A novel composite nanofiltration (NF) membrane prepared from graft copolymer of trimethylallyl ammonium chloride onto chitosan (GCTACC)/poly(acrylonitrile) (PAN) by epichlorohydrin cross-linking

Article abstract of DOI:10.1016/j.carres.2006.09.011

A novel composite nanofiltration (NF) membrane was prepared by over-coating the PAN ultrafiltration (UF) membrane with a GCTACC thin layer. The effects of membrane preparation techniques and operating conditions on the performance of the composite membrane were studied. The results indicate that a composite NF membrane from 1.0 wt % GCTACC casting solution, vaporized for 2 h at 50 °C, cross-linked for 20 h at 50 °C and pH ≈ 12 with ethanol/epichlorohydrin (50/0.45 wt/wt) had optimum performance. The resultant GCTACC/PAN composite membrane was positively charged. Scanning electron microscopy showed its asymmetric and composite features. At 25 °C and 30 L/h of cycling flow, the permeability of pure water through this membrane is 6.3 L/h m² MPa. At 25 °C, 1.2 MPa and 30 L/h of cycling flow, the rejection of 1000 mg/L MgCl₂, CaCl₂, MgSO₄, Na₂SO₄, and NaCl solutions is 0.976, 0.972, 0.897, 0.65, and 0.407, respectively, with fluxes of 6.8, 6.12, 6.12, 5.57, and 5.51 L/h m², respectively. The order of rejection of different salts follows the decreasing order of MgCl₂, CaCl₂, MgSO₄, NaCl, KCl, Na₂SO₄, and K₂SO₄, which reveals the characteristics of the positively charged NF membrane. In addition, the curve for the streaming potential also illustrates the positively charged characteristics of this membrane, with a pressure osmotic coefficient of 11.7 mV MPa^-1.

Full text of DOI:10.1016/j.carres.2006.09.011

Carbohydrate Research 341 (2006) 2777–2784
A novel composite nanofiltration (NF) membrane prepared
from graft copolymer of trimethylallyl ammonium chloride onto
chitosan (GCTACC)/poly(acrylonitrile) (PAN) by
epichlorohydrin cross-linking
Ruihua Huang, Guohua Chen,* Mingkun Sun and Congjie Gao
School of Chemistry and Chemical Engineering, Ocean University of China, Qingdao 266003, China
Received 19 May 2006; received in revised form 2 September 2006; accepted 17 September 2006
Available online 11 October 2006
Abstract—A novel composite nanofiltration (NF) membrane was prepared by over-coating the PAN ultrafiltration (UF) membrane
with a GCTACC thin layer. The effects of membrane preparation techniques and operating conditions on the performance of the
composite membrane were studied. The results indicate that a composite NF membrane from 1.0 wt % GCTACC casting solution,
vaporized for 2 h at 50 ꢀC, cross-linked for 20 h at 50 ꢀC and pH ꢀ 12 with ethanol/epichlorohydrin (50/0.45 wt/wt) had optimum
performance. The resultant GCTACC/PAN composite membrane was positively charged. Scanning electron microscopy showed its
asymmetric and composite features. At 25 ꢀC and 30 L/h of cycling flow, the permeability of pure water through this membrane is
6.3 L/h m2 MPa. At 25 ꢀC, 1.2 MPa and 30 L/h of cycling flow, the rejection of 1000 mg/L MgCl2, CaCl2, MgSO4, Na2SO4, and
NaCl solutions is 0.976, 0.972, 0.897, 0.65, and 0.407, respectively, with fluxes of 6.8, 6.12, 6.12, 5.57, and 5.51 L/h m2, respectively.
The order of rejection of different salts follows the decreasing order of MgCl2, CaCl2, MgSO4, NaCl, KCl, Na2SO4, and K2SO4,
which reveals the characteristics of the positively charged NF membrane. In addition, the curve for the streaming potential also
illustrates the positively charged characteristics of this membrane, with a pressure osmotic coefficient of 11.7 mV MPaꢁ1
.
ꢁ 2006 Elsevier Ltd. All rights reserved.
Keywords: Graft copolymer of trimethylallyl ammonium chloride onto chitosan; Epichlorohydrin; Positively charged nanofiltration membrane;
Membrane preparation
1. Introduction
Chitosan is readily obtained by deacetylation of chi-
tin, which is abundant in nature, and it shows good
membrane forming characteristics. Tan et al.10 reported
the preparation of NF membranes by ultraviolet irradi-
ation using poly(acrylonitrile) (PAN) membranes as
support and chitosan as a modifier. The resultant mem-
brane was used in the treatment of E. globulus in waste-
water from BCHMP (blue-gum chemical hotgrinding
machine plasm) and it was found that the retention to
soda increased from 2.5% to 40.1% and the rejections
to chroma, COD (chemical oxygen demand) and BOD
(biologic oxygen demand) also increased greatly in com-
parison with a PAN matrix membrane. Musale and Ku-
mar11 studied the effect of surface cross-linking on
sieving characterization of a chitosan/PAN compo-
site NF membrane, and found that the cross-linked
Nanofiltration (NF) membranes have been recognized for
having properties in between those of ultrafiltration (UF)
and reverse osmosis (RO) and thus have found applications
in many areas including water softening and removal of
natural matter, treatment of waste water, etc.1–6
At present, internationally developed NF membranes
are generally neutral or negatively charged.7,8 There are
a few studies on positively charged NF membranes. In
fact, a positively charged membrane is actually needed9
in some cases such as for the retention of multivalent cat-
ions and the recovery of cathode electrophoresis lacquer.
*
Corresponding author. Tel.: +86 532 82032223; e-mail:
huangrh20022002@yahoo.com.cn
0008-6215/$ - see front matter ꢁ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carres.2006.09.011

2778
R. Huang et al. / Carbohydrate Research 341 (2006) 2777–2784
membranes were stable over a 10-h operating period for
pure water permeation and the stability increased with
increasing glutaraldehyde concentration. But chitosan
only dissolves in dilute acid, not in water, so that the
application of chitosan was limited to some extent.
Quaternization is applied to modify chitosan to in-
crease the hydrophilicity and ionic character, and the
chitosan derivative, quaternized chitosan, is obtained
by grafting trimethylallyl ammonium chloride onto
chitosan. It is used as a novel film-forming material
for the NF membrane. In this study, a composite NF
membrane was prepared using quaternized chitosan as
the film-forming material and epichlorohydrin as the
cross-linking reagent. The effects of preparation tech-
niques and operation factors on membrane properties
have been investigated.
2.2.2. Preparation of graft copolymer of trimethylallyl
ammonium chloride onto chitosan (GCTACC). Two
grams of chitosan was dispersed in 85 g of 5% HAC solu-
tion at rt. After chitosan had dissolved completely, oxy-
gen was removed for 30 min in order to avoid side
reaction, then the temperature rose to 60 ꢀC. Subse-
quently, 1.0 g ammonium ceric nitrate and 23 g 15 wt %
aq trimethylallyl ammonium chloride were added to the
chitosan solution with stirring. The reaction was carried
out with stirring for 9.5 h at 60 ꢀC. The residue was
poured into ethanol, and the precipitate (graft copolymer)
was filtered, washed with ethanol, and dried at 60 ꢀC.
2.3. Preparation of GCTACC/PAN composite NF
membranes
The casting solution was prepared by dissolving a cer-
tain amount of GCTACC in deionized water. After
the solution was filtered with a G3 sand filter, it was
over-coated on a surface-dried PAN UF membrane.
Then the composite membrane was vaporized at a cer-
tain temperature, and cross-linked with epichlorohydrin
in ethanol at a certain pH in an airtight container. The
cross-linked membrane was heat-treated for 5 min at
50 ꢀC again, then washed thoroughly with deionized
water and immersed in deionized water for 24 h.
2. Experimental
2.1. Materials and apparatus
Chitosan [MW = 5.4 · 105 Da, degree of deacetylation
(DD = 90%)] was purchased from Haihui Bioengineer-
ing (Qingdao, China); trimethylamine alcohol solution,
allyl chloride, ammonium ceric nitrate, acetic acid,
NaCl, KCl, MgCl2, CaCl2, Na2SO4 were of analytical
grade; quaternized chitosan was prepared by a modi-
fied method according to the literature;12 PAN UF
membrane with MWCO of 1.0 · 105 Da was provided
by the Development Center of Water Treatment Tech-
nology, State Oceanic Administration (Hangzhou,
China).
2.4. Permeation experiments
The flux of the membrane was determined by the weight
of the permeated fluid through the membrane during a
certain period of time and calculated with the following
equation:
Apparatus: A model DDS-11A conductivity meter
(Shanghai Leici Instrument, China) was used; mem-
brane evaluation apparatus was provided by the Devel-
opment Center of Water Treatment Technology, State
Oceanic Administration (Hangzhou, China); a JEOL
JMS-840 scanning electron microscope was applied to
investigate the structure characteristic of the membrane
prepared.
F ¼ W =A ꢂ t
ð1Þ
where F is the flux, A is the effective area of the mem-
brane; t is the time for permeation and W is the weight
of the permeating fluid passing through the membrane.
Rejection was calculated with the following equation:
R ¼ 1 ꢁ Cp=Cm
ð2Þ
where Cp and Cm are the concentrations of the perme-
ated fluid and the feed concentration, respectively. In this
study, salt concentrations are replaced by the conductiv-
ity of salt solutions. Two membrane disks were cut from
the same sheet, and the data presented are the averages of
these measurements with a standard deviation of 5%.
2.2. Preparation of graft copolymer of trimethylallyl
ammonium chloride onto chitosan (GCTACC)
2.2.1. Synthesis of trimethylallyl ammonium chloride
(monomer). Twenty milliliter trimethylamine alcohol
solution was placed in a 250-mL flask with three
mouths, then 10 mL allyl chloride was dripped into the
above solution with stirring at rt. After the addition,
the temperature rose to 40 ꢀC and the reaction pro-
ceeded with stirring for 12 h at this temperature under
circumfluence. Finally, the residue was poured into ace-
tone, and the precipitate (trimethylallyl ammonium
chloride) was filtered, washed with acetone, and dried
at 40 ꢀC.
3. Results and discussions
3.1. Effect of preparation conditions on the rejection
and flux of GCTACC/PAN composite membranes
The performance tests for the prepared membrane were
carried out at 25 ꢀC and 0.40 MPa and the concentra-

R. Huang et al. / Carbohydrate Research 341 (2006) 2777–2784
2779
tion of feed was 1000 mg/L MgCl2 without specification.
The recirculation rate of feed was kept at 30 L/h. Due to
operating at rather low pressure, the concentration
polarization through the membrane test cell may be
minimized and ignored. The effective membrane area
was 19.6 cm2, and the membrane was pre-pressurized
at 0.4 MPa for 0.5 h before testing. Both the retentate
and the permeate were re-circulated to the feed tank to
maintain a constant feed concentration during the per-
meation experiments.
together, a mass ratio of ethanol to epichlorohydrin of
50:0.45 is the optimum ratio.
3.1.2. Effect of evaporation time on the rejection and flux
of GCTACC/PAN composite membranes. As noted
earlier in the membrane preparation, the membranes
are allowed to vaporize solvent to form a dense skin
layer. To ensure that the GCTACC casting solution is
vaporized completely at 50 ꢀC, it needs at least 2 h. As
shown in Figure 2, obviously both rejection and flux
decrease with prolongation of evaporation time. In fact,
the evaporation process of GCTACC solution is also a
conversion process from sol to gel, and the molecules
of polymers rearrange continually with the evaporation
of solvent. When the arrangement of molecules is in an
optimal state, the membrane prepared will show a high
rejection and allow moderate flux. According to the
results obtained, a curing time of 2 h at 50 ꢀC was
adopted for further experiments.
3.1.1. Effect of epichlorohydrin concentration on the
rejection and flux of GCTACC/PAN composite NF
membranes. Epichlorohydrin was chosen as the cross-
linking reagent because it could react with not only
–OH, but also –NH2. After the reaction had taken place,
the hydrophile –OH was introduced, which strengthened
the hydrophilicity of the membrane. In addition, the
chains of 2-OH-1,3 propylene introduced were compar-
atively short, which was beneficial for the formation of
the membrane with pore sizes suitable for NF. To inves-
tigate the effect of epichlorohydrin concentration, the
same preparation technique discussed in Section 2.3
was used and a series of GCTACC/PAN composite
membranes with 1.0 wt % casting solution were pre-
pared by changing the weight ratio of ethanol to epi-
chlorohydrin from 50:0.17 to 50:0.75.
As shown in Figure 1, rejection increases with the in-
crease in epichlorohydrin concentration until the mass
ratio of ethanol to epichlorohydrin is 50:0.45, above
which ratio rejection virtually remains constant. How-
ever, the flux decreases all of the time. This phenomenon
may be explained as follows: when the epichlorohydrin
concentration increases, the cross-linking reaction tends
to be complete, which makes the membrane surface less
permeable. Thus, the rejection increases and the flux
decreases. Considering the flux and rejection of MgCl2
Figure 2. Effect of evaporation time on the rejection and flux of
GCTACC/PAN composite membranes.
Figure 1. Effect of epichlorohydrin concentration on the rejection and flux of GCTACC/PAN composite membranes.

2780
R. Huang et al. / Carbohydrate Research 341 (2006) 2777–2784
3.1.3. Effect of different vaporizing methods on the
rejection and flux of GCTACC/PAN composite mem-
branes. To investigate the effect of the different vapor-
izing methods on GCTACC/PAN composite membrane
properties, a series of GCTACC/PAN composite mem-
branes was prepared by vaporizing at 50 ꢀC for 2 h, at
60 ꢀC for 1 h and at 18 ꢀC for 18 h, respectively. The
results are shown in Table 1. Obviously, vaporizing for
a long time at a low temperature (18 ꢀC) or for a short
time at a high temperature (60 ꢀC) is unfavorable for
the preparation of composite membranes with good per-
formance. Compared with them, vaporizing at 50 ꢀC for
2 h is a satisfactory method, and this membrane has a
rejection of 0.946 and flux of 4.4 L/h m2 at 0.4 MPa.
Figure 3. Effect of GCTACC concentration on the rejection and flux
of GCTACC/PAN composite membranes.
3.1.4. Effect of GCTACC concentration on the rejection
and flux of GCTACC/PAN composite NF mem-
branes. Because the method of over-coating was
adopted to form the membrane, the concentration of
GCTACC casting solution was a key factor in mem-
brane preparation. It would fail when the concentration
of GCTACC solution was too low or too high. There-
fore, the concentration of casting solution was chosen
to be in the range of 0.25–1.25 wt %. As shown in Figure
3, the rejection increases from 0.844 to 0.946 when the
concentration changes from 0.25 to 1.0 wt %, and above
this concentration the rejection begins to decrease.
However, the flux decreases all of the time from 9.8 to
1.3 L/h m2. Considering the flux and rejection of MgCl2
together, the optimal GCTACC concentration is
1.0 wt %.
Figure 4. Effect of cross-linking temperature on the rejection and flux
of GCTACC/PAN composite membranes.
3.1.5. Effect of cross-linking temperature on the rejection
and flux of GCTACC/PAN composite membranes. For
this test, a series of GCTACC/PAN composite mem-
branes from cross-linking temperatures of 20–60 ꢀC
was prepared and measured. The results are shown in
Figure 4. The rejection increases from 0.857 to 0.946,
whereas the flux decreases from 6.73 to 3.98 L/h m2,
then it remains constant at 4.0 L/h m2 or so when the
cross-linking temperature changes from 20 to 50 ꢀC.
They show the same decreasing trend above 50 ꢀC. Con-
sidering the flux and rejection of MgCl2 together, 50 ꢀC
is the optimum cross-linking temperature.
times in the range of 18–26 h. As shown in Figure 5,
the rejection increases with the increase in cross-linking
time until 20 h, which results from the increase in cross-
linking degree, pore contraction and tortuosity.13,14
When the cross-linking time was longer than 20 h, the
rejection begins to decrease due to the decrease in degree
of cross-linking. However, the flux decreases all of the
time. The optimal cross-linking time is therefore 20 h.
3.1.7. Effect of pH of the cross-linking system on the
rejection and flux of GCTACC/PAN composite mem-
branes. The cross-linking reaction is carried out under
alkaline conditions. pH in the cross-linking system is
controlled by adding different weights of 50% KOH to
the solution containing 50 g of ethanol and 0.45 g of epi-
chlorohydrin. The results are shown in Figure 6. Rejec-
tion increases with the increase in alkalinity until 0.32 g
of 50% KOH has been added to adjust the solution to
pH 12. But when the addition is more than 0.32 g, that
is to say, pH is higher than 12, the rejection begins to
3.1.6. Effect of cross-linking time on the rejection and flux
of GCTACC/PAN composite membranes. To investi-
gate this effect, a series of GCTACC/PAN composite
membranes were prepared with different cross-linking
Table 1. Effect of vaporizing methods on the rejection and flux of
GCTACC/PAN composite membranes
Different vaporizing methods
Rejection
Flux/L mꢁ2 hꢁ1
Vaporizing for 2 h at 50 ꢀC
Vaporizing for 1 h at 60 ꢀC
Vaporizing for 18 h at 18 ꢀC
0.946
0.9
0.907
4.41
3.06
2.85

R. Huang et al. / Carbohydrate Research 341 (2006) 2777–2784
2781
3.2.1. Permeability of pure water through the GCTACC/
PAN composite membrane. The permeability of pure
water was measured at different pressures. As shown
in Figure 7, the flux of the membrane increased linearly
with operating pressure. This linear behavior is de-
scribed by a slope close to the permeability of pure water
according to the Spiegler–Kedem Model:16,17
Jv ¼ LpðDP ꢁ rDPÞ
ð3Þ
where Jv is the water flux, Lp is the permeability of pure
water, DP is the operating pressure difference, r is the
reflection factor of the membrane, and DP is the osmo-
sis pressure. Because pure water is the permeating fluid,
rDP may be ignored. The permeability of pure water is
calculated at 6.3 L/h m2 MPa for this composite NF
membrane by the linear fit of experimental data.
Figure 5. Effect of cross-linking time on the rejection and flux of
GCTACC/PAN composite membranes.
3.2.2. Effect of feed concentration on the GCTACC/PAN
composite membrane properties. Figures 8 and 9 show
the rejection and flux of MgCl2 with different concentra-
tions at operating pressures in the range 0.5–1.4 MPa,
respectively. The rejection shows the same trend of
increasing markedly with the increase in operating pres-
sures lower than 1.2 MPa, then an approximately con-
stant value is attained around 0.975; the flux increases
almost linearly with the increase in operating pressure
regardless of MgCl2 concentration.
decrease, as a result of the decrease in degree of cross-
linking due to the auto-agglutination of epichlorohydrin
under strongly alkaline conditions.15 However, the flux
increased gradually during the whole process. Therefore,
pH in the cross-linking system should be maintained at
pH 12.
3.2. Effect of operating conditions on the GCTACC/PAN
composite membrane properties
Salt concentrations were determined with a Model
DDS-11 A conductivity meter. The membrane perfor-
mance tests were carried out at 25 ꢀC, and the cycling
flow of the feed was kept at 30 L/h. The membrane
was pre-pressurized for 30 min prior to the test. The flux
and rejection of the membrane were determined accord-
ing to the equations given in Section 2.3. The membrane
for further experiments was prepared from 1.0 wt %
casting solution, vaporized at 50 ꢀC for 2 h, and cross-
linked for 20 h at 50 ꢀC and pH ꢀ 12 with ethanol/
epichlorohydrin (50:0.45 wt/wt).
3.2.3. Effect of different inorganic electrolytes on the
GCTACC/PAN composite membrane properties. In the
salt transport process, the electric properties of the NF
membrane have a great effect on ion permeation and
rejection. Here the concentration of the single salt solu-
tions tested is about 1000 mg/L. The rejection and flux
of different inorganic electrolytes by the GCTACC/
PAN composite membrane are shown in Figures 10 and
11, respectively. Obviously the rejections to inorganic
Figure 6. Effect of pH on the rejection and flux of GCTACC/PAN
composite membranes.
Figure 7. Effect of operating pressure on the flux of pure water.

2782
R. Huang et al. / Carbohydrate Research 341 (2006) 2777–2784
Figure 8. The curves about the rejections on feed solution with a
different concentration against operating pressure.
Figure 10. The rejection on different inorganic electrolyte solution
against operating pressure.
Figure 9. The curves about the fluxes on feed solution with a different
concentration against operating pressure.
Figure 11. The flux on different inorganic electrolyte solutions against
operating pressure.
electrolytes follow the order of MgCl2 ꢀ CaCl2 >
MgSO4 > NaCl ꢀ KCl > Na2SO4 > K2SO4, which is
typical of a positively charged NF membrane and is ex-
plained by the electrostatic effect.18,19 The orders of the
rejections: MgCl2 ꢀ CaCl2 > NaCl ꢀ KCl and MgSO4 >
Na2SO4 > K2SO4, corresponding to the increasing order
of the cation charge densities, because the active layer
consisting of GCTACC has quaternary ammonium
group distribution and allows a stronger repulsion of
Mg2+, Ca2+ than Na+, K+. The order of rejection:
MgCl2 > MgSO4, NaCl > Na2SO4 and KCl > K2SO4,
goes opposite with the increasing order of anion charge
densities, because the active layer of the composite mem-
branes contains quaternary ammonium groups as men-
tioned above and has stronger attraction forces for
SO4 than Clꢁ. The order of the rejection MgSO4 >
2ꢁ
NaCl may be because the repulsion forces on the cation
are stronger than the dominant attraction. In addition,
the fluxes increase almost linearly with the increase in
Figure 12. The curves about streaming potential of the GCTACC/
PAN composite NF membrance.

R. Huang et al. / Carbohydrate Research 341 (2006) 2777–2784
2783
Figure 13. The surface (left) and cross-section (right) images of the composite nanofiltration membrane.
operating pressure, but they are lower than the flux of
pure water, which results from the solvent effect.
graft copolymer of trimethylallyl ammonium chloride
onto chitosan (GCTACC) as membrane-forming mate-
rial, PAN UF as the support layer and epichlorohydrin
as the cross-linking reagent. The effect of preparation
conditions including GCTACC concentration, epichlo-
rohydrin concentration, cross-linking temperature, etc.
on the performance of the composite membrane has
been evaluated. The membrane with optimal perfor-
mance has been determined by adjusting these factors
and its structure has been characterized by scanning
electron microscopy.
In addition, some performance tests have been con-
ducted for the composite NF membrane. The perfor-
mances are affected by the feed concentration and the
type of feed, etc. The rejections of electrolytes with diva-
lent cations (Mg2+, Ca2+) reach >0.9, whereas rejections
of electrolytes with divalent anions and low-valency cat-
ions are lower than 0.7, so this membrane has the poten-
tial for selective separation of different electrolytes. The
rejections follow the decreasing order of MgCl2, CaCl2,
MgSO4, NaCl, KCl, Na2SO4, and K2SO4, which is
typical of a positively charged NF membrane. In
addition, the curve for the streaming potential also
illustrates the positively charged characteristics of the
composite membrane.
3.3. Streaming potential of the GCTACC/PAN composite
NF membrane20,21
The streaming potential of the GCTACC/PAN compos-
ite NF membrane was measured at different pressures in
0.1 M KCl solution, which was filled between the sides
of the membrane. The curve for the streaming potential
against pressure is shown in Figure 12. It can be seen
that the streaming potential increases linearly with the
increase in pressure and all potentials are above zero,
suggesting the characteristics of a positively charged
membrane. The pressure osmotic coefficient is attained
at 11.7 mV MPaꢁ1 for this NF membrane by the linear
fit of experimental data.
3.4. Structure characteristic of the GCTACC/PAN
composite NF membrane
The cross-section and surface of this membrane were
characterized with a JDEL JMS-840 scanning electron
microscope and the results are shown in Figure 13.
There is a thin active functional layer with a compact
surface on a finger-like supporting layer of PAN UF
base membrane, revealing the asymmetric and compos-
ite structure of this membrane.
References
1. Wu, L. H. Membr. Sci. Technol. 1997, 17, 11–15.
2. Cadotte, J.; Forester, R.; kin, M.; Petersen, R., et al.
Desalination 1988, 70, 77–88.
3. Raman, L. P.; Cheryan, M.; Rajagopalan, N. Chem. Prog.
1994, 90, 68–74.
4. Conclusions
Positively charged composite NF membranes have been
successfully prepared by using chitosan derivative, a

2784
R. Huang et al. / Carbohydrate Research 341 (2006) 2777–2784
4. Frank, M.; Brageman, G.; Zwijnenburg, A.; Wessling, M.
Sep. Purif. Technol. 2001, 22–23, 499–506.
13. Miao, J.; Chen, G. H.; Gao, C. J. Desalination 2005, 181,
173–183.
5. Schaep, J.; Vander Bargeman, B.; Vandecasteele, C.;
Wilms, D. Sep. Purif. Technol. 1998, 14, 155–162.
6. Xu, Y.; Lebrun, R. E. J. Membr. Sci. 1999, 158, 93–104.
7. Raman, L. P.; Cheryan, M.; Rajagaopalan, N. Chem. Eng.
Prog. 1994, 90, 68–74.
14. Krajewska, B. O.; Olech, A. Poly. Gels Networks 1996, 4,
33–43.
15. Zheng, H.; Du, Y. M.; Yu, J. H.; Xiao, L. Chem. J. Chin.
Univ. 2000, 21, 809–812.
16. Xu, Y. Z.; Lebrun, R. E. Desalination 1999, 122, 95–106.
17. Song, Y. J.; Sun, B. H. Technol. Water Treat. 1997, 23, 78–
82.
8. Matsuyama, H.; Kitamura, Y.; Naramura, Y. J. Appl.
Polym. Sci. 1999, 72, 397–404.
410
9. Xu, T. W.; Yang, W. H. J. Membr. Sci. 2003, 215, 25–32.
10. Tan, S. Z.; Chen, Z. H.; Chen, Z. H. Trans. China Pulp
Paper 2002, 17, 63–66.
11. Musale, D. A.; Kumar, A. Sep. Purif. Technol. 2000, 21,
27–38.
18. Xiaolin, W.; Tsuru, T.; Nakao, S.; Kimura, S. J. Membr.
Sci. 1995, 103, 117.
19. Linde, K.; Jonsson, A. S. Desalination 1995, 103, 223–232.
20. Mo, JX.; Liu, S. M. Technol. Water Treat. 1991, 17, 153–
161.
12. Shu, H. Y.; Tang, X. H.; Fu, R. H. Chin. J. Appl. Chem.
2004, 21, 734–736.
21. Wang, Y.; Fan, Y. S.; Du, Q. Y. J. Tianjing Polytech.
Univ. 2001, 20, 79–83.