You are on page 1of 13

Characterization of nanofiltration membranes by the solute transport method:

Some practical aspects in determining of mean pore size and pore size distributions

A. R. Hassan, A. F. Ismail*

Membrane Research Unit, Faculty of Chemical Engineering & Natural Resources Engineering,
Universiti Teknologi Malaysia, 81310 UTM Skudai, Johor Darul Taazim, Malaysia.


Abstract

Mean pore size and pore size distributions of asymmetric nanofiltration
membranes were characterized by using the solute transport methods. The nanofiltration
experiment was conducted using the polyethylene glycol feed solutions at the various
molecular weight. Based on the solute rejection data, the membrane properties in terms of
molecular weight cut-off (MWCO), intrinsic viscosity and solute diameter were
calculated using the Stokes-Einstein equation. The linear correlation between solute
diameter and solute rejection allowed the estimation of the mean pore size, p and
geometric standard deviation, p which were in the ranged from about 1.31 to 2.01 nm
and about 0.08 to 1.02, respectively. The membrane pore size distribution had been
evaluated using the log-normal probability density function at different shear rates. The
Hagen-Poiseulle equation was employed to calculate the surface porosity, S
p
and pore
density, N and in order to study the influences of preparation condition that is shear rate
to the membrane structural properties. In general, the results obtained indicated that the
asymmetric nanofiltration membrane prepared posses an acceptable separation
performance.


Keywords : Mean pore size, pore size distributions, intrinsic viscosity, surface porosity
and pore density.

1. Introduction
The membranes characteristics were based on flux and rejection determined by the
morphology of the skin layer. The flux depends on the pore density and the thickness of
the skin layer while the selectivity depends on the membrane pore size and its
distribution. The membrane separation capabilities were closely depends on the
membrane structural parameters that insight the knowledge of the structural
characteristics was so important in order to control their quality, characteristic and
transport mechanism [1]. It is fundamentally very important to know the structure-
performance relationship for the membrane to achieve any progress in the membrane
technologies [2]. The separation performance of asymmetric nanofiltration membranes
had been described using the theoretical models including of the phenomenological
equation by the irreversible thermodynamic model, steric-hindrance pore (SHP) model
and the Teorell-Meyer-Sievers (TMS) model [3].
________________________________________________________________________
* Corresponding author: Tel:+607-5535592; Fax:+607-5581463; E-mail: afauzi@utm.my

Determination of the membrane molecular weight cut-off (MWCO) and the mean pore
size alone are not sufficient to predict the membrane performance. A large number of
different membrane characterization techniques have been investigated in the last
decades, but none can give any decisive information when they are used individually [2].

There are several well establish techniques to determine the membrane pore size and
pore size distribution. The summarily presented below are the several methods for
determining pore statistic [4,17,18]. The major ones are:
Bubble pressure breakthrough: This method is based on the measurement of the
pressure necessary to blow air through a liquid-filled porous membrane. This
method commonly called as liquid displacement techniques.
Mercury porosimetry: This method used mercury (a non-wetting fluid) to fill a
dry membrane based on the same principles as the bubble pressure method but at
higher pressure.
Electron microscopy: This method uses several electronic microscopy techniques
that are available to view the top and bottom or the cross section of the
membranes, such as SEM (scanning electron microscopy), TEM (transmission
electron microscopy), STEM (scanning transmission electron microscopy) and et.
Atomic force microscopy (AFM): This is a novel technique that allows the
surface study of non-conducting materials, down to the nanometers scale.
Solute transport method: This method is based on the permeation test in terms of
flux and percentage rejection of solute.
Adsorption-desorption methods: This technique used the Kelvin equation, which
relates the reduced vapour pressure of the liquid with a curved surface to the
equilibrium vapour pressure of the same liquid in a plane surface.
Thermoporometry: Based on the fact that the solidification point of the vapour
condensed in the pores is a function of the interface curvature. By using the
differential scanning calorimeter (DSC), the phase transition can be easily
monitored and the pore size distribution calculated.
Permporometry: Based on the controlled blocking of pores by condensation of
vapour, present as a component of a gas mixture, and the simultaneous
measurement of the gas flux through the membrane.
NMR Measurement: The NMR (nuclear magnetic resonance) measurement must
be calibrated using a known pores population materials.

There have been a number of studies in which the relationship between the solute
separation and the size of the solute has been examined in an attempt to obtain the
information about the pore size and its distribution of the membrane [5]. Polyethylene
glycols (PEG) have been widely used for the determination of the nominal molecular
weight cut-off and polyethylene glycol of membranes [6]. This is because of their
colloidal properties are only slightly sensitive to chemical environment giving low-
fouling levels [7]. Cho and co-workers have been proposed the novel technique that is
fractional rejection of nonionic and charged macromolecules for determination of
membranes pore size distributions [6]. Matsuura and co-workers used the polyethylene
glycol (PEG) and polyethylene oxide (PEO) with different molecular weight to
characterize the ultrafiltration membrane and its distributions [5].

Kassotis et al. used dextrans to measure the rejection coefficient of polyacrylonitrile
membranes in order to find the pore size distribution [8]. The solute separation was
dependent on the ratio of solute molecular size to the pore size [9]. Aimar et al. [10]
measured the rejection coefficients of membranes with dextrans and the data were fitted
to a log-normal pores size distributions. Aimar, Meireles, and Sanchez proposed a
method for obtaining the log-normal pore size distribution of ultrafiltration membranes
based upon the normalization of the curves of sieving coefficient against molecular
weight with an experimentally measured solute rejection. Solute rejection was simplified
to a purely steric mechanism and this method was successfully applied in a study of the
changes of pore structure caused by protein fouling [11]. Log-normal probability function
might be used to describe a complete sieving curve for a given membrane from only two
experimental values of the sieving coefficient for two different solutes of known Stokes-
Einstein radius [12].

The complexity of a log-normal distribution function into models of solute transport,
some researchers have adopted a different approach in which pore size distribution
effects are taken into accounted implicitly [13]. Recent work on the theoretical effects of
pore size distributions on uncharged solute transport has attempted to quantify solute
rejection and flux using log-normal and Gaussian distributions [14]. Bowen and Welfoot
[13] had been proposed a detailed analysis of solute transport and membrane properties to
quantify the relative importance of pore radius-dependent nanofiltration separation
phenomena when included in a pore size distribution by using a log-normal and
truncation of the log-normal distribution function.

In this work, the mean pore size was obtained from the solute transport method while
the pore size distribution had determined using the log-normal distribution. The aim of
this paper is to propose some practical aspects in determining of membrane mean pore
size and pore size distribution by combining the experimental data and using the
established theory of the solute transport method.


2. Theoretical background

From the methods reviewed, the solute transport and the gas permeation techniques
were correlated with permeation parameters (gas flux, liquid flux, solute retention) and
allowed to determine the pore size for the pores open to flux, obtaining the minimal size
of the pore con-striction present along the whole pore [1]. These two techniques are very
important in characterizing the thin layer in asymmetric membranes, but not give any
insight into the remaining membrane pore size structure. In this study, the solute transport
method was applied to determine bulk pore size and its distributions.


2.1. Solute transport method

The cellulose acetate flat-sheet membrane prepared at different shear rates were tested
for nanofiltration process using an aqueous solutions containing of polyethylene glycol
(PEG) and was performed with PEG of increasing molecular weights. The feed solute
concentration was 500 ppm. The solute rejection was calculated using the following
equation,

R (%) = % 100 1 x
C
C
f
p
|
.
|

\
|
(1)

where, C
p
and C
f
are the solute concentration in the permeate and feed solution,
respectively. The Eq. 1 is not considered the effect of the concentration polarization.
The Stokes radius of a macromolecule can be obtained from its diffusivity in a solution
by using the following Stokes-Einstein equation [2,5],

D
AB
=
a
kT
6
(2)

where, D
AB
is the diffusivity, k is Boltzmanns constant, is the solvent viscosity and
is the Stokes radius. The diffusivity can also be calculated using the following equation
[5],
a
D
AB
=
| | ( ) { }
3 / 1
6
10 5 . 2
M
kT x
(3)

where, M and | | is the molecular weight and the intrinsic viscosity of the polymer,
respectively. Intrinsic viscosity of the polyethylene glycol (PEG) of known molecular
weight can be calculated using equation as below,

| | = 4.9 x 10
-4
M
0.672
(4)

By combining the Eqs. (2) and (3) the solute diameter was becomes as,

a = 2.122 x 10-8(M| | )
1
(5)
3 /

By substituting the intrinsic viscosities of PEGs at various molecular weight, the Stokes
radius of PEG, a (cm), can be obtained from its molecular weight, M, as follows [1,3,
5,14],

a (PEG) = 16.73 x 10
-10
M
0.557
(6)

to facilitate the calculation of solute diameter, d
s
= 2a (nm), the Stokes-Einstein equation
was rearranged as follow,

d
s
(PEG) = 2 [16.73 x 10
-10
M
0.557
] x 10
7
(7)






2.2. Probability density function

The pore size distribution was evaluated from the real permeation coefficients versus
molecular weight plot (standard selectivity curve) by assuming that the permeation of
solute solution is mainly due to a sieving mechanism but taking into account corrections
attribute to the hindered transport. Pore size distribution of the membrane can be
expressed by the probability density function using the log-normal distribution. It is the
key element for the description of the permeation of organic molecules and the log-
normal model can be adapted by taking the correlation between molecular weight and the
diameter of the molecule into account [16].

The correlation between solute separation and solute diameter according to the log-
normal probability function were and can be expressed as below,

R(%) = erf(z) =

Z
u
e

2
2
1
2
du (8)

where;

z =
g
s s d

ln
ln ln
(9)

and d
s
is the solute diameter, s is the geometric mean diameter of solute at

R (%) = 50%
and g is the geometric standard deviation about the mean diameter. According to the
Eqs. (4) and (5), a straight line in the form of,

F(R%) = A
0
+ A
1
(ln d
s
) (10)

where, A
0
and

A
1
are intercept and the slope, respectively on the log-normal plot. By
ignoring the dependence of solute separation on the steric and hydrodynamic interaction
between solute and pore sizes [5], the mean pore size ( p ) and the geometric standard
deviation ( p ) of the membrane can be considered to be the same as of solute mean size
and solute geometric standard deviation.

From the p and p , the pore size distribution of the nanofiltration membrane can be
expressed by the following probability density function [1,5,15],

(


=
2
2
) (ln 2
) ln (ln
exp
2 ln
1 ) (
p
P p
p p
p
p d
d
dd
d df


(11)

where, d
p
is the pore size.







2.3. Number of pores and surface porosity

The number of pores per unit area, N also known as pore density, can be calculated
from the permeability data of the membrane using the Hagen-Poiseulle equation. Based
on this equation, solvent flux (J
v
) through the pores of the diameter d
i
can be expressed
as,

J
v
=

128
4
P d N i i
(12)

where N
i
is the number of pores (per unit area) having diameter of d
i
, is the length of
the pores, is the solvent viscosity and P is the pressure difference across the pores
[3,5] . From the Eq. (8), the total number of can be obtained by rearranging this equation
as,

N
i
=

max
min
4
128
d
d
i Rd P
J

(13)

where pore length, is considered equivalent to the skin layer thickness of asymmetric
nanofiltration membrane. Similarly, the expression for surface porosity (S
p
), which is
defined as the ratio between the area of pores to the total membrane surface area, can be
derived as [3,5,15],

S
p
= |
.
|

\
|

max
min
2
4
d
d
i Rd
N
x 100% (14)


3. Experimental

3.1. Membranes and solutes

In brief, a 23.5% cellulose acetate, 9.5% formamide and 67.0% acetone were cast by a
simple dry/wet casting technique using our pneumatically-controlled casting machine. In
order to study the influence of shear rates to the mean pore size and pore size
distributions, the asymmetric nanofiltration membranes were cast at various casting speed
and various shear rates (152.00s
-1
- 506.67s
-1
). The membranes were cast on a glass plate
at ambient temperature with casting knife notch of 200 m. An inert nitrogen gas stream
was flushed across the as-cast membrane surface for about 30s to induce forced-
convective evaporation. The membranes were immersed into an aqueous bath and
remained there for 1 day. These asymmetric nanofiltration membranes were then tested
using the PEG solutions at various molecular weights up to 35 000. The solutes feed
concentrations were kept at 500 ppm for this experiment. The test of each fabricated flat
membrane was repeated at minimum of three times to ensure that the resultants were
reproducible [19].

3.2. Preparation of feed solutions

In order to study nanofiltration membranes, aqueous solutions of some linear polymers
could be chosen, such as polyethylene glycol or dextranes. Polyethylene glycols should
be preferred, as their colloidal properties are only slightly sensitive to the chemical
environment giving low-fouling levels. In this work, the polyethylene glycol (molecular
weight up to 35 000) was used as solutes in the feed solution. The feed concentration was
kept at 500 ppm by weight.


3.3. Nanofiltration experiment

In this study, the nanofiltration experiment was conducted using the laboratory
permeation cell with an effective area of 13.2 cm
2
, details of which were described
elsewhere. Prior to testing, the pure water fluxes were measured as in Table 1 and at the
same time were to ensure that the membranes used were stable [2]. The PEG permeated
was collected under an operating pressure of 7 bar and at ambient temperature. The PEG
contents in the feed and in the permeated were measured in terms of total organic carbon
(TOC) by using a total organic analyzer.

Table 1
Membranes pure water fluxes

Shear rates, s
-1
Pure water fluxes (m/s)
152.00 7.28 x 10
-4
217.14 7.66 x 10
-4

304.00 7.41 x 10
-4

506.67 8.02 x 10
-4




3.4. Characterization of membranes pore

The membrane transport mechanism has been described using the theoretical model as
mentioned [2]. The solute transport method was applied by using From PEG testing data,
the solute transport method was applied was to determine the influence of mean pore size
and pore size distributions on structural properties and performance of asymmetric
nanofiltration membranes.


3.4.1 Determination of mean pore size and geometric standard deviation

The solutes rejection for these experimental works is shown in Table 2. Mean pore size
and its distribution can be obtained when the solute separation (%) of a nanofiltration
membrane is plotted versus the solute diameter on the log-normal probability paper.
Based on the solutes rejection in Table 2, the solutes properties in terms of intrinsic
viscosity, Stokes radius and solutes diameter was calculated using the Eqs. 2 - 7 and were
details in Table 3. From the linear correlation of these two parameters, according to the
log-normal probability function in Figure 1, the mean pore size ( p ), can be calculated as
d
s
corresponding to f = 50% and the geometric standard deviation ( p ), can be
determined from the ratio of d
s
at f = 84.13% and at 50%. The results of the mean pore
size ( p ) and the geometric standard deviation ( p ) for the fabricated nanofiltration
membranes were tabulated in Table 4.


Table 2
Experimental solutes separation (%) data obtained at different shear rates

Solutes at different Solutes separation (%) at different shear rates
molecular weight 152.00s
-1
217.14s
-1
304.00s
-1
506.67s
-1


PEG 1000 2.4 3.4 3.9 3.3
PEG 2000 5.0 6.9 8.3 7.7
PEG 6000 23.7 28.7 31.5 28.9
PEG 12000 40.8 67.1 75.1 68.9
PEG 35000 85.6 89.8 95.4 92.3


Table 3
Solutes properties: Intrinsic viscosity, Stokes radius and solutes diameter

Solutes (MW) Intrinsic Stokes Solute diameter,
viscosity radius, (cm) d
s
(nm)

PEG 1000 0.05 7.84 x 10
-8
1.57
PEG 2000 0.08 11.54 x 10
-8
2.31
PEG 6000 0.17 21.28 x 10
-8
4.23
PEG 12000 0.27 31.30 x 10
-8
6.26
PEG 35000 0.55 5.68 x 10
-8
11.37



3.4.2. Pore size distributions (PSD)

Basically, the membranes pore size distributions was closely related to the membranes
pore size, mean pore size and geometric standard deviation. Usually, this distributions
was presented in terms of probability density function, differential and cumulative
distributions. In this study, according to the solutes transport method, the probability
density function was choose as a way to present the membranes pore size distributions for
the prepared asymmetric nanofiltration membranes. Based on the data in Table 4, the
membranes pores size distributions had been calculated and were clearly viewed in a
Figure 2.




Table 4
MWCO, mean pore size and standard deviation at different shear rates

Shear MWCO Mean pore Standard
rates, s
-1
(kDalton) size, p (nm) deviation, p

152.00 1.57 7.84 x 10
-8
1.57
217.14 4.20 11.54 x 10
-8
2.31
304.00 4.85 21.28 x 10
-8
4.23
506.67 4.13 31.30 x 10
-8
6.26




3.4.3. Determination the total number of pores and surface porosity

The membrane pore density and surface porosity could be calculated, respectively,
based on the Eqs. 13 and 14 and summarized in Table 5. The skin layer thickness
substituted in the mentioned above equations were obtained and generated by using the
theoretical model. According to the modeling results, the obtained skin layer thickness
was found to be in the range of about 2.9x10
-5
m to 4.52x10
-5
m [2].

Table 5
Membranes surface porosity and pore density at different shear rates

Shear Surface porosity, Pore density,
rates, s
-1
S
p
(%) N (pores/ m
2
)

152.00 0.241 8.13
217.14 0.252 10.72
304.00 0.297 12.11
506.67 0.231 6.71



4. Results and discussion

The pure water permeation flux (PWP) of the prepared membranes as in Table 1
showed that, the pure water flux is increased with increasing of shear rates. It is
interesting to note that, the prepared asymmetric nanofiltration membranes (at different
shear rates in the range of about 152.00s
-1
to 506.67s
-1
produces both high pure water
permeation fluxes and solute rejection. The solutes rejection data obtained indicated that
asymmetric nanofiltration membrane prepared posses an acceptable separation
performance.

Based on to the solutes rejection in Table 2, the solutes properties in terms of intrinsic
viscosity, Stokes radius and solute diameter were calculated using the Eqs. 2 7 as
mentioned in the theoretical section. In order to measure the values of the geometric
mean pore size and geometric standard deviation, the calculated solutes diameter were
correlated to the solutes rejection data into the log-normal probability paper as shown in
Figure 1. A straight line was obtained with reasonably high correlation coefficient (r
2

0.95). From Figure 1, the above membranes geometric values were determined and were
summarized in Table 4. Mean pore size was the smallest for the membranes fabricated at
shear about 304.00s
-1
while it was the largest for the membranes fabricated at shear rate
of about 152.00s
-1
. The values of mean pore size and geometric standard deviation, were
not much different and were very close to each other for the membranes fabricated at
shear rates of about 217.14s
-1
to 506.67s
-1
. In general, for the prepared asymmetric
nanofiltration membranes, the mean pore size was higher for the membrane having lower
MWCO (Table 4). According to the previous study [2], the mean pore size and geometric
standard deviation for the fabricated membrane at critical shear were found to be at about
2.01 nm and 0.80 nm, respectively.

The probability density function curves were generated from Eq. 11 by using the
values of mean pore size and geometrical standard deviation given in Table 4. The values
of the probability functions were plotted versus to the membranes pore diameter as
shown in Figure 2. PSD 1 was refer to the pore size distribution at shear rate of about
152.00s
-1
while PSD 2 was refer to the pore size distribution at shear rate of about
217.14s
-1
. PSD 3 and PSD 4 were refers to the pore size distributions at shear rates about
304.00s
-1
and 506.67s
-1
, respectively. For example, PSD 3 membrane was showed the
highest pore size distribution at about 0.68 nm
-1
with only about 65% of the pores were
less than 1.0 nm in diameter while for the PSD 2 and PSD 4 membranes, as much as 70%
of the pores were higher than the latter values. On the other hand, for the PSD 1
membrane it is found that 95% of the pores were higher than 1.0 nm in diameter whereas
the pore distribution was about 0.63 nm
-1
. From the graph, we also identified that, at
critical shear rate (304.00s
-1
), the membrane exhibited the highest pore size distribution
and were in good correlation due to the highest solute (sodium chloride) rejection in the
previous study [2,19].

0
10
20
30
40
50
60
70
80
90
100
1 10 100
Solute diameter,d
s
(nm)
S
o
l
u
t
e

r
e
j
e
c
t
i
o
n

(
%
)
Shear rate=152.00s-1
Shear rate=217.14s-1
Shear rate=304.00s-1
Shear rate=506.67s-1

Figure 1: Solutes rejection vs. solute diameter for NF membranes at different shear rates
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
Pore size, d
p
(nm)
P
r
o
b
a
b
i
l
i
t
y

f
u
n
c
t
i
o
n
,

d
f
(
d
p
)
/
d
d
p

(
n
m
-
1
)
PSD 1
PSD 2
PSD 3
PSD 4

Figure 2: Probability density function curve for NF membranes at different shear rates

The membranes pores density and surface porosity was calculated from Eqs. 13 and
14, respectively, and given in Table 5. The skin layer thickness using in this calculation
were in the range of about 2.9x10
-5
m to 4.52x10
-5
m. However, the skin layer thickness
may vary, depending on the membrane preparation conditions. It was found that, as the
pore size is increased the membranes pore density and surface porosity were decreases
until a critical shear is reached. It is also found that, this calculation results were shrunk
beyond the critical shear. Among the prepared nanofiltration membranes, the membrane
prepared at shear rate of 304.00s
-1
had the highest pore density of about 12.11
pores/ m
2
; whereas for the membrane prepared at shear rate of 506.67s
-1
, had the lowest
pore density of 6.71 pores/ m
2
. The membranes surface porosity was also in the same
trend where the membrane at critical shear having the highest surface porosity of about
0.30% and the membrane cast beyond the critical shear was having the lowest surface
porosity at about 0.23%.


5. Conclusion

The solute transport method allowed the determining of the solutes properties (intrinsic
viscosity, Stokes radius and solute diameter) which is an important parameters for the
determination of membrane mean pore size and pore size distributions. This study
showed that, the membranes molecular weight cut-off (MWCO) and geometric mean
pore size were increased with increasing of shear rates. The highest mean pore size (2.01
nm) was found for the membrane prepared at shear rate of about 152.00s
-1
while the
highest pore size distribution is about 0.68 nm
-1
was found to be at the shear rate of about
304.00s
-1
. In general, the experimental results showed that increasing of shear rates
causes the reduction of mean pore size and its distributions. The same trend was occurred
for the membranes pore density and surface porosity where the membrane prepared at
critical shear rate (304.00s
-1
) generated the highest values of pore density and surface
porosity. In this circumstance, the asymmetric nanofiltration membrane could be
prepared at shear rate of about 304.00s
-1
in order to produce high performance
nanofiltration membrane for liquid-liquid separation. The pure water permeation flux and
the solute rejection data achieved showed that the prepared asymmetric nanofiltration
membranes were within the acceptable limit.


References

[1] M. Khayet, and T. Matsuura. Determination of surface and bulk pores sizes of flat-
sheet and hollow fiber membranes by atomic force microscopy, gas permeation and
solute transport methods. Desalination, 158 (2003) 57 64.
[2] A. F. Ismail, and A. R. Hassan. The deduction of fine structural details of asymmetric
nanofiltration membranes using theoretical models. J. Membrane Sci., 231 (2004) 25
36.
[3] T. Gumi, M. Valiente, K. C. Khulbe, C. Palet and T. Matsuura. Characterization of
activated composite membranes by solute transport, contact angle measurement, AFM
and ESR. J. Membranes Sci., 212 (2003) 123 134.
[4] A. Hernandez, J. I. Calvo, P. Pradanos and F. Tejerina. Pore size distributions of
track-etched membranes; comparison of surface and bulk porosities. Colloid and Surface
A, 138 (1998) 391 401.
[5] S. Singh, K. C. Khulbe, T. Matsuura, and P. Ramamurthy. Membrane characterization
by solute transport and atomic force microscopy. J. Membrane Sci., 142 (1998) 111 -
127
[6] S. Lee, G. Park, G. Amy, S. K. Hong, S. H. Moon, D. H. Lee, and J. Cho.
Determination of membrane pore size distribution using the fractional rejection of
nonionic and charged macromolecules. J. Membranes Sci., 201 (2002) 191 201.
[7] N. A. Ochoa, P. Pradanos, L. Palacia, C. Pagliero, J. Marchese, and A. Hernandez.
Pore size distributions based on AFM imaging and retention of multidisperse polymer
solutes: Charactersation of polyethersulfone UF membranes with dopes containing
different PVP. J. Membranes Sci., 187 (2001) 227 237.
[8] J. Kassotis, J. Shmidt, L. T. Hodgins, and H. P. Gregor. Modelling of the pore size
distribution of ultrafiltration membranes. J. Membranes Sci., 22 (1985) 61 - 76.
[9] P. L. Paine, and P. Scheer. Drag coefficient for the movement of rigid sphere through
liquid filler pores. J. of Biophysical, 15 (1975) 1087 1091.
[10] P. Aimar, M. Meireles, and V. Sanchez. A contribution to the translation of retention
curves into pore size distribution for sieving membranes. J. Membranes Sci., 54 (1990)
321 338.
[11] M. Meireles P. Aimar, and V. Sanchez. Effects of protein fouling on the apparent
pore size distribution of sieving membranes. J. Membranes Sci., 56 (1991) 13 - 28
[12] M. Meireles, A. Bessieres, I. Rogissart, P. Aimar, and V. Sanchez. An appropriate
molecular size parameter for porous membranes calibration. J. Membrane Sci., 103
(1995) 105 115.
[13] W. R. Bowen and J. S. Welfoot. Modelling of membrane nanofiltration-pore size
distribution effects. J. Chemical Eng. Sci., 57 (2002) 1393 1407.
[14] S. Mochizuki and A. L. Zydney. Theoretical analysis of pore size distributions
effects on membrane transport. J. Membranes Sci., 82 (1993) 211 227.
[15] M. Khayet, C. Y. Feng and T. Matsuura. Morphological study of fluorinated
asymmetric polyetherimide ultrafiltration membranes by surface modifyimg
macromolecules. J. Membranes Sci., 5502 (2002) 1 - 22.
[16] M. Gholami, S. Nasseri, C. Y. Feng, T. Matsuura and K. C. Khulbe. The effect of
heat-treatment on the ultrafilration performance of polyethersulfone (PES) hollow fiber
membranes. Desalination, 155 (2003) 293 301.
[17] A. Hernandez, J. I. Calvo, P. Pradanos and F. Tejerina. Pore size distributions in
microporous membranes. A critical analysis of the bubble point extended method. J.
Membranes Sci., 112 (1996) 1 - 12..
[18] J. I. Calvo, A. Hernandez, P. Pradanos, L. Martinez and W. R. Bowen. Pore size
distributions in microporous membranes II. Bulk characterization of track-etched filters
by air porometry and mercury porosimetry. J. of Colloid and Interface Sci., 176 (1995)
467 - 478.
[19] A. F. Ismail, A. R. Hassan and B. C. Ng. Effect of shear rate on the performance of
nanofiltration membrane for water desalination. Songklanakarin J. Sci. Technol., 24
(2002) 879 889.

You might also like