ISSN 09655441, Petroleum Chemistry, 2013, Vol. 53, No. 8, pp. 567–571. © Pleiades Publishing, Ltd., 2013.

Original Russian Text © E.P. Ageev, N.N. Matushkina, N.L. Strusovskaya, 2013, published in Membrany i Membrannye Tekhnologii, 2013, Vol. 3, No. 3, pp. 231–236.

Concentration History of Pervaporation

E. P. Ageev, N. N. Matushkina, and N. L. Strusovskaya
Faculty of Chemistry, Moscow State University, Moscow, 119991 Russia
email: ageev@phys.chem.msu.ru
Received December 5, 2012

Abstract—The permeability and selectivity of the separation of aqueous organic solutions by pervaporation
have been studied on semicrystalline poly(ethylene terephthalate) membranes and asymmetric amorphous
polyvinyltrimethylsilane membranes in the presence of structural rearrangements initiated by the develop
ment of relaxation processes. Attention was focused on the concentration history, which lies in the fact that
the permeability and the permeate composition depend on the concentration of the previous solution passed
through the membrane at a fixed composition of the feed solution. It has been shown that micropores are
formed in both membranes under certain pervaporation conditions and mass transfer occurrs through two
competing transport channels. As a result, the oscillatory kinetics of pervaporation and the inversion of sep
aration selectivity in a steady state are observed experimentally.

Keywords: separation processes, selective flux, target component, transport channels

DOI: 10.1134/S0965544113080021

INTRODUCTION appear and the films lose mechanical strength and

A method for actively controlling mass transfer
through a polymer membrane consists in changing the
properties of the membrane as a result of its structural
rearrangements. These processes include the crystalli
zation of semicrystalline polymers under the action of
liquids whose solubility parameters are close to the
solubility parameter of the polymer. It was repeatedly
noted in the literature [1, 2] that an increase in the
degree of crystallinity decreases the permeability of
membranes, thereby suggesting that this line of
research does not hold much promise. Indeed, the
crystalline phase is denser than the amorphous one;
therefore, an increase in crystallinity decreases the
fractional free volume and, correspondingly, trans
membrane mass transfer coefficients.
However, a directly opposite situation—a sharp
increase in the permeability of crystallized mem
branes—cannot be excluded. The fact is that a local
increase of polymer density, which decreases the
transmembrane flow, is only one side of the process.
The other side consists in a change in the properties of
amorphous regions. It is likely that a local density
increase due to the appearance of a crystalline phase
can occur only upon the local loosening (a decrease of
density) of amorphous regions in a number of cases,
for example, in the separation of liquid media; this
leads to the formation of the microheterogeneous
defect structure of the polymer. The crystallization of
amorphous poly(ethylene terephthalate) (PET) films
on contact with solvents that crystallize the polymer is
an extreme manifestation of the occurrence of such
processes. In this case, a great number of defects
undergo degradation.
These experimental observations gave us the idea to
regulate the number and size of defects in semicrystal
line polymer films by varying active component con
centration.
The formation of a defect structure can be inter
preted as a local microfracture of the material. In this
case, the ratios between the rates of crystallization and
other relaxation processes, internal tensile stresses,
and the presence of surfactants that stabilize the dis
persed polymer structure play an important role. For
the purity of the experiment, the concentration of an
active constituent should be varied in a solution with
an inactive component, which is not sorbed by the
polymer and does not crystallize it.
The aim of this work was to study the permeability
and selectivity of pervaporation separation in the pres
ence of structural rearrangements initiated by the
development of relaxation processes in membranes.
Attention was focused on the concentration history of
pervaporation. The concentration history lies in the
fact that at a fixed composition of the initial solution,
the permeability and permeate composition depend
on the concentration of the previous solution passed
through the membrane.
EXPERIMENTAL RESULTS AND DISCUSSION
Let us consider the pervaporation properties of the
PET–acetone–water system. PET is taken as films
with a thickness of 40 µm and a density of 1.34–1.35.
This polymer is characterized by low crystallinity

567
568
5 100
2 c'
4 80
G, kg (10 µm) m ⎯2 h⎯1
3 60
2 40
1 20
1
0 0
0 20 40 60 80 100
c
Fig. 1. Pervaporation in the PET−acetone−water system at
293 K. The concentration dependence of the (1) flux and
(2) permeate composition. Each point in a curve corre
sponds to a new membrane.
because the calculated density of fully crystalline PET
is 1.45–1.48 [3]. The sorption of acetone by the poly
mer is accompanied by its crystallization, whose
intensity can be regulated using an acetone solution in
water, as water not only does not crystallize PET but
also is not sorbed by it. In this case, the concentration
dependence of permeability and selectivity in the per
vaporation process can be studied in three versions:
first, a new membrane is used for each concentration
of the feed solution; second, one membrane is used
and the active component concentration is succes
sively increased; the third version differs from the sec
ond in that the concentration of the inactive compo
nent is successively increased. In these versions of the
concentration history of the process, the conditions of
polymer crystallization and, hence, its structure relax
ation are varied.
Pervaporation was carried out at 293 K by evapo
rating the test solution through a membrane into a
vacuum. The flux was normalized to a membrane with
a thickness of 10 µm.
Figures 1 and 2 show the transmembrane flux and
the concentration of acetone in the permeate in the
first and second process versions, respectively. On the
membranes that contacted with the separated solution
of only one concentration (in Fig. 1, each experimen
tal data point was obtained on a new membrane) at an
initial acetone concentration to 40 vol %, it increased
to 90 vol % in the permeate. As the active constituent
content was further increased, polymer crystallization
came into play, and it was accompanied by an increase
in the permeability and the inversion of selectivity.
Water began to predominantly pass through the hydro
phobic membrane! In the initial equivolumetric solu
tion passed through the membrane (see Fig. 1 and the
AGEEV et al.
3 100
1 c'
80
G, kg (10 µm) m ⎯2 h⎯1
2
60
40
1
2 20
0
20 40 60 80 100
c
Fig. 2. Pervaporation in the PET−acetone−water system at
293 K. The concentration dependence of the (1) flux and
(2) permeate composition. Successive increase in acetone
concentration on the same membrane.
table), the acetone content of the permeate decreased
by a factor of 2! Then, as the concentration of the
active constituent was increased and the polymer crys
tallization was complete, the structural rearrange
ments of the polymeric matrix, which accompanied
relaxation processes, led to a decrease in the perme
ability and a weakening of the partial flow of water.
The crystallization of PET and the relaxation of a
formed new polymer structure occurred differently in
the first and second versions of the concentration his
tory. In the first version, the active constituent in any
concentration crystallized and modified the initial
PET structure, whereas the above processes occur
stepwise in small doses in the second version. As a
result, in the first version, we obtained a more non
equilibrium defect structure, which exhibited
increased permeability, as compared with that in the
second version.
For revealing the reasons for the inversion of selec
tivity in the first version of the concentration prehis
tory of the pervaporation process, we additionally
determined the concentration dependences, the equi
librium degrees of swelling, porosimetry data, changes
in the relative rotational mobility of a spin (paramag
netic) probe, and a relative change in the crystallinity
of the polymer.
The presence or absence of porosity in PET sam
ples after the pervaporation of solutions with different
concentrations was determined by a hydrodynamic
method [4]. The method consists in that a pressure on
the surface of a penetrating liquid can be specified and
increased at a chosen rate with the aid of a special
device. The liquid penetrates through pores of a cer
tain diameter depending on pressure. The flow passed
through a liquid film was measured with a highly sen
PETROLEUM CHEMISTRY Vol. 53 No. 8 2013
 CONCENTRATION HISTORY OF PERVAPORATION
Properties of PET films treated with aqueous acetone solutions at 293 K
4 kg (10 μm)
c0, vol % c, vol % G × 10 , 
 α∞
m /h
2 N × 10–14, m–2
0 0 0.8 0 Pores were not detected
30 84 0.9 1.0
50 25 3.8 4.6
60 31 3.9 8.0
70 56 1.9 8.6
sitive (to 10–5 mL/s) flow meter. The readings of the
flow meter pressure gage were converted into an elec
trical signal and arrived at a data processing and con
1 sorption, it decreases the surface energy to facilitate
trol unit and then at a digital printer .
The paramagnetic probe used was the stable radical
2,2,6,6tetramethylpiperidine1oxyl, which was
introduced into the polymer films from the vapor
phase at a temperature of 318–333 K. The concentra
tion of the radical in the polymer was 10–3 mol/L. The
EPR spectra were measured on an EPR8 spectrome
ter at 293 K. The rotational mobility of the probe
(τ/τ0) was calculated from these spectra. The degree of
crystallinity of PET (ϕ) was calculated from a change
in the density of film samples, which was determined
by a flotation method in an aqueous solution of cal
cium nitrate with an accuracy of ±0.05%. The table
summarizes the rotational mobility of the probe and
the degrees of crystallinity normalized to their initial
values. The first row in the table corresponds to zero
concentration of acetone in the solution, that is, to
pure water. PET does not sorb water (α∞ = 0); there
fore, all of the properties of the initial polymer are
retained. The solution containing 30% acetone does
not crystallize PET, but it slightly modifies the poly
mer, as follows from an increase in the spin probe rota
tional mobility. As expected, the permeate was
enriched in the active component. As a result of the
contact of the polymer with equivolumetric solutions,
the degree of crystallinity in PET increases and
micropores appear. In this case, the selectivity inver
sion take place, which is manifested in that water
began to predominantly penetrate through the hydro
phobic polymer. At higher concentrations of the active
component, the presence of pores was not detected,
but the selectivity inversion was retained and the struc
ture continued to remain defective, as follows from not
only an increase in the rotational mobility of the radi
cal probe, but also a change in the pattern of the EPR
spectrum. In the EPR spectra of the films treated by
solutions with acetone concentrations to 50 vol %, the
radical had two rotational degrees of freedom or three
degrees of freedom at higher concentrations of the
active component; this manifested itself in a loosening
1 The porosimetric measurements were performed by N.S. Orlov.
PETROLEUM CHEMISTRY Vol. 53 No. 8
569
Number of pores
Pore radius r × 1010, m τ/τ0 ϕ
1.0 1.0
1.3 1.0
8.5 1.3 × 1017 2.2 2.0
Pores were not detected 5.0 2.2
4.3
of the third component in the triplet EPR spectrum
characteristic of the radical probe.
Acetone is a surfactant with respect to PET. Upon
local polymer degradation by internal stress tensile
forces with the formation of a microheterogeneous
defect structure and, in some cases, a microporous
polymer structure. Furthermore, the amphiphilic ace
tone molecule sorbed on PET via its hydrophobic
moiety (methyl groups) hydrolyzes the surface
formed. As a result, a new mass transfer channel—
hydrophilic pores—appears in the hydrophobic poly
mer. Thus, on the pervaporation of aqueous acetone
solutions through a PET membrane, polymer crystal
lization occurs, and a channel of transfer through
pores is added to the channel of mass transfer through
a continuous structure; the transport properties of this
additional channel are different to cause the inversion
of selectivity.
It was of interest to study the kinetics of mass trans
fer complicated by structural changes in the polymer
membrane. The PET–acetone–inactive component
systems are unsuitable for this purpose because of
small transmembrane flows under the nonsteady
state conditions of membrane operation.
We used commercial asymmetric membranes
made from polyvinyltrimethylsilane (PVTMS) of the
PA160. They consisted of an active nonporous diffu
sion layer ~0.2 µm in thickness and a macroporous
substrate 200 µm in thickness, which possessed low
masstransfer resistance. The porous substrate con
sisted of open pores with sizes from fractions of a
micron to several microns.
The average molecular weight of PVTMS was 1.5 ×
106 D; its density, glasstransition temperature, and
solubility parameter were 0.84–0.89, 443 K, and
8.2 (cal/cm3)1/2, respectively. The capability of
PVTMS for crystallization is very weakly pronounced;
this is most likely due to branching in its macromole
cules.
The asymmetric membranes were formed from
PVTMS by solvent evaporation from a solution of the
polymer and its subsequent precipitation with a mod
ifying solution, which forms micropores. A field of
internal stresses appears in the active layer of the asym
2013
570 AGEEV et al.

сsurfactant, vol % сsurfactant, vol %

сsurfactant, vol % 60
2
5' 40
95 6' I, kg m ⎯2 h ⎯1
3', 4'
7'
1
3
90 2'
90
1
20 40 t, min
0 100 200 t, min

Fig. 4. (1) Kinetics of pervaporation of an equivolumetric

aqueous solution of isopropanol through an asymmetric
80 PVTMS membrane and (2) changes in the alcohol con
centration in the permeate. The membrane contacted with
I, kg m ⎯2 h ⎯1 the liquid on the macroporous substrate at 303 K.

tration of the active component, its concentration in

the permeate was 90 vol %. At higher concentrations
2 of acetone in the feed solution, its concentration in the
permeate increased to 95 vol % or higher. For the solu
tion containing 80 vol % acetone, antibatic regular
I, kg m ⎯2 h ⎯1 fluctuations of permeability and selectivity were
7 observed and the steady state of pervaporation was
6 reached in more than 200 min. In this case, the con
1 0.3 5 centration history of the process consists in the occur
4 rence of a critical surfactant concentration, at which
3 membrane operation conditions dramatically
0.1 2 changed.
1
20 40 t, min Figure 4 shows the results of pervaporation in the
PVTMS–isopropanol–water system with an equivo
lumetric concentration of solution. The periodic
100 200 t, min inversion of selectivity can be observed in this system;
that is, the solution enriched in either the alcohol or
water passed through the membrane at regular inter
Fig. 3. Flux during the pervaporation of aqueous acetone vals of about 30 min. The oscillatory regime of mem
solution through an asymmetric PVTMS membrane and
concentration changes in the permeate (vol %, acetone):
brane operation can be explained as follows: The
(1) 0; (2, 2 ') 20; (3, 3 ') 30; (4, 4 ') 40; (5, 5 ') 50; (6, 6 ') 60; hydrophilization of the surface of micropores facili
(7, 7 ') 70; (8, 8 ') 80. The membrane contacted with the tates the predominant transfer of small and more polar
liquid on the active layer at 293 K. water molecules and an increase in the transmem
brane flow as a whole. An increased concentration of
water in pores caused surfactant desorption and
metric membrane due to the formation conditions. As increased the surface stress and, hence, cohesion
in the crystallized PET, this circumstance allows ten forces. These forces tend to decrease the surface area,
sile stresses in the presence of a surfactant to locally and the structure is converted again into the continu
destroy the PVTMS membrane material and to form a ous one, but, possibly, with surface defects because a
microheterogeneous defect structure of the active constant surfactant concentration in the solution does
layer. not make it possible to completely heal a microscopic
Figure 3 shows the concentration dependence of crack. The collapse of micropores increases tensile
the kinetics of permeability and pervaporation selec stresses, and the mechanism of mass transfer changes
tivity in the PVTMS–acetone–water system. At ace to an activation mechanism, which leads to the pre
tone concentrations to 80 vol %, we have smooth ferred surfactant penetration, and the process is
kinetic curves for the flux and the acetone concentra repeated once again. Thus, the oscillations of flow,
tion in the flow. The process reached a steady state in concentration, and surface philicity are accompanied
approximately 20 min. For an initial 20 vol % concen by the oscillations of internal stresses. The oscillatory

PETROLEUM CHEMISTRY Vol. 53 No. 8 2013

 CONCENTRATION HISTORY OF PERVAPORATION 571

сsurfactant, vol % conditions (see Fig. 4) was much longer than that in
the PVTMS–acetone–water system. This is related to
60 an additional face of the concentration history of per
vaporation, namely: the orientation of an asymmetric
membrane relative to the surface of the feed solution.
If the active layer of the membrane directly contacts
with the feed solution, the sorption and, correspond
ingly, relaxation processes occur more intensely to
2 facilitate the achievement of a steady state. If the
30
membrane is overturned and the macroporous sub
strate contacts with the solution, whereas the evacu
I, kg m ⎯2 h ⎯1 ated chamber contacts with the active layer, the con
1 centration of the penetrating solution in the active
3 layer is relatively small and all of the relaxation pro
cesses occur more slowly. For this reason, the time
taken to reach a steady state was longer than 300 min
2 in the PVTMS–isopropanol–water system in which
the solution contacted with the macroporous sub
strate, whereas the process reached steadystate con
1 ditions in 200 min in the PVTMS–acetone–water sys
tem in which the membrane was overturned. In order
to be convinced of the validity of the above judgments,
we carried out an experiment with the opposite orien
100 200 t, min tation of the membrane relative to the solution in the
pervaporation system. Ass can be seen in Fig. 5, the
Fig. 5. (1) Kinetics of pervaporation of an equivolumetric time taken to reach a steady state in the PVTMS–iso
aqueous solution of isopropanol through an asymmetric propanol–water system shortened to 200 min, as com
PVTMS membrane and (2) changes in the alcohol con
centration in the permeate. The membrane contacted with pared with the data (Fig. 4) obtained with the opposite
the liquid on the active layer at 303 K. orientation of the membrane.

period is related to structural transitions, which are REFERENCES

controlled by the kinetics of surfactant absorption and
desorption.
The proposed physical picture of the process is the
reversible manifestation of the Rehbinder effect in
polymer systems. In this case, reversible (in a kinetic
sense) structural transitions play the role of a feedback
relay, which gives commands for turning on and off the
stages of flow through pores; thereby leading to the
oscillatory modes of permeability and selectivity.
In the PVTMS–isopropanol–water system, the
time taken to reach steadystate membrane operating
1. Yu. I. Dytnerskii, Membrane Processes for Separation of
Liquid Mixtures (Nauka, Moscow, 1975) [in Russian].
2. S. A. Reitlinger, Permeability of Polymer Materials
(Khimiya, Moscow, 1974) [in Russian].
3. Encyclopedia of Polymers, Ed. by V. A. Kabanov,
M. S. Akutin, N. F. Bakeev, et al. (Sovetskaya Entsiklo
pediya, Moscow, 1977), Vol. 3 [in Russian].
4. N. S. Orlov and S. M. Shebanov, in Proceedings of
II AllUnion Conference on Membrane Separation Meth
ods (Vladimir, 1977), p. 133 [in Russian].
Translated by V. Makhlyarchuk

PETROLEUM CHEMISTRY Vol. 53 No. 8 2013