Você está na página 1de 10

Available online at www.sciencedirect.

com

Separation and Purification Technology 60 (2008) 5463

Performance of a nanofiltration membrane for removal


of ethanol from aqueous solutions by pervaporation
Adrian Verhoef a, , Alberto Figoli b , Bram Leen a , Ben Bettens a ,
Enrico Drioli b , Bart Van der Bruggen a
a

Department of Chemical Engineering, Laboratory for Applied Physical Chemistry and Environmental Technology,
Katholieke Universiteit Leuven, W. de Croylaan 46, B-3001 Leuven, Belgium
b Research Institute on Membrane Technology (ITM-CNR), c/o University of Calabria,
via P. Bucci, Cubo 17/C, 87030 Rende, Cosenza, Italy
Received 30 December 2006; received in revised form 10 July 2007; accepted 26 July 2007

Abstract
In this study, the performance of a hydrophobic nanofiltration membrane (SolSep 3360) for treating alcoholic solutions by pervaporation
is investigated and compared to a conventional pervaporation membrane (PV 1070, Sulzer Chemtech and Pervatech PDMS). Both binary
ethanol/water mixtures and common multicomponent mixtures (alcoholic beverages) are examined. The experiments were performed at feed
ethanol concentrations up to 50 vol% and at temperatures up to 45 C.
The effects of feed ethanol content and temperature were studied in terms of: (1) fluxes and permeances of individual components, and (2)
separation factor, enrichment factor and selectivity of ethanol to water. Using permeance and selectivity instead of flux and separation/enrichment
factor allows the effects on performance evaluation of operating conditions, such as temperature and swelling, to be decoupled. In this way the
contribution by nature of the membrane to separation performance can be clarified and quantified. In addition, previous analyses indicate that
the aqueous activity coefficient and the saturated vapour pressure play an important role when evaluating the membrane performance in terms of
permeance and selectivity. This is confirmed by this study.
Furthermore, it was found that multicomponent alcoholic beverages behave in exact the same manner as binary ethanol/water mixtures. Using a
nanofiltration membrane for pervaporation purposes is a suitable possibility, because of the higher fluxes and permeances, while remaining a good
separation factor and selectivity. The difference between nanofiltration and pervaporation membranes is explained by the influence of swelling,
making the membrane more dense, and the different interactions between permeating molecules and the membrane.
2007 Elsevier B.V. All rights reserved.
Keywords: Pervaporation; Nanofiltration membrane; Alcoholic solutions; PDMS membranes

1. Introduction
In the beginning of the 20th century, Kober [1] first observed
that a membrane could efficiently separate two liquid chemicals mixed together, by applying a vacuum on the other side of
the membrane, resulting in a gradient of chemical potential. In
reaction to this gradient, the components of the mixture penetrate
into the membrane and evaporate on the other side. Kober named
this phenomenon pervaporation. Separation is ensured by differ-

Corresponding author. Tel.: +32 16 32 23 64; fax: +32 16 32 29 91.


E-mail addresses: adrian.verhoef@cit.kuleuven.be (A. Verhoef),
a.figoli@itm.cnr.it (A. Figoli).
1383-5866/$ see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.seppur.2007.07.044

ences in solvents sorption affinity and diffusion coefficients in


the membrane.
Industrially, pervaporation is particularly useful to separate
mixtures hard to separate by conventional techniques, such as
distillation or extraction. Examples are azeotropic mixtures,
such as alcohol/water, or chemical products with close boiling
points, such as acetic acid/water.
For every membrane process, a good membrane must be
found, in order to obtain optimal separation. Many studies prove
the membrane has significant influence. The selectivity and performance of the membrane is not only determined by obvious
features such as thickness, porous or dense nature of the top
layer [25], pore size and geometry, and porosity [2,3,6,7],
but also by less obvious material properties such as glass

A. Verhoef et al. / Separation and Purication Technology 60 (2008) 5463

Nomenclature
A
F
J
l
Mw
m
p
S
t
x
x
y
y

membrane area (m2 )


membrane permeability (kg m1 h1 bar1 )
flux (kg m2 h1 )
membrane thickness (m)
molar mass (kg/mol)
permeate mass (kg)
partial pressure (bar)
ideal membrane selectivity
measurement time (s)
feed weight fraction
feed molar fraction
permeate weight fraction
permeate molar fraction

Greek letters

separation factor

enrichment factor

activity coefficient
Subscripts
i
component i
j
component j
tot
total parameter
Superscripts
p
permeate
sat
saturated vapour phase

transition temperature [2,8], composition [2,9,10], hydrophobicity/hydrophilicity [2,3,6,11,12] and membrane surface charge
[2,13]. For each membrane process, optimal membrane properties are determined separately in terms of structure and material.
For pervaporation, rather thick membranes with a dense nature
are used, in view of chemical stability. In contrast, nanofiltration membranes can be dense or nanoporous and are as thin as
possible, because the flux through the membrane is inversely
proportional to its thickness. In literature, specific membranes
are reported to be suitable for both nanofiltration and pervaporation [14,15]. Both polymeric and inorganic membrane materials
were used for this purpose.
Polymers are a commonly used material for membranes.
However, upon wetting, they swell, altering the structure of the
membrane [12]. Swelling occurs because a solvent enters and
passes through the membrane, due to a chemical potential gradient. This increases permeability, but decreases selectivity, since
another component in the feed mixture can benefit from the now
available free volume inside the membrane, and permeate as
well [16].
This property can be used as an advantage. The swelling
phenomenon can make the structure of a polymeric micro- or
nanoporous nanofiltration membrane more dense [2]. In this
study, the effect of swelling is examined to see whether a
hydrophobic nanofiltration membrane can be used for separating
alcohol/water mixtures by pervaporation. In order to investigate

55

the behaviour of the membrane with common multicomponent solutions, alcoholic beverages were used for pervaporation
experiments.
Since polymeric materials have a lower cost, the use of
nanofiltration membranes in pervaporation can be expected to
broaden the application range of membrane processes in industry.
2. Materials and methods
2.1. Membranes
All membranes in this study have a PDMS (poly dimethyl
siloxane)-based top layer. This is a silicon elastomer with a
typical hydrophobic character. Three different membranes were
used: SolSep 3360 is a hydrophobic nanofiltration membrane
manufactured by SolSep BV (Apeldoorn, the Netherlands); Pervatech PDMS is a dense hydrophobic pervaporation membrane
manufactured by Pervatech BV (Enter, the Netherlands); and PV
1070 is a zeolite-filled, dense pervaporation membrane manufactured by Sulzer Chemtech (Neunkirchen, Germany).
Experiments were performed on the SolSep 3360 nanofiltration membrane. These results were compared to those of the two
other traditional pervaporation membranes. The results for the
Pervatech membrane were obtained by Ranieri et al. [17] on the
same set-up.
2.2. Chemicals
Measurements were performed with ethanol/water mixtures.
The ethanol concentration varies up to 50 vol%. Ethanol was of
analytical purity (>99.8%) and was obtained from Carlo Erba.
In this study, it was also tested if common multicomponent mixtures show identical pervaporation behaviour as
ethanol/water mixtures. To investigate this, alcoholic beverages
were measured. Lager beer (Becks, 5 vol% alcohol), white wine
(Marino, Le Contrade 2004, 11.5 vol% alcohol) and gin (Argia
Gin, extra dry, 38 vol% alcohol) were used as common multicomponent mixtures containing ethanol. These beverages were
chosen since they do not contain too many impermeable components, to prevent fouling [18], and are not too viscous, in order to
avoid fluid mechanical phenomena such as concentration polarisation.
Earlier research [19,20] showed that having a non-viscous
feed not necessarily avoids concentration polarisation, but slow
sorption of permeants in the membrane ensures that permeation
fluxes are independent of feed membrane concentration. The
relatively low separation factor in this work indicates this is the
case.
2.3. Pervaporation experiments
In the pervaporation system, schematically shown in Fig. 1,
the feed temperature is controlled by a thermostat (T). A recirculation pump circulates the feed through the set-up at a flow
rate of 0.7 l/min. The pervaporation module contains a circular
flat sheet membrane with a membrane area of 56.74 cm2 .

56

A. Verhoef et al. / Separation and Purication Technology 60 (2008) 5463

Fig. 1. Schematic diagram of the pervaporation set-up.

At the permeate side, a vacuum is maintained below 5 mbar


by a vacuum pump, and the permeate pressure is measured with
a manometer (P). Two condensers are used, cooled with liquid
nitrogen. The permeate is collected in the first condenser, and
the second protects the vacuum pump.
Several experiments are performed at different concentrations up to 50 vol% and temperatures below 50 C, to avoid
problems with the membrane module, such as blocking of the
cold trap. For each experiment, at least three measurements of
about 1 h are performed and averaged to calculate fluxes. After
the completion of each experiment, the permeate collected inside
the cold trap was warmed up to room temperature. The permeate
was then weighed to determine the total flux.
For each measurement, samples are taken from feed and
permeate. The concentration of ethanol in these samples is determined by means of gas chromatography. The apparatus used is
an Agrilent Technologies 6890N. The capillary column consists
of a stainless steel Porapack Q 80/20s mesh coating. The furnace
temperature is kept constant at 120 C, N2 is used as carrier gas
and the pressure is held constant at 1.29 bar. The detector used
is a TCD detector.

Here, [Fi /l] is the membrane permeability Fi divided by the


membrane thickness l, also called the permeance; x i the mole
fraction in the feed; i the activity coefficient; psat
i the saturated
vapour pressure; y i the permeate mole fraction and pp is the permeate pressure. The activity coefficients were calculated using
the UNIFAC equation; the saturated vapour pressures using the
ClausiusClapeyron equation [21]. These two equations give a
driving force based on molar fractions, but are used since no
correct equations based on weight fractions are available to calculate these parameters. Probably this difference will not lead
to large errors in the calculations.
The ideal membrane selectivity Si/j is now defined as the ratio
of the permeances:
Si/j =

[Fi / l]
[Fj / l]

(4)

The separation factor is determined by Eq. (5). Here, xi is the


feed concentration of the preferentially permeating component
(ethanol in this study), and yi the permeate concentration of this
component.
=

yi /(1 yi )
xi /(1 xi )

(5)

The enrichment factor is calculated by dividing the permeate


concentration by the feed concentration, as illustrated in Eq. (6).
yi
=
(6)
xi
For the membranes in this study, i means ethanol, and j means
water, since ethanol is the preferentially permeating component.
Attention must be paid that, although the driving force is
based on molar fractions, all parameters used in this research
are calculated in mass units.

2.4. Theoretical analysis of the performance parameters


3. Results and discussion
Pervaporation, as well as other membrane separation processes, can be characterized by performance parameters like
flux and selectivity, which are indicators of the ability of the
process for the extraction of a chosen component. In this study,
the membrane efficiency and selectivity is defined by the separation factor , the enrichment factor and the ideal membrane
selectivity S.
Total fluxes were derived from measurements using Eq. (1),
where m is the mass of the permeate, A is the membrane area
and t is the measurement time.
m
Jtot =
(1)
At
Partial fluxes were determined by Eq. (2), with xi the weight
fraction of component i in the membrane.
Ji = Jtot xi

(2)

From these partial fluxes, the permeance can be calculated from


Eq. (3).
 
Fi
Ji
(3)
=  sat
l
(xi i pi yi pp )

3.1. Effects of feed composition on uxes and permeances


of ethanol and water
Fig. 2a illustrates the variations of total flux, partial ethanol
flux and partial water flux for a binary system at 33 C, whereas
Fig. 2b shows the corresponding ethanol and water permeance versus feed water concentration at 33 C. Fig. 2a indicates
that the partial ethanol flux increases rapidly with feed ethanol
concentration. The flux versus feed ethanol concentration relationship can be elaborated by the interaction between polymer
molecules in the membrane and permeating molecules. In
the SolSep 3360 membrane, the selective layer is made of
poly dimethyl siloxane (PDMS), a silicon elastomer with a
typical hydrophobic character. Thus, it is non-polar and experiences strong interactions with non-polar compounds. In an
ethanol/water mixture, ethanol is the least polar component,
and will therefore have more interaction with the PDMS membrane. At higher ethanol concentrations in the feed, more ethanol
molecules are in contact with the selective layer of the membrane. Therefore, more ethanol molecules are sorbed in the
membrane, causing a greater degree of swelling in the top

A. Verhoef et al. / Separation and Purication Technology 60 (2008) 5463

Fig. 2. (a) Total flux (Jtot ), ethanol flux (JEtOH ) and water flux (JH2 O ) vs. feed

ethanol composition at 33 C for SolSep 3360. (b) Ethanol permeance (FEtOH
)
and water permeance (FH 2 O ) vs. feed ethanol composition at 33 C for SolSep
3360.

layer. Consequently, more ethanol molecules are able to pass


through the ethanol-swollen membrane and ethanol permeation
flux increases as the feed ethanol concentration increases.
At the same time however, because of this swelling, water is
also able to permeate through the membrane, because of drag
effects or possibly as a coupling unit of alcohol and water. This
causes the water flux to remain constant, while a decrease would
be expected, since the ethanol flux has increased. At low ethanol
concentrations, the water flux is even higher than the ethanol
flux. This can be explained by the fact that at low ethanol
concentrations the difference in polarity of water and ethanol
is counterbalanced by the larger amount of water molecules
present. Therefore, the water flux will be higher than the ethanol
flux.
The permeance versus feed ethanol concentration plot
(Fig. 2b) is not fully in agreement with the previous analysis
for ethanol. When comparing the permeance and flux increment
percentages with increasing feed concentration for both water
and ethanol, it can be seen that for water these are comparable.
However, for ethanol the increment percentage permeance is
lower than that of flux. When looking at the formula for permeance (Eq. (3)), the denominator shows that the effect of fugacity
difference as the driving force is significantly decoupled from the

57

Fig. 3. (a) Ethanol flux (JEtOH ) vs. feed composition at 23, 33 and 44 C for
SolSep 3360. (b). Water flux (JH2 O ) vs. feed composition at 23, 33 and 44 C
for SolSep 3360.

overall membrane transport. The fugacity difference can be written in permeant-specific terms. Therefore, permeance exhibits
more accurate permeant-specific transport properties with feed
concentration. This explains the difference in increment percentages.
3.2. Effects of temperature on uxes and permeances of
ethanol and water
Trends of ethanol and water fluxes versus feed ethanol concentration as a function of temperature are presented in Fig. 3a
and b, respectively, for 23, 33 and 44 C. Both ethanol and
water flux increase with temperature. This phenomenon can
be explained by a thermally induced increased motion of the
polymer chains and hence an expansion of the free volume.
In addition, the increased thermal motions of the permeating
molecules also promote their diffusion. The combination of
these effects brings about the rapid increase in permeation flux.
In Fig. 4a and b, respectively, variations as a function of temperature in ethanol and water permeances versus feed ethanol
concentration are given for 23, 33 and 44 C. The effects of temperature on ethanol and water permeances are quite opposite
to what is expected and to the permeance versus temperature

58

A. Verhoef et al. / Separation and Purication Technology 60 (2008) 5463


Fig. 4. (a) Ethanol permeance (FEtOH
) vs. feed composition at 23, 33 and 44 C
for SolSep 3360. (b). Water permeance (FH 2 O ) vs. feed composition at 23, 33
and 44 C for SolSep 3360.

relationship observed in most gas separation membranes, i.e.


permeance does not increase with temperature. The highest permeance is found for the system at 23 C, followed by the system
at 33 C, and then 44 C (when extrapolating to higher feed
ethanol concentrations). The negative dependence of ethanol
and water permeances on temperature may arise from the fact
that permeance is defined as permeant flux divided by permeant driving force. The driving force combines two temperature
dependent factors: i and psat
i , which are external factors outside
the membrane. This is independent from the fact whether they
are based on molar or weight fractions. As shown in Table 1,
the values of ethanol and water activity coefficients are quite
close at different temperatures. Thus, psat
i plays a more important role than the activity coefficient. Mathematically, a higher

temperature results in a higher psat


i , a larger denominator of Eq.
(3) (since yi pp usually is negligible), and a smaller permeance.
The predicted order of permeance versus temperature for both
ethanol and water will follow the order 23 C > 33 C > 44 C.
In addition, the negative and positive temperature dependence
of i and psat
i , respectively, may also counterbalance each others
effect on permeance. Thermodynamically, the indifference of
permeance to temperature at lower feed ethanol concentrations
indicates that there is a high degree of counterbalance of the negative temperature effect on sorption and the positive temperature
effect on diffusion. An increase in temperature tends to reduce
sorption, but increase diffusion. However, the degree of counterbalance is reduced at high ethanol concentrations, possibly
because of swelling.
In the past, similar approaches have been used. Wijmans and
Baker [20] are the first to use permeances, called pressurenormalized fluxes, for performance analysis to quantify the
contribution of the vapour liquid equilibrium. Ten and Field [22]
used a similar approach. Although that is more complicated, it
can be reduced to the equations used in this research.
These analyses demonstrate the importance of taking the
combined effects of activity coefficient and saturation vapour
pressure into consideration when analysing the temperature
effect on membrane performance. The external factors seriously
affect the apparent membrane performance of membranes in
pervaporation. Without using the permeance as a membrane performance parameter, these effects may be simply overlooked
from the flux plots. Permeance may reflect the true separation
potential of the membrane because flux comprises the effects of
external operational factors, such as temperature and swelling,
and the intrinsic membrane properties.
3.3. Separation and enrichment factors versus selectivity of
ethanol to water
Fig. 5ac illustrates, respectively, the separation factor,
enrichment factor and selectivity of ethanol to water for the
SolSep 3360 membrane at different temperatures. Both separation and enrichment factor decrease with an increase in feed
ethanol concentration. Increasing feed ethanol content enhances
membrane swelling, resulting in an enlarged interstitial space
between the polymer chains and declined separation performance.
The selectivity (defined as the ratio of the permeances) shows
a tendency completely different from that of the separation and
enrichment factor: with increasing feed ethanol concentration,
the selectivity also increases. This can also be concluded from

Table 1
A comparison of ethanol and water activity coefficients at different temperatures (ethanol concentration = 10, 30 or 50 vol%), calculated with help of the UNIFAC
equation [21]
Activity coefficient

23 C
33 C
44 C

Ethanol (10%)

Water (90%)

Ethanol (30%)

Water (70%)

Ethanol (50%)

Water (50%)

5.6716
5.6070
5.5385

1.0046
1.0045
1.0044

3.5829
3.5758
3.5660

1.0348
1.0340
1.0333

2.5794
2.5874
2.5938

1.0835
1.0819
1.0804

A. Verhoef et al. / Separation and Purication Technology 60 (2008) 5463

59

Fig. 6. Permeate ethanol concentration ([EtOH]) vs. feed ethanol concentration


at 23, 33 and 44 C for SolSep 3360.

rate permeant-specific transport properties with feed ethanol


concentration. Since the SolSep 3360 membrane is a nanofiltration membrane, under non-swollen conditions, the interstitial
spaces in and between the polymer chains are larger than in
a pervaporation membrane. Upon swelling, the membrane free
volume is altered. This alteration causes lower separation and
enrichment factors because of drag and coupling effects. However, also a more dense membrane structure [3] is formed.
Therefore, the permeant-specific transport properties of a highly
swollen nanofiltration membrane will be comparable to those
of a dense PDMS pervaporation membrane. This explains why
an increased feed ethanol concentration (and thus increased
swelling) results in a small increment in membrane selectivity,
which is the ratio of selectivities of the components.
Fig. 5a and b shows no temperature dependence for both separation and enrichment factor. This is in agreement with the results
obtained for the permeate ethanol concentration versus the feed
ethanol concentration at different temperatures, as depicted in
Fig. 6. With the given definitions for both factors (Eqs. (5) and
(6)), this equal tendency comes as no surprise, since the separation factor can be written as the ratio of enrichment factors of
both components:
=

Fig. 5. (a) Separation factor () vs. feed ethanol concentration at 23, 33 and 44 C
for SolSep 3360. (b) Enrichment factor () vs. feed ethanol concentration at 23,
33 and 44 C for SolSep 3360. (c) Selectivity (S) vs. feed ethanol concentration
at 23, 33 and 44 C for SolSep 3360.

the plot of the permeances versus feed ethanol concentration, as


shown in Fig. 2b, since water permeance remains constant and
ethanol permeance increases.
From the influence of feed composition on permeances, it
was concluded that the effect of fugacity difference as the driving force is significantly decoupled from the overall membrane
transport, thereby causing permeance to exhibit more accu-

yi /yj
i
=
xi /xj
j

(7)

For the selectivity no distinct temperature dependence can be


seen (Fig. 5c), though this is not as definite as for the separation
and enrichment factors. This small difference can be explained
by looking at the definitions of selectivity (Eq. (4)) and separation factor (Eq. (5)). With the help of some general assumptions,
very common in chemical engineering, a correlation between the
two parameters can be derived.
If a vacuum is applied at the permeate side of the membrane and all permeate is condensed immediately, the permeant
concentration near the membrane surface can be written as
the permeate flux. Then the separation factor can be rewritten (Eq. (8)) as a combination of both membrane transport
properties (ethanol and water permeances) and component properties (activity coefficients and saturated vapour pressures of
both ethanol and water). The selectivity is mainly dominated by

60

A. Verhoef et al. / Separation and Purication Technology 60 (2008) 5463

membrane transport properties (Eq. (9)).


=
=

JEtOH
x H2 O

xEtOH
JH 2 O


p
EtOH psat
x H2 O
[FEtOH / l] (xEtOH
EtOH yEtOH p )



xEtOH
[FH2 O / l]
(xH
psat yH
pp )
2 O H2 O H 2 O
2O

=S

S=

factor and selectivity. However, neither separation and enrichment factor, nor the selectivity, show significant temperature
dependence.

Mw,H2 O EtOH psat


EtOH
Mw,EtOH H2 O psat
H2 O

[FEtOH / l]
[FH2 O / l]

3.4. Common multicomponent mixtures

(8)

(9)

Since the downstream pressure is normally very low compared


to the upstream pressure, it may be neglected, and the separation
factor can be simplified to the selectivity multiplied by the ratio
of the activity coefficient and saturated vapour pressure divided
by the molecular mass of ethanol to water. With the temperature
variation, the ratio of the activity coefficient and saturated vapour
pressure of ethanol will change accordingly and counterbalance
temperature dependence of the permeances. Therefore, there is
a small difference in temperature dependence of the separation

With the SolSep 3360 membrane it was also tested whether


common multicomponent systems behave in a similar way to
binary ethanol/water mixtures. Fig. 7ad gives the results for
both the partial fluxes (Fig. 7a and b) and permeances (Fig. 7c
and d) for ethanol and water. Fig. 8ac shows the results
for the separation factor, enrichment factor and selectivity,
respectively.
In the past, Lipnizki showed that the presence of impermeable components in the feed can change the selectivity of the
membrane [18]. As can be seen in this research, the tested
alcoholic beverages behave exactly in the same way as the
binary ethanol/water mixtures. Apparently, the unknown extra
components in the mixtures have no influence on the pervaporation properties or counterbalance each others effects,
ensuring these mixtures behave like regular ethanol/water
mixtures.

Fig. 7. (a) Ethanol flux vs. feed ethanol concentration of ethanol/mixtures and common multicomponent mixtures at different temperatures for SolSep 3360. (b) Water
flux vs. feed ethanol concentration of ethanol/mixtures and common multicomponent mixtures at different temperatures for SolSep 3360. (c) Ethanol permeance
vs. feed ethanol concentration of ethanol/mixtures and common multicomponent mixtures at different temperatures for SolSep 3360. (d) Water permeance vs. feed
ethanol concentration of ethanol/mixtures and common multicomponent mixtures at different temperatures for SolSep 3360.

A. Verhoef et al. / Separation and Purication Technology 60 (2008) 5463

Fig. 8. (a) Separation factor vs. feed ethanol concentration of ethanol/mixtures


and common multicomponent mixtures at different temperatures for SolSep
3360. (b) Enrichment factor vs. feed ethanol concentration of ethanol/mixtures
and real multicomponent mixtures at different temperatures for SolSep 3360.
(c) Selectivity vs. feed ethanol concentration of ethanol/mixtures and real multicomponent mixtures at different temperatures for SolSep 3360.

3.5. Comparison with pervaporation membranes


In the past, different studies on nanofiltration membranes
[2325] showed that whether convection or diffusion is the most
important transport mechanism, the solute transport rate is influ-

61

enced by the effective pore size (in case of convection) or the


available free volume (in case of diffusion) and the effective
solute size. When looking at molecule size, it can be noticed that
both water and ethanol are relatively small molecules. Therefore
the effect of sterical hindrance will have a minor influence. By
using a swollen nanofiltration membrane for pervaporation purposes, the pore size of the membrane is altered. The effect of this
change on the pervaporation performance can be checked when
comparing the results of the PDMS nanofiltration membrane
with results of PDMS pervaporation membranes.
Therefore, the results of the SolSep 3360 nanofiltration membrane are compared to the results of two PDMS pervaporation
membranes, namely PV 1070 and Pervatech PDMS. For these
membranes, pervaporation experiments were performed with
binary mixtures of varying ethanol concentrations and at different temperatures.
For the fluxes, similar tendencies are found for all three
membranes, although the order of magnitude differs. At higher
temperatures, the fluxes increase and at higher feed ethanol concentrations, the partial ethanol fluxes increase, whereas the water
fluxes remain constant. One difference can be found for the pervaporation membranes, where the break-even point for dominant
partial flux (at lower feed ethanol concentrations the water flux
dominates) lies at higher feed ethanol concentrations than for the
SolSep 3360 nanofiltration membrane. This indicates stronger
interactions between ethanol molecules and membrane for the
nanofiltration membrane than for the pervaporation membranes.
Another difference is found in the order of magnitude of the
fluxes. For the SolSep 3360 membrane, fluxes in the range of
03.5 kg m2 h1 were found, whereas the fluxes of the PV 1070
membrane staid below 0.3 kg m2 h1 . This can be explained
by the fact that SolSep 3360 is a nanofiltration membrane with
larger free volume, but also because it has a more recent and
advanced chemical structure.
The SolSep 3360 membranes fluxes are nearly twice as large
as the fluxes through the Pervatech membrane at low temperatures. However, at higher temperatures this difference has almost
completely vanished. Apparently, at large temperatures the thermally induced movement of polymer chains in the pervaporation
membrane and the better diffusion of the permeant counterbalance the advantages of the larger pore size of the nanofiltration
membrane.
For the permeances, similar differences in order of magnitude
were found. The permeances of the SolSep 3360 membrane are
several factors larger than of the PV 1070 membrane, for the
same reasons as explained above. For the Pervatech membrane
the order of magnitude of the permeances is smaller than for the
SolSep 3360 membrane at lower temperatures, just as for the
fluxes. At higher temperatures, the permeances approach the
same order of magnitude.
Although the permeances show the same increasing tendency
with feed ethanol concentration, a striking difference when comparing the permeances of the Pervatech and the SolSep 3360
membrane, is the fact that for the SolSep 3360 membrane the
ethanol permeance dominates, while for the Pervatech membrane an opposite behaviour is observed. Since permeance shows
more accurate permeant-specific transport properties, it appears

62

A. Verhoef et al. / Separation and Purication Technology 60 (2008) 5463

water has more interaction with the Pervatech membrane than


with the SolSep 3360 membrane. The obtained higher water
fluxes than ethanol fluxes at low feed ethanol concentration also
manifest this.
When looking at the temperature influence, it was found the
ethanol permeance of the PV 1070 membrane hardly changes,
whereas it was expected it would decrease. Possibly, the swelling
has no effect here on the degree of counterbalance between the
negative temperature effect on sorption and the positive temperature effect on diffusion.
For the Pervatech membrane, with increasing feed ethanol
concentration a decreasing separation factor is found, while the
selectivity shows an increasing tendency, just as for the SolSep
3360 membrane. The selectivity of the PV 1070 membrane
shows the same trend. For this membrane however, constant separation and enrichment factors are reported at low temperatures.
Apparently, the swelling of the membrane has no influence on the
separation and enrichment factor of a pervaporation membrane
under these conditions. At higher temperatures, the separation
and enrichment factors decrease, as do they in the SolSep 3360
membrane.
Where there is little or no temperature dependence for the
SolSep 3360 membrane for separation and enrichment factors
and selectivity, the PV 1070 membrane shows larger values for
all three parameters at higher temperatures. This is probably
because the ethanol permeance for this membrane is higher than
expected, resulting in a larger ratio in Eqs. (8) and (9).

4. Conclusion
This research confirms findings of earlier studies, that plots of
flux and permeance versus feed ethanol content may give different results and conclusions on effects of processing conditions
on separation performance. Using permeance and selectivity, instead of flux and separation and enrichment factor, can
significantly decouple the effect of operating conditions on
performance evaluation, while clarifying and quantifying the
contribution by the nature of the membrane to the separation
performance. Aqueous activity coefficient and saturated vapour
pressure appear to play an important role when using permeance
and selectivity in the performance analysis.
Furthermore, it was found that the membrane performance
in terms of permeance and selectivity, as well as in terms of
flux, separation and enrichment factors, is similar when alcoholic beverages are used as feed instead of synthetic solutions.
Apparently, the extra components in these mixtures have no
influence on the pervaporation properties, or their influences
are counterbalanced.
Compared to two PDMS-based pervaporation membranes,
both fluxes and permeances are larger in the nanofiltration membrane, while the separation factor and selectivity are comparable
or even better. The difference between the nanofiltration and
pervaporation membranes can be explained by the influence
of swelling and the different interactions between permeating
molecules and the membrane. This makes a nanofiltration membrane very useful for pervaporation.

Acknowledgments
The Research Council of the K.U. Leuven is gratefully acknowledged for financial support (OT/2002/33 and
OT/2006/37). SolSep and Pervatech are thanked for kindly
supplying membrane samples. The Socrates program is also
gratefully acknowledged for the grant given to Bram Leen for
performing the experimental work at ITM-CNR, ITALY.
References
[1] P.A. Kober, Pervaporation, perstillation and percrystallization, J. Am.
Chem. Soc. 39 (5) (1917) 944948.
[2] B. Van der Bruggen, J.C. Jansen, A. Figoli, J. Geens, K. Boussu, E. Drioli,
Characteristics and performance of a universal membrane suitable for gas
separation, pervaporation and nanofiltration applications, J. Phys. Chem.
B 110 (28) (2006) 1379913803.
[3] B. Van der Bruggen, J.C. Jansen, A. Figoli, J. Geens, D. Van Baelen, E.
Drioli, C. Vandecasteele, Determination of parameters affecting transport in
polymeric membranes: parallels between pervaporation and nanofiltration,
J. Phys. Chem. B 108 (35) (2004) 1327313279.
[4] J.P.G. Villaluenga, M. Khayet, P. Godino, B. Seoane, J.I. Mengual, Analysis of the membrane thickness effect on the pervaporation separation of
methanol/methyl tertiary butyl ether mixtures, Sep. Purif. Technol. 47 (12)
(2005) 8087.
[5] X.-P. Wang, Y.-F. Feng, Z.-Q. Shen, Pervaporation properties of a threelayer structure composite membrane, J. Appl. Polym. Sci. 75 (6) (2000)
740745.
[6] J. Caro, M. Noack, P. Kolsch, Chemically modified ceramic membranes,
Microporous Mesoporous Mater. 22 (13) (1998) 321332.
[7] S.S. Madaeni, The effect of surface characteristics on RO membrane performance, Desalination 139 (13) (2001) 371.
[8] J. Meier-Haack, W. Lenk, S. Berwald, T. Rieser, K. Lunkwitz, Influence of
thermal treatment on the pervaporation separation properties of polyamide6 membranes, Sep. Purif. Technol. 19 (3) (2000) 199207.
[9] S. Ray, S.K. Ray, Effect of copolymer type and composition on separation characteristics of pervaporation membranesA case study with
separation of acetonewater mixtures, J. Membr. Sci. 270 (12) (2006)
7387.
[10] S.V. Satyanarayana, A. Sharma, P.K. Bhattacharya, Composite membranes
for hydrophobic pervaporation: study with the toluenewater system,
Chem. Eng. J. 102 (2) (2004) 171184.
[11] S.V. Satyanarayana, P.K. Bhattacharya, Pervaporation of hydrazine
hydrate: separation characteristics of membranes with hydrophilic to
hydrophobic behaviour, J. Membr. Sci. 238 (12) (2004) 103115.
[12] S.I. Semenova, H. Ohya, K. Soontarapa, Hydrophilic membranes
for pervaporation, an analytical review, Desalination 110 (3) (1997)
251286.
[13] J. Meier-Haack, T. Rieser, W. Lenk, D. Lehmann, S. Berwald, S. Schwarz,
Effect of polyelectrolyte complex layers on the separation properties and
the fouling behavior of surface and bulk modified membranes, Chem. Eng.
Technol. 23 (2) (2000) 114118.
[14] J. Liu, W.K. Teo, C.H. Chew, L.M. Gan, Nanofiltration membranes
prepared by direct microemulsion copolymerization using poly(ethylene
oxide) macromonomer as a polymerizable surfactant, J. Appl. Polym. Sci.
77 (12) (2000) 27852794.
[15] J. Sekulic, J.E. ten Elshof, D.H.A. Blank, A microporous titania membrane for nanofiltration and pervaporation, Adv. Mater. 16 (17) (2004)
15461550.
[16] B. Barri`ere, L. Leibler, Permeation of a solvent mixture through an elastomeric membraneThe case of pervaporation, J. Polym. Sci. Part B 41
(2) (2003) 183193.
[17] R. Ranieri, A. Figoli, A. Criscuoli, E. Drioli, La pervaporazione e i membrane contactors come sistemi alternative per la determinazione dellalcool
etilico in soluzioni idroalcoliche, Master Thesis, ITM-CNR, Rende (CS),
Italy, 2005.

A. Verhoef et al. / Separation and Purication Technology 60 (2008) 5463


[18] F. Lipnizki, S. Hausmanns, R.W. Field, Influence of impermeable components on the permeation of aqueous 1-propanol mixtures in hydrophobic
pervaporation, J. Membr. Sci. 228 (2) (2004) 129138.
[19] P. Cote, C. Lipski, Mass transfer limitations in pervaporation for water and
wastewater treatment, in: Proceedings of the 3rd Conference on Pervaporation Processes in Chemical Industry, 1988, pp. 449462.
[20] J.G. Wijmans, R.W. Baker, A simple predictive treatment of the permeation process in pervaporation, J. Membr. Sci. 79 (1) (1993) 101
113.
[21] VLECalc v1.1. http://www.softlookup.com/display.asp?id=3122 (accessed August 2006).
[22] P.K. Ten, R.W. Field, Organophilic pervaporation: an engineering science
analysis of component transport and the classification of behaviour with

63

reference to the effect of permeate pressure, Chem. Eng. Sci. 55 (8) (2000)
14251445.
[23] L. Braeken, R. Ramaekers, Y. Zhang, G. Maes, B. Van der Bruggen, C.
Vandecasteele, Influence of hydrophobicity on retention in nanofiltration
of aqueous solutions containing organic compounds, J. Membr. Sci. 252
(12) (2005) 195203.
[24] J. Geens, A. Hillen, B. Bettens, B. Van der Bruggen, C. Vandecasteele,
Solute transport in non-aqueous nanofiltration: effect of membrane material, J. Chem. Technol. Biotechnol. 80 (12) (2005) 13711377.
[25] J. Geens, K. Peeters, B. Van der Bruggen, C. Vandecasteele, Polymeric
nanofiltration of binary wateralcohol mixtures: influence of feed composition and membrane properties on permeability and rejection, J. Membr.
Sci. 255 (12) (2005) 255264.

Você também pode gostar