You are on page 1of 10

Separation and Purification Technology 64 (2008) 38–47

Contents lists available at ScienceDirect

Separation and Purification Technology


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

Effect of silica particles on cellulose acetate blend


ultrafiltration membranes: Part I
G. Arthanareeswaran ∗ , T.K. Sriyamuna Devi, M. Raajenthiren
Department of Chemical Engineering, A. C. College of Technology, Anna University, Chennai 600025, India

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

Article history: This paper deals with the preparation of organic–inorganic ultrafiltration (UF) membranes by solution
Received 26 April 2007 casting followed by phase separation method. The silica (SiO2 ) particles were added to the cellulose acetate
Received in revised form 22 July 2008 (CA) polymer with the increment of 10 wt.% from 0 to 40% by weight using N,N-dimethyl formamide (DMF)
Accepted 24 August 2008
as polar solvent. The prepared organic–inorganic membranes were characterized for UF performance such
as compaction, pure water flux, % water content, membrane hydraulic resistance, molecular weight cut-
Keywords:
off (MWCO) and pore statistics. MWCO and pore statistics were investigated using protein solutions of
Organic–inorganic membrane
different molecular weights. It is observed that by increasing the concentration of SiO2 in CA polymer, the
Pore statistics
Membrane morphology
MWCO, pore radius, surface porosity and pore density has been increased. The mechanical stability of the
Protein separation CA/SiO2 blend membranes increased initially and then declined with the addition of inorganic particle
Membrane fouling-resistant ability above 10 wt.% to the casting solution. The morphological structure was changed with the addition of SiO2
particles in the casting solution. Further the permeate flux and % rejection of different molecular weight of
proteins were investigated which showed an increased protein permeate flux with the decreased solute
rejection. Meanwhile, the effect of SiO2 content in the CA membranes on fouling-resistant ability was
studied using BSA solution. The results indicated that increasing the SiO2 content in the casting solution,
the reversible fouling resistance dominated the total fouling resistance thereby improving the fouling
resistance ability of the blend membranes.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction In general, Cellulose acetate (CA) is one of the best membrane


materials and is very important in the field of Reverse osmosis
Membrane process at present time is extensively applied in (RO), UF, Gas permeation [3]. Asymmetric UF membranes based on
many sectors of the manufacturing industries including water CA were prepared and studied extensively as a function of cast-
desalination, ultra-pure water production, oil–water separation, ing solution composition and membrane formation mechanism
production of beverages, electro-coat paint recovery, etc. [1]. Mem- [4]. CA membranes has been prepared by many of the membrane
brane processes have gained more importance than the other sep- researchers and characterized for their compaction, hydraulic per-
aration processes due to low energy consumption, easy scale-up, meability and osmotic permeability [5]. CA and its derivative
less or no use of chemicals and no harmful by-product formation. membranes made them suitable as membrane materials because
Ultrafiltration (UF) is the most commonly used membrane sepa- of their moderate flux and moderate salt rejection, has good film
ration method used to separate a solution containing desirable and forming properties, they are relatively easy to manufacture, cost
undesirable components. Typical rejected species include sugars, effective, and highly non-toxic. However, CA is not suitable for more
bio-molecules, polymers, and colloidal particles [2]. UF membrane aggressive cleaning, has low oxidation, has poor resistance to chlo-
process depends on physical properties of the membrane, such as rine, can be used only in the narrow temperature range (maximum
hydraulic permeability, membrane thickness, process and system 30 ◦ C) and has poor mechanical strength, the modification of CA
variable like system pressure, filtration time and feed concentra- gains importance [6].
tion. In practice, CA is blended with other polymers to improve the
performance of the membranes [7–12]. The blend membranes
have shown improved permselectivity and permeability than those
∗ Corresponding author. Present address: Department of Chemical Engineering,
made of individual polymers. Few researchers have investigated the
National Institute of Technology, Tiruchirappalli 620015, India.
blend of organic polymer with inorganic materials like alumina,
Tel.: +91 431 2501811; fax: +91 431 2500133. silica, zirconia, titania, etc. [13–18]. Bottino et al. [16] prepared the
E-mail address: arthanaree10@yahoo.com (G. Arthanareeswaran). novel organic–inorganic membranes formed by uniform dispersion

1383-5866/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.seppur.2008.08.010
G. Arthanareeswaran et al. / Separation and Purification Technology 64 (2008) 38–47 39

of fine silica particles in the porous matrix of polyvinylidenefluo- the CA (17.5 wt.%, by weight of the solution) in the DMF as sol-
ride (PVDF) and observed that with the increasing concentration of vent at room temperature and adding fumed SiO2 particles to the
SiO2 in PVDF, the permeate flux increased with the lowered pro- casting dopes while stirring. For the organic–inorganic membranes
tein retention. Yan et al. [18] prepared blend membranes based (CA–SiO2 ), the CA polymer was first dissolved in the solvent and
on PVDF and alumina (Al2 O3 ) materials by phase inversion pro- then thoroughly mixed with the SiO2 . Special care was taken to
cesses and characteristic studies like membrane hydrophilicity, ensure homogeneous dispersion of the SiO2 particles. The concen-
porosity, protein retention and surface morphologies were inves- tration of SiO2 added to CA varied from 0 to 40 wt.%, beyond the
tigated. It was observed that the permeation flux increase of the 40 wt.% resulted in the turbid form and did not form the mem-
membrane is attributed to surface hydrophilicity increase due to brane. This may due to concentration of CA in the casting solution
the hydrophilic inorganic nano-sized Al2 O3 particles addition. Yang was reduced below 60 wt.%, in other words the polymer rich phase
et al. [13] prepared PSf/TiO2 organic–inorganic composite mem- became very weak in the membrane formation, which in turn lead
branes via phase-inversion process and observed that with the to the inhibition of membrane formation. In order to obtain optimal
addition of TiO2 particles to polymer casting solution resulted in dispersions of the particles in the polymer solutions, agitation was
the increased pore numbers in the skin layer and the hydrophilic- required for at least 24 h. The casting solutions were then kept for
ity also improved apparently that enhanced the pure water flux. The 24 h to remove air bubbles. The polymer solution with SiO2 parti-
mechanical strength of membrane was enhanced through adding cles was kept in dehumidifying chamber with a relative humidity
inorganic fillers, especially, at 2 wt.% filler concentration, the burst- of 60 ± 5%. The solution was then casted over a glass plate with the
ing strength and the breaking strength increased 50% and 26.7%, help of doctor blade. The thickness of prepared membranes was
respectively. Wara et al. [14] reported the fabrication of nano- measured in different places of membrane, the average membrane
composite membranes of Al2 O3 particles in CA by using the solution thickness was 0.2 ± 0.02 mm. This thickness is maintained for all
blending. We have examined consist of inorganic domains that are the membranes. The casted solution was then allowed to be in the
randomly dispersed with the polymer matrix. The presence of the chamber for 30 s for the evaporation of solvent from top surface of
inorganic phase can also serve to restrict the molecular motions the membrane. The casted membrane was then immersed in the
of the polymer chains and to induce an increase in the mean dis- gelation bath with water as non-solvent and DMF as solvent with
tance (free volume) between polymer chains. We postulate that the the addition of SLS as surfactant for 30 min for the leaching out of
restricted molecular motions and a favorable increase in the mean solvent from the membrane surface [20]. This method is commonly
distance between chains can lead to a simultaneous improvement known as phase inversion technique [21]. The formed membranes
in selectivity and permeability. were stored in the 0.1 wt.% formalin solution to avoid microbial
Hence, for the present study, SiO2 particles are blended with attack. The stepwise preparation of the polymeric membrane is
CA with an increment of 10 wt.% from 0 to 40 wt.% to obtain mem- shown in Fig. 1.
branes with improved quality in terms of UF characteristics such
as compaction, pure water flux, percent water content, membrane 2.3. Experimental set up
hydraulic resistance, and in terms of % solute rejection, MWCO,
pore statistics, mechanical stability and membrane morphology. The characterization methods were carried out in a batch type,
The fouling-resistance ability and the recycling of the CA/SiO2 blend dead end cell (ultrafiltration cell-S76-400-Model, Spectrum, USA)
membranes were evaluated through UF experiments using BSA as with a diameter of 76 mm and effective membrane filtration area of
model protein. Further, increase of SiO2 in CA is above 40 wt.%, 38.5 cm2 fitted with Teflon coated magnetic paddle. While carry-
resulted in the turbid form and the polymer phase separation ing out the protein rejection studies, a constant agitation of speed
occurred. Hence, the threshold limit of SiO2 is maintained to a 300 rpm was used in to order to reduce concentration polarization
maximum of 40 wt.% in CA blend composition. of the membranes. The cell was connected to a nitrogen cylinder
with the pressure control valve and gauge through a feed reservoir.
2. Experimental procedure
2.4. UF characterization
2.1. Materials
The prepared membranes were characterized for compaction,
The CA used was obtained from Mysore Acetate and Chemi- pure water flux, membrane hydraulic resistance, protein rejec-
cals Company Limited, India, and was used, after recrystallatization tion studies, fouling-resistant ability and recycling of the blend
from acetone. Analar grade N,N-dimethyl Formamide (DMF) was membranes using dead end UF experiment. The % water content,
obtained from was and sieved through molecular sieves for MWCO, pore statistics, mechanical stability and morphological
removing moisture and stored in dried condition. Fumed sil- studies were also investigated.
ica powders (hydrophilic silica powder, average primary particle
size = 0.014 mm) was purchased from Central Drug House, India 2.4.1. Compaction
and dried at 80 C before use. Sodium lauryl sulfate (SLS) used as The thoroughly washed membrane was cut into desired shape
surfactant was obtained from Sisco Research Limited, India. Pro- and fitted in UF kit. The distilled water was fed into the UF kit from
teins like trypsin, pepsin, and bovine serum albumin (BSA) were the pressure reservoir and the initial water flux was taken, 20 s after
also obtained from Sisco Research Limited, India. Egg albumin was the pressurization at 414 kPa [22]. The permeate was collected for
obtained from Central Drug House, India. Phosphate buffers like every 1 h of time interval till it attains steady state.
Mono-sodium di-hydrogen ortho phosphate and di-sodium hydro-
gen ortho phosphate were procured from Central Drug House, India. 2.4.2. Pure water flux (PWF)
After compaction, the experiment was carried out at a differ-
ential system pressure of 345 kPa and the permeate was collected
2.2. Solution blending and membrane formation
[22]. The PWF was calculated using the equation:

CA–SiO2 flat sheet membranes were prepared by the phase- Q


Jw = (1)
inversion method. A casting solution was prepared by dissolving t × A
40 G. Arthanareeswaran et al. / Separation and Purification Technology 64 (2008) 38–47

Table 1
Average solute radius of the proteins

Protein Molecular weight (kDa) Average solute radius (Å)a

Trypsin 20 21.5
Pepsin 35 28.5
Egg albumin 45 33.0
BSA 69 45.0
a
Values given by Sarbolouki [24].

2.4.5. Molecular weight cut-off (MWCO)


MWCO characterization involves finding inert solutes, which
has the lowest molecular weight that is more than 80% and less
than 100% retained by the membrane in a steady state RO or UF
experiments. Here the percent solute rejection, % SR is found using
the equation:

Cp
%SR = 1 − × 100 (4)
Cf

where Cp and Cf are the concentrations of the solute in perme-


ate and feed solutions. This method is applied to the homogenous
and heterogeneous membranes, and the molecular weight of the
solute retained depends only on the structure of the upstream sur-
face layer of the membrane [23]. The feed solutions were prepared
using the proteins of different molecular weight such as trypsin
(20 kDa), pepsin (35 kDa), egg albumin (45 kDa) and BSA (69 kDa).
The protein solutions of 0.1 wt.% were prepared by dissolving it in
(0.5 M, pH 7.2) phosphate buffer that were used as standard feed
solutions. The pH of the buffer solution was maintained at 7.2, since
any change in the pH may lead to adsorptive fouling of the mem-
brane surface [24]. The concentration of the feed and the permeate
collected was found using UV–visible Double beam Spectrometer
(Elica, SL 164, Double Beam) at max = 280 nm.

Fig. 1. Scheme showing the preparation method of CA/SiO2 UF membrane.


2.4.6. Pore statistics
For studying the pore statistics, the proteins of different molecu-
where Jw is the pure water flux (l m−2 h−1 );
Q is the amount of
lar weight such as trypsin, pepsin, egg albumin and BSA were used.
permeates collected (l); t is the sampling time (h) and A is the
From the protein rejection studies described below, the average
membrane area (m2 ).
pore radius, surface porosity and pore density of the membrane
were studied. The pores statistics [23] like average pore radius was
2.4.3. % Water content
calculated using the equation:
Water content of the membranes was obtained after soaking
membranes in water for 24 h and the membranes were weighed ˛
followed by mopping it with blotting paper. The wet membranes R= × 100 (5)
SR%
were placed in vacuum drier at 75 ◦ C for 48 h and the dry weights of
the membranes were determined [22]. The percent water content where R is the average pore radius (Å) of the membrane; ˛ is the
was calculated given by the equation: average solute radius (Å) and is constant for each molecular weight.
The average solute radii is known as “stoke radii” and the value of ˛
(wet sample weight − dry sample weight)
% Water content = can be found from the plot between the solute radius and molecular
wet sample weight
weight of the solute given by Sarbolouki which is shown in Table 1.
×100 (2) Assuming the membrane to be asymmetric skin type, the surface
porosity of the membrane was found using the equation:
2.4.4. Membrane hydraulic resistance (Rm )
3Jw
The membrane hydraulic resistance is the resistance offered by ε= (6)
the membrane to the flow of the feed. It is an indication of the R × P
tolerance of the membrane towards hydraulic pressure and is deter- where ε is the surface porosity;  is the viscosity of the permeate
mined by subjecting the membranes to varied pressure of 345, 275, (g/cm s); Jw is the pure water flux (cm/s) and P is the applied
207, 138, 69 kPa, respectively. It is evaluated from the inverse of pressure (dyn/cm2 ).
the slope of transmembrane pressure vs. pure water flux. It is also From the known values of ε and R (in cm), the pore density in
calculated by the equation: the membrane surface can be calculated using equation:
p
Rm = (3) ε
Jw n= 2
(7)
×R
where Rm is the membrane hydraulic resistance (kPa/l m−2 h−1 ) and
p is the pressure difference (kPa). where n is number of pores/cm2 .
G. Arthanareeswaran et al. / Separation and Purification Technology 64 (2008) 38–47 41

2.5. Mechanical stability (Jw1 − Jw2 )


rir = (11)
Jw1
The tensile intensity, tensile stress and break elongation ratio of
Obviously, rt was the sum of rr and rir :
the membranes were examined to investigate the mechanical sta-
bility. Two dumbbell shaped specimen of 5 mm wide and 10 mm rt = rr + rir (12)
long, were punched out of the membrane film. Mechanical studies
In order to test the recycling potential of CA/SiO2 membranes,
such as tensile strength, tensile stress and percentage break elonga-
three repetitive UF operations were carried out using BSA solution
tion ratio were carried out using Instron 4500 model tensile testing
at 345 kPa. The BSA solutions of 0.1 wt.% were prepared by dissolv-
system at an extension rate of 2 mm/min, after fixing the samples
ing in (0.5 M, pH 7.2) phosphate buffer that were used as standard
in the holders [13,18].
feed solutions. This recycling process includes four times run of
pure water flux and three times run of BSA solution flux using dead
2.6. Morphological studies end UF experiment cell. The water cleaning processes were carried
out after every time of BSA solution flux, and again the pure water
The top surface and cross-sectional morphology of the flux of cleaned membranes was measured. The process was stopped
CA/SiO2 blend membranes were studied using Scanning Electron with four runs, because the BSA molecules foul on the membrane
Microscopy (SEM) (JEOL JSM-6360). The membranes were cut into surface and the flux remained constant.
pieces of various sizes and mopped with filter papers and immersed
in liquid nitrogen for few seconds and were frozen. The sam- 3. Results and discussion
ples were mounted on the platinum sputtered sample holders to
provide electrical conductivity to very thin layers of polymeric 3.1. Compaction
membranes and photomicrographs were taken in very high vac-
uum conditions operating at 20 kV with a magnification of 5000× In this investigation, CA was modified by blending with SiO2 at
for top surface and 250× for cross-section of the membranes. varying composition from 0 to 40 wt.%. The effect of SiO2 particle
of blend membranes can be visualized from Fig. 2 where the
2.7. Protein separation studies graph is plotted between compaction time vs. pure water flux of
blend membranes. The initial flux that was collected after 20 s of
The protein rejection study was carried out using the protein pressurization, showed a value of 38.95 l m−2 h−1 for 100 wt.% CA.
solutions of different molecular weight such as trypsin (20 kDa), As the compaction time increased, the value of PWF decreased to
pepsin (35 kDa), egg albumin (45 kDa) and BSA (69 kDa). The exper- 20.54 l m−2 h−1 after 5 h and attained the steady state value, which
imental pressure was maintained at 345 kPa to carry out the further confirms the completion of compaction process. This is due to the
study. The protein of lowest molecular weight such as trypsin was fact that during compaction of membrane under pressure, the reor-
used first in order to reduce the fouling of the membranes with ganization of polymeric chains occurs which leads to the change
higher molecular weight protein molecules. The % solute rejection in the membrane structure with the formation of uniform pores.
was found for each protein solution using the prepared membranes When the concentration of SiO2 particle added to CA increased
individually and was calculated using the Eq. (4) and their individ- from 0 to 40 wt.%, the initial flux was increased from 38.95 to
ual protein flux was also calculated using the Eq. (1). 140.22 l m−2 h−1 . This increase in trend may be due to the increased
free volume between CA and SiO2 in the polymer matrix which
2.8. Fouling-resistant ability and recycling of blend membranes resulted in the formation of pores on the membrane surface. Simi-
lar results were obtained by Yan et al. [18] for the alumina particles
After 4 h of ultrafiltration, the membranes were washed with in the PVDF blend membranes. The steady state flux of 100 wt.%
deionized water for 20 min and the water flux of the cleaned mem- CA and 40 wt.% SiO2 in the CA blend membrane resulted in an
branes was measured (Jw2 ) at 345 kPa. In order to evaluate the increased value from 20.54 to 59.204 l m−2 h−1 . Arthanareeswaran
fouling-resistant ability of the blend membranes, flux recovery et al. [26] reported similar steady state observations for CA/PEG
ratio (FRR) was introduced [25] and calculated using the following
expression:

Jw2
%FRR = × 100 (8)
Jw1

To analyze the fouling process in details, we defined several


ratios to describe the fouling-resistant ability of the blend mem-
brane. The first ratio was rt as in equation:

Jp
rt = 1 − (9)
Jw1

Here, rt was the degree of total flux loss caused by total fouling.
rr and rir were also defined to distinguish reversible fouling and
irreversible fouling. Reversible fouling ratio (rr ) describes the foul-
ing caused by concentration polarization and irreversible fouling
ratio (rir ) describes the fouling caused by adsorption or deposition
of protein molecules on the membrane surface. They are defined
by equations:

(Jw2 − Jp )
rr = (10)
Jw1 Fig. 2. Effect of compaction on pure water flux of CA/SiO2 blend UF membrane.
42 G. Arthanareeswaran et al. / Separation and Purification Technology 64 (2008) 38–47

Table 2
Effect of SiO2 concentration on CA blend membranes

Polymer (17.5 wt.%) Solvent (wt.%) PWF at 345 kPa (l m−2 h−1 ) Water content (%) Rm (kPa/l m−2 h−1 ) MWCO (kDa)

CA SiO2 DMF

100 0 82.5 15.58 76.59 21.65 20


90 10 82.5 35.83 78.75 9.37 35
80 20 82.5 42.07 79.54 8.22 45
70 30 82.5 45.18 81.32 7.54 45
60 40 82.5 46.74 82.56 7.15 45

blend ultrafiltration membranes. It is believed to form uniform, membrane to the feed flow. From Table 2, it is evident that the pure
rigid pores during compaction on the membrane surface and to 100 wt.% CA exhibited a higher value of 21.65 kPa/l m−2 h−1 due to
obtain a steady state flux. Similar trend was obtained for the mem- low porosity of the CA membrane. In the CA/SiO2 blend membranes,
branes with varying composition of 90/10, 80/20 and 70/30 wt.% when the concentration of SiO2 increased with 10 wt.% increments
of CA/SiO2 blend composition. Further, if any trace quantity of to a maximum of 40 wt.% in the casting solution, Rm value gradu-
solvent present in the membrane surface, which is pass through ally decreased to 7.15 kPa/l m−2 h−1 . The introducing of silica in the
the membrane along with the permeate during compaction. polymer chains is expected to be effective in promoting interface
void formation during the membrane preparation. The enhanced
3.2. Pure water flux (PWF) pore size of the CA/SiO2 membrane due to the extended free vol-
ume between the polymer chains by addition SiO2 resulted in the
After completion of compaction, the pure water flux of blend decrease of Rm value. Yang et al. [13] studied PSf/TiO2 blend mem-
membranes were studied at 345 kPa pressure and shown in Table 2. branes in terms of numerous pores on the membrane surface, which
The low pure water flux value of 15.58 l m−2 h−1 is obtained for is inversely proportional to the water flux of the respective mem-
100 wt.% CA membranes. This is because for the pure 100 wt.% CA, branes. Similar decrease in Rm values was obtained for the other
a very tight polymer matrix is formed, thus the pores formed on blend membranes with varying composition of CA/SiO2 as 90/10,
the membrane surface will be in smaller size. When the concentra- 80/20, 70/30 wt.%.
tion of SiO2 is increased from 10 to 40 wt.% in the pure CA blend
membrane, the PWF showed an increased value from 35.83 to 3.5. Molecular weight cut-off (MWCO)
46.74 l m−2 h−1 .
The increase in the pure water flux may be due addition of fumed The MWCO of a membrane corresponds to molecular weight of
silica particles to CA has been confirmed to systematically increase solutes that has the rejection of more than 80% [23]. MWCO of all
the average size of the free volume. Similar observations were made the membranes of varying composition of CA/SiO2 was determined
by He and Morisato [27], non-porous, nano-size fumed silica, was individually based on the study of protein rejection using proteins
incorporated into a PMP matrix. of different molecular weight and tabulated in Table 2. MWCO of the
100 wt.% CA is compared with the blend membranes with compo-
3.3. % Water content sition 90/10, 80/20, 70/30 and 60/40 wt.% of CA/SiO2 . It is evident
from Table 2 that the MWCO of the pure 100 wt.% CA membrane
Water content of the membrane relates hydrophilicity of the is 20 kDa. This is because the membrane formed of pure CA has
membrane [28]. The % water content was calculated using the Eq. smaller pores size due to the tight polymer matrix. Malaisamy et al.
(2) and the value of the membranes were shown in Table 2. From [30] obtained similar results for CA/Sulfonated polysulfone blend
Table 2, the % water content of 100 wt.% CA membrane showed a membranes. When considering the MWCO of the newly formed
value of 76.59%. In case of 60/40 wt.% of CA/SiO2 blend membrane, blend membrane of 90/10 wt.% CA/SiO2 showed a higher value of
the percent water content was 82.56%. When the 40 wt.% SiO2 par- 35 kDa. With the further increase of 40 wt.% SiO2 to CA membrane,
ticles were present in the pure CA membrane, the water content of the MWCO further increased to 45 kDa. The increase in MWCO of
blend membrane was increased. Similar increase in water content the blend membranes may be due to the increase in free volume and
was interpreted by Lv et al. [12] when small amount of Pluronic a decrease in polymer chain segmental mobility by the presence of
F127 was incorporated into CA membranes. The increased water silica material in the polymer casting solution [31].
content may be due to the detachment of polymer chains from
the silica surface, causing the interface voids. Further, this leads 3.6. Pore statistics
to increase in void volume resulting in the formation of bigger
size pores on the membrane surface and increase the water uptake Pore statistics includes average membrane pore radius (Å), sur-
in the pores. Similar results were obtained by Yan et al. [18] for face porosity, and pore density. Average pore radius (R) of the
PVDF/Al2 O3 blend membranes which reveal that with the addi- membrane is found from the % solute rejection using the protein
tion of Al2 O3 particle to the PVDF blend membrane, the contact solutions prepared from trypsin, pepsin, EA, and BSA of 0.1 wt.%
angle decreased which showed an increase in hydrophilicity of the in phosphate buffers. The pure 100 wt.% CA membrane was com-
blend membranes which confirms an increased uptake of the water pared with the CA/SiO2 blend composition of 90/10, 80/20, 70/30,
through the membrane pores. and 60/40 wt.%. It is observed that for pure 100% CA, the value of
R is found to be 26.88 Å. The relation between the average mem-
3.4. Membrane hydraulic resistance (Rm ) brane pore radius and the concentration of silica in CA is plotted
in Fig. 3(a) and observed that the membrane average pore radius
In order to determine the membrane hydraulic resistance, the is 41.25 Å when the concentration of SiO2 is increased to 40 wt.% in
pure water flux was measured by varying the transmembrane pres- the CA membrane [16].
sure of the system to 414, 345, 276, 207, 138, 69 kPa. The value of Rm Surface porosity (ε) gives the detail about total pore area per
is calculated by Eq. (3), which reveals the resistance offered by the unit surface area of the membrane and calculated using Eq. (6).
G. Arthanareeswaran et al. / Separation and Purification Technology 64 (2008) 38–47 43

The value of ε is found to be 5.72 × 10−5 when the concentration


of SiO2 is 0 wt.% in CA membrane which is observed from Fig. 3(b).
The ε value showed comparatively a higher value of 10.89 × 10−5 for
60/40 wt.% CA/SiO2 blend membrane. It was concluded that silica
particles yield more polymer/particle interfacial area and provide
more opportunity to disrupt polymer chain packing [29]. The result
on membrane pore density of CA/SiO2 blend membrane is shown in
Fig. 3(c) where the graph is drawn between the silica concentration
vs. pore density (number of pores/cm2 ) and calculated using Eq.
(7). When 100 wt.% CA membrane is compared with the 60 wt.%
CA blend membranes, the pore density showed an increased value
from 6.77 × 109 to of 8.4 × 109 which is apparent from the graph
mentioned above. While adding silica (40%) to the casting solution
enhances the pore density (8.4 × 109 ) caused by the silica particles
aggregate phenomenon, which leads to a pure water flux (Fig. 2).
These results demonstrate that adding if SiO2 particles to CA matrix
can improve its porosity and increase the number of small pores.
The effect of inorganic materials on polymer matrix in terms of pore
statistics was investigated elsewhere [18,32].

3.7. Mechanical stability

The testing results of mechanical stability including tensile


intensity, tensile stress and break elongation ratio are listed in
Table 3. It is clear that the tensile intensity, tensile stress and
break elongation ratio of pure 100 wt.% CA membrane is 3.07 N,
2.05 N/mm2 , 8.84%, respectively. The mechanical stability enhances
with the increase of the filler concentration, especially, at 10 wt.%
SiO2 concentration to CA membrane, tensile intensity, tensile stress
and break elongation ratio reach a peak value of 3.79 N, 2.53 N/mm2 ,
16.29%, respectively, and then decline with the further increase
of filler concentration. These trend can be appreciated that the
facts caused by addition of SiO2 particles such as the thicker skin-
layer, the suppression of macrovoids, and the interaction between
inorganic filler and polymers, result in the increase of mechanical
strength of membrane. However, an excessive filler concentration
can cause the particles aggregation and make them not to disperse
uniformly in polymeric matrix, which will decline the membrane
elasticity, accordingly leading to weaken the mechanical stability of
membrane. Yang et al. [13] made similar observations for PSf/TiO2
blend UF membranes. Further, Yan et al. [18] also observed the
mechanical stability in terms of tensile intensity and break elon-
gation ratio for Al2 O3 /PVDF blend UF membranes.

3.8. Morphological studies

To attain a better performance of membranes, the morphologi-


cal structure of the membrane has to be manipulated. Here in this
study, the morphology of the membrane top surface and cross-
section of 100 wt.% CA is compared with the membranes made of
80/20 and 60/40 wt.% of CA/SiO2 blend composition observed at a
magnification of 5000× and 250× using SEM. The top surface of
pure 100 wt.% CA membrane is observed from Fig. 4(a) and seen
that with 0 wt.% of SiO2 in blend composition, the smooth mem-
brane surface is formed and the smaller size pores are formed.
Similar observations were made by Lv et al. [12] for novel blend
membranes with CA and Pluronic F127 using phase separation tech-
nique. Fig. 4(b) and (c) shows that the aggregates SiO2 particles are
packed by polymers or adsorbing on the surface of CA/SiO2 blend
Fig. 3. (a) Effect of SiO2 concentration on average pore radius of CA/SiO2 blend UF membrane. When the concentration of SiO2 content increased to
membranes; (b) effect of SiO2 concentration on surface porosity of CA/SiO2 blend UF 20 wt.% in CA, the number of pores on the surface also increased.
membranes; and (c) effect of SiO2 concentration on pore density of CA/SiO2 blend Similarly, with the further addition of 40 wt.% SiO2 to CA mem-
UF membranes.
brane, the membrane top surface is clearly visible with increased
pore density. The higher silica concentration (40 wt.%) induces an
44 G. Arthanareeswaran et al. / Separation and Purification Technology 64 (2008) 38–47

Table 3
Mechanical stability of the CA/SiO2 blend UF membranes

Composition (17.5 wt.%) Solvent (wt.%) Tensile intensity (N) Tensile stress (N/mm2 ) Break elongation ratio (%)

CA SiO2 DMF

100 0 82.5 3.07 2.05 8.84


90 10 82.5 3.79 2.53 16.29
80 20 82.5 3.32 2.21 13.25
70 30 82.5 2.45 1.63 10.16
60 40 82.5 2.01 1.38 8.01

apparent aggregate phenomenon, and produces a number of large


surface pores (see Fig. 3(b)). These facts indicate that there exist
interfacial stresses between polymer and silica particles, which
build up and finally are comfortable by forming interfacial pores
due to the shrinkage of organic phase during the demixing pro-
cess. Lin et al. [33] made similar observations when the lithium
chloride was added to the PVDF membrane. These facts can be con-
firmed from the pore statistics of the blend membranes of varying
composition of CA/SiO2 as described above.
The cross-section of the CA/SiO2 blend membranes was
observed at a magnification of 250×. Fig. 5(a–c) shows the cross-
section of CA/SiO2 blend membranes that with the increase in the
SiO2 concentration, the asymmetric structure of the 100 wt.% CA
membrane (Fig. 5(a)) becomes faint, the thickness of skin layer
increases with the addition of SiO2 of 20 wt.% in CA (Fig. 5(b)) and
the structure of sub-layer undergoes a transition from finger like
to sponge like structure for 60/40 wt.% of CA/SiO2 blend membrane
(Fig. 5(c)). This shows the distribution of the SiO2 on the mem-
brane surface that forms finger like pores linked by the sponge walls
which confirmed that a large number of micropores allow the fin-
ger like pores to communicate with each other. The similar results
were observed by Arthanareeswaran et al. [9] for CA/SPEEK blend
membranes.

3.9. Separation of proteins

3.9.1. Rejection of protein


The rejection of proteins like BSA, EA, pepsin and trypsin were
studied separately with the blend membranes of CA/SiO2 of com-
position 100/0, 90/10, 80/20, 70/30 and 60/40 wt.%. The proteins for
the rejection study was used in the increasing order of their molec-
ular weight hence, the lowest molecular weight trypsin was used
first to avoid the blocking of the pores by higher molecular weight
protein molecules that might lead to fouling of the membranes.
The protein rejection of blend composition of 90/10, 80/20, 70/30
and 60/40 wt.% of CA/SiO2 membranes were compared with pure
100 wt.% CA membrane. The efficiency of protein rejection on blend
composition of membranes is shown in Fig. 6 where the graph is
plotted between silica concentration vs. % solute rejection. From
the graph it is evident that the rejection for 100 wt.% CA membrane
showed a value of 94% for BSA and 80% for trypsin. The higher rejec-
tion of protein may be due to the higher molecular weight of the
BSA solutes (69 kDa) when compared with lower molecular weight
trypsin solutes (20 kDa). Malaisamy and Mohan [34] interpreted a
similar increase in protein rejection with higher molecular weight
(BSA) for CA/SPSf blend UF membranes. As the composition of the
SiO2 content increased from 10 to 40 wt.% in the casting solution,
the % solute rejection of BSA decreased from 92 to 81%. Similarly,
a decreasing trend was obtained for all other blend membranes
with other proteins like EA, pepsin, and trypsin, respectively. This
may be due to the increased inhomogeneity between the polymer
Fig. 4. (a) SEM of the top surface view of pure CA membrane; (b) SEM of the top
matrices, resulting in the formation of aggregate pores in the mem-
surface view of 80/20 wt.% of CA/SiO2 blend UF membrane; and (c) SEM of the top branes. Bottino et al. [16] obtained a similar decreasing trend of
surface view of 60/40 wt.% of CA/SiO2 blend UF membrane. protein rejection for PES blend membrane.
G. Arthanareeswaran et al. / Separation and Purification Technology 64 (2008) 38–47 45

Fig. 6. Effect of SiO2 concentration on % solute rejection of CA/SiO2 blend UF mem-


branes.

meate flux may be due to the fact that solutes of lower molecular
weight pass through the smaller pore size membrane. When the
concentration of the SiO2 is decreased to 40 wt.% in the casting solu-
tion, the permeate flux for the given protein (e.g. BSA) increased to
19.79 l m−2 h−1 . The addition of fumed silica in CA pushes the per-
meability of the CA membrane towards a bigger pore size with high
permeability. Further incorporation of a small volume fraction of
inorganic fillers into the polymer matrix can result in a significant
increase in overall separation efficiency. The newly formed blend
membranes of 90/10, 80/20, 70/30 wt.% of CA/SiO2 showed a similar
increase in protein flux when other proteins like EA, pepsin, trypsin
were used individually. Sivakumar et al. [28] obtained decreased
protein flux when proteins of increasing molecular weights were
used for CA/PSf blend membrane.

3.10. Analysis of membrane fouling during ultrafiltration

Under constant pressure, the effects of membrane fouling and


concentration polarization were usually observed by a considerable
decline in the flux with time. In the present work, the concentration

Fig. 5. (a) SEM of the cross-sectional view of pure CA membrane; (b) SEM of the
cross-sectional view of 80/20 wt.% of CA/SiO2 blend UF membrane; and (c) SEM of
the cross-sectional view of 60/40 wt.% of CA/SiO2 blend UF membrane.

3.9.2. Permeate flux of protein


Protein flux is the measure of the product rate of the membrane
for given protein solution. The effect of varying SiO2 concentration
in the blend composition on the protein flux is shown in Fig. 7. The
permeate flux of pure 100 wt.% CA showed the lower value when
compared with the membranes made of increasing SiO2 concen-
tration in CA membrane. The permeate flux of 100 wt.% CA showed
a value of 5.19 l m−2 h−1 for BSA and with other proteins showed
comparatively higher permeate flux of 8.72, 11.95, 14.17 l m−2 h−1 Fig. 7. Effect of SiO2 concentration on protein permeate flux of CA/SiO2 blend UF
for EA, pepsin and trypsin, respectively. The increase in the per- membranes.
46 G. Arthanareeswaran et al. / Separation and Purification Technology 64 (2008) 38–47

Table 4
The total fouling resistance (rf ), reversible resistance (rr ), irreversible resistance (rr ) and FRR value of different blend membranes using BSA solution

Composition (17.5 wt.%) Solvent (wt.%) Jw1 (l m−2 h−1 ) Jw2 (l m−2 h−1 ) Jp (l m−2 h−1 ) Fouling resistance ratio FRR %

CA SiO2 DMF rr rir rf

100 0 82.5 15.58 9.50 5.19 0.37 0.39 0.76 61


80 20 82.5 42.07 29.02 14.15 0.35 0.31 0.66 69
60 40 82.5 46.74 35.06 19.79 0.38 0.25 0.63 75

polarization was minimized because of the high molecular weight 100 wt.% CA membrane, the initial water flux was 15.58 l m−2 h−1 .
of BSA molecules and rigorous stirring near the membrane surface. As the operation time increased, the water flux reached a steady
Therefore, the membrane fouling mostly caused the flux decline of state value of 9.1 l m−2 h−1 after three runs of BS solution flux. When
the membranes. Membrane fouling is caused by the adsorption or the concentration of SiO2 content in the CA membrane is increased
deposition of the protein molecules on the membrane surface. The to 40 wt.%, the initial water flux increased to 46.74 l m−2 h−1 . The
flux changes due to the fouling and cleaning of BSA were observed increase in permeation-flux of the CA/SiO2 blend membrane was
for 100/0, 80/20 and 60/40 wt.% of CA/SiO2 blend membranes is due to increase in the surface hydrophilicity and the efficient fil-
shown in Table 4. As seen from Table 4, the irreversible fouling tration area due to the addition of inorganic SiO2 particles [19].
ratio of 100 wt.% CA membrane is 0.39 and this value decreased After four runs of water flux and three runs of BSA solution flux,
to 0.25 with 40 wt.% of SiO2 in CA membrane which indicates that the water flux reached a steady state value of 27 l m−2 h−1 . This
the fouling of BSA on the membrane surface decreased. This may decrease in flux may be due to the fact that BSA particles foul the
be due to the fact that with the addition of inorganic particle to membrane surface. From the graph, it is evident that pure 100 wt.%
CA membrane, the hydrophilicity of the CA membrane is enhanced CA membrane can be used for two runs of BSA solution flux and
which reduces the interactions of the BSA particles on membrane 60/40 wt.% of CA/SiO2 blend membrane can be used for three runs
surface [19]. of BSA solution flux. This increase may be due to the increased
Since long time operation with the product rate was of great hydrophilicity of CA/SiO2 blend membranes that reduces the mem-
importance for practical application of membranes, FRR was intro- brane fouling [18]. The outstanding flux recovery property of blend
duced to evaluate the recycling property of the blend membranes. membranes indicates the long-run utilization and operation reli-
The FRR values for the CA/SiO2 blend membranes were calculated ability. Wang et al. [25] made similar observations for PES blend
using Eq. (8) and are presented in Table 4. It is observed that the membranes.
FRR values increased with the increase of SiO2 content in the casting
solution, and the maximum FRR value reached 75% for 60/40 wt.% 4. Conclusions
of CA/SiO2 blend UF membrane. This may be due to the increased
hydrophilicity [18] with addition of SiO2 content in CA blend mem- In this study, CA/SiO2 UF membranes were formed by phase
brane. inversion technique using DMF as solvent. The concentration of
SiO2 particles increased from 0 to 40 wt.% to CA membrane, above
3.11. Recycling of the blend membranes 40 wt.% did not form membrane. The prepared membranes were
studied for the UF performance such as compaction, PWF, mem-
The excellent flux recovery property of blend membranes will brane hydraulic resistance and percent water content. The UF
provide the long time membrane run without significant decline performance of CA/SiO2 membranes illustrates that the pure water
of separation performances. To study the recycling property of the flux and water content were increased, while the membrane
CA/SiO2 blend membranes, the graph is plotted between operation hydraulic resistance was decreased, as the concentration of SiO2
time vs. flux and the results are shown in Fig. 8. In case of pure in the casting solution is increased.
The MWCO was increased from 20 to 45 kDa with the increase
from 0 to 40 wt.% SiO2 concentration in the casting solution. The
gradual increase in average membranes pore radius, surface poros-
ity and pore density of CA/SiO2 membranes were observed when
compared with pure CA membrane. The mechanical stability of
the CA/SiO2 blend UF membranes were found using tensile testing
machine and observed that with the presence of inorganic parti-
cle of 10 wt.% in the casting solution, the mechanical stability in
terms of tensile intensity, tensile stress and break elongation ratio
increased, respectively. With the further increase of SiO2 particles
in the casting solution resulted in the decrease of the mechanical
stability of the UF membrane. The morphological studies were con-
ducted using SEM. It is observed that the pore formation increased
with the addition of inorganic grains (SiO2 ) to the polymer (CA)
membrane and change in morphological structure also observed.
The separation of protein studies also investigated using prepared
pure CA and CA/SiO2 UF membranes and observed that the rejec-
tion of protein molecules decreased with the enhanced permeate
flux as the concentration of SiO2 particles in the casting solu-
tion increased. The SiO2 content in CA/SiO2 blend membranes
enhanced the fouling-resistant ability thereby improving the recy-
Fig. 8. Flux variation of CA/SiO2 UF membranes during three times of BSA aqueous
cling ability of the CA/SiO2 blend membrane with increased FRR
solution. value.
G. Arthanareeswaran et al. / Separation and Purification Technology 64 (2008) 38–47 47

References [19] L. Yan, Y.S. La, C.B. Xiang, S. Xianda, Effect of nano-sized alumina particle
addition of PVDF UF membrane performance, J. Membr. Sci. 276 (2006) 162–
[1] M. Cheryan, Ultrafiltration, Microfiltration Handbook, Technomic Pub. Co. Inc., 167.
Lancaster, PA, USA, 1998. [20] M. Sivakumar, D. Mohan, R. Rangarajan, Preparation and performance of Cel-
[2] R. Mahendran, R. Malaisamy, D. Mohan, Cellulose acetate and polyether sul- lulose acetate-polyurethane Blend membranes and their applications. Part I,
fone blend ultrafiltration membranes. Part I. Preparation and characterizations, Polym. Int. 47 (1998) 311–316.
Polym. Adv. Technol. 15 (2004) 149–157. [21] P.V. Witte, P.J. Dijkstra, J.W.A. Vanden Berg, J. Feijen, Phase separation process in
[3] O. Kuttowy, S. Sourirajan, Cellulose acetate UF membranes, J. Appl. Polym. Sci. polymer solution in relation to membrane formation, J. Membr. Sci. 117 (1996)
19 (1975) 1449–1460. 1–31.
[4] B. Kunst, D. Skevin, G. Dezolic, J.J. Peters, A light scattering and membrane [22] M. Sivakumar, A.K. Mohansundaram, D. Mohan, K. Balu, R. Rangarajan, Modifi-
formation study on concentrated CA solution, J. Appl. Polym. Sci. 20 (1976) cation of CA: its characterization and application as an UF membranes, J. Appl.
1339–1353. Polym. Sci. 67 (1998) 1939–1946.
[5] S. Prabhakar, B.M. Misra, Studies on structural, kinetic and thermodynamic [23] M.N. Sarbolouki, General diagram for estimating pore size of UF and RO mem-
parameters of CA membranes, J. Membr. Sci. 29 (1986) 143–153. branes, Sep. Purif. Technol. 17 (1982) 381–386.
[6] R.E. Kesting, Synthetic Polymeric Membranes, McGraw Hill, New York, 1971. [24] J. Brinck, A.S. Jönsson, B. Jönsson, J. Lindau, Influence of pH on the adsorptive
[7] G.M. Shashidhara, K.H. Guruprasad, A. Varadarajulu, Miscibility studies on fouling of ultrafiltration membranes by fatty acid, J. Membr. Sci. 164 (2000)
blends of cellulose acetate and nylon 6, Eur. Polym. J. 38 (2002) 611–614. 187–194.
[8] C.J. Sajitha, R. Mahendran, D. Mohan, Studies on cellulose acetate–carboxylated [25] Y.Q. Wang, Y.L. Su, X.L. Ma, Q. Sun, Z.Y. Jiang, Pluronic polymers and
polysulfone blend ultrafiltration membranes––Part I, Eur. Polym. J. 38 (2002) polyethersulfone blend membranes with improved fouling-resistant ability
2507–2511. and ultrafiltration performance, J. Membr. Sci. 283 (2006) 440–447.
[9] G. Arthanareeswaran, K. Srinivasan, R. Mahendran, D. Mohan, M. Rajendran, V. [26] G. Arthanareeswaran, P. Thanikaivelan, J. Abdoul Rabuime, M. Raajenthiren, D.
Mohan, Studies on cellulose acetate and sulfonated poly (ether ether ketone) Mohan, Metal ion separation and protein removal from aqueous solutions using
blend ultrafiltration membranes, Eur. Polym. J. 40 (2004) 751–762. modified cellulose acetate membranes: role of polymeric additives, Sep. Purif.
[10] Z. Chen, M. Deng, Y. Chen, G. He, M. Wu, J. Wang, Preparation and performance Technol. 55 (2007) 8–15.
of cellulose acetate/polyethyleneimine blend microfiltration membranes and [27] P.Z. He, A. Morisato, Nanostructured poly (4-methyl-2-pentyne)/silica hybrid
their applications, J. Membr. Sci. 235 (2004) 73–86. membranes for gas separation, Desalination 146 (2002) 11–15.
[11] R. Mahendran, R. Malaisamy, D. Mohan, Preparation, Characterization and effect [28] M. Sivakumar, D. Mohan, R. Rangarajan, Studies on cellulose acetate-
of annealing on performance of cellulose acetate/sulfonated polysulfone and polysulfone ultrafiltration membranes II. Effect of additive concentration, J.
cellulose acetate/epoxy resin blend UF membranes, Eur. Polym. J. 40 (2004) Membr. Sci. 268 (2006) 208–219.
623–633. [29] S.B.T. Ersolmaz, Effect of zeolite particle size on the performance of
[12] C. Lv, Y. Su, Y. Wang, X. Ma, Q. Sun, Z. Jiang, Enhanced permeation performance polymer–zeolite mixed matrix membrane, J. Membr. Sci. 175 (2000) 285–
of cellulose acetate ultrafiltration membrane by incorporation of Pluronic F127, 288.
J. Membr. Sci. 294 (2007) 68–74. [30] R. Malaisamy, R. Mahendran, D. Mohan, Cellulose acetate and sulfonated
[13] Y. Yang, H. Zhang, P. Wang, Q. Zhang, J. Li, The influence of nano-sized TiO2 fillers polysulfone blend ultrafiltration membranes. Part II. Pore statistics, molecu-
on the morphologies and properties of polysulfone UF membranes, J. Membr. lar weight cutoff, and morphological studies, J. Appl. Polym. Sci. 84 (2002)
Sci. 288 (2007) 231–238. 430–444.
[14] M.N. Wara, L.F. Francis, B.V. Velamakanni, Addition of alumina to cellulose [31] C.J. Cornelius, E. Marand, Hybrid silica-polyimide composite membranes: gas
acetate membranes, J. Membr. Sci. 104 (1995) 43–49. transport properties, J. Membr. Sci. 202 (2002) 97–118.
[15] R.P. Castro, Y. Cohen, H.G. Monbouette, Silica-supported PVP filtration mem- [32] C.D. Jones, M. Fidalgo, M.R. Wiesner, A.R. Barron, Alumina ultrafiltration mem-
branes, J. Membr. Sci. 115 (1996) 179–190. branes derived from carboxylate–alumoxane nanoparticles, J. Membr. Sci. 193
[16] G. Bottino, V. Capannelli, P. D’Asti, Piaggio, preparation, and properties of novel (2001) 175–184.
organic–inorganic porous membranes, Sep. Purif. Technol. 22 (2001) 269–275. [33] D.-J. Lin, C.-L. Chang, F.-M. Huang, L.-P. Cheng, Effect of salt additive on
[17] S. Liu, R. Highes, Preparation of porous aluminum oxide hallow fibre mem- the formation of microporous PVDF membranes by phase inversion from
branes by a combined phase-inversion and sintering method, Ceram. Int. 29 LiCl4 /Water/DMF/PVDF system, Polymer 44 (2003) 413–422.
(2003) 875–881. [34] R. Malaisamy, D. Mohan, Cellulose acetate and sulfonated polysulfone blend
[18] L. Yan, Y.S. Li, C.B. Xiang, Preparation of PVDF UF membrane modified by nano- ultrafiltration membranes. Part III. Application studies, Ind. Eng. Chem. Res. 40
sized alumina and its antifouling research, Polymer 46 (2005) 7701–7706. (2001) 4815–4820.

You might also like