You are on page 1of 15

Int. J. Nuclear Desalination, Vol. 1, No.

4, 2005

Direct contact membrane distillation for nuclear


desalination. Part I: Review of membranes used in
membrane distillation and methods for their
characterisation
M. Khayet* and J.I. Mengual
Department of Applied Physics I,
Faculty of Physics, University Complutense of Madrid,
Avda. Complutense s/n, 28040, Madrid, Spain
Fax: +34-91-39445191 E-mail: khayetm@fis.ucm.es
E-mail: mengual@fis.ucm.es
*Corresponding author

G. Zakrzewska-Trznadel
Department of Nuclear Methods in Process Engineering,
Institute of Nuclear Chemistry and Technology,
Dorodna 16, Warsaw, Poland
Fax: +48-22-8111532
E-mail: gzakrzew@orange.ichtj.waw.pl
Abstract: Membrane distillation (MD) is a relatively novel membrane
technology considered by the researchers as a potential method for seawater
desalination. In the first issue of this journal, the possibility of nuclear
desalination by MD was discussed comparing various MD configurations. The
present paper firstly reviews the membranes used in MD and the methods of
their characterisation, among them wet/dry flow method, gas permeation test,
scanning electron microscopy, field emission scanning electron microscopy
and atomic force microscopy. The membrane parameters, which have to be
known before selection of the membranes for some specific applications such
as liquid entry pressure of water, pore size, pore size distribution, porosity and
pore tortuosity, were determined. The knowledge on membranes themselves,
membrane materials and morphology is very important for engineering of
polymer barriers for MD and development of industrial membrane units. The
availability of the industrial MD modules is up to now one of the limitations
for further process implementation.
Keywords: membrane distillation; nuclear desalination; membrane; membrane
characterisation.
Reference to this paper should be made as follows: Khayet, M., Mengual, J.I.
and Zakrzewska-Trznadel, G. (2005) Direct contact membrane distillation for
nuclear desalination. Part I: Review of membranes used in membrane
distillation and methods for their characterisation, Int. J. Nuclear
Desalination, Vol. 1, No. 4, pp.435449.

Copyright 2005 Inderscience Enterprises Ltd.

435

436

M. Khayet, J.I. Mengual and G. Zakrzewska-Trznadel


Biographical notes: Mohamed Khayet graduated from University Caddi
Ayyad, Marrakech, Morocco in 1990 and completed his PhD degree in Physics
at the Department of Atomic, Molecular and Nuclear Physics of University
Complutense of Madrid (UCM), Spain in 1997. He was an assistant professor,
at the Department of Applied Physics I at UCM from 1999 to 2002 and has
been an associate professor at the same university since 2002. He worked in the
Industrial Membrane Research Institute IMRI, Ottawa, Canada, from 2000 to
2001 on a UCM post-doctoral grant. His scientific research interests are in
membrane preparation and characterisation, heat and mass transport
phenomena, membrane processes, separation of organic compounds,
desalination, water and wastewater treatment.
Juan Ignacio Mengual completed his MSc in Physics (1970) and PhD in
Physics (1976) at University Complutense, Madrid (UCM), Spain. He was an
assistant professor and later an associate professor at UCM from 1970-1983.
During this period he spent 2 years as an associate professor at the University
of Valencia, Spain, and later became Director of the Department of Physics at
that university (1986-1989). He has been Professor, since 1983, and Director,
since 1989, at the Department of Applied Physics I, University Complutense,
Madrid, Spain. His research interests are in heat and mass transport
phenomena, desalination and membrane processes such as membrane
distillation, thermo-osmosis and pervaporation.
Grazyna Zakrzewska-Trznadel graduated from Warsaw University of
Technology with MSc in Chemical Engineering and completed her PhD at the
Institute of Nuclear Chemistry and Technology. She was at the Institute of
Nuclear Sciences from 1978 to 1982 and since then has been at the Institute of
Nuclear Chemistry and Technology, Department of Nuclear Methods in
Process Engineering as a senior researcher. Her research interests are in
process engineering and environmental protection problems. Her expertise lies
in membrane permeation processes such as concentration and purification of
hazardous streams, radioactive waste processing, gas separation and separation
of isotopes. She is the author and co-author of over 40 papers, seven patents
and over 80 conference presentations.

Introduction

Membrane distillation (MD) technique has been known for about forty years and is still
being developed. The first MD patent was obtained by Bodell (1963) in 1963 and the
first MD publication was made by Findley (1967) in 1967. In the previous years, interest
in MD is increasing significantly and direct contact membrane distillation (DCMD) is the
configuration most used. DCMD refers to a thermally driven transport of water through
microporous hydrophobic membranes. The membrane is maintained between a hot
solution (i.e., feed side) and cold pure water (i.e., permeate side). Due to the hydrophobic
nature of the membrane, liquid water cannot penetrate inside the dry membrane pores
unless a trans-membrane hydrostatic pressure exceeding the liquid entry pressure of
water (LEPw), which is characteristic of each membrane, is applied. In this manner the
trans-membrane vapour pressure, which is the driving force in MD, is created by
maintaining a temperature difference between both liquids. Under these conditions,
evaporation takes place at the hot feed interface and, after water vapour is transported

Direct contact membrane distillation for nuclear desalination. Part I

437

through the membrane pores, condensation takes place at the cold permeate interface
inside the membrane module.
DCMD is successfully applied for production of high-purity water, concentration of
non-volatile aqueous solutions and removal of trace volatile organic compounds (VOCs)
from dilute wastewater (Lawson and Lloyd, 1997; Mengual and Pea, 1997). In fact,
desalination is the best-known DCMD application. It must be stated that in the case of a
solution with non-volatile components, only water molecules flow through the membrane
pores. Therefore, the retention degree of solutes is very close to 100%. The membrane
acts only as a physical support for the liquidliquid interfaces and the MD separation is
based on the liquidvapour equilibrium state of the feed solution. In a previous paper, the
possibility of nuclear desalination by MD was discussed by comparing various MD
configurations (Khayet, Godino and Mengual, 2003).
It must be pointed out that the first membranes used in MD processes were
commercial membranes manufactured for micro-filtration. This means that there are few
membranes manufactured specifically for MD studies. In other words, a lot of effort has
been devoted in recent years to theoretical MD transport models and less attention has
been paid to the preparation and characterisation of MD membranes (Lawson and Lloyd,
1997). The successful application of MD may be aided by a good knowledge of the
properties of the membranes. For the same commercial membranes used in MD, different
values were associated to the membrane parameters. One of the objectives of this study is
to first give an overview of commercial membranes used in MD and laboratory-made
MD membranes, together with the applied characterisation techniques to obtain the
necessary parameters in MD and comparison between the results obtained from the
different methods. The possibility of application of MD for nuclear desalination warrants
the broad study of the membrane parameters, which are key for the design of big
industrial units.

Experimental method

2.1 Membranes
Commercial micro-porous hydrophobic membranes made of polypropylene (PP),
polyvinylidene fluoride (PVDF) and polytetrafluoroethylene (PTFE, Teflon), available in
capillary or flat sheet forms have been used in MD experiments. Table 1 summarises the
commercial membranes used in MD studies together with their principal characteristics
as specified by the manufacturers. Among the membranes studied are commercial
module units, formed in shell-and-tube, plate-and-frame and spiral wound configurations
that were tested in laboratory and pilot plant experiments. Table 2 shows laboratorymade micro-porous hydrophobic flat-sheet membranes developed in Polish laboratories
together with their experimental characteristics (Chlubek and Tomaszewska, 1989;
Chmielewski et al., 1995; Buczkowski et al., 1998, 2001). Further prepared membranes
for MD may be found mentioned in Khayet and Matsuura (2001) and Khayet, Khulbe
and Matsuura (2004).

438

M. Khayet, J.I. Mengual and G. Zakrzewska-Trznadel

2.2 Membrane characterisation


2.2.1 Liquid entry pressure of water (LEPw)
LEPw, also called wetting pressure, is an important parameter to be determined in MD
due to the fact that the pores must not be wetted by the feed liquid. This is the pressure
that must be applied onto pure water before water penetrates into dry membrane pores.
LEPw depends on the maximum pore size and on the hydrophobicity of the membrane. In
other words, LEPw decreases as the pore size increases and/or the water contact angle
decreases. The procedure and apparatuses used to measure the LEPw of flat-sheet and
hollow-fibre membranes were extensively explained in Khayet and Matsuura (2001) and
Khayet et al. (2002).
Table 1

Commercial membranes commonly used in membrane distillation:


membrane thickness, ; liquid entry pressure of water, LEPw, porosity and ; mean
pore size, dp

Membrane Trade Name

Manufacturer

Material

dp

(m) (m) (%)

LEPw
(kPa)

0.20

282

TF200
Gelman

TF450

PTFE/PP

178

80

1.00

TF1000
GVHP
Millipore

HVHP
FGLP

PVDF b
PTFE/PP

Accurel S6/2
MD020CP2N

Microdyn
Enka A.G.

BFMF 06-30-33 d

Euro-Sep

Sartocon-Mini SM 3031 750701W e

Sartorius

G-4.0-6-7 f
a

AkzoNobel
c

Accurel

Notes:

0.45

GoreTex
Sep GmbH

110

0.22

140

0.45

175

0.20

450

0.2

PP

48
75
70

280
140

Polyolefine
100

0.20

0.22

0.20

80

Flat-sheet polytetrafluoroethylene membranes supported by polypropylene net


Flat-sheet polyvinylidene fluoride membranes
c
Shell-and-tube capillary membrane module: filtration area: 0.1 m2
inner capillary diameter: 1.8 mm; length of capillaries: 470 mm
d
Shell-and-tube capillary membrane module: filtration area : 0.3 m2
inner capillary diameter: 0.33 mm, length of capillaries: 200 mm
e
Plate-and-frame module, dimensions: 138/117/7 mm, filtration area: 0.1 m2
f
Spiral-wound module, SEP Gesellschaft fr Technische Studien, Entwicklung
Planung mbH, filtration area: 4 m2
b

70
200

PTFE

138

Direct contact membrane distillation for nuclear desalination. Part I

439

2.2.2 Mean pore size, pore size distribution and porosity


The applied characterisation techniques of membranes used in MD are physical methods,
which can be divided into two main groups: (i) those which are related to membrane
permeation such as liquid and gas flow tests and (ii) those which permit obtaining the
morphological properties of the membranes such as scanning electron microscopy
(SEM), field emission scanning electron microscopy (FESEM) and atomic force
microscopy (AFM) directly.

Gas permeation test: This method allows determination of the mean pore size (dp)
and the effective porosity (/Lp), defined as the ratio of the porosity and the effective
pore length (Lp) that takes into account the tortuosity () of the membrane pores. Air
and nitrogen were employed as standard gases. The gas permeation flux of each dry
membrane is measured at various trans-membrane pressures. The theoretical
procedure and the experimental set-ups used were reported elsewhere (Khayet and
Matsuura, 2001; Khayet et al., 2002). It must be mentioned that one of the
limitations of this method is the determination of pore size distribution.

Wet/dry flow method: The bubble point together with the gas permeation test known
as the wet/dry flow method can be employed for determining the maximum pore
size, the mean pore size and the pore size distribution of MD membranes. First, the
gas permeation is measured though a dry membrane. Generally a straight line is
observed between the gas permeation and the trans-membrane pressure difference.
Subsequently, the membrane is wetted by a liquid with low surface tension such as
isopropyl alcohol (IPA) and again the gas permeation is measured at different
trans-membrane pressures. In this case the dependence of the gas flux with the
applied trans-membrane pressure is not lineal. At low trans-membrane pressure, the
pores remain filled with IPA and the gas flux is practically zero, whereas for higher
pressure than the bubble point of the membrane, the largest pores will be empty and
the gas flux starts to increase with the pressure because smaller pores are opened
with the increase of the pressure until all pores become empty at the pressure
corresponding to the minimum pore size. In this method, air was used for liquid
displacement through the membrane pores and the theoretical procedure used to
determine the pore size distribution and the mean pore sizes together with the
experimental set-up used may be found in Khayet and Matsuura (2001) and Khayet,
Velzquez and Mengual (2004). It must also be mentioned here that a Coulter
Porometer II (Coulter Electronics Ltd., GmbH) based on wet/dry flow method has
been used for characterisation of the membranes given in Table 2 (Chlubek and
Tomaszewska, 1989; Buczkowski et al., 2001).

Scanning electron microscopy (SEM) provides a very convenient method for


characterising and studying the structure of porous membranes. The images of the
membranes used in MD experiments were performed with Zeiss DSM 942 and
S-4500 HITACHI scanning electron microscopes with magnification from 200 to
90,000. The samples of the membranes were prepared by coating the surface with
gold layer of 200 , in JEOL JEE-4X coater, to avoid charging up of the polymer
surface. SEM is a simple method enabling a clear view of the overall structure of the
membranes: the top and the bottom surfaces as well as the cross section of the

440

M. Khayet, J.I. Mengual and G. Zakrzewska-Trznadel


membrane. Good micrographs can not only allow visualisation of the geometry and
all irregularities of the pores and asymmetries in membrane structure but also enable
the assessment of pore size radius, pore size distribution and surface porosity.
However, one of the limitations of this technique is the heavy metal coating, required
for membrane sample preparation, which gives some artefacts and tends to damage
the membrane surface.

Atomic force microscopy (AFM) is a newly developed high-resolution technique to


study the surface morphology of the membranes. Three-dimensional images of the
membrane surface can be obtained directly without special sample preparation. In
other words, a more true surface structure of a polymeric membrane can be observed
by AFM. The mode of operation of taking AFM pictures, together with the
theoretical approach to obtain the membrane characteristics needed in MD process
(mean pore size, pore size distribution, geometric standard deviation, pore density
and surface porosity), were extensively described in a previous paper (Khayet,
Khulbe and Matsuura, 2004).

Porosity: The ratio between the volume of the pores and the total volume of the
membrane is the membrane porosity, which also called membrane void volume ().
For hydrophobic membranes, can be determined by measuring the density of the
membrane using two types of liquids. A wetting liquid such as isopropyl alcohol
(IPA), which penetrates inside the pores and non-wetting liquid such as water, which
does not enter the pores because of the hydrophobicity of the membrane material. In
this method, a pyknometer and a balance are necessary. The method was explained
in detail in previous papers (Khayet and Matsuura, 2001; Khayet et al., 2002).

Table 2

Laboratory-made membranes and their characteristics determined by Coulter


Porometer II. Mean pore size, dp; minimum pore size, dmin; maximum pore size, dmax;
membrane thickness, ; liquid entry pressure of water, LEPw; porosity, and pore
tortuosity, ,determined from gas permeation measurements
Pore Size (m)

(m)

dmin
(m)

dp
(m)

dmax
(m)

(%)

62

0.16

0.21

0.27

81

2.1 (2.7)

60

0.34

65

(2.2)

Tarflen5

84.5

0.25

59

(2.7)

Tarflen6

72.7

0.19

21

(3.2)

12

0.22

0.28

0.31

10

Membrane Trade Name

Material

Tarflen1
Tarflen2

PTFE a

Estrofol
Notes:

PET
a

PTFE membranes of different porosity, pore diameter and thickness, prepared


in Szczecin University of Technology
b
Polyethylene terephthalate track membrane prepared by heavy ion
bombardment by INCT in collaboration with Dubna
c
Values in brackets were estimated by using dmax

Direct contact membrane distillation for nuclear desalination. Part I

441

2.2.3 Pore tortuosity and membrane thickness


The thickness () of flat-sheet membranes was measured with a micrometre Millitron
(Mahr Feinprf, type 1202 IC), with a precision of 0.1 m, at more than ten spots for
each sample. Due to the fact that the membranes (TF200, TF450 and TF1000) are
supported by backing polypropylene net, the membrane layer was peeled off from the
support and the thickness was measured. It must be pointed out that if the membrane
thickness and the void volume or porosity () are known, the pore tortuosity may be
estimated from the effective porosity (/Lp) determined from the gas permeation test as
reported in Khayet, Khulbe and Matsuura (2004). From the surface porosity (s) obtained
from the wet/dry flow method or from AFM technique, the pore tortuosity can also be
evaluated as stated in Khayet, Khulbe and Matsuura (2004) and Khayet, Velzquez and
Mengual (2004).

Results and discussion

3.1 Results on morphological characterisation of membranes


Tables 3 and 4 show the characteristics of flat-sheet membranes obtained by the methods
cited previously. In Tables 1 and 3 it can be observed that the membranes having small
pore size exhibit higher LEPw values than the membranes having large pore size, and for
membranes with almost identical pore size the LEPw is lower for PVDF membranes
(GVHP, HVHP) compared to PTFE membranes (TF200, TF450). In fact, the LEPw
depends on the maximum pore size and on the hydrophobicity of the membranes
(the membranes made from PTFE polymer are more hydrophobic than those prepared
from PVDF). It is worth pointing out that the discrepancy between the LEPw given by
the manufacturers for Gelman membranes and those given in Table 3 varies between
2.2% for TF200 membrane and 17.2% for TF1000 membrane.
Membrane characteristics: membrane thickness, ; liquid entry pressure of water,
LEPw, porosity, ; mean pore size, dp and effective porosity, /Lp

Table 3

Gas Permeation

Membrane Trade
Name

a
(m)

(%)

LEPw
(kPa)

dp (nm)

/Lp (m-1)

TF200

54.8

68.7

276

198.96

7878.1

TF450

60.0

64.3

149

418.82

7439.0

TF1000

58.4

67.1

58

844.32

9744.5

GVHP

117.7

70.1

204

283.15

2781.9

HVHP

115.8

71.3

105

463.86

2904.7

Note:

Thickness of the PP support of Gelman membranes (TF200, TF450, TF1000):


110.4 m

The mean pore size was determined from the gas permeation test, wet/dry flow method
and AFM technique. The results are shown in Tables 24. It can be noted that for
Gelman membrane the mean pore size determined from the gas permeation test (Table 3)
is similar to the one given by the manufacturer for TF200 membrane (Table 1) and is

442

M. Khayet, J.I. Mengual and G. Zakrzewska-Trznadel

6.9% and 15.6% smaller for TF450 and TF1000 membranes, respectively. Nevertheless,
for Millipore membranes such as GVHP and HVHP, the mean pore size obtained from
the gas permeation test is larger than that given by the manufacturer (28.7% for GVHP
and 3% for HVHP). From the wet/dry flow method, the mean pore size is up to 17.4%
larger than that obtained from the gas permeation test (that is for TF450 membrane) and
12.7% smaller for TF1000 membrane.
Three-dimensional AFM pictures of GVHP, HVHP and TF1000 membranes are
shown in Figure 1 as example. More AFM images may be found in Khayet, Khulbe and
Matsuura (2004). As expected, a difference in the morphology of the membrane surfaces
can be observed and the pores are seen as dark depressions. The pore sizes were
measured on various portions of the membrane surface and image analysis was done by
the built-in computer program. The mean pore size and the geometric standard deviation
were calculated for each membrane. The results are presented in Table 4. For all
membranes, the mean pore size evaluated from the AFM technique is greater than
that given by the manufacturers. For Gelman membranes the deviation is up to 25.4%
(that is for TF200 membrane) and for Millipore membranes the deviation is up to 125.3%
(that is for GVHP membrane). These results may be attributed to the fact that the pores
have maximum openings at the surface entrance and an amalgamate of a few small pores
could easily be misinterpreted as one large pore, resulting in an overestimation of the
pore size. Similar results were found for flat-sheet and hollow-fibre ultra-filtration
membranes (Khayet and Matsuura, 2003). It was observed that the pore sizes at the
membrane surface are larger than the bulk membrane pore sizes and the mean pore sizes
determined by AFM analysis were 2.1 and 1.7 times larger than those determined from
the gas permeation test for flat-sheet and hollow-fibre membranes, respectively.
Figure 1

Three-dimensional AFM images of the membranes: (a) GVHP, (b) HVHP and (c)
TF1000

Direct contact membrane distillation for nuclear desalination. Part I


Table 4

443

Membrane characteristics determined from wet/dry flow method and AFM technique:
mean pore size, dp; geometric standard deviation, p; surface porosity, s, pore
tortuosity,
Wet/Dry Flow Method

AFM Technique

dp
(nm)

(%)

dp
(nm)

(%)

TF200

233.38

1.07

43.18

1.59

250.72

1.23

33.97

2.02

TF450

491.67

1.10

44.65

1.44

538.67

1.29

30.86

2.08

TF1000

736.84

1.09

56.90

1.18

1185.34

1.27

34.24

1.96

GVHP

265.53

1.12

32.74

2.14

495.70

1.69

25.28

2.77

HVHP

451.23

1.19

33.64

2.12

644.67

1.65

24.63

2.90

Membrane Trade Name

As stated earlier, pore size distribution can also be obtained from wet/dry flow method
and AFM technique. Figure 2 shows both the cumulative pore size and probability
density curves of Gelman and Millipore membranes determined from wet/dry flow
method. When compared to TF200 membrane, it can be observed that the pore size
distribution curves of the other membranes shift to the right and are lower and broader
around their corresponding mean pore size. It can be seen that the pore size given by the
manufacturer is included in the pore size distribution of each membrane. Moreover, in
Figure 3, a comparison between the pore size distribution obtained from wet/dry flow
method and AFM technique is presented for the membranes GVHP and TF450 as an
example. Similar figures were plotted for the other membranes (HVHP, TF200 and
TF450). For all these membranes the pore sizes determined from AFM technique were
larger than those obtained from wet/dry flow method.
Figure 2(a)

Cumulative pore size (a) and probability density curves

444

M. Khayet, J.I. Mengual and G. Zakrzewska-Trznadel

Figure 2(b)

Figure 3

Cumilative pore size of Gelman and Millipore membranes determined from


wet/dry flow method

Pore size distribution of GVHP and TF-450 obtained from wet/dry flow method and
AFM technique

From the field emission scanning electron microscopy (FESEM), Phattaranawik,


Jiraratananon and Fane (2003) obtained mean pore sizes of 251 nm for GVHP
membrane, 414 nm for HVHP membrane and 253 nm for TF200 membrane. Compared
to the values obtained from AFM technique, the low values determined from FESEM
may due to pore contraction during metal coating of the membrane sample, which is

Direct contact membrane distillation for nuclear desalination. Part I

445

required to take the FESEM images. For the same membranes, by using the Coulter
Porometer system, Martnez et al. (2002, 2003) found 310 nm for TF200, 470 nm for
TF450, 650 nm for TF1000, 320 nm for GVHP and 660 nm for HVHP. The deviation
between these values and the values reported in Table 4 for wet/dry flow method are
between 4.4% (that is for TF450 membrane) and 46.2% (that is for HVHP membrane).
The examples of micrographs of Tarflen, Estrofol and FGLP membranes are shown
in Figure 4. The Tarflen membranes, which were made by the stretching method, have
different appearance depending on selected sample. Some of them have a sponge-like
structure while others have more fibrous structure. The differences depend not only on
manufacturing conditions, but also on the initial material, which is an industrial product
(PTFE tape). The surface of the membrane is not homogenous. It must be stated that
during microscopic observation some regions free of pores are noticed, which is
disadvantageous and needs further improvement of manufacturing technique. The pores
are irregular in shape and the pore size distribution is variable depending on the
membrane. In DCMD experiments, the membranes of different mean pore size and
different pore size distribution were used.
The commercial FGLP membrane has more homogenous, sponge-like structure and
uniform surface, without void regions. The pores of the membrane are irregular, but as
similar to Tarflen membranes, the pore size distribution of FGLP is rather small.
The images of cross sections also show sponge-like symmetrical structure of the
membranes (Figure 4).
The images of Estrofol membranes, which were prepared by another technique, i.e.
by heavy ion irradiation, show uniform arrangement of the pores, which are regular and
circular in shape. The pore size distribution assessed from the micrographs is expected to
be small. An example of the pore size distribution curves obtained with Coulter
Porometer II is given in Figure 5 for two membrane samples (FGLP and Tarflen1) and
the statistical parameters such as maximal (dmax) minimal (dmin) and mean (dp) pore
diameters are presented in Table 2. Four Tarflen membranes characterised in Table 2
have different dmax and, as reported in the second part of this paper, their permeabilities
measured in DCMD experiments were different. As was expected, Estrofol track
membranes exhibit the most narrow pore size distribution curves. The pore diameters are
very uniform; some differences are caused by variation of glancing angle of bombarding
heavy ions (that also changes the tortuosity), or possibilities of joining the neighbouring
pores together.
In addition, the obtained values of the geometrical standard deviation, p, determined
from the wet/dry flow method (Table 4) were close to each other (near unity) for all
membranes. p is higher for PVDF membranes (i.e. GVHP, HVHP membranes) than for
PTFE membranes (TF200, TF450 and TF1000 membranes). It was found that p
determined by AFM is higher than that obtained by wet/dry flow method. This is the
reason why the pore size distribution from AFM was broader around the mean pore size
than that determined from wet/dry flow method. p obtained by AFM for PVDF
membranes was also greater than that of PTFE membranes. Based on FESEM study,
lower p values, 1.04 and 0.94, were reported for GVHP and TF200 membranes,
respectively (Phattaranawik, Jiraratananon and Fane, 2003).
The measured porosity of the membranes (TF200, TF450, TF1000, GVHP and
HVHP) is given in Table 3. For all these membranes, the measured values are lower than
those given by the manufacturers. This may be attributed to the different measurement

446

M. Khayet, J.I. Mengual and G. Zakrzewska-Trznadel

techniques used. Moreover, the effective porosity of these membranes was evaluated
from the gas permeation test and the results are also shown in Table 3. The effective
porosity is higher for PTFE membranes having lower thickness than for PVDF
membranes (GVHP, HVHP) although the measured porosities of both membrane types
are not so different. It must be pointed out that the same observations can be made for the
effective porosity obtained by other authors using the Coulter Porometer (7900 m1 for
TF200, 8900 m1 for TF450, 11100 m1 for TF1000, 3350 m1 for GVHP and 3790 m1
for GVHP) (Martnez et al., 2002, 2003).
Figure 4

SEM micrographs of flat-sheet membranes: (a) FGLP, (b) FGLP cross section, (c)
Tarflen2, (d) Tarflen1, (e) Estrofol and (f) Estrofol cross section

Direct contact membrane distillation for nuclear desalination. Part I

447

The pore density was obtained by counting the number of pores on the AFM images of
the membranes and the surface porosity was calculated using the pore size distribution
obtained from AFM (Khayet, Khulbe and Matsuura, 2004). The surface porosity can also
be determined from the pore size distribution obtained from wet/dry flow method as
stated in (Khayet, Velzquez and Mengual, 2004). The results are given in Table 4.
The surface porosities of PVDF membranes are lower than that of PTFE membranes,
whereas the void volume of the PVDF membranes is higher (Table 3). This may be
attributed to the larger thickness of the PVDF membranes used. When comparing surface
porosity from wet/dry flow method and AFM technique, the one obtained from AFM was
lower for all membranes. Moreover, it is worth noting that the surface porosity (s) values
obtained from the AFM analysis is lower than the void volume () given in Table 3 even
though the membranes are the same.
Figure 5

Pore size distribution curves determined by Coulter Porometer II: (a) FGLP,
(b) Tarflen1

The tortuosity factor was calculated from the void volume, , and the effective porosity,
/Lp, determined by the gas permeation test (i.e. Lp = ). The obtained values are given
in Table 4. It can be observed that values calculated based on AFM results are higher
than those calculated from wet/dry flow method and the values of PVDF membranes
are higher than those of PTFE membranes. This may be attributed partly to the higher
membrane thickness of the PVDF membranes. In MD studies, a pore tortuosity factor of
2 is frequently assumed (Phattaranawik, Jiraratananon and Fane, 2003). Furthermore, it
was found that for the same membrane, the tortuosity factor calculated from AFM is
higher than that calculated from wet/dry flow method. The tortuosities determined from
gas permeation tests for the laboratory made Tarflen membranes are shown in Table 2.
For those membranes for which dp was not determined the tortuosities were calculated
with dmax values, and the reported values are certainly a little overestimated.
The FGLP membrane was also characterised by Coulter Porometer II, the obtained
pore sizes are 0.29, 0.35 and 0.39 m for the minimum, mean and maximum pore sizes.
The mean pore size specified by manufacturer 0.2 m is lower than dmin measured by
Coulter porometer. This discrepancy may be due to the different measurement technique
used by the manufacturer. The thickness of this membrane without the PP supporting

448

M. Khayet, J.I. Mengual and G. Zakrzewska-Trznadel

grid is 65 m, its porosity is 70% and the calculated pore tortuosity from gas permeation
measurements is 1.44.
For the PP-Accurel membrane, the minimum, mean and maximum obtained values
of pore sizes are 0.22, 0.27 and 0.45 m, respectively. The thickness of this membrane is
160 m and its porosity is 80%.

Concluding remarks

Although membrane distillation (MD) is a relatively novel membrane technology,


number of applications of the process is expected, among them seawater desalination or
wastewater treatment. The possibility of various applications warrants further study of
the process and intensive development of industrial membrane units. Detailed studies
concerning the design of membranes for MD are still lacking and different values were
associated to the commercial membrane parameters depending on the characterisation
technique used.
At present one of the limitations of further process implementation is lack of big
industrial modules dedicated especially for MD applications, however the membranes,
which proved their usability, are available on the market. The second barrier for
implementation is low, comparing with pressure-driven membrane processes,
productivity of the MD units. It can be improved by proper process parameters
adjustment and modules operation, but also by appropriate membranes engineering,
based on extensive study of their materials and morphology.

Acknowledgement
The authors of this paper acknowledge the bilateral SpanishPolish Project, No 004/R98
002/R97.

References
Bodell, B.R. (1963) Silicone rubber vapour diffusion in saline water distillation, United States
Patent Serial No. 285,032.
Buczkowski, M., Sartowska, B., Wawszczak, D. and Starosta, W. (2001) Radiation resistance of
track etched membranes, Radiation Measurements, Vol. 34, pp.597599.
Buczkowski, M., Starosta, W., Wawszczak, D. and towski, T. (1998) Application of particle
track membranes Polish experiences, Oganesjan, Y.T. and Kalpakcheva, R. (Eds.): Proc.
VI Int. School-Seminar on Heavy Ion Physics, 2227 Sept., 1997, World Scientific, Dubna,
Russia, pp.761769.
Chlubek, N. and Tomaszewska, M. (1989) Some properties of hydrophobic membranes for
membrane distillation, Environ. Prot. Eng., Vol. 15, pp.95103.
Chmielewski, A.G., Zakrzewska-Trznadel, G., Miljevi, N. and Van Hook, W.A. (1995)
Membrane distillation employed for separation of water isotopic compounds, Sep. Sci.
Technol., Vol. 30, pp.16531667.
Findely, M.E. (1967) Vaporization through porous membranes, Ind. & Eng. Chem. Process Des.
Dev., Vol. 6, pp.226230.

Direct contact membrane distillation for nuclear desalination. Part I

449

Khayet, M. and Matsuura, T. (2001) Preparation and characterization of polyvinylidene fluoride


membranes for membrane distillation, Ind. Eng. Chem. Res., Vol. 40, pp.57105718.
Khayet, M. and Matsuura, T. (2003) Determination of surface and bulk pore sizes of flat-sheet and
hollow-fiber membranes by atomic force microscopy, gas permeation and solute transport
methods, Desalination, Vol. 158, pp.5764.
Khayet, M., Feng, C.Y., Khulbe, K.C. and Matsuura, T. (2002) Preparation and characterization of
polyvinylidene fluoride hollow fiber membranes for ultrafiltration, Polymer, Vol. 43,
pp.38793890.
Khayet, M., Godino, M.P. and Mengual, J.I. (2003) Possibility of nuclear desalination through
various membrane distillation configurations: a comparative study, Int. J. of Nuclear
Desalination, Vol. 1, pp.3046.
Khayet, M., Khulbe, K.C. and Matsuura, T. (2004) Characterization of membranes for membrane
distillation by atomic force microscopy and estimation of their water vapour transfer
coefficients in vacuum membrane distillation process, J. Membrane Sci., Vol. 238,
pp.199211.
Khayet, M., Velzquez, A. and Mengual, J.I. (2004) Modelling mass transport through a porous
partitions: effect of pore size distribution, J. Non-Equilibrium Thermodynamics, Vol. 29,
pp.279299.
Lawson, K.W. and Lloyd, D.R. (1997) Membrane distillation: review, J. Membrane Sci.,
Vol. 124, pp.125.
Martnez, L., Florido-Daz, F.J., Hernndez, A. and Prdanos, P. (2002) Characterization of three
hydrophobic porous membranes used in membrane distillation: modelling and evaluation of
their water vapour permeabilities, J. Membrane Sci., Vol. 203, pp.1527.
Martnez, L., Florido-Daz, F.J., Hernndez, A. and Prdanos, P. (2003) Estimation of vapour
transfer coefficient of hydrophobic porous membranes for applications in membrane
distillation, Sep. Purif. Technol., Vol. 33, pp.4555.
Mengual, J.I. and Pea, L. (1997) Membrane distillation, Colloid & Interface Sci., Vol. 1,
pp.1729.
Phattaranawik, J., Jiraratananon, R. and Fane, A.G. (2003) Effect of pore size distribution and air
flux on mass transport in direct contact membrane distillation, J. Membrane Sci., Vol. 215,
pp.7585.

You might also like