Journal of Membrane Science 460 (2014) 99–109

Contents lists available at ScienceDirect

Journal of Membrane Science

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

Organic fouling in forward osmosis membranes: The role of feed

solution chemistry and membrane structural properties
Machawe M. Motsa a,c,n, Bhekie B. Mamba a, Arnout D’Haese c,
Eric M.V. Hoek a,b, Arne R.D. Verliefde c,d
a
Department of Applied Chemistry, University of Johannesburg, P.O. Box 17011, Doornfontein 2028, South Africa
b
Department of Civil and Environmental Engineering, California NanoSystems Institute, University of California, 5732-G Boelter Hall, PO Box 951593,
Los Angeles, CA 90095-1593, USA
c
Department of Applied Analytical and Physical Chemistry, University of Gent, Coupure links 653, B-9000 Ghent, Belgium
d
Delft University of Technology, Faculty of Civil Engineering and Geosciences, Stevinweg 1, 2628CN Delft, The Netherlands

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

Article history: This study was aimed at investigating the effect solution total ionic strength, divalent ion concentration
Received 20 September 2013 (Ca2 þ and Mg2 þ ) and membrane structural properties on the fouling propensity of alginate. The fouling
Received in revised form reversibility of the resulting fouling layer was also determined through the use of ultrapure water and
3 February 2014
salt solutions as cleaning agents. Sodium alginate was used as a model foulant to represent extracellular
Accepted 24 February 2014
Available online 4 March 2014
polymeric substance (EPS) that are present in seawater and wastewater. In order to understand the
influence of all possible interactions the membrane was tested using solutions of two ionic strengths
Keywords: (0.1 M and 0.5 M) with varying concentrations of calcium and magnesium ions. Experimental results
Forward osmosis suggested that membrane orientation had an impact on fouling behavior since the membrane fouled
Fouling
more easy when operated in PRO mode than in FO mode. There was severe permeate flux decline in PRO
Ionic strength
mode mainly due to the calcium–alginate complexes blocking the pores in the support layer. The
Interfacial energy
Pressure retarded osmosis interfacial free energies obtained from advanced contact angle measurements correlated strongly with
Sodium alginate the rates of membrane fouling and further predicted the membrane fouling trends. Initial adhesion of
alginate particles on the membrane surface was dictated by the chemical interactions between the
membrane and alginate. It was also found that once the membrane surface is covered with the foulant,
fouling becomes less sensitive to the changes in hydrodynamic conditions (permeation drag and cross-
flow velocity) but rather depends on the foulant–foulant interactions. The resulting fouling layer was
easily removed by shear force resulting from increased cross-flow velocity during membrane cleaning.
& 2014 Elsevier B.V. All rights reserved.

1. Introduction
High pressure membrane filtration processes such as nanofiltration
(NF) and reverse osmosis (RO) are presently the most widely applied
processes for seawater desalination and wastewater reclamation to
augment potable water supplies. These processes have proved to be
capable of rejecting most dissolved constituents, including trace
organic chemicals [1,2]. Provided that adequate pre-treatment is
implemented, these processes can produce high quality potable water
from reclaimed water, seawater, or from surface or ground water
impacted by discharge from wastewater treatment plants [3]. How-
ever, since they require high hydraulic pressure to drive water across
n
Corresponding author at: Department of Applied Analytical and Physical
Chemistry, University of Gent, Coupure links 653, B-9000 Ghent, Belgium.
Tel.: þ 32 9 264 6001; fax: þ 32 9 264 6242.
E-mail addresses: mxolisimachawe.motsa@UGent.be,
machawemmvulane@gmail.com (M.M. Motsa).
the membrane, they are regarded as energy intensive and costly,
especially with the current increasing energy prices. In addition to the
high energy demand, the efficiency of NF and RO membranes is also
thwarted by the inherent problem of membrane fouling that usually
requires intensive chemical cleanings which can further escalate the
treatment costs and reduce membrane lifetime [4–6].
Therefore it became imperative to investigate other possible
alternative technologies for seawater desalination and wastewater
reclamation, amongst which forward osmosis (FO) membrane filtra-
tion has become one of the most promising in the past years. Indeed,
forward osmosis has over the past years received worldwide attention
as promising alternative technology due to its claimed high resistance
to fouling and wide range of pollutant rejection [7–10]. The distin-
guishing feature for FO when compared to RO is that FO utilizes
the natural osmotic pressure gradient generated by a concentrated
draw solution as the driving force for the net movement of water
molecules through a semi-permeable membrane. The concentrated
draw solution is diluted by water permeating from the feed water to

http://dx.doi.org/10.1016/j.memsci.2014.02.035
0376-7388 & 2014 Elsevier B.V. All rights reserved.
100 M.M. Motsa et al. / Journal of Membrane Science 460 (2014) 99–109
desalination, and the resulting diluted draw solution can then be re-
concentrated to recycle the draw solutes as well as to produce purified
water. When compared to pressure-driven membrane processes, FO is
claimed to possess a number of potential advantages including high
rejection of a wide range of contaminants, low energy consumption
and brine discharge, simple configurations and equipment as well as
low membrane fouling propensity [11–13]. Nevertheless, similar to
other membrane processes, the efficiency of FO processes is still
thwarted by the enduring problem of membrane fouling and despite
the numerous publications on the topic, there is still a lack of full
systematic and mechanistic understanding of the fouling behavior for
FO processes.
Membrane fouling in FO membrane filtration processes is hugely
impacted by various factors ranging from hydrodynamic operating
conditions to physical and chemical interactions between foulants and
the membrane. Previous studies have confirmed lower fouling pro-
pensity for FO membranes compared to RO, mainly due to the lack of
hydraulic pressure in driving water across the semi-permeable mem-
brane [14–16]. Mi and Elimelech [17] investigated the role of calcium
binding, initial permeate flux and membrane orientation in organic
fouling of forward osmosis membranes using alginate, bovine serum
albumin (BSA) and humic acid (HA). It was reported that calcium bin-
ding, permeation drag and hydrodynamic shear force were the major
factors governing the development of a fouling layer on the mem-
brane surface. They also performed subsequent membrane cleaning to
determine the fouling reversibility behavior, whereby the results illus-
trated that alginate fouling in FO membrane processes is almost fully
reversible by simple physical cleaning for a relatively short time [18].
Hong et al. [19] reported that the key mechanism of flux decline in
FO is rather accelerated cake-enhanced osmotic pressure (CEOP) due
to reverse salt diffusion from draw solution to the feed than an
increase in fouling layer resistance, after conducting fouling experi-
ments with alginate, humic acid and BSA, as well as silica colloids
(SiO2). A more recent study by Yong Ng and Parid [20], focused on the
impact of lower organic loads (10, 30, 50 ppm) in secondary effluents
with calcium inclusion on the fouling characteristics of FO membranes
both in the FO and PRO modes. In their work, they demonstrated that
the FO mode (active layer facing feed water) had lower fouling
compared to the PRO mode (active layer facing draw solution), which
was also seen by other authors [17–22]. This was attributed to the
denser, smoother and tighter structure of the membrane active layer
which prevented the adhesion and accumulation of foulants on the
membrane surface, while the porous support layer, being a looser
structure, allowed the accumulation and deposition of the foulants on
its surface and inside the membrane, by the mechanisms of direct
interception and subsequent pore plugging. It was also found that the
presence of 5 mM of Ca2þ in the feed solution containing NOM, humic
acid and biopolymers caused severe membrane fouling in the PRO
configuration (about 85% water flux reduction in 20 h), due to the
binding of calcium to carboxylic groups in natural organic matter
result in large NOM clusters that settled easier on the loose sup-
port layer.
The internal concentration polarization (ICP) phenomenon has also
been reported in several studies as another contributing factor to flux
decline in FO membrane processes due to the reduction of the osmotic
drive force. Tang et al. [21] investigated the coupled effect of organic
fouling and internal concentration polarization (ICP) on FO flux
behavior. They reported that ICP played a dominant role on FO flux
behavior at higher draw solution concentrations and/or greater mem-
brane fluxes due to their exponential dependence of ICP on flux levels.
They observed that the active layer-facing draw solution (PRO)
configuration was highly prone to flux reduction due to combined
effects of internal clogging of the support layer and the resulting
enhanced ICP in the support layer; the latter being caused by reduced
porosity and reduced mass transfer coefficient of the support layer.
Similar observations were made by Zhang et al. [22], who studied the
complex fouling phenomenon of FO membranes in a FO membrane
bioreactor, where they discovered that a biofouling layer covered the
substrate surface of the FO membrane with a combined structure of
bacterial clusters and extracellular polymeric substances (EPS), which
contributed to a massive drop in membrane support layer mass trans-
fer coefficient.
In addition to solution chemistry and hydrodynamic conditions,
membrane surface properties also greatly influence flux decline in
membrane based filtration systems, since they impact membrane
surface related processes such as wetting, adhesion and adsorption
[23]. Foulant rejection potential is an another process that can be
influenced by the membrane surface due to its affinity towards cer-
tain foulants. The magnitude of the chemical interactions between
the membrane surface and foulant can be better predicted from
determining the membrane surface free energy, which is generally
explained by the extended Derjaguin–Landau–Verwery–Overbeek
(XDLVO) theory. The XDLVO describes surface free energy as the
sum of the apolar Lifshitz–van der Waals (LW) component and the
polar or Lewis acid–base (AB) component [24]. Traditionally, this
theory has been used to characterize membrane–colloid/aggregate
interactions. Childress et al. [25] evaluated the surface energetics of
three RO membranes and silica colloids using the XDLVO approach.
The membranes were found to have low surface energies compared
to the colloid and the interaction energy between the membranes
and colloids was primarily dictated by the surface energies of the
colloids.
In another work Lee and co-workers [26] used the XDLVO
approach to describe NOM fouling in terms of interaction forces
between NOM macromolecules and a membrane surface. They
reported that acid–base (AB) interaction forces were dominant at
short distance (o5 nm) and determined the short-range foulant–
membrane interactions. The obtained negative cohesion energy
values predicted strong attractive hydrophobic interactions amongst
the NOM macromolecules. While the similar negative adhesion
energy values predicted strong attractive interactions between the
NOM macromolecules and the membrane surface.
Jin et al. [27] examined the impacts of major ions in seawater on
the acid–base properties of seawater RO membranes and alginic acid.
They also evaluated correlations between (alginate–membrane and
alginate–alginate) acid–base free energies and fouling. It was reve-
aled that in membrane processes, initial organic adhesion is
mediated by interfacial interactions between the organic molecules
and clean membrane surface. After the membrane surface is covered
by a layer of absorbed organics, subsequent organic adhesion is
controlled by interfacial interactions between newly deposited and
already deposited organics.
These studies have given more insight onto the cryptic mechan-
isms governing RO and FO membrane fouling by various foulants.
However, since forward osmosis processes are aimed at purifying
impaired water sources such as seawater more work needs to be
done in elucidating the influence of the presence of cationic species
on the foulant–membrane chemical interactions. It has been hypo-
thesized that the acid–base properties of membranes and alginate
are a function of the ionic composition of the aqueous media they
occur in [27], therefore, developing a systematic understanding of FO
membrane fouling behavior and the relation with foulant–mem-
brane interactions in different ionic environments is of paramount
importance.
The main focus of this work was at establishing a complete
systematic understanding of the underlying fouling mechanisms of
FO membranes by organic foulants at ionic concentrations as high as
observed in seawater desalination processes. This was achieved by
exposing the FO membrane to feed solutions of varying ionic
concentrations of NaCl, CaCl2 and MgCl2. Alginic acid (sodium algi-
nate) was used as a model foulant to represent extracellular poly-
meric substance (EPS) found in both seawater and wastewater. The
 M.M. Motsa et al. / Journal of Membrane Science 460 (2014) 99–109
influence of the structural membrane characteristics on fouling
behavior was also studied by running tests in both FO and PRO
modes. To determine the impact of solution chemistry and mem-
brane surface properties on fouling, and relate fouling to foulant–
membrane interactions, membrane samples and alginate were
characterized through advanced contact angle measurements. The
performance of non-chemical intensive solutions (ultrapure water
and salt solution) in reversing the fouling on the membrane surface
was also examined.
2. Material and methods
2.1. Membrane
The membrane used in this study was supplied by Hydration
Technologies (Albany, OR). It was a flat sheet FO membrane,
asymmetrical in structure and composed of a cellulose triacetate
(CTA) active layer embedded in polyester mesh to render mechanical
support. The CTA membrane has been extensively used in the pre-
vious research and has been found to have a total thickness of
approximately 50 mm [28]. It was previously tested and recorded a
salt rejection of above 95%, a pure water permeability (A) of
0.4470.12 L m  2 h  1 bar  1 and a NaCl permeability coefficient (B)
of 0.26170.0261 L m  2 h  1 [29]. The membrane structural para-
meter (S) is a factor that describes how the membrane structure
change with different conditions and is around 481 mm. The mem-
brane was stored in ultrapure water at 4 1C prior to use.
2.2. Organic foulant
Alginic acid (provided as sodium alginate) was used as a model
organic foulant to represent common polysaccharides which are the
main constituents of organic matter in wastewater and seawater. This
model foulant was provided by Sigma-Aldrich, St Louis, MO and was
received in powder form and was used as received. The molecular
weight of the alginate is between 12 and 80 kDa. Stock solutions of
2 g/l were prepared by dissolving alginate powder in deionised (DI)
water by mixing vigorously for 24 h and then kept at 4 1C.
2.3. Feed and draw solution chemistries
Sodium alginate is a representative of polysaccharides in waste-
water and seawater has been previously identified as a major organic
foulant [16]. It is also known to form highly arranged complexes in
the presence of divalent cations resulting in severe membrane flux
decline. It is therefore for these reasons that in this work it was
exposed to varying concentrations of cationic species (Na þ , Ca2 þ ,
and Mg2 þ ) to prepare the different feed water types presented in
Table 1. The foulant concentration was fixed at 200 mg/l. No further
pH adjustments were conducted; the solutions were tested at their
ambient pH (6.9070.45). No large pH variations were observed
during experiments. The two ionic strengths were prepared to mimic
brackish and seawater desalination processes. The total ionic
strengths for feed solutions were calculated based on Eq. (1):
1 the process loosening the fouled layer. The cross-flow velocity was
I¼ ∑C i z2i
2 maintained at the same rate on both sides of the membrane.
where Ci is the molar concentration of ion i (M mol/L) in the solution
and zi is the charge number of that ion. Sodium chloride (NaCl) was
used to prepare draw solutions of different concentrations depend-
ing on the feed ionic strength and the experiment mode (i.e. FO or
PRO). Draw solution ionic strength was adapted to obtain similar
starting fluxes in all experiments (Table 1).
101
2.4. Forward osmosis membrane test unit and fouling protocol
The FO membrane fouling tests were performed using a bench-
scale cross-flow system. It consists of two closed loops for the feed
and draw solution streams. The solutions were pumped through a
cross-flow membrane cell and around the loops using variable speed
pumps (Cole-Palmer, USA). The cross-flow membrane cell was cus-
tom built with equally structured channels on both sides of the
membrane. Each channel had the dimensions of 250, 50 and 1 mm
for length, width and depth respectively. A polypropylene spacer
mesh was added on both sides of the membrane to create turbulence
and mimic real membrane filtration processes. The change in feed
solution weight was monitored over time through a weighing
balance (Ohaus, USA) connected to a computer for data logging.
The changes in feed weight overtime were used to compute the
water flux during fouling. The assembled laboratory FO unit used in
this work is depicted in Fig. 1. During filtration the draw solution gets
diluted and gradually decreases in concentration which turns to
reduce the osmotic drive force. To prevent this from occurring, the
concentration of the draw solution was maintained at a constant
value using a real time conductivity based program using a Consort
conductivity meter (Turnhout, Belgium). Varying amounts of a highly
concentrated salt solution (5 M) were dosed into the draw solution
triggered by a decline in conductivity.
Prior to each set (i.e. I¼0.1 M and 0.5 M) of fouling experiments a
baselines experiment was performed; a solution containing only
the background electrolytes (without the foulant) was filtered for the
same duration as the fouling experiments. The presented flux decline
curves were corrected using the baseline to isolate the effect of
fouling. Fouling experiments were performed for a duration of 10 h
and a new membrane coupon was used for each experiments. The
cross-flow velocity was fixed at 10 cm/s (400 ml/min) and initial
permeate flux was 16.8470.75 L m  2 h  1 for all conducted experi-
ments. Membrane fouling behavior was tested on both sides of the
asymmetrical CTA FO membrane (i.e. in PRO and FO configuration).
2.5. FO membrane cleaning
Membrane cleaning was done immediately after fouling runs were
stopped. The feed solution with the most fouling potential (solution 6)
was then used to determine the reversibility of the fouling layer. The
fouled membrane was cleaned using three conditions: (1) osmotic
backwash with ultrapure water, (2) cleaning by cross-flow flushing
with salt solution of NaCl and CaCl2 and (3) cleaning by cross-flow
flushing with ultrapure water. Cleaning was done for 15 min and at a
fixed cross-flow velocity of 20 cm/s. Membrane cleaning was per-
formed through these conditions to determine whether the lack of
hydraulic pressure during fouling in osmotic driven processes and the
use of non-chemical intensive methods during membrane cleaning
influences the reversibility of fouling on the membrane surface.
2.5.1. Osmotic back wash
During osmotic back wash, the concentrated draw solution is
replaced with ultrapure water such that there is a net movement of
water into the concentrated feed solution. Therefore, the then feed
solution acts as a draw solution and pure water permeates from the
back side of the membrane towards the fouled membrane surface, in
ð1Þ
2.5.2. Cleaning with salt solution and ultrapure water
To determine the role of cleaning solution chemistry on fouling
reversibility and membrane surface charge was determined by clean-
ing the membrane with a salt solution and ultrapure water. The salt
solution, composed of NaCl and CaCl2 with a total ionic strength of
102 M.M. Motsa et al. / Journal of Membrane Science 460 (2014) 99–109

Table 1
A list of the different feed water types, their composition and draw solutions.

Feed solutions (FS) FS composition Total ionic strength (mM) Draw solutions (M) NaCl

FO PRO

Solution 1 Alginateþ 100 mM NaCl 100 3.1 1.674

Solution 2 Alginateþ 94 mM NaClþ 2 mM CaCl2
Solution 3 Alginateþ 70 mM NaClþ 10 mM MgCl2
Solution 4 Alginateþ 64 mM NaClþ10 mM MgCl2 þ 2 mM CaCl2
Solution 5 Alginateþ 500 mM NaCl 500 3.5 2.1
Solution 6 Alginateþ 476 mM NaClþ8 mM CaCl2
Solution 7 Alginateþ 350 mM NaClþ 50 mM MgCl2
Solution 8 Alginateþ 326 mM NaClþ 50 mM MgCl2 þ 8 mM CaCl2

Fig. 1. Schematic diagram of the bench-scale forward osmosis (FO) experimental unit.

500 mM, was used for cleaning a fouled FO membrane by cross-flow
flushing at double the filtration cross-flow. Both the feed and draw
solutions were replaced with the salt solution. During this process
there was limited water permeation and salt diffusion on both sides of
the membrane. Ultrapure water was used in a similar procedure as the
salt solution: both feed and draw solution were replaced with Milli-Q
water. Therefore, there was also limited water permeation and the salt
accumulated on the membrane interface was washed away in the
process.
2.6. Foulant characterization
2.6.1. Particle sizing
The mean hydrodynamic diameter of the alginate macromole-
cules in the different feed solutions was determined by dynamic light
scattering (DLS) using a Malvern photon correlation spectrometer
(Malvern Instruments, UK). The concentration of alginate (0.5 g/l)
used in the size determination was higher than that used in the
fouling experiments to enhance sample detection, since 200 mg/l
was too dilute. It is worth mentioning that this concentration can
contribute/influence the observed changes in alginate particle size
due to multiple lights scattering resulting in over-estimated particle
size. Therefore, to determine this effect the alginate particle size was
measured at 0.3 g/l, 0.75 g/l and 1 g/l and the resulting values were
not significantly different from each other and the chosen 0.5 g/l that
was used for this work. Thus we assumed that the size and zeta
potential trends observed could be extrapolated down to 0.2 g/l.
2.6.2. Zeta potential measurements
The changes on the alginate aggregate surface charge when
exposed to the different electrolyte solutions were determined
from electrophoretic mobility, measured using a Malvern Zetasi-
zer300 HS series (Malvern Instruments, UK). The concentration of
alginate was increased to 0.5 g/l during this analysis to enhance
instrument detection.
2.7. Membrane characterization
2.7.1. Membrane zeta potential measurements
To determine the influence of different specific cationic species on
the membrane surface properties, streaming potential was measured
using the different salt that make-up the different feed solutions. In
the experimental set-up, the electrolyte solution was pumped through
the cell which consists of two membranes with a spacer in between.
In this way, only the charge characteristics on the exterior surface of
the membrane are determined. The membrane samples were taped to
glass slides. The applied pressure difference was 0.2 bar. The cell
potential was measured continuously by two Pt electrodes. When the
valve at the inlet of the cell was closed, the solution stopped flowing
through the cell (non-flow), the flow started again when the valve
 M.M. Motsa et al. / Journal of Membrane Science 460 (2014) 99–109
opened (flow mode). The difference in potential flow and non-flow
mode is equal to the streaming potential. This potential difference was
measured more than 5 times and a mean value was calculated. The
relationship between the measurable streaming potential and the zeta
potential is given by the Helmholtz–Smoluchowski equation (Eq. 2)
ΔV:η:s
ζ¼ ð2Þ
ΔP:ε:ε0
where ΔV is the measured streaming potential, η is the electrolyte
viscosity (Pa.s), electrolyte0 s electrical conductivity (s/m), ΔP is the
applied pressure and ε is the permittivity of water (C2 N  1 m  2).
The permittivity is defined as ε¼ε0.D, where ε0 is the permittivity of
vacuum¼8.85  10–12 (C2 N  1 m  2) and D is the dielectric constant of
water¼78.55 at 25 1C.
2.8. Alginate and membrane surface tension measurements
Contact angle (θ) measurements are generally used to determine
the surface free energy through the Young–Dupre equation (Eq. (3))
which links the contact angle of a drop of a liquid placed on a flat
solid surface with the surface tension of a liquid, according to van Oss
(2007) [30].
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffi
γ l ð1 þ cos θÞ ¼ 2ð γ LWs γl
LW þ γ sþ γ l þ γ s γ lþ Þ ð3Þ
where θ is the measured contact angle; γLW is the Lifshitz–van der
Waals free energy component; γ þ is the electron-acceptor compo-
nent; γ  is the electron donor component; and the subscripts s and l
designate the solid and liquid phases, respectively. Surface tension
components of a membrane can be determined by measuring
contact angles between the membrane and three well-charac-
terized probe liquids with known surface tension components, such
as deionised water, glycerol and diiodomethane. The contact angles
of these probe liquids were measured using a computerized Krüss
DSA 10-MK2 (Germany) contact angle goniometer.
The surface tensions of alginate due to the different electrolyte
solutions were determined by first dissolving it in ultrapure water
and deposited onto a NF membrane using a dead-end membrane
filtration system for a period of 8 h a method which was also imple-
mented by Jin et al. [27]. The filtered lawns of alginate were then
allowed to dry for 12 h in a dessicator. Contact angle measurements
of the different salt solutions, deionised water and diiodomethane
were then performed using the sessile drop method.
However when determining the clean membrane surface ten-
sions two methods were employed: (1) sessile drop on a dry mem-
brane surface and (2) sessile drop measurement on a hydrated
membrane surface as previously conducted by Botton and workers,
[31] on their work with NF membranes.
After drying the membrane samples in a sealed desiccator for 12 h,
they were then attached to a glass slide and a minimum of 10 drops
were measured per membrane sample for all three probe liquids.
The average contact angle was used for computation of surface
tensions and interfacial free energies. The hydrated method aimed
at delaying water evaporation during contact angle measurement as
well as simulating the liquid–membrane contact during the filtration
process. Membrane samples were placed on a wet filter paper to curb
the effect of water evaporation. Sessile drops of the three probe liquids
were placed on the hydrated membrane surface with the contact
angle measured within seconds. Drops were placed at intervals of
3–5 min for a period of about 1.5 h. For each membrane sample a
minimum of 10 drops were measured. Measuring the liquid drop
contact angle on the dry CTA membrane gave an over-estimation of
the contact angles since the contact angle is quantified while the
membrane is still wetting leading to higher values that are not
accurate representatives of the membrane surface tensions. Therefore
103
only the sessile drops on the hydrated membrane surface were used
for the computation of membrane surface tensions.
The total interfacial free energies (interfacial free energy of cohe-
sion and adhesion) per unit area between solid materials 1 and
2 immersed in liquid medium 3 were determined from the sum of the
Lifshitz–van der Waals (LW) and acid–base (AB) components as
expressed by Eq. (4).
ΔGTOT
132 ¼ ΔG132 þ ΔG132
LW AB
ð4Þ
qffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffi
ΔGLW
132 ¼ 2ð γ 3 
LW γ LW
1 Þð γ 2 
LW γ LW
3 Þ ð5Þ
qffiffiffiffiffiffiffi pffiffiffiffiffiffiffi pffiffiffiffiffiffiffi pffiffiffiffiffiffiffi pffiffiffiffiffiffiffi qffiffiffiffiffiffiffi
ΔGAB
132 ¼ 2 γ 3 ð γ 1 þ
þ  γ 2  γ 3 Þ þ 2 γ 3 ð γ 1þ
qffiffiffiffiffiffiffi qffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffi
þ γ 2þ  γ 3þ Þ  2 γ 1þ γ 2  2 γ 1 γ 2þ ð6Þ
If surfaces 1 and 2 are the same, then ΔGTOT131 indicates the interfacial
free energy of cohesion. The interfacial free energy of cohesion,
ΔG131, describes energetic favorability of a solid material (1) interact-
ing through a liquid medium (3) with itself (1). Cohesive free energy
offers insight into particles stability as well as particle deposition
onto surfaces already covered by the same particles. The interfacial
free energy of adhesion, ΔG132, describes the attraction or repulsion
of solid material (1) interacting with another solid material (2)
through a liquid medium (3). The adhesive free energy relates to
adsorption and adhesion of dissimilar materials.
3. Results and discussion
3.1. Membrane and foulant zeta potential determination
The electrochemical properties of the membrane, especially the
selective layer, exert a great influence on the type and extent of
interactions between the membrane and the foulants in the liquid
feed as well as the separation properties such as permeability and
permselectivity [32]. The surface charges of both the membrane
and alginate macromolecules when exposed to different ionic solu-
tions are presented in Table 2. The CTA membrane was found to be
weakly negatively charged in the presence of 0.1 M NaCl background
electrolyte and in the pH range of 6–8. However, the inclusion Ca2þ
and Mg2 þ slightly reduced the membrane charge. The decrease in
membrane charge can be attributed to the compression of the electric
double layer thickness which arises from short distance screening of
surface charge resulting in charge neutralization. When the total ionic
composition was increased to 0.5 M the calculated membrane zeta
potential was found to be positive, an observation that is likely to be
accounted for by the high concentration of cations which tend to
adsorb on the membrane surface influencing the measured streaming
potential. This is more evident on elevated presence of divalent cations
such as calcium ions which have the capability to attach to the mem-
brane's carboxylate functionality surface making it more conductive.
The alginate macromolecules remained negatively charged the entire
range of electrolyte solutions. However, the presence of cationic spe-
cies at increased concentrations greatly reduced the negative charge.
This can be explained by the same process of reduced electric double
layer thickness leading to reduced electrostatic repulsion among
the alginate macromolecules prompting their easy deposition on the
membrane surface. The obtained zeta potential data suggest that it
can be tentatively concluded that more membrane fouling can be
expected at higher ionic strength due to attractive electrostatic forces
between oppositely charged phases (positive membrane and negative
alginate macromolecules). However, there are other factors that
further influence the interaction between the membrane and the
foulant, as will be discussed in paragraphs to follow.
104 M.M. Motsa et al. / Journal of Membrane Science 460 (2014) 99–109

Table 2
Alginate and membrane zeta potentials determined with the different feed salt solutions.

Test solutions Alginate zeta potential (mV) Membrane zeta potential (mV)
Membrane orientation

Deionised water  68.70 72.15 FO (active layer) PRO (support layer)

100 mM NaCl  29.707 2.05  3.80  10.78 7 0.67
94 mM NaClþ2 mM CaCl2  25.2 70.95  1.70  9.717 0.75
70 mM NaClþ 10 mM MgCl2  20.80 70.20  4.137 0.22  9.39 7 0.58
64 mM NaClþ10 mM MgCl2 þ 2 mM CaCl2  23.80 71.15  2.067 0.41  5.42 7 0.59
500 mM NaCl  14.7 7 1.35 18.02 7 0.99 28.377 1.99
476 mM NaClþ8 mM CaCl2  11.2 71.37 6.687 0.64 19.83 7 1.22
350 mM NaClþ 50 mM MgCl2  10.677 1.10 16.75 7 1.52 25.317 1.75
326 mM NaClþ50 mM MgCl2 þ 8 mM CaCl2  9.7 70.97 5.58 7 0.74 17.38 7 0.86

3.2. Effect of mono- and divalent cations on bulk alginate Table 3

complexation and its fouling capacity Alginate average hydrodynamic diameters in the various salt solutions and its
corresponding permeate flux declines.

The structure of the fouling layer on a membrane surface is Feed solution Aggregate/particle size Flux
affected by both chemical (i.e. pH, ionic strength, and divalent (nm) decline
cations) and physical (initial permeate flux, cross-flow velocity and (%)
osmotic pressure gradient) conditions. Among these factors, it has
FO PRO
been documented that the presence of divalent ions such as calcium
plays a predominant role in membrane fouling by alginate [33]. Alg þDI water 1257 4.38
Calcium has the ability to bind preferentially to the carboxylate Alg þ100 mM NaCl 1537 2.44 19 35
groups of alginate in a uniquely arranged manner to result in highly Alg þ94 mM NaClþ2 mM CaCl2 1857 1.86 31 39
Alg þ70 mM NaCl þ 10 mM MgCl2 2477 7.53 28 31
arranged gel networks [34]. Table 3 presents hydrodynamic diameter
Alg þ64 mM NaClþ 10 mM 281 7 5.55 35 43
of alginate in the different salt solutions and their fouling propensity, MgCl2 þ 2 mM CaCl2
determined as total flux decline after 10 h fouling experiments. The Alg þ500 mM NaCl 1977 2.55 25 47
average aggregate size of alginate varied among the range of Alg þ476 mM NaClþ 8 mM CaCl2 4137 6.12 28 57
electrolyte solutions. When dissolved in deionised water, the alginate Alg þ350 mM NaCl þ 50 mM MgCl2 1977 5.66 29 49
Alg þ326 mM NaClþ50 mM 1095 7 18.5 30 57
aggregates had a size of 125 nm. However, a notable increase in MgCl2 þ 8 mM CaCl2
the size was observed with the addition of Na þ , Ca2 þ and Mg2 þ and
the largest size was recorded as a result of the combined presence of
calcium and magnesium ions in solution at lower ionic strength
(0.1 M). When raising the total ionic strength (to 0.5 M) as well as the
divalent concentration (8 mM and 50 mM), the influence of Ca2 þ highly dependent on the magnitude of internal concentration polar-
inclusion was clearly revealed. The aggregate size was increased ization acting on the two operational modes. Therefore, to eliminate
more than twice when the Ca2 þ concentration was increased from the impact of different initial fluxes on the fouling behavior, the same
2 mM to 8 mM confirming the intermolecular complexation bet- osmotic pressure difference was maintained on both modes. Fig. 2
ween alginate macromolecules with calcium resulting in larger compare the flux decline profiles in FO and PRO mode for one calcium
aggregate sizes. The deposited alginate particles on the membrane containing feed solution at lower and high total ionic strengths (0.1
surface resulted in a loose and less compact fouling layer that and 0.5 M). The feed facing active layer experiment showed relatively
allowed continued water permeation. The persistent permeation stable flux decline recording less than 10% (17.44–16.20 L m  2 h  1)
provided a constant permeation drag force that led to the high rate of flux drop in the first 3 h (Fig. 2), while flux decline was rapid in the
alginate particle deposition on the membrane surface [35]. There feed facing porous support layer experiment with 22% (17.81–
were no clear correlations between particle size and permeate flux 13.74 L m  2 h  1) flux lost in an equivalent time duration. The overall
decline. The presence of MgCl2 at lower ionic strength resulted in flux losses over the 10 h period were however not very different with
large alginate particles than the alginate–calcium combination but recordings of 31 and 39% respectively. A similar observation was made
resulted in lower flux decline an observation that can be to the less when the feed solution's total ionic strength was raised to 0.5 M; a
tortuous path of larger particles which results in a cake layer with high rate of flux decline at the initial stages was recorded for the PRO
larger pore volume. A similar observation was recorded at higher mode with an even larger overall loss of 57% compared to the 28% of
ionic strength where the presence of both Ca2 þ and Mg2 þ resulted the active layer. The higher fouling rate observed in the PRO mode can
in largest particles (1095 nm) but recorded the same flux decline as be attributed to the larger pores of the porous support layer and
the calcium containing feed solution. The highest recorded permeate consequently its rough structure [36]. This surface allowed the easy
flux declines were 35% and 57% for FO and PRO mode, respectively, deposition and adhesion of alginate aggregates on the membrane
and were due to feed solutions containing calcium ions. surface. The aggregates were trapped within the porous structure
during the initial stages of fouling causing rapid flux decline. On
3.3. The influence of membrane orientation on the alginate fouling the other hand, the tighter and smoother structure of the dense layer
rate was resilient to foulant deposition thus lowering the fouling propen-
sity in FO mode [37]. It can therefore be concluded that the hydro-
As already mentioned before, the FO membrane is asymmetric and dynamic conditions (permeation drag and shear force) were dominant
characterized by a smooth, dense active layer and a porous support in governing the fouling behavior of alginate on the two membrane
layer. The effective osmotic driving force in feed facing active layer surfaces. The membrane's structural arrangement also influenced the
(FO) and the draw solution facing active layer (PRO) configuration is extent of foulant settling on the surface.
 M.M. Motsa et al. / Journal of Membrane Science 460 (2014) 99–109
Fig. 2. Comparison between flux decline curves of FO and PRO at the two tested
feed ionic strengths (0.1 M and 0.5 M). The feed solution was composed of
alginateþ 0.094 M NaClþ 0.002 M CaCl2 at 0.1 M and alginateþ 0.476 M
NaClþ0.008 M CaCl2 at higher ionic strength.
3.4. Effect of feed solution ionic strength and composition on alginate
fouling
Membrane fouling results from the interaction of a matrix of
factors and from this work it has been demonstrated that calcium
binding, divalent ion concentration, membrane structural arrange-
ment and hydrodynamic parameters were predominant in influen-
cing alginate fouling in PRO mode. Fouling rate remained similar
when the total ionic strength was increased five times to 0.5 M in FO
mode as illustrated by the flux decline curves in Fig. 3. There was
generally less membrane fouling in FO mode due to the resilient
active layer and as observed the increase in ionic strength had no
obvious effect on the fouling. This observation is complemented by
the fouling trends depicted in Fig. 3. It shows that increasing the total
ionic strength from 0.1 M to 0.5 M resulted to a similar total flux
decline as well as raising the calcium concentration from 0.002 M to
0.008 M yielded a very similar fouling behavior. In the case of
calcium containing solutions, the final flux decline in the high ionic
strength solution was even 3% lower (only 28%, compared to 31%
recorded at lower ionic strength) a phenomenon also highlighted by
Brink et al. [38] in their work on the effect of calcium and ionic
strength on alginate fouling on ultrafiltration membranes. This
observation might be due to electrostatic interactions between the
alginate macromolecules and the charged membrane surface.
3.5. Membrane surface tensions and interfacial free energies
3.5.1. Alginate–alginate free energy of cohesion, ΔG131
The measured contact angles and calculated surface tension
parameters for alginate due to the different salt solutions are listed
in Table 4. Measured droplet contact angles varied amongst the
different probe solutions, especially at higher ionic strength where
the value of the measured contact angle increased with the addition
of divalent ions. This was indicative of the changes on the alginate's
surface functionality. The Lifshitz–van der Waals component did not
change significantly amongst the salt solutions. This is understand-
able due to its dependence on the apolar liquid (diiodomethane)
contact angle, which gave relatively similar contact angle values
( 401) on the alginate and membrane surface. The alginate exhib-
ited a mono-polar electron-donor (γ  ) functionality with the
electron-acceptor (γ þ ) functionality reduced significantly to very
small values (0). The electron donor functionality increased in the
presence of (multivalent) cationic species. The NaCl solution had the
highest electron donor (γ  ) component due to the “charging-up”
105
Fig. 3. Comparison of the effect of ionic strength on alginate's fouling capacity. The
effect was tested between feed solutions containing: alginateþ0.1 M NaCl and
alginateþ 0.5 M NaCl as well as those containing alginateþ 0.094 M NaClþ 0.002 M
CaCl2 and alginateþ 0.476 M NaClþ0.008 M CaCl2. These fouling runs were all
performed in FO mode.
effect on alginate as referred by Jin and co-workers [27], on a similar
work with alginate fouling on RO membranes. However, the addition
of divalent cations reduced the electron-donor component and
slightly increased the electron-acceptor component. The pKa value
of carboxylic acids is dependent on the pH and ionic strength of the
solution they exist in and has shown a decreasing trend with
increase in ionic strength [39]. Therefore, the decrease in the electron
donor functionality in the presence of divalent cations (Ca2 þ and
Mg2 þ ) suggest interaction (binding) of these ions with the already
deprotonated carboxylic groups. Calcium ions were the most effec-
tive in reacting more specifically with the negatively charged
carboxylate groups, indicated by an alginate γ  of 30% and 8% lower
than that of alginate in sodium and magnesium solutions at lower
ionic strength, in the presence of calcium. In the presence of both
divalent ions (Mg2 þ and Ca2 þ simultaneously), high electron donor
functionality was obtained suggesting competition between Ca2 þ
and Mg2 þ off-setting the Ca2 þ effect. At higher ionic strength
(0.5 M) the effect of calcium was dominant, reducing the electron
donor functionality by a factor of 4 and 2 compared to that of sodium
and magnesium respectively.
The interfacial free energy of cohesion reveals that the alginate
aggregates in most of the electrolyte solutions are unstable (negative
cohesion energy values) indicating favorable coagulation, thus allow-
ing particles to form even larger aggregates. The exceptions were the
solutions of NaCl and the combined presence of Ca2 þ and Mg2 þ
(0.1 M), which show stable states (positive cohesion energy values),
indicating that the aggregates were repulsive towards each other.
Therefore, alginate is expected to cause more fouling in Ca2 þ contai-
ning feed solutions, particularly at seawater ionic levels as predicted
by the high magnitude (more negative energy of cohesion) of the
cohesion energies.
3.5.2. Alginate–membrane interfacial free energy of adhesion
Table 5 displays the calculated alginate–membrane interfacial free
energies of adhesion, ΔG132, in different salt solutions. The alginate–
membrane interfacial free energy of adhesion gives an insight into
the likelihood of alginate being attracted or repelled by the mem-
branes. It relates to processes such as adsorption and adhesion of
dissimilar materials. The increasingly negative values signify greater
foulant–membrane attraction. As previously reported, calcium pre-
ferentially bind to the oxygen atoms of the deprotonated alginate
106 M.M. Motsa et al. / Journal of Membrane Science 460 (2014) 99–109

Table 4
Contact angles, surface free energy and interfacial free energy of cohesion for alginate complexes.

Testing solution Alginate contact angle γ γþ γ γTOT Energy of cohesion, ΔG131 (mJ/M2)

Deionised water 37.12 72.19 40.51 0.014 44.54 42.09 26.29

0.1 M NaCl 53.31 73.58 41.03 0.16 34.19 44.10 9.66
0.094 M NaClþ0.002 M CaCl2 52.83 73.68 41.03 0.31 23.67 46.47  9.35
0.07 M NaClþ 0.01 M MgCl2 54.95 74.21 41.03 0.25 25.93 46.12  5.25
0.064 M NaClþ 0.01 M MgCl2 þ 0.002 M CaCl2 46.86 72.48 41.03 0.13 35.50 43.90 11.76
0.5 M NaCl 37.48 72.92 40.14 0.0025 49.28 40.85 33.84
0.350 M NaClþ 0.05 M MgCl2 53.12 73.64 40.14 0.30 26.34 45.72  4.07
0.476 M NaClþ0.008 M CaCl2 64.51 71.76 40.14 0.95 12.08 46.93  31.21
0.326 M NaClþ0.05 M MgCl2 þ 0.008 M CaCl2 63.96 72.89 40.14 0.91 12.68 46.94  29.96

Table 5 3.5.3. Correlation of permeate flux decline and interfacial free

Alginate–membrane interfacial free energy of adhesion in different salt solutions.
Testing solution Energy of adhesion,
ΔG132 (mJ/M2)
FO
Deionised water 26.29
0.1 M NaCl 17.80
0.094 M NaClþ0.002 M CaCl2 9.31
0.07 M NaClþ 0.01 M MgCl2 6.88
0.064 M NaClþ 0.01 M MgCl2 þ 0.002 M CaCl2  5.50
0.5 M NaCl 30.02
0.350 M NaClþ 0.05 M MgCl2 9.80
0.476 M NaClþ0.008 M CaCl2  9.16
0.326 M NaClþ0.05 M MgCl2 þ 0.008 M CaCl2  7.04
carboxylate groups in solution to form highly arranged complexes/
aggregates that get deposited on the membrane surface and result in
a labile gel layer responsible for the severe permeate flux decline in
membrane fouling [40]. Similar observations were made in this
work; greater attractions between the alginate–calcium complexes
and the membrane were observed in both operational modes. The
active layer (FO mode) had a greater resilience to foulant deposition.
The alginate aggregates and the membrane active surface layer were
repellent of each other (positive adhesion values) at low ionic
strength but slightly declining with the addition of divalent ions
and increasing ionic composition. Therefore, less alginate fouling was
supposed to occur on the active layer except for the solutions
containing calcium and the combination of calcium and magnesium
which revealed intermediate attractive forces (slightly negative
values) between the membrane and the foulant. The attraction was
due to the binding of calcium and magnesium to the oxygen atom of
the negative acetate groups on the membrane surface enhancing
attachment of the already complexed (calcium-doped) alginate
macromolecules.
The energy data also revealed that the support layer was less
resilient (tolerant) towards alginate aggregate deposition on the
surface; negative interaction energy values were recorded for almost
all electrolyte solutions. The adhesion energy for calcium containing
solutions increased by a factor of three and two compared to that of
sodium and magnesium at 0.1 M total ionic strength. The recorded
energy of adhesion further increased suggesting stronger alginate–
membrane attractions when the calcium ion concentration was
increased to 8 mM. This was confirmed by the increased presence
of alginate aggregates on the membrane surface for feed solutions
containing 8 mM Ca2 þ ions since free calcium ions formed bridges/
bonds between the membrane and the already complexed alginate
macromolecules. For the CTA membrane the free energy of adhesion
with alginate in different electrolytes followed this ascending order:
NaCloMgCl2 oCaCl2 rCaCl2 þ MgCl2.
energies on membrane fouling
Hydrodynamic drag forces bring the particles close to the mem-
brane surface, but the chemical interactions cause the actual binding
of the particles to the membrane. The membrane–alginate interaction
PRO energies are thus expected to dictate the initial fouling behavior.
Overtime, the membrane surface eventually becomes covered with
0.45
 5.33 the alginate particles, and consequently the interaction energy
 10.86 between approaching alginate molecules and the already deposited
 16.94 alginate governs the long-term fouling. At this phase of the membrane
 15.68
fouling, the continued deposition of foulants is controlled by the
3.95
0.24
energy of cohesion between the alginate molecules rather than
 29.04 the energy of adhesion between the approaching alginate molecules
 32.14 and the membrane surface. In Fig. 4, the initial fouling rate is shown to
correlate strongly with the interfacial free energies of adhesion
(alginate–membrane interactions) for both FO and PRO modes with
a correlation factor of 0.954. The two operational modes produced
different initial fouling rates due to the distinct membrane sur-
face structures; however, the strong correlation between the flux
decline rates and the interfacial free energy of adhesion indicates that
the fouling mechanism is similar for both modes. The free energy of
cohesion also correlated with the later fouling rates (Fig. 5) for both
experimental modes resulting in a good correlation factor of 0.827. The
observed correlation further reinforce that the later deposition and
accumulation of alginate aggregates is mainly dependent on the inter-
actions between on-coming alginate particles and those already
covering the membrane surface (foulant–foulant interactions). The
initial fouling rate was determined from the slope of relative flux
decline curve within the first 2 h. The later fouling rate was quantified
from the slope of relative flux decline curve between 8 and 10 h. All
the examined parameters (calcium binding, membrane surface func-
tionality, variations in ionic strength) gave an insight on the mem-
brane fouling behavior, it is however clear that the non-electrostatic
forces dominantly govern fouling behavior, particularly in PRO mode.
This is a plausible observation since the CTA membrane is weakly
charged.
3.6. Fouling reversibility
Flux decline was completely reversible by all cleaning conditions
for fouling with alginate in the presence of sodium and magnesium
ions. Therefore, the reversibility of the fouling layer on the mem-
brane surface was determined using the feed solution with the
highest fouling propensity (8 mM CaCl with alginate in the presence
of NaCl). As it is displayed in Table 6, all cleaning conditions recorded
an efficiency of more than 80% after the three consecutive fouling–
cleaning cycles. The high reversibility can be attributed to the loosely
formed alginate gel layer on the membrane surface which was easily
removed by an increase in shear force generated from increased
cross-flow velocity as also observed by Mi et al. and Amy et al. [41].
During membrane cleaning the fouling layer was broken down to
smaller pieces that were easily removed from the membrane surface.
 M.M. Motsa et al. / Journal of Membrane Science 460 (2014) 99–109 107

Table 6
0.14 Permeate flux recovery through non-chemical methods.

0.12 Cleaning method Flux recovery, J/Jo (%)

Cleaning cycles
Initial Fouling Rate (h )
-1

0.10 1st 2nd 3rd

0.08 Osmotic back-wash 100 99 93

Salt solutions 88 81 81
Deionised water 98 93 91
0.06

0.04
of 0.5 M (without alginate) on both sides of the membrane. This
2
R = 0.954 cleaning condition recorded an efficiency 10% lower than that the
0.02
two other conditions. The results suggest that the ultrapure water
had an effect on the hydrolysis of the alginate–calcium complexes
0.00
-40 -30 -20 -10 0 10 20 30 and also on depleting the concentration polarization profile. These
2 results further suggest that the major cleaning mechanism under our
Free Energy of Adhesion (mJ/m )
performed test conditions was physical removal of the fouling layer.
Fig. 4. Correlations between the initial fouling rate of flux decline and alginate–
membrane interfacial free energy of adhesion for both operational modes (FO and PRO).
4. Conclusions

0.035
The membrane surface and alginate charge density was altered
due to the exposure to electrolyte solutions of varying Ca2þ and Mg2þ
0.030
composition. The free calcium ions formed bridges with the alginate
Later Fouling Rate (h )
-1

0.025
0.020
0.015
0.010
2
R = 0.827
0.005
0.000
-40 -30 -20 -10
Free Energy of Cohesion (mJ/m )
Fig. 5. Correlations between the late fouling rate of flux decline and alginate–alginate
interfacial free energy of cohesion for both operational modes (FO and PRO).
Membrane cleaning through osmotic back wash with ultrapure
(Milli-Q) water recorded the highest performance; whereby the
recovered permeate flux was higher than 93% of the initial permeate
flux in all cases. This could be a coupled effect of the permeate drag
(net water movement from the draw side to the feed side) and cross-
flow velocity where the gel layer is loosened and segregated into small
particles that are flushed back to the bulk solution. The concentration
polarization profile generated during fouling is also reversed, restoring
the charge on the membrane surface which had the abundance of
cationic species. To determine the role of the ultrapure water in
restoring the membrane surface charge, a cleaning condition different
from osmotic back wash was also tested on its cleaning performance,
where by ultrapure water was installed on both sides of the mem-
brane and during cleaning there was limited permeation as well as
salt diffusion. Flushing with ultrapure water recorded cleaning effi-
ciencies of more than 90%, close to that of osmotic back wash. It is
clear that ultrapure water thus plays a significant role in restoring
membrane surface charge. Ultrapure water also diluted and removed
the salt from the cake layer allowing the electrical double layer of
charged species to enlarge resulting in repulsion between the alginate
particles forming the fouling cake layer an observation also made
earlier by Li et al., for backwashing of UF membranes with ultra-
pure water.
To clearly show this effect, we cleaned the fouled membrane by
flushing with salt solutions of NaCl and CaCl2 at total ionic strength
macromolecules resulting in alginate–Ca complexes that got deposited
on the membrane surface reducing permeate flux especially at high
calcium concentrations. The fouling propensity of alginate was not
influenced by an increase in solution total ionic strength, the fouling
rate remained similar at raised total ionic composition (0.5 M). The
membrane active layer was more resistant towards alginate aggre-
gates settling compared to the visibly thick fouling layer noticed on
the support layer. Permeate flux decline in PRO mode was found to be
hugely influenced parameters such as calcium binding, hydrodynamic
conditions (cross-flow velocity and permeate drag) and membrane
surface properties (porous structure of support layer). The alginate and
0 10 20 30 40 membrane exhibited a mono-polar electron-donor functionality with
2
high values of the electron-donor component (γ  ) and negligible
electron-acceptor component (γ þ ). The interaction between alginate
and calcium ions increased the hydrophobicity of alginate (more
negative free energy of cohesion values) and further reduced the
electrostatic repulsions between the alginate macromolecules. The
rate of initial adhesion of alginate particles on the membrane was
controlled by the extent of chemical interactions between the mem-
brane surface functionality and the alginate. Strong correlation bet-
ween the initial rate of fouling and the free energy of adhesion
confirmed the above observation. Long term fouling of the membrane
was governed by the interactions between the already laid alginate
layer and the approaching particles; this was further complemented
by the good correlation between the later fouling rate and the free
energy of cohesion. A high magnitude of permeate flux recovery was
recorded (480%) when the membrane was cleaned through osmotic
back-wash and with ultrapure pure water and salt solutions. Cleaning
was mainly due to the physical removal of the fouling layer by shear
force generated from the increased cross-flow velocity as well as the
dilution and removal of salt from the cake layer leading to its
depletion to smaller easily removed particles.
Probing the membrane and alginate surface energetics helped
elucidate the influence of the electron acceptor and electron donor
(AB) interactions on alginate fouling. This knowledge and the
understanding of the specific ion interactions with foulants can
provide more insight into strategies of controlling organic fouling
of FO membranes. Modification of membrane support layer and
specific reduction in calcium ion concentration in pre-treatment
can lower the fouling propensity of FO membranes in forward
osmosis seawater desalination.
108 M.M. Motsa et al. / Journal of Membrane Science 460 (2014) 99–109
Fig. A1. SEM micrographs of a fresh CTA membrane surface: (a) dense active layer; (b) loose support layer.
Acknowledgments
This research work was financially supported by the University
of Gent, Belgium and the University of Johannesburg, South Africa.
The authors would also like to thank Emile Cornelissen at KWR
Water cycle Research Institute for the interesting discussion and
HTI for the supply of FO membranes.
Appendix A. Scanning electron microscopy characterization
The morphologies of the active and support layer of a fresh
membrane were studied using a JOEL (JSM 5600) scanning electron
microscopy. The dried membrane samples were carbon coated then
analyzed at varying magnifications. The SEM micrographs in Fig. A1
show the surface morphologies of the dense smooth active layer
and the rough loosely arranged support layer of a fresh membrane.
The support layer shows the presence of thin interwoven
polyester mesh.
References
[1] D. Mattai, K.P. Lee, T.C. Arnot, A review of reverse osmosis membrane materials
for desalination – development to date and future potential, J. Membr. Sci. 370
(2011) 1–22.
[2] J. Dong, L. Li, T.M. Neoff, R. Lee, Desalination by reverse osmosis using MFI
zeolite membranes, J. Membr. Sci. 243 (2004) 401–404.
[3] T.Y. Cath, Osmotically and thermally driven membrane processes for enhance-
ment of water recovery in desalination processes, Desalin. Water Treat. 15
(2010) 279–286.
[4] C.Y. Tang, T.H. Chong, A.G. Fane, Colloidal interactions and fouling of NF and RO
membranes: a review, Adv. Colloid Interface Sci. 164 (2011) 126–143.
[5] D. Paul, A.R.M. Abanmy, RO membrane fouling – the final frontier, Ultrapure
water 7 (3) (1990) 25–36.
[6] Y.C. Yang, Y. Kwon, J.O. Leckie, Fouling of reverse osmosis nanofiltration
membranes by humic acid – effects of solution composition and hydrody-
namic conditions, J. Membr. Sci. 290 (2007) 86–94.
[7] O.A. Bamaga, A. Yokochi, B. Zabara, A.S. Babaqi, Hybrid FO/RO desalination
system: preliminary assessment of osmotic energy recovery and designs of
new FO membrane module configuration, Desalination 268 (2011) 163–169.
[8] E.M.V. Hoek, M.C.Y. Wong, K. Martinez, G.Z. Ramon, Impacts of operating
conditions and solution chemistry on osmotic membrane structure and
performance, Desalination 287 (2012) 340–349.
[9] D. Wang, M. Wang, L. Lui, C. Gao, Current patents of forward osmosis
membrane process, Recent Pat. Chem. Eng. 2 (2009) 76–82.
[10] M. Elimelech, W.A. Philip, The future of seawater desalination: energy,
technology, and the environment: review, Science 333 (2011) 712–717.
[11] T.Y. Cath, A.E. Childress, E. Elimelech, Forward osmosis: principles, applications
and recent developments, J. Membr. Sci. 281 (2006) 70–87.
[12] A. Subramani, M. Badruzzaman, J. Oppenheimer, J.G. Jacangelo,Energy, mini-
mization strategies and renewable energy utilization for desalination: a
review, Water Res. 45 (2011) 1907–1920.
[13] Z. Liu, H. Bai, J. Lee, D.D. Sun, A low-energy forward osmosis process to
produce drinking water, Energy Environ. Sci. 4 (2011) 2582.
[14] M. Elimelech, J.R. McCutcheon, R.L. McGinnis, Desalination by ammonium–
carbon dioxide forward osmosis: influence of draw and feed solutions on
process performance, J. Membr. Sci. 278 (2006) 114–123.
[15] W.C.L. Lay, J. Zhang, C. Tang, R. Wang, Y. Liu, A.G. Fane, Factors affecting flux
performance of FO systems, J. Membr. Sci. 394–394 (2012) 151–168.
[16] C.Y. Tang, X. Peng, Q.S. Fu, S. Nie, Effect of draw solution concentration and
operating conditions on FO and PRO performance in spiral wound module,
J. Membr. Sci. 348 (2010) 298–309.
[17] B. Mi, M. Elimelech, Chemical and physical aspects of organic fouling of
forward osmosis membranes, J. Membr. Sci. 320 (2008) 292–302.
[18] B. Mi, M. Elimelech, Organic fouling of forward osmosis membranes: Fouling
reversibility and cleaning without chemical reagents, J. Membr. Sci. 348 (2010)
337–345.
[19] S. Hong, S. Lee, C. Boo, M. Elimelech, Comparison of fouling behaviour in forward
osmosis (FO) and reverse osmosis (RO), J. Membr. Sci. 365 (2010) 34–39.
[20] H. Yong Ng, V. Parida, Forward osmosis organic fouling: effects of organic
loading, calcium and membrane orientation, Desalination 312 (2012) 88–98.
[21] C.Y. Tang, Q. She, W.C.L. Lay, R. Wang, A.G. Fane, Coupled effects of internal
concentration polarization and fouling on flux, J. Membr. Sci. 354 (2010)
123–133.
[22] J. Zhang, W.L. Loong, C. Chou, C. Tang, A.G. Fane, Membrane biofouling and
scaling in forward osmosis membrane bioreactor, J. Membr. Sci. 403–404
(2012) 8–14.
[23] A.J. Brant, A.E. Childress, Assessing short-range membrane–colloid interac-
tions using surface energetic, J. Membr. Sci. 203 (2002) 257–273.
[24] G. Hurwitz, G.R. Guillen, E.M.V. Hoek, Probing polyamide membrane surface
charge, zeta potential, wettability and hydrophobicity with contact angle
measurements, J. Membr. Sci. 349 (2010) 349–357.
[25] A.E. Childress, J.A. Brant, Colloidal adhesion to hydrophilic membrane surfaces,
J. Membr. Sci. 241 (2004) 235–248.
[26] S. Lee, S. Kim, J. Cho, E.M.V. Hoek, Natural organic matter fouling due to
foulant–membrane physicochemical interactions, Desalination 202 (2007)
377–384.
[27] X. Jin, X. Huang, E.M.V. Hoek, Role of specific ion interactions in seawater RO
membrane fouling by alginic acid, Environ. Sci. Technol. 43 (2009) 3580–3587.
[28] J.R. McCutcheona, L.R. McGinnisb, M. Elimelech, A novel ammonia–carbon dioxide
forward (direct) osmosis desalination process, Desalination 174 (2005) 1–11.
[29] W.A. Phillip, J.S. Yon, M. Elimelech, Reverse draw solute permeation in forward
osmosis: modeling and experiments, Environ. Sci. Technol. 44 (2010)
5170–5176.
[30] C.J. Van Oss, Acid–base interfacial interactions in aqueous media, Colloids Surf.
A: Physicochem. Eng. Asp. 78 (1993) 1–9.
[31] S. Botton, A.R.D. Verliefde, N.T. Quach, E.R. Cornelissen, Influence of biofouling
on pharmaceuticals rejection in NF membrane filtration, Water Res. 46 (2012)
5848–5860.
[32] T.Y. Chiu, A.E. James, Electrokinetic characterisation techniques on asymmetric
microfiltration membranes, Colloids Surf. A: Physicochem. Eng. Asp. 301
(2007) 281–288.
[33] K. Katsoufidiou, S.G. Yiantsios, A.J. Karabelas, Experimental study of ultrafil-
tration membrane fouling by sodium alginate and flux recovery by back-
washing, J. Membr. Sci. 300 (2007) 137–146.
[34] K. Katsoufidiou, S.G. Yiantsios, A.J. Karabelas, UF membrane fouling by
mixtures of humic acids and sodium alginate: fouling mechanisms and
reversibility, Desalination 264 (2010) 220–227.
[35] Z. Li, V. Yangali-Quintanilla, V.R. Valladares-Linares, Q. Li, T. Zhan, G. Amy, Flux
patterns and membrane fouling propensity during desalination of seawater by
forward osmosis, Water Res. 46 (2012) 195–204.
[36] Y. Mo, K. Xiao, Y. Shen, X. Huang, A new perspective on the effect of
complexation between calcium and alginate on fouling during nanofiltration,
J. Membr. Sci. 82 (2011) 121–127.
 M.M. Motsa et al. / Journal of Membrane Science 460 (2014) 99–109
[37] Y. Lui, B. Mi, Combined fouling of forward osmosis membrane: synergistic
foulant interaction and direct observation of fouling layer formation, J. Membr.
Sci. 407–408 (2012) 136–144.
[38] P. van den Brink, A. Wzijnenburg, G. Smith, H. Temmink, M. van Loosdrecht,
Effect of free calcium concentration and ionic strength on alginate fouling in
cross-flow membrane filtration, J. Membr. Sci. 345 (2009) 207–216.
[39] K. Listiarini, L. Tan, D.D. Sun, J.O. Leckie, Systematic study on calcium–alginate
interaction in a hybrid coagulation–nanofiltration system, J. Membr. Sci. 370
(2011) 109–115.
109
[40] 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.
[41] R.V. Valladares-Linares, V. Yangali-Quintanilla, Z. Li, G. Amy, NOM and TEP
fouling of forward osmosis (FO) membrane: Foulant identification and clean-
ing, J. Membr. Sci. 421–422 (2012) 217–224.