You are on page 1of 13

This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy

Journal of Membrane Science 354 (2010) 150–161

Contents lists available at ScienceDirect

Journal of Membrane Science


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

Mixed matrix blend membranes of poly(vinyl alcohol)–poly(vinyl pyrrolidone)


loaded with phosphomolybdic acid used in pervaporation dehydration of ethanol
Veeresh T. Magalad a , Gavisiddappa S. Gokavi a,∗ , K.V.S.N. Raju b , Tejraj M. Aminabhavi a,1
a
Department of Chemistry, Shivaji University, Kolhapur 416004, India
b
Organic Coatings and Polymers Division, Indian Institute of Chemical Technology, Hyderabad 500607, India

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

Article history: Mixed matrix blend membranes of poly(vinyl alcohol)–poly(vinyl pyrrolidone) (PVA–PVP) loaded with
Received 19 December 2009 4, 8 and 12 wt.% of phosphomolybdic acid (PMA) were prepared and characterized by FTIR, XRD, SEM,
Received in revised form 15 February 2010 DSC, dynamic mechanical testing analysis (DMTA) and contact angle measurements. The extent of PMA
Accepted 19 February 2010
loaded in the membrane affected the PV performance during ethanol dehydration. Selectivity calculated
Available online 25 February 2010
from permeability ratio of two components of a membrane was higher for 4 wt.% PMA-loaded mem-
brane than those of 8 and 12 wt.% loaded membranes. At the azeotropic mixture composition of 4% of
Keywords:
water and remaining ethanol, the 4 wt.% PMA-loaded membrane gave a selectivity of 8206 and a flux of
Pervaporation
Phosphomolybdic acid
3.75 g/m2 h kPa. Accordingly, this membrane was further investigated in detail to understand the effect
PVA–PVP blend of variations in feed water composition and temperature on its PV performance. Sorption and diffusion
Ethanol–water mixture results in conjunction with PV data were analyzed in terms of modified Flory–Huggins theory for three-
component system. Heat of sorption, calculated from Arrhenius relationship was negative, suggesting
Langmuir mode of sorption.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction membranes of PVA with other polymers [11–17]. PVP is another


important biopolymer used in PV dehydration studies [18]. In order
Aqueous ethanol solutions are formed during the fermentation to improve PV separation characteristics of water–ethanol mix-
of disposed biomass, wherein ethanol is used as a renewable source tures, in this work, blending of PVA with PVP is considered to limit
of energy to replace conventional petroleum products. Purifica- the excessive swelling of PVA. Since PVA and PVP polymers are
tion and concentration of aqueous ethanol in these processes are completely miscible in all proportions, due to hydrogen bond for-
generally carried out by distillation, but at azeotropic composi- mation between donor groups of PVA and acceptor groups of PVP,
tion, the method is not efficient without recourse to benzene as an in the blend system, it is thought that selectivity of PVA to water
entrainer, a deadly carcinogen. Alternatively, the membrane-based might be enhanced thus, favoring dehydration of ethanol [19].
pervaporation (PV) separation can be used to separate azeotropic Apart from the above considerations and to attain better stabil-
mixtures due to its intrinsic savings in capital and energy costs [1,2]. ity of the membranes, efforts have been made to develop MMMs
PV is thus an attractive method of separating azeotropic mixtures by physical mixing of different types of inorganic fillers [20–23]. In
[3–10]. In this study, mixed matrix membranes (MMMs) prepared the past, phosphomolybdic acid (PMA) has been rarely chosen as a
by loading different quantities of phosphomolybdic acid (PMA) filler [24,25]. PMA is a heteropolyacid, having one heteroatom sur-
into the blend systems of poly(vinyl alcohol) (PVA) and poly(vinyl rounded by four oxygens to form a tetrahedron. The heteroatom is
pyrrolidone) (PVP) were used in the PV separation of water–ethanol located centrally and caged by 12 octahedral MO6 -units linked to
mixtures, and compared with the unfilled blend membrane. one another by the neighboring oxygen atoms, thus with a total of
In the literature, PVA has been widely used in PV dehydration of 24 bridging oxygen atoms that link 12 addenda atoms. The metal
organics [5] because of its good membrane forming ability, but its centers in the 12 octahedra are arranged in a spherical geometry,
excessive swelling in aqueous media tend to lower the selectivity almost equi-distant from each other, in four M3 O13 units, giving
to water. In order to achieve its membrane stability and improve an overall tetrahedral symmetry. The bridging and terminal oxy-
selective characteristics, efforts have been made to develop blend gen atoms on the periphery of the structure are thus available to
associate with water molecules to form hydrates that are thought
to enhance selectivity to water. Also, hydrogen-bonding is likely
∗ Corresponding author. Tel.: +91 231 2609167; fax: +91 231 2692333. to be established between PMA and PVA/PVP blend system. Hence,
E-mail address: gsgokavi@hotmail.com (G.S. Gokavi). in the present study, PMA-loaded PVA/PVP blend membranes are
1
CSIR Emeritus Scientist, New Delhi, India. fabricated and used in PV dehydration of ethanol.

0376-7388/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.memsci.2010.02.055
Author's personal copy

V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161 151

In all, four membranes were fabricated; of these, one was mechanical testing analysis (DMTA) was done (Indian Institute
unfilled nascent blend membrane of PVA/PVP and three were of Chemical Technology, Hyderabad) using Rheometric Scientific,
MMMs prepared by incorporating 4, 8 and 12 wt.% of PMA into USA, DMTA IV instrument operated in a tensile mode at the fre-
the blend membranes of PVA/PVP, all prepared by solution cast- quency of 1 Hz and at the heating rate of 6 ◦ C/min. Static contact
ing method. Thus formed membranes were characterized by FTIR, angles between water and membrane surfaces were measured
XRD, SEM, DSC, DMTA and contact angle measurements. The effect using contact anglemeter (Rame-hart, Model 500-F1, USA) at 25 ◦ C.
of PMA content on swelling and PV performance was investigated
with reference to PV dehydration of ethanol. The results of this 2.4. Swelling and sorption studies
study are analyzed by Flory–Huggins theory [30–32] extended
for three-component system. Sorption, diffusion and permeation Swelling experiments on the membranes were performed gravi-
properties of the permeating liquids have been assessed in the metrically at 27 ◦ C in 10, 20, 30 and 40 wt.% water containing
temperature interval from ambient (27 ◦ C) to 70 ◦ C. Arrhenius feed mixtures. Initial masses of the circularly cut (dia = 3 cm)
parameters have been evaluated to discuss the data in terms of membranes were taken by placing them on a single-pan digi-
temperature effects and also to understand the separation abilities tal microbalance (model AE 240, Mettler, Switzerland) sensitive
of the membranes over the range of temperature interval investi- to ± 0.01 mg. Samples were placed inside the specially designed
gated. airtight test bottles containing 30 cm3 of the test solvent and these
were kept in a hot-air oven maintained at a constant desired tem-
2. Experimental perature. Dry membranes were equilibrated by soaking in different
compositions of the feed mixture in a sealed vessel at 27 ◦ C for
2.1. Materials 48 h. Swollen membranes were weighed immediately after care-
fully blotting them on a digital single pan microbalance. The sorbed
Poly (vinyl pyrrolidone), (PVP), of weight average molecular liquids were recovered in a liquid nitrogen trap by desorbing the
weight 40,000, was purchased from SRL, Mumbai, India. Poly equilibrated sample in the purge and trap apparatus, and analyz-
(vinyl alcohol), (PVA), of weight average molecular weight 125,000, ing by gas chromatography (Model: Ultima-2100, Netel India Pvt.
phosphomolybdic acid (PMA), glutaraldehyde and ethanol were Ltd., Mumbai, India). The percentage degree of swelling, DS was
all purchased from s.d. Fine Chemicals, Mumbai, India. All the calculated as:
chemicals were of reagent grade samples used without further W − W 
s d
purification. Double-distilled water was used throughout the study. Degree of swelling (%) = × 100 (1)
Wd
where Ws and Wd are the weights of the swollen and dry mem-
2.2. Membrane preparation
branes, respectively. Sorption selectivity, ˛s , was calculated as:
M   F 
PMA-filled blend membranes of PVA/PVP were prepared by w e
˛s = × (2)
solution casting method by dissolving 6 g of PVA in 100 mL of dis- Me Fw
tilled water at 90 ◦ C. Similarly, 6 wt.% of PVP solution was prepared where Mw and Me are the mass fraction of water and ethanol in
at ambient temperature (27 ◦ C) by mixing with PVA solution in the the membrane, Fw and Fe are those of water and ethanol in the
ratio of 9:1 (volume basis) and stirring the mixture for 30 min to feed, respectively. Diffusion selectivity, ˛d , was calculated using
form a homogeneous solution. After uniform mixing, the solution solution–diffusion theory [26]:
was filtered to remove any suspended particles. In situ crosslink-
˛ij
ing was done by adding 0.3 mL of glutaraldehyde (GA) to the above ˛d = (3)
solution and respective amounts (4, 8 and 12 wt.% with respect to ˛s
weight of the polymer) of PMA were added as catalyst for crosslink- where ˛ij and ˛S refer to real selectivity and sorption selectivity,
ing. The resulting solution was cast onto a clean glass plate in respectively. Considering the volatile nature of ethanol, all sorption
a dust-free environment, the membranes obtained were dried at experiments were repeated thrice, but experimental errors were
ambient temperature and peeled off from the glass plate. Mem- within ±3%.
brane thickness was measured using a micrometer screw gauge and
was found to be around 50 ± 3 ␮m. Membranes were designated 2.5. Pervaporation experiment
as: M-0 (0.2 mL of conc. HCl as catalyst for crosslinking with pure
PVA and PVP blend system of 9:1 ratio); PMA-filled membranes PV experiments were performed in an indigenously built appa-
were prepared in the same manner by dispersing 4 wt.%, 8 wt.%, and ratus as described before [13]. Effective area of the membrane in the
12 wt.% of PMA, designated as M-4, M-8 and M-12, respectively. PV cell was 26.43 cm2 with a liquid volume capacity of 200 cm3 . PV
apparatus was made of a stainless steel cell in which the feed stock
2.3. Membrane characterization solution was maintained at the required constant temperature con-
trolled thermostatically with a water jacket. PV cell consists of an
FTIR spectra of PMA, M-0, M-4, M-8 and M-12 were all scanned efficient three-blade stirrer powered by DC motor in the feed com-
in the range 400–4000 cm−1 using Nicolet, Impact-410 (Milwaukee, partment. Feed mixture was stirred at 3.334 Hz (200 rpm) speed by
WI), FTIR spectrophotometer by the KBr pellet method. A Siemens maintaining the downstream pressure at 5 mbar using a vacuum
D 5000 powder X-ray diffractometer was used to study the solid pump (Model: ED-21, Hindhivac, Bangalore, India).
state morphology of the membranes in powdered form. The X-rays Before starting the PV experiment, the test membrane was
of 1.5406 Å wavelength were obtained by Cu K␣ radiation source. equilibrated for 3 h with the feed mixture and after establish-
Surface morphology of the membranes was obtained at 10 kV with a ment of steady state equilibrium, liquid permeate was collected
JSM-840A scanning electron microscope (JEOL, Tokyo, Japan). Since and condensed in traps under liquid nitrogen atmosphere. Per-
these films were non-conductive, gold coating of thickness 10 nm meate was collected up to 8–10 h and tests were carried out at
was done on the samples. DSC thermograms were obtained using the required constant temperature. The collected permeate was
SDT 2960 (TA Instruments, USA). Measurements were performed weighed after allowing it to attain the ambient temperature using
at the heating rate of 10 ◦ C/min. The sample pan was conditioned in a digital microbalance sensitive to ±0.01 mg and was then ana-
the instrument before performing the actual experiment. Dynamic lyzed by gas chromatography (Model: Ultima-2100, Netel India,
Author's personal copy

152 V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161

Ltd., Mumbai) to calculate permeation rate, J and selectivity, ˛ij . It calculated as:
may be noted that in all our PV experiments, even after continu- P   F 
w o
ously using the present membranes for up to 8 h, they remained ˇij = × (14)
Po Fw
in-tact without showing any damage and hence, the same mem-
branes were used for repeated PV runs. where Pw and Po are wt.% of water and ethanol in permeate, Fw and
Permeation rate or flux (J) of water (w) and ethanol (e) were Fo are wt.% of water and ethanol in feed, respectively. Separation
calculated as per the theory of Wijmans and Baker [26]: factor does not reflect the real property of a membrane in many
Pi situations. In order to facilitate readers, we have also presented
p
Jw = (pfi − pi ) (4) both the sets of data, but discuss our results mainly based on real
l
selectivity, since it reveals the real effect of the amount of PMA
p Pj loading on membrane property.
Je = (pfj − pj ) (5)
l
Here the superscripts, f and p refer to feed and permeate, respec- 2.6. Thermodynamic analysis
tively; pi and pj are partial vapor pressures of water and ethanol,
respectively; Pi and Pj are membrane permeability coefficients of Thermodynamic analysis of sorption in terms of PV performance
water and ethanol, respectively, which are the product of diffusion is attempted based on the Flory–Huggins theory [30] as suggested
(Dij ) and solubility coefficients (Sij ). Diffusion and solubility coef- before [33] for a three-component system (polymer in mixed liq-
ficients are calculated using the procedure described elsewhere uid media) using Gibbs free energy of mixing (Gmix ); a similar
[17]; l is thickness of the membrane, while Pi /l and Pj /l are the approach was also attempted earlier for PV analysis [32]. The equa-
permeances of water and ethanol, respectively. The relationship tion for sorption selectivity, ˛s of a three-component system is
between partial vapor pressure (pfij ), molar concentration of water given as:
(xi ), ethanol (xj ) and activity coefficients of the individual compo- ˚  v  V  ˚ 
1 1 1 2
nents of the feed mixture obtained from the van Laar Eq. (20) of ln˛S = ln − ln = −1 ln
this paper are given as ˚2 v2 V2 v2
 V1

pfi = xi i psi (6) − 12 (˚2 − ˚1 ) − 12 (v1 − v2 ) − ˚P 1P − 2P (15)
V2
pfj = xj j psj (7) Here ˚i is volume fraction of the ith-component in the swollen
polymer membrane, vi is volume fraction of the ith component in
Saturated vapor pressures of water (psi ),
ethanol (psj )
were cal-
the external liquid phase and Vi is the respective molar volume(s).
culated using the Antoine equation [27] given in the form:
Subscripts 1, 2 and P refer, respectively to water, ethanol and poly-
A−B mer. The values of Vi at 27 ◦ C for water and ethanol were taken from
logpoij = (8)
T +C the literature [34], while volume fraction, P of the polymer in the
swollen state was calculated as [35]:
where poij are the vapor pressures of water and ethanol, respec-
tively; A, B, and C are Antoine constants were taken from the
 P
 M    −1
a P
literature [28], T is the temperature in Kelvin. By considering the ˚P = 1 + − (16)
S Mb S
activity coefficient and molar concentrations of individual com-
ponent, the saturated vapor pressure of water and ethanol were where P and S are the densities of polymer and solvent,
calculated: respectively; Mb and Ma are the masses of polymer before and
after swelling. Density of the polymer was measured by ben-
psi = xi i poi (9) zene displacement method using specific gravity bottle. Initially,
benzene-filled bottle and empty bottle weights were taken and a
psj = xj j poj (10)
weighed quantity of polymer was introduced into the bottle. Excess
Combining Eqs. (5)–(8), we obtain: benzene was wiped off with soft filter papers and mass of the bottle
P  along with benzene and polymer was taken. Volume of the polymer
i p was used to calculate density of the polymer; the molar volume of
Jw = (xi i psi − pi ) (11)
l water + ethanol mixture was calculated using [36]:
 
Pj p x1 M1 + x2 M2
Je = (xj j psj − pj ) (12) V= (17)
l m

In order to compute membrane selectivity, ˛ij , we prefer to use where x1 , x2 are respectively, mole fractions of components 1 and 2
the most recent paper of Baker et al. [29], according to which for of the mixture, M1 , M2 are the corresponding molecular weights, m
a binary mixture, selectivity is the permeability ratio of two com- is density of the water + ethanol mixture. The interaction parame-
ponents of a membrane, an intrinsic property of the membrane, ter, 12 between water and ethanol was calculated using [37]:
under certain conditions or simply, it is the ratio of permeabilites
or permeances of components, i and j through the membrane and x1 ln(x1 /v1 ) + x2 ln(x2 /v2 ) + (GE /RT )
12 = (18)
is calculated as: x1 v2
Pi /l where GE (J mol−1 ) is excess Gibbs free energy of mixing, R is
˛ij = (13)
Pj /l gas constant (J mol−1 K−1 ) and T is absolute temperature (Kelvin).
Values of GE were calculated using activity coefficients,  as:
On the other hand, some authors use the separation factor
(ˇij ), calculated as the ratio of molar component concentrations GE = RT (x1 ln1 + x2 ln2 ) (19)
in the fluids on either side of the membrane. This quantity not only
depends on the selectivity of a membrane, but also on other param- In the absence of direct experimental data on  1 and  2 , we have
eters like feed composition and volatility of each component and is used van Laar equation at 30 ◦ C to compute activity coefficient,  i
Author's personal copy

V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161 153

edge oxygens, respectively [39]. A strong and broad band appear-


ing around 3400 cm−1 corresponds to O–H stretching vibrations of
hydroxyl groups of PMA.
The spectrum of unfilled PVA–PVP blend membrane (M-0) and
those of the MMMs (i.e., M-4, M-8 and M-12) are displayed in
Fig. 1b. The M-0 membrane has characteristic peaks at 3440,
2900–2800, 1630 and 1020 cm−1 corresponding to peaks of –OH,
C–H, C O and C–O stretching vibrations, respectively. In case of M-
4, M-8 and M-12 membranes, no characteristic bands of Keggin unit
appeared, indicating homogeneous distribution of PMA particles
in PVA–PVP blend membranes [40]. For M-4, the peak inten-
sity corresponding to –OH around 3440 cm−1 shows a decrease
in intensity compared to unfilled M-0 membrane, indicating the
interaction between PMA and blend membrane through (Mo Ot ),
(Mo–Oc –Mo) and (Mo–Oe –Mo) (see Fig. 2). With the progressive
increase in PMA as in the case of M-8 and M-12, the peak intensity
of–OH becomes higher than that of M-4, due to large number of
hydroxyl groups of PMA.

3.1.2. XRD studies


The microstructure of PVA–PVP blend membrane along with
MMMs was studied by XRD. In this study, dried samples of 50 ␮m
thickness were used. Fig. 3 displays the XRD patterns of the dried
samples of M-0, M-4, M-8 and M-12. From the spectra, a moderately
sharp peak can be noticed for M-0 around 14◦ of 2, indicating its
semi-crystalline nature, whereas all the XRDs for MMMs showed
broad peaks with diminishing intensities at increasing loadings of
PMA particles; this also indicates the average intermolecular dis-
tance of the amorphous part of the matrix around 14◦ of 2. On the
other hand, relatively sharp intensity peaks are seen at 2 of 20◦ ,
indicating the existence of higher free volume in the unfilled blend
Fig. 1. FTIR spectra of PMA, M-0, M-4, M-8 and M-12 membranes.
membrane (M-0), which would give a decrease in selectivity with
an increase of flux [41]. Thus, interactions of PMA with the poly-
of component, i in the mixture as: mer blend seem to hinder the ordered packing of polymeric chains,
 2 resulting in a reduced crystallinity of the blend. The loss of crys-
Aji xj tallinity due to the addition of PMA is evident in Fig. 3, wherein a
lni = Aij (20)
Aij xi + Aji xj progressive decrease in peak intensity is observed from 12 wt.% to
4 wt.% PMA loadings; this supports the observed improvement in
van Laar parameters, Aij for water and Aji for ethanol were taken
selectivity as the PMA loadings are lowered. As observed in Fig. 3
from the literature [28,38].
that there are two peaks one at 14◦ of 2 and another at 2 of 20◦ ,
The polymer–solvent interaction parameter, iP was calculated
which are related to two types of crystals: crystal 1 and crystal 2. Of
as [31]:
these, crystal 1, corresponding to peak at 14◦ of 2, is responsible
Vi (ıP − ıi )
2 for the separation, since it comprises of the hydrophilic–OH group.
iP = (21)
RT
3.1.3. Scanning electron microscopy
where ıi is solubility parameter of the ith component. Solubility SEM pictures of PVA–PVP mixed matrix membranes contain-
parameters of polymer, water and ethanol were taken from the ing 4, 8 and 12 wt.% PMA were taken to inspect the distribution
literature [34], but the solubility parameter, ı of the blend was of PMA particles. As can be seen from Fig. 4a–c, specially in the
calculated using the additive relation [30]: case of M-4 membrane, we observe a smooth surface with uniform
ı = w1 ı1 + w2 ı2 (22) distribution of PMA particles. Such a homogenous mixing of PMA
particles in the bulk of the polymer phase would facilitate higher
Here w1 and w2 are weight fractions, ı1 and ı2 are solubility param- water transport through the membrane due to the creation of chan-
eters of PVA and PVP, respectively. These data were fed into Eq. (15) nels that are more favorable to the transport of water molecules
to compute the sorption selectivity, ˛s . than ethanol. With increasing loading of PMA, as in M-8 and M-12
membranes, some surface roughness can be seen, but with not so
3. Results and discussion much of uniform distribution of PMA particles on the surfaces of
the membranes.
3.1. Characterization of membranes
3.1.4. DSC analysis
3.1.1. FTIR studies From the DSC curves of the unfilled (M-0) and filled (M-4, M-
FTIR spectra of PMA and PMA-loaded blend membranes are 8 and M-12) matrix membranes shown in Fig. 5, we find that
illustrated in Fig. 1. The spectrum of PMA shown in Fig. 1a M-0 has a Tg of 82.3 ◦ C, which is shifted to higher temperatures
shows four characteristic bands (Keggin type) around 1064, 965, of 83.9◦ , 98.3◦ and 102.7 ◦ C for M-4, M-8 and M-12, respectively
868 and 785 cm−1 that are, respectively attributed to as (P–O), after the addition of PMA; this can be attributed to intermolecu-
as (Mo Ot ), as (Mo–Oc –Mo) and as (Mo–Oe –Mo) stretching lar hydrogen-bonding interactions between the blend polymer and
vibrations, wherein Ot , Oc and Oe refer to terminal, corner and the PMA particles. This would decrease molecular chain mobility
Author's personal copy

154 V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161

Fig. 2. Schematics of the interaction between PMA and PVA–PVP blend.

of PVA–PVP polymer with increasing PMA loading. Notice that Tg 3.1.5. DMTA analysis
values of M-0 and M-4 membranes are quite identical, indicating DMTA measurements give information about the micro-
the compatibility between organic and inorganic phases, whereas structural as well as thermal and viscoelastic properties of the
Tg of M-8 and M-12 are considerably higher than M-0 due to the polymeric systems over a wide temperature interval. In DMTA,
micro-phase separation during the membrane formation process EI represents the energy stored and recovered per cycle. The loss
at >4 wt.% loading [42]. modulus, EII measures the energy dissipated per cycle during defor-
mation. The ratio EII /EI is loss tangent. The tan ı of DMTA curves
shown in Figs. 6–8 suggest that with increasing PMA content, a
shift of tan ı peak at higher temperature occurs, such that higher
storage modulus is evident in the rubbery region. This behavior is
attributed to an impediment of the polymeric segmental motion
induced due to the presence of PMA particles. The observed EI
for M-0 at 35 ◦ C is 1.6 × 102 N/cm2 (1.6 × 107 dynes/cm2 ), which
got enhanced to 7.9 × 104 N/cm2 (7.9 × 109 dynes/cm2 ) after incor-
porating PMA (4 wt.%) into the blend. However, incorporation of
higher amount of PMA increased EI values, but the observed val-
ues for MMMs are quite higher than the unfilled (M-0) membrane.
It is certain that introduction of PMA particles between polymeric
chains would reduce the polymer chain mobility with a consequent
increase in its modulus. Such an increase in modulus gives reduced
free volume in the matrix.

3.1.6. Contact angle measurement


In PV separation, the surface free energy of the membrane is
Fig. 3. XRD curves of M-0, M-4, M-8 and M-12 membranes. important to understand the transport process, since surface free
Author's personal copy

V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161 155

Fig. 5. DSC thermograms of M-0, M-4, M-8 and M-12, membranes.

Fig. 6. Temperature dependence of loss modulus (EII ) for M-0, M-4, M-8 and M-12
membranes.

3.2. Pervaporation performance

3.2.1. Effect of filler particles


Fig. 10 displays flux of water, ethanol and water selectivity cal-
culated using Eqs. (11)–(13) at 27 ◦ C for all the membranes at the
azeotropic composition i.e., 4 wt.% water plus remaining ethanol.
Initially, selectivity of 4 wt.% PMA-containing membrane (i.e., M-
4) increased considerably compared to nascent (M-0) membrane.
However, at higher concentrations of PMA i.e., M-8 and M-12 mem-

Fig. 4. Surface SEM images of (a) M-4, (b) M-8 and (c) M-12 membranes.

energy determines the surface and interfacial behavior in wetting


and adhesion. Even though surface energy cannot be measured
directly, few indirect methods have been proposed, of which use
of contact angle measurement of liquid and membrane surface has
been done in this study. Contact angle between water and mem-
brane surface gives a measure of membrane hydrophilicity; smaller
the contact angle, greater is its hydrophilicity. As shown in Fig. 9,
lower contact angle observed for M-0 than M-4, M-8 and M-12 indi-
cates that due to the presence of PMA particles in polymer phase,
hydrophilicity of the MMMs has changed. In this study, M-4 has
the highest contact angle at a lower concentration of PMA, sug-
gesting a reduction in hydroxyl groups of PVA of the blend. In the
presence of high PMA content, a large number of hydroxyl groups
tend to enhance the hydrophilicity of membranes, thus increasing
Fig. 7. Temperature dependence of dynamic storage modulus (EI ) for M-0, M-4, M-8
selectivity to water, with a concomitant decrease in water flux. and M-12 membranes.
Author's personal copy

156 V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161

Fig. 8. Tan ı curves of M-0, M-4, M-8 and M-12 membranes. Fig. 10. Effect of PMA content on PV performance at azeotropic feed composition.

in high density of functional groups of PMA as per the interaction


model suggested in Fig. 2. Moreover, the resistance to molecular dif-
fusion and/or tortuosity of the diffusion pathway, resulting from the
lower degree of swelling does not change evidently when higher
amount of PMA particles are present in the membranes. Similar
results were observed by Wang et al., [42], who studied the clay-
modified polyimide/SDS-clay membrane for the PV separation of
water–ethanol mixture.
As per the principles of solution-diffusion [26], permeability
coefficient is the product of solubility and diffusivity. In order to
test this empirical hypothesis that is similar to gas transport pro-
cess [29], we have computed the values of sorption selectivity and
diffusion selectivity using Eqs. (2) and (3), respectively. These data
summarized in Table 1 suggest that sorption selectivity is much
higher than diffusion selectivity supporting the fact that more of
Fig. 9. Effect of PMA content on contact angle between water and membranes. water molecules are transported through the membrane. It may be
noted that the values of ˇij calculated from Eq. (14), also included
branes, the selectivity values decreased considerably, but are still in Table 1, display similar trends as those of ˛ij , but in general,
higher than that observed for M-0. Such a rapid decrease in mem- ˛ij values are lower than those of ˇij . These data suggest that water
brane selectivity due to the addition of higher amounts of PMA molecules are easily dissolved and transported into and through the
may be due to the change in pore texture of the membrane matrix barrier membrane, resulting in an increase of sorption selectivity.
in the presence of PMA. This would also increase the void volume Consequently, sorption effect dominates the PV behavior, because
of the membrane phase, a fact that was supported by water sorp- the real membrane selectivity follows the same trend as that of
tion/swelling studies as well as SEM images (Fig. 4) and contact sorption selectivity, but not diffusion selectivity. Here, of all the
angle measurements (Fig. 9). Initially, the high density of functional membranes studied, M-4 exhibits higher sorption selectivity com-
groups dominated the PV performance, but at certain concentration pared to the remaining membranes and hence, M-4 was considered
of PMA in the membrane phase (viz., at 4 wt.%), the PV performance for detailed studies on investigating the effect of temperature and
becomes optimal. In this work, we have decided to investigate in feed water composition on its PV performance.
more details on M-4 membrane that gave the optimum selectivity
compared to all the membranes. 3.2.2. Effect of feed water composition
As regards flux of water for M-4 membrane, it is lower than those In this study, PVP was considered for blending with PVA because
observed for nascent (M-0) as well as M-8 and M-12 membranes, the rigidity and bulkiness of the derived blend membrane is as
but the flux curve for ethanol is independent of PMA loading and important a factor as its hydrophilicity for an effective dehydra-
the values are much lower than those observed for water as well tion of ethanol. While hydrophilicity is needed for preferential
as follow the linear trend. An achievement of trade-off between attraction of water to achieve high selectivity, the high rigidity
flux and selectivity has been a formidable task even though, it is and bulkiness of the membrane are also equally necessary for an
evident from Fig. 10 that both M-8 and M-12 membranes exhibit easy transport of permeating water molecules. The performance of
lower selectivity and higher fluxes than M-4; this could be due to nascent PVA/PVP blend membrane was investigated in comparison
the fact that linear polymer clusters in M-4 might have resulted to M-4 membrane as a function of feed water composition in order

Table 1
Flux of water and ethanol, total flux, separation factor, selectivity, sorption and diffusion selectivities of different membranes at 27 ◦ C.

Membrane Jw (g/m2 h kPa) Je (g/m2 h kPa) Total flux (g/m2 h kPa) ˇij ˛ij ˛s ˛d

M-0 10.20 2.541 12.74 96 73 80 0.22


M-4 3.74 0.008 3.75 9,976 8,206 529 354
M-8 16.49 0.053 16.55 6,800 5,717 421 212
M-12 18.08 0.162 18.25 2,431 2,051 344 33
Author's personal copy

V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161 157

Table 2
Effect of feed water composition on degree of swelling, flux of water and ethanol, total flux, separation factor and selectivity of M-0 and M-4 membranes at 27 ◦ C and 5 mbar
pressure.

Membrane Feed water composition (wt.%) DS (%) Jw (g/m2 h kPa) Je (g/m2 h kPa) Total flux (g/m2 h kPa) ˇij ˛ij

M-0 10 38 30.3 1.0 31.3 71 109


20 80 35.7 1.8 37.5 45 48
30 139 51.8 5.1 57.0 23 18
40 155 91.7 22.1 113.8 9 6

M-4 10 19 12.8 0.02 12.8 3,324 5,327


20 32 19.1 0.09 19.2 1,246 1,363
30 51 35.1 0.5 35.6 414 351
40 92 71.0 3.6 74.6 104 73

to show improved PV performance over that of the nascent (M-0) the bulk due to H-bond formation between PMA protons and water
membrane. molecules that exist as guarded protons viz., H3 O+ or H5 O2 + [43,44].
Table 2 displays the effect of feed water composition rang- With increasing feed water composition from 10 to 40 wt.%, more
ing from 10 to 40 wt.% in the mixture on PV performance of of water molecules are replaced with lesser number of ethanol
M-0 and M-4 membranes at 27 ◦ C under the constant pressure molecules for protonation to take place. After protonation, ethanol
of 5 mbar. For the unfilled M-0 membrane, percentage degree of in the pseudo-liquid phase of PMA with its reduced size will perme-
swelling is considerably higher than that observed for M-4. With ate through the membrane, thereby giving high flux with a reduced
increasing feed water composition, for both the membranes (M- selectivity. The present results demonstrate the positive role played
0 and M-4), the percentage degree of swelling as well as flux by the presence of PMA particles embedded into PVA–PVP blend
of water and ethanol both increased with increasing feed water to enhance the membrane performance over that of the unfilled
composition from 10 to 40 wt.% due to increased preferential inter- PVA–PVP blend membrane.
action of water molecules with the membrane polymers. However,
extremely smaller values of flux are observed for ethanol compared 3.2.3. Diffusion coefficient
to water. Notice that total flux is almost close to the value observed Transport of binary liquid mixtures through PV membranes
for water flux, indicating the water-selective nature of the mem- can be conventionally explained by the solution-diffusion mech-
brane. Conversely, selectivity decreased considerably for both the anism [26], which occurs in three steps: sorption, diffusion, and
membranes with increasing feed water composition. Particularly, evaporation. Permeation and selectivity are rather governed by
in case of M-0 (no filler), lower selectivity to water than M-4 is due solubility and diffusivity of the components of the feed mix-
to the fact that blend polymer molecular segmental motion might ture. Thus, in order to understand permeation, we have calculated
be retarded by the presence of PMA particles, resulting in a lesser concentration-independent diffusion coefficient, Di of the perme-
degree of swelling in M-4 than M-0 membrane. See for instance, the ating molecules using Fick’s diffusion equation:
highest selectivity of 5327 and separation factor of 3324 observed
 dC 
for M-4 at 10 wt.% water-containing feed mixture, which is a clear i
Ji = −Di (23)
indication of the effect of added PMA particles into PVA/PVP blend dx
matrix. The decreased ˛ij and ˇij values with increasing water com-
positions of the feed mixture is due to the probable leaching of PMA where Ji is permeation flux/unit area (kg/m2 s), Di is diffusion coef-
particles from the matrix polymer. Therefore, the present mem- ficient (m2 /s), Ci is concentration of permeate (kg/m3 ), subscript i
branes may not be extremely useful for separating aqueous ethanol stands for water or ethanol, and x is diffusion path length. For sim-
mixtures at higher amount of water. plicity, we assume that concentration profile along the diffusion
It is likely that physico-chemical nature of the blend polymer pathway is linear and hence, concentration-independent, Di can be
due to the presence of PMA particles might have changed. Het- calculated using:
eropolyacids are like mineral acids having protons in the protected Ji h
environment, allowing them to interact with hydroxyl groups of Di = (24)
Ci
the blend polymer reducing the free volume channels in the bulk
of the polymer matrix, due to the reduction in degree of swelling. where h is membrane thickness (50 ␮m in this study).
There is also a possibility that with increasing feed water composi- Calculated values of diffusion coefficients of water and ethanol
tion, more number of water molecules are sorbed by the filler PMA at temperatures between 40 ◦ C and 70 ◦ C are compiled in Table 3.
particles due to induced plasticization effect of the membrane poly- Notice that with increasing temperature from 40 ◦ C to 70 ◦ C, flux
mer (even in the presence of small amount of ethanol) and also due of water as well as ethanol increased, but selectivity (˛ij ) as well
to increased chain mobility. as separation factor (ˇij ) values decreased. The D values of water
PMA is known to exhibit pseudo-liquid phase behavior, where are higher than those of ethanol by two orders of magnitude, fur-
the polar water molecules will enter into the bulk phase by expand- ther signifying the water-selective nature of the membranes. It
ing or contracting the distance between Keggin type anions in is observed that diffusion coefficients of water as well as ethanol
the crystal lattice; thus, lesser number of ethanol molecules are increased systematically with increasing temperature, a trend that
sorbed onto the membrane surface without actually entering into follows the conventional Arrhenius trend.

Table 3
Effect of temperature on diffusion coefficients of water and ethanol, flux, separation factor and selectivity for M-4 membrane at 10 wt.% water containing mixture.

Temperature (◦ C) Dw × 108 (m2 /s) De × 1010 (m2 /s) Jw (g/m2 h kPa) Je (g/m2 h kPa) Total flux (g/m2 h kPa) ˇij ˛ij

40 4.9 3.4 7.9 0.13 8.04 366 577


50 6.4 6.2 28.7 0.26 28.95 262 412
60 10.9 13.9 53.6 0.66 54.29 197 311
70 14.5 21.6 81.2 1.16 82.38 169 265
Author's personal copy

158 V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161

Table 4
Interaction parameters and sorption selectivity at different feed water compositions for M-4 membrane for water/ethanol mixture at 27 ◦ C.

Feed water composition (wt.%) Interaction parameters ˛s Expt. ˛s Cal.

Water/ethanol 12 Water/polymer 1p Ethanol/polymer 2p

10 1.61 0.020 2.48 110 116


20 1.36 0.018 1.23 34 35
30 1.19 0.017 0.80 12 14
40 1.05 0.015 0.68 6 7

3.2.4. Effect of temperature on pervaporation 2p values at all feed water compositions, due to water-selective
Temperature affects the PV performance. Flux increases with nature of M-4. Majority of water molecules are, therefore, adsorbed
increasing temperature, and hence, the overall temperature effect by the hydrophilic PMA particles, making it more water-selective,
on flux or diffusivity can be described through Arrhenius-type rela- thereby extracting more of water molecules on permeate side, thus
tionship, from which activation parameters can be computed. The enhancing the membrane selectivity to water. One can observe that
effect of temperature on PV performance was studied only in case of experimental and theoretical values of sorption selectivity are quite
M-4 membrane at a fixed feed water composition of 10 wt.% under comparable (see Table 4) and confirming that the proposed thermo-
5 mbar pressure. The temperature dependencies of permeation and dynamic treatment based on Flory–Huggins theory is satisfactory
diffusion have been investigated using the Arrhenius-type relation- to explain the PV results, and also, it gives accurate predictions.
ship:
 −E  3.4. Comparison of PV results with vapor–liquid equilibrium
a
X = Xo exp (25) (VLE) data
RT
where X represents the diffusion coefficient (D) or permeation flux As stated in the beginning of this paper, conventional distilla-
(Jp ), Xo is a constant representing the pre-exponential factors, Do or tion has drawbacks compared to the PV separation, particularly
Jo , and Ea represents the activation energy of permeation or diffu- for the separation of azeotropes and no complete separation can
sion and RT is the energy term. As the feed temperature increases, be achieved without adding an entrainer. To prove this anomaly,
vapor pressure in the feed side compartment increases, but vapor we have compared permeate concentration obtained in PV experi-
pressure on permeate side is not affected, resulting in increased ment for M-4 membrane with vapor–liquid equilibrium (VLE) curve
driving force. obtained from distillation [47] in Fig. 12. PV separation is thus a
Arrhenius plots of ln J vs 1000/T and ln D vs 1000/T shown, better option than distillation to break ethanol–water azeotropic
respectively in Fig. 11a and b for permeation flux and diffusion
exhibit linear trends, suggesting that permeability and diffusiv-
ity follow the Arrhenius rule. From the least squares fit of the
linear plots of Fig. 11a and b, activation energies of permeability
(Ep ) and diffusivity (Ed ) have been estimated. Permeation activa-
tion energy value of water (Epw ) obtained is 45.5 kJ/mol, which is
close to 46.4 kJ/mol found for total activation energy (Ep ), whereas
activation energy of ethanol (Epe ) is 69.15 kJ/mol, which is signifi-
cantly higher than Epw (45.5 kJ/mol). Diffusion activation energy of
water (Edw ) is 48.5 kJ/mol, while that of ethanol (Ede ) is 69.2 kJ/mol,
suggesting that both permeating and diffusing molecules require
higher activation energies for the molecular transport to take place
through the membrane. Notice that Ed values are higher than Ep
values, yet the difference is not significant, indicating that both per-
meation and diffusion would contribute to the overall PV process.
Using the values of Ep and Ed , heat of sorption, Hs was calcu-
lated as:

HS = Ep − Ed (26)

The resulting Hs is negative (−2.4 kJ/mol), which suggests that


the mode of transport involves contributions from Langmuir’s type
of sorption. Langmuir’s sorption requires the pre-existence of a
site in which sorption occurs by hole-filling mechanism giving an
exothermic contribution [45].

3.3. Thermodynamic interpretation

Experimental and theoretical values of sorption selectivity for


M-4 are given in Table 4. From the values of interaction param-
eters of M-4 calculated at different feed water compositions, we
see that the interaction parameter of water-membrane (1p ) and
ethanol-membrane (2p ) decreases with increasing feed water
composition. Usually, stronger interaction between the mixture
components would result in a smaller value of interaction parame-
ter [46]. In the present study, M-4 has lower 1p values and higher Fig. 11. Arrhenius plot (a) ln J vs. 1000/T and (b) ln D vs. 1000/T for M-4 membrane.
Author's personal copy

V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161 159

Table 5
Comparison of flux and separation factor of the present mixed matrix membranes with literature for ethanol dehydration at 27 ◦ C.

Membrane Feed water composition (%) ˇij J (kg/m2 h 10 ␮m) Reference

Two ply composite CS/NaAlg 5.0 1,110 0.07 [48]


CS/NaAlg 13.5 436 0.22 [49]
P-NaAlg 5.2 2,182 0.24 [2]
CS 10.0 8,000 0.26 [50]
NaAlg 10.0 120 0.29 [51]
PVA/PVS 6.2 700 0.50 [52]
M-4 4.0 9,976 0.10 Present work
M-4 10.0 3,324 0.23 -do-

CS: Chitosan; NaAlg: sodium alginate; HEC: hydroxyethyl cellulose; P-NaAlg: phosphorylated sodium alginate; PVA/PVS: poly(vinylalcohol)/poly(styrene sulfuric acid).

composition, since here, the membrane acts as a third phase to of 0.22 kg/m2 h 10 ␮m. The lowest value of separation factor of
break such an azeotrope. Particularly, high selectivity of the mem- 120 is observed for pure NaAlg with a somewhat high flux of
brane would allow more of water molecules to transport across 0.29 kg/m2 h 10 ␮m [51]. Thus, overall, it can be inferred that after
the membrane due to its preferential affinity towards membrane, the addition of PMA particles into the PVA–PVP blend matrix, one
thereby overcoming the azeotropic barrier. can drastically increase membrane separation factor and/or selec-
tivity to water, but not flux.
3.5. Comparison of present data with literature
4. Conclusions
A number of studies have been made in the literature concern-
ing the PV separation of water–ethanol mixture. Therefore, it is It is demonstrated in this study that after incorporating 4 wt.%
necessary to compare our data with those of the published reports. of PMA particles into PVA–PVP blend matrix, one can enhance
Table 5 summarizes such data in comparison to the published the PV performance of the filled matrix membranes over that of
reports. However, a direct comparison of our PV data with litera- the unfilled PVA–PVP blend membrane during ethanol dehydra-
ture findings is somewhat difficult, since such results are obtained tion. However, higher loadings (8 and 12 wt.%) of PMA particles
at varying feed water compositions and temperatures, yet some did not result in any improvement of PV performance. Mem-
data that are obtained under similar set of experimental conditions brane characterization by DSC and DMTA revealed an adequate
for flux and separation factor are compared in Table 5 to assess the thermal and mechanical stability of the membranes, which is essen-
usefulness of the present membranes with those of the published tial for PV dehydration above the ambient temperature. Sorption
reports. selectivity of such membranes being much higher than diffusion
In the case of M-4 membrane, for the separation of azeotropic selectivity, favors the water separation selectivity during ethanol
mixture, the highest separation factor of 9976 for water with a dehydration, since the process is dominated by sorption mecha-
lower flux of 0.10 kg/m2 h 10 ␮m is the result effect of added PMA nism. Sorption selectivity of M-4 is higher than M-8 and M-12,
particles. In the case of novel composite chitosan membrane pre- due to better compatibility of PMA particles into M-4 at 4 wt.%
pared by Wang et al. [50], the separation factor is 8000, while loading than with M-8 and M-12. Selective sorption by the filler
flux is 0.26 kg/m2 h 10 ␮m for 10 wt.% of water containing ethanol particles increased the preferentially permeating water molecules
feed. On the other hand, chitosan and sodium alginate blend mem- from ethanol-containing mixtures. Flory–Huggins theory enabled
brane [49] has a much lower separation factor of 436 and flux the accurate predictions of the parameters and also the theory is
useful in satisfactorily explaining the thermodynamics of disso-
lution and permeation processes. With the increasing feed water
composition, membrane performance was affected markedly in
accordance with the swelling and flux data, which also supports
the fact that increase of selectivity and separation factor with a
reduction in flux. Azeotropic composition of water–ethanol mix-
ture was separated by the highly water-selective M-4 membrane,
which exhibited significantly lower activation energies for water
than ethanol, and with a negative heat of sorption, suggesting its
higher separation ability. Furthermore, sorption is dominated by
Langmuir’s mode of sorption. Finally, as suggested by one of the
referees, it may be useful to the compute real selectivity instead
of separation factor. However, in this work, we have explained our
data in terms of both selectivity and separation factor.

Acknowledgements

G.S.G. and V.T.M. thank Department of Science & Technology,


New Delhi, India (No. SR/S1/PC-31/2006) for a financial support.
T.M.A. thanks the CSIR, New Delhi, India [21/(0760)/09/EMR-II] for
awarding CSIR Emeritus Scientist. We also thank Profs. A.V. Rao and
P.S. Patil, Department of Physics, Shivaji University for providing
contact angle and SEM facilities. Finally, we immensely thank the
suggestion of an unknown reviewer to recast the PV data in terms
Fig. 12. Comparison of vapor liquid equilibrium curve with PV data for
water–ethanol mixtures at 27 ◦ C.
of real selectivity, which led to the novelty of this paper.
Author's personal copy

160 V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161

[18] J. Lu, Q.T. Nguyen, L. Zhou, B. Xu, Z. Ping, Study of the role of water in the
Nomenclature transport of water and THF through hydrophilic membranes by pervaporation,
J. Membr. Sci. 226 (2003) 135.
[19] Z.H. Ping, Q.T. Nguyen, J. Neel, Investigation of poly(vinyl alcohol) poly(n-vinyl-
A effective membrane area (m2 )
2-pyrrolidone) blends. 3. Permeation properties of polymer blend membranes,
J flux (kg/m2 h) Macromol. Chem. Phys. 195 (1994) 2107.
Jo pre-exponential factor for permeation [20] R. Mahajan, R. Burns, M. Schaeffer, W.J. Koros, Challenges in forming successful
Ms and Md weights of the swollen and dry membranes mixed matrix membranes with rigid polymeric materials, J. Appl. Polym. Sci.
86 (2002) 881.
P and F weight% of permeate and feed [21] Z. Huang, Y. Shi, R. Wen, Y. Guo, J. Su, T. Matsuura, Multilayer poly(vinyl
R gas constant alcohol)–zeolite 4A composite membranes for ethanol dehydration by means
T temperature (K) of pervaporation, Sep. Purif. Technol. 51 (2006) 126.
[22] Z. Huang, H. Guan, W. Tan, X. Qiao, S. Kulprathipanja, Pervaporation
Wi weight of the permeate (kg) study of aqueous ethanol solution through zeolite-incorporated multilayer
poly(vinyl alcohol) membranes: effect of zeolites, J. Membr. Sci. 276 (2006)
Greek letters 260.
[23] T.T. Moore, W.J. Koros, Non-ideal effects in organic–inorganic materials for gas
˛d diffusion selectivity separation membranes, J. Mol. Struct. 739 (2005) 87.
˛ij real selectivity [24] I.K. Song, S.K. Shin, W.Y. Lee, Catalytic activity of H3 PMo12 O40 blended poly-
˛s sorption selectivity sulfone film in the oxidation of ethanol to acetaldehyde, J. Catal. 144 (1993)
348.
ˇij separation factor
[25] J.K. Lee, I.K. Song, W.Y.J. Lee, J. Kim, Modification of 12-molybdophosphoric acid
ı solubility parameter catalyst by blending with polysulfone and its catalytic activity for 2-propanol
Hs heat of sorption (kJ/mol) conversion reaction, J. Mol. Catal. A 104 (1996) 311.
iP Flory–Huggins interaction parameter between sol- [26] J.G. Wijmans, R.W. Baker, The solution-diffusion model: a review, J. Membr. Sci.
107 (1995) 1.
vent and polymer [27] W.F. Guo, T.S. Chung, T. Matsuura, R. Wang, Y. Liu, Pervaporation study of water
and tert-butanol mixtures, J. Appl. Polym. Sci. 91 (2004) 4082.
[28] M.J. Holmes, M.V. Winkle, Prediction of ternary vapor–liquid equilibria from
binary data, Ind. Eng. Chem. 62 (1970) 21.
[29] R.W. Baker, J.G. Wijmans, Y. Huang, Permeability, permeance and selectivity: a
References preferred way of reporting pervaporation performance data, J. Membr. Sci. 348
(2010) 346.
[1] S. Amnuaypanich, N. Kongchana, Natural rubber/polyacrylic acid semi- [30] B.V.K. Naidu, T.M. Aminabhavi, Pervaporation separation of water/2-propanol
interpenetrating polymer network membranes for the pervaporation of mixtures by use of the blend membranes of sodium alginate and hydroxyethyl
water–ethanol mixtures, J. Appl. Polym. Sci. 114 (2009) 3509. cellulose: roles of permeate-membrane interactions, zeolite filling, and mem-
[2] S. Kalyani, B. Smitha, S. Sridhar, A. Krishnaiah, Pervaporation separation of brane swelling, Ind. Eng. Chem. Res. 44 (2005) 7481.
ethanol–water mixtures through sodium alginate membranes, Desalination [31] P.J. Flory, Principles of Polymer Chemistry, Cornell University Press, Ithaca, New
229 (2008) 68. York, 1953.
[3] T.M. Aminabhavi, R.S. Khinnavar, S.B. Harogoppad, U.S. Aithal, Q.T. Nguyen, [32] T.M. Aminabhavi, P. Munk, Preferential adsorption onto polystyrene in mixed
K.C. Hansen, Pervaporation separation of organic–aqueous and organic–organic solvent systems, Macromolecules 12 (1979) 607.
binary mixtures, J. Macromol. Sci. Rev. Macromol. Chem. Phys. C 34 (1994) [33] M.H.V. Mulder, C.A. Smolders, On the mechanism of separation of
139. ethanol/water mixtures by pervaporation. Part I. Calculations of concentration
[4] S.D. Bhat, T.M. Aminabhavi, Pervaporation separation using sodium alginate profiles, J. Membr. Sci. 17 (1984) 289.
and its modified membranes—a review, Sep. Purif. Rev. 36 (2007) 203. [34] J. Brandrup, E.H. Immergut, E.A. Grulke, Polymer Hand Book,
[5] R.Y.M. Huang, Pervaporation Membrane Separation Processes, Elsevier Science Wiley/Interscience, New York, 1999.
Publishers, Amsterdam, Netherlands, 1991. [35] U.S. Aithal, T.M. Aminabhavi, P.E. Cassidy, Interactions of organic halides with
[6] S.D. Bhat, N.N. Mallikarjuna, T.M. Aminabhavi, Microporous aluminophosphate a polyurethane elastomer, J. Membr. Sci. 50 (1990) 225.
(AlPO4 -5) molecular sieve-loaded novel sodium alginate composite mem- [36] M.I. Aralaguppi, T.M. Aminabhavi, R.H. Balundgi, S.S. Joshi, Thermodynamic
branes for pervaporation dehydration of aqueous–organic mixtures near their interactions in mixtures of bromoform with hydrocarbons, J. Phys. Chem. 95
azeotropic compositions, J. Membr. Sci. 282 (2006) 473. (1991) 5299.
[7] S.D. Bhat, T.M. Aminabhavi, Zeolite K-LTL-loaded sodium alginate mixed matrix [37] T.M. Aminabhavi, P. Munk, Diffusion coefficients of some nonideal liquid mix-
membranes for pervaporation dehydration of aqueous–organic mixtures, J. tures, J. Phys. Chem. 84 (1980) 442.
Membr. Sci. 306 (2007) 173. [38] DECHEMA Chemistry Data Series by Gmehling, Onken, and Arlt., vol. 1, part 1b,
[8] J.H. Chen, Q.L. Liu, X.H. Zhang, Q.G. Zhang, Pervaporation and characterization of 1977.
chitosan membranes cross-linked by 3-aminopropyltriethoxysilane, J. Membr. [39] G. Lakshminarayana, M. Nogami, Synthesis and characterization of pro-
Sci. 292 (2007) 125. ton conducting inorganic–organic hybrid nanocomposite membranes based
[9] X.H. Zhang, Q.L. Liu, Y. Xiong, A.M. Zhu, Y. Chen, Q.G. Zhang, Pervaporation dehy- on tetraethoxysilane/trimethylphosphate/3 glycidoxypropyl trimethoxysi-
dration of ethyl acetate/ethanol/water azeotrope using chitosan/poly(vinyl lane/heteropoly acids, Electrochim. Acta 54 (2009) 4731.
pyrrolidone) blend membranes, J. Membr. Sci. 327 (2009) 274. [40] J.H. Chen, Q.L. Liu, A.M. Zhu, Q.G. Zhang, J. Fang, Pervaporation separation
[10] Y. Ma, J. Wang, T. Tsuru, Pervaporation of water/ethanol mixtures through of MeOH/DMC mixtures using STA/CS hybrid membranes, J. Membr. Sci. 315
microporous silica membranes, Sep. Purif. Tecnol. 66 (2009) 479. (2008) 74.
[11] H. Zheng, Y. Du, J. Yu, R. Huang, L. Zang, Preparation and characterization [41] K. Sunitha, Y.V.L. Ravi Kumar, S. Sridhar, Effect of PVP loading on pervapora-
of chitosan/poly(vinyl alcohol) blend fibers, J. Appl. Polym. Sci. 80 (2001) tion performance of poly(vinyl alcohol) membranes for THF/water mixtures, J.
2558. Mater. Sci. 44 (2009) 6280.
[12] B.V.K. Naidu, M. Sairam, K.S.V.N. Raju, T.M. Aminabhavi, Pervaporation of [42] Y. Wang, S. Fan, K.R. Lee, C.L. Li, S.H. Huang, H.A. Tsai, J.Y. Lai,
water + IPA mixture using novel nanocomposite membranes of poly(vinyl alco- Polyamide/SDS-clay hybrid nanocomposite membrane application to
hol) and polyaniline, J. Membr. Sci. 260 (2005) 131. water–ethanol mixture pervaporation separation, J. Membr. Sci. 239 (2004)
[13] T.M. Aminabhavi, H.G. Naik, Synthesis of graft copolymeric membranes of 219.
poly(vinyl alcohol) and polyacrylamide for the pervaporation separation of [43] S. Shanmugam, B. Vishwanathan, T.K. Varadarajan, Esterification by solid
water/acetic acid mixtures, J. Appl. Polym. Sci. 83 (2002) 244. catalysts—a comparison, J. Mol. Catal. A 223 (2004) 143.
[14] M.D. Kurkuri, U.S. Toti, T.M. Aminabhavi, Synthesis and characterization of [44] I.V. Kozhevnikov, Catalysis by heteropoly acids and multicomponent polyox-
blend membranes of sodium alginate and poly(vinyl alcohol) for pervapora- ometalates in liquid-phase reactions, Chem. Rev. 98 (1998) 171.
tion separation of water–isopropanol mixtures, J. Appl. Polym. Sci. 86 (2002) [45] T.M. Aminabhavi, M.B. Patil, S.D. Bhat, A.B. Halgeri, R.P. Vijayalakshmi, P. Kumar,
3642. Activated charcoal-loaded composite membranes of sodium alginate in perva-
[15] T.M. Aminabhavi, U.S. Toti, Pervaporation separation of water–acetic acid poration separation of water–organic azeotropes, J. Appl. Polym. Sci. 113 (2009)
mixtures using polymer membranes, Des. Monomers Polym. 6 (2003) 966.
211. [46] Q.G. Zhang, Q.L. Liu, Z.Y. Jiang, L.Y. Ye, X.H. Zhang, Effects of annealing on
[16] B.V.K. Naidu, S.D. Bhat, A.C. Wali, D.P. Sawant, N.N. Mallikarjuna, S.B. Hal- the physico-chemical structure and permeation performance of novel hybrid
ligudi, T.M. Aminabhavi, Comparative study on the pervaporation separation membranes of poly(vinyl alcohol)/␥-aminopropyl-triethoxysilane, Microp-
of water + acetonitrile mixtures using zeolite-filled sodium alginate and poly- orous Mesoporous Mater. 110 (2008) 379.
(vinyl alcohol)–polyaniline IPN membranes, J. Appl. Polym. Sci. 96 (2004) [47] D. Bahrens, R. Eckermann (Eds.), Vapor–liquid Equilibrium Data Collection, vol.
1968. 1, Part 6C, DECHEMA, 1983.
[17] S.G. Adoor, B. Prathab, L.S. Manjeshwar, T.M. Aminabhavi, Mixed matrix [48] G.Y. Moon, R. Pal, R.Y.M. Huang, Novel two-ply composite membranes of chi-
membranes of sodium alginate and poly(vinyl alcohol) for pervaporation dehy- tosan and sodium alginate for the pervaporation dehydration of isopropanol
dration of isopropanol at different temperatures, Polymer 48 (2007) 5417. and ethanol, J. Membr. Sci. 156 (1999) 17.
Author's personal copy

V.T. Magalad et al. / Journal of Membrane Science 354 (2010) 150–161 161

[49] P. Kanti, K. Srigowri, J. Madhuri, B. Smitha, S. Sridhar, Dehydration of ethanol [51] C.K. Yeom, J.G. Jegal, K.H. Lee, Characterization of relaxation phenomena and
through blend membranes of chitosan and sodium alginate by pervaporation, permeation behaviors in sodium alginate membrane during pervaporation sep-
Sep. Purif. Technol. 40 (2004) 259. aration of ethanol–water mixture, J. Appl. Polym. Sci. 62 (1996) 1561.
[50] X.P. Wang, Z.Q. Shen, Y.F. Zhang, Zhang, A novel composite chitosan mem- [52] A. Toutianoush, L. Krasemann, B. Tieke, Polyelectrolyte multilayer membranes
brane for the separation of alcohol–water mixtures, J. Membr. Sci. 119 (1996) for pervaporation separation of alcohol/water mixtures, Colloids Surf. A 198
191. (2002) 881–889.

You might also like