

[67] J.A. Howell, T.C. Arnot, H.C. Chua, P. Godino andS. Metsamuuronen, Controlled flux behaviour ofmembrane processes. Macromol. Symp., 188 (2002)23.

[68] S. Nakatsuka, L. Nakate and T. Miyano, Drinkingwater treatment by using ultrafiltration hollow fibremembranes. Desalination, 106 (1996) 55–61.

[69] A. Nabe, E. Staude and G. Belfort, Surface modi-fication of polysulfone ultrafiltration membrane and


fouling by BSA solutions. J. Membr. Sci., 133(1997) 57–72.

[70] A.E. Childress and M. Elimelech, Effect of solutionchemistry on the surface charge of polymeric reverseosmosis and nanofiltration membranes. J. Membr.,Sci., 119 (1996) 253–268.

[71] I.C. Soung, L.P. Clech, J. Bruce and S. Judd, Mem-brane fouling in membrane bioreactors for waste-water. ASCE., 128(11) (2002) 1018.

[72] M. Hiraide, Heavy metals complexes with humicsubstances in fresh water. Anal. Sci., 8 (1992) 47–54.

[73] J.C. Vickers, M.A. Thompson and U.G. Kelkar, Theuse of membrane filtration with coagulation pro-cesses for improved NOM removal. Desalination,102 (1995) 57–61.

[74] J.G. Jacangelo, R.R. Trusell and M. Watson, Role ofmembrane technology in drinking water treatment inUnited States. Desalination, 113 (1997) 119–127.

[75] K. Konieczny and K. Grzegovz, Using activatedcarbon to improve natural water treatment by porousmembranes. Desalination, 147 (2002) 109–116.

[76] J.A. Howell, Sub-critical flux operation of micro-filtration. J. Membr. Sci., 107 (1995) 165–171.

[77] L. Defrance and M.Y. Jaffrin, Comparison betweenfiltrations at fixed transmembrane pressure and fixed

permeate flux: application to a membrane bioreactorused for wastewater treatment. J. Membr. Sci., 152(1999) 203–210.

[78] R.W. Field, D. Wu, J.A. Howell and B.B. Gupta,Critical flux concept for microfiltration fouling.J. Membr. Sci., 100 (1995) 259–272.

[79] A. Seidel and M. Elimelech, Coupling betweenchemical and physical interactions in natural organicmatter (NOM) fouling of nanofiltration membranes:implications for fouling control. J. Membr. Sci., 203(2002) 245.

[80] H. Thomas, S. Judd and J. Murrer, Fouling charac-teristics of membrane filtration in membrane bio-reactors. Membr. Technol., 122 (2000) 10–13.

[81] E.S. Tarleton and R.J. Wakeman, Understanding fluxdecline in cross-flow microfiltration. 1. Effect ofparticle and pore size. Chem. Eng. Res. Des., 71(1993) 399–410.

[82] E.H. Bouhabila, B.R. Aim and H. Buisson, Foulingcharacterization in membrane bioreactors. Sep. Purif.Tech., 22–23 (2001) 123–132.

[83] J.A. Howell, Future of membrane and membranereactors in green technologies and for water reuse.Desalination, 162 (2004) 1–11.

behaviours of natural organic matter in membrane filtration for surface water treatment

Documents