Desalination 250 (2010) 236–248

Contents lists available at ScienceDirect

Desalination
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / d e s a l

Inorganic fouling of pressure-driven membrane processes — A critical review

Saqib Shirazi a, Che-Jen Lin b,c,⁎, Dong Chen b,1
a
Texas Water Development Board, Austin, TX 78701, United States
b
Department of Civil Engineering, Lamar University, Beaumont, TX 77710, USA
c
College of Environmental Science & Engineering, South China University of Technology, Guangzhou University Town, Guangzhou, 510006, China

a r t i c l e i n f o a b s t r a c t

Article history: One of the major limitations of the application of membrane processes in water and wastewater treatment is
Accepted 22 February 2009 inorganic fouling. Despite the extensive studies on concentration polarization and inorganic scaling in
Available online 13 October 2009 membrane filtration, the fundamental mechanisms and processes involved in inorganic fouling are not fully
understood. This paper critically reviews the mechanisms and models of concentration polarization and
Keywords:
inorganic fouling in pressure-driven membrane processes. Effects of operating parameters and membrane
Membrane filtration
Concentration polarization
properties on the formation of inorganic scale at the membrane surface are also evaluated. Future research
Inorganic fouling areas that need to be pursued to alleviate inorganic fouling problems in membrane installations are discussed.
Filtration model © 2009 Elsevier B.V. All rights reserved.
Scaling

1. Introduction membranes differ in their pore sizes, operating pressures, and

Membrane is a selective barrier between two phases that restricts
the transport of particulate, colloidal, and dissolved chemical species
other than solvent or water. The selective transport is achieved based on
the differences in the physical and/or chemical properties of permeating
components across the membrane. In recent years, membrane
processes are extensively employed in the textile, pharmaceutical,
pulp and paper, semi-conductor, tanning and leather, mining, electro-
plating, dairy, food and beverage processing, and biotechnology
industries as well as for water and wastewater treatment. In water
and wastewater industries, four types of membranes (microfilter,
ultrafilter, nanofilter and reverse osmosis membrane) are widely used
for the removal of hardness, color, biogenics, and disinfection by-
products and their precursors to produce required quality of processed
water.
Membranes used in water and wastewater industry can be broadly
classified into two major groups: porous membranes and non-porous
membranes. Porous membranes separate particles based on sieving,
straining, or size exclusion. Microfiltration, ultrafiltration and loose
end nanofiltration membranes are examples of porous membranes
[1]. Non-porous membranes separate molecules based on the
differences in solubility or diffusivity between the solvent and the
solute in the membranes [2,3]. Tight end nanofiltration and reverse
osmosis membranes are typical non-porous membranes. These
⁎ Corresponding author. Department of Civil Engineering, Lamar University, Beaumont,
TX 77710, USA.
E-mail address: Jerry.Lin@lamar.edu (C.-J. Lin).
1
Current address: Department of Engineering, Indiana University - Purdue University
Fort Wayne, Fort Wayne, IN 46805, USA.
applications, which are summarized in Table 1.
Traditional methods to remove pollutants from water and waste-
water are coagulation, flocculation, sedimentation, sand filtration, ion
exchange, electrodeposition, extraction, precipitation and biological
degradation, etc. Most of which have disadvantages of operating in a
successive steps of heterogeneous reactions, or distribution of sub-
stances between different phases that usually require a lengthy
operating period and a large area [4,5]. Membrane processes are of
great interest because they reduce the number of unit operations,
recycle process water, and recover valuable products for other
applications [5]. Additional inherent advantages, such as selective
separation, continuous and automatic operation, relatively low capital/
operating cost, easy scale-up and low space requirement, made
membrane filtration an attractive alternative compared to conventional
treatment [6–9].
Because of the potential of the membrane applications, many
membranes have been and are being developed by different manu-
facturers (e.g., Nitto Denko Corp., Koch, Toray Industries, Dow/FILMTEC,
GE Osmonics, etc). Most of the membranes are composed of organic
polymers, such as aromatic polyamide, polysulphonates, polyvinyl
alcohol, piperazineamide, polyimide, and polyacetylene [10,11]. Table 2
shows some of the commercially available membranes. Inorganic
membranes are also used for the rejection of organic molecules and
ion separation in aqueous solutions. The major strength of inorganic
membranes is their higher chemical, thermal, and mechanical stability
compared to organic membranes [12]. Applicability of inorganic
membranes is of great importance in non-aqueous filtration due to
their stability in organic solvents [13]. High temperature-resistant
inorganic membranes have been prepared from titania, silica–zirconia,
and alumina [13].

0011-9164/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.desal.2009.02.056
 S. Shirazi et al. / Desalination 250 (2010) 236–248 237

Symbols α̂ Specific cake resistance per unit thickness for incom-

pressible cake (m− 2)
a Numerical constant characterizing the flow channel α′ Reynolds number exponent in the mass transfer
geometry
correlation
ap Particle radius (m) β Area occupied per unit mass (m2/kg)
Ab Membrane area occupied by surface crystals (m2)
β′ Schmidt number exponent in the mass transfer
Ac Cross-sectional area (m2) correlation
At Total membrane surface area (m2)
δ Film thickness (m)
C Concentration of solute (kg/m3) δc Cake thickness (m)
Cb Concentration of solute in bulk (kg/m3)
δm Membrane thickness (m)
Cg Concentration of solute at gel layer (kg/m3) ε Cake porosity
Cm Concentration of solute at the membrane surface (kg/m3)
εc Void fraction of the cake
Cp Concentration of solute in permeate (kg/m3) μ Solution viscosity (N s/m2)
D Diffusion coefficient (m2/s)
ν Kinematic viscosity (m2/s)
D* Diffusion coefficient of salt in hindered layer (m2/s) ρ Particle density (kg/m3)
dh Hydraulic diameter (m)
πb Osmotic pressure of feed (KPa)
dp Particle diameter (m) πp Osmotic pressure of permeate (KPa)
f Degree of concentration polarization
σ Reflection coefficient
i Symbol to identify a solute τ Cake tortuosity
j Factor for mole increase due to dissociation for a solute
J Permeate flux (m3/h/m2)
J0 Initial flux (m3/h/m2)
Jb Permeate flux estimated from bulk crystallization (m3/
Despite its potential in water treatment, certain limitations prohibit
h/m2)
membrane process from large-scale and continuous operation ([14].
JCrit Critial flux (m3/h/m2)
One of the major limitations arises from membrane fouling caused by
Js Permeate flux due to surface crystallization (m3/h/m2)
different inorganic salts [15], which reduces permeate flux, increases
Jf Flux of fouled membrane (m3/h/m2)
feed pressure, decreases product quality and ultimately shortens
ji Factor for mole increase due to dissociation of solute i
membrane life [16,17]. Consequently, membrane fouling increases the
(Jv)H2O Permeate flux emanating from salt-free water (m3/h/m2)
costs by increasing (1) energy consumption, (2) system down time,
(Jv)salt Permeate flux emanating from saline water (m3/h/m2)
(3) necessary membrane area, and (4) construction, labor, time, and
K Boltzman constant
material costs for backwashing and cleaning processes. Successful
K′ Constant
application of membrane technology, thus, requires efficient control of
k Mass transfer coefficient (m/s)
membrane fouling ([18]). However, lack of clear understanding of the
k′ Cake-hindered mass transfer coefficient (m/s)
fouling mechanisms is a challenge in water treatment using membrane
L Channel length (m)
technology. In the past decade, numerous studies had been performed
ms Weight of scale formed directly on the membrane
to investigate concentration polarization (CP), cake or gel layer
surface (kg)
formation, membrane pore blocking, and other fouling mechanisms
n Number of moles
on different membranes ([20,21] and the reference cited therein). These
NF Filtration number
studies suggested that membrane fouling due to inorganic salts is
np Number of pores per unit membrane area
dependent on several factors, including but not limited to, membrane
Ps Solute permeability (m/s)
characteristics, module geometry, feed solution characteristics and
Qf Feed flow rate (m3/s)
operating conditions [22–24]. Moreover, several hypotheses have been
R Universal gas constant (J K− 1 mol− 1)
introduced to explain the observed fouling behaviors. The objective of
Rc Cake resistance (m− 1)
this paper is to provide a state-of-science review of membrane fouling
Rcp Resistance due to concentration polarization (m− 1)
caused by inorganic salts in water and wastewater treatment. This
Re Reynolds number
critical review addresses the strengths and deficiencies of the models
Ri Irreversible resistance (m− 1)
and theories proposed to date, summarizes the experimental investiga-
Rm Membrane resistance (m− 1)
tions, and projects future research needs on inorganic fouling.
rp Radius of membrane pores
Rp Resistance due to pore blocking (m− 1)
RT Total resistance (m− 1) 2. Concentration polarization (CP)
Sc Schmidt number
Sh Sherwood number By definition, CP is a phenomenon that the solute or particle
Ss Solids surface area per unit volume of solids in the concentration in the vicinity of the membrane surface is higher than
cake (m2/m3) that in the bulk [25]. The phenomena of CP are inherent to all
T Absolute temperature (K) membrane filtration processes [1]. It causes elevated concentration of
V Cross-flow velocity (m/s) solutes and/or particles at the membrane surface and increases their
VB Potential barrier between particles (m) break through into the permeate stream. This not only increases the
Vi Volume of solution (m3) risk of fouling and deteriorates the quality of the permeate, but also
v Volume (m3) decreases the permeation rate due to increased osmotic pressure in
ΔPp Pressure drop through the accumulated particle layer RO and NF. CP occurs due to the difference in the permeability
(KPa) between the solvent and the particle/solute. The increased concen-
ΔP Applied filtration pressure (KPa) tration of the solutes and particles at the membrane surface results in
Δπ Osmotic pressure difference (KPa) a greater back diffusion into the bulk until a steady state is reached,
α Specific cake resistance for compressible cake (m/kg) when the rate of back diffusion balances the rate of accumulation near
the membrane surface [26]. The factors that enhance the back
238 S. Shirazi et al. / Desalination 250 (2010) 236–248

Table 1
Pore sizes, operating pressures and application of typical membranes in water and wastewater treatment.

Membrane type Pore size (nm) Operating pressure (KPa) Applications

Microfilter 50–2000 10–50 Separate particles and bacteria from other smaller solutes
Ultrafilter 2–50 50–200 Separate colloids from solutes such as sugar or salts
Nanofilter <2 or non-porous 200–1000 Separate multivalent salts, pesticides, herbicides, etc., from water
Reverse osmosis membrane Non-porous 1000–10,000 Separate monovalent salts, small molecules and solvents, etc., from water

diffusion reduce CP during the filtration process, such as an increased The ratio of the diffusion coefficient (D) and the boundary layer
cross-flow velocity, a greater diffusion coefficient of the solute/ (film) thickness (δ) is the mass transfer coefficient k, i.e.:
particle, and a higher temperature. In contrast, an increased filtration
pressure or permeate flux increase CP. D
k= ð3Þ
δ
2.1. CP models
Cm
The ratio of is known as the degree of CP (f). The ratio and thus
Cb
Several quantitative models have been developed to describe CP the concentration at the membrane surface increases with increasing
during membrane filtration. Table 3 summarizes the application and permeate flux, increasing boundary layer thickness, and decreasing
limitation of different CP models. These models are briefly described mass transfer coefficient.
below. For the membrane with 100% solute rejection, the degree of CP
becomes:
2.1.1. Film theory

The solute concentration at the membrane surface can be esti- Cm J
f = = exp ð4Þ
mated based on the film theory of mass transfer. The model assumes Cb k
that the permeation driving force is the gradient in chemical potential
of the solute [27]. Film theory provides a simple analytical approach The major deficiency of this model is that k is assumed to be
that works well for most RO/NF separations. Therefore, film theory constant for all cases, but the diffusion coefficient of macromolecular
serves as the design basis for most modern reverse osmosis processes solute is often concentration-dependent [29]. Moreover, the concen-
[28]. tration increase in the boundary layer is not linear. The film theory
During cross-flow filtration, there is a formation of a boundary also neglects the concept of localized CP near the membrane surface.
layer near the membrane surface where the mass transfer occurs. The
concentration in the layer varies from maximum at the membrane 2.1.2. Spiegler–Kedem model
surface to the minimum in the bulk. Complete mixing caused by feed The Spiegler–Kedem model [30] treats NF membranes as a black
flow turbulence keeps the bulk concentration homogeneous. At box and characterizes CP in terms of solute permeability, Ps, and
steady state, the convective flux of the solute to the membrane is reflection coefficient, σ, both obtained from the experimental data of
equal to the sum of the permeate flux of solute and the diffusive flux of rejection versus permeate flux by a best-fit method [31]. The model
the solute back into the bulk, i.e.: yields the following expression:
" #
dC 1−σ σCp
JC + D = JCp ð1Þ J = ln ð5Þ
dx Ps Cp −Cm ð1−σÞ

Where JC is the solute flux towards the membrane due to convection; The primary difference between this model and the film theory
JCp is the solute flux through the membrane and D dC is the solute flux model is that the Spiegler–Kedem model incorporates the reflection
dx
from the membrane surface back to the bulk. coefficient σ with a value ranging between 0 and 1. The Spiegler–Kedem
Integration of Eq. (1) with the boundary conditions of model conforms to the film theory model when σ = 1, which means
x = 0 ⇒ C = Cm and x = δ ⇒ C = Cb yields: that the convective solute transport does not take place. This is the case
for ideal RO membranes where the membranes have no pores available

 for the convective transport. In an entirely unselective membrane, a
Cm −Cp Jδ
ln = ð2Þ concentration gradient does not cause volumetric flow at all (σ = 0). For
Cb −Cp D
MF, and UF, which have pores, σ is a positive value smaller than 1 [32].

where J is the permeate flux, δ is the boundary layer thickness, Cm, Cb 2.1.3. Gel layer model
and Cp are the solute concentrations at the membrane surface, in the Gel layer model is based on the fact that a solute boundary layer is
bulk (feed) and in the permeate, respectively. developed near the membrane surface during the filtration of macro-
molecules. After polarization, macromolecules polymerize and accu-
Table 2 mulate as gel next to the membrane. Thickness of the gel will continue to
Examples of commercially available membranes for water treatment. increase until a steady state condition is developed. The concentration of
Membrane Manufacturer Material
this gel is dependent on morphological, physical and chemical
properties of the feed solution. Gel formation may be reversible or
NF 70 Filmtec Polyamide
irreversible, a very important factor in membrane cleaning.
NF 255 Filmtec Piperazineamide
NTR-7450 Nitto Denko Synthetic polymer composite When the solute is completely retained by the membrane, the
NTR-7410 Nitto Denko Synthetic polymer composite solvent flux through the membrane increases with pressure until a
Desal-5 DL Osmonics Piperazineamide critical concentration is reached (Cg). If the pressure is increased further,
HL Osmonics Polyamide
the solute concentration at the membrane surface is not capable of any
TFC-S Fluidsystems Polyamide
further increase so the gel may become thicker. This implies that the
 S. Shirazi et al. / Desalination 250 (2010) 236–248 239

Table 3
Application and limitations of different concentration polarization models.

CP Model Application Limitation

Film theory The model determines permeate flux based on chemical potential gradient. It assumes a constant mass transfer coefficient for all cases.
Spiegler–Kedem Similar to solution diffusion theory but incorporates reflection It neglects the phenomenon that CP increases along the membrane
model coefficient as an additional term. surface.
Gel layer model The model determines permeate flux based on the constant gel layer It assumes a fixed surface gel concentration and adapts a mass transfer
resistance (developed by gels of macromolecules) and membrane coefficient from theories of convective heat transfer to impermeable
resistance. surface.
Osmotic pressure model The model determines the osmotic pressure near membrane surface It cannot be applied to MF and UF since osmotic pressure is negligible in
that reduces trans-membrane pressure and permeates flux. these cases.
Resistance in series model The model estimates permeate flux during different fouling stage. It only predicts the fouling behavior of colloids and mono-disperses particles.
Theory of non-interacting The model determines the average permeate velocity of uniform non- It cannot be used for multi-component system.
particles interacting spherical particles.
Cake-enhanced concentration The model is a conceptual analysis of the solute transport and CP in Its performance in the multi-component system is unknown.
polarization cross-flow membrane filtration.

resistance of the gel layer (Rg) to solvent transport increases so that the
gel layer becomes the limiting factor in determining flow. The total
resistance can then be represented by two resistances in series, i.e. the
gel layer resistance and the membrane resistance. For the gel layer
region, flux can be described by:
 
Cg
J = k ln
Cb
where J is flux, Cg is concentration of the solute in the gel, Cb is the
concentration of the solute in the bulk, k is the mass transfer
coefficient. Eq. (6) suggests that if J is plotted against of ln(Cb), the
result must be a straight line with a negative slope. This model has
been used as a tool for directly interpreting experimental data in
porous membranes [25]. Gel layer model is also applicable to non-
porous membranes, which have a higher rejection of macromolecules
than porous membranes.
Although this model may be considered to be a significant
contribution to the theory of CP and limiting flux behavior, there are
also drawbacks of this theory. The theory is a semi-empirical model as it
assumes a fixed surface macromolecular concentration and adapts a
mass transfer coefficient from theories of convective heat transfer to
impermeable surfaces [25]. Neither has been proven theoretically for
cross-flow filtration. Earlier studies have indicated that the gel
concentration (Cg) is not a constant but depends on the bulk
concentration and the cross-flow velocity [1]. Also, k is assumed to be
constant while the diffusivity of macromolecular solute is often
concentration-dependent. More importantly, based on this model an
incorrect conclusion may be drawn that a higher permeate flux (J)
occurs under an elevated concentration of the solute in the gel (Cg).
the ideal gas constant and T is the absolute temperature. Evidently,
this model cannot be applied to porous membranes such as
microfiltration, ultrafiltration and loose end nanofiltration systems
since osmotic pressure is negligible [25,33].
2.1.5. Resistance-in-series model
The resistance-in-series model states that the total resistance of a
ð6Þ
membrane consists of two parts, namely the resistance of a clean
membrane and the resistance of fouling layers. While the membrane
resistance is constant, the fouling layer resistance increased with time.
Other factors, such as membrane compaction, CP, and absorption may
also affect the flux or resistance of the membrane system. However,
these phenomena usually occur at much shorter time span than that
of membrane fouling. The effect of these factors on process
performance can be included by determining the membrane resis-
tance after a period of operation. This model can be applied to both
porous and non-porous membranes. The primary deficiency of this
model is that the existing model can only predict the fouling behavior
induced by feed water containing relatively simple foulants, such as
mono-dispersed colloids, calcium sulfate or calcium phosphate [34].
2.1.6. Theory of non-interacting particles
Song and Elimelech [25] developed a theory for CP based on the
hydrodynamics and thermodynamics of particle suspensions. In the
model they showed that in addition to mass balance (which is used in
gel layer model to identify the permeate velocity), another funda-
mental relationship (energy balance) is required to completely
describe CP in cross-flow filtration. The extent of CP and the behavior
of the permeate flux can be characterized by a dimensionless filtration
number (NF):

2.1.4. Osmotic pressure model

During reverse osmosis filtration, osmotic pressure may become
very high at the membrane surface where CP significantly increases
the wall (membrane) concentrations, especially when filtration is
performed at a high permeate flux with high rejection levels. The
relation among flux, hydraulic pressure difference and osmotic
pressure difference at the membrane surface is expressed as:
ΔP−Δπ
J=
μRT
where, ΔP is the applied filtration pressure, Δπ is the osmotic pressure
difference, RT is the total resistance, μ is the solution viscosity. The
osmotic pressure of an inorganic solute can be calculated as:
ni
Δπ = ∑ji RT
Vi
where ji is the factor for mole increase due to dissociation for solute i,
ni is the number of moles, Vi is the volume of the ideal solution, R is
4πa3p ðΔPp Þ
NF = ð9Þ
3KT
where ap is the particle radius, ΔPp is the pressure drop through the
accumulated particle layer, K is the Boltzman constant, T is the
absolute temperature. There is a critical value of the filtration number
for a given suspension and operational conditions. Song and Elimelech
ð7Þ [25] demonstrated that a cake layer will form between the CP layer
and the membrane surface when NF is greater than the critical value.
CP exists when NF is smaller than the critical value. This model can be
applied to both porous and non-porous membranes. Typical values of
NF for porous and non-porous membranes at different operating
conditions and cross-flow filtrations are shown in Table 4. Although
the average permeate velocity of uniform non-interacting spherical
ð8Þ
particles can be determined using this model, it cannot be used for a
multi-component system where organics and/or inorganics interact
with each other.
240 S. Shirazi et al. / Desalination 250 (2010) 236–248

Table 4 obtained from Deissler correlation:

Typical values of filtration number (NF) at different operating parameters [25].
0:875 0:25
Filtration system Pressure (KPa) Particle size (m) Filtration number Sh = 0:023 Re Sc ð14Þ
Reverse osmosis 103–4 × 103 3.6 × 10− 10 0.049–0.20
Ultrafiltration 102–103 10− 9–10− 7 0.10–106 Lee and Lee [16] expressed mass transfer coefficient (k) for different
Microfiltration <30 5 × 10− 8 – 10− 5 5 × 103–5 × 1010 flow regimes:
" #
2 0:33
VD
k = 1:86 for laminar flow ð15Þ
dh L
2.1.7. Cake-enhanced CP
Hoek and Elimelech [63] made a conceptual analysis of the solute " #
transport and CP phenomena in cross-flow membrane filtration. They V 0:8 D0:67
k = 0:023 for turbulent flow ð16Þ
showed if the CP layer was thick relative to the cross-flow channel d0:2
h v
0:47

height, mass transfer may be affected through both hindered

diffusivity of salt ions and altered tangential shear due to hydrody- where, L is the channel length, and V, the cross-flow velocity can be
namic drag caused by the stationary particles in the fluid. During cross- calculated as:
flow filtration, the bulk tangential flow velocity determines the wall
shear rate, which combines with the salt diffusion coefficient to Qf
V= ð17Þ
determine mass transfer in the salt CP layer. Tangential flow and salt Ac
ion back diffusion are hindered within the colloid deposit layer, thus where, Q f is the feed flow rate and Ac is the cross-sectional area of the
increasing the membrane surface salt concentration and transmem- membrane cell.
brane osmotic pressure. They derived cake-hindered mass transfer Sutzkover et al. [37] developed a model to estimate mass transfer
coefficient k′ from a one-dimensional convective-diffusive mass coefficient based on the permeate flux decline induced by addition of
balance, integrated separately across the cake layer and the salt CP a salt solution to an initially salt free water feed. Since the net pressure
layer above it: driving force is influenced by the level of the osmotic pressure
prevailing on the membrane surface, the magnitude of flux decline
" ! #−1
1 1 1 enables evaluation of membrane surface concentration Cm and hence
0
k = δc − + ð10Þ the determination of the mass transfer coefficient (k):
D* D k
ðJv Þsalt
k=    ð18Þ
where D is the diffusion coefficient of the salt in the bulk and D* is ΔP ðJ Þ
ln πb −πp 1− ðJ vÞ salt
diffusion coefficient of the salt in the hindered layer. The hindered v H2 O

diffusion coefficient can be expressed as [63]:

where (Jv)salt and (Jv)H2O are the permeate flux emanating from saline
water and salt-free water respectively. πb and πp are osmotic pressure
−1
D* = Dετ ð11Þ of the feed and permeate, respectively.
Sutzkover et al. [37] obtained the following mass transfer correlation
2
by performing experiments in a tubular RO system under turbulent flow
τ = 1–ln ðε Þ ð12Þ conditions (ranging the Reynolds number between 2,600 and 10,000):

where ε and τ are the cake porosity and tortuosity, respectively. Using kdh 0:91 0:25
Sh = = 0:020Re Sc ð19Þ
Eq. (10), Hoek and Elimelech [63] concluded that the salt mass D
transfer coefficient decreases with increasing cake thickness (or
This expression is practically identical with the theoretically predicted
mass) and decreasing cake porosity. This model is applied to describe
Deissler correlation [Eq. (14)]. However, if the feed is a supersaturated
both losses of flux and salt rejection accompanying colloidal fouling at
or near-saturated solution, flux could be declined due to both CP and
the surface of RO and NF membranes [35].
scale formation. Scaling may start instantaneously after completion of
CP. In that case, it is almost impossible to correctly identify (Jv)salt to
2.2. Estimating solute mass transfer coefficient in the CP layer
determine the value of k.

To determine the permeate flux through the CP layer, the mass

2.3. In-situ monitoring of CP
transfer coefficient must be determined. A number of formulations
have been developed for this purpose. The most widely used methods
During the past few decades, several in-situ monitoring techniques
are discussed below.
have been developed to better understand the physico-chemical
A generalized expression of the mass transfer coefficient in a fully
processes governing the development of a polarized layer of solutes
developed flow is of the form [36]:
near the membrane surface. Such techniques enable the testing of
theoretical models, and provide mechanistic information on the

 0 development of CP. Most of the studies of in-situ monitoring of CP
d V α  ν β
0
kdh α0 β0
Sh = =a h = aRe Sc ð13Þ were conducted in porous membranes during filtration of colloids
D ν D
because colloids are easier to monitor and detect than soluble salts.

where Sh is the Sherwood number, Re is the Reynolds number, SC is
the Schmidt number, dh is the hydraulic diameter of the feed channel,
V is the cross-flow velocity, ν is the kinematic viscosity, D is the solute
diffusivity in water, a is a numerical constant characterizing the flow
channel geometry, α′ and β′ are the exponents in the mass transfer
correlation. The values of a, α′ and β′ for turbulent flow can be
2.3.1. Light deflection technique
One important optical property of a solution is that its refractive
index changes with concentration. Thus, the change in deflection when
light passes through a solution provides information about the
concentration gradient along the light pathway. One of the most widely
used methods that make use of light deflection by the CP layer is
 S. Shirazi et al. / Desalination 250 (2010) 236–248 241

shadowgraphy. This technique is based on the principle that light is Table 5

deflected when passing through a medium of continuously varying Solubility product of the common inorganic salts that cause scaling on the membrane
surface [88].
refractive index, with the deflection patterns determined by the gradient
of the refractive index. Vilker et al. [38] utilized a shadowgraph optical Fouling salt Solubility product (25°C)
method to measure solute (bovine serum albumin, BSA) concentration CaCO3 2.8 × 10− 9
profiles in the polarization layer adjacent to an ultrafiltration membrane CaHPO4 1 × 10− 7
during dead-end filtration. During filtration, the induced CP layer created CaSO4 4.93 × 10− 5
Ca3(PO4)2 2.07 × 10− 29
a one-dimensional refractive index gradient above the membrane. The
MgCO3 · 3H2O 2.38 × 10− 6
linear correlation between the refractive index and the BSA concentra- Mg3(PO4)2 1.04 × 10− 24
tion was found to be valid up to 580 g/l, which set a limit for the AlPO4 9.83 × 10− 21
maximum measurable concentration. There are, however, certain Al(OH)3 1.3 × 10− 33
deficiencies for the shadowgraph technique. This technique does not Ca(OH)2 5.5 × 10− 6
CaHPO4 1.0 × 10− 7
consider polarization layer profile in the presence of gel/cake formation.
Fe(OH)3 2.79 × 10− 39
Moreover, this method can only measure concentration profiles beyond FePO4 · 2H2O 9.92 × 10− 29
200 µm above the membrane surface, due to deflection of light near the
membrane, where the boundary layer thickness for most of the inorganic
foulants are about 10 µm [14].
3.1. Reversibility of inorganic fouling

2.3.2. Radio isotope labeling Flux decline caused by membrane fouling may be generally classified
Radio-isotope-labeled macromolecules were first used to study CP as reversible or irreversible, depending on the effectiveness of the
by McDonogh et al. [39]. The accumulation of radio-labeled macro- fouling control and cleaning technology. For porous membranes,
molecules on the membrane surface due to CP was detected as an reversible fouling can be described as the portion that can be recovered
increase in voltage by a scintillation detector (a device for measuring the by backwashing/backflushing. However, backwashing or backflushing
number and energy of gamma rays). The detected signal is converted to is normally not available for non-porous membranes. The flux decline
a corresponding mass of the macromolecule above the membrane, so
that the development of CP can be monitored. However, the radio
isotope labeling technique described by McDonogh et al. [39] did not
provide quantitative measurements of the polarization layer thickness
or the concentration profile within it.

2.3.3. Electron diode array microscope (EDAM)

McDonogh et al. [39] presented a method for observing CP in ultra-
and microfiltration of BSA and dextran blue solutions using an EDAM. A
collimated, near infrared light parallel to the membrane surface, but
perpendicular to the flow direction, was channeled through the
filtration cell to a photodetector by microscopic lenses. Owing to
absorption of the incident light, the concentration gradient in the
polarization layer resulted in an intensity pattern on the detector, which
was monitored by an oscilloscope. EDAM technique can measure the CP
layer as close as 20 µm from the membrane surface. This method is thus
superior to shadowgraphy in terms of measurable range, as the later
technique can only measure 200 µm beyond the membrane surface.

3. Inorganic fouling

Membrane fouling is the accumulation of materials at the surface, or

in the pores of a membrane, which decreases the permeate flux of the
membrane [40]. The proceeding of fouling is perceived as a multistage
process, of which adhesion of the fouling agents to the membrane
surface is an essential step [41]. According to the definition of membrane
fouling, CP is not considered as fouling, although it is also responsible for
the flux decline. Because once the filtration process is stopped, the
phenomenon of CP disappears.
The term ‘mineral scale’ is used to differentiate fouling due to
inorganic salt deposits from organic fouling and biofouling. Presence of
high concentration of inorganic salts in the raw water is mainly
responsible for inorganic scaling. In NF and RO systems, the dissolved
salts are normally concentrated by 4–10 times, causing high concentra-
tions exceeding the solubility at the membrane surface [42]. CaSO4,
CaCO3, SiO2 and BaSO4 are some of the most common inorganic salts
responsible for scaling on the membrane surface. [14,42,43]. The salt
precipitates when the solubility product of the constituent ions is
reached or exceeded. Table 5 shows the solubility products of the Fig. 1. Schematics of reversible and irreversible fouling of porous membranes [1]:
common inorganic salts that cause scaling on the membrane surface. (a) reversible fouling, and (b) irreversible fouling.
242 S. Shirazi et al. / Desalination 250 (2010) 236–248
caused by irreversible fouling cannot be recovered unless the
membrane is replaced or cleaned by chemical reagents [43–45]. Fig. 1
(a and b) shows the filtration cycles of reversible and irreversible fouling
in porous membranes. Unlike reversible membrane fouling, irreversible
fouling normally caused by strongly adherent films or material trapped
within the porous substructure of the membrane [46]. Several methods
have been developed to quantify the resistance to filtration caused by
membrane fouling.
3.2. Resistance models used to quantify membrane fouling
Fouling increases resistance, which in turn reduces permeate flux.
Resistances that are responsible for decreasing flux are, membrane
resistance (Rm), CP resistance (Rcp), cake resistance (Rc), and pore-
blocking resistance (Rp). Therefore, total resistance during membrane
filtration can be expressed as:
RT = Rm + Rcp + Rc + Rp ð20Þ
The type of the membranes (i.e. porous versus non-porous) plays an
important role in determining the resistances for flux decline caused
by inorganic fouling. For example, all four types of resistance (Rm, Rcp,
Rc, and Rp,) could be responsible for flux decline for porous
membranes, while pore-blocking resistance (Rp) is not applicable
for non-porous membranes. Fig. 2 shows a schematic of major
resistances of a porous membrane during membrane filtration. Each
of the resistance terms is discussed below.
Fig. 2. Overview of the major resistances of a porous membrane during filtration.
3.2.1. Membrane resistance
Membrane resistance is caused by the membrane itself. Ideally, in
the absence of fouling and CP, only membrane resistance is involved
during filtration and can be calculated as [44]:
ΔP
Rm = ð21Þ
μJ0
where μ is the viscosity of the solution, ΔP, and J0 are the applied
filtration pressure and permeate flux of the clean membrane,
respectively. Membrane resistance is common for both porous and
non-porous membranes. The extent of resistance depends on the
membrane thickness and various morphological features such as the
tortuosity, porosity, and pore size distribution. For a membrane with
pores assumed to be cylindrical capillaries of uniform radius
perpendicular to the face of the membrane, the resistance can be
calculated using Hagen–Poiseuille equation:
8δm
Rm = ð22Þ
np πrp4
where np is the number of pores per unit membrane area, δm is the
membrane thickness, rp is the radius of the membrane pores. This
equation indicates that membrane resistance increases with increas-
ing the membrane thickness and decreases with increasing the pore
size and pore density.
 S. Shirazi et al. / Desalination 250 (2010) 236–248
3.2.2. Concentration polarization (CP) resistance
CP develops a highly concentrated layer near the membrane surface,
which exerts a resistance towards mass transfer, i.e. the CP resistance.
Due to CP, the accumulated solute and particle concentrations become
so high that a scale or cake layer can be formed near the membrane
surface, which exerts the cake resistance. Resistance exerted by CP
triggers fouling on the membrane surface. It is important to distinguish
between CP and fouling resistance, although both are not completely
independent of each other [47–49]. CP of particles is common for both
porous and non-porous membranes. However, CP of soluble ions is
applicable for non-porous membrane only.
3.2.3. Cake resistance
During membrane fouling, foulants (the materials that cause fouling)
produce a cake layer at the membrane surface, which generates
resistance to permeate transportation and causes permeate flux decline.
Cake resistance exists for both porous and non-porous membranes.
Different particles form different types of cakes; such as, particles
of metal hydroxides (e.g. ferric hydroxide, cupric hydroxide, alumi-
num hydroxide) produce cakes that are readily deformed or rear-
ranged under pressure (compressible cakes). Crystals of carbonates of
calcium and sodium produce cakes that are not readily deformed or
rearranged under pressure (incompressible cakes) [50].
Incompressible cake resistance is often estimated by the Carman–
Kozeny equation:
K 0 ð1−εc Þ2 S2s δc
Rc = ð23Þ
ε3c
where K′ is a constant, δc is the cake thickness, εc is the void fraction
of the cake, and Ss is the solids surface area per unit volume of solids in
the cake. For rigid spherical particles of radius r, the specific surface
area is Ss = 3/r, the void fraction εc of a randomly packed cake is
approximate 0.4, and the constant K′ reported by Grace [51] is 5.
Noting that the cake resistance is proportional to its thickness, a
specific cake resistance per unit thickness (α̂) is defined as:
^ = Rc
α ð24Þ
δc
Compressible cakes exhibit a decrease in void volume and an
increase in the specific resistance as the compressive pressure is
increased. For compressible cakes, specific cake resistance (α) is a
function of particle diameter (dp), porosity of cake (ε), and particle
density (ρ). A well-established empirical relationship for α is
Carman's equation [52]:
180ð1−εÞ
α= ð25Þ
ρd2p ε3
According to Eq. (25), the smaller the floc size, the greater the cake
resistance is. The typical range of specific cake resistance varies from
0.27 to 2.7 × 103 m/kg [47].
3.2.4. Pore-blocking resistance
For porous membranes, if the size of the foulant is less than the
pore size of the membrane, the foulant may completely or partially
block the membrane pores by adsorption onto the inner wall of the
membrane pores. The resistance caused by pore blocking is referred to
pore blocking resistance (Rp in Fig. 2).
3.3. Mechanism of inorganic fouling
Mechanism of scale growth on the membrane surface is regarded as
one of the key-problems and its solution provides a guideline for
243
membrane operation. Two major mechanisms, crystallization and
particulate fouling, play important roles during scale formation on
membrane surface. Both crystallization and particulate fouling are direct
results of inverse solubility of salts [53]. During crystallization, deposition
at the membrane surface occurs due to precipitation of ions; whereas
during particulate fouling, deposition occurs due to convective transpor-
tation of colloidal particulate matter from the bulk solution to the
membrane surface [16]. For non-porous membranes, particulate fouling is
caused by the accumulation of particles on the membrane surface
resulting in the formation of cake layer, which creates a hydraulic
resistance to permeate flow through the membrane [54]. For porous
membranes, pore plugging by particulate matters smaller than the
membrane pore size is an important mechanism for particulate fouling, in
addition to the formation of cake on the membrane surface [54].
Most of earlier studies used CaSO4 and CaCO3 as the fouling salts to
determine the mechanism of scale deposition on the membrane surface
(e.g., [19,55]). These studies showed that in contrast to CaCO3 scaling, for
which surface crystallization is the dominant mechanism [53,56], CaSO4
deposition on the membrane surface starts after an induction period
characterized as a “delayed” period before flux decline occurs [15,16].
Lin et al. [43] proposed a four-stage mechanistic model to describe the
permeate flux decline process during CaSO4 scaling in NF. The length of
each stage is dependent on the operating parameters (such as cross-
flow velocity, operating pressure, feed concentration etc.). First, the flux
was reduced due to concentration polarization. At the second stage, flux
was not reduced; instead, nucleation of CaSO4 occurred. The major
permeate flux decline occurred at the third stage due to CaSO4 cake
formation. At the final stage, the system reached the steady state, where
the rate of CaSO4 deposition on the membrane was balanced by shearing
caused by the increase of crossflow velocity. Beyond this stage, the flux
did not decrease significantly.
Hasson et al. [57] developed an ionic diffusion model based purely on
precipitation which accounted for water chemistry for calcium
carbonate scaling. The “diffusion model” assumes that (1) the diffusion
resistance is much greater than the surface reaction resistance, (2) a
significant concentration gradient exists between the bulk and the solid
surface, and (3) water at the solid surface is saturated with the solid.
Okazaki and Kimura [58] also emphasized the role of precipitation
kinetics in studying the fouling of CaSO4. They used basic concepts from
nucleation and crystallization kinetics in conjunction with a cake
filtration model to interpret CaSO4 flux decline from a tubular RO
membrane under turbulent conditions. However, the model was
theoretically developed without any “experimental evidence” [59]. In
a later study, Gilron and Hasson [60] examined the cake filtration fouling
model for RO and concluded that the model was physically unrealistic
for flux decline.
Lee et al. [14] combined flux decline model by Gilron and Hasson [60]
with Okazaki and Kimura's [58] cake filtration model to describe the
scale formation mechanisms in NF membranes. They found that during
cross-flow filtration both bulk and surface crystallization were
responsible for flux decline during inorganic scaling on NF. Lee and
Lee [16] described that during inorganic scaling of NF membrane, cake
layer can be formed in one of three ways: by bulk crystallization
(homogeneous), surface crystallization (heterogeneous), or simulta-
neous bulk and surface crystallization. They suggested that the
crystallization process might be affected by operating conditions, and
concluded that surface crystallization is favored at low cross-flow
velocity and high operating pressure while bulk crystallization is
favored at high cross-flow velocity and intermediate to high operating
pressure.
Bulk crystallization is the phenomenon that crystal particles are
formed in the bulk phase through homogeneous crystallization then
deposit on membrane surface as particles. In the bulk phase,
supersaturated solute gives rise to agglomeration of scale-forming
ions due to random collision of ions in motion; or the secondary
crystallization occurs on the surface of foreign particles present in the
244 S. Shirazi et al. / Desalination 250 (2010) 236–248
bulk [58,59]. The cluster of ions coalesces to form crystal. After growing
above the critical size, precipitation occurs. The flux decline due to bulk
crystallization can be expressed as [14]:
ΔP−Δπ
Jb = ð26Þ
μðRm + Rc Þ
where Jb is the permeate flux estimated from bulk crystallization, ΔP
is applied pressure, Δπ is osmotic pressure, Rm and Rc are the
resistance due to the membrane and the cake, respectively.
Surface crystallization produces solid crystals directly on mem-
brane surface. The scale formation occurs on the membrane surface
through the lateral growth of crystals. As active sites are present on
the membrane surface, nucleation originates on the membrane active
sites and grows further. The flux decline due to surface crystallization
can be expressed as follows [14]:
ΔP−Δπ At −Ab
Js = ð27Þ
μRm At
where Js is the flux due to surface crystallization, At is the total
membrane area, Ab is the membrane area occupied by surface crystals
estimated as [61]:
Ab = βms ð28Þ
where β is the area occupied per unit mass; and ms is the weight of
the scale formed directly on the membrane surface.
Fig. 3 depicts both bulk and surface crystallization. When the bulk
phase becomes supersaturated, both mechanisms of crystallization
simultaneously proceed. Under such condition, the permeate flux can
be expressed as [14]:
ΔP−Δπ At −Ab
J= ð29Þ
μðRm + Rc Þ At
Although in the past several decades extensive studies have been
performed to investigate the factors (including operating parameters
and membrane properties) that promote bulk crystallization or
surface crystallization, there is not much information regarding the
properties of the cake (such as density, thickness, compactness etc.)
formed by these crystals. As bulk crystals are formed in the bulk rather
than on the surface, there could be some differences of the properties
Fig. 3. Schematic representation of inorganic scaling mechanism.
between the cake formed by bulk crystals and surface crystals.
Therefore, the resistance produced by bulk crystals may not be the
same as the resistance by surface crystals.
The crystallization and deposition of solids on the membrane
surface can be affected by a number of physical and chemical
parameters.
a) Surface roughness. It is well known that polished surface has fewer
crystals at the end of a crystallization experiment compared to a
rough surface. This is because an increase in surface roughness
increases surface free energy, contributes to an increase in
adhesiveness, and enhances the trapping of crystal [62]. Hoek et
al. [63] measured the surface roughness of four types of RO/NF
membranes and showed surface roughness of a membrane can
affect the salt deposition as well as flux decline during membrane
fouling. Rough membrane surfaces are more prone to particle
deposition compared to a smooth surface. AFM images of their
study revealed that particles preferentially accumulate in the
valleys of rough membranes, leading to “valley clogging” and
causing more severe flux declines [64].
b) Supersaturation. The primary cause of inorganic scaling is
supersaturation. When the solubility of a salt is exceeded, the
salt precipitates and forms scale. Increasing supersaturation
accelerates the nucleation rate [65–67]. If the region of supersat-
uration is at the membrane–liquid interface, precipitation on the
membrane surface is likely. However, if the supersaturation region
is away from the membrane surface, crystals form in the bulk and
migrate to the surface as particles to form a solid deposit [68].
There is usually a period of time which elapses between the
achievement of supersaturation and the appearance of the first
crystals, which may be referred to as the “induction” or the
“initiation” period. The magnitude of the induction period is
governed by the degree of supersaturation, the temperature level,
and turbulence [68].
c) Shear rate. Bulk or surface crystallization will dominate under
certain velocity regimes of system operation. As the velocity
increases, the probability for aggregation at the membrane surface
decreases, because the CP phenomenon is less significant at a high
crossflow velocity. In addition, increasing cross-flow velocity
increases shear or drag across the membrane surface and thus
causes greater back diffusion and transportation of the foulants,
and thereby reduces membrane fouling.
 S. Shirazi et al. / Desalination 250 (2010) 236–248 245

d) Transmembrane pressure. Increasing transmembrane pressure
reduces the effect of shear stress caused by cross-flow, mainly
due to the relatively greater convective flux across the membrane,
and thus increases the possibility of surface crystallization [69].
Higher transmembrane pressure causes rapid permeate flux
decline and membrane fouling. The rapid flux variation produces
changing conditions with respect to the concentration and the
solubility in the boundary layer throughout a filtration [70].
Therefore, it is necessary to control transmembrane pressure and
boundary conditions to avoid or minimize the fouling of
membranes when the “critical flux” is exceeded.
In membrane filtration, critical flux is defined as the flux below
which no membrane fouling occurs [71]. Bacchin et al. [72]
developed a theoretical model that describes solid deposition on
the membrane surface. A mass transfer equation links the
deposition rate to hydrodynamic conditions (permeation and
tangential flow through a film thickness, δ) and to physicochem-
ical properties of the suspension (diffusion, D, and potential
barrier between particles, VB). This equation predicts the existence
of a critical flux, JCrit, for porous and non-porous filtration of large-
sized colloids.
D VB
JCrit = ln
δ δ
e) Solution chemistry (ionic composition in the solution). Solution
chemistry has an important effect on the surface charge char-
acteristics of polymeric surfaces. Taborek et al. [73] illustrated the
effect of crystallization from solutions containing high concentra-
tion of large variety of salts. Each of the salts exhibits different
crystalline formations, and consequently, crystalline clusters build
up in irregular patterns, forming cavities between them, which
permits deposition of suspended particles, thus further decreasing
crystalline cohesion. In the presence of an indifferent (1:1)
electrolyte, all polymer surfaces (including membrane) are
positively charged in the lower pH range, and negatively charged
in the higher pH range [74]. Therefore, solution pH, ionic strength,
and membrane surface charge affects the electrostatic interactions
between the membrane and the foulants. As a result, insignificant
membrane fouling was observed when charge repulsion occurred
between the membrane and the foulants at a low ionic strength
[75]. For the effect of ionic strength, because the same type of
particles normally has the same characteristics of charge, the
distance between each particle in the cake layer is shorter at high
ionic strength due to the charge screening of the repulsive forces
among particles. As a result, the cake formed in a high ionic
strength solution is more compact and has higher resistance [76].
Therefore, the permeate flux drops faster and reaches the steady-
state flux faster at high ionic strength than at low ionic strength.
f) Particle size and concentration. Increasing particle size reduced
fouling during membrane filtration [77–79]. Larger particles are
sterically hindered from depositing onto the membrane surface or
inside the membrane pores, and they are more easily swept away
by axial forces. A higher feed concentration brings greater
convective flux of particles onto the membrane surface. Conse-
quently, more significant membrane fouling was observed with
particles of smaller sizes and higher concentrations [80].
g) Surface and bulk temperatures. Few studies have been conducted to
investigate the effect of bulk and surface temperatures on the
inorganic fouling of membranes. Temperature affects the viscosity
of water, the solubility and the mass transfer coefficient of
inorganic salts or particles. The viscosity of water decreases with
temperature. As a result, the permeate flux increases in higher
temperature, and thus the convective transportation of foulants to
the membrane is more significant. Consequently, membrane
fouling is expected to be severe. However, the mass transfer
coefficient also increases with temperature. Therefore, back
diffusion of CP and the fouling layer to the bulk solution should
be faster in high temperature. In addition, temperature affects the
solubility of salts and their degree of supersaturation as well. More
studies need to be conducted to determine the temperature effect
on membrane fouling.
h) Membrane hydrophilicity. Hydrophilicity allows a surface to be
wetted forming a water film or coating. Hydrophilic materials
possess an ability to form hydrogen bonds with water. The greater
ð30Þ
the tendency for a material to associate with water through
hydrogen bonding, the more hydrophilic the material is. As a
general hypothesis, hydrophobic particles are expected to have a
higher fouling potential to hydrophobic membranes. For both
hydrophilic particles and membranes, their interactions likely
depend on both electrostatic and hydrophobic interactions.
3.4. Visual confirmation of fouling mechanism
Membrane fouling is generally associated with a cake or gel
formation on the membrane surface, or blocking membrane pores by
macromolecules, colloids or particulate matters. At present, the
mechanisms of cake formation and fouling are not yet well
understood. Thus, in-situ measurements of fouling and direct
observation of cake layer formation are of paramount importance in
efforts to understand the fundamental processes governing mem-
brane fouling. Table 6 shows the applications and principles of
different methods of visual observation of fouling. In this section, the
methods of membrane surface observation are reviewed.
3.4.1. Direct observation of the membrane
The most straightforward method to directly observe particle
deposition is visualization by an optical microscope. One of the first
applications of this technique was developed by Li et al. [78] and was
termed direct observation through membrane (DOTM). In this method,
a microscope objective was positioned at the permeate side of a
transparent membrane and particle deposition was observed in real

Table 6
Different methods of fouled membrane observation.

Method Principle Application

Direct observation of A microscope objective is positioned at the permeate side of a transparent membrane to observe particle To directly observe particle deposition by an
membrane [78] deposition in real time by microscope. optical microscope
Optical laser sensor [81] The formation of deposit layer absorbs lights from a bypassing laser beam. The variation of the signal To investigate the thickness of cake layer
intensity after the laser beam traversed through the cake layer corresponds to the deposit layer during microfiltration
thickness.
Ultrasonic time-domain This technique uses sound waves to measure the location of a moving or stationary interface and can To investigate in-situ measurement of
reflectometry [82] provide information on the physical characteristics of the media through which the waves travel. membrane fouling
Electrical impedance An alternating current is injected directly into the membrane. Capacitance dispersion changes are To characterize membrane properties and
spectroscopy [84,85] measured to monitor in-situ accumulation of particulates. to investigate membrane fouling
Scanned electron SEM shows 3D images of cake and membrane at much higher magnification. To investigate the membrane surface and
microscopy (SEM) fouling
246 S. Shirazi et al. / Desalination 250 (2010) 236–248
time by the microscope. A key to this technique is the use of an
appropriate membrane. Li et al. [78] used an Anopore (Whatman, UK)
anodized aluminum membrane with high porosity (60%) and straight
through pores. When wet, the membrane is transparent, which
facilitates the observation of the membrane surface from the permeate
side. The membrane module was fabricated to allow the transmission of
light from the feed side. Images of particle deposition were recorded by a
video camera. The technique showed that the motion of particles near
the membrane surface subsequent to deposition depends on the cross-
flow velocity, the particle size and the particle size distribution.
Adjustment of operating variables including permeate flux and cross-
flow induces particle motion along the membrane surface. The DOTM
technique can also yield the size distribution of the deposited particles.
Comparison of particle size distributions of the deposited particles
under various operating conditions showed that under high cross-flow
velocity, the deposited particles were smaller than those deposited
under low cross-flow conditions. This observation supports the theory
that cross-flow shear forces deter the deposition of large particles and
thus the cake layer that formed under high cross-flow conditions has a
compact structure consisting of small particles. The major drawbacks of
the DOTM technique are the need to use a transparent membrane and
the positioning of the microscope objective below the membrane at the
permeate side. The first requirement is a major disadvantage as it
confines the use of the technique to only a limited number of inorganic
porous membranes, mostly microfiltration membranes. The second
limitation does not allow the observation of particle accumulation
beyond a monolayer as observation through the deposited particle layer
is not readily attained.
3.4.2. Optical laser sensor
Hamachi and Mietton-Peuchot [81] developed an optical laser
sensor method for investigating the thickness of the cake layer during
microfiltration of bentonite suspensions. The principle of this
technique is that the formation of the fouling layer absorbs light
from a bypassing laser beam. The variation of the signal intensity
detected after the laser beam has traversed through the cake layer
corresponds to the deposit thickness. A calibration procedure with
known values of cake layer thickness can be performed to extract the
correlation. The cake thickness was shown to increase with increasing
transmembrane pressure, increasing bulk concentration and decreas-
ing cross-flow velocity. Thicker cake layers were formed with more
concentrated bulk suspensions for the same volume of filtrate. Further
analysis was performed by using Darcy′s law to calculate cake layer
resistance. The specific cake layer resistance of the deposit was shown
to be higher when filtering more dilute suspensions. This might be
due to the fact that dilute particle suspensions contain fewer large
particles and thus form less permeable deposits. In other words,
particle aggregation may occur in concentrated suspensions. At a
given cake thickness, the resistance of the cake layer was also found to
increase with increasing transmembrane pressure, which was
attributed to cake compaction. However, the maximum concentration
that can be filtered in the study described above was 375 mg/l, above
which the turbidity of the suspension obstructs the photo detection.
3.4.3. Ultrasonic time-domain reflectometry (UTDR)
Mairal et al. [82] used the UTDR technique for in-situ measurement
of fouling during reverse osmosis of CaSO4 solutions. UTDR is a
technique that uses sound waves to measure the location of a moving
or stationary interface. It provides information on the physical
characteristics of the media through which the waves travel. When
ultrasonic waves encounter an interface between two media, energy
will generally be partitioned such that a reflected wave occurs and is
detected by an ultrasonic transducer. The amplitude of the reflected
wave relative to the incident wave is determined by the difference of
the acoustic impedance between the two media as well as the
topography of the interface. The time interval between the initiation
of the incident wave and the detection of the reflected wave is used
for measuring the distance between interfaces. Li and Sanderson [83]
performed an experiment to determine the cake layer thickness on
the membrane surface using UTDR technique. A comparison of UTDR
measurements with post experimental SEM images affirmed that the
realized signal differences did reflect the formation of an appreciable
deposit layer.
3.4.4. Electrical impedance spectroscopy (EIS)
Chilcott et al. [84] and Gaedt et al. [85] proposed the use of electrical
impedance spectroscopy (EIS) for characterizing membrane properties
and investigating membrane fouling. In this technique, an alternating
current is injected directly into the membrane via external electrical
contacts with the edges of the membrane. A metal layer sputtered onto
the membrane surface is used to enhance conduction properties. The
flow of the current across the membrane surface leads to dispersion of
the current into the bulk solution and the membrane pores. This
dispersion phenomenon is characterized by the capacitance and
conductance of various components of the system, such as the
membrane material and the bulk solution, and possible polarization of
the fouling layer. The dispersion of the current changes as foulants
accumulate on the membrane surface leading to alteration of the
capacitive and conductive properties of the membrane interfacial
region. Measuring the changes in the capacitance dispersion of the
system therefore becomes a means of monitoring in-situ accumulation
of particulates that could potentially foul the membrane. One major
limitation of the proposed method is that it requires the surface of the
membrane to be coated with thin metal films. The coating of the
membrane surface not only departs from a true representation of the
system but may also occlude membrane pores and alter the experi-
mental conditions. Thus, while the technique may be used to evaluate
properties of fouled membranes, its application for in-situ observation
of the dynamics of fouling behavior is questionable.
3.4.5. Scanning electron microscopy (SEM)
SEM provides topographical information at magnifications of 10–
100,000 times (Photometrics, 2001). SEM creates the magnified
images by using electrons instead of light waves, thus the SEM shows
very detailed images at much higher magnifications than is possible
with an optical microscope. The images created without light waves
are rendered black and white.
For a SEM examination, samples have to be prepared carefully to
withstand the vacuum environment of the instrument. SEM samples are
coated with a very thin layer of carbon or gold by a sputter coater. The
sample is placed inside the vacuum column through an air-tight door.
After the air is pumped out of the column, an electron gun emits a beam
of high energy electrons. This beam travels downward through a series
of magnetic lenses designed to focus the electrons to a very fine spot.
Near the bottom, a set of scanning coils moves the focused beam back
and forth across the specimen, row by row. As the electron beam hits
each spot on the sample, secondary electrons are scattered from its
surface. A detector counts these electrons and sends the signals to an
amplifier. The final image is build up from the number of electrons
emitted from each spot on the sample. One drawback of SEM technique
is that it requires the sample to be partially or totally dehydrated. The
consequence of dehydration may alter the original structure and the
morphology of the foulant layer. In addition, the highly energetic
electron beam may damage some kinds of organic membrane.
4. Conclusions and future research needs
This paper summarizes the concentration polarization, reversible and
irreversible fouling phenomena of membranes. Different models of CP and
inorganic fouling, as well as the factors that affect CP and inorganic fouling
are also discussed. Most of the studies on inorganic fouling mechanisms
were performed to investigate the effect of operating parameters (such as
 S. Shirazi et al. / Desalination 250 (2010) 236–248
shear rate, and operating pressure) and membrane properties (e.g.
porosity, roughness, etc.) on the crystallization of inorganic salts [86].
However, the precipitation of inorganic salts can also be affected by
module geometry and membrane materials [87], and there is limited
information on the effect of these factors on fouling mechanisms. In
addition, there is not much information available to describe inorganic
fouling on the membrane surface in a multi-component system. Given
that multiple fouling ions are usually present in raw waters, study of
inorganic fouling in a multi-component system is needed to reveal the
interactions between salts/ions and the membranes during scale
formation. Moreover, it will be important to observe if a multi-component
system has any effect on crystallization and deposition mechanisms of
inorganic fouling on membranes.
Acknowledgements
The study is sponsored in parts by the Texas Hazardous Waste
Research Center (Project No.: 068LUB0966 and 069LUB2966) and the
Sustainable Agricultural Water Conservation (SAWC) program of the
United States Department of Agriculture through a subcontract from
Sul Ross State University. The funding support is greatly appreciated.
References
[1] M. Mulder, Basic Principles of Membrane Technology, Kluwer Academic Publish-
ers, Boston, 1996.
[2] J.C. Crittenden, R.R. Trussell, D.W. Hand, K.J. Howe, G. Tchobanoglous, Water
Treatment: Principles and Design, MWH, 2nd edition, John Wiley and Sons,
Hoboken, New Jersey, 2005, p. 958, (Revised).
[3] U. Merten, Desalination by Reverse Osmosis, MIT Press, Cambridge, MA, 1966.
[4] S. Alami-Younsii, A. Larbot, M. Persin, J. Sarrazin, L. Cot, Rejection of mineral salts
on a gamma alumina nanofiltration membrane: application to environmental
process, Journal of Membrane Science 102 (1995) 123–129.
[5] N. Saffaj, H. Loukili, A. Alami-Younssi, A. Albizane, M. Bouhria, M. Persin, A. Larbot,
Filtration of solution containing heavy metals and dyes by means of ultrafiltration
membranes deposited on support made of Morrocan clay, Desalination 168 (2004)
301–306.
[6] R.W. Bowen, F. Jenner, Theoretical description of membrane filtration of colloids
and fine particles; an assessment and review, Advances in Colloid and Interface
Sciences 56 (1995) 141.
[7] J.G. Jacangelo, S. Chellam, R.R. Trussell, The membrane treatment, Civil Engineering 68
(9) (1998) 42–45.
[8] R.R. Sharma, R. Agrawal, S. Chellam, Temperature effects on sieving characteristics
of thin-film composite nanofiltration membranes: pore size distributions and
transport parameters, Journal of Membrane Science 223 (2003) 69–87.
[9] M.R. Wiesner, S. Chellam, The promise of membrane technology, Environmental
Science and Technology 33 (17) (1999) 360A–366A.
[10] D. Bessarabov, Z. Twardiwski, Industrial application of nanofiltration-new
perspectives, MembraneTechnology September (2002) 6–9.
[11] M. Thanuttanavong, J. Oh, K. Yamamoto, T. Urase, Comparison between rejection
characteristics of natural organic matter and inorganic salts in ultra low pressure
nanofiltration for drinking water production, Water Science and Technology:
Water Supply 1 (2001) 77–90.
[12] S. Benfer, U. Popp, H. Richter, C. Siewert, G. Tomandl, Development and
characterization of ceramic nanofiltration membranes, Separation and Purifica-
tion Technology 22–23 (2001) 231–237.
[13] T. Tsuru, T. Sudou, S. Kawahara, T. Yoshioka, M. Asaeda, Permeation of liquids
through inorganic nanofiltration membranes, Journal of Colloid and Interface
Science 228 (2000) 292–296.
[14] S. Lee, C. Lee, J. Kim, Analysis of CaSO4 scale formation mechanism in various
nanofiltration modules, Journal of Membrane Science 163 (1999) 63–74.
[15] S. Bhattacharjee, G. Johnston, A model of membrane fouling by salt precipitation
from multicomponent ionic mixtures in cross-flow nanofiltration, Environmental
Engineering Science 19 (2002) 399–412.
[16] S. Lee, C. Lee, Effect of operating conditions on CaSO4 scale formation mechanism
in nanaofiltration for water softening, Water Research 34 (2000) 3854–3866.
[17] A. Seidel, M. Elimelech, Coupling between chemical and physical interactions in
natural organic matter (NOM) fouling of nanofiltration membranes: implication
for fouling control, Journal of Membrane Science 203 (2002) 245–255.
[18] R. Sheikholeslami, Fouling mitigation in membrane processes, Desalination 123
(1999) 45–53.
[19] Y.A. Le Gouellec, M. Elimelech, Calcium sulfate (gypsum) scaling in nanofiltration
of agricultural drainage water, Journal of Membrane Science 205 (2002) 279–291.
[20] G. Belfort, R.H. Davis, A.L. Zydney, The behavior of suspensions and macromolecular
solutions in cross-flow microfiltration, Journal of Membrane Science 96 (1994) 1–58.
[21] L.J. Zeman, A.L. Zydney (Eds.), Microfiltration and Ultrafiltration: Principles and
Applications, Marcel Dekker, New York, 1996.
247
[22] H.R. Rabie, P. Cote, N. Adams, A method for assessing membrane fouling in pilot
and full scale systems, Desalination 141 (2001) 237–243.
[23] H.F. Shaalan, Development of fouling control strategies pertinent to nanofiltration
membranes, Desalination 153 (2002) 125–131.
[24] S.H. Shirazi, C.J. Lin, S. Doshi, S. Agarwal, P. Rao, Comparison of fouling mechanism
by CaSO4 and CaHPO4 on nanofiltration membranes, Separation Science &
Technology 41 (2006) 2861–2882.
[25] L. Song, M. Elimelech, Theory of concentration polarization in cross-flow filtration,
Journal of Chemical Society 91 (1995) 3389–3398.
[26] T. Matsuura, Synthetic membranes and membrane separation processes, CRC
Press, Boca Raton, 1994.
[27] J.G. Wijmans, R.W. Baker, The solution-diffusion model — a review, Journal of
Membrane Science 107 (1995) 1–21.
[28] E.M.V. Hoek, A.S. Kim, M. Elimelech, Influence of cross-flow membrane filter
geometry and shear rate on colloidal fouling in reverse osmosis and nanofiltration
separations, Environmental Engineering Science 19 (6) (2002) 357–372.
[29] E. Caldin (Ed.), Mechanisms of Fast Reaction in Solution, IOS Press, Amsterdam,
The Netherlands, 2001, pp. 51–52.
[30] K.S. Spiegler, O. Kedem, Thermodynamics of hyperfiltration (reverse osmosis):
criteria for efficient membranes, Desalination 1 (1966) 311–326.
[31] H. Al-Zoubi, N. Hilal, N.A. Darwish, A.W. Mohammad, Rejection and modelling of
sulphate and potassium salts by nanofiltration membranes: neural network and
Spiegler–Kedem model, Desalination 206 (2007) 42–60.
[32] J. Schaep, B. Van der Bruggen, C. Vandecasteele, D. Wilms, Influence of ion size and
charge in nanofiltration, Separation and Purification Technology 14 (1998)
155–162.
[33] A. Fane, T.D. Waite, A.I. Schafer (Eds.), Nanofiltration Principles and Applications,
Elsevier Publications, 2004, pp. 169–240.
[34] K.L. Chen, L. Song, S.L. Ong, W.J. Ng, The development of membrane fouling in full-
scale RO processes, Journal of Membrane Science 232 (2004) 63–72.
[35] E.M.V. Hoek, M. Elimelech, Cake-enhanced concentration polarization: a new
fouling Mechanism for salt rejecting membranes, Environmental Science and
Technology 37 (2003) 5581–5588.
[36] E. Matthiasson, B. Sivik, Concentration polarization and fouling, Desalination 35
(1980) 59–103.
[37] I. Sutzkover, D. Hasson, R. Semiat, Simple technique for measuring the
concentration polarization level in a reverse osmosis system, Desalination 131
(2000) 117–127.
[38] V.L. Vilker, C.K. Colton, K.A. Smith, Concentration polarization in protein
ultrafiltration. Part II: Theoretical and experimental study of albumin ultrafiltered
in an unstirred batch cell, AIChE Journal 27 (1981) 637–645.
[39] R.M. McDonogh, H. Bauser, N. Stroh, Grauschopf, Experimental in situ measure-
ment of concentration polarization during ultra- and micro-filtration of bovine
serum albumin and dextran blue solutions, Journal of Membrane Science 104
(1995) 51–63.
[40] T.F. Speth, R.S. Summers, A.M. Gusses, Nanofiltration foulants from a treated
surface water, Environmental Science and Technology 32 (1998) 3612–3617.
[41] R. Oliviera, Understanding adhesions: a means for preventing fouling, Experi-
mental Thermal and Fluid Science 14 (1997) 316–322.
[42] C.A.C. Van de Lisdonk, J.A.M. Van Paassen, J.C. Schippers, Monitoring scaling in
nanofiltration and reverse osmosis membrane systems, Desalination 132 (2000)
101–108.
[43] C.J. Lin, S. Shirazi, P. Rao, Mechanistic model for CaSO4 fouling on nanofiltration
membrane, Journal of Environmental Engineering 131 (10) (2005) 1387–1392.
[44] S.R. Fabish, Y. Cohen, Fouling and rejection behavior of ceramic and polymer-
modified ceramic membranes for ultrafiltration of oil-in-water emulsions and
microemulsions, Colloids and Surfaces A: Physicochemical and Engineering
Aspects 191 (2001) 27–40.
[45] B. Van der Bruggen, C. Vandecasteele, Modelling of the retention of uncharged
molecules with nanofiltration, Water Research 36 (2002) 1360–1368.
[46] R.G. Gutman, Membrane filtration: the technology of pressure-driven crossflow
processes, Adam Hilger, Bristol, England, 1987 Chapter 4.
[47] L.F. Song, Flux decline in cross-flow microfiltration and ultrafiltration- mechanisms and
modeling of membrane fouling, Journal of Membrane Science 139 (1998) 183–200.
[48] G.B. van den Berg, C.A. Smolders, Flux decline in ultrafiltration processes,
Desalination 77 (1990) 101–133.
[49] J.G. Wijmans, S. Nakao, C.A. Smolders, Flux limitation in ultrafiltration-osmotic pressure
model and gel layer model, Journal of Membrane Science 20 (1984) 115–124.
[50] N.P. Cheremisinoff, Liquid Filtration Elsevier, 1998, pp. 59–87.
[51] H.P. Grace, Resistance and compressibility of filter cake, Chemical Engineering
Progress 49 (1953) 303–318.
[52] P.C. Carman, Fundamental principles of industrial filtration, Transactions of the
Institution of Chemical Engineers 16 (1938) 168–176.
[53] R. Sheikholeslami, Calcium sulfate fouling — precipitation or particulate: a
proposed composite model, Heat Transfer Engineering 21 (2000) 24–33.
[54] X. Zhu, M. Elimelech, Colloidal fouling of reverse osmosis membranes: measure-
ments and fouling mechanisms, Environmental Science and Technology 31 (1997)
3654–3662.
[55] D. Hasson, A. Drak, R. Semiat, Inspection of CaSO4 scaling on RO membranes at
various water recovery levels, Desalination 139 (2001) 73–81.
[56] P. Dydo, M. Turek, J. Ciba, Scaling analysis of nanofiltration systems fed with
saturated calcium sulfate solutions in the presence of calcium carbonate ions,
Desalination 159 (2003) 245–251.
[57] D. Hasson, H. Sheramn, M. Biton, Prediction of calcium carbonate scaling reates,
Proceedings of 6th International Symposium on Fresh Water Flow From the Sea,
vol. 2, 1978.
248 S. Shirazi et al. / Desalination 250 (2010) 236–248
[58] M. Okazaki, S. Kimura, Scale formation on reverse osmosis membranes, Journal of
Chemical Engineering Japan 17 (1984) 145–151.
[59] A.G. Pervov, Scale formation prognosis and cleaning procedure schedules in
reverse osmosis systems operation, Desalination 83 (1991) 77–118.
[60] J. Gilron, D. Hasson, Analysis of laminar precipitation fouling on RO membranes,
Desalination 60 (1986) 9–24.
[61] J. Gilron, D. Hasson, Calcium sulphate fouling of reverse osmosis membranes: Flux
decline mechanism, Chemical Engineering Science 42 (1987) 2351–2360.
[62] B. Larsson, Ultrafiltration membranes and applications, Plenum Press, New York,
1980.
[63] E.M.V. Hoek, S. Bhattacharjee, M. Elimelech, Effect of membrane surface roughness
on colloid-membrane DLVO interactions, Langmuir 19 (2003) 4836–4847.
[64] M. Elimelech, X. Zhu, A.E. Childress, S. Hong, Role of membrane surface
morphology in colloidal fouling of cellulose acetate and composite aromtic
polyamide reverse osmosis membranes, Journal of Membrane Science 127 (1997)
101–109.
[65] M. Brusilovsky, J. Borden, D. Hasson, Flux decline due to gypsum precipitation on
RO membranes, Desalination 86 (1992) 187.
[66] N. Epstein, Thinking about heat transfer fouling: a 5 × 5 matrix, Heat Transfer
Engineering 4 (1997) 43–56.
[67] D. Hasson, D. Bramson, B. Limoni-Relis, R. Semiat, Influence of the flow system on the
inhibitory action of CaCO3 scale prevention additives, Desalination 108 (1996) 67–79.
[68] T.R. Bott (Ed.), Fouling Science and Technology, 1998, pp. 251–260.
[69] O.D. Linnikov, Investigation of the initial period of sulphate scale formation Part 3.
Variations of calcium sulphate crystal growth rates at its crystallization on a heat-
exchange surface, Desalination 128 (2000) 47–55.
[70] P. Aimar, J.A. Howell, Effects of concentration polarization boundary layer
development on the flux limitations in ultrafiltration, Chemical Engineering
Research Design 67 (1989) 255–261.
[71] R.W. Field, D. Wu, J.A. Howell, B.B. Gupta, Critical flux concept for microfiltration
fouling, Journal of Membrane Science 100 (1995) 259–272.
[72] P. Bacchin, P. Aimar, V. Sanchez, Model for colloidal fouling of membranes, AIChE
Journal 41 (2) (1995) 368–376.
[73] J. Taborek, T. Aoku, R.B. Ritter, J.W. Paeln, J.G. Knudsen, Fouling the major
unresolved problem in heat transfer, Chemical Engineering Progress 68 (2) (1972)
59–67.
[74] A.E. Childress, M. Elimelech, Effect of solution chemistry on surface charge of
polymeric RO and NF membranes, Journal of Membrane Science 119 (1996)
253–268.
[75] D. Chen, L.K. Weavers, H.W. Walker, J.J. Lenhart, Ultrasonic control of ceramic
membrane fouling by natural organic matter and silica particles, Journal of
Membrane Science 276 (2006) 135–144.
[76] P. Aimar, S. Baklouti, V. Sanchez, Membrane-solute interactions: influence on pure
solvent transfer during ultrafiltration, Journal of Membrane Science 29 (1986)
207–224.
[77] D.Y. Kwon, S. Vigneswaran, Influence of particle size and surface charge on critical
flux of cross-flow microfiltration, Water Science and Technology 38 (4–5) (1998)
481–488.
[78] H. Li, A.G. Fane, H.G.L. Coster, S. Vigneswaran, Direct observation of particle
deposition on the membrane surface during cross-flow microfiltration, Journal of
Membrane Science 149 (1998) 83–97.
[79] S.S. Madaeni, The effect of operating conditions on critical flux in membrane filtration
latexes, Process Safety and Environmental Protection (75) (1997) 266–269.
[80] D. Chen, L.K. Weavers, H.W. Walker, Ultrasonic control of ceramic membrane
fouling: effect of particle characteristics, Water Research 40 (2006) 840–850.
[81] M. Hamachi, M. Mietton-Peuchot, Experimental investigation of cake character-
istics in cross-flow microfiltration, Chemical Engineering Science 54 (1999)
4023–4030.
[82] A.P. Mairal, A.R. Greenberg, W.B. Krantz, L.J. Bond, Realtime measurement of
inorganic fouling of RO desalination membranes using ultrasonic time-domain
reflectometry, Journal of Membrane Science 159 (1999) 185–196.
[83] J. Li, R.D. Sanderson, In situ measurement of particle deposition and its removal in
micro-filtration by ultrasonic timedomain reflectometry, Desalination 146 (2002)
169–175.
[84] T.C. Chilcott, M. Chan, L. Gaedt, T. Nantawisarakul, A.G. Fane, H.G.L. Coster,
Electrical impedance spectroscopy characterization of conducting membranes I.
Theory, Journal of Membrane Science 195 (2002) 153–167.
[85] L. Gaedt, T.C. Chilcott, M. Chan, T. Nantawisarakul, A.G. Fane, H.G.L. Coster,
Electrical impedance spectroscopy characterization of conducting membranes II.
Experimental, Journal of Membrane Science 195 (2002) 169–180.
[86] C.J. Lin, P. Rao, S. Shirazi, Effect of operating parameters on flux decline caused by
cake formation — a model study, Desalination 171 (1) (2005) 95–105.
[87] W.Y. Shih, A. Rahardianto, R.W. Lee, Y. Cohen, Morphometric characterization of
calcium sulfate dihydrate (gypsum) scale on reverse osmosis membranes, Journal
of Membrane Science 252 (2005) 253–263.
[88] R.J. Xie, M.J. Gomez, Y.J. Xing, P.S. Klose, Fouling assessment in a municipal water
reclamation reverse osmosis system as related to concentration factor, Journal of
Environmental Engineering Science 3 (2004) 61–72.