Journal of Membrane Science 187 (2001) 255–269

Removal of dilute volatile organic compounds in water

through graft copolymer membranes consisting of
poly(alkylmethacrylate) and poly(dimethylsiloxane) by
pervaporation and their membrane morphology
Tadashi Uragami∗ , Hiroshi Yamada, Takashi Miyata
Chemical Branch, Faculty of Engineering and High Technology Research Center, Kansai University, Suita, Osaka 564-8680, Japan
Received 14 March 2000; received in revised form 17 January 2001; accepted 17 January 2001

Abstract
Removal of volatile organic compounds (VOCs) such as benzene and chloroform in aqueous benzene and chloroform
solutions through the poly(methylmethacrylate)-poly(dimethylsiloxane) (PMMA-g-PDMS), poly(ethylmethacrylate)-PDMS
(PEMA-g-PDMS), and poly(n-butylmethacrylate)-PDMS (PBMA-g-PDMS) graft copolymer membranes was investigated
by pervaporation. When aqueous solutions of dilute VOCs were permeated through the PMMA-g-PDMS and PEMA-g-PDMS
membranes, these membranes showed a benzene- and chloroform-permselectivity. Permeation and separation characteristics
of the PMMA-g-PDMS and PEMA-g-PDMS membranes changed drastically at a DMS content of about 40 and 70 mol%,
respectively. Permeation rate and VOC-permselectivity of the PBMA-g-PDMS membrane, however, increased gradually with
increasing DMS content, unlike those of the PMMA-g-PDMS and PEMA-g-PDMS membranes. Furthermore, the transmis-
sion electron microscope observations revealed that the PMMA-g-PDMS and PEMA-g-PDMS membranes had microphase
separations consisting of a PDMS phase and a poly(alkylmethacrylate) phase, but the PBMA-g-PDMS membrane was ho-
mogeneous. The permeability and permselectivity for aqueous VOCs solutions through their graft copolymer membranes in
pervaporation are discussed in detail from the viewpoint of their membrane structure, and permselectivity was analyzed by
the solution–diffusion theory. © 2001 Elsevier Science B.V. All rights reserved.
Keywords: Graft copolymer membrane; Pervaporation; Permselectivity of VOCs; Microphase-separated structure; Membrane morphology

1. Introduction and ponds. Drinking water sources are often depen-

Volatile organic compounds (VOCs) such as aro-
matic hydrocarbons and chlorinated hydrocarbons
used as washing solvents in the chemical industry
cause soil pollution. Such pollution contaminates
subterranean water and consequently water in rivers
∗ Corresponding author. Tel.: +81-6-6368-0805;
fax: +81-6-6330-3770.
E-mail address: uragami@ipcku.kansai-u.ac.jp (T. Uragami).
dent on such a contaminated water. Because VOCs
are known to be chemical materials disturbing in-
ternal secretion (environmental hormones) which are
potentially dangerous to human. As such, these chem-
icals have to be removed from a contaminated water.
An activated carbon method, ozone processing and
other methods have been developed to remove VOCs
from water [1]. However, these methods do not ef-
ficiently remove VOCs and the cost of treatment is
high.

0376-7388/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 6 - 7 3 8 8 ( 0 1 ) 0 0 3 5 5 - 6
256 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269

Pervaporation is a promising membrane technique in order to develop pervaporation membranes for the
for the separation of VOCs/water mixtures [2–6], as removal of VOCs in water.
pervaporation can be used to separate organic liquid
mixtures such as azeotropic mixtures or close-boiling
point mixtures. Poly(dimethylsiloxane) (PDMS) 2. Experimental
membranes have been studied as VOC-permselective
membranes, because they have a high permselec- 2.1. Materials
tivity and permeability for organic chemicals in
water [7–10]. Brooks and Livingston [11] demon- A PDMS macromonomer of a pendant PDMS
strated removal of chloroform in tap water through a (81 units) (1) was supplied by Toray Dow Corning
hollow-fiber PDMS membrane, which showed high Silicone Co., Ltd. Methylmethacrylate (MMA) (2),
chloroform-permselectivity and permeability. The ethylmethacrylate (EMA) (3) and n-butylmethacrylate
high VOCs-permselectivity of the PDMS membrane (BMA) (4) as comonomers were purified by dis-
is attributable to its stronger affinity for VOCs than tillation under reduced pressure in nitrogen gas.
for water. Another advantage of the PDMS membrane 2,2-Azobis(2-methylpropionitrile) (AIBN) recrystal-
is high diffusivity of VOCs in the membrane due to its lized from methanol solution was used as an initiator.
low Tg . The PDMS membrane has some disadvantages All other solvents and reagents were of analytical
however, most notably weak mechanical strength. grade from commercial sources and were used with-
Nakagawa et al. have introduced cross-linking into the out further purification.
PDMS membrane to improve its mechanical strength.
The introduction of cross-linking, however, lead to
lower diffusivity of the permeants in the membrane
[12].
Multicomponent polymers have certain proper-
ties that cannot be achieved in a homopolymer. It
is well known that multicomponent polymers con-
sisting of immiscible components are characterized
by a microphase separation. Many researchers have
studied the relationship between microphase separa-
tion and the material properties of multicomponent
polymers. There are, however, few reports revealing
the relationship between membrane permselectivity
and microphase separation. Previously, we reported
that graft and block copolymer membranes consisting
of poly(methylmethacrylate) (PMMA) and PDMS
undergo microphase separation, and that their perms-
electivity is strongly influenced by their morphology
[13–17]. The microphase separation phases, consist- 2.2. Copolymerization of PDMS macromonomer
ing of a PMMA phase and a PDMS phase, cause with MMA, EMA and BMA
the high permselectivity and permeability of the graft
and block copolymer membranes without lowering The PDMS macromonomer and alkylmethacrylate
mechanical strength. were dissolved in benzene to make a 40 wt.% solu-
In this study, we focused on the microphase- tion with AIBN, and placed in a glass tube. Copo-
separated structure of two-component polymer lymerizations were performed for 6 h to synthes-
membranes consisting of PDMS and poly(alkylmeth- ize poly(methylmethacrylate)-poly(dimethylsiloxane)
acrylate), and investigated the relationship between (PMMA-g-PDMS), poly(ethylmethacrylate)-PDMS
microphase separation and membrane permselectiv- (PEMA-g-PDMS), and poly(n-butylmethacrylate)-
ities for aqueous benzene and chloroform solutions PDMS (PBMA-g-PDMS) under nitrogen gas. The
 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269 257

structures of the obtained copolymers (5) are as chromatograph (GPC), equipped with a TSK-GEL
follows: column (Tosoh Co. Ltd.; G2000HXL, G3000HXL,
G5000HXL). Characterizations of the resulting graft
copolymers are summarized in Table 1.

2.3. Membrane preparation

A prescribed amount of the graft copolymer was

These copolymers were purified by reprecipita-
tion from benzene solution into a 1:2 mixture of
n-hexane and ethanol, and dried in vacuo. The DMS
content in PMMA-g-PDMS, PEMA-g-PDMS, and
PBMA-g-PDMS were determined from 1 H NMR
(JEOL; GSX-400spectra) by measuring the integrals
of the peaks assigned to the methyl protons (3.5 ppm)
of the methacrylate and DMS protons (0 ppm) of the
PDMS macromonomer. Average molecular weights of
the copolymers were determined by a gel permeation
dissolved in benzene at 25◦ C at a concentration of
4 wt.% for the preparation of casting solutions. Their
membranes were prepared by pouring the casting so-
lution onto rimmed glass plates and then allowing the
solvent to evaporate completely at 25◦ C. The result-
ing membranes were transparent and their thickness
were 80–150 ␮m.
2.4. Glass transition temperature measurement
The glass transition temperatures (Tg ’s) of the
PMMA-g-PDMS, PEMA-g-PDMS, and PBMA-g-
PDMS membranes were determined by a differential
scanning calorimeter (DSC) (Rigaku; TAS-200). The
specimens were heated from about −160 to 180◦ C,
with a heating rate of 20◦ C/min.

Table 1
Result of the copolymerization of MMA, EMA, and BMA with PDMS macromonomer
Monomer Molecular weighta Polymerb Tg c

Methacrylate Mol% of DMS M̄w × 10−5 M̄n × 10−5 M̄w /M̄n Mol% of DMS Low Middle High
units in feed units in feed
MMA 0 0.89 0.51 1.75 0 – – 128
50 1.74 0.66 2.62 34 −128 – 120
60 1.14 0.60 1.90 52 −125 – 118
70 1.52 0.74 2.04 68 −126 – 112
EMA 0 2.82 2.21 1.28 0 – – 75
10 1.96 1.36 1.43 17 −119 −37 76
30 2.11 1.46 1.43 56 −120 −40 72
50 1.95 1.95 1.55 72 −120 −45 74
60 1.94 1.94 1.50 75 −119 −45 74
70 2.40 1.69 1.42 85 −121 −45 69
BMA 0 2.22 2.12 1.05 0 – – 38
10 1.67 1.24 1.35 16 – −44 –
30 1.63 1.21 1.35 41 – −40 –
60 2.05 1.76 1.17 73 – −40 –
70 2.13 1.83 1.17 86 – −39 –
a Determined by GPC.
b Determined by 1 H NMR.
c Determined by DSC.
258 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269
2.5. Permeation measurements
Pervaporation was carried out under the following
conditions: permeation temperature, 40◦ C; pressure of
permeate side, 1 × 10−2 Torr. The effective membrane
area was 13.8 cm2 . Aqueous solutions of 0.05 wt.%
benzene and 0.02 wt.% chloroform were used as feed
solutions. The permeate was collected in a U-tube
at liquid nitrogen temperature. The permeation rate
(kg/m2 h) was determined from the weight of the per-
meate, permeation time, and effective membrane area
(A). In order to facilitate comparison of permeation
rates of membranes with different thickness, a normal-
ized permeation rate (kg/m h) which is the product of
the permeation rate and the membrane thickness, was
used. The benzene or chloroform concentrations in
the feed solutions and permeants were determined by
gas chromatography (Shimadzu GC-17A) using flame
ionization detector (FID) and capillary column (Shi-
madzu Co. Ltd.: PorapacQ) that was heated to 180◦ C.
The separation factor, αVOC/H2 O , was calculated from
Eq. (1)
(WVOC /WH2 O )P
αVOC/H2 O = (1)
(WVOC /WH2 O )F
where (WVOC and WH2 O )F are the weight frac-
tions of benzene and water in the feed solution,
and (WVOC and WH2 O )P are those in the permeate,
respectively.
2.6. Contact angle measurement
Contact angles for water on the surface of the
air-side of the PMMA-g-PDMS, PEMA-g-PDMS,
and PBMA-g-PDMS membranes were measured by
a contact angle meter (Erma, Model G-I) at 25◦ C.
2.7. Degree-of-swelling of membrane
The PMMA-g-PDMS, PEMA-g-PDMS, and
PBMA-g-PDMS membranes, which were completely
dried under reduced pressure at room temperature
and weighed, were immersed into aqueous solutions
of 0.05 wt.% benzene and 0.02 wt.% chloroform in a
sealed vessel at 40◦ C. After the weight of the mem-
branes became constant, they were taken out of the
vessel, wiped quickly with filter paper and weighed.
The degree-of-swelling (DS) of the membrane was
determined from Eq. (2):
Ws
DS = (2)
Wd
where W s is the weight of the membrane swollen in
an aqueous solution of benzene or chloroform and W d
is the weight of the dried membrane.
2.8. Sorption selectivity
The composition of benzene or chloroform/water
absorbed by the PMMA-g-PDMS, PEMA-g-PDMS,
and PBMA-g-PDMS membranes was determined
by a previous method [18]. The compositions of
VOC/water mixtures in the membrane and in the feed
gave the sorption selectivity, αsorp.VOC/H2 O , expressed
in Eq. (3);
(WVOC /WH2 O )M
αsorp.VOC/H2 O = (3)
(WVOC /WH2 O )F
where (WVOC and WH2 O )F are the weight frac-
tions of VOC and water in the feed solution and
(WVOC and WH2 O )M are those in the membranes,
respectively.
2.9. Transmission electron micrographs (TEM)
The PMMA-g-PDMS, PEMA-g-PDMS, and
PBMA-g-PDMS membranes were vapor-stained with
an aqueous solution of 5 wt.% RuO4 . The stained
membranes were embedded in epoxy resin and sliced
into thin films (thickness approximately 50 nm) with a
microtome (Leica; Reichert Ultracut E). The structure
of these membranes was observed with a transmission
electron microscope (TEM) (JEOL JEM-1210) at an
accelerating voltage of 80 kV.
3. Results and discussion
3.1. Characterization of graft copolymer membranes
Table 1 summarizes the molecular weights and com-
positions of the PMMA-g-PDMS, PEMA-g-PDMS,
and PBMA-g-PDMS graft copolymers as deter-
mined by GPC and 1 H NMR, and the Tg ’s of their
 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269 259

membranes as determined by DSC. Number-average have microphase-separated structures. In PBMA-g-

molecular weights of PMMA-g-PDMS were in the
range of 50,000–75,000, those of PEMA-g-PDMS
were between 130,000 and 200,000, and those
of PBMA-g-PDMS were between 120,000 and
210,000. The ratio of the weight-average molecu-
lar weight to the number-average molecular weight
(M̄w /M̄n ) of PMMA-g-PDMS was greater than two
and that of PEMA-g-PDMS and PBMA-g-PDMS
was less than two. The molecular weight distribu-
tion of PBMA-g-PDMS was very narrow. Two Tg ’s
were observed at about 128 and −127◦ C in the
PMMA-g-PDMS membranes and at about 73 and
−120◦ C in the PEMA-g-PDMS membranes. The
observation of two Tg ’s in the PMMA-g-PDMS
and PEMA-g-PDMS membranes is attributable to
the fact that the Tg ’s of PMMA or PEMA and
PDMS homopolymers are about 128 or 73◦ C and
−120 ∼ −127◦ C, respectively. The higher Tg
is due to the MMA or EMA component and the
lower Tg to the DMS component. The presence of
these two Tg ’s implies that the PMMA-g-PDMS
and PEMA-g-PDMS membranes had a heteroge-
neous structure. Therefore, it is suggested that the
PMMA-g-PDMS and PEMA-g-PDMS membranes
PDMS membranes, however, Tg ’s based on the BMA
and DMS components were not observed. Such obser-
vations suggest that the PBMA-g-PDMS membranes
do not form the microphase separation structure and
that the BMA and DMS components may be misci-
ble. On the other hand, in the PEMA-g-PDMS and
PBMA-g-PDMS membranes a peak, which is not due
to Tg , appeared at about −39 ∼ −45◦ C on the DSC
curves of those membranes. This peak is not due to
the second transition caused by the Tg but rather is
due to the first transition as shown in Fig. 1.
3.2. Pervaporation characteristics of the graft
copolymer membranes
Fig. 2 shows the effect of DMS content on the
benzene concentration in the permeate and normal-
ized permeation rate for an aqueous solution of
0.05 wt.% benzene through the PMMA-g-PDMS,
PEMA-g-PDMS, and PBMA-g-PDMS membranes
by pervaporation. Because the permeation rate is sig-
nificantly influenced by the membrane thickness, it is
difficult to compare the permeability of membranes
of different thickness. Therefore, to evaluate the

Fig. 1. DSC curves of the PMMA-g-PDMS, PEMA-g-PDMS, and PBMA-g-PDMS membranes: (a) PMMA-g-PDMS (DMS content:
68 mol%); (b) PEMA-g-PDMS (DMS content: 75 mol%); (c) PBMA-g-PDMS (DMS content: 73 mol%).
260 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269
Fig. 2. Effect of the DMS content on the benzene concentration
in the permeate and normalized permeation rate for an aqueous
solution of 0.05 wt.% benzene through the PMMA-g-PDMS (䊊),
PEMA-g-PDMS (䊉), and PBMA-g-PDMS (䊐) membranes by
pervaporation.
permeability of the graft copolymer membranes,
we used the normalized permeation rates. All
of the graft copolymer membranes increased
benzene-permselectivity and permeability with in-
creasing DMS content. The benzene-permselectivity
and normalized permeation rate of the PMMA-g-
PDMS membrane increased dramatically at a DMS
content of more than about 40 mol%, and that of the
PEMA-g-PDMS membrane increased at a DMS con-
tent of more than about 70 mol%. On the other hand,
the benzene concentration in the permeate and the nor-
malized permeation rate through the PBMA-g-PDMS
Fig. 3. Effect of the DMS content on the chloroform concentration
in the permeate and normalized permeation rate for an aqueous
solution of 0.02 wt.% chloroform through the PMMA-g-PDMS
(䊊), PEMA-g-PDMS (䊉), and PBMA-g-PDMS (䊐) membranes
by pervaporation.
membranes were gradually enhanced with increas-
ing DMS content. The PBMA-g-PDMS membrane
did not show dramatic increases in its permselec-
tivity and permeability, differing from those in the
PMMA-g-PDMS and PEMA-g-PDMS membranes.
Fig. 3 shows the results of pervaporation for
an aqueous solution of 0.02 wt.% chloroform
through the graft copolymer membranes. Similar
to the results for an aqueous benzene solution, the
chloroform-permselectivity of PMMA-g-PDMS and
PEMA-g-PDMS membranes increased dramatically
at a DMS content of more than about 40 and 70 mol%,
 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269
respectively. It is noticeable that the benzene- and
chloroform-permselectivity, and the normalized per-
meation rates for the graft copolymer membranes
changed remarkably at the same DMS content.
These results suggest that the PMMA-g-PDMS and
PEMA-g-PDMS membrane structure changed at the
corresponding DMS content. On the other hand, the
PBMA-g-PDMS membrane showed a gradual increase
in chloroform-permselectivity and normalized per-
meation rate with increasing DMS content. From the
above results, it was found that the structural change
in the PBMA-g-PDMS membrane with variations in
DMS content was quite different from that of the
PMMA-g-PDMS and PEMA-g-PDMS membranes.
3.3. Chemical structure of the graft copolymer
membrane
In general, the permselectivity of a polymer mem-
brane is dependent on its chemical and physical struc-
tures, which influences the solubility of permeants into
the membrane and diffusivity of permeants through
a membrane. Therefore, we investigated the chemi-
cal and physical structures of the PMMA-g-PDMS,
PEMA-g-PDMS, and PBMA-g-PDMS membranes to
elucidate the permeation behavior.
The contact angles for water on the air-side sur-
face of the PMMA-g-PDMS, PEMA-g-PDMS and
PBMA-g-PDMS membranes as a function of the
DMS content are shown in Fig. 4. The contact angles
for water on the air-side surface of all graft copolymer
membranes increased with increasing DMS content.
A higher contact angle means that the surfaces of
the membranes have become more hydrophobic. The
high contact angle is attributable to the fact that the
hydrophobic PDMS component is localized to the
air-side surface. Therefore, it is expected that graft
copolymer membranes with a high DMS content have
a stronger affinity for benzene and chloroform than
those with a low DMS content due to their hydropho-
bic surface. As can be seen in Fig. 4, contact angles
for water on the air-side surface became nearly con-
stant at a DMS content of more than about 50 mol%.
It has been reported that the contact angle for wa-
ter of PDMS membrane was about 106–107◦ [19].
The observation in Fig. 4 suggests that the surfaces
of graft copolymer membranes with a DMS content
of more than 50 mol% were covered with a PDMS
261
Fig. 4. Relationship between the DMS content and the contact
angle for water on the air-side surface of the PMMA-g-PDMS
(䊊), PEMA-g-PDMS (䊉), and PBMA-g-PDMS (䊐) membranes.
component. Therefore, it is presumed that the air-side
surfaces of membranes with a DMS content of more
than 50 mol% have a constant affinity for benzene
and chloroform.
Fig. 5(a) and (b) show the effect of DMS content
on the benzene and chloroform concentrations incor-
porated into the PMMA-g-PDMS, PEMA-g-PDMS,
and PBMA-g-PDMS membranes in aqueous solu-
tions of 0.05 wt.% benzene and 0.02 wt.% chloroform.
The benzene and chloroform concentrations in the
PMMA-g-PDMS and PEMA-g-PDMS membranes
were highest at particular DMS contents. On the other
hand, the benzene and chloroform concentrations in
the PBMA-g-PDMS membrane decreased uniformly
with increasing DMS content. The difference in the
VOC concentrations in the graft copolymer mem-
branes is significantly dependent upon the affinities
of those graft copolymer membranes for benzene and
chloroform. To discuss the affinity of the membranes
for the permeants, we tried to calculate the difference
in the solubility parameter between VOCs and graft
copolymers.
The effect of DMS content on differences in the
solubility parameter (SP) between VOCs and the
graft copolymers is shown in Fig. 6. The difference in
SP, ∆, was calculated from the solubility parameters
of benzene (δ benzene ) and the membrane (δ membrane )
262 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269

Fig. 5. Effect of the DMS content on the benzene and chloroform concentration in the PMMA-g-PDMS (䊊), PEMA-g-PDMS (䊉), and
PBMA-g-PDMS (䊐) membranes for aqueous solutions of 0.05 wt.% benzene (a) and 0.02 wt.% chloroform (b) and the degree-of-swelling of
the PMMA-g-PDMS (䊊), PEMA-g-PDMS (䊉), and PBMA-g-PDMS (䊐) membranes in 0.05 wt.% benzene (c) and 0.02 wt.% chloroform
(d) solutions at 40◦ C.

using Eq. (4):

∆ = |δbenzene − δmembrane | (4) Table 2

Hansen solubility parameters of solvents and polymers
The solubility parameter, δ membrane , of each graft
Material Solubility parameters ((MPa)1/2 )
copolymer membrane was calculated from the solubil-
ity parameters and volume ratios of poly(alkylmeth- δd δp δh δ
acrylate) and PDMS. The solubility parameters of H2 O 15.5 16.0 42.4 47.9
homopolymers, water and benzene are summarized Benzene 18.4 0 2.0 18.6
in Table 2. Small ∆ means that the membrane PMMA 18.8 10.2 8.6 23.1
has a strong affinity for benzene. The ∆s of the PEMA 18.8 10.2 4.3 22.1
PBMA – – – 17.8
PMMA-g-PDMS and PEMA-g-PDMS membranes
PDMS 15.9 0.10 4.7 16.5
showed a minimum at a DMS content of about 50 and
 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269
Fig. 6. Effect of the DMS content on the difference in solu-
bility parameter between benzene and the PMMA-g-PDMS (䊊),
PEMA-g-PDMS (䊉), and PBMA-g-PDMS (䊐) membranes.
70 mol%, respectively. These results support the sug-
gestion that the affinity of the PMMA-g-PDMS and
PEMA-g-PDMS membranes for benzene is strongest
at about 50 and 70 mol% DMS content, respectively.
On the other hand, the ∆ of the PBMA-g-PDMS
membrane increased uniformly with increasing DMS
content. Changes in the ∆ of the PMMA-g-PDMS,
PEMA-g-PDMS, and PBMA-g-PDMS membranes
with changes in DMS content correspond to changes
in the benzene concentration incorporated into their
graft copolymer membranes, as shown in Fig. 5(a).
These results strongly suggest that the concentration
of VOCs incorporated into the membranes is signif-
icantly dependent on the affinity between the VOCs
and the membranes.
In Fig. 5(a) and (b) it is obvious that is a huge differ-
ence between PEMA-g-PDMS and PBMA-g-PDMS
at low DMS content. It seems that this does not cor-
relate with the measurements of contact angles in
Fig. 4. However, it may be able to explain as follows.
The measurement of contact angles gives information
on only the membrane surface. On the other hand,
the liquid composition adsorbed into the membrane
is governed by both the affinity of permeants to the
membrane and the diffusion of permeants in the
membrane matrix. As mentioned below a huge dif-
ference between the morphology of PEMA-g-PDMS
263
and PBMA-g-PDMS membrane was observed. The
difference in such membrane structure may give a
big difference in the liquid composition in the graft
copolymer membrane. We will continuously investi-
gate to make these results clear.
3.4. Physical structure of the graft copolymer
membrane
Fig. 5(c) and (d) show the DS of the PMMA-g-
PDMS, PEMA-g-PDMS, and PBMA-g-PDMS mem-
branes in aqueous solutions of 0.05 wt.% benzene
and 0.02 wt.% chloroform at 40◦ C. The degrees of
swelling of all graft copolymer membranes were
very low, regardless of the DMS content. Gener-
ally, the polymer membranes having a strong affinity
for permeants become swollen in the feed solution
[20]. Such swelling of polymer membranes causes a
change in membrane structure during pervaporation.
The graft copolymer membranes used in this study,
however, have a high affinity for benzene and chloro-
form but a very low affinity for water. Therefore, as
aqueous solutions of dilute benzene and chloroform
were used as feed solutions in this study, the degrees
of swelling of the graft copolymer membranes were
very low. These facts suggest that membrane struc-
ture did not change during pervaporation. As can be
seen in changes in Fig. 5, changes in the DS of the
graft copolymer membranes with the DMS content
were very small. Therefore, increases in the permse-
lectivity and permeability with an increasing of DMS
content in the graft copolymer membranes may be
independent of swelling behavior.
The effect of DMS content on the density of the
graft copolymer membranes is shown in Fig. 7. The
density of the PMMA-g-PDMS and PEMA-g-PDMS
membranes decreased dramatically at a DMS content
of about 40 and 70 mol%, respectively, and that of
the PBMA-g-PDMS membrane decreased uniformly
with increased DMS content. In general, the degree
of packing of the molecules in a membrane strongly
influences the permeation rate. For example, when
the degree of packing decreases, the permeation rate
increases [21]. The normalized permeation rate of the
PMMA-g-PDMS and PEMA-g-PDMS membranes
increased dramatically at the corresponding DMS
content as their density decreased. As the density of
the PBMA-g-PDMS membranes decreased uniformly,
264 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269
Fig. 7. Effect of the DMS content on the density of the
PMMA-g-PDMS (䊊), PEMA-g-PDMS (䊉), and PBMA-g-PDMS
(䊐) membranes.
the permeation rate gradually increased as DMS con-
tent increased. Changes in membrane density with
changes in DMS content correspond to changes in both
benzene-permeability and chloroform-permeability as
shown in Figs. 2 and 3. These results suggest that the
density of the graft copolymer membranes strongly
governs their permeability.
In order to discuss membrane structure in more
detail, the morphologies of the graft copolymer mem-
branes were observed by TEM. Fig. 8 shows the TEM
images for cross-sections of the PMMA-g-PDMS,
PEMA-g-PDMS, and PBMA-g-PDMS membranes.
The dark region represents the PDMS compo-
nent stained by RuO4 . In the PMMA-g-PDMS and
PEMA-g-PDMS membranes, distinct microphase
separation was observed, but PBMA-g-PDMS mem-
branes shown no microphase separation. The TEM
observations agree with the results of Tg ’s in the
DSC measurements of the graft copolymer mem-
branes. Furthermore, it seemed that the PDMS com-
ponent formed a continuous phase at a DMS content
of more than 52 and 72 mol% in PMMA-g-PDMS
and PEMA-g-PDMS membranes, respectively.
As shown in Figs. 2 and 3, the benzene- and
chloroform-permselectivity of the PMMA-g-PDMS
and PEMA-g-PDMS membranes was enhanced dra-
matically at a DMS content of more than about
50 and 70 mol%, respectively, but the benzene-
and chloroform-permselectivity of PBMA-g-PDMS
membranes was enhanced gradually with increas-
ing DMS content. These results suggest that the
morphology of microphase separation in the graft
copolymer membranes governs their benzene- and
chloroform-permselectivity significantly. Because
benzene and chloroform are mainly permeated through
the continuous poly(alkylmethacrylate) phases in the
PMMA-g-PDMS membranes with less than about
50 mol% DMS and PEMA-g-PDMS membranes with
less than about 70 mol% DMS, the diffusivity of
benzene and chloroform molecules, which are larger
in molecular size, in these graft copolymer mem-
branes is much lower than the diffusivity of water
molecules. After the PDMS component forms a con-
tinuous phase as DMS content increases, the such
graft copolymer membranes become more benzene-
and chloroform-permselective due to the enhanced
diffusivity of benzene and chloroform molecules in
the continuous PDMS phase. This is the reason why
the benzene- and chloroform-permselectivity of the
graft copolymer membranes was strongly enhanced
with increased DMS content.
3.5. Mechanism of permselectivity in the graft
copolymer membranes
In general, the separation and permeation charac-
teristics for organic liquid mixtures through polymer
membranes by pervaporation are based on the solubil-
ity of the permeants into the polymer membrane (sorp-
tion process) and the diffusivity of the permeants in
the polymer membrane (diffusion process) [9,22–24].
It is very important to determine the sorption selectiv-
ity and diffusion selectivity to elucidate the separation
mechanism of aqueous benzene and chloroform solu-
tions through the graft copolymer membranes. In gen-
eral, the separation factor for mixtures of VOCs and
water in pervaporation was defined by Eq. (5) [25].
αVOC/H2 O =αsorp.VOC/H2 O · αdiff.VOC/H2 O · αevap. (5)
where α evap. is defined as follows:


∗
PVOC γVOC
αevap. = ∗
PH2 O γH2 O
 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269 265

Fig. 8. TEM micrographs of the PMMA-g-PDMS (a), PEMA-g-PDMS (b), and PBMA-g-PDMS (c) membranes.

where PVOC∗ and PH∗2 O are vapor pressures for VOC aqueous solution of 0.05 wt.% benzene is approx-
and H2 O, respectively, and γ VOC and γH2 O are their imately 29.3 (see Appendix A), the αdiff.VOC/H2 O
activity coefficients. In Eq. (5), since the αsorp.VOC/H2 O can be calculated from the overall separation factor,
can be determined by Eq. (3) and the α evap. for an αVOC/H2 O , αsorp.VOC/H2 O , and α evap. by rearranging
266 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269

Eq. (5). Even though the αdiff.Bz/H2 O for all graft

copolymer membranes is less than unity (for example
at ∼68 mol% DMS in Fig. 9 for the PMMA-g-PDMS
point, αBz/H2 O and αsorp.Bz/H2 O are about 3730 and
1055, respectively, so αevap. · αdiff.Bz/H2 O = 3.54, thus
αdiff.Bz/H2 O = 3.54/29.3 = 0.121 (see Appendix A)),
due to the favorable α evap. , a good αevap. · αdiff.Bz/H2 O
product still results.
In Fig. 9, the αBz/H2 O , αsorp.Bz/H2 O , and αevap. ·
αdiff.Bz/H2 O for an aqueous solution of 0.05 wt.%
benzene as a function of the DMS content in the
PMMA-g-PDMS, PEMA-g-PDMS, and PBMA-g-
PDMS membranes is shown. In this figure, the
αdiff.Bz/H2 O of the PMMA-g-PDMS and PEMA-g-
PDMS membranes increased drastically at DMS con-
tents of about 40 and 70 mol%, respectively. These
observations suggest that the relative diffusivity of
benzene in the PMMA-g-PDMS and PEMA-g-PDMS
membranes increased at the corresponding DMS
contents. The PMMA-g-PDMS and PEMA-g-PDMS
membranes with low DMS content showed low dif-
fusion selectivity for benzene. This is attributable
to the larger molecular size of benzene than water.
In addition, the DMS content at which αdiff.Bz/H2 O
of the PMMA-g-PDMS and PEMA-g-PDMS mem-
branes changed drastically agreed with the DMS
content at which the PDMS component changed
from a discontinuous phase to a continuous phase.
This indicates that morphological changes in the mi-
crophase separation enhanced the relative diffusivity
of benzene in their membranes. On the other hand,
the αdiff.Bz/H2 O of the PBMA-g-PDMS membrane
increased monotonously with increasing DMS con-
tent. The monotonous increase in the αdiff.Bz/H2 O of
the PBMA-g-PDMS membrane is due to the fact that
the PDMS component with low Tg increased in the
membrane, and no microphase separation occurred.
Collectively these observations suggest that the
morphology of the graft copolymer membranes
significantly influences membrane performance in
pervaporation of aqueous VOC solutions. Namely,
the benzene and chloroform-permselectivity of the
PMMA-g-PDMS and PEMA-g-PDMS membranes
increased dramatically at a particular DMS content,
and that of the PBMA-g-PDMS increased gradu- Fig. 9. Effect of the DMS content in the PMMA-g-PDMS (䊊),
PEMA-g-PDMS (䊉), and PBMA-g-PDMS (䊐) membranes on
ally with DMS content. The PMMA-g-PDMS and
their separation factor (a), sorption selectivity (b), and diffusion
PEMA-g-PDMS membranes, in particular, showed a selectivity (c) for an aqueous solution of 0.05 wt.% benzene.
higher benzene- and chloroform-permselectivity as
 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269
the PDMS component changed from a discontinuous
phase to a continuous phase.
3.6. Permeation mechanism
The permeation mechanism of aqueous VOCs solu-
tions through the graft copolymer membranes can be
explained by a tentative model shown in Fig. 10. Ben-
zene and chloroform are more preferentially incorpo-
rated into the graft copolymer membranes with every
DMS contents, because PMMA, PEMA, PBMA, and
PDMS have a much stronger affinity for benzene and
chloroform than water. Since the surfaces of the graft
copolymer membranes with a DMS content of more
than 50 mol% are covered with the PDMS compo-
nent, the concentrations of benzene and chloroform
on the surfaces of the graft copolymer membranes are
Fig. 10. Tentative model for separation process (a) and diffusion process (b) of dilute aqueous VOCs solutions through the PMMA-g-PDMS,
PEMA-g-PDMS, and PBMA-g-PDMS membranes.
267
saturated at a DMS content of more than 50 mol%.
However, the benzene- and chloroform-permselectivity
of the PMMA-g-PDMS and PEMA-g-PDMS mem-
branes changed dramatically at DMS contents of about
40 and 70 mol%, respectively. This is attributable to a
drastic change in the diffusivity of permeants based on
structural changes in the graft copolymer membranes.
The TEM observation demonstrated that the selectiv-
ity and the permeability of the PMMA-g-PDMS and
PEMA-g-PDMS membranes are strongly enhanced
as the PDMS component in the PMMA-g-PDMS
and PEMA-g-PDMS membranes begins to form a
continuous phase. Our previous studies have made
it clear that graft copolymer membranes for the sep-
aration of aqueous ethanol solutions changed from
water-permselective to ethanol-permselective at the
DMS content at which their microphase-separated
268 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269
structure changed from a discontinuous PDMS phase
to a continuous PDMS phase [13–17]. The pervapo-
ration results of the graft copolymer membranes for
aqueous VOC solutions in this study are supported
by similar observations in previous studies. Table 2
suggests that the PDMS component has a stronger
affinity for benzene than the PMMA or PEMA com-
ponents. Moreover, as the Tg of PDMS is lower
than that of either PMMA or PEMA, it is presumed
that the permeants diffuse more easily in the PDMS
phase than in the PMMA or PEMA phase [13–17].
When the PDMS component in their graft copolymer
membranes forms a continuous phase, benzene and
chloroform are preferentially incorporated into a con-
tinuous PDMS phase with a high affinity for benzene
and chloroform. Consequently, the high benzene and
chloroform-permselectivity for the PMMA-g-PDMS
and PEMA-g-PDMS membranes is significantly
dependent on both the sorption of benzene and
chloroform into a continuous PDMS phase and the
diffusivity of these molecules in this PDMS phase.
Although the PBMA-g-PDMS membranes showed
benzene- and chloroform-permselectivity, they did
not demonstrate a microphase separation. As a result,
the permeation and separation characteristics of the
PBMA-g-PDMS membranes were quite different from
those of the PMMA-g-PDMS and PEMA-g-PDMS
membranes. This study revealed that the morphology
of multicomponent polymer membranes has a strong
influence on their permeability and permselectivity.
4. Conclusions
The graft copolymer membranes consisting of
PMMA, PEMA, and PBMA with PDMS showed
benzene and chloroform-permselectivity in the sepa-
ration of aqueous solutions containing a very small
amount of benzene and chloroform. Benzene- and
chloroform-permselectivity was enhanced with in-
creasing DMS content. In particular, the permselec-
tivity of the PMMA-g-PDMS and PEMA-g-PDMS
membranes changed drastically at DMS con-
tents of 40 and 70 mol%, respectively. Investiga-
tions into the chemical and physical structure of
these graft copolymer membranes revealed that
their permselectivity was significantly influenced
by the microphase-separated structure. When the
PDMS component in the graft copolymer mem-
branes with microphase separation formed a con-
tinuous phase, the membranes showed a high
benzene- and chloroform-permselectivity. The high
benzene and chloroform-permselectivity of the
microphase-separated membranes was due to high
diffusivity of benzene and chloroform in the continu-
ous PDMS phase.
Acknowledgements
This work was supported by a Grant-in-Aid for Sci-
entific Research (C) (no. 12650892) from the Ministry
of Education, Science, Sports and Culture, Japan.
Appendix A
We took the azeotropic composition at 69◦ C for
benzene/water and fit the Carlson and Colburn modi-
fication to the Van Laar equation:
A xH2 2 O
log γBz =  2
(A /B  )xBz + xH2 O
and
B  xBz
2
log γBz =  2
(B  /A )xH2 O + xBz
for A = 0.946 and B  = 1.48 as effectively constant
values over the range 69 → 25◦ C, where A and B
are constants. γ Bz and γH2 O are activity coefficients
for benzene and water, respectively. xBz and xH2 O are
mole fraction for them.
∗ and that of wa-
The vapor pressure of benzene, PBz
∗
ter, PH2 O are 182 and 54.8 mmHg, respectively, and
a mole fraction of benzene for the weight fraction of
0.0005 is 0.000115. The γ Bz and γH2 O at that com-
position can be estimated by the above numbers as
follows:
γBz = 8.828
γH2 O = 1.00017
These numbers seem reasonable, since benzene
does not like water and is essentially at “infinite
dilution” where log γ Bz = A (1.0), or γ Bz = 8.83.
 T. Uragami et al. / Journal of Membrane Science 187 (2001) 255–269
The α evap. in Eq. (5) can be calculated as follows:
∗ /γ
PVOC VOC ∼ 182(8.83)
αevap. = = = 29.3
PH∗2 O /γH2 O 54.8(1.0001)
References
[1] R.W. Baker, in: R.W. Baker, E.L. Cussler, W. Eykamp,
W.J. Koros, R.L. Riley, H. Strathmann (Eds.), Membrane
Separation Systems, Recent Developments and Future
Directions, Noyes Data Corp., Park Ridge, NJ, 1991, p. 151.
[2] R. Psaume, Ph. Aptel, Y. Aurelle, J.C. Mora, J.L. Bersillon,
Pervaporation: importance of concentration polarization in the
extraction of trace organics from water, J. Membr. Sci. 36
(1988) 373–384.
[3] I. Blume, J.G. Wijmans, R.W. Baker, The separation of
dissolved organics from water by pervaporation, J. Membr.
Sci. 49 (1990) 253–286.
[4] Y. Fang, V.A. Pham, T. Matsuura, J.P. Santerre, R.M. Narbaitz,
Effect of surface-modifying macromolecules and solvent
evaporation time on the performance of polyethersulfone
membranes for the separation of chloroform/water mixtures
by pervaporation, J. Appl. Polym. Sci. 54 (1994) 1937–1943.
[5] D. Yang, S. Majumdar, S. Kovenklioglu, K.K. Sirkar, Hollow
fiber contained liquid membrane pervaporation system for
the removal of toxic volatile organics from waste water, J.
Membr. Sci. 103 (1995) 195–210.
[6] W.W.Y. Lau, J. Finlayson, J.M. Dickson, J. Jiang, M.A.
Brook, Pervaporation performance of oligosilylstyrene-poly-
dimethylsiloxane membrane for separation of organics from
water, J. Membr. Sci. 134 (1997) 209–217.
[7] T. Miyata, J. Higuchi, H. Okuno, T. Uragami, Preparation
of polydimethylsiloxane/polystyrene interpenetrating polymer
network membranes and permeation of aqueous ethanol
solutions though the membranes by pervaporation, J. Appl.
Polym. Sci. 61 (1996) 1315–1324.
[8] T. Miyata, Y. Nakanishi, T. Uragami, Ethanol permselectivity
of poly(dimethylsiloxane) membranes controlled by
simple surface modifications using polymer additives,
Macromolecules 30 (18) (1997) 5563–5565.
[9] T. Uragami, K. Tukamoto, K. Inui, T. Miyata,
Pervaporation characteristics of a benzoylchitosan membrane
for benzene–cyclohexane mixtures, Macromol. Chem. Phys.
199 (1998) 49–54.
[10] K. Inui, H. Okumura, T. Miyata, T. Uragami, Permeation
and separation of benzene/cyclohexane mixtures through
cross-linked poly(alkyl methacrylate) membranes, J. Membr.
Sci. 132 (1997) 193–202.
[11] P.R. Brooks, A.G. Livingston, Aqueous–aqueous extraction
organic pollutants through tubular silicone rubber membranes,
J. Membr. Sci. 104 (1995) 119–137.
[12] C. Dotremont, B. Brabants, K. Geeroms, J. Mewis,
C. Vandecasteele, Sorption and diffusion of chlorinated
hydrocarbons in silicalite-filled PDMS membranes, J. Membr.
Sci. 104 (1995) 109–117.
269
[13] S. Mishima, T. Nakagawa, The characterization of
UV modified poly(dimethylsiloxane) membrane and their
permselectivity for chlorinated hydrocarbons, Kobunshi
Ronbunshu 54 (6) (1997) 375–383.
[14] S. Schnabel, P. Moulin, Q.T. Nguyen, D. Roizard, P. Aptel,
Removal of volatile organic components (VOCs) from water
by pervaporation: separation improvement by Dean vortices,
J. Membr. Sci. 142 (1998) 129–141.
[15] A. Shimazu, K. Tani, M. Nasu, M. Kondo, Removal of
halogenated organic compounds from drinking water by
pervaporation, Kankyogakkaishi 6 (1993) 1–7.
[16] M. Hoshi, M. Kobayashi, T. Saitoh, A. Higuchi, T. Nakagawa,
Separation of organic solvent from dilute aqueous solutions
and from organic solvent mixtures through cross-linked
acrylate copolymer membranes by pervaporation, J. Appl.
Polym. Sci. 69 (1998) 1483.
[17] T. Miyata, T. Takagi, T. Kadota, T. Uragami, Characteristics
of permeation and separation for aqueous ethanol
solutions through methyl methacrylate-dimethylsiloxane graft
copolymer membranes, Macromol. Chem. Phys. 196 (1995)
1211–1220.
[18] T. Miyata, T. Takagi, T. Uragami, Microphase separation
in graft copolymer membranes with pendant oligodimethyl-
siloxanes and their permselectivity for aqueous ethanol
solutions, Macromolecules 29 (1996) 7787–7794.
[19] T. Miyata, J. Higuchi, H. Okuno, T. Uragami, Preparation
of polydimethylsiloxane/polystyrene interpenetrating polymer
network membranes and permeation of aqueous ethanol
solutions through the membranes by pervaporation, J. Appl.
Polym. Sci. 61 (1996) 1315–1324.
[20] T. Miyata, S. Obata, T. Uragami, Morphological effects of
microphase separation on the permselectivity for aqueous
ethanol solutions of block and graft copolymer membranes
containing poly(dimethylsiloxane), Macromolecules 32
(1999) 3712–3720.
[21] T. Miyata, S. Obata, T. Uragami, Annealing effect of
microphase-separated membranes containing poly(dimethyl-
siloxane) on their permselectivity for aqueous ethanol
solutions, Macromolecules 32 (1999) 8465–8475.
[22] K. Inui, K. Okazaki, T. Miyata, T. Uragami, Effect of
mesogenic groups on characteristics of permeation and
separation for benzene/cyclohexane mixtures of side-chain
liquid-crystalline polymer membranes, J. Membr. Sci. 143
(1998) 98–104.
[23] T. Uragami, M. Saito, Analysis of permeation and separation
characteristics and new technique for separation of aqueous
alcoholic solutions through alginic acid membranes, Sep. Sci.
Technol. 24 (1989) 541–554.
[24] T. Uragami, M. Saito, K. Takigawa, Studies on syntheses and
permeabilities of special polymer membranes, 69. Comparison
of permeation and separation characteristics for aqueous
alcholic solutions by pervaporation and new evapomeation
methods through chitosan membranes, Macromol. Chem.
Rapid Commun. 9 (1988) 361–365.
[25] J.G. Wijmans, R.W. Baker, A simple predictive treatment of
the permeation process in pervaporation, J. Membr. Sci. 79
(1993) 101–113.