University of Wollongong

Research Online
Faculty of Engineering - Papers (Archive) Faculty of Engineering and Information Sciences

2011

Effects of membrane fouling and scaling on boron

rejection by nanofiltration and reverse osmosis
membranes
Kha Tu
University of Wollongong, lkt883@[Link]

Allan Chivas
University of Wollongong, allan_chivas@[Link]

Long D. Nghiem
University of Wollongong, longn@[Link]

Publication Details
Tu, K. Le., Chivas, A. Nghiem, L. D. (2011). Effects of membrane fouling and scaling on boron rejection by nanofiltration and reverse
osmosis membranes. Desalination, 279 (1-3), 269-277.

Research Online is the open access institutional repository for the

University of Wollongong. For further information contact the UOW
Library: research-pubs@[Link]
 Effects of membrane fouling and scaling on boron rejection
by nanofiltration and reverse osmosis membranes

Revised Manuscript Submitted to

Desalination

May 2011

Kha L. Tu 1, Allan R. Chivas 2 and Long D. Nghiem 1,*

1
Strategic Water Infrastructure Laboratory and GeoQuEST Research Centre

School of Civil Mining and Environmental Engineering

University of Wollongong, NSW 2522, Australia

2
GeoQuEST Research Centre, School of Earth & Environmental Sciences

University of Wollongong, NSW 2522, Australia

_______________________

* Corresponding author: Long Duc Nghiem, Email: longn@[Link], Ph +61 2 4221 4590

1
Abstract
The effects of membrane fouling on the performance of nanofiltration and reverse osmosis
membranes with respect to boron rejection and permeate flux were investigated in this study.
A nanofiltration (NF270) membrane and a reverse osmosis (BW30) membrane were used in
this investigation. Four typical membrane fouling conditions were simulated under controlled
laboratory conditions in a cross-flow membrane system using four model foulants including
humic acid, sodium alginate, colloidal silica and CaSO4. Among these model foulants, humic
acid was found to increase boron rejection whereas the other foulants led to a decrease in
boron rejection. Properties of foulants were found to be an important factor that determined
the transport of boron through the fouling layer. Results reported in this study also indicate
that the extent of flux decline caused by different model foulants differed substantially from
one another. The impact of membrane fouling on permeate flux decline was found to be
dependent on the initial permeate flux and hydrophobicity of the foulant. On the other hand,
membrane scaling was found to be governed by the salt rejection efficiency of the membrane.
Cake-enhanced concentration polarisation appears to be a major mechanism that affects
boron rejection efficiency of fouled membranes.

Keywords: boron, cake-enhanced concentration polarisation, fouling, scaling,

nanofiltration, reverse osmosis.

1. Introduction

The rejection of boron by nanofiltration (NF) and reverse osmosis (RO) membranes has been
extensively investigated in recent years due to a growing interest in seawater desalination to
supplement potable water supply [1-2]. However, most studies currently available in the open
literature focus on the rejection of boron under virgin membrane conditions. A notable
exception is a study by Huertas et al. [3] who reported a significant decrease in boron
rejection by NF/RO membranes under biologically fouled conditions. This observation was
later confirmed by a mathematical model developed by Oh et al. [4]. Given that membrane
fouling is inevitable for any membrane desalination plant, it is essential to investigate the
boron removal efficiency under fouled membrane conditions.

Membrane fouling can substantially decrease permeability, increase operational cost and
shorten membrane life [5-8]. Organic fouling typically exhibits a range of behaviour due to

2
the complex interactions between chemical functional groups of organic foulants and those of
the polymeric membrane skin layer [9-11]. Alginate, a polysaccharide compound, was
reported to be able to block the “valleys” made by roughness on the membrane surface, and
also form a dense cake layer composed of cross-links between carboxylate functional groups
on neighbouring alginate molecules by calcium ions bridging [12]. Therefore, severe flux
decline is usually observed when the membrane is fouled by alginate substances.
Physiochemical properties of membranes also play an important role in the extent of organic
fouling. A study by Nghiem and Hawkes [13] revealed that permeate flux decline due to
membrane fouling would be more severe with membranes having larger pore size. The
authors also argued that pore blocking was the predominant fouling mechanism at the first
stage of fouling, and the latter stage is governed by cake-enhanced mechanism. Several
studies found that higher negative zeta potential and hydrophilicity of the membrane surface
should lead to less fouling by organic macromolecules due to higher electrostatic repulsion
and lower hydrophobic interactions between the foulant and membrane surface [12, 14]. In
addition, it is reported that solution conditions such as pH and ionic strength also contribute
to the extent of membrane fouling by organic molecules [7, 14-17].

The impact of colloidal fouling on permeate flux decline has been extensively reported in the
literature [8, 18-20]. Cohen and Probstein [21] found a linear connection between permeate
flux decline and foulant layer thickness during the initial stages of fouling. This observation
was attributed to the enhanced osmotic pressure caused by colloidal cake layer which
obstructs the back diffusion of salt ions from the membrane surface to the bulk solution [22-
24]. This phenomenon was also utilised to elucidate the increasing salt passage through the
membrane which was usually observed when colloidal fouling occurred. Furthermore, Lee et
al. [20] reported that the decrease in salt rejection of NF membranes was more severe than
that of RO membranes because the cake layer could reduce the membrane charged exclusion
which was more important in NF than RO membranes. The extent of colloidal deposition on
the membrane surface was also found to be affected by properties of colloid (size and
hydophobicity) [25] and of the membrane surface (membrane roughness and hydrophobicity)
[26-28].

During the membrane filtration process, sparingly soluble salts such as CaCO3 and CaSO4
may become supersaturated and precipitate on the membrane surface to cause membrane
scaling. According to Lee et al. [29], the flux decline in cross-flow NF could be attributed to

3
not only surface crystallisation but also due to bulk crystallisation. In fact, in a similar study,
Gilron and Hasson [30] attributed bulk rather than surface crystallisation to the observed flux
decline. Pervov [31] described the scaling process as the crystal formation took place in the
bulk solution due to strong oversaturation in the deadlocks, and then the crystals approached
and precipitated on the membrane surface. The impact of membrane scaling on salt rejection
has not been extensively investigated. Scaling of divalent cations such as Ca2+ may induce
more positive charge to the membrane surface, and consequently reduce the rejection of
charge solutes [32].

This study aims to investigate the effects of membrane fouling on the rejection of boron by
NF/RO membranes. Four typical membrane fouling conditions were simulated under
controlled chemical and physical conditions in a laboratory-scale cross-flow membrane
system using four model foulants. Membrane fouling was systematically related to the
membrane and foulant characteristics. The separation behaviours of boron under different
solution pH were subsequently described and discussed in detail.

2. Materials and methods

2.1. Chemicals and reagents

Unless otherwise stated, all experiments were conducted based on an electrolyte background
that includes NaCl, CaCl2, and NaHCO3 at concentrations of 10 mM, 1 mM, and 1 mM,
respectively. Boron in the form of boric acid (H3BO3) was spiked into the feed solution at a
level of 0.43 mM (or 4.6 mgL-1 as boron) to represent the average boron concentration in
seawater [33]. NaHCO3 was used as a buffer reagent. Adjustment of the feed water pH was
carried out by adding an appropriate volume of 1M NaOH or 1M HCl. Humic acid, sodium
alginate, colloidal silica and CaSO4 were used as model foulants and scalant to simulate
organic substance, colloidal matter, and inorganic divalent salt that are ubiquitous in natural
water sources. These model foulants were purchased from Sigma-Aldrich (Castle Hill,
Australia) except CaSO4 which was purchased from Ajax Chemicals (Auburn, Australia).
The Aldrich humic acid and alginate have molecular weights in the range of 12 – 80 kDa and
4 – 100 kDa, respectively. Colloidal silica (Ludox HS30) is monodispersed colloid particles
which are negatively charged at neutral or high pH rendering them somewhat hydrophilic.
The colloidal particle has an average hydrodynamic diameter of 18 nm [28] and was supplied

4
at 30% weight suspension in water and was stored at 4 oC. Suprapur grade nitric acid from
Merck Co. (Darmstad, Germany) was used for dilution and sample preparation prior to
inductively-coupled plasma mass spectrometry (ICP-MS) analysis. All other chemicals used
in this study are of analytical grade.

2.2. Membrane filtration system and experimental protocol

A NF membrane (NF270) and a RO membrane (BW30) were used. Both membranes were
supplied by Dow FilmTecTM (Minneapolis, MN, USA) as flat sheet samples and were stored
dry.

A laboratory-scale NF/RO filtration system equipped with a cross-flow stainless steel

membrane cell was used in this study. The cell could hold a flat sheet membrane sample with
an effective membrane area of 40 cm2 (4 cm x 10 cm). The channel height of the cell was 2
mm. The unit utilized a Hydra-Cell pump (Wanner Engineering Inc., Minneapolis, MN)
capable of providing pressures up to 6,800 kPa and a flow rate of 4.2 Lmin-1. Feed pressure
and cross-flow velocity were controlled by a bypass valve and a back-pressure regulator. The
temperature of the test solution was kept constant using a chiller/heater (Neslab RTE 7)
equipped with a stainless steel heat exchanger coil, which was submerged directly into a
stainless steel reservoir. Permeate flow was measured by a digital flow meter (Optiflow 1000,
Agilent Technologies, Palo Alto, CA) connected to a computer, and the cross-flow rate was
monitored by a rotameter. Permeate and retentate flows were recycled back to the feed
reservoir.

The fouling and subsequent rejection test protocols were performed in three stages:
compacting, fouling development, and rejection test. At first, the membrane was compacted
using 9 L deionized water at a pressure of approximately 500 kPa higher than the rejection
test pressure. Membrane compaction was conducted for at least 1 h until a stable baseline flux
was obtained. The electrolyte solution was then added to the feed reservoir, and made up to
the total feed volume of 10 L. The fouling layer was then allowed to develop by dosing one
of the nominated model foulants (humic acid, sodium alginate, and colloidal silica) at a
concentration of 20 mgL-1 (in total mass). Membrane scaling was simulated by adding CaSO4
to the electrolyte solution to make up 1 gL-1 of CaSO4. The permeate flux was then adjusted
to be 84 Lm-2h-1 for NF270 membrane and 60 Lm-2h-1 for the BW30 membrane. The fouling
development was carried out for 18 h and the feed solution pH was kept at 8.2. After the

5
fouling development step, boric acid was spiked to the feed solution at the concentration of
0.43 mM. The cross-flow velocity and permeate flux were then adjusted to be 30.4 cms-1 and
40 Lm-2h-1, respectively. The temperature of the feed solution was kept constant at 20 ± 0.1
o
C during the experiment. To assess the impact of solution pH on the rejection of boron, the
solution pH was raised to 11 by adding an appropriate volume of 1M NaOH, and then the pH
was incrementally dropped to pH 6 by adding 1M HCl. At each pH value, the system was
stablised for 1 h prior to the collection of feed and permeate samples for electrical
conductivity and ICP-MS analysis. The system was operated under a full recirculation mode
where both permeate and retentate were re-circulated to the feed tank. The permeate flux was
kept constant during the experiment to minimize the effect of flux on rejection.

2.3. Zeta potential measurement

The zeta potential of the membrane surface was measured by a SurPASS electrokinetic
analyser (Anton Paar GmbH, Graz, Austria). The zeta potential of the membrane surface was
calculated from the measured streaming potential using the Fairbrother-Mastin approach [34].
All streaming potential measurements were conducted in the background electrolyte which
was previously described in Section 2.1. HCl (1 M) and KOH (1 M) solutions were used to
adjust the pH by manual titration. The test solution was used to thoroughly flush the cell prior
to the pH adjustment for each measurement. All streaming potential measurements were
performed at room temperature of approximately 25 oC, which was monitored by the
temperature probe of the instrument.

2.4. Surface roughness morphology

Surface roughness of the membranes was characterised by a multimode atomic force

microscope (AFM) (Digital Instruments, Santa Barbara, CA). Imaging was taken in air in
tapping mode operation using an oxide sharpened SiN probe. The membrane average surface
roughness was determined in triplicate mode by AFM image analysis over a 2 µm × 2 µm
surface area.

2.5. Contact angle measurement

Contact angle measurements were conducted by a Rame-Hart Goniometer (Model 250,

Rame-Hart, Netcong, NJ) using the standard sessile drop method. Milli-Q water was used as
the reference solvent. The membranes were dried in the air before the measurements. Contact
6
angles on both sides of the droplet were measured. At least 5 droplets on each membrane
sample were measured, and twenty counts of each droplet were recorded.

2.6. Scanning electron microscope

The clean and fouled membranes were visually characterised with a JSM-6490LA (JEOL
Japan) scanning electron microscope (SEM). Before introduction to the electron beam, the
membrane sample was coated with a thin layer of carbon using a carbon sputter. SEM images
were taken at a magnification of 2,500 fold at 20kV. The elemental analysis was determined
using an integrated energy-dispersive spectrometer (EDS).

2.7. Analytical methods

The concentrations of boron, sodium and calcium were analysed using an Agilent 7500CS
ICP-MS (Agilent Technologies, Wilmington, DE, USA). A lithium internal standard (BDH
Spectrosol, Poole, U.K.) was spiked to all samples at the concentration of 40 µgL-1. Sample
dilution was carried out with 5% Suprapur nitric acid with a dilution factor of up to 20. To
avoid contamination, all apparatus related to preparing samples were plastic materials, and
were soaked in 5% Suprapur nitric acid for at least 24 h before being used. Calibration was
conducted prior to each batch of analysis. The linear regression coefficients for all calibration
curves were greater than 0.99 for all elements. Prior to each batch of analyses, the ICP-MS
was tuned by a multi-element tuning solution that contained 10 µgL-1 of Li, Y, Ce, Tl and Co.
Each analysis was conducted in triplicate and the variation was always less than 5%.
Conductivity and pH were measured using an Orion 4-Star Plus pH/conductivity meter
(Thermo Scientific, Beverly, MA).

3. Results and discussion

3.1. Characterization of virgin and fouled membranes

According to the manufacturer, the two membranes used in this study are made of an ultra
thin polyamide skin layer on top of a microporous polysulfone supporting layer. The former
governs the separation characteristics of the membrane whereas the later provides mechanical
strength to the membrane. The NF270 membrane has an average pore diameter of 0.84 nm
[35], whereas the BW30 is considered to have a non-porous active skin layer. The rejection
of typical salts (sodium and calcium) and pure water permeability of the membranes directly

7
correspond to the membrane pore sizes (Table 1). The virgin NF270 membrane has a very
smooth surface with a surface roughness of 4.1 nm compared to that of the virgin BW30
membrane with a surface roughness of 62.6 nm (Table 1). The high surface roughness of the
BW30 membrane could render this membrane more susceptible to fouling because foulant
particles could accumulate in the valleys on the membrane surface due to higher local flux
over valley regions [27]. Although the contact angle (which measures the hydrophobicity of
the membrane surface) does not directly affect the rejection of solute, this parameter can
determine the transport of water through membranes which governs the overall rejection of a
membrane system [36].

Table 1. Properties of the membranes used in this study.

Average pore Na+ Ca2+ Pure water Contact Surface

Membrane diameter rejection b rejection b permeability angle roughness c
(nm) (%) (%) (Lm-2h-1bar-1) (◦) (nm)
a
NF270 0.84 35.8 57.5 14.0 28.8 ± 2.4 4.1
BW30 not applicable 96.2 97.7 3.5 54.6 ± 3.0 62.6
a
Ref [35].
b
Rejection data were recorded at pH 8.
c
Scanning area 2 µm × 2 µm.

Because the polyamide layer that makes up the membrane active skin contains both
carboxylic and amine functional groups that can ionize in an aqueous solution [14], the
membrane surface zeta potential can vary as a function of the solution chemistry, such as pH
and ionic strength. Schäfer et al. [37] reported that a more negative membrane zeta potential
could lead to a higher salt rejection due to an enhanced electrostatic interaction between the
negatively charged membrane surface and charged solutes. The membranes selected in this
study have negative charge in the investigated pH range (Figure 1). In addition, their negative
surface charge density increases as the solution pH increases. This phenomenon suggests that
electrostatic interaction can be an important rejection mechanism of charged solutes,
particularly for the NF membrane.

8
 10
BW30
NF270
0

Zeta potential (mV)

-10

-20

-30

2 4 6 8 10 12
pH

Figure 1. Zeta potential of the selected membranes (measured at 25 oC, in a background

electrolyte solution containing NaCl, CaCl2, and NaHCO3 at concentrations of 10 mM, 1
mM, and 1 mM, repectively; pH was adjusted using HCl or KOH solutions).

The contact angle data which reflect the hydrophobicity of the virgin and fouled membranes
are illustrated in Figure 2. The virgin BW30 membrane appeared to be more hydrophobic
than the virgin NF270 membrane. Higher hydrophobicity could make the BW30 membrane
become vulnerable to fouling due to hydrophobic interaction between membrane surface and
hydrophobic foulants. However, despite having different hydrophobicity in virgin condition,
these two membranes showed very similar contact angle values once they were fouled by the
same foulant (Figure 2). This observation suggests that hydrophobicity, and probably other
physiochemical properties of the fouled membranes are governed by the fouling layer rather
than the aromatic polyamide active layer of the membrane surface.

9
 80 BW30

Contact angle ( )
o
60

40

20

0
Clean Humic acid Silica colloid Na alginate CaSO4

80
NF270
Contact angle ( )
o

60

40

20

0
Clean Humic acid Silica colloid Na alginate CaSO4

Figure 2. Contact angle values of the investigated clean and fouled membranes. Milli-Q
water was used as the reference solvent. The error bars show the standard deviation of five
replicate measurements.

SEM analysis confirmed that at the completion of the fouling development stage, the
membrane surface was completely covered by the fouling layer (Figure 3). Because the
model foulants are significantly larger than the pore size of the NF270 membrane, permeation
of the organic and silica colloids through the membrane did not occur as could be confirmed
by the low turbidity of the permeate samples. Surface morphology and composition of the
virgin membrane (Figure 3a) differed distinctively from that of the fouled membranes (Figure
3b, 3c, 3d, 3e). Humic acid, alginate and colloidal silica formed a dense and uniform cake
layer on the membrane surface as a result of hydrophobic interactions between the foulants
and membrane surfaces. On the other hand, CaSO4 crystallised and precipitated on the
membrane surface to establish a spongy and coarse layer that highlighted the foulant-foulant
interactions. This CaSO4 scaling layer is expected to affect the membrane permeate flux by a
lesser extent than that by the humic, alginate and colloidal silica foulants. Elementary
compositions of the fouling layer obtained from qualitative EDS analysis were consistent
with key signature elements of the corresponding foulants (Figure 3). In addition to the model
foulants, carbon, oxygen and sulfur were parts of the membrane polymeric composition and
thus were detected in all samples, including the virgin membrane (Figure 3a). A high level of
calcium was found in the alginate fouling layer (Figure 3c). This result is consistent with

10
previous studies that calcium could make cross-links with alginate molecules and accumulate
in the alginate fouling layer [12]. Silica was the most abundant element of the colloidal silica
fouling layer (Figure 3d). Similarly, the CaSO4 scaling could also be confirmed by the
dominant presence of sulfur and calcium on the membrane surface (Figure 3e).

Figure 3. SEM images and EDS data of the (a) virgin NF270 membrane, membrane surfaces
fouled by (b) humic acid, (c) sodium alginate, (d) colloidal silica, and (e) CaSO4 scaling. The
membrane samples were coated with a thin layer of carbon using a carbon sputter.

11
3.2. Membrane fouling behaviour

Figure 4 shows the alteration of permeate flux as a function of fouling development time
(Figure 4a) and as a function of accumulated mass of foulant on the membrane surface
(Figure 4b). The latter was calculated based on a simple mass-balance assuming that the
foulant was completely retained by the membrane and that tangential transport of the foulant
away from the membrane surface was negligible. Membrane fouling of the BW30 was less
severe in comparison to that of the NF270 membrane, with CaSO4 being the only exception
(Figure 4). This proportional behaviour between the initial permeate flux decline rate and
membrane pore size was reported previously [13]. The greater membrane fouling observed
with the NF270 membrane could be attributed to the higher operational initial permeate flux
(84 Lm-2h-1) in comparison to that (60 Lm-2h-1) of the BW30 membrane. The higher initial
permeate flux introduced more foulant to the membrane surface, and subsequently
encouraged fouling on the NF270 membrane (Figure 4b). It is noteworthy that the virgin
NF270 membrane possesses a lesser surface roughness, lower hydrophobicity (Table 1) and
higher negative zeta potential (Figure 1) which could subsequently lead to lower fouling
potential compared to the BW30 membrane. The data reported here (Figure 4) suggest that
the impact of surface roughness, hydrophobicity and zeta potential was overwhelmed by the
high initial permeate flux that makes the NF270 membrane more susceptible to fouling
compared to the BW30 membrane. On the other hand, the BW30 membrane was more
severely influenced by CaSO4 scaling than the NF270 membrane (Figure 4). In this case,
higher salt rejection efficiency of the BW30 membrane compared to that of the NF270
membrane generated a greater concentration polarisation at the membrane surface.
Consequently, membrane scaling caused by the precipitation of CaSO4 was more severe for
the BW30 than for the NF270 membrane. In addition, Figure 4b shows that at the same
accumulated masses, different foulants caused different extent of flux decrease. This implies
that the properties of foulants play the key role in governing the fouling extent.

12
 1.0 1.0
Normalised permeate flux (J/J0)

Normalised permeate flux (J/J0)

0.8 0.8

0.6 0.6

0.4 0.4
NF270 BW30
Humic acid Humic acid
0.2 Na alginate 0.2 Na alginate
Silica colloid Silica colloid
CaSO4 (a1) CaSO4 (a2)
0.0 0.0
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
Time (hour) Time (hour)

1.0 1.0

Normalised permeate flux (J/J0)

Normalised permeate flux (J/J0)

0.8 0.8

0.6 0.6

0.4 0.4
NF270 BW30
Humic acid Humic acid
0.2 0.2 Na alginate
Na alginate (b2)
(b1) Silica colloid
Silica colloid
CaSO4 CaSO4
0.0 0.0
0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5
-2 -2
Accumulated mass of foulant (mgcm ) Accumulated mass of foulant (mgcm )

Figure 4. Normalised permeate flux as a function of (a) time and (b) accumulated mass of
foulant on the membrane surface. Initial permeate flux: 84 Lm-2h-1 for NF270 membrane and
60 Lm-2h-1 for BW30 membrane, respectively. Feed solution: 10 mM NaCl, 1 mM CaCl2, 1
mM NaHCO3, and 20 mgL-1 of each foulant, except CaSO4 was 1 gL-1. For data presentation
purposes, the accumulated mass of CaSO4 shown in the figure has been divided by 50.

Membrane fouling can occur in two successive stages where foulant-membrane interactions
determine fouling mechanisms at the first stage, and the latter stage is governed by foulant-
foulant interactions [12-13, 38]. The rapid decrease in permeate flux at the first stage implied
that foulant-membrane interactions played a dominant role in fouling development. This is
the case of the NF270 membranes fouled by humic acid, sodium alginate and colloidal silica
(Figure 4-1). A very sharp and substantial flux drop was observed with these fouled
membranes within the first hour, and then the decrease rate became noticeably lower, even
inconsiderable in case of the colloidal silica fouling. The foulant-membrane interactions

13
could be the hydrophobic interaction between the organic foulants and the membrane surface
[12, 14] and the clogging impact of the foulant particles to the valleys on the membrane
surface [12]. This phenomenon seemed not occur at the CaSO4 scaling membrane because of
the large size of the CaSO4 crystals. The later stage of fouling development, which is
governed by foulant-foulant interactions, caused less effect on permeate flux decline. In
particular, the presence of Ca2+ in the feed solution could cause severe impact on the flux
decline because Ca2+ could bridge carboxylate functional groups on neighbouring alginate
molecules and so make the fouling layer thicker and denser [10, 12, 39-40]. This finding is
verified by the high level of calcium found in the fouling layer (Figure 3c). In general, cake-
enhanced concentration polarisation may occur and reduce the permeate flux of all the fouled
membranes. In this phenomenon, back diffusion of solutes at the membrane surface to bulk
solution is hindered by the fouling layer, and results in a substantial increase in solute
concentration at the membrane surface [23]. The increase in solute concentration led to an
increase in osmotic pressure and subsequently a decrease in permeate flux [41].

3.3. Effects of fouling and pH on the rejection of boron and salts

Boron rejection by both virgin and fouled membranes was greatly affected by solution pH
(Figure 5). An increase in the solution pH could substantially increase boron rejection by
NF/RO membranes, and the NF270 membrane was more sensitive to this effect than the
BW30 membrane (Figure 5). In particular, boron rejection by the NF270 membrane under
both virgin and fouled conditions was negligible at pH lower than 8, and the rejection
increased up to 40 – 60% when the solution pH increased to 11. As a notable exception, the
humic acid fouled NF270 membrane showed a rejection value of almost 30% below pH 8.
This observation will be further discussed in a later section. On the other hand, boron
rejection by the BW30 membrane in virgin and fouled conditions increased by only about
35% when the solution pH increased from 6 to 11 (Figure 5). Mechanisms governing boron
rejection by virgin NF/RO membranes have been elucidated in previous studies as size
exclusion (at low pH) and charged repulsion (at high pH) [42-44]. The similar behaviour of
the fouled and virgin membranes as solution pH increased (Figure 5) implied that
mechanisms governing boron rejection by virgin membranes did not change substantially
when the membranes were fouled, and therefore these mechanisms could be applied to
explain boron rejection by fouled membranes.

14
 100

BW30 NF270
Clean Clean
Humic acid Humic acid
80 Na alginate Na alginate
Colloidal silica Colloidal silica
CaSO4 CaSO4

Boron rejection (%)

60

40

20

0
6 7 8 9 10 11
pH

Figure 5. Boron rejection by virgin and fouled NF270 and BW30 membranes as a function of
solution pH. Feed solution: 10 mM NaCl, 1 mM CaCl2, 1 mM NaHCO3, 0.43 mM B(OH)3
and 20 mgL-1 of each foulant, except CaSO4 was 1 gL-1; feed temperature = 20 oC; permeate
flux = 40 Lm-2h-1; cross flow velocity = 30.4 cms-1.

Fouled membranes responded to changes in solution pH at a lower degree than virgin

membranes (Figure 5). In other words, the fouling layer seemed to reduce the impact of
solution pH on boron rejection. For example, when the solution pH increased from 6 to 11,
boron rejection by the virgin NF270 membrane increased 65%, whereas it was only
approximately 45% for the membranes fouled by sodium alginate and by colloidal silica
(Figure 5). In this case, the fouling layer acts as a physical barrier that shields the contact
between the bulk solution and membrane surface. Consequently, the zeta potential and
double-charged layer of the membrane would not be substantially affected by solution pH
changes, and lower boron rejection attained by fouled membranes as a result. Indeed, Tang et
al. [45] reported a constant zeta potential of the NF270 membrane fouled by Aldrich humic
acid when solution pH varied in the range of 3.5 – 9.5. In addition, the colloidal silica and
CaSO4 fouling layers could not only shield the contact between the membrane surface and the
bulk solution, but also neutralise the membrane surface charge which explains the lower
rejection of boron by membranes fouled by colloidal silica and CaSO4. The charged
neutralisation feature of these two foulants has been reported in the literature [20, 32]. On the
other hand, solution pH changes affected boron rejection more apparently at the fouled
NF270 than at the fouled BW30 membrane. Size exclusion was thought to be the dominant

15
rejection mechanism of boron by the BW30 membrane and this rejection mechanism was not
significantly affected by solution chemistry [42]. Indeed, the role of size exclusion and its
relationship with solution pH apparently revealed in the rejection of conductivity and sodium
(Figure 6). Conductivity and sodium rejection by the fouled NF270 membranes were almost
invariable with increasing solution pH, whereas their rejection by the virgin NF270
membrane significantly improved (Figure 6). Similar observations were reported by several
previous studies [45-47].

100 100

BW30 NF270
Conductivity rejection (%)

BW30 NF270 Clean Clean

80 Clean Clean 80
Humic acid Humic acid

Sodium rejection (%)

Humic acid Humic acid Na alginate Na alginate
Na alginate Na alginate Colloidal silica Colloidal silica
Colloidal silica Colloidal silica CaSO4 CaSO4
60 CaSO4 CaSO4 60

40 40

20 20

0 0
6 7 8 9 10 11 6 7 8 9 10 11
pH pH

Figure 6. The rejection of conductivity and sodium by virgin and fouled NF270 and BW30
membranes as a function of solution pH. Feed solution: 10 mM NaCl, 1 mM CaCl2, 1 mM
NaHCO3, 0.43 mM B(OH)3 and 20 mgL-1 of each foulant, except CaSO4 was 1 gL-1; feed
temperature = 20 oC; permeate flux = 40 Lm-2h-1; cross flow velocity = 30.4 cms-1.

Under high pH conditions, the difference in boron rejection by virgin and fouled membranes
was apparent. The boron rejection efficiency of virgin membranes was considerably higher
than that of the CaSO4, alginate and colloid-fouled membranes. One possible explanation for
the lower boron rejection observed with the membranes fouled by sodium alginate, colloidal
silica and CaSO4 under high pH conditions is the impact of cake-enhanced concentration
polarisation caused by fouling layers. Cake-enhanced concentration polarisation has been
extensively reported as a major cause of decrease in solute rejection by NF/RO membranes
[9, 20, 22-23]. In fact, colloidal fouling resulted in a severe drop in conductivity and sodium
rejection by the NF270 membrane (Figure 6), which is in good agreement with previous
studies [18, 20-22, 24, 26-28, 48]. The considerable increase in boron concentration at the

16
membrane surface coupled with the decline in permeate flux resulted in a significant decrease
in boron rejection by the fouled membranes as observed in Figure 5. It is noteworthy that
cake-enhanced concentration polarisation could occur and reduce boron rejection under either
low or high pH conditions. However, because the boron rejection efficiency of the virgin
NF270 membrane was negligible at low pH conditions, the effect of cake-enhanced
concentration polarisation on boron rejection was not apparent at this low pH condition. The
occurrence and influence of cake-enhanced concentration polarisation were further confirmed
by the lower sodium rejection by the colloidal silica fouled membrane observed in Figure 6.

The humic acid fouling layer could significantly increase boron rejection by both NF270 and
BW30 membrane at all pH values within the investigated pH range (Figure 5). This
phenomenon was more apparent at the NF270 membrane. At pH lower than 9, boron
rejections by the humic acid fouled NF270 and BW30 membranes were 30% and 15% higher
than for the virgin membranes, respectively. It was found that size exclusion is a dominant
removal mechanism for the neutral boric acid species [42]. The data reported in Figure 5
implied that the adsorption of humic acid on the membrane surface would enhance both size
exclusion and charge repulsion mechanisms. The enhancement of size exclusion might be
attributed to the phenomenon where the ‘hot spots’ on membrane surface with high local flux
and low salt rejection were plugged by humic material [27, 45], which resulted in the increase
in the steric-hindrance impact. For the BW30 membrane, the increase in size exclusion could
also be driven by the clogged valleys on the rough surface of this membrane. In addition, the
greatly negative charge of the humic layer resulted in a significant increase in boron rejection
by charge repulsion mechanism under high pH conditions. The effect of the humic acid
fouling layer to improve solute rejection was further confirmed by the higher conductivity
and sodium rejection obtained by the humic acid fouled NF270 membrane as illustrated in
Figure 6. This result is also consistent with several previous studies [7, 14, 49]. Cake-
enhanced concentration polarisation might occur and decrease boron rejection by the humic
acid fouled NF270 and BW30 membranes. However, the decrease in boron rejection caused
by this effect was probably compensated by the significant increase in boron rejection caused
by the ‘hot spots’ clogging and membrane surface charge increasing as discussed above [27,
45].

17
4. Conclusion
The impacts of four different model foulants on the performance of NF/RO membranes were
investigated in this study. Organic foulants (humic acid and sodium alginate) caused the most
severe drop in permeates flux and followed by colloidal silica and CaSO4 scaling. Various
extents of flux decline caused by different model foulants implied different fouling
mechanisms involved. All membrane fouling experiments investigated in this study,
including with different foulants and membranes used, appeared to be subjected to the cake-
enhanced concentration polarisation phenomenon which not only caused severe permeate
flux decline, but also decreased the rejection efficiency of boron and inorganic salts. In
addition, the fouling layer could also play the role of a physical barrier that inhibited the
impact of solution pH changes on membrane surface charged properties. Consequently, the
effect of high solution pH to increase membrane surface negative charge that encouraged
boron rejection was inhibited. In addition to cake-enhanced concentration polarisation, boron
rejection efficiency could also be reduced by the membrane charge neutralisation effect of the
fouling layer, particularly of colloidal silica and CaSO4 scaling layer. In contrast, boron
rejection could be improved by the adsorption of humic acid on the membrane surface
because of the highly negatively charge property of the humic substance.

5. Acknowledgements
Tony Nicholas, Laurent Devriendt and Helen Price are thanked for their technical advice
regarding ICP-MS analysis. Alex Simon is thanked for the measurement of the membrane
zeta potential. Dow FilmTec (Minneapolis, MN, USA) is gratefully acknowledged for the
donation of membrane samples to this research.

18
6. References
1. N. Hilal, G.J. Kim, and C. Somerfield, Boron removal from saline water: A
comprehensive review, Desalination 273 (2011) 23-35.
2. K.L. Tu, L.D. Nghiem, and A.R. Chivas, Boron removal by reverse osmosis
membranes in seawater desalination applications, Separation and Purification
Technology 75 (2010) 87-101.
3. E. Huertas, M. Herzberg, G. Oron, and M. Elimelech, Influence of biofouling on
boron removal by nanofiltration and reverse osmosis membranes, Journal of
Membrane Science 318 (2008) 264-270.
4. H.-J. Oh, T.-M. Hwang, and S. Lee, A simplified simulation model of RO systems for
seawater desalination, Desalination 238 (2009) 128-139.
5. M.F.A. Goosen, S.S. Sablani, H. Ai-Hinai, S. Ai-Obeidani, R. Al-Belushi, and D.
Jackson, Fouling of reverse osmosis and ultrafiltration membranes: A critical review,
Separation Science and Technology 39 (2004) 2261-2297.
6. Y. Kaiya, Y. Itoh, K. Fujita, and S. Takizawa, Study on fouling materials in the
membrane treatment process for potable water, Desalination 106 (1996) 71-77.
7. C.Y. Tang, Y.-N. Kwon, and J.O. Leckie, Fouling of reverse osmosis and
nanofiltration membranes by humic acid--Effects of solution composition and
hydrodynamic conditions, Journal of Membrane Science 290 (2007) 86-94.
8. L.D. Nghiem and P.J. Coleman, NF/RO filtration of the hydrophobic ionogenic
compound triclosan: Transport mechanisms and the influence of membrane fouling,
Separation and Purification Technology 62 (2008) 709-716.
9. Q. Li and M. Elimelech, Synergistic effects in combined fouling of a loose
nanofiltration membrane by colloidal materials and natural organic matter, Journal of
Membrane Science 278 (2006) 72-82.
10. S. Hong and M. Elimelech, Chemical and physical aspects of natural organic matter
(NOM) fouling of nanofiltration membranes, Journal of Membrane Science 132
(1997) 159-181.
11. J.A. Nilson and F.A. DiGiano, Influence of NOM composition on nanofiltration,
Journal / American Water Works Association 88 (1996) 53-66.
12. Q. Li, Z. Xu, and I. Pinnau, Fouling of reverse osmosis membranes by biopolymers in
wastewater secondary effluent: Role of membrane surface properties and initial
permeate flux, Journal of Membrane Science 290 (2007) 173-181.
13. L.D. Nghiem and S. Hawkes, Effects of membrane fouling on the nanofiltration of
trace organic contaminants, Desalination 236 (2009) 273-281.
14. A.E. Childress and M. Elimelech, Effect of solution chemistry on the surface charge
of polymeric reverse osmosis and nanofiltration membranes, Journal of Membrane
Science 119 (1996) 253-268.
15. S.-H. Yoon, C.-H. Lee, K.-J. Kim, and A.G. Fane, Effect of calcium ion on the
fouling of nanofilter by humic acid in drinking water production, Water Research 32
(1998) 2180-2186.
19
16. A.I. Schäfer, A.G. Fane, and T.D. Waite, Nanofiltration of natural organic matter:
Removal, fouling and the influence of multivalent ions, Desalination 118 (1998) 109-
122.
17. C. Jarusutthirak, S. Mattaraj, and R. Jiraratananon, Influence of inorganic scalants and
natural organic matter on nanofiltration membrane fouling, Journal of Membrane
Science 287 (2007) 138-145.
18. H.Y. Ng and M. Elimelech, Influence of colloidal fouling on rejection of trace organic
contaminants by reverse osmosis, Journal of Membrane Science 244 (2004) 215-226.
19. P. Xu, J.E. Drewes, T.-U. Kim, C. Bellona, and G. Amy, Effect of membrane fouling
on transport of organic contaminants in NF/RO membrane applications, Journal of
Membrane Science 279 (2006) 165-175.
20. S. Lee, J. Cho, and M. Elimelech, Influence of colloidal fouling and feed water
recovery on salt rejection of RO and NF membranes, Desalination 160 (2004) 1-12.
21. R.D. Cohen and R.F. Probstein, Colloidal fouling of reverse osmosis membranes,
Journal of Colloid and Interface Science 114 (1986) 194-207.
22. E.M.V. Hoek, A.S. Kim, and M. Elimelech, Influence of crossflow membrane filter
geometry and shear rate on colloidal fouling in reverse osmosis and nanofiltration
separations, Environmental Engineering Science 19 (2002) 357-372.
23. E.M.V. Hoek and M. Elimelech, Cake-Enhanced Concentration Polarization: A New
Fouling Mechanism for Salt-Rejecting Membranes, Environmental Science and
Technology 37 (2003) 5581-5588.
24. S. Lee, J. Cho, and M. Elimelech, Combined influence of natural organic matter
(NOM) and colloidal particles on nanofiltration membrane fouling, Journal of
Membrane Science 262 (2005) 27-41.
25. J.A. Brant and A.E. Childress, Membrane-colloid interactions: Comparison of
extended DLVO predictions with AFM force measurements, Environmental
Engineering Science 19 (2002) 413-427.
26. X. Zhu and M. Elimelech, Colloidal fouling of reverse osmosis membranes:
Measurements and fouling mechanisms, Environmental Science and Technology 31
(1997) 3654-3662.
27. E.M. Vrijenhoek, S. Hong, and M. Elimelech, Influence of membrane surface
properties on initial rate of colloidal fouling of reverse osmosis and nanofiltration
membranes, Journal of Membrane Science 188 (2001) 115-128.
28. K. Boussu, A. Belpaire, A. Volodin, C. Van Haesendonck, P. Van der Meeren, C.
Vandecasteele, and B. Van der Bruggen, Influence of membrane and colloid
characteristics on fouling of nanofiltration membranes, Journal of Membrane Science
289 (2007) 220-230.
29. S. Lee, J. Kim, and C.-H. Lee, Analysis of CaSO4 scale formation mechanism in
various nanofiltration modules, Journal of Membrane Science 163 (1999) 63-74.
30. J. Gilron and D. Hasson, Calcium sulphate fouling of reverse osmosis membranes:
Flux decline mechanism, Chemical Engineering Science 42 (1987) 2351-2360.

20
31. A.G. Pervov, Scale formation prognosis and cleaning procedure schedules in reverse
osmosis systems operation, Desalination 83 (1991) 77-118.
32. C. Bellona and J.E. Drewes, The role of membrane surface charge and solute physico-
chemical properties in the rejection of organic acids by NF membranes, Journal of
Membrane Science 249 (2005) 227-234.
33. P. Argust, Distribution of boron in the environment, Biological Trace Element
Research 66 (1998) 131-143.
34. F. Fairbrother and H. Mastin, Studies in electro-endosmosis, Journal of the Chemical
Society 125 (1924) 2319-2330.
35. L.D. Nghiem, A.I. Schäfer, and M. Elimelech, Removal of Natural Hormones by
Nanofiltration Membranes: Measurement, Modeling and Mechanisms, Environmental
Science and Technology 38 (2004) 1888-1896.
36. J.R. McCutcheon and M. Elimelech, Influence of membrane support layer
hydrophobicity on water flux in osmotically driven membrane processes, Journal of
Membrane Science 318 (2008) 458-466.
37. A.I. Schäfer, A. Pihlajamäki, A.G. Fane, T.D. Waite, and M. Nyström, Natural
organic matter removal by nanofiltration: effects of solution chemistry on retention of
low molar mass acids versus bulk organic matter, Journal of Membrane Science 242
(2004) 73-85.
38. S. Lee and M. Elimelech, Relating organic fouling of reverse osmosis membranes to
intermolecular adhesion forces, Environmental Science and Technology 40 (2006)
980-987.
39. M. Beyer, B. Lohrengel, and L.D. Nghiem, Membrane fouling and chemical cleaning
in water recycling applications, Desalination 250 (2010) 977-981.
40. W.S. Ang, S. Lee, and M. Elimelech, Chemical and physical aspects of cleaning of
organic-fouled reverse osmosis membranes, Journal of Membrane Science 272 (2006)
198-210.
41. C. Park, Y.H. Lee, S. Lee, and S. Hong, Effect of cake layer structure on colloidal
fouling in reverse osmosis membranes, Desalination 220 (2008) 335-344.
42. K.L. Tu, L.D. Nghiem, and A.R. Chivas, Coupling effects of feed solution pH and
ionic strength on the rejection of boron by NF/RO membranes, Chemical Engineering
Journal 168 (2011) 700-706.
43. M.H. Oo and L. Song, Effect of pH and ionic strength on boron removal by RO
membranes, Desalination 246 (2009) 605-612.
44. A. Sagiv and R. Semiat, Analysis of parameters affecting boron permeation through
reverse osmosis membranes, Journal of Membrane Science 243 (2004) 79-87.
45. C.Y. Tang, Y.-N. Kwon, and J.O. Leckie, Characterization of Humic Acid Fouled
Reverse Osmosis and Nanofiltration Membranes by Transmission Electron
Microscopy and Streaming Potential Measurements, Environmental Science &
Technology 41 (2006) 942-949.

21
46. L.D. Nghiem and S. Hawkes, Effects of membrane fouling on the nanofiltration of
pharmaceutically active compounds (PhACs), Separation and Purification Technology
57 (2007) 176-184.
47. A.R.D. Verliefde, E.R. Cornelissen, S.G.J. Heijman, I. Petrinic, T. Luxbacher, G.L.
Amy, B. Van der Bruggen, and J.C. van Dijk, Influence of membrane fouling by
(pretreated) surface water on rejection of pharmaceutically active compounds
(PhACs) by nanofiltration membranes, Journal of Membrane Science 330 (2009) 90-
103.
48. S.G. Yiantsios and A.J. Karabelas, The effect of colloid stability on membrane
fouling, Desalination 118 (1998) 143-152.
49. J. Cho, G. Amy, and J. Pellegrino, Membrane filtration of natural organic matter:
comparison of flux decline, NOM rejection, and foulants during filtration with three
UF membranes, Desalination 127 (2000) 283-298.

22