Journal of Membrane Science 433 (2013) 112–120

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science

0376-73

http://d

n Corr

Environ

Singapo

E-m

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

The fouling potential of colloidal silica and humic acid and their mixtures

Amir Hooshang Taheri a,b, Lee Nuang Sim a,b, Chong Tzyy Haur a, Ebrahim Akhondi a,b,Anthony Gordon Fane a,b,n

a Singapore Membrane Technology Centre, Nanyang Environment & Water Research Institute, Nanyang Technological University, Singapore 639798, Singaporeb School of Civil and Environmental Engineering, Nanyang Technological University, Singapore 639798, Singapore

a r t i c l e i n f o

Article history:

Received 21 October 2012

Received in revised form

17 January 2013

Accepted 22 January 2013Available online 29 January 2013

Keywords:

MFIUF

Specific cake resistance

Particle concentration

Mixed foulants

88/$ - see front matter & 2013 Elsevier B.V. A

x.doi.org/10.1016/j.memsci.2013.01.034

esponding author at: Singapore Membrane

ment & Water Research Institute, Nanyan

re 639798, Singapore. Tel.: þ65 6790 5272;

ail address: AGFane@ntu.edu.sg (A.G. Fane).

a b s t r a c t

The fouling propensity of an inorganic colloid (silica) and an organic (humic acid) and their mixture was

studied in this paper. Fouling propensity was characterized as the modified fouling index (MFIUF) and

specific cake resistance in dead-end UF at constant pressure. Experimental results for individual

foulants demonstrated that MFIUF increased linearly with increasing particle concentration for both

humic acid and colloidal silica over most of the concentration range, with deviations at the lowest

concentrations. In terms of specific cake resistance, the individual foulants tended to have higher values

at low concentrations that gradually declined to steady values at higher concentrations. These trends

could be due to differences in cake formation and packing density with foulant flux. For the mixed

foulants the presence of modest amounts of humic acid (o15 mg/L) tended to reduce both MFIUF and

specific resistance below that of the colloidal silica alone (50–150 mg/L). This is attributed to the effect

of the humic acid on silica packing density. At higher concentrations of humic acid (420 mg/L) or

lower concentrations of silica the mixture MFIUF started to exceed that of the colloid alone, possibly due

to interstitial humic acid effects on cake resistance. Very low adsorption of humic acid on colloidal silica

was also observed using Quartz crystal microbalance with dissipation (QCM-D) and zeta potential

measurements. The results confirm that knowledge of the individual MFIUFs could not be used reliably

to predict the fouling potential of the mixture.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

Ultrafiltration processes (UF) are popular for water and waste-water treatment with a relatively lower cost compared with othermethods like nanofiltration (NF) or reverse osmosis (RO) [1,2].It is also used increasingly as a pretreatment step for RO units inseawater desalination and water reclamation [3]. However, one ofthe most challenging problems in membrane processes is foulingwhich decreases the performance and increases the operationalcosts [4]. Membrane fouling can be affected by many parameterslike feed water characteristics and solution chemistry (pH, salinity,etc.), membrane properties and operational conditions [5].

There are several methods to determine the particulate foulingpotential of feed water. The most frequent and widely usedmethod is the silt density index (SDI) which is based on0.45 mm microfiltration (MF) membranes [6]. This method is anempirical method without a good theoretical basis [7]. Themodified fouling index (MFI0.45) has advantages over the SDI as

ll rights reserved.

Technology Centre, Nanyang

g Technological University,

fax: þ65 6791 0676.

it is based on filtration theory and is typically linear with thefoulant load [7–9]. However the MF based MFI0.45 may notcapture all potential foulants, and the modified fouling indexultrafiltration (MFIUF) has been proposed as an alternative tocapture smaller colloidal particles than that by the MFI0.45

[10–12]. To capture even smaller particles and macromoleculesthe MFI-NF was proposed, but found to take longer and producesimilar results to MFIUF [13]. Ultrafiltration membranes appear tobe a good compromise for measurement of fouling potential byMFIUF. The MFIUF responds to fine particles, colloids and macro-molecular organics provides and an indication of the foulingpropensity for UF membranes and RO membranes (other foulingtypes, such as scale and biofouling require other fouling monitors)[14–16]. During colloidal fouling of porous membranes, particlesaccumulate on the surface or within pores [17–20]. Humic sub-stances, which are considered as a main component of dissolvednatural organic matter (NOM) in surface waters, have a significantrole in organic fouling of membranes. Organic fouling causes bothreversible and irreversible flux decline, thus removing NOM is animportant step in many pre-treatment processes [21]. Investiga-tions of the influence of NOM on membrane fouling havebeen carried out by several researchers [22–25]. For example,the ultrafiltration of natural water with various organic

A.H. Taheri et al. / Journal of Membrane Science 433 (2013) 112–120 113

macromolecules and inorganic colloids indicated that organicmatter had a much greater effect on the flux decline than inorganiccolloids which can be attributed to the different fouling mechan-isms, including pore plugging or cake formation [26].

Some studies on membrane filtration of combined foulantshave been reported recently and suggested that NOM plays animportant role for particle aggregation or stabilization duringfiltration of combined foulants [27–29].

Most fouling index studies have concentrated on individualorganic or inorganic foulants and the effects of combined foulantson membrane filtration remain relatively poorly understood. There-fore the objective of this paper is to study the influence ofconcentration for both colloidal (silica) and organic (humic) fou-lants, alone and in mixtures, on membrane fouling potential underconstant pressure filtration. The trends in MFIUF and specific cakeresistances are also compared.

2. Theory

MFIUF has been developed based on cake filtration theory [7].Schippers et al. [7] proposed that dead-end filtration at constantpressure takes place in three steps in sequence: pore blocking,cake filtration without compression and formation of a compres-sible cake layer.

According to the membrane blocking laws, there is a linearrelationship between t/V and V during the cake filtration periodwhere V refers to the filtrate volume and t is the filtration time.At the beginning of filtration, membrane resistance (Rm) controls theflux, however after cake formation; the cake resistance (RC) becomesincreasingly important. For filtration under constant pressure thefollowing equation shows the relationship between t/V and V:

t

V¼mRm

DPAþ

mI

2DPA2� V ð1Þ

where A is membrane area, DP is the transmembrane pressure (TMP)and m is filtrate solution viscosity.

The slope of the linear region of t/V vs. V is defined as theMFIUF:

MFIUF ¼mI

2DPA2ð2Þ

Fig. 1. Schematic diagram of dead-end

where ‘I’ is the fouling index or resistivity, which is a measure ofthe fouling potential of the feedwater for the membrane, andcombines the specific cake resistance (a) and the foulant con-centration (Cb), so [30]:

IðtÞ ¼ aðtÞCb ð3Þ

The specific cake resistance under constant pressure is con-stant for incompressible cakes and can be estimated using theCarman–Kozeny relationship [31]:

a¼ 180ð1�eÞrpd2

pe3ð4Þ

Here e is the porosity of the cake layer and dp and rp are thediameter and density of the particles, respectively. Eq. (4) isqualitatively useful as it illustrates the sensitivity of a (and henceMFIUF) to foulant size and porosity (e) of the formed cake. It alsohighlights the challenge to determine specific cake resistance formixed foulants as it is difficult to assume a characteristic ‘‘dp’’value or porosity for a mixture.

By substituting Eq. (3) in Eq. (2), the relationship betweenspecific cake resistance and MFIUF can be measured by followingequation:

a¼ 2MFIUFDPA2

mCbð5Þ

3. Materials and methods

3.1. Filtration apparatus

Fig. 1 is a schematic diagram of the dead-end filtration setupused to measure the MFIUF. A stainless steel feed reservoir (2 lcapacity) was connected to a dead-end filtration cell (maximumcapacity of 120 ml) with an active membrane area of 0.0012 m2.The required trans-membrane pressure (TMP) in constant pres-sure filtration was supplied via a nitrogen gas cylinder. Thepermeate volume and TMP were measured and recorded withan electronic weighing balance (Mettler Toledo, model PB3002-L)and pressure transducer, respectively, which were connected tothe data acquisition system (National Instrument, model PCI 6014and LabView 7). All filtration tests were conducted at standardconditions; 20 1C and 210 kPa.

constant pressure filtration device.

A.H. Taheri et al. / Journal of Membrane Science 433 (2013) 112–120114

Several experiments were repeated up to 3 times to ensure thereproducibility of the measurements. Relative errors were within710% for both MFIUF and specific cake resistance measurements.

3.2. UF membranes

The pure water (Milli-Q) flux was first measured in thefiltration experiments to stabilize the membrane and to obtainthe intrinsic membrane resistance (Rm). Ultrafiltration mem-branes (regenerated cellulose—RC) from Millipore (PLTK15005),with molecular weight cut off (MWCO) of 30 kDa were used asthe MFIUF test membranes with intrinsic resistance of 9.11�1011m�1 (standard deviation¼3.74�1010m�1). A new mem-brane was used for each filtration test.

3.3. Model foulants

Colloidal silica (Sigma Aldrich, Ludox TMA) and humic acid(Sigma-Aldrich) were used as fouling agents in the experiments.The colloidal silica was supplied in the form of 34 wt% suspensionin deionised water, with a particle size of 20 nm as reported bythe manufacturer. All suspensions were prepared using Milli-Qwater at pH 6.5.

3.4. Zeta potential measurement

The surface zeta potentials of the humic acid, colloidal silicaand their mixtures were measured using a Malvern ZetaSizerNano ZA. To avoid any interference during measurement bybubbles, samples were slowly injected into the zeta potential celland each test was repeated three times.

3.5. Measurement of humic acid adsorption on silica

Quartz crystal microbalance with dissipation (QCM-D) mea-surements were conducted to study and quantify any interaction

Table 1Surface zeta potential of model foulantsn.

Model foulants Zeta potential (mV)

Colloidal silica (100 mg/L) �39.672.3

Humic acid (15 mg/L) �59.972.2

Colloidal silica (100 mg/L)þhumic acid (15 mg/L) �52.471.4

n Reported values are the average of three measurements with standard deviation.

0

2

4

6

8

10

12

14

16

0 10 20 30

Ads

orbe

d am

ount

(ng/

m2 )

Tim

Fig. 2. Amounts of humic acid adsorbed on silica-coa

between humic acid and colloidal silica. QCM-D can provide a realtime measurement of organic adsorption on inorganic surfaces[32,33]. Silica-coated quartz crystals sensors with 14 mm dia-meter (Q-Sense, QSX 301, Sweden) and fundamental resonantfrequency of 4.95 MHz, were used to simulate the surface of thecolloidal silica foulant. Before each experiment, the sensor waspre-cleaned using 2% sodium dodecyl sulfate (SDS) solution andthen rinsed by Milli-Q water followed by drying with ultrapureN2 gas.

To verify the fundamental frequency of each crystal, experi-ments were first conducted under dry air conditions followed byMilli-Q water for 10 min to stabilize the sensor. Then humic acidsolution (15 mg/L) was fed into the flow chamber at a flow rate of10 ml/min for 60 min and any changes of resonance frequency andenergy dissipation measured. According to the Sauerbrey Eq. (6)[34], the change in frequency is related to the adsorbed amount ofhumic acid on the sensor per unit surface.

DF ¼�2f 2

0

Affiffiffiffiffiffiffiffiffiffiffirqmq

p Dm ð6Þ

where Dm is change in mass adsorbed (kg) on the crystal sensor, f

is the frequency (Hz), f0 is the resonant frequency (Hz) of thesensor, Aq is the piezoelectrically active crystal area (m2), rq is thedensity of quartz (kg/m3), and mq is the shear modulus of thequartz (Pa).

4. Results and discussion

The effect of various parameters on MFIUF and specific cakeresistance under constant pressure were investigated and ana-lyzed using the equations shown in the previous section.

4.1. Characterization of model foulants

4.1.1. Zeta Potentials

Surface zeta potentials of humic acid, colloidal silica and theirmixture were measured under the same solution conditions usedin the experiments (pH�7) and are summarized in Table 1.

Silica (100 mg/L) had a negative surface zeta potential of�39.672.3 mV whereas humic acid (15 mg/L) had a significantlyhigher negative surface zeta potential -59.972.2 mV. The surfacezeta potential for the mixture of humic acid (15 mg/L) andcolloidal silica (100 mg/L) was measured as �52.471.4 whichis more negatively charged than colloidal silica, but less nega-tively charged than humic acid alone. This is assumed to be due to

40 50 60 70e (min)

ted quartz crystals sensor as a function of time.

A.H. Taheri et al. / Journal of Membrane Science 433 (2013) 112–120 115

the interaction of humic acid and silica that resulted in changes tothe surface properties of the colloidal silica.

4.1.2. Adsorption by QCMD

Fig. 2 shows the amounts of humic acid adsorbed onto thesilica-coated quartz crystal sensor.

According to the surface zeta potential measurements, both thesilica and the humic acid had significant negative charge, soelectrostatic repulsion between these model foulants should limitadsorption of HA on silica. This finding is in good agreement withContreras et al. [35] who observed very low adsorption of humicacid on the surface of colloidal silica. To put the amount adsorbedinto perspective a monolayer of humic acid on the silica particlescould be equivalent to about 1–10 mg/m2, which is orders ofmagnitude higher than the measured value (�10 ng/m2). Howeverthe data in Table 1 implies some surface interaction.

4.2. Individual foulants

4.2.1. Determination of MFIUF value

The typical results from plotting t/V as a function of V for100 mg/L silica and 15 mg/L humic acid are plotted in Fig. 3.

According to Eq. (2), MFIUF at constant pressure refers to theslope of (t/V) vs. (V). In this case, silica and humic acid have MFIUF

values of 186,058 s/L2 and 16,713 s/L2 respectively.

4.2.2. Inorganic fouling experiments with silica

Filtration experiments were conducted with UF membranes(RC 30 kDa) to study inorganic fouling by MFIUF. The silicaconcentration was in the range of 15–800 mg/L, and a constantpressure of 210 kPa was applied.

The effect of silica concentration on the MFIUF is presented inFig. 4, and the results confirmed the direct relation of themodified fouling index (MFIUF) value to the concentration ofparticles as anticipated by Eq. (3). Similar trends for colloidalsilica particles, humic acid and formazine under constant pressurefiltration have been observed by others [7,16,25,36,37]. Howeverthe trend in the lower concentration ranges (below 100 mg/L)was non-linear, and this is more evident in the specific resis-tances. Using the MFIUF values, it is possible to obtain specificcake resistance (a) based on Eq. (5). Fig. 5 shows specific cakeresistance as a function of silica concentration. The specific cakeresistance shows a rapid increase and then steadily decreases to aplateau value when particle concentration increases. The initial

y = 186058x + 3132.8R² = 0.9993

0

2000

4000

6000

8000

10000

12000

14000

0 0.01 0.02 0.03

t / V

(s/L

2 )

Permea

Fig. 3. Filtration curve (t/V) vs. (V) for 15 mg/L HA a

value for specific cake resistance is anomalously low, probablydue to the poorly dispersed cake formation at very low particleconcentration where the cake layer is not mature due to insuffi-ciency of the convected solids in the bulk. The subsequent trendfrom low to raised a to a lower constant value can be explained bychanges in effective cake structure for different solids concentra-tions. At low/moderate solid concentrations, the deposition pro-cess can be orderly producing layers with few ‘defects’. This lowporosity structure would have raised a. As the concentrationincreases, the cake layering becomes more disordered with morerandom voids, causing higher porosity and lower a [38]. The cakelayering process is influenced by net solids convection as deter-mined by flux and concentration leading to the observed ‘plateau’in a.

4.2.3. Organic fouling experiments with humic acid

Ultrafiltration experiments were performed with humic acidsolution in the range of 1–50 mg/L. The effect of humic acid on theMFIUF values and specific cake resistance are presented in Figs. 6and 7, respectively.

In common with the colloidal data the MFIUF values for humicacid showed a generally linear relationship with the concentra-tion of humic acid Fig. 6. However at the low concentration range(o10 mg/L) the values were above trend, as observed with thesilica. This is reflected in the calculated specific resistance shownin Fig. 7. In agreement with colloidal fouling experiments, there isa significant decline in specific resistance as concentrationincreases from 1–10 mg/L and then negligible increase in a from10–50 mg/L humic acid.

4.3. Combined fouling experiments

4.3.1. Effect of humic acid in the presence of colloidal silica

In these experiments the humic acid concentration wasincreased from 2.5 mg/L to 50 mg/L while the silica concentrationwas kept constant at 100 mg/L. Fig. 8 compares the MFIUF fororganic, colloid and combined fouling experiments as well as thesum of individual organic and colloidal foulants. The calculatedMFIUF refers to sum of individual MFIUF for HA and SiO2(MFIcalculated¼MFISiO2þMFIHA). However, the measured MFIUF

refers to the MFIUF value obtained from filtration using mixedfoulants.

The measured MFIUF increased as the concentration of humicacid increased, and the MFIUF for the mixture was higher than the

y = 16713x + 4116R² = 0.9982

0.04 0.05 0.06 0.07 0.08

te Volume (L)

100 mg/L SiO2

15 mg/L HA

nd 100 mg/L SiO2, constant pressure of 210 kPa.

y = 0.1298xR² = 0.9848

0

20

40

60

80

100

120

0 100 200 300 400 500 600 700 800 900

MFI

-UF

(s/L

2 ) ×

10-4

Concentration of SiO2(mg/L)

Fig. 4. MFI-UF as a function of silica concentration in the feed water, constant pressure of 210 kPa.

0

5

10

15

20

25

0 100 200 300 400 500 600 700 800 900

Spec

ific

Cak

e R

esis

tanc

e ×1

0-14

α (m

/kg)

Concentration of SiO2(mg/L)

Fig. 5. Specific cake resistance of silica deposit as a function of silica concentrations, constant pressure of 210 kPa.

y = 0.1216xR2 = 0.9864

0

1

2

3

4

5

6

7

0 5 10 15 20 25 30 35 40 45 50 55

MFI

-UF

(s/L

2 ) ×

10-4

Concentration of Humic Acid (mg/L)

Fig. 6. MFI-UF as a function of humic acid concentration, constant pressure of 210 kPa.

A.H. Taheri et al. / Journal of Membrane Science 433 (2013) 112–120116

fouling index of individual humic acid at all concentrations.However, at low concentration of humic acid (o15 mg/L), themeasured MFIUF of the combined fouling was lower than the silica

alone or the summation of the fouling indices for silica and humicacid. At higher concentrations (420 mg/L) the measured MFIUF

was greater than the MFIUF of the silica alone. To explain this

0

5

10

15

20

25

30

0 5 10 15 20 25 30 35 40 45 50 55

Spec

ific

Cak

e R

esis

tanc

e ×1

0-14

α (m

/kg)

Concentration of Humic Acid (mg/L)

Fig. 7. Specific cake resistance as a function of humic acid concentration, constant pressure of 210 kPa.

Fig. 8. MFI-UF for different feed: humic acid (various concentrations), colloidal silica (100 mg/L), combined foulants (various concentrations of HA and 100 mg/L silica),

constant pressure of 210 kPa.

Fig. 9. Schematic of cake layer on the membrane surface: (a) silica; (b) silica in the

presence of low concentration of humic acid; (c) silica in the presence of high

concentration of humic acid.

A.H. Taheri et al. / Journal of Membrane Science 433 (2013) 112–120 117

behavior there is a need to consider the likely effects of foulantinteractions during filtration of complex solutions containingboth organic and colloidal particles.

According to surface zeta potential values (Table 1), adsorptionof HA on the membrane surface would increase the electrostaticrepulsion between the membrane and colloidal silica. This wouldhinder deposition of colloidal silica on the membrane and as aresult create a more porous initial cake layer. However, the t/V vs.V plot used to provide MFIUF uses the slope generated by multiplelayers of foulant and therefore characterizes foulant–foulantdeposition. Although QCM-D experiments revealed low adsorp-tion of HA on colloidal silica, it could still increase the netrepulsion either by increasing the surface charge or by penetrat-ing between silica particles. Humic acid has higher negativecharge compared to silica and could increase the net repulsiveforces within the deposit leading to higher cake porosity withlower fouling potential. However, the data in Fig. 8 suggests thatbeyond a certain load of humic acid the interparticle voids start tobecome blocked and the cake becomes less porous, giving ahigher MFIUF. Fig. 9 illustrates these ideas for the mixed foulantcake. The organic-filled void model is analogous to a bacterialcake, where the presence or absence of extra-cellular organicsubstances has been shown to make a significant difference incake resistance [39].

Li and Elimelech [28] also studied the fouling mechanism ofcolloids and natural organic matter and suggested that there is a

A.H. Taheri et al. / Journal of Membrane Science 433 (2013) 112–120118

synergistic effect due to hindered back diffusion. They claimedthat the back diffusion of one model foulant can be hindered bythe presence of other types of foulants and thereby increase the

0

2

4

6

8

10

12

14

16

0 10 20 30 40 50 60

Spec

ific

Cak

e R

esis

tanc

e ×1

0-14

α (m

/kg)

Concentration of humic acid (mg/L)

HA

SiO2 (100 mg/L)

Mixture

Fig. 10. Specific cake resistance for different feed: humic acid (various concentra-

tions), colloidal silica (100 mg/L), combined foulants (various concentrations of HA

and 100 mg/L silica), constant pressure of 210 kPa.

0

5

10

15

20

25

30

15 25 50 100 150

MFI

-UF

(s/L

2 )×1

0-4

Concentration of Silica (mg/L)

SiO2Measured MFI (Combined foulants)Calculated MFI (MFI SiO2 + MFI HA)HA (15 mg/L)

Fig. 11. MFI-UF results: humic acid (15 mg/L), colloidal silica (various concentra-

tions), combined foulants (various concentrations of silica and 15 mg/L HA),

constant pressure of 210 kPa.

0

5

10

15

20

25

0 20 40 60

Spec

ific

Cak

e R

esis

tanc

e ×1

0-14

α (m

/kg)

Concentratio

Fig. 12. Specific cake resistance results: humic acid (15 mg/L), colloidal silica (various co

constant pressure of 210 kPa.

cake resistance. Thus possible explanations for the trends athigher humic acid content might be either due to hindered backdiffusion or filing the voids between colloidal particles andinfiltration of humic acid in the silica cake layer. It is also possiblethat both mechanisms apply. The combined effect would be acake with a low porosity, a greater resistance and therefore ahigher MFIUF. Not surprisingly the sum of the individual MFIUFsdoes not match the mixture MFIUF very well. This highlights thefact that knowledge of the fouling propensity of the individualfoulants does not lead to the fouling propensity of the mixture.

Fig. 10 shows the specific cake resistance results obtained fromexperiments for the mixture of silica at different humic acidconcentrations. The trends mirror those for MFIUF, and emphasizethe significant effect a small amount of humic acid can have onthe specific resistance of the silica cake (a 33% drop at 2 mg/Lhumic).

4.3.2. Effect of colloidal silica in the presence of humic acid

In these experiments the humic acid concentration was fixedat 15 mg/L and the silica concentration increased from 15 to150 mg/L. Fig. 11 presents the MFIUF data and Fig. 12 the specificresistances from dead-end filtration under constant pressure forhumic acid, silica and their mixture.

The measured MFIUF value increased by increasing the silicaconcentration in the presence of humic acid. At low concentra-tions of silica (o25 mg/L) the MFIUF for the mixture was greaterthan that for silica alone. At higher values of silica, the mixtureMFIUF was less than for silica alone. Whether the MFIUF isenhanced or diminished appears to depend on the relativeamounts of humic and colloid. When the proportion of humic islow the MFIUF is diminished; this is also evident in Fig. 8. Underthese conditions humic acid may promote interparticle repulsionand increasing porosity. With relatively more humic acid presentthe results imply a loss of cake porosity, presumably due tointerstitial loading by humic acid. The data also confirm thatknowledge of the individual MFIUFs cannot be used reliably topredict the fouling potential of the mixture. The trends in specificcake resistance are similar. Over most of the silica concentrationrange the specific cake resistance of the mixture of foulants issignificantly smaller than that of the silica alone. However, thistrend is dependent on the selected humic concentration (here15 mg/L) as evident from Fig. 10.

80 100 120 140 160n of Silica (mg/L)

HA (15 mg/L)

Mixture

Silica

ncentrations), combined foulants (various concentrations of silica and 15 mg/L HA),

A.H. Taheri et al. / Journal of Membrane Science 433 (2013) 112–120 119

5. Conclusions

The effects of different factors on membrane fouling withcolloidal silica and humic acid under constant pressure wereinvestigated using ultrafiltration membranes. Based on theexperimental results, the following conclusions can be made:

–

A linear response of MFIUF was observed for both humic acidand colloidal silica, except at the lowest concentrations. Arapid increase was observed for the specific cake resistance ofcolloidal silica followed by a steady decrease when particleconcentration increased. The initial low value of the specificcake resistance is assumed to be due to dispersed cakeformation at low particle concentration and subsequentdecline was due to inhomogeneous deposition of particles onthe membrane at higher concentration.

–

QCM-D and surface zeta potential tests provided insights intothe mechanisms responsible for the fouling propensity ofmixed colloids and organics. The results revealed that humicacid has a higher surface zeta potential compared to thecolloidal silica and also very low adsorption of humic acid oncolloidal silica was observed using QCM-D experiments.

–

By increasing humic acid at a constant concentration ofcolloidal silica the MFIUF value increased. At low concentra-tions of humic acid the MFIUF of the combined foulants wasless than that of silica alone. This may be due to repulsion ofhumic acid and silica and creation of a more porous cake layer.At higher concentrations of humic the MFIUF of the combinedfoulants exceeded that of silica alone. This may be due toinfiltration of humic acid into the silica cake layer, or hinderedback diffusion effects, creating a lower porosity cake.

–

Similar behavior was observed when the concentration ofcolloidal silica increased and the HA was kept constant. Overmost of the silica concentration range the MFIUF value waslower than that of the silica alone.

–

Our results show that the fouling propensities of mixtures of acolloid (silica) and macrosolute (humic acid) are related to thefouling propensity of the individual components in complexways that are not readily predicted.

Acknowledgments

The authors would like to thank the Environment and WaterIndustry Programme Office (EWI) of Singapore for the support ofthis work (Project Ref. EWI RFP 09/01). We also acknowledge thesupport from the Singapore Economic Development Board toSingapore Membrane Technology Centre.

Nomenclature

MFI0.45 modified fouling index determined using 0.45 mmmembrane

MFIUF modified fouling index ultrafiltration measuredunder constant pressure

SDI silt density indexQCM-D quartz crystal microbalance with dissipationRm membrane resistance (m�1)RC cake resistance (m�1)A membrane area (m2)Aq piezoelectrically active crystal area (m2)DP transmembrane pressure (Pa)

I resistivity (m�2)Cb bulk concentration (kg/m3)dp particle diameter (m)t filtration time (s)V permeate volume (L)n compressibility coefficientf frequency (Hz)f0 resonant frequency (Hz)m filtrate solution viscosity (Pa s)a specific cake resistance (m/Kg)a0 constantrp Particle density (kg/m3)e cake porosityrq density of quartz (kg/m3)

References

[1] J. Davey, A.I. Schafer, Ultrafiltration to supply drinking water in internationaldevelopment: a review of opportunities, 2009.

[2] W. Gao, H. Liang, J. Ma, M. Han, Z.L. Chen, Z.S. Han, G.B. Li, Membrane foulingcontrol in ultrafiltration technology for drinking water production: a review,Desalination 272 (2011) 1–8.

[3] K.T. Chua, M.N.A. Hawlader, A. Malek, Pretreatment of seawater: results ofpilot trials in Singapore, Desalination 159 (2003) 225–243.

[4] N. Porcelli, S. Judd, Chemical cleaning of potable water membranes: the costbenefit of optimisation, Water Res. 44 (2010) 1389–1398.

[5] C.Y. Tang, T.H. Chong, A.G. Fane, Colloidal interactions and fouling of NF andRO membranes: a review, Adv. Colloid Interface Sci. 164 (2011) 126–143.

[6] A.S. D4189-07, standard test method for silt density index (SDI) of water,D19.08 on membranes and ion exchange materials, 2007.

[7] J.C. Schippers, J. Verdouw, The modified fouling index, a method of determiningthe fouling characteristics of water, Desalination 32 (1980) 137–148.

[8] J.C. Schippers, J.H. Hanemaayer, C.A. Smolders, A. Kostense, Prediction of fluxdecline of reverse osmosis membranes, Desalination 38 (1981) 339–348.

[9] A. Mosset, V. Boonnelye, M. Petry, M.A. Sanz, The sensitivity of SDI analysis:from RO feed water to raw water, in: Proceedings of the Conference onDesalination and the Environment, Halkidiki, GREECE, 2007, pp. 17–23.

[10] S.F.E. Boerlage, M.D. Kennedy, M.P. Aniye, E.M. Abogrean, D.E.Y. El-Hodali,Z.S. Tarawneh, J.C. Schippers, Modified fouling index((ultrafiltration)) to comparepretreatment processes of reverse osmosis feedwater, vol. 131, p. 201, 2000;Desalination, vol. 133, 2001, p. 89.

[11] S.F.E. Boerlage, M.D. Kennedy, M.R. Dickson, D.E.Y. El-Hodali, J.C. Schippers,The modified fouling index using ultrafiltration membranes (MFI-UF):characterisation, filtration mechanisms and proposed reference membrane,J. Membr. Sci. 197 (2002) 1–21.

[12] S.F.E. Boerlage, M. Kennedy, Z. Tarawneh, R. De Faber, J.C. Schippers, Devel-opment of the MFI-UF in constant flux filtration, Desalination 161 (2004)103–113.

[13] S. Khirani, R. Ben Aim, M.H. Manero, Improving the measurement of themodified fouling index using nanofiltration membranes (NF-MFI), in:Proceedings of the International Congress on Membranes and MembraneProcesses, Seoul, South Korea, 2005, pp. 1–7.

[14] K.J. Howe, M.M. Clark, Fouling of microfiltration and ultrafiltration mem-branes by natural waters, Environ. Sci. Technol. 36 (2002) 3571–3576.

[15] G. Amy, Fundamental understanding of organic matter fouling of mem-branes, Desalination 231 (2008) 44–51.

[16] H.A. Mousa, Investigation of UF membranes fouling by humic acid, Desalina-tion 217 (2007) 38–51.

[17] C. Visvanathan, R. Benaim, Studies on colloidal membrane fouling mechanismsin crossflow microfiltration, J. Membr. Sci. 45 (1989) 3–15.

[18] G. Belfort, R.H. Davis, A.L. Zydney, The behavior of suspensions and macro-molecular solutions in crossflow microfiltration, J. Membr. Sci. 96 (1994)1–58.

[19] E.M. Vrijenhoek, S. Hong, M. Elimelech, Influence of membrane surfaceproperties on initial rate of colloidal fouling of reverse osmosis and nanofil-tration membranes, J. Membr. Sci. 188 (2001) 115–128.

[20] S.G. Yiantsios, D. Sioutopoulos, A.J. Karabelas, Colloidal fouling of ROmembranes: an overview of key issues and efforts to develop improvedprediction techniques, Desalination 183 (2005) 257–272.

[21] A.S. Al-Amoudi, A.M. Farooque, Performance restoration and autopsy of NFmembranes used in seawater pretreatment, Desalination 178 (2005)261–271.

[22] A. Seidel, M. Elimelech, Coupling between chemical and physical interactionsin natural organic matter (NOM) fouling of nanofiltration membranes:implications for fouling control, J. Membr. Sci. 203 (2002) 245–255.

[23] A.S. Al-Amoudi, Factors affecting natural organic matter (NOM) and scalingfouling in NF membranes: a review, Desalination 259 (2010) 1–10.

A.H. Taheri et al. / Journal of Membrane Science 433 (2013) 112–120120

[24] A.G. Fane, W. Xi, W. Rong, Membrane filtration processes and fouling, in:N. Gayle, D. David (Eds.), Interface Science and Technology, Elsevier, 2006,Chapter 7, pp. 109–132.

[25] S.H. Chuang, W.C. Chang, M.C. Chang, M.A. Sung, The effects of solubleorganic matters on membrane fouling index, Biores. Technol. 100 (2009)1875–1877.

[26] V. Lahoussine-Turcaud, M.R. Wiesner, J.-Y. Bottero, Fouling in tangential-flowultrafiltration: The effect of colloid size and coagulation pretreatment, J.Membr. Sci. 52 (1990) 173–190.

[27] S. Lee, J. Cho, M. Elimelech, Combined influence of natural organic matter(NOM) and colloidal particles on nanofiltration membrane fouling, J. Membr.Sci. 262 (2005) 27–41.

[28] Q. Li, M. Elimelech, Synergistic effects in combined fouling of a loosenanofiltration membrane by colloidal materials and natural organic matter,J. Membr. Sci. 278 (2006) 72–82.

[29] D. Jermann, W. Pronk, R. Kagi, M. Halbeisen, M. Boller, Influence of interac-tions between NOM and particles on UF fouling mechanisms, Water Res. 42(2008) 3870–3878.

[30] S.F.E. Boerlage, M.D. Kennedy, M.P. Aniye, E.M. Abogrean, G. Galjaard,J.C. Schippers, Monitoring particulate fouling in membrane systems, Desali-nation 118 (1998) 131–142.

[31] P.C. Carman, Fundamental principles of industrial filtration, Trans. Inst.Chem. Eng. 16 (1938) 168–188.

[32] X. Wang, C. Ruengruglikit, Y.-W. Wang, Q. Huang, Interfacial interactions ofpectin with bovine serum albumin studied by quartz crystal microbalancewith dissipation monitoring: effect of ionic strength, J. Agric. Food Chem. 55

(2007) 10425–10431.[33] S.X. Liu, J.T. Kim, Study of adsorption kinetics of surfactants onto polyether-

sulfone membrane surface using QCM-D, Desalination 247 (2009) 355–361.[34] G. Sauerbrey, Verwendung von Schwingquarzen zur Wagung dunner Schich-

ten und zur Mikrowagung, Z. Phys. A Hadrons Nuclei 155 (1959) 206–222.[35] A.E. Contreras, A. Kim, Q. Li, Combined fouling of nanofiltration membranes:

mechanisms and effect of organic matter, J. Membr. Sci. 327 (2009) 87–95.[36] C. Park, H. Kim, S. Hong, S.I. Choi, Variation and prediction of membrane

fouling index under various feed water characteristics, J. Membr. Sci. 284

(2006) 248–254.[37] L.N. Sim, Y. Ye, V. Chen, A.G. Fane, Crossflow sampler modified fouling index

ultrafiltration (CFS-MFIUF)—an alternative Fouling index, J. Membr. Sci. 360(2010) 174–184.

[38] K. Parameshwaran, A.G. Fane, B.D. Cho, K.J. Kim, Analysis of microfiltrationperformance with constant flux processing of secondary effluent, Water Res.35 (2001) 4349–4358.

[39] P.H. Hodgson, G.L. Leslie, A.G. Fane, R.P. Schneider, C.J.D. Fell, K.C. Marshall,Cake resistance and solute rejection in bacterial microfiltration: the role of

the extracellular matrix, J. Membr. Sci. 79 (1993) 35–53.