Separation Science and Technology

ISSN: 0149-6395 (Print) 1520-5754 (Online) Journal homepage: http://www.tandfonline.com/loi/lsst20

Membrane distillation and pervaporation for

ethanol removal: are we comparing in the right
way?

Bernardo A. Cinelli, Denise M. G. Freire & Frederico A. Kronemberger

To cite this article: Bernardo A. Cinelli, Denise M. G. Freire & Frederico A. Kronemberger (2018):
Membrane distillation and pervaporation for ethanol removal: are we comparing in the right way?,
Separation Science and Technology, DOI: 10.1080/01496395.2018.1498518

To link to this article: https://doi.org/10.1080/01496395.2018.1498518

Published online: 18 Jul 2018.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=lsst20
SEPARATION SCIENCE AND TECHNOLOGY
https://doi.org/10.1080/01496395.2018.1498518

Membrane distillation and pervaporation for ethanol removal: are we

comparing in the right way?
Bernardo A. Cinellia, Denise M. G. Freireb, and Frederico A. Kronemberger a

a
Chemical Engineering Program, COPPE, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil; bInstitute of Chemistry, Federal University
of Rio de Janeiro, Rio de Janeiro, Brazil

ABSTRACT ARTICLE HISTORY

A new glance on the comparison between membrane distillation (MD) and pervaporation is Received 17 January 2018
performed. There is a difficulty in this comparison, mainly due to different hydrodynamic condi- Accepted 5 July 2018
tions reported in the literature. Pervaporation and MD are similar although the comparison KEYWORDS
between these two processes can be tricky. In that way, how can we make a proper comparison Membrane distillation;
between results of these two processes? This study proposes a reasonable comparison between vacuum membrane
MD and pervaporation for ethanol–water separation. Two very distinct regions of results in terms distillation; pervaporation;
of selectivity and flux are presented. Feed temperature and composition and concentration concentration polarization;
polarization effects were also investigated. ethanol recovery

Introduction reducing those effects in the metabolic activity of the

microorganism.
Ethanol is the main biofuel used in the world, and its use
There are several proposals for ethanol removal, such as
is becoming more widespread, with prospects of expand-
solvent extraction;[5,6] gas stripping, for example, with CO2;
ing production and consumption worldwide.[1] Biofuels
and vacuum evaporator flash tank[7,8] although a great
technology leads to significant reduction on greenhouse
attention has been directed toward membrane processes
gas emissions compared to fossil fuels. Despite all ben-
for this application. There are essentially two different
efits, ethanol production also raises concerns about
membrane technology processes described in scientific lit-
excessive use of water and pollution, among other envir-
erature for ethanol removal: pervaporation[9–11] and mem-
onmental impacts. One of the biggest challenges for this
brane distillation (MD).[12–14] In pervaporation, liquid
industry is making improvements in process efficiency,
mixtures are separated by selective interaction of com-
thereby promoting reduction in production costs,
pounds with a dense membrane. The components are
aligned with environmental issues. Among the new
selectively transported through the membrane and then
research trends in this field, process integration has a
vaporized due to lower partial pressure in the permeate
key role for reducing costs and increasing ethanol com-
side, achieved using a vacuum pump or an inert gas stream.
petitiveness related to gasoline.[2] Process intensification
The separation is accomplished by relative permeation of
offers a possible solution that increases competitiveness
solution through the membrane, which depends on both
of the process industry, making industrial processes
thermodynamic (adsorption) and kinetic (diffusion)
more efficient, more productive and faster, also decreas-
aspects. MD is similar to pervaporation process. The main
ing energy consumption and waste generation using
difference between the two processes is the role of the
innovative technical solutions.[3]
membrane. The hydrophobic microporous membrane
Several techniques have been applied for ethanol
acts only as a support to the liquid–vapor interface and
extraction during fermentation, in order to improve the
does not chemically distinguish solution components. The
process performance. The accumulation of ethanol in the
process selectivity depends on the vapor–liquid equilibrium
fermentation medium ultimately inhibits the growth and
phase separation.
viability of ethanol producing microorganisms[4] causing
In this work, we evaluate and propose a reasonable
a toxic effect in yeast. Therefore, while removing it, the
comparison between MD and pervaporation for ethanol
concentration can be kept below inhibitory levels,
separation from water, taking into account experimental

CONTACT Bernardo A. Cinelli bernardo_cinelli@hotmail.com Chemical Engineering Program, Membrane Processes Lab., COPPE/UFRJ, Federal
University of Rio de, P.O. Box: 68502, Janeiro 21941-972, Brazil
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/lsst.
© 2018 Taylor & Francis
2 B. A. CINELLI ET AL.

data and most significantly data reported in the literature, Unlike MD, pervaporation showed a growth in num-
carefully selected in terms of temperature and ethanol ber of publications throughout the 1980s. Therefore, it is
concentration. Experiments of MD and pervaporation a technology that has been studied for a long time,
were performed under same operational and hydrody- although MD has already reached values of 350 publica-
namic conditions, in order to allow a fair comparison tions in 2017, in comparison with the 264 works of
between two different processes. A critical analysis on pervaporation published in the same year. It is noticed
experimental and literature compiled data of different that the curve of publications of MD presents a high
MD settings was carried out, aiming to analyze concentra- growth rate year by year, while the curve of publications
tion polarization effects. A model of mass transport phe- of pervaporation has been relatively stagnant in recent
nomena involved in vacuum membrane distillation years, from 2007 to 2017. When combining MD with
(VMD) process was developed, and reconciled with ther- ethanol, there are very few works, much less than per-
modynamic vapor–liquid equilibrium, composition in vaporation. Important to highlight that the knowledge
interfaces was evaluated. A comparison between actual with MD can be applied for the removal of ethanol, and
and theoretical selectivities allows identifying, in a qualita- there is now enormous potential for growth in this area.
tive and quantitative way, how closely reported results of
MD were of their theoretical maximum, evaluating the
MD principles
effect of hydrodynamic conditions and calculating concen-
tration polarization layer and polarization modulus for In MD, a hydrophobic microporous membrane acts
each case. Finally, taking into account the boundary layer only as support of the vapor–liquid interface, and the
film model, using all MD experimental data, it was possible driving force for transport is maintained by the vapor
to estimate the separation achieved specifically by the pressure difference of a particular component between
selective membrane for the pervaporation experiments. the sides of membrane, which may result from an
imposed temperature difference, or using a vacuum
pump or a sweep gas in the permeate side. The low
MD and pervaporation principles vapor pressure of the permeate side may be configured
in several ways:[1] by direct contact MD (DCMD),[2]
MD and pervaporation background
MD with an “air gap” (AGMD),[3] MD with sweeping
The MD technology was first patented in 1963 while gas (SGMD)[4] and VMD.[16,17] For the MD process, it
the first MD article published in a scientific journal is essential that liquid water does not pass through the
appeared 4 years later, in 1967.[15] In the 1990s, new pores. In this sense, the role of membranes is different
membranes and modules became available and interest from other membrane processes since it acts as a phy-
in MD was renewed. Efforts have been made in devel- sical support for the liquid–vapor interface.
oping this technology, and especially after 2000, the There are a wide variety of membranes that may be
number of published works has increased significantly used in MD. However, polytetrafluoroethylene (PTFE),
(Fig. 1). polypropylene (PP) and polyvinylidene fluoride (PVDF)

350

300
Number of Publications

250

200

150

100

50

0
1980 1985 1990 1995 2000 2005 2010 2015
Years

Figure 1. Publications per year of MD (full diamonds) and pervaporation (empty diamonds) and “MD and ethanol” (full square) and
“pervaporation and ethanol” (empty square) mentioned in the title, keywords, abstract and Scopus database.
 SEPARATION SCIENCE AND TECHNOLOGY 3

polymeric materials are more commonly used due to the bulk of the phase, resulting in higher transport
their low surface tension values.[16,18] Recently, new resistance.
membrane materials, such as hydrophobized ceramic The temperature polarization occurs when tempera-
membranes and carbon nanotubes,[19] which are ture at membrane surface is lower than in bulk of the feed
known to have high porosity and hydrophobicity, are solution because evaporation occurs at the membrane
being used as membrane materials to improve the per- surface and latent heat is removed.[24] Nevertheless, in
formance of MD but are still in the early stages of VMD temperature polarization, it is considerably
development for low cost use and applications. reduced compared with that in DCMD configuration.[25]
Currently, PTFE membrane dominates applications in Izquierdo-Gil and Jonsson[26] indicate that the tempera-
commercial and pilot MD modules because of its high ture polarization effects under their operating conditions
hydrophobicity and good mechanical resistance toward in VMD are not significant and could be neglected. It is
harsh operation conditions.[20] important to highlight that temperature polarization,
According to the current commercial membranes depending on the system, affects MD and pervaporation
usually in MD, some typical values of morphological in terms of total flux but does not strongly affect selectiv-
characteristics are mentioned, such as porosity of mem- ity. While concentration polarization has a greater influ-
brane, which is in the range of 60–95%; the pore size, in ence on flux and also dramatically affects selectivity.
the range of 0.2–1.0 μm; and membrane thickness Several studies in the literature have sought to eval-
between 0.06 and 0.25 mm.[21,22] uate operating parameters that influence flux and selec-
The hydrophobic nature of the material is essential. tivity of MD.[26–29] Bandini et al.[27] investigated VMD
Pore wetting must be avoided, in the first place, in order for ethanol removal using PTFE and PP membranes,
to reduce transport resistance inside pores, and ulti- and they found that the separation factor was limited
mately, if liquid feed passes directly through the pores, by concentration polarization, whereas permeate flux
no separation occurs at all. Thus, the transmembrane was usually accompanied by decreases in selectivity.
hydrostatic pressure must be kept below minimum feed Banat et al.[30] conclude that polarization occurs to a
liquid entry pressure, and according to Young–Laplace significant extent, and the model version which neglects
equation,[21] it can be expressed as follows: these effects does not adequately predict the experi-
mental data at lower feed flow rates.
2Θγ cos θ
ΔP ¼ (1) Regardless of the membrane system configuration
r under evaluation, there is a convergence in their con-
where ΔP is the entry pressure difference, γ is the sur- clusions, noting that the factors that most affect the
face tension of the liquid, Θ is the geometric factor process are temperature; permeate pressure; concentra-
related to the pore structure (equal to 1 for cylindrical tion of solutes; and hydrodynamic conditions, for
pores), θ is the solid–liquid contact angle and r is the example, agitation and feed flow.
pore size.
The MD flux increases with feed temperature since
Pervaporation principles
there is an exponential relationship between the vapor
pressure and temperature which can be described by Pervaporation is a process that a liquid stream contain-
Antoine equation. The vapor pressure difference across ing two or more miscible species is placed in contact
the membrane is the driving force of the MD process, with one side of a dense polymer membrane, while
and so the flux increases exponentially with feed tem- vacuum is typically applied on the other side. The
perature as related by an Arrhenius type of driving force for transport through pervaporation
dependence.[23] The vapor diffusion is also higher at membrane is given by the chemical potential difference
higher feed temperatures and higher will be the flux of species between feed fluid and permeate vapor. The
obtained through the membrane. components sorb and diffuse, permeating through the
The transport resistance is usually low in MD, membrane, and change to the vapor phase after deso-
resulting in higher fluxes. In many cases, exactly rption and so the term “pervaporation.”[31]
due high flux of permeant molecules, concentration A schematic representation of pervaporation mem-
polarization becomes more significant. Concentration brane and the membrane pore in VMD process is
polarization phenomenon is characterized by the shown in Fig. 2.
development of a boundary layer due to decreased The transport mechanism through the membrane pro-
concentration of the component close to the mem- posed by Binning et al.[32] considers that the transport of
brane surface. The concentration of the molecule of permeant components occurs in three successive stages:
interest is lower in the membrane interface than in selective sorption of feed components in the surface layer
4 B. A. CINELLI ET AL.

Figure 2. Schematic representation of pervaporation (left) and VMD (right), where T1 and P1 are the temperature and pressure on
the liquid feed side, respectively; T0 and P0 are the temperature and pressure on the permeate side, respectively; and J is the flux.
Source: Adapted from Drioli and Criscuoli.[21].

of the membrane; diffusion of the dissolved species Also, in MD, concentration polarization can be very
through a swollen film, across the membrane matrix; significant, mainly when the most permeable component
and desorption of components on the permeate side. is at very low concentrations. In general, due to the lower
In this case, the polymeric material of the membrane flux values, its extension is not as high as in the MD process.
has a key influence on selectivity due to its affinity with
the component of interest. Thus, different diffusion
rates through the membrane are observed, and even a Literature data for ethanol separation using MD
component at a low concentration in the feed may be and pervaporation
highly enriched in the permeate. In order to evaluate the ethanol separation by MD and
Therefore, both selectivity and permeate flux are pervaporation processes, Table 1 and Table 2 shows an
controlled substantially by the membrane. The affinity extensive literature review, including flux and selectivity
difference between the polymer that constitutes the data, some operating parameters and experimental con-
selective layer of the membrane and the solute is ditions, for ethanol separation from water or fermenta-
responsible for the selectivity. The flux depends on tion medium. Moreover, almost all selected works present
the mobility of solutes in polymeric material. Thus, ethanol concentrations close to 5 wt% in the feed, similar
the thicker the selective layer, the lower the permeate to typical values for ethanol fermentation broth.
flux. Several membrane materials have been studied ð1CpÞ
Selectivity is defined as: α ¼ Cp=
Cb=ð1CbÞ , where Cp and
aiming at recovering organic compounds from water. Cb the concentrations by mass fraction on the permeate
Poly(dimethylsiloxane; PDMS) is the main reference and bulk phases, respectively.
material for ethanol separation from water, and several Based on the presented literature survey, it is clear
companies have been manufacturing thin film compo- that MD shows much higher fluxes than pervaporation.
site PDMS membranes over the years.[31] Reported values of MD total flux were in the range of

Table 1. Literature survey of ethanol separation processes by MD.

Experimental condition Total flux (kg/m2 h) Selectivity (α) Notes (membrane material, area, pore size, pressure) Reference
DCMD: 37°C (ΔT = 19°C) 6.0 1.6 PTFE, NA, 0.45 μm; NA [14]
DCMD: 36°C (ΔT = 16°C) 0.94 4.6 PP, 490 cm2, NA [13]
DCMD: 37°C (ΔT = 17°C) 0.67 6.6 PP, 370 cm2, 0.2 μm, NA [33]
DCMD: 37°C (ΔT = 17°C) 0.78 2.9 PP, 183 cm2, 0.2 μm, NA [34]
DCMD: 37°C (ΔT = 17°C) 2.80 5.5 PP, 190 cm2, NA [29]
DCMD: 32°C (ΔT = 17°C) 0.64 4.3 PP, 366 cm2, 0.2 μm, NA [35]
VMD: T = 30°C 5.7 5.4 PTFE, 200 cm2, NA, 5 mbar [24]
VMD: T = 35°C 1.8 5.3 PTFE, NA, 0.2 μm, 66 mbar [27]
VMD: T = 30°C 1.2 5.6 PP, 0.5 m2, 0.02 μm, NA [36]
VMD: T = 45°C 2.9 7.7 PTFE, 20 cm2, 0.22 μm, 80 mbar [25]
VMD: T = 50°°C 3.6 5.4 PP, 0.1 m2, 0.2 μm, 50 mbar [37]
VMD: T = 23°C 0.30 4.4 PP, 389 cm2, 0.2 μm, 40 mbar [38]
VMD: T = 23°C 0.51 2.9 PTFE, 266 cm2, 0.2 μm, NA [39]
VMD: T = 30°C 17.1 6.5 PVDF, 37.05 cm2, 0.2 μm, NA [26]
VMD: T = 32°C 0.17 5.9 PP, 400 cm2, 0.46 μm, NA [40]
SGMD: T = 30°C 0.23 6.7 PTFE, 189 cm2, 2 μm, NA [41]
AGMD: 50°C (ΔT = 30°C) 2.2 3.0 PTFE, 160 cm2, 0.45 μm, NA [30]
AGMD: 35°C (ΔT = 15°C) 4.3 3.2 PTFE, 259.6 cm2, 0.2 μm, NA [42]
AGMD: 41°C (ΔT = 35°C) 2.9 1.6 PTFE, 13.9 cm2, 0.45 μm, NA [43]
DCMD: direct contact membrane distillation; VMD: vacuum membrane distillation SGMD: sweep gas membrane distillation; AGMD: air gap membrane
distillation NA: data not available; PTFE: polytetrafluoroethylene; PP: polypropylene.
 SEPARATION SCIENCE AND TECHNOLOGY 5

Table 2. Literature survey of ethanol separation processes by pervaporation.

Experimental condition Total flux (kg/m2 h) Selectivity (α) Notes (membrane material, area, pressure) Reference
T = 30°C 0.02 10.8 PDMS, NA, 1 mbar [44]
T = 30°C 0.06 8.1 PDMS, 200 cm2, 5 mbar [24]
T = 30°C 0.39 6.3 PDMS, 0.17 m2, 27 mbar [45]
T = 22°C 0.34 10.0 PDMS with silicalite, NA, 3 mbar [46]
T = 35°C 0.05 10.5 PDMS with zeolites, NA, 2 mbar [47]
T = 35°C 0.37 4.1 PDMS, 0.1 m2, 5 mbar [9]
T = 25°C 0.15 9.9 PTMSP, 55 cm2, 4 mbar [48]
T = 50°C 0.13 13.4 PDMS with zeolites, 28 cm2 0.3 mbar [49]
T = 34°C 0.15 10.3 PDMS, 0.22 m2, 5 mbar [50]
T = 35°C 0.14 9.0 PDMS with zeolites, NA, 3 mbar [51]
T = 32°C 0.01 9.4 PDMS, 0.21 m2, NA [40]
T = 25°C 0.02 10.0 MFI with silicalite, 5.2 cm2, 3 mbar [52]
T = 35°C 0.09 8.2 PDMS, 0.22 m2, 5 mbar [53]
T = 40°C 0.25 10.0 PDMS, 100 cm2, 13 mbar [54]
T = 65°C 1.6 7.8 PDMS, 170 cm2, 1 mbar [55]
T = 28°C 0.39 6.4 PDMS in PA support, 0.08 m2, 50 mbar [56]
T = 30°C 0.55 8.4 PDMS, 20 cm2, 3 mbar [11]
T = 43°C 0.52 9.1 POMS, 170 cm2, 1 mbar [57]
T = 41°C 0.17 5.5 PDMS, 8.8 cm2, 1 mbar [58]
T = 40°C 0.50 8.3 PDMS, 22.4 cm2, 10 mbar [59]
T = 40°C 0.23 6.8 PDMS/PEI, 0.9 m2, 7 mbar [60]
T = 27°C 0.004 10.6 PDMS, 400 cm2, 7 mbar [61]
T = 40°C 0.10 14.9 PDMS with silicalite, 15.9 cm2, 1 mbar [62]
T = 35°C 0.40 8.5 PDMS, 0.16 m2, 46 mbar [63]
T = 35°C 0.21 7.2 PDMS/PEI, 19.6 cm2, 40 mbar [64]
NA: data not available; PTMSP: poly[(1-trimethylsilyl)-l-propyne]; POMS: polyoctylmethyl siloxane; PA: polyamide; PEI: polyethyleneimine.

0.17–17.1 kg/m2 h with 2.9 kg/m2 h in average, ethanol reported for VMD and 5 wt% ethanol–water mixtures at
flux were between 0.04 and 1.27 kg/m2 h with 0.34 kg/ 35°C, separation factor between 5.3 and 8.8.
m2 h in average and ethanol-to-water selectivity, Kujawska et al.[43] compared the efficiency of
between 1.6 and 7.7, with a mean value of 4.7. AGMD with commercial PDMS membrane in perva-
Whereas in pervaporation, selectivity values were in poration, and for the case of ethanol–water, mixture it
the range between 4.1 and 14.9, with 8.9 in average, was found that AGMD seems to be more suitable
and much lower fluxes were observed in comparison to mainly due to higher flux than pervaporation, almost
the MD, varying from 0.004 to 1.6 kg/m2 h, with 5 times higher. The authors also highlighted that such
0.27 kg/m2 h in average, and an ethanol flux from comparison does not allow discussing overall processes’
0.001 to 0.31 kg/m2 h, with 0.07 kg/m2 h in average. efficiency in organic removal from water.
Since pervaporation is based on diffusive process In fact, MD process exhibited higher total and etha-
mechanism with dense membranes, the lower flux nol fluxes than pervaporation, but more than that, it
values when compared to the MD were expected. But showed a comparable performance to pervaporation
it is important to highlight that although pervaporation with polymeric selective membranes, even in terms of
shows more results with higher selectivities than MD, overall ethanol–water selectivities in some cases. There
the highest values were close. Relying on the influence is a cloudy region with overlapping results between MD
of polymeric material in ethanol-to-water selectivity, it and pervaporation. It was not possible to clearly point
was expected that pervaporation data presented higher out the best method for ethanol removal and make a
values of selectivity. fair comparison because of the diverse conditions in
According to Han et al.,[59] using a series of composite which each experiment was performed.
silicone rubber membranes to investigate pervaporation Based on these data, it can be partially concluded
of 5 wt% ethanol–water mixtures, the experimental that there is a difficulty in this comparison, among
separation factors ranged from 7.9 to 9.4. While Garcia other factors, especially due to different hydrodynamic
et al.[57] in their pervaporation study evaluated different conditions. It should be noted that pervaporation selec-
ethanol feed compositions from 1 to 11 wt%, and the tivity values were not as high as expected. Thus, a
experimental separation factors ranged from 7.2 to 10.7 critical analysis was performed over these collected
at three different feed temperatures (30°C, 43°C and data from literature review, comparing pervaporation
53°C). Izquierdo-Gil and Jonsson[26] investigated the with MD and calculating the effects of concentration
influence of concentration polarization on flux and etha- polarization on different MD settings, assessing the
nol selectivity using VMD and obtained a separation effect of hydrodynamic conditions and polarization
factor in a range of 5.2–7.9. While Bandini et al.[27] phenomena in each case.
6 B. A. CINELLI ET AL.

Materials and methods reported in the literature for each mixture of ethanol
and water.
Membranes and modules
The thermodynamic relationship between a binary
Both pervaporation and MD systems consisted in a sub- liquid mixture can be represented by the modified
merged membrane module with PDMS (MedicOne®, Raoult’s law,[65] which includes the activity coeffi-
Brazil) and PVDF (homemade) hollow fibers, respec- cient, taking into account nonidealities in the liquid
tively. The PDMS dense hollow fibers have internal and phase:
external diameters of 0.59 and 0.99 mm, respectively, and
PVDF microporous hydrophobic hollow fibers have yi P ¼ Pi sat xi γi (2)
internal and external diameters of 0.54 and 0.92 mm, where yi and xi are the vapor and liquid mole fractions
respectively. Both modules were constructed to ensure of component i, respectively, P is the total vapor pres-
the same permeation area of 200 cm2, in order to keep sure and γi is the activity coefficient of component i in
similar overall dimensions, maintaining same flask solution. While Pisat is the saturation pressure of pure
hydrodynamics. component i which can be estimated by Antoine equa-
tion. In this work, activity coefficients (γ) were esti-
mated by Margules activity model:[66,67]
Separation experiments
ln γ1 ¼ ½A12 þ 2ðA21  A12 Þx1 x2 2
Separation experiments were carried out using syn- (3)
thetic solutions of ethanol and water, with two different lnγ2 ¼ ½A21 þ 2ðA12  A21 Þx2 x2 1
concentrations, 5 and 10 wt% in 0.3-L flasks.
where A12 and A21 are binary interaction Margules
The experiments were conducted at different tem-
parameters for the system ethanol[1] and water.[2]
peratures, 25°C, 30°C, 35°C, 40°C and 50°C, with con-
The Antoine equation, used to estimate the satura-
tinuous stirring using a magnetic stirrer in order to
tion vapor pressure, is given by
maintain a minimum homogeneity of the feed. The
vacuum applied to the permeate side was provided by Bi
log Pi sat ¼ Ai  (4)
a vacuum pump at 10 mbar. T þ Ci
where T is the temperature expressed in degree Celsius
Analytical methods and Ai, Bi and Ci are constants based on Poling et al.[65]
The mass transfer through the boundary layer plays
Ethanol was analyzed by high-performance liquid chro- an important role in the performance of an MD system.
matography (HPLC; Agilent Technologies, USA), using The polarization boundary layer may increase the over-
a HPX-87H column (Bio-Rad Laboratories, USA). The all resistance to mass transfer. The concentration dif-
mobile phase was 0.005 M sulfuric acid in ultrapure ference between the membrane surface and bulk,
water, injected at a flow rate of 0.6 mL/min, and oven caused by concentration polarization, can be described
temperature was 65°C. by the boundary layer film model:[27,68]
 
C0  Cp
J ¼ kf ln (5)
Theory for thermodynamic equilibrium calculation Cb  Cp
The effect of concentration polarization in different where J is the total mass flux; C0, Cb and Cp are the
MD settings was evaluated, comparing reported concentrations by mass fraction on the membrane
selectivities to the ones given by vapor–liquid equili- interface, the feed bulk phase and permeate, respec-
brium. For this purpose, it was considered that these tively; and kf is the boundary layer mass transfer coeffi-
differences in selectivity were only due to external cient that can be written as follows:
concentration polarization in the feed side. The effect
Di
of temperature polarization was neglected; although kf ¼ (6)
it may be expressive, it is difficult to estimate without δ
having full knowledge of the system in each case. where Di is the diffusion coefficient of permeating
Besides, no internal concentration polarization was species that can be predicted from the Wilke–Chang
considered since the pores are filled with vapor and correlation[69] adjusted by actual viscosity of mixture
thus less susceptible to this phenomenon. The follow- according to the equation proposed by Khattab et al.[70]
ing models and equations were used to calculate and δ is the thickness of the concentration polarization
thermodynamic equilibrium in the conditions layer. For all experimental data, the following
 SEPARATION SCIENCE AND TECHNOLOGY 7

Sherwood correlation was used to estimate the bound- Comparative analysis of MD and pervaporation
ary layer mass transfer coefficients of submerged VMD:
Compiled data analysis
 1=3
δv0 Based on the data collected in the literature review for
Sh ¼ 1:4 (7) ethanol removal by MD and pervaporation, Fig. 3 pre-
Di
sents a graphic containing ethanol flux and selectivity
where Sh is the Sherwood number and v0 is the super- for both processes. The gray region represents the
ficial velocity.[71] range of theoretical selectivity (αVLE) for all reported
If the information of xi (Cb) is provided in all revised conditions, which is between 8 and 12.
works, then the concentration of ethanol in the vapor The lower fluxes observed for pervaporation tests, in
phase in thermodynamic equilibrium (yiVLE ) can be comparison to MD, were expected due to higher trans-
calculated. Assuming that this concentration in the port resistance caused by dense membranes. However,
vapor phase will be the same concentration in the since the nature of polymeric material that takes a role
permeate, then the theoretical selectivity (αVLE), dic- in selectivity favoring ethanol permeation, higher selec-
tated by the thermodynamic vapor–liquid equilibrium, tivity values were expected. It is interesting to note that
can be calculated. For all cases, this difference (α/αVLE) most of the pervaporation results are in the range or
and yi, the ethanol concentration in the theoretical below the thermodynamic limit illustrated by the gray
vapor phase, were calculated. Neglecting the resistance area. For those above, it is important to mention that
in the permeate side and with the consideration that they present very low flux[44,61] or they work with
this difference in selectivity was only due to the con- mixed matrix membranes with the incorporation of
centration polarization in the feed, using experimental silicalite.[46,49,62]
data of Cp, the “actual” concentration at the interface of It is important to stress that the comparison was
the membrane pores can be calculated, thus estimating carried out using data obtained under similar condi-
the C0 value. tions: ethanol–water solutions or ethanol removal from
The decrease in ethanol concentration at membrane fermentation broths, considering approximately 5 wt%
surface C0, compared to bulk concentration Cb, deter- ethanol in the feed, and temperature range of 22–65°C,
mines the extent of concentration polarization, and the with an average of 35°C.
ratio of C0/Cb is called the concentration polarization However, several other factors must be taken into
modulus and is a useful measure of the extent of con- account before pointing MD as the most appropriate
centration polarization.[68] solution for this application. The MD, considering the
Therefore, with all the data collected and solving Eq. hydrophobic nature of microporous membranes
(3)–(7), the thickness of the boundary layer and con- applied, could present higher tendency to fouling pro-
centration polarization modulus can be estimated. blems than pervaporation, especially when using micro-
organisms (biofouling). The possibility of pore wetting

16

14

12
Selectivity (α)

10

0
0.001 0.01 0.1 1 10 100
Total flux (kg/m2.h)

Figure 3. Selectivity and total flux results for ethanol–water separation using pervaporation (○) and MD (♦).
8 B. A. CINELLI ET AL.

and liquid permeation with disruption of the vapor– 2020 μm can be observed, while concentration polariza-
liquid interface should also be considered, according to tion modulus varied from 0.15 to 0.82.
feed characteristics. Also, exactly due to higher flux The hydrodynamic conditions have a very important
values, polarization effects may be more pronounced. role in MD process efficiency. The effect of stirring in
While it is important to note that the highest selectivity the liquid phase, for example, with a bad agitation,
of pervaporation also accentuates the effects of concen- results in water accumulating close to membrane sur-
tration polarization, in addition to the effects caused by face, consequently less ethanol in the interface, and this
the hydrodynamic conditions, the lower fluxes could would be the actual concentration for thermodynamic
somehow hide this effect. equilibrium, resulting in selectivity loss.
Gryta et al.[13] working with a continuous removal of
ethanol from fermentation media in a tank of 5.5 L
The vapor–liquid equilibrium calculation for MD
with DCMD obtained 16.8 wt% of ethanol in the
The goal of the vapor–liquid equilibrium calculations permeate from 4.2 wt% in the feed at 36°C, and
for MD is to identify qualitatively and quantitatively 30.3% was the maximum expected which represents a
how close the results reported were to its theoretical maximum selectivity of 9.9 in comparison with 4.6
maximum, assessing the effect of hydrodynamic condi- achieved; for this work, a concentration polarization
tions and hence the concentration polarization in each layer thickness of 690 μm and concentration polariza-
case. These calculations were used to determine the tion modulus of 0.48 were calculated.
boundary layer thickness for all literature data. García-Payo et al.[42] using an AGMD apparatus in
The results presented in Table 3 summarize all the crossflow mode resulted in 24.0% ethanol in the
reported data from the most relevant MD works, show- permeate when 44.1% maximum was expected, result-
ing total flux, estimated vapor composition in vapor– ing in a 262 μm concentration polarization layer thick-
liquid equilibrium, the thickness of the concentration ness and 0.42 of polarization modulus. Recently, Zhang
polarization layer and polarization modulus. et al.[39] used VMD with PTFE module in a 0.5-L
The selectivity values obtained experimentally were reactor and obtained 21.5% when 44.0% was expected
lower than the theoretical maximum for all cases, in theory, which represents 34.8% of the theoretical
equivalent to one theoretical distillation stage, mainly selectivity, thus a concentration polarization layer
due to additional mass transfer resistances. Among 18 thickness of 1943 μm was calculated.
evaluated values, 16 experimental selectivity results were
considerably lower than thermodynamic equilibrium,
from 14.8% to 63.3% of the theoretical maximum and Experimental results and discussion
only two cases above 70%. Therefore, in all 18 MD,
MD
works showed polarization phenomenon effect. The con-
centration polarization layer thickness values were cal- Ethanol–water VMD experiments
culated and expressed in micrometers, in Table 3. Table 4 shows the results obtained using the PVDF
Among the cases studied, values ranging from 22.9 to microporous hollow fiber membranes in the MD tests.

Table 3. Comparison of the thermodynamic equilibrium with actual selectivities assessed and thickness of the polarization layer
calculations for the MD literature data.
Reference Temperature (°C) Cb (%) Cp (%) C0 (%) yi ðVLEÞ (%) α αVLE δ (μm) Concentration polarization modulus
[24]
30 5.0 22.2 2.87 33.7 5.4 9.6 71.5 0.57
[41]
30 6.0 30.0 4.4 37.0 6.7 9.2 896 0.74
[14]
37 2.3 3.71 0.36 20.6 1.6 10.8 742 0.15
[27]
35 5.0 21.8 2.8 33.5 5.3 9.6 155 0.57
[36]
30 4.0 18.8 2.2 29.6 5.6 9.4 340 0.56
[30]
50 5.0 13.6 1.7 33.0 3.0 9.4 826 0.33
[13]
36 4.2 16.8 2.0 30.3 4.6 9.9 690 0.48
[42]
35 9.0 24.0 3.8 44.1 3.2 8.0 262 0.42
[26]
30 0.3 1.6 0.13 2.9 6.5 12.0 22.9 0.54
[38]
23 3.0 12.0 1.3 24.9 4.4 10.7 1972 0.42
[33]
37 3.0 17.0 1.9 24.4 6.6 10.5 519 0.64
[29]
50 2.8 13.7 1.5 23.3 5.5 10.6 192 0.53
[34]
37 3.7 10.1 1.1 27.9 2.9 10.1 2020 0.30
[25]
45 5.0 28.7 4.1 33.1 7.7 9.4 54.2 0.82
[37]
50 5.0 22.1 3.0 33.0 5.4 9.3 167 0.59
[43]
41 2.7 4.2 0.4 22.0 1.6 10.2 1280 0.16
[35]
37 1.4 5.9 0.6 12.5 4.3 9.9 1163 0.44
[39]
23 8.7 21.5 3.20 44.0 2.9 8.0 1943 0.37
 SEPARATION SCIENCE AND TECHNOLOGY 9

Table 4. Experimental VMD results with synthetic solution with 5% and 10% of ethanol at different feed temperatures.
Temperature (°C) Ethanol in the permeate (%) Selectivity Total flux (kg/m2 h) Ethanol flux (kg/m2 h)
5% Ethanol 25 18.7 4.5 0.3073 0.0575
30 18.9 4.7 0.5606 0.1060
35 17.8 5.0 0.7984 0.1419
40 17.9 4.9 0.9917 0.1775
50 17.9 4.6 1.4620 0.2613
10% Ethanol 25 26.2 3.2 0.2236 0.0586
30 31.4 4.1 0.2653 0.0834
35 32.2 4.3 0.3459 0.1114
40 29.9 3.9 0.5309 0.1586
50 23.8 3.1 1.2349 0.2938

The results show that the MD system with submerged This behavior was expected according to the Antoine
PVDF hollow fibers was able to selectively remove etha- equation, which represents the dependence of vapor pres-
nol from the synthetic mixture. Ethanol, initially with sure with temperature—the water and ethanol vapor pres-
5% in the feed, was collected in the permeate stream sures increase exponentially with temperature (Fig. 4).
with 17.8% and 18.9%, resulting in a selectivity value
varying from 4.5 to 5.0, while total flux varied between Evaluation of concentration polarization
0.31 and 1.46 kg/m2 h. For 10% of ethanol in the feed, a Similar to the literature compiled data, an analysis of
permeate of 23.8–32.2%, selectivity values between 3.1 experimental results was carried out to identify effects
and 4.3, total flux up to 1.2 kg/m2 h and 0.29 kg/m2 h of of concentration polarization phenomenon. Table 5
ethanol flux at 50°C were obtained. Furthermore, the shows the results obtained by MD comparing reported
temperature increase resulted in no significant increase actual and theoretical selectivities and calculations of
in selectivity. Gryta et al.[13] investigated for a range of the thickness of the polarization layer.
ethanol concentration in fermentation medium and con- It can be observed that all selectivity values experimen-
cluded that the selectivity depends on feed temperature, tally obtained in this work were lower than the theoretical
whereas increasing from 30°C to 60°C, the enrichment values, between 40.3% and 55.8% of the theoretical max-
factor varied from 4 to 6. Banat et al.[30] reported that imum, with a mean of 48.5%. The calculated values of
with increasing feed temperature from 40°C to 60°C, the thickness of the concentration polarization layer were
selectivity increased from 2.5 to 3. Also, according to the between 380.5 and 3223.7 μm and concentration polar-
authors, the selectivity increases with a rise in tempera- ization modulus varying from 0.43 to 0.58. The calculated
ture until 60°C at different ethanol concentrations values are within the range to the others in the literature.
although at 70°C, the negative effect of flux increase on Tomaszewska and Białończyk[34] obtained similar results
concentration polarization, a decrease in selectivity was at 37°C, where 2020 μm of polarization layer and 0.30 of
observed. Additionally, at higher feed temperature, the polarization modulus were calculated, while for Calibo
viscosity is lower, and there will be a higher ethanol et al.,[41] a polarization modulus of 0.74 and a polarization
diffusion rate, but the increase in the ethanol flux layer of 896 μm were estimated. A similar result was also
through the membrane increases the effect concentration reported by Diban et al.[38] for VMD experiments at 23°C,
polarization, which may lead to a loss in selectivity. where a polarization layer of 1972 μm and 0.42 of polar-
Regarding the effect of feed temperature on permeate ization modulus were estimated.
flux, it may be noted that the increase in temperature also It can also be observed that the thickness of the polar-
leads to an increase in both total and ethanol fluxes ization layer decreases as feed temperature increases, as
through the membrane. According to El-Bourawi et al.,[15] expected, due to the reduction in liquid viscosity caused
the effect of feed temperature on the permeate flux has by the temperature raising, which improves fluidity of
been widely investigated in different MD settings. feed, and enhanced turbulent movement, which decreases
Evaluating different MD settings, with temperatures ran- the mass transfer resistance in the boundary layer of the
ging from 20°C to 80°C and keeping all other parameters feed side. This behavior can be observed in Fig. 5.
constant, increasing the temperature enhanced the Nevertheless, when ethanol concentration in the feed
permeate fluxes. Rom et al.[37] reported that higher tem- increased from 5 to 10 wt%, the thickness of the polariza-
peratures result in higher flow rates, and higher ethanol tion layer increased to all evaluated feed temperatures.
feed concentrations also result in higher ethanol fluxes Figure 6 shows schematic diagrams of the concentration
through the membrane. According to the authors, for profiles in the pores of PVDF membranes in VMD tests
feed temperatures of 20°C, 35°C and 50°C, flux values with ethanol solution of concentration of 5 and 10 wt% at
were approximately 100, 500 and 700 g/m2 h, respectively. 30°C.
10 B. A. CINELLI ET AL.

0.5

0.5 Water Ethanol

0.4

0.4

Vapor pressure (bar)

0.3

0.3

0.2

0.2

0.1

0.1

0.0
0 10 20 30 40 50 60 70
Feed temperature (°C)

Figure 4. Ethanol and water vapor pressures as a function of feed temperature.

Table 5. Comparison of experimental results obtained in this work and thermodynamic equilibrium and calculation of the
polarization layer thickness.
Temperature (°C) Cb (%) Cp (%) C0 (%) y1 ðVLEÞ(%) α αVLE Concentration polarization layer (μm) Concentration polarization modulus
25 4.9 18.7 2.3 33.4 4.5 9.8 1181.3 0.47
30 4.7 18.9 2.3 32.5 4.7 9.8 628.4 0.50
35 4.1 17.8 2.1 30.1 5.0 9.9 429.5 0.51
40 4.2 17.9 2.2 30.3 4.9 9.8 396.5 0.51
50 4.5 17.9 2.2 31.1 4.6 9.6 380.5 0.49
25 10.0 26.2 4.4 46.4 3.2 7.8 3223.7 0.44
30 10.0 31.4 5.6 46.1 4.1 7.7 1477.3 0.56
35 10.0 32.2 5.8 46.0 4.3 7.6 1126.5 0.58
40 9.9 29.9 5.3 45.6 3.9 7.6 1094.7 0.53
50 9.0 23.8 3.9 43.7 3.1 7.8 1023.5 0.43

3500 1.4
Polarization layer thickness (µm)

3000 1.2

2500 1.0
Viscosity (10−3 Pa·s)

2000 0.8

1500 0.6

1000 0.4

500 0.2

0 0.0
20 25 30 35 40 45 50 55
Temperature (°C)

Figure 5. Effect of feed temperature in the polarization layer thickness: 5 wt% (♦) ethanol and 10 wt% (●) ethanol in the feed
solution. Solid line is feed solution viscosity with 5% ethanol, and dashed line is the viscosity with 10% ethanol in the feed solution.

The concentration profiles shown in Fig. 6 illustrate polarization layer increases when feed ethanol concen-
that ethanol concentration in the feed boundary layer tration increases. At higher ethanol concentration,
region decreases in similar shape for 5% and 10%. there will be different hydrodynamic conditions due
Although the thickness of the concentration to higher viscosity and consequently lower diffusion
 SEPARATION SCIENCE AND TECHNOLOGY 11

(a) Porous membrane (b) Porous membrane

Boudary layer Cp = 31.4
Cp = 18.9
δ = 1477 µm
Boudary layer

δ = 628 µm

Concentration (wt %)
Concentration (wt%) Cb = 10.0

2
J = 0.56 kg/m .h J = 0.27 kg/m2.h

Cb = 4.7

C0 = 5.6

Bulk phase Vapour phase Bulk phase Vapour phase

(Feed side) (Permeate side) (Feed side) (Permeate side)
C0 = 2.3

Figure 6. Schematic diagrams of concentration profiles at 30°C for (a) 5 wt% and (b) 10 wt% feed ethanol concentrations.

coefficient and thus increases the concentration polar- between 43.3% and 50.7%, which results in a selectivity
ization effects. ranging from 6.9 and 9.3. The total flux varied between
According to García-Payo et al.,[42] concentration 13.0 and 40.5 g/m2 H, and ethanol flux was also the
polarization is dependent on membrane material, its lowest at 25°C (6.4 g/m2 h) with a maximum value at
roughness and its pore size distribution and feed tem- 50°C (17.6 g/m2 h).
perature. They also observed that thickness of the bound- It is observed that the increase in temperature resulted
ary layer slightly increases from 122 to 138 μm as ethanol in a significant increase in total and ethanol fluxes. As
concentration in the feed increases from 6 to 18 wt%. stated in other studies,[58,72] the temperature increase
enhanced the total permeate flux due to increased mole-
cular diffusion rate that permeates the free volume pro-
Pervaporation
duced due to the increase in segmental mobility of
Ethanol–water pervaporation experiments polymeric chains. Bello et al.[61] observed that both total
Table 6 shows the pervaporation results obtained using and ethanol fluxes increased when increasing feed tem-
PDMS hollow fiber membranes for synthetic solutions perature from 22°C to 30°C. Total flux increases from 3.5
containing 5% and 10% of ethanol. to 8 g/m2 h with an increase in ethanol flux by a factor of
The performance of pervaporation system obtained 1.5. Dobrak et al.[58] have also shown that increase in
in experimental tests with PDMS hollow fiber mem- temperature caused a significant increase in total perme-
brane showed relatively low fluxes and good selectivity ate flux, when increasing temperature from 41°C to 51°C
values, comparable to published pervaporation perfor- and the ethanol flux up from 20 to 35 g/m2 h.
mance results with PDMS membrane. Moreover, with higher temperature, a slight decrease
For 5% ethanol in the feed, permeate concentrations in selectivity was observed, and even with some oscilla-
were between 31.4% and 41.2%, which represents selec- tory behavior, showing an antagonistic effect of tem-
tivity values from 8.7 to 13.3. The total flux varied perature, in general, a downward trend in selectivity
between 11.2 and 24.3 g/m2 H, and ethanol flux was may be noted. Actually, different effects were related in
the lowest for the feed temperature of 25°C (4.6 g/m2.h) the literature, with some studies demonstrating that the
and the highest at 50°C (8.3 g/m2 h). At 10% ethanol selectivity increases when temperature decrease and
feed, concentrations in the permeate stream varied others showing the opposite effect.[47,61]

Table 6. Experimental pervaporation results with synthetic solution with 5% and 10% of ethanol at different feed temperatures.
Temperature (°C) Ethanol in the permeate (%) Selectivity Total flux (kg/m2 h) Ethanol flux (kg/m2 h)
5% Ethanol 25 41.2 13.3 0.0112 0.0046
30 40.4 12.9 0.0123 0.0050
35 35.9 10.6 0.0159 0.0057
40 31.4 8.7 0.0181 0.0057
50 34.2 9.9 0.0243 0.0083
10% Ethanol 25 49.4 8.8 0.0130 0.0064
30 49.9 9.0 0.0162 0.0081
35 50.7 9.3 0.0230 0.0117
40 46.8 8.0 0.0287 0.0134
50 43.3 6.9 0.0405 0.0175
12 B. A. CINELLI ET AL.

The effect of temperature on the selectivity in perva- way, considering the same value of MD polarization
poration seems to be dependent on the membrane layer for pervaporation when submitted to the same
material and the operating conditions, thus varying process conditions, it was possible to calculate the con-
greatly between each case. According to Lee et al.,[60] centration at membrane interface (C0). Also, according
the ethanol selectivity decreased to 28% when the tem- to Baker,[68] the separation achieved in pervaporation is
perature increased from 20°C to 60°C. While Li et al.[72] equivalent to the product of the separation achieved by
and Bello et al.[61] reported that ethanol selectivity evaporation of the liquid (βevap) and the separation
increased slightly with increasing temperature. achieved by selective membrane (βmem):
According to Dobrak et al.,[58] it was reported that
α ¼ βpervap ¼ βevap  βmem (9)
the increase in temperature was possible to promote
phenomena of membrane swelling. This conclusion is Therefore, since α was obtained experimentally and βevap
in agreement with the observations made by Vane was calculated for each condition (same value of the
et al.[73] for pervaporation processes for ethanol theoretical selectivity αVLE), it was possible to estimate
removal with PDMS membranes at elevated tempera- βmem. Table 7 shows pervaporation experimental results
tures (30–70°C). According to the authors, higher tem- comparing the selectivity and βevap and estimated βmem.
peratures may increase total flux, but at the same time It can be observed that the selectivity of membrane
cannot guarantee increase in selectivity. Liang and decreases with the increase in temperature, whereas βevap
Ruckenstein[74] observed this effect and explained does not change significantly but decreases significantly
through temperature dependence on the interactions from 5% to 10% of ethanol in the feed, what it was expected
between water molecules and ethanol, and it turns out considering the thermodynamic equilibrium. Furthermore,
that less water is stimulated to permeate through the with higher temperature, greater the mobility of polymeric
membrane, which is swollen by ethanol. If membrane chains, greater will be flux for both ethanol and water and,
swelling is dominant in transport through pervapora- consequently, lower selectivity. The value of the βmem varied
tion membrane, a decrease in selectivity with increasing from 1.46 to 0.94. PDMS membrane selectivity factor
temperature will be expected, which will depend on the decreases with increase in temperature, which was
characteristics of the membrane polymer. Also, in this expected, since at higher temperatures increases the mobi-
work, for a mixture containing 5% ethanol, the selec- lity of the polymeric chains and improves the diffusivity of
tivity decreased to 26% with the temperature rising ethanol and water molecules, thus enhances flux. Because
from 25°C to 50°C, while at 10% ethanol, the selectivity the radius of water molecule (0.37 nm) is smaller than that
decreased to 22% for the same range of temperature. of ethanol molecule (0.52 nm),[67] the enhancement of
water flux is larger than that of ethanol flux, therefore
Determination of concentration polarization and decreasing the selectivity.
membrane selectivity
It is essential in order to allow a fair comparison between
different processes, to maintain exactly the same test Comparison of the efficiency of VMD and
setup conditions. In this way, it was possible to compare pervaporation processes
different results with MD and pervaporation, eliminating Based on all experiments carried out under same opera-
the effects of flow and hydrodynamics conditions, better tional and hydrodynamic conditions, it is possible to
evaluating the real potential for each process, both for compare the different results with MD and pervapora-
literature data and for results generated in this work. It is tion, better evaluating the real potential of each process.
also important to highlight that the choice for poor
hydrodynamic conditions, that is, bad agitation, for
Table 7. Pervaporation experimental results obtained in this
both MD and pervaporation experiments was intentional
work and thermodynamic equilibrium and calculation of the
to facilitate the identification of the effects of tempera- membrane selectivity.
ture and concentration. In ideal conditions of perfect Temperature (° Cb Cp C0 yi ðVLEÞ
agitation, none of these effects of polarization could be C) (%) (%) (%) (%) α βevap βmem
seen. This practice was done to emphasize the impor- 25 5.0 41.2 4.3 29.1 13.3 9.1 1.46
30 5.0 40.4 4.6 30.7 12.8 9.3 1.39
tance of looking at literature data very carefully. 35 5.0 35.9 4.7 31.3 10.6 9.3 1.14
According to Baker[68] approach, the hydrodynamic 40 5.0 31.4 4.7 31.6 8.7 9.3 0.94
50 5.0 34.2 4.7 31.0 9.9 9.1 1.08
conditions of the medium, such as turbulence, influ- 25 10.0 49.4 8.1 37.6 8.8 6.9 1.29
ence the thickness of the polarization layer, while the 30 10.0 49.9 8.8 40.5 9.0 7.1 1.27
35 10.0 50.7 8.8 40.4 9.3 7.0 1.32
flux and the effect of membrane influence the polariza- 40 10.0 46.8 8.8 40.4 8.0 7.0 1.13
tion module, that is, consequence changes in C0. In this 50 10.0 43.3 8.9 40.3 6.9 7.0 0.99
 SEPARATION SCIENCE AND TECHNOLOGY 13

The two main response parameters for measuring Despite presenting a lower result in terms of selectivity,
efficiency of each separation process are selectivity and MD has a much lower resistance to transport through the
flux. Figure 7 shows the comparison of ethanol con- membrane compared to pervaporation allowing much
centration in the permeate obtained by pervaporation higher flux, and all VMD experimental results presented
and MD and its comparison with vapor–liquid equili- in this work, as well as all literature data, corroborate this
brium curve for T = 30°C and T = 50°C. fact. The flux obtained with MD was 15 times higher for
It can be noted that ethanol concentration in the the case of 35°C and 10% ethanol and up to 60 times
permeate by pervaporation in all conditions were higher higher for the case of 50°C and 5% of ethanol. In addition,
than in MD, as expected. There are just two cases as expected, according to literature review, due to the
observed in pervaporation: the ethanol composition in nature of each of the processes, pervaporation results
the permeate was below vapor–liquid equilibrium, and enable higher selectivities than in MD. It can be observed
for other results, the experimental selectivity was higher that selectivity in pervaporation was 1.8 times higher in
than theoretical selectivity, while in MD, all results were case of 35°C and 5% ethanol and up to 3 times higher in
below vapor–liquid equilibrium, by 48.5% of this max- case of 25°C and 5% when compared to MD tests under
imum. Moreover, it is important to add that the selec- experimental conditions.
tivity is very sensitive with composition of ethanol in the The effect of feed temperature was positive for both
feed. The lower the ethanol fraction in the feed, the MD and pervaporation but presented different beha-
higher the ethanol composition in the vapor phase and, viors. When temperature increased from 25°C to 50°C,
consequently, higher the αVLE. As discussed previously, the total flux increased almost 5 times for MD, while
ethanol and water form a nonideal mixture, for feed for the same range, flux in pervaporation increased
solution with low concentration of ethanol (<10 wt%), around 2 times. The variation in concentration of etha-
the curve is very distant from the diagonal (y = x), so the nol in the feed was positive for pervaporation with an
relative volatility, or selectivity, between ethanol and expressive increase in flux, whereas for MD, it caused a
water is high in this range of concentration. For mix- slight decrease in flux. As previously mentioned, the
tures with higher ethanol concentrations, relative volati- increase in temperature causes loss in the selectivity of
lity decreases, clarifying the loss in selectivity when pervaporation, while for MD, it is almost constant.
ethanol feed concentration increases. Thus, while increasing ethanol concentration is more

Figure 7. Ethanol composition in the vapor phase (yeVLE ) as a function of the ethanol composition in the feed, for T = 30°C and
T = 50°C. The curve x = y just as a reference. In (a) the highlighted area of the chart.
14 B. A. CINELLI ET AL.

beneficial for pervaporation, the increase in tempera- (66%) lower than 2 kg/m2 h with a PSI of 1.9 kg/m2 h
ture is more positive for MD. in average for 26 works. The PSI values in VMD are
In order to compare the separation effectiveness of significantly higher than in pervaporation at tested feed
different membrane possesses, Kujawska et al.[43] sug- composition and temperature, which can suggest that
gest the utilization of a parameter that combines total in this case, the VMD with PTFE membranes is a more
flux (J) and selectivity (α) into the so-called process efficient method for ethanol removal. Kujawska et al.[43]
separation index (PSI). also conclude that based on PSI values, the AGMD,
which were significantly higher than in pervaporation,
PSI ¼ J ðα  1Þ (8)
is a more efficient method for ethanol recovery; how-
The PSIs for experimental results of MD were in between ever, they mention that in case of water–butanol mix-
0.5 and 5.3 kg/m2 h, while for pervaporation, they range ture, higher pervaporation presented higher PSI values,
from 0.10 to 0.24 kg/m2 h, as shown in Fig. 8. The gray suggesting that pervaporation process is a more effec-
area corresponds to PSI value higher than 2 kg/m2 h. tive method in butanol recovery.
A comparison of PSI values for ethanol–water system Figure 9 shows the comparison of experimental results
through porous PTFE membranes in VMD and through of total flux and selectivity between MD and pervaporation.
dense PDMS membrane in pervaporation is presented. Comparing results obtained from pervaporation and
For MD just two conditions, the PSI value was lower than MD, it can be said that the well-known premise of high
1 kg/m2 h, and three conditions presented a PSI value of flux for MD and high selectivity for pervaporation is
above 3 kg/m2 h. The best condition for ethanol removal valid and that in fact can be clearly observed according
was obtained by MD at 50°C and 5 wt% of ethanol in the to the results exhibited in Fig. 9.
feed solution which resulted in a PSI value of 5.3 kg/m2 h. In this work, we have two very distinct regions of results
Equivalent results were reported by Tomaszewska and for MD and pervaporation, with high flux against high
Białończyk[35] using DCMD with PP membranes, where selectivity, being fairly compared in terms of performance
a PSI of 2.1 kg/m2 h was obtained in ethanol removal of ethanol removal in same operational conditions.
from fermentation broth. Bandini et al.[30] working with Although it is tough to weight these two responses, the
VMD of ethanol–water mixture reported results equiva- PSI is one alternative to help in the comparison. In that
lent to a PSI of 7.6 kg/m2 h with 5 wt% ethanol in the feed way, it can be observed that all MD experimental results
solution at 35°C, and Diban et al.[38] with VMD obtained presented a PSI value higher than pervaporation, 50% of
a PSI value of 1 kg/m2 h with 3 wt% of ethanol in the feed MD results were above a PSI of 2 kg/m2 h, while pervapora-
solution at 23°C. tion results were far from the line of PSI equal to 2 kg/m2 h.
Most of MD literature (70%) presents results with The results presented evidence for the excellent potential
PSI value higher than 2 kg/m2 h, with a mean value of of both MD and pervaporation processes for ethanol
11 kg/m2 h in 19 works and most of pervaporation removal from fermentation medium. However, the need

10.0
PSI (kg/m2.h)

1.0

0.1
20 25 30 35 40 45 50 55
Temperature (°C)

Figure 8. Effect of feed temperature on process separation index (PSI) in ethanol–water separation for MD with 5% of ethanol in the
feed solution (♦), MD with 10% ethanol (▲), pervaporation with 5% ethanol (○) and pervaporation with 10% ethanol (□).
 SEPARATION SCIENCE AND TECHNOLOGY 15

Figure 9. Experimental results of total flux and selectivity for pervaporation (○) and MD (♦). Gray area represents the vapor–liquid
equilibrium selectivity range. While circled area with dotted line indicates two very distinct regions of MD and pervaporation. Solid
line corresponds to PSI equal to 2 kg/m2 h.

to assess the conditions very carefully, focusing mainly on dependent not only on membrane characteristics but
higher flux, is evident. Even though, with a trade-off analy- extremely on operational and hydrodynamic conditions
sis, it can be said that MD has apparently more advantages of reactor. The feed temperature plays a very important
in comparison to pervaporation process, each has its advan- role over thickness of the polarization layer and flux
tages and disadvantages. The PSI also supports this conclu- although the temperature increase was more positive for
sion. Furthermore, it has been observed that the higher the MD than for pervaporation. Moreover, the increase in
feed temperature, the more advantageous the process ethanol concentration in the feed was positive for perva-
apparently becomes, at least in terms of flux and selectivity poration but had a negative effect in MD performance.
responses; however, some trade-off in terms of increasing We strongly recommend that care should be taken, to
operational costs are expected. ensure that there is no misinterpretation of the experi-
mental data, when comparing MD and pervaporation
processes, especially due to different hydrodynamic tested
Conclusion conditions. Therefore, the question proposed in the title
of this article was discussed and answered: many times,
MD is an emerging technology with excellent potential to pervaporation and MD are not being correctly compared
efficiently recover ethanol and other biofuels from fer- because each published datum was obtained in a different
mentation broths with higher fluxes and consequently way, mainly due of the limitations in terms of mass
lower membrane area required. However, pervaporation transfer of each experimental condition. Therefore, the
is a more mature technology with higher selectivity but direct comparison of the literature data is only illustrative
lower flux and, consequently, an increased footprint area. and shows the potential of each technique.
As reviewed in this work, several studies have investigated
this application, reporting membrane material and prop-
erties, module configurations and operational conditions. Acknowledgements
Experiments of MD and pervaporation were per-
The authors gratefully acknowledge Dr Cristina Cardoso
formed under the same operational and hydrodynamic Pereira for the PVDF membrane fiber fabrication. This
conditions. Two very distinct regions of results for MD research did not receive any specific grant from funding
and pervaporation in terms of selectivity and flux were agencies in the public, commercial or not-for-profit sectors.
presented, being fairly compared in terms of performance
of ethanol removal in same operational conditions.
Highlights
Corroborating the well-known premise of high flux for
MD and high selectivity for pervaporation is valid. The ● Pervaporation and membrane distillation (MD)
present results show that the concentration polarization is were compared for ethanol/water separation;
16 B. A. CINELLI ET AL.

● MD would be more adequate by considering the Campinas. Doctorate Thesis, Universidade Estadual
experimental condition; de Campinas, Campinas, Brazil.
● Concentration polarization cannot be neglected; [9] O’Brien, D.J.; Craig Jr., J.C. (1996) Ethanol production
in a continuous fermentation/membrane pervapora-
● Literature data cannot be directly compared due
tion system. Applied Microbiology and Biotechnology,
to different hydrodynamics. 44 (6): 699–704. doi:10.1007/BF00178605
[10] Groot, W.J.; Kraayenbrink, M.R.; Van Der Lans, R.G.J.
M.; Luyben, K.C.A.M. (1993) Ethanol production in an
Nomenclature integrated fermentation/membrane system. Process
simulations and economics. Bioprocess Engineering, 8
J mass flux (kg/m2 h) (5–6): 189–201. doi:10.1007/BF00369829
C0 concentrations by mass fraction on the membrane [11] Chovau, S.; Gaykawad, S.; Straathof, A.J.J.; Van der
interface Bruggen, B. (2011) Influence of fermentation by-pro-
Cb concentrations by mass fraction on the feed bulk ducts on the purification of ethanol from water using
phase pervaporation. Bioresource Technology, 102 (2): 1669–
Cp concentrations by mass fraction on the permeate 1674. doi:10.1016/j.biortech.2010.09.092
δ thickness of the concentration polarization layer [12] Lewandowicz, G.; Białas, W.; Marczewski, B.;
α selectivity Szymanowska, D. (2011) Application of membrane
αVLE theoretical selectivity distillation for ethanol recovery during fuel ethanol
production. Journal of Membrane Science, 375 (1–2):
212–219. doi:10.1016/j.memsci.2011.03.045
ORCID [13] Gryta, M.; Morawski, A.W.; Tomaszewska, M. (2000)
Ethanol production in membrane distillation bioreac-
Frederico A. Kronemberger http://orcid.org/0000-0003- tor. Catalysis Today, 56 (1–3): 159–165. doi:10.1016/
1220-6340 S0920-5861(99)00272-2
[14] Udriot, H.; Ampuero, S.; Marison, I.W.; Von Stockar,
U. (1989) Extractive fermentation of ethanol using
References membrane distillation. Biotechnology Letters, 11 (7):
509–514. doi:10.1007/BF01026651
[1] Bastos, V.D. (2007) Etanol, alcoolquímica e biorrefi- [15] El-Bourawi, M.S.; Ding, Z.; Ma, R.; Khayet, M. (2006) A
narias. BNDES Setorial, 25: 5–38. framework for better understanding membrane distilla-
[2] Cardona, C.A.; Sánchez, Ó.J. (2007) Trends in biotech- tion separation process. Journal of Membrane Science, 285
nological production of fuel ethanol from different (1–2): 4–29. doi:10.1016/j.memsci.2006.08.002
feedstocks. Bioresource Technology, 99 (13): 5270– [16] Onsekizoglu, P. (2012) Membrane distillation: princi-
5295. doi:10.1016/j.biortech.2007.11.013 ple, advances, limitations and future prospects in food
[3] Drioli, E.; Curcio, E. (2007) Membrane engineering for industry. Advancement Model Applications, pn 233.
process intensification: a perspective. Journal of 266. ISBN: 978-953-51-0428-5
Chemical Technology and Biotechnology (Oxford, [17] Tomaszewska, M. (2000) Membrane Distillation -
Oxfordshire : 1986), 82 (3): 223–227. doi:10.1002/ Examples of Applications in Technology and
(ISSN)1097-4660 Environmental Protection, 9 (1): 27–36.
[4] Stanley, D.; Bandara, A.; Fraser, S.; Chambers, P.J.; [18] Alkhudhiri, A.; Darwish, N.; Hilal, N. (2012)
Stanley, G.A. (2010) The ethanol stress response and Membrane distillation: A comprehensive review.
ethanol tolerance of Saccharomyces cerevisiae. Journal Desalination, 287: 2–18. doi:10.1016/j.desal.2011.08.027
of Applied Microbiology, 109 (1): 13–24. doi:10.1111/ [19] Peng, F.; Hu, C.; Jiang, Z. (2007) Novel ploy(vinyl
j.1365-2672.2009.04657.x alcohol)/carbon nanotube hybrid membranes for per-
[5] Daugulis, A.; Axford, D.; Ciszek, B.; Malinowski, J. vaporation separation of benzene/cyclohexane mix-
(1994) Continuous fermentation of high-strength glu- tures. Journal of Membrane Science, 297 (1–2): 236–
cose feeds to ethanol. Biotechnology Letters, 16 (6): 242. doi:10.1016/j.memsci.2007.03.048
637–642. doi:10.1007/BF00128614 [20] Wang, P.; Chung, T.-S. (2015) Recent advances in
[6] Chang, H.N.; Yang, J.W.; Park, Y.S.; Kim, D.J.; Han, K. membrane distillation processes: membrane develop-
C. (1992) Extractive ethanol production in a mem- ment, configuration design and application exploring.
brane cell recycle bioreactor. Journal of Biotechnology, Journal of Membrane Science, 474: 39–56. doi:10.1016/j.
24 (3): 329–343. doi:10.1016/0168-1656(92)90041-7 memsci.2014.09.016
[7] Nguyen, V.D.; Auresenia, J.; Kosuge, H.; Tan, R.R.; [21] Drioli, E.; Criscuoli, A. (2013) Membrane Distillation.
Brondial, Y. (2011) Vacuum fermentation integrated Water and Wastewater Treatment Technologies,
with separation process for ethanol production. Encyclopedia of Life Support Systems (EOLSS).
Biochemical Engineering Journal, 55 (3): 208–214. e-ISBN: 978-1-84826-250-8. ISBN: 978-1-84826-700-8.
doi:10.1016/j.bej.2011.05.001 [22] Nene, S.; Patil, G.; Raghavarao, K. (2009) 19membrane
[8] Atala, D.I.P. (2004) Montagem, instrumentação, con- Distillation in Food Processing. In: Handbook of mem-
trole e desenvolvimento experimental de um processo brane separations: chemical, pharmaceutical, food, and
fermentativo extrativo de produção de etanol. In: biotechnological applications. Pabby, A.K, SSH Rizvi,
Engenharia de Alimentos; Universidade Estadual de AMS Requena. CRC press, 2008.
 SEPARATION SCIENCE AND TECHNOLOGY 17

[23] Mengual, J.I.; Khayet, M.; Godino, M.P. (2004) Heat situ separation of ethanol from aqueous solutions.
and mass transfer in vacuum membrane distillation. Chemical Engineering, 39. doi:10.3303/CET1439165.
InternatIonal Journal of Heat and Massachusetts p 985–990.
Transfer, 47 (4): 865–875. doi:10.1016/j. [38] Diban, N.; Voinea, O.C.; Urtiaga, A.; Ortiz, I. (2009)
ijheatmasstransfer.2002.09.001 Vacuum membrane distillation of the main pear aroma
[24] Nakao, S.I.; Saitoh, F.; Asakura, T.; Toda, K.; Kimura, compound: experimental study and mass transfer mod-
S. (1987) Continuous ethanol extraction by pervapora- eling. Journal of Membrane Science, 326 (1): 64–75.
tion from a membrane bioreactor. Journal of doi:10.1016/j.memsci.2008.09.024
Membrane Science, 30 (3): 273–287. doi:10.1016/ [39] Zhang, Q.; Nurhayati,; Cheng, C.-L.; Lo, Y.-C.;
S0376-7388(00)80123-4 Nagarajan, D.; Hu, J.; Chang, J.-S.; Lee, D.-J. (2017)
[25] Shi, J.Y.; Zhao, Z.P.; Zhu, C.Y. (2013) Studies on simu- Ethanol production by modified polyvinyl alcohol-
lation and experiments of ethanol-water mixture immobilized Zymomonas mobilis and in situ membrane
separation by VMD using a PTFE flat membrane mod- distillation under very high gravity condition. Applied
ule. Separation and Purification Technology, 123: 53– Energy, 202: 1–5. doi:10.1016/j.apenergy.2017.05.105
63. doi:10.1016/j.seppur.2013.12.015 [40] Di Luccio, M. (2001) Produção de Etanol e Frutose em
[26] Izquierdo-Gil, M.A.; Jonsson, G. (2003) Factors affect- Biorreator Integrado a Processos com Membrana. In:
ing flux and ethanol separation performance in Programa de Engenharia Química; COPPE/UFRJ.
vacuum membrane distillation (VMD). Journal of Doctorate thesis. Universidade Federal do Rio de
Membrane Science, 214 (1): 113–130. doi:10.1016/ Janeiro, Rio deJaneiro, Brazil.
S0376-7388(02)00540-9 [41] Calibo, R.L.; Matsumura, M.; Takahashi, J.; Kataoka, H.
[27] Bandini, S.; Gostoli, C.; Sarti, G.C. (1992) Separation (1987) Ethanol stripping by pervaporation using por-
efficiency in vacuum membrane distillation. Journal of ous PTFE membrane. Journal of Fermentation
Membrane Science, 73 (2–3): 217–229. doi:10.1016/ Technology, 65 (6): 665–674. doi:10.1016/0385-6380
0376-7388(92)80131-3 (87)90009-4
[28] Tomaszewska, M.; Gryta, M.; Morawski, A.W. (1994) [42] Garcı́a-Payo, M.C.; Izquierdo-Gil, M.A.; Fernández-
A study of separation by the direct-contact membrane Pineda, C. (2000) Air gap membrane distillation of aqu-
distillation process. Separations Technology, 4 (4): 244– eous alcohol solutions. Journal of Membrane Science, 169
248. doi:10.1016/0956-9618(94)80028-6 (1): 61–80. doi:10.1016/S0376-7388(99)00326-9
[29] Gryta, M. (2013) Effect of flow-rate on ethanol separa- [43] Kujawska, A.; Kujawski, J.K.; Bryjak, M.; Cichosz, M.;
tion in membrane distillation process. Chemical Papers, Kujawski, W. (2016) Removal of volatile organic com-
67 (9): 1201–1209. doi:10.2478/s11696-013-0382-0 pounds from aqueous solutions applying thermally
[30] Banat, F.A.; Simandl, J. (1999) Membrane distillation driven membrane processes. 2. Air gap membrane dis-
for dilute ethanol: separation from aqueous streams. tillation. Journal of Membrane Science, 499: 245–256.
Journal of Membrane Science, 163 (2): 333–348. doi:10.1016/j.memsci.2015.10.047
doi:10.1016/S0376-7388(99)00178-7 [44] Ishihara, K.; Matsui, K. (1987) Pervaporation of etha-
[31] Vane, L.M. (2005) A review of pervaporation for pro- nol-water mixture through composite membranes
duct recovery from biomass fermentation processes. composed of styrene-fluoroalkyl acrylate graft copoly-
The Journal of Chemical Technology & Biotechnology, mers and cross-linked polydimethylsiloxane mem-
80 (6): 603–629. doi:10.1002/(ISSN)1097-4660 brane. Journal of Applied Polymer Science, 34 (1):
[32] Binning, R.; Lee, R.; Jennings, J.; Martin, E. (1961) 437–440. doi:10.1002/app.1987.070340135
Separation of liquid mixtures by permeation. [45] Blume, I.; Wijmans, J.G.; Baker, R.W. (1990) The
Industrial & Engineering Chemistry, 53 (1): 45–50. separation of dissolved organics from water by perva-
doi:10.1021/ie50613a030 poration. Journal of Membrane Science, 49 (3): 253–
[33] Barancewicz, M.; Gryta, M. (2012) Ethanol production 286. doi:10.1016/S0376-7388(00)80643-2
in a bioreactor with an integrated membrane distilla- [46] Jia, M.D.; Pleinemann, K.V.; Behling, R.D. (1992)
tion module. Chemical Papers, 66 (2): 85–91. Preparation and characterization of thin-film zeolite-
doi:10.2478/s11696-011-0088-0 PDMS composite membranes. Journal of Membrane
[34] Tomaszewska, M.; Białończyk, L. (2013) Production of Science, 73 (2): 119–128. doi:10.1016/0376-7388(92)
ethanol from lactose in a bioreactor integrated with 80122-Z
membrane distillation. Desalination, 323: 114–119. [47] Vankelecom, I.F.J.; Depre, D.; De Beukelaer, S.;
doi:10.1016/j.desal.2013.01.026 Uytterhoeven, J.B. (1995) Influence of zeolites in
[35] Tomaszewska, M.; Bialonczyk, L. (2016) Ethanol pro- PDMS membranes: pervaporation of water/alcohol
duction from whey in a bioreactor coupled with direct mixtures. The Journal of Physical Chemistry, 99 (35):
contact membrane distillation. Catalysis Today, 268: 13193–13197. doi:10.1021/j100035a024
156–163. doi:10.1016/j.cattod.2016.01.059 [48] Schmidt, S.; Myers, M.; Kelley, S.; McMillan, J.;
[36] Kaseno, I.; Miyazawa, I.; Kokugan, T. (1998) Effect of Padukone, N. (1997) Evaluation of PTMSP membranes
product removal by a pervaporation on ethanol fer- in achieving enhanced ethanol removal from fermenta-
mentation. Journal of Fermentation and Bioengineering, tions by pervaporation. In Davison, B.; Wyman, C.;
86 (5): 488–493. doi:10.1016/S0922-338X(98)80 Finkelstein, M. (Eds) Biotechnology for Fuels and
157-8 Chemicals; Humana Press. doi:10.1007/BF02920447
[37] Rom, A.; Strommer, M.; Friedl, A. (2014) Comparison [49] Chen, X.; Ping, Z.H.; Long, Y. (1998) Separation prop-
of sweepgas and vacuum membrane distillation as in- erties of alcohol-water mixture through silicalite-I-
18 B. A. CINELLI ET AL.

filled silicone rubber membranes by pervaporation. [61] Bello, R.H.; Linzmeyer, P.; Franco, C.M.B.; Souza, O.;
Journal of Applied Polymer Science, 629–636. Sellin, N.; Medeiros, S.H.W.; Marangoni, C. (2014)
doi:10.1002/(SICI)1097-4628(19980124)67:4<629:: Pervaporation of ethanol produced from banana
AID-APP5>3.0.CO;2-4. waste. Waste Management, 34 (8): 1501–1509.
[50] O’Brien, D.J.; Roth, L.H.; McAloon, A.J. (2000) Ethanol doi:10.1016/j.wasman.2014.04.013
production by continuous fermentation-pervaporation: [62] Naik, P.V.; Kerkhofs, S.; Martens, J.A.; Vankelecom, I.
a preliminary economic analysis. Journal of Membrane F.J. (2016) PDMS mixed matrix membranes containing
Science, 166 (1): 105–111. doi:10.1016/S0376-7388(99) hollow silicalite sphere for ethanol/water separation by
00255-0 pervaporation. Journal of Membrane Science, 502: 48–
[51] Moermans, B.; Beuckelaer, W.D.; Vankelecom, I.F.J.; 56. doi:10.1016/j.memsci.2015.12.028
Ravishankar, R.; Martens, J.A.; Jacobs, P.A. (2000) [63] Fan, S.; Xiao, Z.; Li, M. (2016) Energy efficient of
Incorporation of nano-sized zeolites in membranes. ethanol recovery in pervaporation membrane bioreac-
Chemical Communicable, (24): 2467–2468. tor with mechanical vapor compression eliminating the
doi:10.1039/b007435g cold traps. Bioresource Technology, 211: 24–30.
[52] Tuan, V.A.; Li, S.; Falconer, J.L.; Noble, R.D. (2002) doi:10.1016/j.biortech.2016.03.063
Separating organics from water by pervaporation with [64] Sun, W.; Jia, W.; Xia, C.; Zhang, W.; Ren, Z. (2017)
isomorphously-substituted MFI zeolite membranes. Study of in situ ethanol recovery via vapor permeation
Journal of Membrane Science, 196 (1): 111–123. from fermentation. Journal of Membrane Science, 530:
doi:10.1016/S0376-7388(01)00590-7 192–200. doi:10.1016/j.memsci.2017.02.034
[53] O’Brien, D.J.; Senske, G.E.; Kurantz, M.J.; Craig, J.C. [65] Poling, B.E.; Prausnitz, J.M.; O’Connell, J.P. (2004) The
(2004) Ethanol recovery from corn fiber hydrolysate Properties of Gases and Liquids, 5th ed.; Mcgraw-hill:
fermentations by pervaporation. Bioresource New York.
Technology, 92 (1): 15–19. [66] Gmehling, V.J.; Onken, U.; Rarey-Nies, J.R. (1991)
[54] Aroujalian, A.; Belkacemi, K.; Davids, S.J.; Turcotte, G.; Vapor-liquid equilibrium data collection (chemistry
Pouliot, Y. (2006) Effect of residual sugars in fermenta- data series, Vol. 1, part 2e). DECHEMA, Frankfurt/
tion broth on pervaporation flux and selectivity for M. 1989. LIII, 648 S., geb., DM 312. Chemie Ingenieur
ethanol. Desalination, 193 (1): 103–108. doi:10.1016/j. Technik, 63 (1): 87.
desal.2005.06.058 [67] Hristova, M.; Damgaliev, D.; Popova, D. (2010)
[55] Lewandowska, M.; Kujawski, W. (2007) Ethanol pro- Estimation of water-alcohol mixture flash point.
duction from lactose in a fermentation/pervaporation Journal of the University of Chemical Technology and
system. Journal of Food Engineering, 79 (2): 430–437. Metallurgy, 45 (1): 19–24.
doi:10.1016/j.jfoodeng.2006.01.071 [68] Baker, R.W. (2004) Membrane Technology and
[56] Ding, W.W.; Wu, Y.T.; Tang, X.Y.; Yuan, L.; Xiao, Z.Y. Applications, 2nd ed.; Wiley Online Library.
(2010) Continuous ethanol fermentation in a closed- doi:10.1002/0470020393.
circulating system using an immobilized cell coupled [69] Wilke, C.R.; Chang, P. (1955) Correlation of diffusion
with PDMS membrane pervaporation. Journal of coefficients in dilute solutions. AIChE Journal, 1 (2):
Chemical Technology & Biotechnology, 86 (1): 82–87. 264–270. doi:10.1002/(ISSN)1547-5905
doi:10.1002/jctb.v86.1 [70] Khattab, I.S.; Bandarkar, F.; Fakhree, M.A.A.; Jouyban,
[57] García, M.; Sanz, M.T.; Beltrán, S. (2009) Separation by A. (2012) Density, viscosity, and surface tension of
pervaporation of ethanol from aqueous solutions and water+ethanol mixtures from 293 to 323K. Korean
effect of other components present in fermentation Journal of Chemical Engineering, 29 (6): 812–817.
broths. The Journal of Chemical Technology & doi:10.1007/s11814-011-0239-6
Biotechnology, 84 (12): 1873–1882. doi:10.1002/jctb.2259 [71] Noble, R.D.; Stern, S.A. (1995) Membrane Separations
[58] Dobrak, A.; Figoli, A.; Chovau, S.; Galiano, F.; Simone, Technology: Principles and Applications, 1st ed.;
S.; Vankelecom, I.F.J.; Drioli, E.; Van der Bruggen, B. Elsevier Science. eBook ISBN: 9780080536187
(2010) Performance of PDMS membranes in perva- [72] Li, L.; Xiao, Z.; Tan, S.; Pu, L.; Zhang, Z. (2004)
poration: effect of silicalite fillers and comparison Composite PDMS membrane with high flux for the
with SBS membranes. The Journal of Colloid and separation of organics from water by pervaporation.
Interface Science, 346 (1): 254–264. doi:10.1016/j. Journal of Membrane Science, 243 (1): 177–187.
jcis.2010.02.023 doi:10.1016/j.memsci.2004.06.015
[59] Han, X.; Wang, L.; Li, J.; Zhan, X.; Chen, J.; Yang, J. [73] Vane, L.M.; Alvarez, F.R.; Rosenblum, L.; Govindaswamy,
(2011) Separation of ethanol from ethanol/water mix- S. (2012) Efficient ethanol recovery from yeast fermenta-
tures by pervaporation with silicone rubber mem- tion broth with integrated distillation—membrane pro-
branes: effect of silicone rubbers. Journal of Applied cess. Industrial & Engineering Chemistry Research, 52 (3):
Polymer Science, 3413–3421. doi:10.1002/app.32991. 1033–1041. doi:10.1021/ie2024917
[60] Lee, H.J.; Cho, E.J.; Kim, Y.G.; Choi, I.S.; Bae, H.J. [74] Liang, L.; Ruckenstein, E. (1996) Pervaporation of etha-
(2012) Pervaporative separation of bioethanol using a nol-water mixtures through polydimethylsiloxane-
polydimethylsiloxane/polyetherimide composite hol- polystyrene interpenetrating polymer network sup-
low-fiber membrane. Bioresource Technology, 109: ported membranes. Journal of Membrane Science, 114
110–115. doi:10.1016/j.biortech.2012.01.060 (2): 227–234. doi:10.1016/0376-7388(95)00319-3