J Porous Mater (2012) 19:721731
DOI 10.1007/s10934-011-9524-3
Preparation and characterization of low-pressure nanofiltration
membranes and the application in the separation process of dye
molecules
Xiu-Zhen Wei Jia Yang You-Yi Xu
Bao-Ku Zhu Guo-Liang Zhang
Published online: 14 October 2011
 Springer Science+Business Media, LLC 2011
Abstract Low-pressure nanofiltratioin (NF) membranes
were prepared with hydroxyl groups ended hyperbranched
polyester (HPE) using polysulfone (PSf) ultrafiltration
membrane as the porous support. The prepared membranes
are characterized using Fourier transforms infrared spec-
trometry, scanning electro microscopy, atom force
microscopy, X-ray photoelectron spectroscopy and water
contact angle. The results indicated that the crosslinked
HPE formed a uniform, ultra-thin and active layer on PSf
support. Water permeability and salt rejection of the
membranes were measured at the trans-membrane pres-
sures as low as 0.3 MPa. NF membranes exhibited high
enhancement in water permeability while maintaining high
rejection of salts. The rejections of the nanofiltration
membrane to Rhodamine B and Xylenol orange were 100
and 98.4%, respectively. The study indicated that these NF
membranes could be applied to separate small organic
molecules.
Keywords Hyperbranched polymers  Nanofiltration
membranes  Polyester  Salt rejection  Dye molecules
X.-Z. Wei (&)  J. Yang  G.-L. Zhang
College of Biological and Environmental Engineering,
Zhejiang University of Technology, Hangzhou 310014,
Peoples Republic of China
e-mail: xzwei@zjut.edu.cn
X.-Z. Wei  Y.-Y. Xu (&)  B.-K. Zhu
Institute of Polymer Science, Zhejiang University,
Hangzhou 310027, Peoples Republic of China
e-mail: opl-yyxu@zju.edu.cn
1 Introduction
It was reported by the World Health Organization (WHO)
that 80 percent disease and 33 percent death of human being
were related to the sanitation of drinking water. Sources of
drinking water have been contaminated by various human
activities [1]. Surface water and groundwater are the dom-
inant resources for drinking water in China and also
throughout the world. Various organic compounds, inor-
ganic salts and endocrine disrupting chemicals derived from
plastic wastes pose a danger to human health as well as the
ecosystem and they are widely present in water resources
[2]. Organic micropollutants, for example pharmaceutical
compounds, which are not quantitatively eliminated by
wastewater treatment, have the potential to influence the
quality of raw water resources [36].
Membrane filtration is a common technique to remove
contaminants from waste water and drinking water. NF
membranes have a potential to remove pesticides with the
molecular weights ranging from 200 to 400 Da as shown
by several researchers [69]. Besides, NF membranes have
the ability to separate ions which have different valences
(monovalent or multivalent) or ionic species from low
molecular-weight organic compounds. NF process is
emerging as a favorable alternative to conventional pro-
cesses used for drinking water treatment since it can
remove a broad range of water contaminants within a
single treatment step [1, 6, 1014].
The operation pressure of NF used in cleaning equip-
ments for family drinking water is usually lower than
0.3 MPa. Now, most commercial NF membranes are thin-
film composite (TFC) membranes prepared with interfacial
polymerization developed in 1972 [1517]. The common
membranes such as polyamide (PA) [18], polyvinyl alcohol
(PVA), sulfonated polysulfone (SPS), sulfonated poly
123
722
(ether ketone) (SPEEK), cellulose acetate (CA) [1922]
etc. were negatively charged and accordingly the operation
pressure was higher than 0.5 MPa. Therefore, it is signifi-
cant to explore the specific extra-low-pressure NF mem-
brane via appropriately adjusting the structure of the
reactive monomer to form a relative looser selective.
Previous studies have been focused on choosing or syn-
thesizing new monomers with special functional groups to
prepare the TFC membrane. Willem and Li presented their
primary study on NF formation with poly (propyl imine) (PPI)
and polyamidoamine (PAMAM) dendrimers polymerized
with trimesoyl chloride (TMC) [23, 24], respectively. Some
other researchers have prepared NF membranes using hy-
perbranched polymers (HBPs). Moore et al. [25, 26] synthe-
sized a series of rigid star amphiphiles (RSAs) with different
end groups. New generation of nanofiltration (NF) mem-
branes have been prepared using these RSAs by direct per-
colation of methanol solutions of the RSAs through an
asymmetric polyethersulfone (PES) support film which was
previously conditioned with methanol and cross-linked
polyvinyl alcohol (PES-MeOH-PVA support). Ruaan et al.
[27] prepared NF membranes using three different amines
(ethylenediamine (EDA), diethylenetriamine (DETA), and
hyperbranched polyethyleneimine (PEI)) and two kinds of
acyl chlorides (trimesoyl chloride (TMC) and terephthaloyl
chloride (TPC)). However, within all these NF membrane
developments, further decrease in the operation pressure is
expected. In this study, the extra-low pressure NF membranes
were prepared by taking advance of the highly branched
structure of HBPs. Our previous work has already prepared
the nanofiltration with trimesoyl chloride (TMC) and glutar-
aldehyde (GA) crosslinked HPE as the selective layer [28,
29]. Specific structure of HPE endow these NF membranes
with expected properties. Nanofiltration membrane cross-
linked with GA allowed higher water flux and salt rejection
than that crosslinked with TMC. The reactive monomer
dosage has been optimized compared with our previous study.
The ultra-thin layers chemical structure was studied by
FTIR-ATR and XPS. However, the performance of the
composite membrane is also determined by the preparation
conditions of the selective layer. In this study, the preparation
techniques have been optimized via the reactive monomer
and the NF membranes were introduced to treat analogue dye
wastewater. It is meaningful to widely perform NF membrane
in drinking water treatment for family daily life.
2 Experimental
2.1 Materials
All chemicals used were AR grade. Glutaraldehyde (GA)
was purchased from Aldrich (USA). Sodium chloride
123
J Porous Mater (2012) 19:721731
(NaCl), sodium sulfate (Na2SO4), magnesium chloride
(MgCl2), magnesium sulfate (MgSO4) were purchased
from Shanghai guoyao company and used as received.
Polysulfone (PSf) ultrafiltration (UF) membrane used as a
support was purchased from Development Center of Water
Treatment Technology, State Oceanic Administration
(Hangzhou, China). The molecular weight cut-off of PSf
membrane is 30,000 (Dalton). HPE (BOLTORN H40) was
purchased from Perstorp Specialty Chemicals AB (Swe-
den). The characteristics of BOLTORN H40 can be found
in the Ref. [28].
2.2 Membrane preparation
HPE-GA nanofiltration membranes were prepared accord-
ing to the method reported by us (the Ref. [28]). Different
amount of HPE was dissolved in 40 mL of waterethanol
mixture (2:3 in volume ratio) to form a homogeneous
solution, which was used as the water phase. Firstly, the
PSf-UF membrane was wetted by ethanol. Then the wetted
support membrane was immersed in the water phase con-
taining HPE for 40 min in which 0.3 wt% of sodium
dodecyl sulfate (SDS) was added as a surfactant. And then
the membrane was taken out. When the excessive solution
drained off, the membrane was submerged in 40 mL glu-
taraldehyde (GA) water solution with different concentra-
tion. Different concentration GA solution was obtained
from diluting commercial 25 wt% GA solution. Thus, a
virgin NF membrane was obtained. The obtained com-
posite membrane then underwent heat treatment in an oven
for 20 min. Finally, the membrane was preserved in water
for further use. The resulted membranes were coded as
HPE-GA NFx (x = 0, 1, 2, 3, 4, 5, 6, 7, 8). Table 1 shows
the sample ID of different membranes and their corre-
sponding interfacial polymerization conditions.
2.3 Membrane characterization
The chemistry difference between PSf support and prepared
HPE-GA NF membrane were characterized by Fourier
transform infrared spectrometer (FTIR-ATR) (Vector 22
FTIR, Bruker Optics, Switzerland) and X-ray photoelectron
spectroscopy (XPS, PHI 5000c, Peking-Elmer instruments).
Surface and cross-section morphologies were imaged by a
field emission scanning electron microscopy (FE-SEM,
JEOL, JSM-5510LV, Japan). Surface roughness of different
membranes was characterized by an atomic force micro-
scope (AFM, SPI3800N, Seiko Instrumental, Japan). The
hydrophilicity of the membrane top surface was character-
ized using a contact angle goniometer (OCA20, Dataphys-
ics, Germany) at room temperature. The membranes were
dried for 24 h in vacuum at 45 C before it was
characterized.
J Porous Mater (2012) 19:721731 723

Table 1 Membrane ID of different PSf-based membranes and their also displayed. As demonstrated in Fig. 1, the distinct band at
corresponding compositions in the interfacial polymerization 1,735 cm-1 for NFX (X = 0, 1, 2, ), is corresponded to
Membrane HPE GA Membrane HPE GA carbonyl stretching frequency of HPE, which indicated that a
ID (g) (g/L) ID (g) (g/L) thin layer of HPE was coated onto PSf membrane surface.
Compared with pure PSf membranes, HPE-GA NF mem-
NF0 0.28 0 NF4 0.62 7.4
branes also exhibited evident absorbance peaks at 3,369 (mO
NF1 0.28 7.4 NF6 0.28 8.5
H) and 2,961(CH), which implied that the active skin layer
NF2 0.39 7.4 NF7 0.28 6.3
of HPE-GA NF membrane was polyester containing car-
NF3 0.49 7.4 NF8 0.28 4.2
bonyl bands and hydroxyl groups. The presence of absorp-
tion band at 3,369 (mOH) for all NF membranes indicated
2.4 Characterization of membrane for salts rejection that the hydroxyl groups were not completely reacted, which
would improve the hydrophilicity of the NF membranes.
Nanofiltration properties of different membranes were
studied in a flat-sheet cross-flow NF test cell with an active 3.2 XPS analysis
membrane area of 23.75 cm2. The membrane was stabi-
lized at 0.6 MPa for at least 30 min with deionized water The chemical composition of the thin layer was further
before testing. NaCl, Na2SO4, MgCl2 and MgSO4 salt characterized by XPS analysis. The XPS survey scans for
solutions of 1 g/L flows across the membrane at a rate of the investigated samples are shown in Fig. 2 and the
0.54 L/min, respectively. All experiments were carried out
at 25 C. Both retention and permeate were recycled back
to the feed tank in order to keep a constant concentration of
salts. The filtration characteristics including permeation
flux (F) and rejection (R) were determined. The permeation
flux, F, is calculated according to formula (1).
F V=A  t; 1
where V is the total volume of the water or solution
permeated during the experiment, A represents the
membrane area, and t denotes the operation time.
NF rejection is defined as R 1  Cp =Cf  100%; 2
where R is the percent solute rejection, Cp and Cf are the
concentrations of solute in permeate and feed, respectively.
Salt concentration in the feed and permeate were
determined by Electrical Conductivity DDS-11A (Shang- Fig. 1 FTIR-ATR of PSf-based membranes
hai Leichi Instrument, China). Salt concentration was cal-
culated according to the electrical conductivity of the
corresponding salt solution. The rejection of dye molecules
were determined by measuring spectrophotometrically the
difference between the concentrations in the feed and
permeate. A calibration curve between the spectrophoto-
metrical absorbance and the dye concentration was estab-
lished to reduce the experimental error. The calculated
wavelength for Rhodamine B, xylenol orange and methyl
orange are 550, 436 and 465 nm respectively.

3 Results and discussion

3.1 FTIR-ATR study

Samples of the thin-film composite membrane were char-

acterized using FTIR-ATR for surface analysis. As a com-
parison, the FTIR spectra of PSf support membranes were Fig. 2 XPS survey scan of PSf-based membranes

123
724 J Porous Mater (2012) 19:721731

Table 2 Elements content of membranes surface from XPS analysis membranes showed lower contact angles than those of the
Membrane C (%) O (%) S (%) -2
S/C (910 ) -2
S/O (910 )
PSf-UF membrane. Water contact angle of HPE-GA NF
membranes decreased with HPE dosage increasing, which
PSf-UF 80.3 15.7 2.5 3.1 15.7 resulted from the increase content of hydroxyl groups.
NF1 75.5 23.0 1.5 1.9 6.5 However, the water contact angles were still relative high
NF2 74.4 24.0 1.5 2.1 6.4 because the hydrophobic segments in the HPE molecules
NF3 73.1 25.5 1.5 1.9 5.7 may migrate onto the membrane surface after drying. On
NF4 72.7 26.0 1.3 1.8 3.8 the other hand, the quickly decrease of the water contact
angle on NF composite membranes indicated that the
crosslinked HPE network formed on the PSf-UF membrane
element contents of membrane surface are presented in surface was relatively hydrophilic.
Table 2. Compared with the virgin PSf membrane, HPE-
GA NF membranes showed higher content of oxygen and 3.4 Morphology structures of PSf NF membranes
lower content of sulfur and carbon. The content of carbon
and sulfur decreased from 80.3 to 72.7%, from 2.5 to 1.3%, FE-SEM images of different membranes are showed in
respectively. Meanwhile, the content of oxygen increased Fig. 4. It can be seen that surface of support PSf membrane
from 15.7 to 26%. It is worthwhile to notice here that the was smooth and featureless. The active layer of NF
O/S ratio reduced from 15.7 to 3.8% and the significant membranes was uniform, smooth and dense too. On the
change of the chemical composition was due to the car- other hand, the cross-section photos showed that the
bonyl groups in the active layer of the crosslinked HPE. membrane had regular finger-like pores, which would
Small amount of sulfur could still be detected since the endow the membrane with high flux. The morphology
selective layer was very thin. The element content changes change of the cross-section between PSf-UF and NF
indicated that PSf membrane surface was covered by a thin membranes could be observed distinctly. The uppermost
layer of crosslinked HPE. Both XPS and FTIR-ATR results layer (such as La and Lb presented in Fig. 4) of NF
indicated that PSf-UF membrane surface were completely membrane was thicker than that of UF membrane, which
covered by crosslinked HPE, but it cannot be inferred indicated that crosslinked HPE molecules were deposited
whether crosslinked HPE was penetrated into PSf-UF onto UF membrane surface. It could be seen from Fig. 4,
membranes pores or deposited onto the PSf-UF membrane the thickness of the selective layer increased with the
surface only. Thus, SEM and AFM characterization were dosage increasing of HPE, which indicated that more and
also attempted. more HPE molecules deposited onto PSf-UF membrane
surface. The results implied that the dosage of reactive
3.3 Hydrophilicity of HPE-GA NF membranes monomer may affect the properties of NF membranes
obviously.
Surface hydrophilicity can provide additional information Preliminary AFM experiments were carried out in air
on morphologies, which was characterized by contact angle using dry membranes. Results obtained form AFM char-
measurements. The water contact angle change with time acterization of the surface topographies for different
was also characterized. As shown in Fig. 3, all NF membranes were generally consistent with the FE-SEM
observations. The 3D AFM images revealed that HPE-GA
NF membranes surface were flat and uniform (Fig. 5).
Roughness of PSf-based membranes generally increased in
the sequence blank PSf \ NF0 \ NF1 \ NF2 \ NF3 \
NF4 (Fig. 6). The roughness of the NF membranes
increased with the dosage increase of HPE because the
quantity of HPE deposited onto the PSf membrane surface
increased. From above results, it was found that the effect
of HPE dosage on the roughness was obvious. However,
the average roughness of PSf-UF (5.2 nm), PSf-Et
(5.4 nm) [28], NF0 (5.5 nm), NF1 (5.6 nm) and NF2
(5.7 nm) was very close, which indicated that HPE mole-
cules penetrated into PSf-UF surface pores firstly, and then
deposited onto the surface of PSf-UF membrane surface
with the increase of HPE dosage. In generally, on the basis
Fig. 3 Water contact angle change of different membranes with time of preceding discussion the following simplifying

123
J Porous Mater (2012) 19:721731 725

Fig. 4 FE-SEM images of PSf-based membranes a PSf-UF, b NF1, c NF2, d NF3, e NF4, 1: the upper surface; 2: the cross-section; 3: upper
part of the cross-section

assumptions were made: the reactive monomer, HPE roughness of HPE-GA NF membranes surface was
penetrated into PSf-UF surface pores firstly, and then increased with HPE dosage increasing, HPE-GA NF
deposited on the UF membrane surface. Although, the membrane surface were rather flat. The flatness of the skin

123
726
Fig. 5 AFM image of PSf-based membranes
layer may endow the NF membrane with relative high flux
and good antifouling properties [25].
3.5 Performance optimization of HPE-GA NF
membranes
In general, the performance of the composite membrane is
determined by the chemistry and preparation conditions of
the selective layer. The chemistry structure is an inherent
property of the reactants taken to form the selective layer
while the later plays an important role in determining the
structure of the interfacial polymerized film and the sub-
sequent membrane performance. HPE has highly branched
and nearly three-dimensional structure, which would
endow the selective layer with high permeability and high
salt rejection. In order to form ultra-thin film selective
layer, the preparation conditions have to be optimized. In
this study, 1 g/L Na2SO4 water solution was used to
evaluate the performance of HPE-GA NF membranes
under 0.3 MPa.
3.5.1 Effect of HPE dosage
In this study, the concentration of GA was fixed at 7.4 g/L,
the reaction time was fixed at 10 min and heat-treatment
temperature was fixed at 80 C. The dosage of HPE from
123
J Porous Mater (2012) 19:721731
0.28 to 0.62 g was investigated firstly under the above-
mentioned membrane preparation conditions. The salt
permeation and rejection of different HPE-GA NF mem-
branes was presented in Fig. 7.
It can be found that Na2SO4 flux and rejection increased
with HPE dosage increasing (from Fig. 7). The possible
explanation of the phenomenon was that GA acted as a cross-
linking point during interfacial polymerization, and the
concentration of GA was fixed. When the dosage of HPE was
lower, such as NF1 membrane, most of hydroxyl groups
were reacted with GA molecules, the cross-linked network
on the NF membrane surface was relative denser. Thus, the
flux was lower. In addition, HPE molecules have near three-
dimensional structure, the thin selective layer could not
effectively reject Na2SO4 molecules, which lead to the lower
rejection (88%). Commercially available HPE is synthesized
from 2, 2-bis(methylol) propionic acid (bis-MPA) as the tri-
functional AB2 monomer and ethoxylated pentaerythritol
(PP50) as the tetra-functional B4 core molecules. HPE
molecule has a hydrophobic core, which could anchor to the
support PSf membrane through noncovalent supramolecular
interactions. Due to the relatively hydrophobic core and low
solubility in water, more HPE molecules would be deposited
onto membrane surface with HPE dosage increasing, which
would make the selective layer relative loose and endow the
NF membrane with high flux and high rejection.
J Porous Mater (2012) 19:721731
Fig. 6 Roughness of PSf-based membranes
Fig. 7 Effect of HPE dosage on NF membranes behavior for Na2SO4
under 0.3 MPa
3.5.2 Effect of GA concentration
To optimize the membrane performance, the influence of
GA concentration were investigated on the premise of
same HPE dosage (0.28 g), reaction time 10 min and heat-
treatment temperature 80 C. The salt flux and rejection of
different membranes was shown in Fig. 8. When GA
concentration increased, the flux decreased and Na2SO4
rejection increased. This phenomenon was because that the
selective layer would be denser with GA concentration
increasing while the dosage of HPE was constant.
3.5.3 Effect of heat-treatment temperature
To fabricate NF with perfect performance, suitable heat-
treatment process is essential because some defects would
disappear and the selective layer structure would be more
perfect by this process. The effect of heat-treatment tem-
perature on NF membrane performance was investigated
727
Fig. 8 Effect of GA concentration on NF membranes behavior for
Na2SO4 under 0.3 MPa
on the premise of keeping heat-treatment time at 20 min.
Figure 9 presented the effect of heat-treatment temperature
on flux and salt rejection. Flux increased first and then
decreased with the increasing of heat-treatment tempera-
ture. However, Na2SO4 rejection increased all along with
heat-temperature increasing. This phenomenon could be
explained as following. With increasing the heat-treatment
temperature, more hydroxyl groups would react with GA
molecules on heat-treatment stage, which lead to the rela-
tively compact selective layer and increased salt rejection.
3.5.4 Effect of heat-treatment time
In this study, the reaction time was fixed at 20 min and the
heat-treatment temperature was fixed at 90 C. The effects
of heat-treat time on the permeate flux and salt rejection
were investigated under the above-mentioned conditions.
The flux decrease was accompanied with rejection increase
when the heat-treatment time increased. That is, when heat-
treatment time increased from 10 to 30 min, the perme-
ation flux decreased from 120.5 to 22.8 L/m2h, and salt
rejection increased from 43.2 to 93.2% (Fig. 10). This
phenomenon was attributed to the increased diffusion
barrier, which is caused by the decrease of pore size due to
the prolonging of heat-treatment time. Accordingly, when
the heat-treatment time prolonged to 30 min, the perme-
ation flux was further decreased and salt rejection was
further increased. However, if the heat-treatment time was
greatly extended, the properties of NF membranes would
be destroyed due to some cross-linked bonds would be
decomposed. In the paper, the support of the substrate
membrane is polyester, the porous membrane on polyester
surface is polysulfone, and the selective layer we synthe-
sized on polysulfone membrane surface is GA crosslinked
polyester. Different polymer would present different
123
728
Fig. 9 Effect of heat-treatment temperature on NF membranes
behavior for Na2SO4 under 0.3 MPa
Fig. 10 Effect of heat-treatment time on NF membranes behavior for
Na2SO4 under 0.3 MPa
thermal expansion coefficient, which may lead to the split
of the active functional layer after heat treatment. At the
same time, the selective layer may tend to create chan-
neling (defect) at prolonged heat-treatment time [30]. That
is why the water flux increased and rejection coefficient
decrease is expected.
3.5.5 Effect of operation pressure
The permeate flux and rejection of Na2SO4, NaCl, MgSO4,
MgCl2 (1 g/L) under different operation pressure was
shown in Figs. 11 and 12, respectively. It can be seen from
Fig. 11, the permeation flux of HPE-GA NF membrane
increased with the increase of operation pressure. The
permeation flux is in direct proportion to the operation
pressure, which is due to the difference of operation pres-
sure and osmotic pressure. Compared with the operation
pressure, the osmotic pressure is so low (53 mmHg for
1 g/L Na2SO4 solution) that it can be neglected for the low
123
J Porous Mater (2012) 19:721731
Fig. 11 Effect of operating pressure on flux of NF membranes
Fig. 12 Effect of operating pressure on salt rejection
concentration of salts. Therefore, the operation pressure is
approximate to the trans-membrane pressure. Figure 12
presented that the salt rejection increased slightly with
operation pressure increasing. This sequence is explained
by Eq. 3:
F BDCs 3
where B is the salt permeation coefficient, DCs is salt
concentration difference on two sides of the membrane,
respectively. As DP increases, water flux increases, DCs
increases too, these effects combine to result in a gradual
increase in salt rejection [24].
3.6 Membrane nanofiltration properties test
Table 3 presented the permeate flux and rejection of HPE-
GA NF membrane to different salts. Here the permeation
experiments were performed for four inorganic salts (NaCl,
J Porous Mater (2012) 19:721731 729

Table 3 Separation behavior of NF membranes to different which would impose extra steric hindrance for separated
molecules molecules. Subsequently, the diffusive and convective ion
Solute Molecular weight Flux (l/m2h) Rejection (%) transports are retarded. Comparing the NF results in this
study with HPE-TMC (trimesoyl chloride) NF membranes
NaCl 58.5 41 2 30 1.5
prepared according to the Ref. [28], we found that the
Na2SO4 142 46 2.3 94 3 rejection of Na2SO4 was higher and flux was decreased. This
MgSO4 120 37 1.8 31 1.5 phenomenon can be explained as following. HPE, the reac-
MgCl2 203 36 1.8 11 1.5 tive monomer, has near three-dimensional structure and
Rhodamine B 479 46 2.3 100 3.5 many reactive terminated hydroxyl groups. TMC has three
Xylenol orange 672.7 39 1.8 98 3 proprietary reactive groups and GA have two reactive
Methyl orange 327.3 42 2 21 1 groups. When TMC was used as the crosslinking agent, the
special structure of HPE and the rigid structure of TMC led
HPE-TMC NF membranes have loose selective layer. And
Na2SO4, MgCl2 and MgSO4) solutions with a concentration HPE-TMC NF membranes showed higher permeated flux. If
of 1 g/L at 25 C and the operation pressure was 0.3 MPa. GA was used as the crosslinking agent, the linear flexible
The objective was to investigate the effect of the salts type structure of GA endowed HPE-GA NF membranes relative
on the performance of NF membranes. The salts rejection denser selective layer. Thus, the structure difference of
order of these NF membranes is Na2SO4 [ MgSO4 & crosslinking agent endowed the NF membranes with differ-
NaCl [ MgCl2. The result is similar to the sequence ent properties.
developed by Dai et al. [31], where the negatively charged
NF membrane was used. As well known, the rejection to
salts for a NF membrane depends on both the pore size of 3.7 Comparative among different nanofiltration
the membrane and the electrostatic action between the membranes
membrane and the ions in solution [32, 33]. For the same
type of cation e.g. Na? or Mg2?, HPE-GA NF membranes NF membranes used in drinking water treatment should pos-
showed higher rejection for high-valence anions than that of sess the following properties. First, NF membranes could
low-valence anions. For example, the rejection of Na2SO4 is reject effectively the nature organic molecules, pesticides,
higher than that of NaCl and the rejection of MgSO4 is by-product and precursor form drinking water disinfection. At
higher than that of MgCl2. The results were attributed to the same time, the rejection of NF membranes to dissolved
that the hydration ion radius of SO42- is larger than that of salts should relatively low. Second, Antifouling NF mem-
Cl- [34]. At the same time, for the same kind of anions branes with high flux, especially could resist organic com-
SO42- and Cl-, membranes showed higher rejection for pounds or microorganisms fouling, should be researched and
Na? than that of Mg2?. Thus, it was considered that the fabricated. All of this could promote the application of NF
sieving effect plays an important role for HPE-GA NF membranes in drinking water treatment process. In recent
membranes. Of course, the special salt rejection behavior years, different NF membranes have been prepared with
may be related with the three-dimension spherical structure hyperbranched polymers (HBPs) (Table 4). Compared with
of HPE. HPE molecules have many hydroxyl groups at the the traditional NF membranes fabricated with small reactive
end of the branched arms. Even it was crosslinked with GA, monomer, NF membranes prepared with HBPs showed rela-
some unreacted hydroxyl groups were still remained. Parts of tively low salt rejection. However, the highly branched
hydroxyl groups could migrate into the center of the pores structures and a great number of functional terminated groups
through flexible alkyl arms. The pendant hydroxyl groups of HBPs endowed these NF membranes with relatively
could easily access to charged salts and formed salt bridge, loose selective layer. And these NF membranes showed

Table 4 NF properties of different membranes prepared with various reactive monomers as selective layer
NF membrane Flux (L/(m2 h MPa)) Rejection (%) Salt rejection order

PEI/TMC 95 80 (MgCl2) MgCl2 [ MgSO4 [ Na2SO4 [ NaCl [27]

PEI/TPC 31 93 (MgCl2) MgCl2 [ MgSO4 [ Na2SO4 [ NaCl [27]
HPE/TMC 263.7 85.4 (Na2SO4) Na2SO4 [ MgSO4 [ MgCl2 [ NaCl [28]
HPE/GA 174 93.2 (Na2SO4) Na2SO4 [ NaCl [ MgSO4 [ MgCl2
PAMAEMA/PSF 10.4 98 (MgCl2) MgCl2 [ MgSO4 [ NaCl [ Na2SO4 typical positively charged NF [35]
SPPESK/PSF 120 91 (Na2SO4) Na2SO4 [ NaCl [ MgSO4 [ MgCl2 typical negatively charged NF [31]

123
730
relative high flux. It can be seen from Table 4, these NF
membranes presented special salt rejection order. This special
salt rejection order may be determined by the special structure
of HPE. As we know, HBPs have many reactive groups, which
may migrate into the center of the pores through flexible alkyl
arms and form salt bridge with charged salts. The salt bridge
would impose extra steric hindrance for separated salt mole-
cules, which would lead the NF membranes to illustrate spe-
cial slat rejection order. Although TMC and TPC both have
rigid structures. TMC has three functional groups and TPC has
two functional groups. We may infer NF membranes cross-
linked with TMC would have relative loose selective layer. SO4 & NaCl [ MgCl2. The rejection of the composite
Thus, PEI/TMC and HPE/TMC presented higher flux and
lower rejection than that of PEI/TPC. However, HPE/GA
showed relatively high flux and rejection, despite that GA also
has two functional groups. We infer that may be due to GA is a
flexible molecule. The flexibility of GA endows the HPE/GA
NF membrane with a reasonable loose and thickness drinking water purification.
selective layer.
3.8 Application of nanofiltration membrane
on separation dye wastewater
Rejection of HPE-GA NF membranes to different dye mol-
ecules were also presented in Table 3 and the dye molecules
structure was shown in Fig. 13. Both positive charged
Rhodamine B and negative charged Xylenol orange have
rigid structures. But, the rejection of Rhodamine B and
Xylenol orange were close to 100%. Methyl orange has the
similar molecular weight with Rhodamine B and carries the
same charge with Xylenol orange. However, the rejection of
Methyl orange was very low (21.3%). The rejection behavior
of dye molecules also indirectly suggested that the nanofil-
tration properties of HPE-GA NF membranes mainly
depended on sieving effect. The results implied that the NF
membranes could reject most of organic micropollutants, for
example pharmaceutical molecules, endocrine disrupting
compounds (EDCs) etc. and have a strong application
potential in drinking water purification.
Fig. 13 Structure of different dye molecules
123
J Porous Mater (2012) 19:721731
4 Conclusion
HPE-GA composite NF membranes were prepared via in
situ interfacial polymerization using hyperbranched poly-
ester (HPE) and GA as reactive monomers and cross-
linkers. FTIR-ATR and XPS spectra indicated that GA
reacted successfully with HPE, which is favorable to build
a relative loose top-layer on the support PSf membrane.
The obtained NF membranes showed both high rejection
and high water permeation under 0.3 MPa. The salts
rejection order of NF membranes is Na2SO4 [ Mg-
membranes for Rhodamine B and Xylenol orange were
close to 100%, which indicated that the obtained NF
membranes could remove the organic micropollutants in
drinking water. The results suggested that the NF mem-
branes could find applications as promising materials in
Acknowledgments Financial support from Natural Science Foun-
dation of China (Grant No. 50903071) and Zhejiang Province Natural
Science Foundation (Grant No. Y4090324) are gratefully acknowl-
edged. The authors also greatly thank Science Foundation of Zhejiang
University of technology (Grant No. 105005729).
References
1. A. Favre-Reguillon, G. Lebuzit, D. Murat, J. Foos, C. Mansour,
M. Draye, Selective removal of dissolved uranium in drinking
water by nanofiltration. Water Res. 42, 11601166 (2008)
2. C. Zwiener, Occurrence and analysis of pharmaceuticals and their
transformation products in drinking water treatment. Anal. Bio-
anal. Chem. 387, 11591162 (2007)
3. C. Bellona, J.E. Drewes, P. Cu, G. Amy, Factors affecting the
rejection of organic solutes during NF/RO treatmenta literature
review. Water Res. 38, 27952809 (2004)
4. K. Kimura, G. Amy, J.E. Drewes, T. Kim, Y. Watanabe, Rejection
of organic micropollutants (disinfection by-products, endocrine
disrupting compounds, and pharmaceutically active compounds)
by NF/RO membranes. J. Membr. Sci. 227, 113121 (2003)
5. C. Causserand, P. Aimar, J.P. Cravedi, E. Singlande, Dichloro-
aniline retention by nanofiltration membranes. Water Res. 39,
15941600 (2005)
6. Y. Zhang, C. Causserand, P. Aimar, J.P. Cravedi, Removal of
bisphenol A by a nanofiltration membrane in view of drinking
water production. Water Res. 40, 37933799 (2006)
7. Y. Kiso, Y. Nishimura, T. Kitao, K. Nishimura, Rejection prop-
erties of non-phenylic pesticides with nanofiltration membranes.
J. Membr. Sci. 171, 229237 (2000)
8. L.D. Nghiem, J. McCutcheon, A.I. Schafer, M. Elimelech, The
role of endocrine disrupters in water recycling: risk or mania?
Water Sci. Technol. 50(2), 215220 (2004)
9. B. Van der Bruggen, J. Schaep, D. Wilms, C. Vandecasteele,
Influence of molecular size, polarity, charge on the retention of
organic molecules by nanofiltration. J. Membr. Sci. 156, 2941
(1999)
10. J. Radjenovic, M. Petrovic, F. Ventura, D. Barcelo, Rejection of
pharmaceuticals in nanofiltration and reverse osmosis membrane
drinking water treatment. Water Res. 42, 36013610 (2008)
J Porous Mater (2012) 19:721731
11. E. Wittmann, P. Cote, C. Medici, J. Leech, A.G. Turneer,
Treatment of a hard borehole water containing low levels of
pesticides by nanofiltration. Desalination 119(13), 347352
(1998)
12. A.I. Schafer, A.G. Fane, T.D. Waite, Cost factors and chemical
pretreatment effects in the membrane filtration of waters con-
taining natural organic matter. Water Res. 35, 15091517 (2001)
13. B. Van der Bruggen, K. Everaert, D. Wilms, C. Vandecasteele,
Application of nanofiltration for removal of pesticides, nitrate and
hardness from ground water: rejection properties and economic
evaluation. J. Membr. Sci. 193(2), 239248 (2001)
14. Y. Kiso, T. Kon, T. Kital, K. Nishimura, Rejection properties of
alkyl phthalates with nanofiltration membranes. J. Membr. Sci.
182, 205214 (2001)
15. Y. Zhou, S.C. Yu, M.H. Liu, C.J. Cao, Preparation and charac-
terization of polyamide-urethane thin-film composite membranes.
Desalination 180(13), 189196 (2005)
16. J. Ji, M. Mehta, Mathematical model for the formation of film
composite hollow fiber, tubular membranes by interfacial poly-
merization. J. Membr. Sci. 192(12), 4154 (2001)
17. J. Ji, J.M. Dickson, R.F. Childs, B.E. Mccarry, Mathematical
model for the formation of thin-film composite membranes by
interfacial polymerization: porous and dense films. Macromole-
cules 33(2), 624633 (2000)
18. G.Y. Chai, W.B. Krantz, Formation, characterization of poly-
amide membranes via interfacial polymerization. J. Membr. Sci.
93, 175192 (1994)
19. J.M. Gohil, P. Ray, Polyvinyl alcohol as the barrier layer in thin
film composite nanofiltration membranes: preparation, charac-
terization and performance evaluation. J. Colloid Interface Sci.
338(1), 121127 (2009)
20. X. Li, S.D. Feyter, I.F.J. Vankelecom, Poly(sulfone)/sulfonated
poly(ether ketone) blend membranes: morphology study and
application in the filtration of alcohol based feeds. J. Membr. Sci.
324, 6775 (2008)
21. T. He, M. Frank, M.H.V. Mulder, M. Wessling, Preparation and
characterization of nanofiltration membranes by coating polye-
thersulfone hollow fibers with sulfonated poly(ether ketone).
J. Membr. Sci. 307, 6272 (2008)
22. R.H. Lajimi, A.B. Abdallah, E. Ferjani, M.S. Roudesli, A. Der-
atani, Change of the performance properties of nanofiltration
cellulose acetate membranes by surface adsorption of polyelec-
trolyte multilayers. Desalination 163(13), 193202 (2004)
731
23. D.R.J. Willem, R.C.J. Nicolaa, Z. Arie, Semipermeable com-
posite membrane, method for the preparation of such a mem-
brane, and its use, EP0780152 (1997)
24. L. Li, B. Wang, H. Tan, T. Chen, J. Xu, A novel nanofiltration
membrane prepared with PAMAM and TMC by in situ interfacial
polymerization on PEK-C ultrafiltration membrane. J. Membr.
Sci. 269, 8493 (2006)
25. Y. Lu, T. Suzuki, W. Zhang, J.S. Moore, B.J. Marinas, Nanofil-
tration membranes based on rigid star amphiphiles. Chem. Mater.
19, 31943204 (2007)
26. T. Suzuki, Y. Lu, W. Zhang, J.S. Moore, B.J. Marinas, Perfor-
mance characterization of nanofiltrationi membranes based on
rigid star amphiphiles. Environ. Sci. Technol. 41, 62466252
(2007)
27. Y.C. Chiang, Y.Z. Hsub, R.C. Ruann, C.J. Chuang, K.L. Tung,
Nanofiltration membranes synthesized from hyperbranched
polyethyleneimine. J. Membr. Sci. 326, 1926 (2009)
28. X.Z. Wei, L.P. Zhu, H.Y. Deng, Y.Y. Xu, B.K. Zhu, Z.M. Huang,
New type of nanofiltration membranes based on crosslinked hy-
perbranched polymers. J. Membr. Sci. 323, 278287 (2008)
29. J.Z. Yu, Y.Y. Xu, L.P. Zhu, X.Z. Wei, B.K. Zhu, Preparation and
characterization of low-pressure nanofiltration membranes based
on crosslinked hyperbranched polyester. Acta. Polym. Sinica. 7,
707711 (2009)
30. S. Qiu, L. Wu, L. Zhang, H. Chen, C. Gao, Effect of process
conditions of interfacial polymerization on performance of
reverse osmosis composite membrane. J. Chem. Ind. Eng. (China)
59(8), 20272034 (2008)
31. Y. Dai, X.G. Jian, S.H. Zhang, M.D. Guiver, Thin film composite
(TFC) membranes with improved thermal stability from sulfo-
nated poly(phthalazinone ether sulfone ketone) (SPPESK).
J. Membr. Sci. 207(2), 189197 (2002)
32. J. Schaep, B. Van der Bruggen, C. Vandecasteele, D. Willms,
Influence of ion size and charge in nanofiltration. Sep. Purif.
Technol. 14, 155162 (1998)
33. R.J. Petersen, Composite reverse-osmosis nanofiltration mem-
branes. J. Membr. Sci. 83(1), 81150 (1993)
34. E.R. Nightingale, Phenomenological theory of ion salvation:
effective radii of hydrated ions. J. Phys. Chem. 63, 13811387
(1959)
35. R. Du, J. Zhao, Properties of poly(N, Ndimethylaminoethyl
methacrylate)/polysulfone positively charged composite nanofil-
tration membrane. J. Membr. Sci. 239, 183188 (2004)
123