Professional Documents
Culture Documents
Abstract
In this paper, flat membrane distillation membranes have been successfully manufactured from PVDF/DMAc
and PVDF/DMF blends by using phase inversion induced by an immersion precipitation technique. The structure
of the membranes is asymmetric with a porous top layer and macrovoids, as assessed by SEM. The existence of
MD fluxes in these membranes is established by performing various pure water flux experiments. A maximum
in the MD fluxes for a particular value of the polymer content in the casting solution from which the membrane
is manufactured has been observed. The dependence of the magnitude of the fluxes on the membrane thickness
is also discussed and the influence of temperature polarizatton evaluated.
recently made in order to manufacture a mem- possible mechanisms (Knudsen, Poiseuille and
brane specifica,'ly for MD [8]. In that paper, diffusion) which may govern the transport pro-
tubular membranes were prepared by the dry-jet cess. Their arguments could be summarized in
wet spinning technique. Also recently [9], a new the following equations:
method to develop tailored membranes for MD
has been reported using surface modification of
~ydrophilic porous partitions. It is clear that the j cA~,v c [dev] aTM (1)
prospects of improvement are very great in this
field.
where J is the molar flux per unit area, ~ is the
It is known that the polymeric material used
membrane thickness, C is a phenomenological
to produce MD membranes must be hydrophobic,
constant called the true membrane mass transfer
and it is generally thought that the magnitude of
coefficient, AP r is the vapor pressure difference
the MD fluxes obtained through such membranes
between the two sides of the membrane, which is
depends mainly on the porosity [10] and, of
expressed in terms of the interfacial temperature
course, on the thickness. Therefore, phase inver-
difference A T u in the second part of Eq. (1). Our
sion with hydrophobic polymers seems to be a
present definition of C is slightly different from
reasonable method to obtain MD membranes. In
that of Schofield because we have extracted from
this work we have used different flat PVDF
C the thickness dependence in order to be able to
porous membranes manufactured by phase inver-
discuss separately this dependence later. So, in a
sion from casting of concentrated solutions of
first approximation, coefficient C should only
that polymer. The solvents used were dimethyl-
depend on the mean temperature at which the
acetamide (DMAc) and dimethylformamide
MD is carried out and on some geometric param-
(DMF). In order to characterize the ability of
eters of the membrane, mainly the porosity (be-
these membranes to give MD fluxes, we have
ing independent of membrane thickness and
measured the water flux for different temperature
temperature difference).
gradients at the same mean temperature. We have
Eq. (1) requires an adequate knowledge of the
also taken various SEM microphotographs to
transmembrane temperature difference (ATM), but
complete the characterization of our membranes.
due to the well-known phenomenon o~"tempera-
The dependence of MD fluxes on two geometric
ture polarization, this may vary markedly from
parameters of the membranes, the thickness and
the measurable bulk temperature (ATB). The
the porosity (related with the polymer content in
analysis of temperature polarization was also
the casting solution), is studied. The effect of
performed in [11]. By studying the heat transfer
temperature polarization is discussed and eval-
problem in steady state it is possible to arrive at:
uated.
2 J M AHv
2. Theory
am h
,,r,,, o (2)
Porous hydrophobic membranes separating
two aqueous phases can give significant fluxes
when a temperature gradient is imposed through
l -rrJ
them. The driving force for this transport process where M is the molar mass of water, AH r the
is the water vapor pressure difference between heat of vaporization of water, LM the effective
the edges of each one of the pores (APt). The thermal conductivity of the membrane and h the
theory of MD has been extensively analyzed by heat transfer coefficient of the unstirred layers
Schofieid et al. [11]; they discussed the three adjacent to the membrane.
J.M. Ortiz de Z~rate et al. / Desalination I00 (1995) 139-148 141
By combining Eqs. (1) and (2), it is possible pared, in some cases from the same solution but
to get the following expression: with different casting thicknesses. After a short
residence time at ambient conditions (295 K,
20% rh) the solfilm was immersed in a water
h c t_~_ j T bath at 295 K. The phase inversion took place
j .... aT B (3) and a polymeric film was obtained, completely
imbibed in water. After v~ashing and prior to use,
h a+2 XM+2 C M ~ z [-'~-i r the membranes were dried by putting them in an
oven at 318 K overnight. The resulting partitions
appeared quite homogeneous and hydrophobie
or, in a shorter form: (drops of water over its surface do not wet it) and
had good mechanical properties, tn order to
J--B ATn (4) check reproducibility, at least two specimens of
membrane were manufactured from each solution
where B is a new phenomenologicai coefficient, to make a total of 13 membranes from the
which is called the global non-isothermal mass PVDF/DMAc solutions and nine from the PVDF/
transport coefficient. Coefficient B measures the DMF ones.
global ability of the membrane to give MD flux- The membrane thicknesses were measured
es, and it is that in which we are interested. Coef- with a micrometer. It was observed that, as ex-
ficient B depends on the porosity of the mem- pected, ti'..e membrane thickness was always
brane (through C), on the temperature polariza- smaller than the cast thickness, and for the same
tion (through h), on the thickness of the mem- cast depth it decreased as the polymer content in
brane and, finally, on the mean temperature at the solution became lower. For instance, for the
which the MD is carried out (through both C and DMAc solution: with a casting depth of 400 lain,
the vapor pressure temperature derivative). the membrane thicknesses varied from 85 pro,
for the membrane with 25% PVDF content, to
54 pan, for that of 10%. A geod reproducibility
3. Membrane preparation and characteri- for the membrane thickness prepared from the
zation same solution and casting depth was found. This
can be seen in Tables 1 and 2 where the mea-
Fiat PVDF membranes were prepared as de- sured thicknesses are explicitly displayed for
scribed elsewhere [12,13] where these kind of each membrane. The membranes for which the
membranes were manufactured for using in mi- casting depth was different from 400 lain are
cro- and ultrafiltration or gas separation process- marked.
es. In the present ease, five PVDF/DMAc and Some SEM microphotographs were taken in
three PVDF/DMF solutions were made. The order to visualize the membranes. The first fea-
polymer was purchased from Aldrich and the ture revealed in the pictures is the membrane
solvents from Merck. The concentration of the asymmetry. Thus, in Fig. 1, the two faces of the
solutions was 10%, 13.2%, 15%, 20% and 25% same membrane (15% PVDF content, DMAc as
w/w for PVDF/DMAc and 10%, 15% and 20% solvent) are shown. There a difference in the
w/w for PVDF/DMF blends. It was observed that pore diameters, around 0.1 ~tm in one face and of
the solubility of the polymer is better for the the order of I ~ in the other, is shown. It was
DMAc solvent than for the DMF one. The dope also observed that the pore diameters on both
was cast onto a glass plate using a doctor blade. sides gradually increase as the polymer content in
The depth of the casting was 400 gtm for most of the casting solution decrease. The SEM inspec-
the membranes, although membranes were pre- tion of membrane sections reveals a structure
142 J.M. Ortiz de Z~rate et al. /Desalination 100 (1995) 139-148
Table !
Measured values o f the global non.-isothermal mass transfer coefficient (B) with the corresp.~nding calculated values
o f the true mass transfer coefficient (C) for the different membranes manufactured from PVDF/DMAc blends. In
all cases the casting depth was 4(i0 p.m except for * where the casting depth was 300 p.m and ** where it was
500 ~tm
Membrane PVDF weight Thichness, Bxl04 measured Cxl07 calculated Mean Cxl07 mol
content, % pm mol s - I m -2 K -I mol s -~ m - I kPa - I s - I m -~ kPa - I
aThe C value was calculated by fitting to Eq. (5). In the other cases the correction was done as indicated in the text.
Table 2
Measured values o f the global non-isothermal mass transfer coefficient (B) with the corresponding calculated values
o f the true mass transfer coefficient (C) for the different membranes manufactured from PVDF/DMF blends. In all
cases the casting depth was 400 p.m except for * where the casting depth was 300 p.m and ** where it was 500 pm
Membrane PVDF weight Thickness, Bxl04 measured Cxl07 calculated Mean Cxl07 mol
content, % ~tm mol s - I m -2 K -I mol s - I m - I kPa - I s - ! m - I kPa - I
DMF 1 10 45 15.96 i. 1
DMF2 IC 51 9.570 0.5 0.8 + 0.4
DMF3 15 68 10.40 4.6
DMF4 15 62 13.89 6.7
DMFS** 15 93 7.674 2.5 3.6 - 2.1
DMF6* 20 55 7.584
DMF7 20 75 4.188
DMF8 20 70 5.114
DMF9 20 105 3.439 i.3 a 1.3 4- 0.4
aThe C value was calculated by fitting to Eq. (5). In the other cases tile correction was Oone as indicated in the text.
brane on the right side was manufactured from a DMAc. The two features mentioned above can be
casting solution with 25% PVDF content, the clearly shown. It is important to notice that the
membrane on the lei~ was from a 10% PVDF two pictures were obtained with different magni-
content solution. In both cases the solvent was fications (see wbite bars in the photos). On the
144 J.M. Ortiz de Zdrate et al. / Desalination 100 (1995) 139-148
Fig. 2. Left: Section of a membrane produced from a casting solution with 10% PVDF content in DMAc. Right:
Sect;.,~r: of a membrane produced from a casting solution with 25% PVDF content in DMAc. The length of the
white bar in the photos corresponds to 10 p.m. The structure: asymmetric with porous top layer and macrovoids
appears evident. In both cases the surface in contact with the glass plate during phase inversion is at the right side
of the pictures.
8.0
Eq. (4)) for various membranes, prepared [3] E. Drioli and Y. Wu, Desalination, 53 (1985)
from the same solution. 339.
• The membrane asymmetry is not correlated to [4] A.S. J6.nsson, R. Wimmerstedt and A.C. Harrys-
the fluxes. The dependence o f fluxes on mem- son, Desalination, 56 (1985) _937.
brane thickness may be used to guess the [5] E. Drioli, B.L. Jiao and V. Calabl6, Proe. Int.
Soe. Citriculture, 3 (1992) 1140.
influence of temperature polarization. [6] J.M. Ortiz de Z~irate, A. Velfizquez, L. Pefia and
- A maximum appears in the true mass trans- J.l. Mengual, Sep. Sci. Tech., 28 (1993) 1421.
port coefficient C (as defined in Eq. (1)) for a [7] R.W. Schofield, A.G. Fane, C.J.D. Fell and R.
particular value of the polymer content in the Maeoun, Desalination, 77 (1990) 279.
casting solution from which the membrane is [8] Y. Fujii, S. Kigoshi, H. lwatani and M. Aoyama,
prepared. J. Membr. Sei., 72 (1992) 53.
[9] Y. Wu, Y. Kong, X. Lin, W, Liu and J. Xu, J.
Membr. Sei., 72 (1992) 189.
Acknowledgments [10] M. Mulder, Basic Principles of Membrane Tech-
nology. Kluwer Academic Press, 1992.
We wish to sincerely acknowledge Alfonso
[! !] R.W. Sehofield A.G. Fane and C,J.D. Fell, J,
Rodrfguez, from the Servicio de Microscopia Membr. Sei., 33 (1987) 299.
Electr6nica - UCM, who took the SEM micro- [12] D. Diiputel and E. Staude, J. Membr. Sci., 78
photographs. We also wish to acknowledge fi- (1993) 45.
nancial support from the Comunidad Aut6noma [13] N. Sehamagl and K.V. Peinemann, Proc. of the
de Madrid. XI ICOM Conference, Heidelberg 1993, p. 3.32,
[14] J.1. Mengual, J. Aguilar and C. Fem~indez
Pineda, J. Membr. Sei., 4 (1978) 209.
References [15] L. Pefia, J.M. Ortiz de Zfirate and J.i. Mengual,
[1] N. Pagliuca, D.G. Mita and F.S. Gaeta, J. J. Chem. Soe. Faraday Trans., 89(24) (1993)
Membr. ScL, 14 (1983) 31. 4333.
[2] K. Schneider and T.J. Van Gassel, Chem. [16] J.M. Ortiz de Zfirate, F. Garela L6pez and J.l.
Ing.Teeh., 56 (1984) 514 (in German). Mengual, J. Membr. Sei., 56 (1991) 181.