Journal of Membrane Science 471 (2014) 94–102

Contents lists available at ScienceDirect

Journal of Membrane Science

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

Control of natural organic matter fouling of ultrafiltration membrane

by adsorption pretreatment: Comparison of mesoporous adsorbent
resin and powdered activated carbon
Kai Li, Heng Liang n, Fangshu Qu, Senlin Shao, Huarong Yu, Zheng-shuang Han,
Xing Du, Guibai Li
State Key Laboratory of Urban Water Resource and Environment (SKLUWRE), Harbin Institute of Technology,
73 Huanghe Road, Nangang District, Harbin 150090, PR China

art ic l e i nf o a b s t r a c t

Article history: This paper focused on the effects of mesoporous adsorbent resin (MAR) and powdered activated carbon
Received 27 March 2014 (PAC) pretreatments on ultrafiltration (UF) membrane fouling caused by natural organic matter (NOM).
Received in revised form Three model foulants, humic acid (HA), bovine serum albumin (BSA) and sodium alginate (SA), were
31 July 2014
adopted to represent different NOM fractions in natural waters. Moreover, the impact of the presence of
Accepted 2 August 2014
Available online 11 August 2014
adsorbent particles in UF feed water on membrane fouling was also evaluated. The results indicated that
MAR adsorption exhibited remarkable performance in alleviating HA and BSA fouling, no matter
Keywords: whether MAR particles were removed before UF or not. In contrast, PAC pretreatment slightly
Membrane fouling ameliorated HA fouling when PAC particles were removed before UF, whereas HA fouling was
Natural organic matter (NOM) exacerbated by PAC pretreatment with PAC particles present in UF feed water. BSA fouling was
Adsorption pretreatment
moderately controlled by PAC adsorption irrespective of the presence or absence of PAC particles in
Mesoporous adsorbent resin (MAR)
UF feed water. However, neither of these two pretreatments visibly influenced SA fouling. Overall, the
Powdered activated carbon (PAC)
results obtained in the current research would provide relevant information on adsorbent selection and
process design of the hybrid adsorption/UF process according to the composition and properties of NOM.
& 2014 Elsevier B.V. All rights reserved.

1. Introduction pretreatment always improved the quality of product water, its

Membrane fouling is one of the major impediments for the
widespread application of ultrafiltration (UF) membrane in water
and wastewater treatment [1–3]. Extensive studies have been
undertaken for more insights into UF membrane fouling, and the
natural organic matter (NOM), including allochthonous humic
substances and autochthonous biopolymers (mainly consisting of
proteins and polysaccharides), has been generally considered as
the major culprit responsible for membrane fouling [4–7]. Adsorp-
tion is an efficient technology for NOM removal, therefore it has
been widely adopted as pretreatment for UF to enhance the
performance of UF process [1,8–15]. However, although adsorption
Abbreviations: BSA, bovine serum albumin; CA, cellulose acetate; EOM, extra-
cellular organic matter; HA, humic acid; MAR, mesoporous adsorbent resin; MW,
molecular weight; MWCO, molecular weight cut-off; NOM, natural organic matter;
PAC, powdered activated carbon; PES, polyethersulfone; SA, sodium alginate; TMP,
trans-membrane pressure; UF, ultrafiltration
n
Corresponding author. Tel./fax: þ 86 451 86283001.
E-mail addresses: likai02007@126.com (K. Li), hitliangheng@163.com (H. Liang),
qufangshu@163.com (F. Qu), shaosenlin@gmail.com (S. Shao),
huarongyu@gmail.com (H. Yu), hit08hzs@163.com (Z.-s. Han),
xingduxing@sina.com (X. Du), hitsteven@gmail.com (G. Li).
impact on membrane fouling is still under debate [1,13,14].
Powdered activated carbon (PAC) is the most common type of
commercially available adsorbent in water treatment, and thus it
has been widely applied in hybrid adsorption/UF process [14]. It
was manifested in several studies that PAC adsorbed a significant
proportion of NOM and efficiently controlled the membrane
fouling [9,10,12,16]. However, in some other studies, PAC adsorp-
tion was reported to exert minor influence on membrane fouling
although it indeed removed some NOM [8,17]. Moreover, when
PAC particles were present in UF feed water, the membrane fouling
in the hybrid PAC/UF process was even found to be exacerbated in
comparison with that in the individual UF process [8,15,17]. The
contradictory influence of PAC on membrane fouling was generally
ascribed to the diversity of NOM characteristics in feed water and/
or the membrane properties (e.g., surface hydrophobicity), but
systematical studies on this subject were very limited [11,14].
Unlike PAC, mesoporous adsorbent resin (MAR) is an adsorbent
specially developed for membrane fouling control [18]. It was
reported that MAR significantly reduced both the reversible and
irreversible fouling of 20 kDa polyethersulfone (PES) membrane
while filtering lake water, although only a small amount of NOM
was adsorbed by MAR [19]. Size fractionation results suggested

http://dx.doi.org/10.1016/j.memsci.2014.08.006
0376-7388/& 2014 Elsevier B.V. All rights reserved.
 K. Li et al. / Journal of Membrane Science 471 (2014) 94–102
that MAR preferentially adsorbed a fraction of NOM that had an
apparent MW of 20–200 kDa. Li et al. [20] found that MAR
mitigated the fouling of 100 kDa PES membrane caused by algal
extracellular organic matter (EOM) much more efficiently than
PAC did because MAR selectively removed high-MW fraction of
EOM. Besides, both of the studies demonstrated that the presence
of MAR particles in UF feed water at concentrations as high as
100 mg/L would not bring negative effects on membrane fouling
[19,20]. The aforementioned studies indicated that MAR might be
a more promising adsorbent in membrane fouling control in
comparison with PAC. But the studies with respect to MAR were
restricted to the fouling of PES membrane caused by NOM in lake
water and algal EOM, and the comparative studies of two types of
adsorbents were limited. The performances of MAR and PAC in
mitigating membrane fouling caused by diverse NOM components
and the underlying mechanisms are still unclear.
The main objective of this study was to obtain a comprehensive
understanding of the effects of MAR and PAC pretreatments on
NOM fouling of UF membrane. Because the complexity of natural
water matrices made it difficult to elaborate the influence of
adsorption pretreatment on membrane fouling by different NOM
fractions, model foulants were employed in this study and caution
should be taken when extrapolating the results obtained here to
natural waters. Two types of commonly used UF membranes with
different surface hydrophobicities were used in the tests. The
adsorption capacities of MAR and PAC towards three foulants were
investigated and UF experiments were carried out with NOM
solutions before and after adsorption pretreatment. Moreover,
the contribution of adsorbent particles in UF feed water to
membrane fouling was also examined.
2. Materials and methods
2.1. NOM solutions
Humic acid (HA), bovine serum albumin (BSA) and sodium
alginate (SA) purchased from Sigma-Aldrich (USA) were used as
representatives of humic substances, proteins and polysaccharides,
respectively. To prepare HA stock solution, 2 g of HA was added to
800 mL of 0.01 M NaOH solution, followed by stirring for 24 h and
adjusting pH of the solution to 7.0 using 1 mol/L HCl. The solution
was then diluted to 1000 mL to get the HA stock solution with a
concentration of 2 g/L. The stock solutions of BSA and SA were
prepared by dissolving 1 g of BSA and SA in 1000 mL Milli-Q water,
respectively, followed by stirring for 24 h. The stock solutions were
all stored in dark at 4 1C.
The stock solutions were diluted with Milli-Q water to obtain the
NOM solutions used for adsorption tests and UF experiments. In order
to simulate the solution chemistry of natural waters, 1 mmol/L
NaHCO3, 6 mmol/L NaCl and 1 mmol/L CaCl2 were added and the
pH was adjusted to 7.570.1 with 0.1 mol/L HCl and NaOH. The
concentrations of HA, BSA and SA employed in UF experiments were
10, 2 and 2 mg/L, respectively. The corresponding dissolved organic
carbon concentrations were 4.3870.11, 0.8170.07 and
0.7670.12 mg C/L, respectively. Unless otherwise specified, the con-
centrations of model foulants reported in this paper were on the basis
of the mass of model foulants rather than the content of carbon. The
concentrations of BSA and SA used in UF experiments were much
lower than that of HA because the concentration levels of proteins and
polysaccharides were usually very low in natural surface water [21,22].
2.2. Adsorbents and adsorption tests
MAR is a type of mesoporous adsorbent synthesized following
the method proposed by Clark et al. [18] and detailed steps of
95
Table 1
Main properties of MAR and PAC.
Adsorbent Average particle Zeta potential BET surface Average pore
size (d50, μm) (mV) area (m2/g) size (nm)
MAR 25.2 7 0.8  22.4 71.0 1087 9 16.4 70.5
PAC 32.17 0.7  23.9 70.9 12197 13 2.2 70.1
Note: values represent average7 standard deviation, n¼ 3.
preparation could be found in reference [20]. Wood-based PAC
was purchased from Bench Chemicals (Tianjin, China) and was
used without further purification. Stock solutions of MAR and PAC
with mass concentrations of 10 g/L were prepared and stored in
refrigerator for use. The main properties of MAR and PAC are listed
in Table 1. It can be seen that they had similar particle size and
surface charge, but they displayed substantial difference in the
pore structure characteristics. The BET surface area of MAR was
less than 10% of that of PAC, while the pore size of MAR was much
larger than that of PAC.
Adsorption tests were conducted to characterize the adsorption of
the NOM fractions onto two types of adsorbents. The concentrations of
HA, BSA and SA employed in adsorption tests were 1–30 mg/L and the
concentration of the adsorbent was 50 mg/L. The adsorbent was
added to NOM solutions and the flasks were shaken in a rotary
shaker at 120 rpm and 20 1C for 12 h. Finally, the samples were filtered
with 0.45 μm mixed cellulose filters (Taoyuan, China) to remove
adsorbent particles and the concentrations of HA, BSA and SA in the
filtrate were measured. The retention of these NOM fractions by this
filter has been proved to be negligible in preliminary tests. The
adsorbed amounts of HA, BSA and SA onto the adsorbents were
calculated by mass balance.
2.3. UF membranes and experimental setup
Two types of flat-sheet UF membranes with the same mole-
cular weight cut-off (MWCO) of 100 kDa, i.e., a PES membrane
(OM100076, Pall, USA) and a cellulose acetate (CA) membrane
(PLHK07610, Millipore, USA), were used in this study. They have
similar surface charge and roughness, but are apparently different
in surface hydrophobicity [23]. The contact angles of the PES and
the CA membrane were 58.21 and 19.31, respectively, indicating
that the PES membrane was much more hydrophobic than the CA
membrane.
UF experiments were performed in a filtration cell (Amicon
8400, Millipore, USA) in dead-end mode at room temperature
(20 71 1C). UF membrane was placed on the bottom of the cell
with its glossy side towards the bulk solution during filtration.
A peristaltic pump was used as the suction pump to maintain a
constant permeate flux of 150 L/(m2 h). The trans-membrane pres-
sure (TMP) was monitored by a pressure transducer (PTP708, Tuopo
Electric, Foshan, China) mounted between the filtration cell and the
suction pump. The pressure transducer was connected to a com-
puter and the data was automatically logged every five seconds.
Average fouling rate was calculated by dividing the increment of
TMP by filtration time.
2.4. UF experiments and fouling resistance analysis
Fouling reversibility (i.e. the response of membrane fouling to
physical cleaning strategies) is of great significance from the
perspective of application, while the analysis of fouling resistance
distribution (external vs. internal fouling) would favor the identi-
fication of fouling mechanisms [3,24,25]. In this study, external
fouling was operationally classified into loosely- and strongly-
attached external fouling by its response to shear stress, and
96 K. Li et al. / Journal of Membrane Science 471 (2014) 94–102
internal fouling was divided into reversible internal fouling and
irreversible fouling through backwashing. Therefore, membrane
filtration resistance was described using the resistance-in-series
model as shown below [26,27]:
Rt ¼ TMP=ðμJÞ ¼ Rm þ Rf ¼ Rm þ Ref þ Rif
Ref ¼ Ref l þ Ref s
Rif ¼ Rif r þ Rirr
1
where, Rt is the total hydraulic resistance (m ); TMP is the trans-
membrane pressure (Pa); μ is the dynamic viscosity of the feed
water (Pa s); J is the permeate flux of the membrane (m/s); Rm is
the clean membrane resistance (m  1); Rf is the total fouling
resistance (m  1); Ref is the external fouling resistance (m  1); Rif
is the internal fouling resistance (m  1); Ref-l is the loosely-
attached external fouling resistance (m  1); Ref-s is the strongly-
attached external fouling resistance (m  1); Rif-r is the reversible
internal fouling resistance (m  1); Rirr is the physically irreversible
fouling resistance (m  1).
During this study, UF experiments were conducted as follows to
obtain the filtration resistance mentioned above.
1) The filtration of 100 mL Mill-Q water was carried out and the
average of TMP values was recorded as TMP0. Rm could be
calculated as follows :
Rm ¼ TMP 0 =ðμJÞ
2) NOM solution with a volume of 350 mL was fed into the cell
and then the filtration begun. When the permeate volume
increased to 300 mL or the TMP reached 80 kPa, the filtration
experiment was terminated and the TMP at the end was
recorded and designated as TMP1. After that, the stirrer in the
filtration cell was stirred at a rate of 200 rpm for 2 min to
remove loosely-attached cake/gel layer on the membrane sur-
face, followed by filtration of 100 mL Mill-Q water with the
average of TMP values recorded as TMP2. Ref-l (including the
resistance caused by concentration polarization) could be
calculated as follows:
Ref l ¼ ðTMP 1  TMP 2 Þ=ðμJÞ
3) To remove the strongly-attached cake/gel layer remaining after
stirring, the membrane was taken out and wiped with a wet
sponge carefully. Afterwards, the membrane was reinstalled
and the filtration of Milli-Q water was conducted again with
the average of TMP values recorded as TMP3. Then, Ref-s derived
from strongly-attached cake/gel layer could be calculated as
follows:
Ref s ¼ ðTMP 2  TMP 3 Þ=ðμJÞ
4) The membrane was backwashed with 20 mL Mill-Q water by
placing the reverse side of the membrane upwards and then
the TMP value of filtering Mill-Q water was determined and
recorded as TMP4. Therefore, Rif-r and Rirr could be calculated by
Eqs. (7) and (8), respectively, as follows:
Rif r ¼ ðTMP 3  TMP 4 Þ=ðμJÞ
Rirr ¼ ðTMP 4 TMP 0 Þ=ðμJÞ
Prior to UF experiments, the adsorbent was added to NOM
solutions at a dose of 50 mg/L and the contact time was set at
30 min. In order to elucidate the effects of the presence of
adsorbent particles in UF feed water on membrane fouling, UF
experiments were conducted in two modes for each NOM solu-
tion: 1) the adsorbent particles were removed before UF by
filtering through 0.45 μm mixed cellulose filters (Taoyuan, China)
and the filtrate, which was denoted as MAR-treated HA/BSA/SA
ð1Þ
and PAC-treated HA/BSA/SA, was used as UF feed water; 2) the
ð2Þ mixed solution of NOM and adsorbent particles was fed into UF
cell and was denoted as MAR þ HA/BSA/SA and PAC þ HA/BSA/SA.
ð3Þ
2.5. Analytical methods
The concentration of HA was determined in terms of UV
absorbance at 254 nm using a UV/vis spectrophotometer (T6,
China). The concentrations of BSA and SA were measured by
dissolved organic carbon analysis using a total organic carbon
analyzer (multi N/C 2100, Jena, Germany). The particle size
distributions of MAR and PAC were characterized using a Master-
Sizer 2000 (Malvern, UK). Zeta potentials of MAR and PAC were
determined using Nano S90 (Malvern, UK) at the pH of 7.570.1.
The surface areas and pore size distributions of MAR and PAC were
measured using a surface area and porosity analyzer (ASAP 2020,
Micromeritics, USA) with a nitrogen adsorption method. All the
analyses were done in triplicate.
3. Results and discussion
ð4Þ
3.1. Adsorption of NOM fractions onto MAR and PAC
Fig. 1 shows the adsorption of different NOM fractions onto
MAR and PAC. It can be observed that the HA adsorption capacity
of MAR was much larger than that of PAC. The adsorbed amount of
HA onto MAR reached 279.6 mg/g at an equilibrium concentration
of 15.95 mg/L while that onto PAC was just 103.4 mg/g at a much
higher equilibrium concentration of 24.8 mg/L. For BSA, the
adsorption capacity of MAR was also larger than that of PAC, but
the gap was not as significant as that for HA. With regard to SA, the
adsorption capacities of MAR and PAC were both quite small. The
different adsorption capacities of MAR and PAC for each NOM
ð5Þ fraction could be attributed to diverse pore structures of the two
adsorbents and various chemical compositions and MW distribu-
tions of NOM fractions. PAC is a typical microporous adsorbent,
whereas the pore size of MAR is much larger (Table 1). HA is a
mixture of polyelectrolytes with a wide MW distribution. As
reported previously, the hydrodynamic size of a large proportion
MAR-HA MAR-BSA MAR-SA
300 PAC-HA PAC-BSA PAC-SA
ð6Þ 250
Adsorbed amount (mg/L)
200
150
100
ð7Þ 50
ð8Þ 0
0 5 10 15 20 25 30
Equilibrium concentration (mg/L)
Fig. 1. Adsorption of NOM fractions onto MAR and PAC (the solid lines represent
the fitting of the data according to Freundlich isotherm model, the equilibrium
concentration and adsorbed amount are on the basis of the mass of model
foulants).
 K. Li et al. / Journal of Membrane Science 471 (2014) 94–102 97

80 60

60 45
TMP (kPa)

TMP (kPa)
40 30

20 15 Raw HA
Raw HA
MAR-treated HA PAC-treated HA MAR-treated HA PAC-treated HA
MAR+HA PAC+HA MAR+HA PAC+HA
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Filtration time (min) Filtration time (min)

15 Ref-l Ref-s Rif-r Rirr 10

Ref-l Ref-s Rif-r Rirr

m -1)
m )
-1

12 8

11
11

Fouling resistance (10

Fouling resistance (10

9 6

6 4

3 2

0 0

A
A

A
A
A

A
A

H
A
H

H
H
H
H

H
H

+H
H

R+

C+
d
w
d

ed

te
w

R+

te
te

Ra
Ra

ea
C

A
at

PA
ea

ea
PA
A

M
tre

tr
-tr

tr
M

R-

C-
C-
R

A
A

PA
PA

M
M

Fig. 2. Effects of MAR and PAC pretreatments on HA fouling of PES membrane: Fig. 3. Effects of MAR and PAC pretreatments on HA fouling of CA membrane:
(a) TMP buildup (only mean values were reported) and (b) fouling resistance (a) TMP buildup (only mean values were reported) and (b) fouling resistance
analysis (error bars represent standard deviations from the means) (n ¼3). analysis (error bars represent standard deviations from the means) (n¼ 3).

of Sigma-Aldrich HA molecules is larger than the size of micro- It can be observed in Fig. 2(b) that raw HA mainly caused
pores (wo2 nm) [28]. As a result, although the BET surface area of external fouling, with the Ref (i.e., the sum of Ref-l and Ref-s)
PAC was much larger than that of MAR, MAR was superior to PAC accounting for 77.5% of the total fouling, and the percentage of
in HA adsorption because micropores of PAC might be unacces- Rirr in Rf was just 16.8%. With adsorbent particles removed before
sible to high-MW HA molecules [20,29]. BSA is a standard model UF, MAR pretreatment dramatically reduced both the external and
protein with MW of  67 kDa [30]. Only parts of the surface pores internal fouling, but PAC pretreatment exerted little influence on
of PAC were accessible to BSA molecules, and thus the BSA the external fouling and just lowered the internal fouling by 35.1%.
adsorption capacity of PAC was lower than that of MAR. Unlike MAR particles present in UF feed water slightly increased the Ref-l
HA and BSA, SA molecules exhibited high hydrophilicity, and the and lowered the Ref-s, whereas the Ref was obviously elevated by
hydrophobic interaction between the adsorbent and SA was weak PAC particles and the increase overwhelmed the decrease in
[31,32]. Therefore, both MAR and PAC displayed very small SA internal fouling due to HA removal. In addition, MAR pretreatment
adsorption capacity. reduced the Rirr more efficiently than PAC pretreatment did, and
the adsorbent particles present in UF feed water showed negligible
3.2. Effects of MAR and PAC pretreatments on HA fouling influence on the Rirr.
Fig. 3 shows the effects of MAR and PAC pretreatments on HA
The effects of MAR and PAC pretreatments on HA fouling of PES fouling of CA membrane. As compared with the fouling of PES
membrane are shown in Fig. 2. As shown in Fig. 2(a), raw HA membrane, the fouling of CA membrane developed more slowly
resulted in significant membrane fouling and the average fouling and the percentages of internal fouling and irreversible fouling in
rate was 1.744 kPa/min. When adsorbent particles were removed total fouling were lower, which might be ascribed to the high
before UF, MAR and PAC pretreatments reduced the fouling rate by hydrophilicity of the CA membrane. Nevertheless, the influence of
61.5% and 9.3%, respectively. The TMP evolution of MARþHA was adsorption pretreatment on fouling of CA membrane was analo-
almost the same as that of MAR-treated HA, indicating that the gous to that on PES membrane.
presence of MAR particles in UF feed water exerted minor For better understanding of the difference between the effi-
influence on membrane fouling. In contrast, PAC particles present ciency of MAR and PAC pretreatments in HA fouling control, HA
in UF feed water obviously contributed to membrane fouling, and concentrations in the feed and permeate of PES and CA mem-
the final TMP of PAC þHA was higher than those of PAC-treated HA branes were measured (Fig. 4(a)). For raw HA, the rejection
and raw HA. coefficients of PES and CA membranes were 81.6% and 67.4%,
98 K. Li et al. / Journal of Membrane Science 471 (2014) 94–102

than the percentage of high-MW HA, indicating that part of HA

Feed water molecules smaller than membrane pore size were also rejected by
10
Permeate of PES membrane the membrane. Several mechanisms, including adsorption, coagu-
HA concentration (mg/L)

Permeate of CA membrane lation, gel layer formation and electrostatic repulsion, might be
8
responsible for the rejection of low-MW HA [5,36]. The higher
rejection coefficient of PES membrane in comparison with that of
6 CA membrane indicated that adsorption might be more pro-
nounced for the more hydrophobic membrane, which was in line
4 with the higher percentage of internal fouling in total fouling for
PES membrane (Figs. 2 and 3).
2 When adsorbent particles were removed before UF, membrane
fouling was determined by the residual HA in the solution after
0
adsorption pretreatment. MAR pretreatment reduced HA concen-
tration in feed water from 9.91 mg/L to 5.05 mg/L, but the
A

A
A

A
H

H
H

H concentrations in PES and CA membrane permeate just declined

H
ed

R+
ed
w

C+
at
Ra

at

by 0.20 and 0.23 mg/L, respectively, demonstrating that most of

tre

PA
tre

M
R-

C-

the HA adsorbed by MAR would have otherwise been rejected by

A

PA
M

the membrane. Thus, MAR pretreatment significantly reduced the

deposition of HA on/within the membrane and mitigated HA
2.5 fouling substantially. In contrast, PAC pretreatment just reduced
Feed water HA concentration in feed water by 1.76 mg/L, and the concentra-
Permeate of PES membrane
BSA concentration (mg/L)

2.0 tions in PES and CA membrane permeate were lowered by 0.91

Permeate of CA membrane and 1.15 mg/L, respectively. The effects of PAC pretreatment on HA
transmission suggested that a large proportion of HA removed by
1.5
PAC adsorption could permeate through the membrane and HA
deposition on the membrane was not reduced significantly, which
1.0 could explain the poor performance of PAC pretreatment in HA
fouling control. In consideration of the pore structure character-
0.5 istics of the adsorbents (Table 1), it was reasonable to speculate
that MAR pretreatment preferentially removed high-MW HA, and
that PAC adsorbed both high- and low-MW of HA though it was
0.0 less efficient in HA removal. The HPSEC analysis results of HA
A
A

A
A

before and after adsorption pretreatment (Fig. S1 in the

BS
BS

BS
BS

BS
ed

R+
w

Supplementary Information) proved the above speculation. When

ed

C+
at
Ra

at

PA
tre

tre

adsorbent particles were not removed before UF, membrane

R-

C-
A

PA

fouling involved the interactions among the adsorbent particles

M

and unadsorbed HA, which will be discussed in Section 3.5.

2.5 Feed water

3.3. Effects of MAR and PAC pretreatments on BSA fouling
Permeate of PES membrane
Permeate of CA membrane
Fig. 5 shows the effects of MAR and PAC adsorptions on BSA
SA concentration (mg/L)

2.0
fouling of PES membrane. As shown in Fig. 5(a), PES membrane
was severely fouled by raw BSA. MAR pretreatment almost
1.5
eliminated BSA fouling, and PAC pretreatment reduced BSA fouling
significantly. The presence of MAR particles in UF feed water
1.0 showed little influence on TMP buildup while PAC particles
slightly contributed to membrane fouling. Unlike the dominance
of external fouling in HA fouling, the internal fouling accounted for
0.5
52.3% of BSA fouling of PES membrane. Regardless of the presence
or absence of MAR particles in UF feed water, MAR pretreatment
0.0 substantially decreased both the external and internal fouling. PAC
SA
SA

SA
SA

SA

pretreatment reduced both the external and internal fouling

ed

R+
ed
w

C+
Ra

at
at

moderately, and PAC particles present in UF feed water increased

A

PA
tre

tre

M
R-

C-

external fouling to some extent. In the case of CA membrane

A

PA
M

which was more hydrophilic, the fouling caused by raw BSA was so
Fig. 4. Effects of MAR and PAC pretreatments on NOM concentrations in the feed slight that the efficiency of the adsorption pretreatments could not
and permeate of PES and CA membranes (a) HA, (b) BSA and (c) SA (error bars be significantly distinguished (Fig. 6).
represent standard deviations from the means, n¼ 3).
BSA concentrations in the feed and permeate of PES and CA
membranes are shown in Fig. 4(b). For raw BSA, the rejection
respectively. According to previous studies regarding Sigma- coefficients of PES and CA membranes were 53.1% and 16.8%,
Aldrich HA and conducted under similar solution environment, respectively. BSA is a homogeneous model protein with MW of
the proportion of HA fraction with MW larger than 100 kDa (the  67 kDa and there are usually some BSA aggregates in the
MWCO of membrane used in this study) was in the range 48–62% solution [37]. The much higher rejection coefficient of PES mem-
[33–35]. This high-MW HA fraction could be rejected by the brane suggested that hydrophobic adsorption played an important
membrane and mainly cause external fouling. But the observed role in BSA rejection and membrane fouling. According to the
rejection coefficients of PES and CA membranes were both higher distribution of fouling resistance shown in Fig. 5, it can be inferred
 K. Li et al. / Journal of Membrane Science 471 (2014) 94–102 99

80 30

25
60
20
TMP (kPa)

TMP (kPa)
40 15

Raw BSA
10
MAR-treated BSA PAC-treated BSA
20
Raw BSA MAR+BSA PAC+BSA
5
MAR-treated BSA PAC-treated BSA
MAR+BSA PAC+BSA
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Filtration time (min) Filtration time (min)

16 Ref-l Ref-s Rif-r Rirr 4

Ref-l Ref-s Rif-r Rirr

m -1)
m -1)

11
3
11

12

Fouling resistance (10

Fouling resistance (10

8 2

4 1

0 0

A
A

A
A

BS
BS
SA

SA
BS
BS

BS
BS

SA
BS
B

+B
d

w
ed

R+

te

te

B
C+
w

te

Ra

C+
ea

ea
at

R
Ra

ea

PA

A
tre

tr
-t r
M
-tr

PA
M
C-
R-

R
C

PA
A
PA
A

M
M

Fig. 5. Effects of MAR and PAC pretreatments on BSA fouling of PES membrane: Fig. 6. Effects of MAR and PAC pretreatments on BSA fouling of CA membrane:
(a) TMP buildup (only mean values were reported) and (b) fouling resistance (a) TMP buildup (only mean values were reported) and (b) fouling resistance
analysis (error bars represent standard deviations from the means) (n ¼3). analysis (error bars represent standard deviations from the means) (n¼ 3).

that BSA fouling of PES membrane was initially caused by pore
constriction due to adsorptive fouling, followed by pore blockage
and cake formation [38,39]. For the hydrophilic CA membrane, the
adsorptive fouling was very limited at the BSA concentration
employed in this study (2 mg/L) and the subsequent pore blockage
and cake formation were also minimized [23,40], thus the BSA
fouling of CA membrane was much lower than that of PES
membrane. As shown in Figs. 1 and 4(b), MAR was more efficient
than PAC in removing BSA. Therefore, it was rational that MAR
pretreatment mitigated BSA fouling more effectively than PAC
pretreatment did. When adsorbent particles were present in UF
feed water, the contribution of MAR and PAC particles to mem-
brane fouling were different and the mechanisms will be
discussed later.
SA is a typical type of hydrophilic polysaccharide [31,39];
according to Kim and Dempsey [34], 86.0% of SA molecules are
larger than 100 kDa and 11.2% of them are in the range of
30–100 kDa. Therefore, SA caused significant external fouling as
a result of cake/gel formation and the percentage of irreversible
fouling was very small. As shown in Fig. 4(c), the rejection
coefficients of PES and CA membranes for raw SA were 90.8%
and 87.5%, respectively, which were in accordance with the
properties of SA. As shown in Fig. 4(c), the SA concentration
hardly declined after MAR and PAC pretreatments, and thus
neither of them were able to reduce SA fouling. Moreover, neither
the presence of MAR nor PAC particles exerted detectable influ-
ence on SA fouling, which was distinctive from the situation of HA
and BSA and will be discussed later.

3.5. Summary of the effects of MAR and PAC pretreatments on

3.4. Effects of MAR and PAC pretreatments on SA fouling membrane fouling caused by different NOM fractions

The effects of MAR and PAC pretreatments on SA fouling of PES
membrane are shown in Fig. 7. The average fouling rate during raw
SA filtration was 2.260 kPa/min, and most of the fouling was
loosely-attached external fouling. The effects of both MAR and
PAC pretreatments on SA fouling were insignificant, irrespective of
the presence or absence of adsorbent particles in UF feed water. As
shown in Fig. 8, SA fouling of CA membrane was less severe than
that of PES membrane, and the influence of adsorption pretreat-
ment was also inconsiderable.
For each NOM fraction investigated in this study, fouling of PES
membrane was more severe than that of CA membrane more or
less. The results were consistent with previous studies and the
differences could be associated with the lower adsorptive fouling
of the CA membrane due to its high hydrophilicity [23,40].
Nevertheless, the fouling of the two types of membranes was
similarly affected by each adsorbent. The varied influence of MAR
and PAC pretreatments on PES membrane fouling caused by
different NOM fractions is summarized in Fig. 9.
100 K. Li et al. / Journal of Membrane Science 471 (2014) 94–102

80 60

60 45
TMP (kPa)

TMP (kPa)
40 30

20 Raw SA 15 Raw SA
MAR-treated SA PAC-treated SA MAR-treated SA PAC-treated SA
MAR+SA PAC+SA MAR+SA PAC+SA
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Filtration time (min) Filtration time (min)

16 Ref-l Ref-s Rif-r Rirr 10

Ref-l Ref-s Rif-r Rirr
m -1)

m )
-1
8
11

12

11
Fouling resistance (10

Fouling resistance (10

6
8
4

4
2

0 0
SA

SA
SA

SA

SA
SA

SA
SA

SA
+S
ed

ed

R+

C+
w

ed

ed
w

C+
at
Ra

at

R
at
A

PA

Ra

at
tre

A
tre

PA
t re
M

tre

M
R-

C-

R-

C-
A

PA

PA
M

Fig. 7. Effects of MAR and PAC pretreatments on SA fouling of PES membrane: Fig. 8. Effects of MAR and PAC pretreatments on SA fouling of CA membrane:
(a) TMP buildup (only mean values were reported) and (b) fouling resistance (a) TMP buildup (only mean values were reported) and (b) fouling resistance
analysis (error bars represent standard deviations from the means) (n¼ 3). analysis (error bars represent standard deviations from the means) (n¼ 3).

As the adsorbent particles were removed before UF, the that of NOM foulants (μm vs. nm), thus it can be assumed that
efficiency of adsorption pretreatment in fouling control depended NOM molecules would pass through the adsorbent particle layer if
on the adsorption capability and preference of the adsorbent. there were no interactions between adsorbent particles and NOM
As for HA which was heterogeneous in MW distribution, MAR molecules. As reported previously, both the resistance of MAR and
pretreatment decreased HA concentration in feed water much PAC layers in the absence of NOM were negligible in comparison
more efficiently than PAC pretreatment did and MAR preferentially with the clean membrane resistance [14,20]. Therefore, the influ-
adsorbed high-MW HA. As a result, MAR pretreatment signifi- ence of the presence of adsorbent particles on membrane fouling
cantly reduced the deposition of HA on the membrane and mainly depended on the interactions between adsorbent particles
mitigated HA fouling substantially, whereas PAC pretreatment just and NOM molecules and the subsequent change in the structure
slightly mitigated HA fouling, which was in accordance with and resistance of the particle layer. Several mechanisms might be
previous studies [8,17]. BSA molecules were relatively uniform in responsible for the change in the structure and resistance of the
MW, therefore MAR pretreatment removed more BSA and con- particle layer: alteration of particle surface properties due to
trolled BSA fouling more efficiently than PAC pretreatment did. For organic adsorption, hindered back diffusion of NOM molecules
SA which was hydrophilic and was hardly adsorbed by both MAR and rejection of NOM molecules by the particle layer [41–44].
and PAC, neither of the pretreatments apparently mitigated its These mechanisms were not exclusive and more than one of them
fouling. might act at the same time [42,45]. As shown in Fig. 4(a), both the
When there was no separation step before UF, adsorbent presence of MAR and PAC particles slightly decreased HA concen-
particles present in UF feed water were retained by the membrane tration in the permeate, suggesting that both MAR and PAC layers
along with part of the unadsorbed NOM. As presented in Sections rejected some HA molecules. But the hindered back diffusion
3.2–3.4, the influence of the presence of PAC particles on NOM mechanism could not be excluded because HA molecules
fouling was specific to the types of NOM fraction while the entrapped in the particle layer might have been adsorbed by the
presence of MAR particles exerted minor influence on all the adsorbent. Nonetheless, the internal pores of MAR were wide
three fractions. On the basis of the SEM images of the particle layer enough to accommodate HA with different MWs, therefore, most
formed on the surface of PES membrane with and without NOM HA molecules adsorbed by MAR entered the internal pores of MAR
foulant (Fig. S2 in the Supplementary Information), one possible and the resistance of MAR layer might be unaffected by the
explanation can be given as the following. The size of adsorbent presence of HA, which means that the resistance of the MAR layer
particles were more than three orders of magnitude larger than formed in the presence of HA was also negligible [14,20]. However,
 K. Li et al. / Journal of Membrane Science 471 (2014) 94–102
Fig. 9. Schematic of the influence of MAR and PAC pretreatments on membrane fouling caused by different NOM fractions.
although high-MW HA was excluded from the internal pores of
PAC, a considerable part of the HA adsorbed by PAC was high-MW
HA (Fig. 4(a) and Fig. S1), implying that this part of HA was
adsorbed in the external pores and/or on the particle surface.
Moreover, there was a large amount of high-MW HA in the
solution pretreated by PAC (Fig. S1). More high-MW HA would
adhere on the surface of PAC particles if hindered back diffusion
and rejection of HA molecules by particle layer occurred. HA
molecules adhered on the surface of PAC particles might interact
with each other, resulting in a PAC layer with lower porosity and
higher resistance in comparison with PAC layer formed in the
absence of NOM. BSA is also a hydrophobic NOM fraction, but its
MW is smaller than high-MW HA and the concentration used in
the experiment was much lower than that of HA. Therefore, the
decrease of PAC layer porosity in BSA fouling was not as significant
as that in HA fouling, and the presence of PAC particles just
resulted in a moderate increase in membrane fouling. For the
reason similar to that stated in HA fouling, the presence of MAR
particles exerted minor influence on BSA fouling. In the case of SA
which could be regarded as a non-interacting foulant, due to the
relatively low concentration of SA used in this study (2 mg/L), the
structures of MAR and PAC layers might remained almost the same
as that formed in the absence of NOM and membrane fouling was
not influenced visibly by the presence of MAR and PAC particles.
Overall, the influence of the presence of adsorbent particles on
membrane fouling was closely related with the textural properties
of the particle and the interactions between the particle and NOM
molecules; further studies are needed to explore the underlying
mechanisms.
It should be noted that the results obtained in this study were
based on the model foulants used in this experiment. There might
be some discrepancies in properties (e.g., MW, hydrophobicity)
between the model foulants used in this study and NOM in natural
waters or other model foulants, therefore, caution should be taken
in extrapolating the results to other systems. For instance, the MW
of HA from IHSS is smaller than that of Sigma-Aldrich HA used in
this study [36], and thus the efficiency of PAC pretreatment in
fouling control might be improved based on the above analysis;
meanwhile, the efficiency of MAR and PAC pretreatments might
vary with the source of feed water to some extent (Fig. S3 in the
Supplementary Information). Nevertheless, this study demonstrated
101
directly that the efficiency of adsorption pretreatment in NOM
fouling control was related to the characteristics of NOM and
adsorbent, and the underlying mechanisms were discussed preli-
minary. The findings reported here would contribute to the under-
standing of membrane fouling in the hybrid adsorption/UF process.
4. Conclusions
Performance of MAR and PAC pretreatments in mitigating UF
membrane fouling caused by different NOM fractions was system-
atically investigated in this study. The following conclusions can be
drawn.
(1) The adsorption capacity of MAR for different NOM fractions
was in the order of HA 4BSA4SA, while that of PAC followed
the order of BSA4 HA4SA. In addition, the adsorption capa-
city of MAR was higher than that of PAC for HA and BSA, and
both MAR and PAC exhibited very small SA adsorption
capacities.
(2) Although the fouling of CA membrane was lower than that of
PES membrane due to the high hydrophilicity of CA mem-
brane, fouling of the two membranes was similarly influenced
by each pretreatment.
(3) MAR pretreatment efficiently removed high-MW HA and
significantly reduced HA fouling, no matter whether MAR
particles were removed before UF or not. In contrast, PAC
pretreatment with PAC particles removed before UF slightly
alleviated HA fouling, whereas HA fouling was exacerbated
when PAC particles were present in UF feed water.
(4) MAR pretreatment mitigated BSA fouling more effectively than
PAC pretreatment did because MAR adsorption removed more
BSA. Moreover, the presence of MAR particles in UF feed water
caused little additional resistance, while PAC particles con-
tributed to membrane fouling moderately. By contrast, regard-
less of the presence or absence of adsorbent particles in UF
feed water, neither MAR nor PAC pretreatments exerted
significant influence on SA fouling.
(5) The efficiency of adsorption pretreatment in NOM fouling control
was associated with the characteristics of NOM and adsorbent.
As a result, from the view of membrane fouling control, the design
102 K. Li et al. / Journal of Membrane Science 471 (2014) 94–102
of a hybrid adsorption/UF system should fully consider the
composition and properties of NOM in the feed water.
Acknowledgment
This research was jointly supported by the National Natural
Science Foundation of China (51138008, 51308146), Program for
New Century Excellent Talents in University (NCET-13-0169), State
Key Laboratory of Urban Water Resource and Environment
(2014DX04), the Funds for Creative Research Groups of China
(51121062) and the Fundamental Research Funds for the Central
Universities (AUGA5710050713). Anonymous referees are greatly
acknowledged for their constructive and helpful suggestions.
Appendix A. Supporting information
Supplementary data associated with this article can be found in the
online version at http://dx.doi.org/10.1016/j.memsci.2014.08.006.
References
[1] H. Huang, K. Schwab, J.G. Jacangelo, Pretreatment for low pressure membranes
in water treatment: a review, Environ. Sci. Technol. 43 (2009) 3011–3019.
[2] P. Xiao, F. Xiao, D.-S. Wang, T. Qin, S.-P. He, Investigation of organic foulants
behavior on hollow-fiber UF membranes in a drinking water treatment plant,
Sep. Purif. Technol. 95 (2012) 109–117.
[3] E. Akhondi, F. Wicaksana, A.G. Fane, Evaluation of fouling deposition, fouling
reversibility and energy consumption of submerged hollow fiber membrane
systems with periodic backwash, J. Membr. Sci. 452 (2014) 319–331.
[4] G. Amy, Fundamental understanding of organic matter fouling of membranes,
Desalination 231 (2008) 44–51.
[5] I. Sutzkover-Gutman, D. Hasson, R. Semiat, Humic substances fouling in
ultrafiltration processes, Desalination 261 (2010) 218–231.
[6] F. Xiao, P. Xiao, W.J. Zhang, D.S. Wang, Identification of key factors affecting the
organic fouling on low-pressure ultrafiltration membranes, J. Membr. Sci. 447
(2013) 144–152.
[7] B. Ma, W. Yu, H. Liu, J. Qu, Effect of low dosage of coagulant on the
ultrafiltration membrane performance in feedwater treatment, Water Res. 51
(2014) 277–283.
[8] C.-F. Lin, T.-Y. Lin, O.J. Hao, Effects of humic substance characteristics on UF
performance, Water Res. 34 (2000) 1097–1106.
[9] M. Tomaszewska, S. Mozia, Removal of organic matter from water by PAC/UF
system, Water Res. 36 (2002) 4137–4143.
[10] J. Kim, Z. Cai, M.M. Benjamin, Effects of adsorbents on membrane fouling by
natural organic matter, J. Membr. Sci. 310 (2008) 356–364.
[11] M. Campinas, M.J. Rosa, Assessing PAC contribution to the NOM fouling control
in PAC/UF systems, Water Res. 44 (2010) 1636–1644.
[12] Z. Cai, M.M. Benjamin, NOM fractionation and fouling of low-pressure
membranes in microgranular adsorptive filtration, Environ. Sci. Technol. 45
(2011) 8935–8940.
[13] S.-K. Kang, K.-H. Choo, Why does a mineral oxide adsorbent control fouling
better than powdered activated carbon in hybrid ultrafiltration water treat-
ment? J. Membr. Sci. 355 (2010) 69–77.
[14] C. Stoquart, P. Servais, P.R. Bérubé, B. Barbeau, Hybrid membrane processes
using activated carbon treatment for drinking water: a review, J. Membr. Sci.
411–412 (2012) 1–12.
[15] M. Zhang, C. Li, M.M. Benjamin, Y. Chang, Fouling and natural organic matter
removal in adsorbent/membrane systems for drinking water treatment,
Environ. Sci. Technol. 37 (2003) 1663–1669.
[16] Y. Zhang, J. Tian, J. Nan, S. Gao, H. Liang, M. Wang, G. Li, Effect of PAC addition
on immersed ultrafiltration for the treatment of algal-rich water, J. Hazard.
Mater. 186 (2011) 1415–1424.
[17] C.-W. Li, Y.-S. Chen, Fouling of UF membrane by humic substance: effects of
molecular weight and powder-activated carbon (PAC) pre-treatment, Desali-
nation 170 (2004) 59–67.
[18] M.M. Clark, W.-Y. Ahn, X. Li, N. Sternisha, R.L. Riley, Formation of polysulfone
colloids for adsorption of natural organic foulants, Langmuir 21 (2005)
7207–7213.
[19] L.C. Koh, W.-Y. Ahn, M.M. Clark, Selective adsorption of natural organic
foulants by polysulfone colloids: effect on ultrafiltration fouling, J. Membr.
Sci. 281 (2006) 472–479.
[20] K. Li, F. Qu, H. Liang, S. Shao, Z.-s. Han, H. Chang, X. Du, G. Li, Performance of
mesoporous adsorbent resin and powdered activated carbon in mitigating
ultrafiltration membrane fouling caused by algal extracellular organic matter,
Desalination 336 (2014) 129–137.
[21] M. Peter-Varbanets, J. Margot, J. Traber, W. Pronk, Mechanisms of membrane
fouling during ultra-low pressure ultrafiltration, J. Membr. Sci. 377 (2011)
42–53.
[22] J.-y. Tian, M. Ernst, F. Cui, M. Jekel, Effect of particle size and concentration on
the synergistic UF membrane fouling by particles and NOM fractions,
J. Membr. Sci. 446 (2013) 1–9.
[23] F. Qu, H. Liang, J. Zhou, J. Nan, S. Shao, J. Zhang, G. Li, Ultrafiltration membrane
fouling caused by extracellular organic matter (EOM) from Microcystis
aeruginosa: effects of membrane pore size and surface hydrophobicity,
J. Membr. Sci. 449 (2014) 58–66.
[24] T. Li, A.W.-K. Law, M. Cetin, A.G. Fane, Fouling control of submerged hollow
fibre membranes by vibrations, J. Membr. Sci. 427 (2013) 230–239.
[25] W.-z. Yu, H.-j. Liu, L. Xu, J.-h. Qu, N. Graham, The pre-treatment of submerged
ultrafiltration membrane by coagulation—effect of polyacrylamide as a coa-
gulant aid, J. Membr. Sci. 446 (2013) 50–58.
[26] R. Jiraratananon, D. Uttapap, P. Sampranpiboon, Crossflow microfiltration of a
colloidal suspension with the presence of macromolecules, J. Membr. Sci. 140
(1998) 57–66.
[27] H. Ma, L.F. Hakim, C.N. Bowman, R.H. Davis, Factors affecting membrane
fouling reduction by surface modification and backpulsing, J. Membr. Sci. 189
(2001) 255–270.
[28] X. Wang, L. Shu, Y. Wang, B. Xu, Y. Bai, S. Tao, B. Xing, Sorption of peat humic
acids to multi-walled carbon nanotubes, Environ. Sci. Technol. 45 (2011)
9276–9283.
[29] C. Moreno-Castilla, Adsorption of organic molecules from aqueous solutions
on carbon materials, Carbon 42 (2004) 83–94.
[30] K. Xiao, X. Wang, X. Huang, T.D. Waite, X. Wen, Analysis of polysaccharide,
protein and humic acid retention by microfiltration membranes using Tho-
mas' dynamic adsorption model, J. Membr. Sci. 342 (2009) 22–34.
[31] K. Katsoufidou, S.G. Yiantsios, A.J. Karabelas, Experimental study of ultrafiltra-
tion membrane fouling by sodium alginate and flux recovery by backwashing,
J. Membr. Sci. 300 (2007) 137–146.
[32] K. Xiao, X. Wang, X. Huang, T.D. Waite, X. Wen, Combined effect of membrane
and foulant hydrophobicity and surface charge on adsorptive fouling during
microfiltration, J. Membr. Sci. 373 (2011) 140–151.
[33] K. Katsoufidou, S.G. Yiantsios, A.J. Karabelas, A study of ultrafiltration mem-
brane fouling by humic acids and flux recovery by backwashing: experiments
and modeling, J. Membr. Sci. 266 (2005) 40–50.
[34] H.-C. Kim, B.A. Dempsey, Membrane fouling due to alginate, SMP, EfOM,
humic acid, and NOM, J. Membr. Sci. 428 (2013) 190–197.
[35] J. Shao, J. Hou, H. Song, Comparison of humic acid rejection and flux decline
during filtration with negatively charged and uncharged ultrafiltration mem-
branes, Water Res. 45 (2011) 473–482.
[36] W. Yuan, A.L. Zydney, Humic acid fouling during ultrafiltration, Environ. Sci.
Technol. 34 (2000) 5043–5050.
[37] S.T. Kelly, W. Senyo Opong, A.L. Zydney, The influence of protein aggregates on
the fouling of microfiltration membranes during stirred cell filtration,
J. Membr. Sci. 80 (1993) 175–187.
[38] D. Jermann, W. Pronk, S. Meylan, M. Boller, Interplay of different NOM fouling
mechanisms during ultrafiltration for drinking water production, Water Res.
41 (2007) 1713–1722.
[39] Y. Ye, P.L. Clech, V. Chen, A.G. Fane, Evolution of fouling during crossflow
filtration of model EPS solutions, J. Membr. Sci. 264 (2005) 190–199.
[40] S. Zhou, Y. Shao, N. Gao, L. Li, J. Deng, C. Tan, M. Zhu, Influence of hydrophobic/
hydrophilic fractions of extracellular organic matters of microcystis aerugi-
nosa on ultrafiltration membrane fouling, Sci. Total Environ. 470–471 (2014)
201–207.
[41] Z. Cai, C. Wee, M.M. Benjamin, Fouling mechanisms in low-pressure mem-
brane filtration in the presence of an adsorbent cake layer, J. Membr. Sci. 433
(2013) 32–38.
[42] 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.
[43] D. Jermann, W. Pronk, M. Boller, Mutual influences between natural organic
matter and inorganic particles and their combined effect on ultrafiltration
membrane fouling, Environ. Sci. Technol. 42 (2008) 9129–9136.
[44] Q. Li, M. Elimelech, Synergistic effects in combined fouling of a loose
nanofiltration membrane by colloidal materials and natural organic matter,
J. Membr. Sci. 278 (2006) 72–82.
[45] A.H. Taheri, L.N. Sim, C.T. Haur, E. Akhondi, A.G. Fane, The fouling potential of
colloidal silica and humic acid and their mixtures, J. Membr. Sci. 433 (2013)
112–120.