1.

1 General Description of Membrane Separation Process

The membrane can be defined essentially as a barrier, which separates two phases and restricts
transport of various chemicals in a selective manner. A membrane can be homogenous or
heterogeneous, symmetric or asymmetric in structure, solid or liquid, can carry a positive or negative
charge or be neutral or bipolar. Transport through a membrane can be effected by convection or by
diffusion of individual molecules, induced by an electric field or concentration, pressure or
temperature gradient. The membrane thickness may vary from as small as 100 micron to several mms.
Membrane filtration is a fluid/fluid and particle/fluid separation technology used to separate dissolved
substances and fine particles from water and wastewater. A general membrane separation process is
shown schematically in Figure 1.1. A thin semi permeable membrane acts as a selective barrier that
separates particles based on molecular or physical size. In general, water and constituents which are
smaller than the membrane pore size are able to pass through the membrane as “permeate” while
particles and molecules larger than the membrane pore size are retained. The retained particles and
molecules, known as “concentrate”, “retentate” or “residual” may be returned to the feed tank to be
concentrated further with time. “Permeate flux” is the term used to describe how fast permeate passes
through a membrane. Flux is generally reported as volume/area-time. Permeate flux generally
depends upon the individual membrane characteristics (e.g., membrane pore size, membrane surface
charge etc.), the characteristics of the feed stream (e.g., viscosity, solute particle size etc.), and
operating parameters (e.g., transmembrane pressure, feed temperature etc.).

Feed
Retentate

Membrane

Permeate
Figure 1.1: Basic membrane separation process.

1.2 Types of Membranes

The proper choice of a membrane should be determined by the specific application objective:
particulate or dissolved solids removal, hardness reduction or ultra pure water production, removal of
specific gases/chemicals etc. The end-use may also dictate selection of membranes for industries such
as potable water, effluent treatment, desalination or water supply for electronics or pharmaceutical
manufacturing. The following section explains the various types of membranes commonly used.
1.2.1 Microporous Membranes

The membrane behaves almost like a fibre filter and separates by a sieving mechanism
determined by the pore diameter and particle size. Materials such as ceramics, graphite, metal oxides,
polymers etc. are used in making such membranes. The pores in the membrane may vary between 1
nm-20 microns.

1.2.2 Homogeneous Membranes

This is a dense film through which a mixture of molecules is transported by pressure,

concentration or electrical potential gradient. Using these membranes, chemical species of similar size
and diffusivity can be separated efficiently when their concentrations differ significantly.

1.2.3 Asymmetric Membranes

An asymmetric membrane comprises a very thin skin layer on a highly porous thick substructure.
The thin skin acts as the selective membrane. Its separation characteristics are determined by the
nature of membrane material or pore size, and the mass transport rate is determined mainly by the
skin thickness. Porous sub-layer acts as a support for the thin, fragile skin and has little effect on the
separation characteristics.

1.2.4 Electrically Charged Membranes

These are necessarily ion-exchange membranes consisting of highly swollen gels carrying fixed
positive or negative charges. These are mainly used in the electrodialysis.

1.2.5 Liquid Membranes

A liquid membrane utilizes a carrier to selectively transport components such as metal ions at
relatively high rate across the membrane interface.

1.3 General Types of Membrane Separation Processes

 It is not possible to produce one classification that covers all membrane processes. We can
classify membrane processes according to the separation effected, but this fails to distinguish between
processes that achieve the same separation, but use different driving forces. For instance, Reverse
Osmosis, Pervaporation and Electrodialysis can all be used to produce pure water from brackish
water, but the manner in which the processes operate are very different. We can also classify the
membrane processes according to the porosity of the membrane, or the minimum size of the retained
particle, but this classification fails to explain the differences between reverse osmosis, gas separation
and pervaporation membranes. Table 1.1 shows the characteristics of different types of membrane
separation processes:

Table 1.1: Characteristics of membrane separation processes.

Membrane Pore size or Feed Permeate Driving force Mechanism

process MWCO*

Microfiltration 0.02-10μm Liquid Liquid Pressure Sieving

(MF) difference

(0.5-5.0 bar)

Ultrafiltration 0.001-0.02μm Liquid Liquid Pressure Sieving

(UF) difference
MWCO 103 -106
(2-10 bar)

Nanofiltration < 2nm Liquid Liquid Pressure Solution-

(NF) difference Diffusion
MWCO 102 -103
( 5-20 bar)

Reverse 0.1-1.1nm Liquid Liquid Pressure Solution-

Osmosis (RO) difference Diffusion

(10-80 bar)

Gas Separation non-porous Gas Gas Partial Solution-

pressure Diffusion
difference

Dialysis 10-30 Å Liquid Liquid Concentration Sieving and

difference Diffusion

Pervaporation non-porous Liquid Vapor Partial Solution-

pressure Diffusion
difference

Electrodialysis MWCO <200 Liquid Liquid Electrical Ion migration

potential

*
MWCO: Molecular Weight Cutoff
 The mechanism governing the separation depends on the relative sizes of the retained and
permeating molecules and the degree of interaction of the permeating species with the membrane. The
useful ranges of the four general membrane separation processes are presented in Figure 1.2 (Perry
and Green, 1997 and Cheryan,1986).

Figure 1.2: Useful ranges of the four general membrane separation processes.

Various types of membrane separation have been developed for specific industrial applications.
Some of the widely used processes are discussed hereunder:

MF and UF are the two processes that are most often associated with the term “membrane
filtration.” MF and UF are characterized by their ability to remove suspended or colloidal particles via
a sieving mechanism based on the size of the membrane pores relative to that of the particulate matter.
However, all membranes have a distribution of pore sizes, and this distribution will vary according to
the membrane material and manufacturing process. When a pore size is stated, it can be presented as
either nominal (i.e., the average pore size), or absolute (i.e., the maximum pore size) in terms of
microns (  m ). MF membranes are generally considered to have a pore size range of 0.1 – 0.2  m ,
although there are exceptions, as MF membranes with pores sizes of up to 10  m are available.
Sometimes MF membranes are used for the pretreatment of UF feed. For UF, pore sizes generally
range from 0.01 – 0.05  m or less, decreasing to an extent at which the concept of a discernable
“pore” becomes inappropriate, a point at which some discrete macromolecules can be retained by the
membrane material. In terms of a pore size, the lower cutoff for a UF membrane is approximately
0.005  m .
 Figure 1.3: Ultrafiltration.

NF and RO constitute the class of membrane processes that is most often used in applications that
require the removal of dissolved contaminants, as in the case of softening or desalination. The typical
range of MWCO levels is less than 100 Daltons for RO membranes, and between 200 and 1,000
Daltons for NF membranes. While NF and RO are sometimes referred to as “filters” of dissolved
solids, NF and RO utilize semi-permeable membranes that do not have definable pores. NF and RO
processes achieve removal of dissolved contaminants through the process of reverse osmosis, as
described below.

Figure 1.4: Conceptual diagram of osmotic pressure.

NF/RO membranes are designed to remove dissolved solids through the process of reverse
osmosis. Osmosis is the natural flow of a solvent, such as water, through a semi permeable membrane
(acting as a barrier to dissolved solids) from a less concentrated solution to a more concentrated
solution.
 Reverse osmosis, as illustrated in Figure 1.4, is the reversal of the natural osmotic process,
accomplished by applying pressure in excess of the osmotic pressure to the more concentrated
solution. This pressure forces the water through the membrane against the natural osmotic gradient,
thereby increasingly concentrating the water on one side (i.e., the feed) of the membrane and
increasing the volume of water with a lower concentration of dissolved solids on the opposite side
(i.e., the filtrate or permeate). The required operating pressure varies depending on the total dissolve
solid (TDS) of the feed water (i.e., osmotic potential), as well as on membrane properties and
temperature, and can range from less than 100 psi for some NF applications to more than 1,000 psi for
seawater desalting using RO.

Figure 1.5: Nanofiltration and Reverse Osmosis.

1.4 Membrane Applications

The list below gives a few examples of the applications to which membranes have been applied
successfully.

• Microfiltration

 Cell separation from fermentation broths.

 Heat sensitive media sterilisation e.g. protein-containing broths for animal cell
culture.
• Ultrafiltration

 Milk concentration.
 Protein concentration.
 Desalting protein solutions.
 Paint recovery.
 Clarification of fruit and vegetable juices.
• Nanofiltration

 Recovery of sugars and amino acids.

  Recovery of bivalent ions.
 Desalination of brackish waters.
 Colour removal from potable water.
• Reverse Osmosis

 Desalination.
 Ultra-pure water.
 Effluent treatment.
 Concentration of fruit and vegetable juices.
• Gas separation

 Small scale oxygen enrichment.

 Carbon dioxide/methane separation.
 Hydrogen recovery.
• Dialysis

 Artificial Kidney.
 Low alcohol beer.
• Pervaporation

 Alcohol dehydration.
 Removal of trace organics from aqueous streams.
• Electrodialysis

 Demineralisation of whey.
 Clarification of fruit juices.

1.5 Limitations of Membrane Processes

Despite the success of many membrane applications, not everything in the membrane garden is
rosy. There remains a range of technical constraints and economic factors that militate against the
universal adoption of membrane processes.

Size discrimination

Membranes can separate large molecules from aqueous media but the same membrane processes
are generally not capable of fractionating one solute from another.

Pressure limitations

The limitations of a membrane process can be seen in the application of reverse osmosis. RO has
been very widely adopted for the large-scale desalination of brackish waters and is now viewed as the
preferred alternative to multi-effect evaporation. However, when applied to seawater or strong brines
only low concentration factors are possible because of the high osmotic pressures generated by the
low molecular weight solutes. To obtain 'reasonable' concentration factors very high transmembrane
pressures have to be used to overcome the osmotic pressure of the retained brine which on the
contrary results excessive membrane compaction and at the extreme causes failure. In addition, the
pumping costs at high transmembrane pressure can become prohibitive rendering the process
uneconomic.

Fouling

Membrane processes are prone to irreversible fouling of the membrane surface, which can lead
over time to a significant drop in the flow of material through the membrane.

1.6 Description of Ultrafiltration Process

Ultrafiltration is considered as a technique with potential for large-scale applications and has
attracted much attention in the research area of fractionation of proteins. In addition to proteins, bulk
of the solid materials of living organisms contains lipids and polysaccharides, which can interact with
proteins, and membranes during filtration. Therefore, studies on ultrafiltration of those mixtures are
also important (Furth, 1980). UF is primarily a size exclusion-based pressure driven membrane
separation process. UF membranes typically have pore sizes in the range from 10 to 1000Å and are
capable of retaining species in the molecular weight range of 300 to 500,000 dalton. Typical rejected
species include sugars, bio-molecules, polymers, and colloidal particles, which makes it highly
suitable for application in industries like pulp and paper, food processing, pharmaceuticals, paints, etc.
Most UF membranes are described by their nominal molecular weight cutoff (MWCO), which is
usually defined as the smallest molecular weight species for which the membrane has more then 90%
rejection. The view that separation by UF is based on relative molecular sizes is true only as a first
approximation. In many instances, the chemistry of the solute/membrane interaction is also important.
The MWCO of any given membrane can vary with changing feed chemistry as well as with factors
such as molecular orientation, molecular configuration, operating conditions, etc (Bhattacharjee,
2004). However, UF membranes remove particulate contaminants via a size exclusion mechanism and
not on the basis of weight or mass; thus, UF membranes used for drinking water treatment are also
characterized according to pore size with respect to microbial and particulate removal capabilities.

1.6.1 Ultrafiltration Membrane Properties:

Several membrane characteristics are important in determining a membrane's suitability for

separation applications. These overall characteristics can be summed up as

(i) porosity
(ii) morphology

(iii) surface properties

(iv) mechanical strength and

(v) chemical resistance.

These characteristics depend on the membrane material as well as the fabrication technique. To a
great extent, these properties are interrelated; for example, a highly porous membrane structure can be
maintained only if the polymer has adequate mechanical strength. Properties such as resistance to
compaction under pressure, cleaning chemicals, bacterial degradation, and temperature are important
for industrial use. Surface properties and pore morphology have a bearing on fouling properties, flux
through the membrane, and solute separation. To be fit for purpose, a membrane should ideally
possess the following properties:

• High mechanical strength

• High permeability

• High selectivity

• Chemical stability

• Thermal stability

• Resistance to fouling

• Defect free production

1.6.2 Ultrafiltration Membrane Materials

The majority of membranes used for ultrafiltration are manufactured by phase inversion. These
membranes typically have an asymmetric structure with a thin dense porous layer supported by a
thicker spongy layer of the same polymer. The thin layer is responsible for the sieving action of the
membrane and has a high intrinsic hydrodynamic resistance. This is due to the surface porosity, which
is considerably less than that for microfiltration membranes.

Typically the surface porosity is in the region of 10-20% and the clean water fluxes are in the
region of 10-50 l/m2.h at 1 bar. Ultrafiltration membranes can be prepared from a wide range of
polymers, but the choice is limited by the operational requirements of the process. Cheryan reports
that over 130 polymers have been used to create membranes, but only a handful of these have made
the transition into the commercial sector. Even fewer of these materials have been given general
regulatory approval for use in applications that treat products for human consumption. Polymeric
materials used for ultrafiltration membranes are as follows:
  Cellulose (Regenerated)
 Cellulose Acetate (CA)
 Cellulose Nitrate
 Polyacrylonitrile (PAN)
 Polyamide (PA)
 Polyetheretherketone (PEEK)
 Polyethersulfone (PES)
 Polyimide (PI)
 Polysulfone (PS)
 Polyvinyl Alcohol (PVA)
 Polyvinylidene Fluoride (PVDF)

Over the past twenty years, new inorganic membranes have been developed for ultrafiltration
applications. The development of inorganic has been driven by the need to produce membranes with
greater chemical and temperature tolerance. The upper temperature limit of organic polymeric
ultrafiltration membranes is a present less than 200 0C but the theoretical limit is closer to 500 0C.
Inorganic materials such as ceramics show promise because they are chemically and structurally
stable even at elevated temperatures. Materials such as titanium oxide and zirconium oxide can
withstand high temperatures and harsh chemical environments.

1.6.3 Manufacturing Process of Polymeric Membranes

Membrane manufacturing is important in understanding the separation and permeation

mechanism of small molecules passing through a membrane. Membranes are generally made of
durable natural polymers, such as cellulose or synthetic polymers such as, polycarbonate or
polysulfone. In Figure 1.6, a schematic of general manufacturing of polymer membranes is presented.
The polymer is first mixed with suitable solvent and swelling agents. The mixture is known as cast
dope. The next phase of the process is to test a small sample of the casting dope under controlled
conditions to assure the casting dope makes a satisfactory membrane. The controlled test is usually
performed in a glass plate. If the cast dope passes the quality control test, the entire batch is placed
into the casting machine. The casting machine is the critical process in the production of membranes.
Casting equipment is designed specifically for each type of membrane or module configuration.
Therefore, the casting machines are usually designed and constructed by the manufacturers
themselves. In generally, membranes are manufactured in flat sheets. However, membranes can be
cast directly onto the module they are made for (e.g. tubular modules). The membranes then under go
an annealing or drying process and are checked for performance characteristics before being
constructed into the final form [Cheryan, 1986]. During the drying process, solvents in the casting
evaporate forming the pores of the membrane. The membranes are then preserved in a wet storage.
 POLYMER SWELLING AGENT SOLVENTS

CASTING DOPE / SOLUTION

CASTING MACHINE SPINNARET

CASTING MACHINE
ANNEALING PROCESS

WET STORAGE

Figure 1.6 Schematic of the general manufacturing process of polymeric membranes.

1.7 Configurations of Conventional Ultrafiltration Membrane Modules

Several membrane configurations are available both commercially and at the laboratory level.
Polymeric membranes can be cast or extruded, as either flat sheets or in cylindrical geometry. Flat-
sheet membranes are used in spiral-wound or plate and frame (or “plate-frame”) modules while other
modules are of the tubular or hollow-fiber type. Inorganic membranes are used commercially only in
tubular or monolith form, flat disks are available for laboratory use. The most common module
configurations used in commercial cross-flow filtration are tubular, hollow fiber, spiral wound, and
plate and frame.

1.7.1 Tubular Membrane Modules

A tubular membrane consists of a porous membrane material cast on a permeable support matrix
housed inside a rigid hollow tube. Feed is recirculated through the tube at a velocity high enough to
produce turbulence at the membrane surface. Permeate is forced radially through the membrane under
pressure and is collected through a port in the rigid outer casing (Belfort, 1988). Tubular membranes
are available in single or multi-tube configurations. Typical commercial tubular membranes can be
operated at pressures up to 90 psig. These modules are generally 10 feet long and have inner tube
diameters ranging from 0.5 to 1 inch. Typical flow rates necessary to achieve a sufficient shear in a 1
inch tubular membrane range from 30 to 40 gal/min (114 to 151 l/min). Due to large channel sizes
(typically 0.5 to 1 inch), tubular membranes are commonly used to treat waste and process streams
with high solids concentrations. Additionally, the large flow channels make it easier to remove
accumulated solute molecules using mechanical and chemical cleaning techniques (MacNeil and
McCoy, 1989). The disadvantages of these membrane modules include a relatively low permeate flow
rate per unit volume and a high volume hold-up per unit area of membrane (Belfort, 1988).

1.7.2 Hollow Fiber Membrane Modules

Hollow fiber membrane modules consists of a bundle of fine tubes (typical inside diameters range
from 20 to 106 mm) made of the membrane material bound at each end of the module by epoxy discs.
Unlike tubular membranes, hollow fiber membranes do not have an underlying support structure.
Feed is forced into the membrane module radially, under pressure, through a perforated tube in the
center of the housing. Permeate flows through the individual fibers and is collected through epoxy
tube sheets at the end of the module (MacNeil and McCoy, 1989). Transmembrane pressures of up to
40 psig are common in hollow fiber modules (Belfort, 1988).

1.7.3 Spiral Wound Membrane Modules

A spiral wound membrane module consists of alternating layers of porous membranes, permeate
carriers, and spacers, wrapped around a central hollow permeate collection tube. The feed solution
flows axially through the wound membrane module. Permeate is forced through the porous membrane
and is collected through the hollow center tube while rejected feed solution exits at the end of the
module. Spiral wound membrane modules range from 5 to 20 cm in diameter and can be operated at
pressures up to 150 psig. Typically, spiral wound membranes foul more easily than tubular
membranes due to smaller waste flow channels (common spacer thicknesses range from 28 to 80
mm). This factor, coupled with the fact that spiral wound modules cannot be mechanically cleaned,
limits the application of these membrane configurations to waste and process streams with low solids
concentrations (MacNeil and McCoy, 1989). However, spiral wound membranes have a significantly
higher packing density than tubular and hollow fiber configurations, making spiral wound membranes
applicable to the treatment of streams with high volume flow rates (and low solids concentrations)
(Eringis et al.,1993). Spiral wound units have primarily been used in laboratory-scale filtration units
to recover proteins in the biotechnology field and have been considered for large-scale commercial
bioprocessing applications (Belfort, 1988).

1.7.4 Plate and Frame Membrane Modules

In these systems, hollow disc-shaped spacers located between porous membranes are used to
direct the feed radially along the membrane surface. A stack of multiple membranes separated by
spacers is placed in series or in parallel inside a cylindrical housing. Permeate is forced through the
membrane under pressure and is collected at the outer edges of the membrane disc plate. Concentrated
solute is collected at the top of the stack. The advantages of this configuration include a relatively low
volume hold-up per unit membrane area and a thin channel height (typically 0.3 to 0.6 mm) which
permits these units to process highly viscous solutions. The disadvantages of this membrane
configuration include susceptibility to channel plugging and difficulties in mechanical cleaning
(Belfort, 1988). Additionally, the commercial application of plate and frame membrane configurations
has been limited due to high capital cost relative to other filtration units (MacNeil and McCoy, 1989).
Table 1.2: Relative merits and demerits of various pressure driven membrane modules.

Different Membrane Modules Advantages Disadvantages

Turbulent flow, easy cleaning, High capital cost, low packing

easy handling of suspended density and limited achievable
Tubular materials and viscous fluids. concentration.

Very high packing density, low Fragility of the fibers, inability

cost, cleaning can be done by to handle suspended solids well,
backflushing. damage of one fiber requires
replacement of the entire
Hollow Fiber
module.

High packing density, relatively Problems in handling suspended

low cost. solids, difficulty in cleaning and
Spiral Wound in high temperature applications
plastic components may deform.

Easy to clean and replace Relatively low packing density,

membranes high initial cost, difficult to
Plate and Frame disassembly for cleaning.

2.1 Defining Membrane Performance

The performance of a membrane can be defined in terms of the following parameters:

 Flux
 Permeate flux ( J ) is the term to describe the performance of membrane filtration system. The
volumetric flux (permeate volume /membrane area.time) is given by

J  Pm  P    / l (2.1)

where, Pm is the permeability coefficient, P is the hydrostatic pressure difference,  is the

osmotic pressure difference between feed and permeate phases, and l is the membrane thickness.
Osmotic effects are usually smaller in UF than in RO. Cases occur in which the osmotic pressure term
in equation 2.1 can be neglected; however, this should be checked for each application since it can
lead to significant error even in the case of large molecules such as polyethylene oxide. Permeate flux
is the volume of permeate that flows through a unit area of membrane in unit time and is represented
by equation 2.2. Two common units are used to report permeate flux are kilogram force per square
centimeter per second (kgf.cm-2s-1) and liters per square meter per hour (LMH).

1 dV
J (2.2)
A dt

where, J is the permeate flux, V is the collected permeate volume, A is the effective membrane
surface area, t is the time of permeate collection.

 Membrane Hydraulic Resistance

Membrane materials must be able to withstand the mechanical, chemical, and biological stresses
imposed during operation and cleaning without suffering appreciable losses in operating efficiency.
Mir et al. (1977) discussed several general operational and feed parameters that must be considered in
the treatment of wastewaters. The factors include pH, ionic strength, temperature, and pressure.
Additionally, membrane surface properties can also affect permeate flux and quality. Electrostatic
repulsion between the membrane and the contaminant can be used to enhance waste solute retention
and increase flux. Membrane resistance is a measure of the hydraulic resistance to flow through a pore
channel and is an intrinsic material property of the membrane due to the material properties of the
membrane. The membrane resistance can be determined by applying Darcy’s Law for a virgin
membrane with clean water feed:

J  P (2.3)
 Rm

where, J is the permeate flux, P is the transmembrane pressure,  is the water viscosity
(absolute), Rm is the membrane hydraulic resistance.

The assumption of well-developed laminar flow through the membrane pore channel is made
when using equation 2.3. The relationship between permeate flux and transmembrane pressure for a
clean water feed is presented in Figure 2.1. When clean water is used as the feed, the slope of a
regression line through the origin on the plot of permeate flux versus transmembrane pressure is
inversely proportional to the product of the feed viscosity and the resistance of the membrane, Rm , as
presented in Figure 2.1 (Cheryan, 1986). In this case, the membrane resistance is:

Rm  1
  slope  (2.4)

A summary of typical resistances for several ceramic (aluminum oxide) micro and ultrafiltration
membranes studied by Nazzal and Wiesner (1996) is presented in Table 2.1. Additionally, a resistance
of 1012 m-1 was measured by Lipp et al. (1988) for a polymeric membrane used to treat oil-in-water
emulsions in a stirred cell UF unit (membrane material and MWCO were not specified). The MF and
UF membrane resistances reported by Nazzal and Wiesner (1996) and Lipp et al. (1988) were
comparable to values measured by other researchers (Gutman, 1987 and Bhave, 1991). Membrane
(material and pore size/MWCO) and operating conditions are selected to achieve adequate rejection of
emulsified solutes while maintaining an acceptable permeate flux. The selection process is based
primarily on empirical data and previous experience. However, Nazzal and Wiesner (1996) have
developed a model for calculating the critical pressure, Pcrit, required to force the entry of an oil
droplet into a membrane pore based on the Young-Laplace equation. A schematic of the passage of an
oil droplet through the pores of a membrane is presented in Figure 2.1. The critical pressure is given
by:

Figure 2.1: The relationship between permeate flux and transmembrane pressure for a clean
water feed.

Table 2.1: Typical membrane resistances for micro and ultrafiltration membranes (Nazzal and
Wisner, 1996).

Pore diameter Membrane resistance (m-1)

(µm)

0.8 1.5 x 1011

 0.5 2.0 x 1011

0.2 5.0 x 1011

0.05 6.0 x 1011

 Permeability: The rate at which material passes through the membrane is known as permeability
of the membrane.
 Selectivity: The ability of a membrane to discriminate between species in the original solution.
Selectivity (S) of A with respect to B is defined as the ratio of concentration of A to B in permeate
to the same ratio in feed, which can be written mathematically as follows:

C A, P CB , P
S (2.5)
C A, F CB , F

 Reflection Co-efficient: The reflection co-efficient ( s) in semipermeable membranes relates to

how such a membrane can reflect solute particles from passing through. A value of zero results in
all particles passing through. A value of one is such that no particle can pass.
 Retention Coefficient / Rejection: The observed solute rejection Ro for a given species i is
given by

cip
Ro  1  (2.6)
cib

where, cip is the concentration in the permeate while cib is the corresponding retentate side value.
It is usually called observed rejection.

When UF is performed in a batch process, the rejection can be calculated from either the permeate
or retentate concentrations and the corresponding volume ratios. If Vo and c f are the initial feed

volume and concentration, respectively, while V p is the total permeate volume and Vr the remaining
retentate volume, the rejection can be calculated as
ln(c r / c f )
Ri  (2.7)
ln(VCF )
or as

ln[VCF  (c p / c f )(VCF  1)]

Ri  (2.8)
ln(VCF )
 Vo
The volume concentration factor (VCF) ( ) is also an important measure of the achievable
Vr
separation.

Real rejection is defined as follows:

cp
Rr  1  (2.9)
cm

 Molecular Weight Cutoff Profile: The rejection characteristics of UF membranes are usually
expressed a nominal MWCO. This number refers to the molecular weight (dalton) of a species,
which would be expected to have an R-value of at least 0.9. A MWCO profile or retention curve
is constructed by measuring the R-value of chemically similar compounds of varying molecular
weights. An average MWCO and a dispersion coefficient D can characterize such a profile, which
is a measure of the sharpness of the separation possible.

2.2 Factors Controlling Permeate Flux

2.2.1 Concentration Polarization

The theoretical relationship between permeate flux, transmembrane pressure, feed concentration,
and hydraulic turbulence is presented in Figure 2.2. Under conditions of low transmembrane pressure,
low feed concentration, and high hydraulic turbulence, permeate flux is linearly proportional to
transmembrane pressure. The linear permeate flux/transmembrane pressure relationship is idealized in
the case of a clean water feed. When clean water feed is ultrafiltered, there are theoretically no
contaminants to either plug the membrane pores or accumulate on the membrane surface. The buildup
of solute at the membrane surface is known as “concentration polarization” and is largely responsible
for the deviation of the permeate flux from the linear flux-pressure model presented in equation 2.3.
Primarily, three competing solute transport mechanisms occur in ultrafiltration: (i) convective
transport of solute to the membrane surface (ii) back diffusion of accumulated solute to the bulk
solution, and (iii) solute permeation through the membrane as presented schematically in Figure 2.3.
Solute is convectively transported to the porous membrane surface and a fraction of the solvent is
removed as permeate which results in a locally higher solute concentration at the membrane surface
relative to the bulk solute concentration.
 Figure 2.2: The theoretical relationship between permeate flux, transmembrane pressure, feed
concentration and hydraulic turbulence.

Figure 2.3: Schematic of the three primary solute transport mechanisms in membrane
separations systems.

As the transmembrane pressure is increased, feed concentration is increased, or hydraulic

turbulence is decreased, the permeate flux becomes independent of pressure. In general, there are four
operational and feed solution parameters which contribute to the formation of the polarized solute
concentration boundary layer and subsequent asymptotic permeate flux response to increased
transmembrane pressure: (1) transmembrane pressure, (2) feed concentration, (3) hydraulic
turbulence, and (4) feed viscosity/temperature. Transmembrane pressure is the principle driving force
in membrane separation systems. However, as transmembrane pressure is increased, there is a
corresponding increase in convective transport of solute molecules to the membrane surface. A
thicker solute boundary layer forms as a result of the enhanced convective transport of solute
molecules to the membrane surface and the permeate flux becomes independent of transmembrane
pressure. Solute concentration in the feed is another parameter which limits permeate flux. As the
solute concentration in the feed increases, permeate flux decreases due to the accumulation of solute
molecules at the membrane surface. At high solute feed concentrations, it becomes difficult to pump a
feed solution at high enough to minimize the thickness of the polarized concentration boundary layer.
Thus, the permeate flux decreases as the solute concentration increases. The hydraulic turbulence in
membrane separation systems also plays an important role in maintaining a satisfactory permeate flux.
As hydraulic turbulence is increased, the permeate flux should theoretically also increase due to a
decrease in the thickness of the polarized concentration boundary layer. Similarly, as the temperature
of the feed solution is increased, permeate flux increases due to: (1) a decrease in the viscosity of the
feed solution (which makes it easier to pump the feed solution at a high cross-flow velocity) or (2) an
increase in the solute diffusivity (which results in an increased diffusion of solute molecules from the
membrane surface back to the bulk solution). Cheryan (1986) and Cheryan (1977) noted three
operational characteristics which can be used to judge the extent of concentration polarization in a UF
system: (1) deviation from the pure water line, (2) relatively rapid attainment of pressure-independent
flux behavior, and (3) marked hysteresis effect observed on lowering the pressure. Although the
deviation from the pure water flux may be attributed to a decrease in the driving force due to
increased osmotic pressure, Cheryan believed the primary mechanism of flux decline was the increase
in hydrodynamic resistance in the concentration boundary layer. Cheryan supported the argument in
favor of flux decline due to the formation of a polarized gel layer. This was based on prior knowledge
of the gel characteristics of soybean proteins and the gel-like behavior of highly structured
macromolecules (at concentrations in excess of 10 wt. %), which were in good agreement with
experimental values. Additionally, the hysteresis effects observed as pressure was varied in these
experiments supported the hypothesis that interactions had occurred between solute molecules which
were approaching a consolidated “close packed” configuration. Citing the marked differences in the
mass transfer properties of feed solutions, Cheryan suggested that mass transfer models will have to
be supplemented with experimental data to be used in UF process design (Cheryan, 1977 and
Cheryan, 1986).

2.2.2 Membrane Fouling

Membrane fouling can also result in a decrease in permeate flux over time. Fouling is a term used
to describe the loss of throughput of a membrane device as it becomes chemically or physically
changed by a process fluid. Fouling is different from concentration polarization. For example, when
the flow rate in a cross-flow membrane separation system is reduced, permeate flux will decrease. If
the decrease in permeate flux is due only to the effects of concentration polarization ( i.e., the
accumulation of solute molecules at the membrane surface), the permeate flux should return to the
initial value when the flow rate is raised. However, if the permeate flux does not increase to the initial
value when the flow rate is increased, membrane fouling is the operative flux decline phenomena.
Membrane fouling is primarily due to (1) pore plugging and/or (2) solute adsorption. In the case of
pore plugging, solute molecules or other minor constituents of a feed stream (e.g., bacteria) are forced
into the membrane pores resulting in a loss of permeate flow channels and thus a decrease in permeate
flux. Pore plugging becomes more prevalent under operating conditions such as low hydraulic
turbulence and high transmembrane pressure. At low hydraulic turbulence, a thicker solute layer
accumulates on the membrane surface. When combined with high transmembrane pressure, the
potential for forcing solute molecules into the membrane pores increases. Additionally, if solute
molecules are small enough to enter the membrane pores, some solute may be adsorbed onto the pore
walls. The adsorbed molecules decrease the effective pore diameter. As the effective pore diameter is
decreased, the overall resistance to hydraulic flow increases and a decreased in permeate flux is
observed. Solute adsorption is due principally to physical and/or chemical properties of the feed
solution (Jonsson and Tragardh, 1990). In both cases, the proper selection of operating conditions,
pretreatment of the feed, and cleaning methods are important aspects to minimizing flux loss due to
membrane fouling. For example, the transmembrane pressure should be balanced to provide
maximum permeate flux at the minimum transmembrane pressure. Thus, an understanding of the flux-
pressure relationship presented schematically in Figure 2.2 is important to the minimization of
membrane fouling. Specifically, a membrane separation system should not be run at transmembrane
pressures in excess of the value necessary to achieve a mass-transfer-controlled permeate flux. Feed
pretreatment is often stream or site-specific and has received little attention in the literature. However,
steps such as removing free oil from a waste or process stream prior to the application of membrane
separation are often used to minimize potential membrane fouling. The primary method of restoring
permeate flux lost to membrane fouling is membrane cleaning. Fouling which can be removed
through membrane cleaning is called “reversible fouling” and fouling which cannot be removed
through cleaning is called “irreversible fouling.” Cleaning procedures are highly stream-specific.
Although cleaning is used to restore permeate flux, the use of aggressive chemicals such as
concentrated acidic and caustic solutions often results in shortened membrane life due to affects on
membrane surface properties (Perry and Green, 1997).

2.3 Flux Decline

Fane has discussed initial ultrafiltration flux decline in terms of the boundary layer theory,
adsorption, and pore plugging. Long term UF flux decline was attributed to deposition/adsorption and
cake consolidation. A similar analysis by Howell and Velicangil divides the UF process into three
time intervals:

ii) First few seconds: A quasi-steady-state concentration polarization layer is set up.
ii) 1 to 10 minutes: Solute adsorption.
iii) Long-term: Gelation.
 Figure 2.4: Flux decline as a result of concentration polarization.

When an ultrafiltration system is first 'switched on', we would anticipate an initial, rapid decline
in the observed flux before levelling out to some steady state value as shown in Figure 2.4. The
difference between the initial flux (which will tend towards the pure solvent flux) and the steady state
flux will be due to the effects of concentration polarisation. In the majority of systems, it was
observed that not a steady state flux but a gradual decline, which takes place over the longer term.
This decline in flux is one of the major problems in pressure driven membrane processes and is
particularly acute in microfiltration and ultrafiltration. The long-term decline in flux is normally due
to fouling; the gradual accumulation of retained species at the membrane interface.

Fouling occurs through a variety of mechanisms including; pore blocking, precipitation (cake
formation) chemical interactions and adsorption. The rate at which fouling changes the trans-
membrane flux is influenced by the nature of the feed the hydrodynamics of the retentate solution and
the surface characteristics of the membrane.

Figure 2.5: The effects of fouling with time.

There are no simple theoretical models which can be applied to describe fouling and its influence
on the long term performance of a membrane system. As a consequence, the designer has to resort to
empirical fouling tests on pilot scale plant using representative solutions to estimate:

• the flux decline over time

• reversible and irreversible fouling

• the cleaning cycle required to restore performance.

In the absence of a unified theory of fouling, a number of qualitative descriptions have been
developed in an attempt to describe the fouling mechanism. It is generally accepted that the species
that are retained at the membrane interface either by deposition on the surface, or by entrapment in the
pores, bring about a change in the effective pore size distribution which not only changes the flux -
transmembrane pressure relationship, but also alters the retention coefficient of the membrane itself.

2.4 Techniques to Increase Flux and Reduce Fouling

Because the modes of fouling dependent on a number of interactions between the membrane
system and the feed solution there are a number of techniques which can help to increase flux and
reduce fouling.

2.4.1 Flow and Turbulence

High shear rates across the membrane surface tend to reduce deposited material but this is not
always the case. At high transmembrane pressures the rate of fouling can increase with increasing
cross flow velocity. This especially marked if the permeation rate is low. The explanation for this
behaviour is thought to be related to the differing effects on large and small particles. For small
particles and solutes the Brownian diffusion is dominant and is independent of shear and inversely
proportional to the size of the particle. For larger particles shear induced diffusion is an important
factor. As the velocity is increased the lift is increased and a greater stratification of the boundary
occur giving rise to a low permeability layer.

2.4.2 Pressure

Increasing transmembrane pressure will first show pressure dependent flux, but as the pressure is
increased a point is reached where further increases produce no increase in flux. If the layer is a result
of concentration polarization from a non-fouling component, a steady state will be reached. If the
layers are fouling i.e. the material binds to the surface, increases in pressure tend to compress the
fouling layer and the permeability decreases which in turn produce a reduction in flux.

2.4.3 Temperature
 Increasing temperature will generally improve flux because of lower viscosity and higher
diffusivity. However for some systems containing salts with inverse temperature solubility. This
occurs especially in whey treatment. The calcium phosphate in whey can precipitate on the membrane
with increasing temperature.

2.4.4 Choice of Membrane

Proper selection of membrane is very important to get desired output in UF process. The use of
hydrophilic membranes for the separation of proteins can improve flux and reduce fouling.

2.4.5 Backflushing, Trans-membrane Pressure Control, Pulsing and Washing

The normal way to operate ultrafiltration to reduce the effects of fouling is to use high retentate
velocities. This increases pumping cost and produces a high pressure-drop across membrane module.
In principle, the high velocity reduces the boundary layer and increases mass transfer. However, as
the pressure-drop on the permeate side is low and the pressure of the permeate phase is usually close
to atmospheric the high pressure-drop gives rise to a very high trans-membranes pressure at the inlet
and a low trans-membrane pressure at the outlet. The high pressure-difference at the inlet can cause
compaction of the fouling layer, which reduces the flux. To overcome this, back flushing can be done.
The flow in the retentate phase is stopped intermittently and the permeate pumped in reverse flow
through the membrane. This helps to remove some of the fouling materials but overtime the benefits
of backflushing tend to reduce and periodically the module has to be taken out of service and cleaned.

2.5 Cleaning of Membrane

Cleaning is necessary to regenerate the fouled membrane. Regeneration is generally measured

in terms of restoration of the initial pure water permeability of the membrane. Cleaning may be
hydraulic cleaning or mechanical cleaning or electrical cleaning or chemical cleaning. The selection
of cleaning method depends on the type of fouling material, membrane configuration, resistance of
the membrane to the applied pressure during backflushing and also resistance to the cleaning
chemicals. Several design modifications in crossflow membrane system had been incorporated in
order to reduce fouling such as backflushing (Mores and Davis, 2002), vortex generation (Costigan et
al., 2002), rotating disk (Lee et al., 1995; Brou et al., 2002), gas sparging into the feed stream (Sur and
Cui, 2005), pulsatile flow (Howell et al., 1993), ultrasonic cleaning (Muthukumaran et al., 2005) etc.
Chemical cleaning is commonly used to restore the fouled membrane. Selection of cleaning chemicals
and cleaning time depends on the membrane materials particularly for polymeric membranes.
Chemical cleaning deteriorates the membrane quality, involves the cost of cleaning chemicals, time
consuming and some chemicals are corrosive to the system (Madaeni et al., 2001).
2.6 Limitations of Conventional Cross-Flow Ultrafiltration Systems

A major limitation in the application of conventional UF membrane modules/systems (e.g.,

tubular, hollow fiber, etc.) for the treatment of industrial wastewaters has been the low permeate flux
observed at high solute concentrations. With time, a solute boundary layer forms as solute molecules
accumulate on the membrane surface. The accumulation of solute molecules at the membrane surface
is referred to as concentration polarization. If the accumulated solute layer is sufficiently thick, a
decrease in permeate flow rate will occur. The buildup of solute molecules at the membrane surface is
reduced by back diffusion of solute molecules from the region of high concentration near the
membrane surface to the region of lower solute concentration in the bulk solution. The solute
boundary layer thickness can be further reduced by providing enhanced hydraulic turbulence to
“scour” the accumulated solute from the membrane surface. In conventional membrane systems the
hydraulically induced turbulence is provided by recycling a large portion of the concentrate back to
the membrane unit producing high liquid velocities near the membrane surface. These large velocities
increase turbulence, which reduces the thickness of the solute concentration boundary layer through a
subsequent increase in the back diffusion of solute molecules away from the membrane surface.
However, conventional systems are unable to maintain the high velocities necessary to minimize the
deleterious effects of the solute boundary layer on permeate flux because of the difficulty in pumping
viscous liquids at high flow rates as the feed solution is concentrated with treatment time (Murkes and
Carlsson, 1988 and Cheryan, 1998).

Figure 2.6: Schematic diagram of crossflow membrane separation.

2.7 Applications of Ultrafiltration

Many ultrafiltration applications are already being practiced in industry and more are being
studied, these included applications in the food industry, pharmaceuticals and biotechnology, water
purification, and waste treatment in the chemical and paper industries. The various application of
ultrafiltration can be summarized as follows:

Table 2.2: Applications of ultrafiltration in different industrial field.

Process Separation
Electrophoretic paint Process rinses water, recycle paint to dip tank, allow safe disposal of
rinse water.
Cheese whey Concentration/ fractionation of proteins from lactose & inorganic.
Juice clarification Remove haze component from apple juice.
Textile sizing agent Recovery of poly vinyl alcohol (PVA) and CMC
Wine clarification Remove haze components from red and white wines.
Oil water emulsions Metal cutting oils (lubricants) concentrated from wastewater for
incineration.
Polymer latex Latex emulsion concentrated from wastewater.
De-waxing Separation of wax components from lowers paraffin.
De-asphalting Solvent recovery/ recycle for de-asphalting of heavy crude.
Egg-white concentration Partial de-watering before spray drying.
Fermentation broth Separates low molecular weight organic therapeutic agents from cell
or cell debris
Kaolin concentration Partial de-watering of clay slurry before centrifugation.
Waste water treatment Concentration before sludge de-watering.
Affinity membrane Retain ligand complex from non-complex protein.
Reverse osmosis pretreatment Retain colloidal silica, bacteria etc.

Forward this mail to all of your friend. Highlighted

topics are important. Read electrodialysis process
from McCabe.