Journal of Membrane Science 491 (2015) 37–44

Contents lists available at ScienceDirect

Journal of Membrane Science

journal homepage: www.elsevier.com/locate/memsci

In-situ combined dual-layer CNT/PVDF membrane

for electrically-enhanced fouling resistance
Shu Wang a, Shuai Liang a,b,n, Peng Liang a, Xiaoyuan Zhang a, Jianyu Sun a, Shijia Wu a,
Xia Huang a,nn
a
State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, China
b
College of Environmental Science and Engineering, Beijing Forestry University, Beijing 100083, China

art ic l e i nf o a b s t r a c t

Article history: Electrically conductive membrane possesses a remarkable performance in antifouling with the
Received 14 January 2015 assistance of applied external electric fields. However, its widespread application is hampered due to
Received in revised form the poor properties of existing conductive membranes, such as low conductivity and inadequate
25 April 2015
strength. The present study demonstrates a novel in-situ combination approach for the fabrication of
Accepted 8 May 2015
Available online 20 May 2015
a highly conductive carbon nanotube (CNT)/polyvinylidene fluoride (PVDF) membrane. During mem-
brane fabrication, the casting solution film was pre-mantled with a CNT layer before the occurrence of
Keywords: polymer precipitation, thereby ensuring a natural and tight in-situ mergence between the CNT layer and
Conductive membrane supporting PVDF. The obtained CNT/PVDF membrane was characterized in terms of physicochemical
PVDF
properties and filtration performance. An electrically-enhanced cross-flow filtration system was devised
Carbon nanotubes (CNT)
to evaluate the antifouling performance of the pristine PVDF and CNT/PVDF membranes. Compared with
In-situ combination
Antifouling the pristine PVDF membrane, the CNT/PVDF membrane possesses a greater electrical conductivity and a
higher permeability. The incorporation of the CNT layer led to a completely different surface composition
in terms of elemental components and functional groups. Being used as a cathode, the CNT/PVDF
membrane exhibited a remarkable antifouling performance in the presence of external electric fields,
and performed better in a 0 V/2 V-pulsed mode as compared with a 2 V-constant mode. This in-situ
combination method provides a simple and practical approach for the fabrication of composite
conductive membranes, suggesting promising applications in various membrane-fabrication areas.
& 2015 Elsevier B.V. All rights reserved.

1. Introduction the commonly negatively charged foulants away from membrane

surface, thus inhibiting membrane fouling [14–16]. However, the
Membrane separation has been widely used in various fields in related configurations were normally complicated and less
the past decades [1,2], such as water treatment [3], food proces- energy-efficient, which limited its widespread application. Recent
sing [4], and drug concentration [5]. However, membrane fouling studies used conductive membranes as electrodes, which shor-
leads to a decrease of membrane permeability and a deterioration tened the distance between cathode and anode thereby reducing
of operation stability. Chemical cleaning processes, which are electric resistance thus lower energy loss [14,17,18]. Consequently,
extensively used for fouling removal, increase operating costs similar antifouling effects were achieved with simplified config-
and shorten membrane lifespan, thereby hindering the wide- urations [14,17–21].
spread application of membrane technology [6–11]. The properties of the conductive membranes have a key effect on
External electric field has been holding a unique position in the antifouling performance. In general, there are two kinds of
mitigating membrane fouling for half a century [12]. During conductive membranes: pure conductive membranes and composite
filtration, an electric field could offer electrostatic forces (reported conductive membranes. The conductive membranes made of pure
to be 1000 times larger than Brownian force [13]) which repulse conductive materials are normally weak in mechanical strength and
poor in filtration performance. In contrast, composite conductive
membranes achieved to unite great electrical conductivity, strong
n
Corresponding author at: State Key Joint Laboratory of Environment Simulation mechanical strength, and efficient filtration performance, demon-
and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, strating a great potential for practical applications. A lot of effort had
China. Tel.: þ86 10 88257606.
nn
Corresponding author. Tel.: þ 86 10 62772324.
been made to develop practical strategies for the fabrication of
E-mail addresses: [email protected] (S. Liang), composite conductive membranes. To date, two main strategies—
[email protected] (X. Huang). blending and surface coating—have been reported for the fabrication

http://dx.doi.org/10.1016/j.memsci.2015.05.014
0376-7388/& 2015 Elsevier B.V. All rights reserved.
38 S. Wang et al. / Journal of Membrane Science 491 (2015) 37–44

of composite conductive membranes [22–25]. However, for the layer also functions as a separation barrier retarding the release of
blending strategy, the conductivity of the resultant membrane was CNT into the environment. During this fabrication, the free area of
normally limited due to the insulating nature of the non-conductive the casting solution which was not covered by the CNT paper
matrix material. For the coating strategy, the durability of the coated formed into pristine PVDF membranes. These membranes were
layer of conductive materials was normally questionable. used as control. All the procedures were carried out at room
Carbon nanotubes (CNTs), owing to their superior electrical temperature (23 71 1C)
conductivity and physicochemical stability, have drawn much
attention as an ideal conductive material [26–32]. Polyvinylidene
2.2. Membrane characterization
fluoride (PVDF), as one of the most extensively used membrane
materials, is well known for its outstanding mechanical strength,
Resistances of the CNT paper, pristine PVDF membranes, and
chemical resistance, and thermal stability [33–35]. A suitable
CNT/PVDF membranes were measured with an Autolab potentio-
integration of CNT and PVDF could give rise to a practical
stat (PGSTAT 128N, Metrohm Autolab, Netherlands). For each
composite conductive membrane which possesses both the advan-
membrane type, at least two samples were taken for measure-
tageous properties of CNT and PVDF.
ments with the same dimension of 15 mm  5 mm. For each
The current study proposed a novel in-situ combination
sample, measurements were performed in two mutually orthogo-
method for the fabrication of a novel dual-layer CNT/PVDF con-
nal directions (each at least in triplicate). Electrical conductivities
ductive membrane. The obtained membrane was systematically
(κ) were calculated using the following equation:
characterized in terms of electrical conductivity, hydrophilicity,
filtration performance, morphology, and chemical compositions. L
κ¼ ð1Þ
Also, the influence of different external electric fields on the RA
antifouling performance of the CNT/PVDF membrane was system- where L is the sample length in the testing direction, A is the
atically evaluated with a series of filtration tests. corresponding cross-sectional area, and R is the measured resistance
(in Ohms). Sample thicknesses were observed by scanning electron
microscope (SEM, Quanta 200F, FEI, USA) at three random positions.
2. Materials and methods Water contact angles of the pristine PVDF and CNT/PVDF
membranes were measured on the basis of the sessile-drop
2.1. Membrane fabrication method with a video-supported contact angle measuring instru-
ment (OCA 20, Dataphysics, Germany). To prepare samples, the
CNT/PVDF membranes were fabricated via phase separation membranes were vacuum-dried (LGJ-12, Beijing Song Yuan Hua
with a modified protocol as illustrated in Fig. 1. Casting solution Xia Co. Ltd., China) and cut into several strips with a size of
was prepared according to the procedures described previously  1 cm  4 cm. Each strip was placed on a separate glass slide with
[36,37], with a recipe containing 17 g of PVDF (Solef 6010, Solvay S. double-side tapes. For contact angle calculations, the Ellipse Fitting
A., Belgium), 0.5 g of polyvinylpyrrolidone (PVP, MW  40 kDa, method was used in the fitting of the shape profile of the water
Amresco), and 82.5 g of N-methy-2-pyrrolidone (NMP, Sinopharm droplet on the sample surface. The final result for each membrane
Chemical Reagent Co., Ltd., China). To fabricate the membranes, a type was averaged from at least 12 measurements at random
liquid film of the casting solution was applied onto a glass plate at locations on at least two separately fabricated membranes.
a thickness of 250 μm using a doctor blade (Elcometer 3530, UK). Membrane permeability and selectivity were evaluated with a
Immediately, a multiwalled CNT paper (areal weight  20 g m  2, dead-end filtration system as previously described in detail [36]. In
diameter  200 nm by scanning electron microscope (SEM), brief, membrane permeability (m s  1 kPa  1) was test with Milli-Q
NanoTechLabs Inc., USA) that served as the conductive layer was water at 15 kPa for 10 min [38]. Molecular cutoff (MWCO) of the
mildly covered onto the casting solution film under the action of membranes was analyzed with sequential filtration tests to eval-
its own gravity (Fig. 1). After a 30 s contact, the plate was uate membrane selectivity. Different poly(ethylene oxide) (PEO,
immersed in a deionized (DI) water (Milli-Q reference ultrapure Aldrich) solutions (1 g L  1, 20 kDa, 100 kDa, 200 kDa, 300 kDa,
purification system, Millipore, USA) base, thereby triggering the 400 kDa and 600 kDa) were carefully prepared and used for
crystallization of PVDF polymers. After a 12 min period of immer- filtration tests. Prior to the tests, all membranes were pre-
sion, the obtained membrane was rinsed with ample DI water and compacted with Milli-Q water at 20 kPa for 20 min. For each
finally transferred to a fresh DI-water bath at 4 1C. During the membrane type, both the permeability and MWCO tests were
period before being immersed in the water base, the casting performed at least in duplicate.
solution would slightly infiltrate into the CNT paper. After the Membrane morphology was observed by SEM. Before observa-
immersion step, the precipitated PVDF polymers from the infil- tion, membrane samples were vacuum-dried and sputter-coated
trated casting solution would act as cramps thereby firmly holding with  4-nm-thick layer of Au/Pd (precision etching and coating
the CNT paper and PVDF membrane together. Such an intensive system, 682, Gatan, USA). To obtain a cross-section view, mem-
physical force is considerably strong and durable, reducing the brane samples were fractured in liquid nitrogen to minimize the
potential risk from CNT release. Moreover, the supporting PVDF damage to membrane structure.

Fig. 1. Schematic protocol for the fabrication of an in-situ combined dual-layer carbon nanotube (CNT)/polyvinylidene fluoride (PVDF) membrane.
 S. Wang et al. / Journal of Membrane Science 491 (2015) 37–44 39

X-ray photoelectron spectroscopy (XPS, ESCALAB 250Xi, Ther- temperature (23 71 1C) for at least 24 h. As the humic acid was
mofisher) and Fourier transform infrared spectrometer (FT-IR, difficult to dissolve, there always exist some undissolved residues
Thermo Scientific, USA) were used to investigate the elemental with different size in the obtained solution, rendering the solution
composition and functional groups of the membrane surface. To heterogeneous. As a result, it is almost impossible to control the
prepare samples, membranes were vacuum-dried at room tem- quality consistency of the solutions that were introduced in each
perature (23 71 1C). filtration tests, thereby impairing the comparison fairness
between different tests. Therefore, prior to experiments, the
foulant solution was pre-filtered with a glass fiber membrane
2.3. Evaluation of antifouling performance
(GF/F, Whatman, UK) to remove the undissolved residues.
2.3.1. Electrically-enhanced cross-flow filtration system
An electrically-enhanced cross-flow filtration cell (as displayed 2.3.3. Filtration tests
in Fig. 2A) was devised by incorporating a conductive module into Filtration tests were performed with the electrically-enhanced
a traditional cross-flow filtration cell. The core component and its cross-flow filtration system. For each membrane type, filtration tests
function principle were illustrated in Fig. 2B. The dual-layer CNT/ were carried out in three different modes of applied external electric
PVDF or pristine PVDF membrane (served as a control) was packed field relating to different applied voltage of 0 V, 1 V DC, pulsed 0 V/1 V
beneath a titanium rectangular ring which was extended outside DC, 2 V DC, and pulsed 0 V/2 V DC. For each mode, three cycles of
the cell and connected to a negative electrode, thereby turning the filtration runs were performed with a fresh unstained membrane
membrane into a cathode. Silicone seal rings were packed sample. Prior to the first cycle, the membrane was installed in the
between the titanium ring and anode to avoid any leaching. For filtration cell and pre-compacted with Milli-Q water at a constant
the filtration runs with the CNT/PVDF membrane, the CNT side is permeation flux of 200 L m  2 h  1 (LMH) until the TMP reached a
set up facing the anode, which is the optimized configuration plateau. After then, foulant solution was introduced into the filtration
according to a previous study [19]. A titanium foil (served as an cell, thereby initiating membrane fouling. After a 1 h period of fouling
anode) was set opposite the membrane surface, allowing a 2-mm- filtration, the foulant solution was replaced with Milli-Q water, and a
high space for the influent passing through. The effective mem- 5 min period of membrane cleaning was performed at the end of
brane area for filtration is 1 mm  4 mm. one filtration cycle with the cross-flow shear of water. Immediately,
To set up an electrically-enhanced cross-flow filtration system, two more cycles of filtration were carried out. Hydrostatic pressures
an Autolab potentiostat (CHI660D, Ch Instruments, China) was of the circulating feed solution and effluent were recorded to
employed as the power (Fig. 2B) to apply controlled electric field calculate TMP.
(via altering applied voltage among 0 V, 1 V DC, 0 V/1 V DC pulsed
with 5 s intervals, 2 V DC, and 0 V/2 V DC pulsed with 5 s intervals)
between the membrane and titanium foil. Peristaltic pumps 3. Results and discussion
(BT100-1F, Longer Pump, China) were used to circulate feed
solution and drive the circulating solution to pass through the 3.1. Electrical conductivity
membrane. The variations of the transmembrane pressure (TMP)
were recorded with a digital recorder (XSR90). Fig. 3 presents the differences of electrical conductivity among
the pristine PVDF membrane, CNT paper, and CNT/PVDF mem-
brane (see detailed calculation data in Supporting information
2.3.2. Model foulants and synthetic foulant solution
Table S1). The pristine PVDF membrane possesses an extremely
To simulate practical feed solutions (e.g., supernatant in an
MBR [39]), three types of model foulants and several inorganic
Table 1
matter were selected to prepare a synthetic foulant solution Chemistry of the synthetic foulant solution (determined on the basis of the
(Table 1) on the basis of an extensive investigation [40,41]. previously published investigation) [40,41].
Sodium alginate (SA, Aldrich), bovine serum albumin (BSA,
Items Component Concentration
66 kDa, Sigma−Aldrich), and humic acid were used to represent
polysaccharides, proteins, and natural organic matter (commonly Model foulants (mg L 1
) Sodium alginate 15
known as the major components of the foulants in wastewater). Bovine serum albumin 4
MgCl2 (hexahydrate, ≥99%, Sigma−Aldrich) and CaCl2 (dihydrate, Humic acid 8
≥99%, Sigma–Aldrich) were added to provide water hardness.
Solution pH was buffered with NaHCO3 (Sigma). Ionic strength of Inorganic components (mM) NaCl 9
the solution was adjusted by the addition of NaCl (crystals, J.T. NaHCO3 2
CaCl2 1
Baker, USA). Microbe growth was inhibited with NaN3 to minimize
MgCl2 0.5
the loss of organic matter during the experiments. All the NaN3 2
above chemicals were vigorously dissolved in DI water at room

Fig. 2. (A) A photograph of the filtration cell. (B) Schematic of the electrically-enhanced cross-flow filtration cell for evaluating the antifouling performance of the PVDF and
CNT/PVDF membranes. A titanium foil was used as an anode.
40 S. Wang et al. / Journal of Membrane Science 491 (2015) 37–44

Fig. 3. Electrical conductivities of the pristine PVDF membrane, CNT paper, and
CNT/PVDF membrane measured in two mutually orthogonal directions (illustrated
as x and y). After measuring the resistance of the membrane, the electrical Fig. 4. Water permeability and surface contact angles of the pristine PVDF and
conductivity was calculated through membrane thickness and sample size. Each CNT/PVDF membranes. Each value is an average of measurements on at least three
value is an average of at least three measurements. separately fabricated membranes. The contact angle measurements for each
membrane were conducted on at least six random locations.

low conductivity (too low to be noticed in Fig. 3) due to its un-

conductive nature. On the contrary, the CNT paper, as an ideal
conductor, possesses a high conductivity of  118 S cm  1 in both
directions (x, y, marked in Fig. 3), which render the dual-layer
CNT/PVDF membrane highly conductive (  105 S cm  1). Com-
pared with the CNT paper, the CNT/PVDF membrane underwent
a slight decrease in conductivity due to the presence of PVDF.
However, the influence of the PVDF layer on the conductivity could
be negligible owing to the special dual-layer structure and the
comparable thickness of the CNT layer (will be further discussed in
Section 3.3), suggesting a remarkable advantage over other fabri-
cation methods for conductive membranes [42]. Moreover, minor
difference was observed between the x and y directions (Fig. 3) for
both CNT paper and CNT/PVDF membrane, demonstrating an
excellent ability to transmit electrons.
Fig. 5. Molecular weight cutoff (MWCO, indicated with the dashed line at 90%
rejection) of the pristine PVDF and CNT/PVDF membranes, evaluated with poly
3.2. Filtration performance and membrane hydrophilicity (ethylene oxide) (PEO) of various MWs ranging from 20 kDa to 600 kDa.

The filtration performance of an asymmetric membrane which

was fabricated by non-solvent induced phase separation was PVDF membrane is fairly comparable in thickness with the PVDF
directly determined by membrane surface properties (e.g., pore layer (see details in Section 3.3), thus completely blocking the
size distribution and porosity). The incorporation of the CNT layer influence of the beneath PVDF and PVP on surface hydrophilicity.
during the CNT/PVDF membrane fabrication would undoubtedly Therefore, the hydrophilicity of the CNT/PVDF membrane was
change the surface property of the original PVDF surface thereby mainly governed by the CNT layer.
giving rise to different filtration performance. Membrane filtration The hydrophobic property of the CNT/PVDF membrane could
performance could be characterized with two indexes: water be attributed to the low surface energy of CNT [43], which could
permeability (representing the performance of solvent permea- benefit the mitigation of membrane fouling in the cross-flow
tion) and MWCO (representing the performance of solute filtration mode. The adsorption of foulants on a low-surface-
rejection). energy surface is normally weak, and could be easily rinsed off
Compared with the pristine PVDF membrane, the CNT/PVDF by the shearing force of the cross-flowing feed solution (see details
membrane obtained an about 10% increase in water permeability in Section 3.5).
(Fig. 4). Meanwhile, the MWCO results (Fig. 5) indicated an MWCO In general, the CNT layer played a key role on determining the
increase due to the incorporation of the CNT layer. Therefore, the overall filtration performance of the CNT/PVDF membrane.
permeability increase of the CNT/PVDF membrane could be
attributed to pore-size enlargement and/or surface porosity 3.3. Membrane morphology
increase due to the presence of the CNT layer. Moreover, the
variation curve of the MWCO for the CNT/PVDF membrane The SEM pictures of the top and cross-section views of the
exhibited a relatively sharper leap than that for the pristine PVDF pristine PVDF membrane and CNT/PVDF membrane are shown in
membrane. It demonstrated that the presence of the CNT layer Fig. 6. The pristine PVDF membrane exhibited an asymmetric
endowed the original PVDF membrane with better membrane structure comprising a dense upper layer and a loose lower layer,
selectivity. However, considering the retention rate of 600 kDa which is a common feature for the membranes fabricated by NIPS.
PEO by pure CNT paper was only  6.5%, we believe that the The super-thin upper layer possessed numerous pores ( 30–
retention ability of the CNT/PVDF membrane was mainly contrib- 50 nm, Fig. 6A) thus serving as a selective layer. The lower
uted by the supporting PVDF layer. supporting layer developed a great amount of finger-like voids
The incorporation of the CNT layer also caused a dramatic (Fig. 6B), thereby contributing to the lower membrane resistance
change in surface hydrophilicity. As illustrated in Fig. 4, the CNT/ thus higher permeability. This finger-like asymmetric structure
PVDF membrane possessed a water contact angle of 1201 due to could be attributed to the use of water as the coagulation bath,
the hydrophobic nature of CNT, much higher than the 581 for the which leaded to a rapid liquid–liquid demixing process during the
pristine PVDF membrane. Additionally, the CNT layer of the CNT/ membrane fabrication [33].
 S. Wang et al. / Journal of Membrane Science 491 (2015) 37–44 41

Fig. 6. Scanning electron microscope (SEM) views of the pristine PVDF and CNT/PVDF membranes: (A) top view, pristine PVDF; (B) cross view, pristine PVDF; (C) top view,
CNT/PVDF; and (D) cross view, CNT/PVDF.

In contrast, the CNT/PVDF membrane exhibited a completely respectively, were detected in the spectra for the pristine PVDF
different top view (Fig. 6C) due to the coverage of the CNT paper. A membrane (Fig. 7A). The distributions of these peaks were
net structure formed by huge quantities of multi-walled carbon associated with the constituents of PVDF molecules, and consis-
nanotubes were observed on the CNT/PVDF membrane surface. tent with the results in our previous study [36]. In contrast, the C
The CNT layer accounted for more than 30% in thickness of the peak was the dominating energy peak in the spectra for the CNT/
whole membrane (Fig. 6D), resulting in a total blockage of the PVDF membrane (Fig. 7B), which is consistent with the CNT
underlying PVDF layer. Moreover, considering the rejection ability composition. Moreover, due to the blockage of the thick CNT layer,
of the CNT layer, it is inferred that the rejection performance of the little signal of F was detected in the CNT/PVDF spectra. This is
CNT/PVDF membrane is probably governed by the interface consistent with all the above analyses.
between the CNT layer and PVDF layer. The PVDF layer of the Surface functional groups of the pristine PVDF and CNT/PVDF
CNT/PVDF membrane also exhibited a different cross-section view. membranes were analyzed through FT-IR. As presented in Fig. 8,
Compared with the pristine PVDF membrane (Fig. 6B), the CNT/ few absorption peaks were detected in the range from 3000 cm−1 to
PVDF membrane developed fewer finger-like voids but more 3800 cm−1 in both spectra, indicating the absence of any aromatic
macro-voids and sponge-like structures. This could be attributed structures or unsaturated fatty chains on both the pristine PVDF and
to the delayed liquid–liquid demixing caused by the hindrance of CNT/PVDF membrane surfaces. For the pristine PVDF membrane,
the CNT layer to water during the membrane fabrication [33]. In several main peaks were observed at 877 cm  1, 1180 cm  1, and
addition, a natural and tight mergence of the CNT and PVDF layers 1402 cm  1, which were associated with the vibration of C–F, C–O,
was observed in Fig. 6D. The CNT layer endowed the final and C–N groups, respectively. These groups were consistent with the
membrane with great electrical conductivity, while the PVDF layer molecular formula of PVDF and PVP. On the contrary, due to the
supported the CNT layer with enhanced mechanical strength. This presence of the CNT layer, no obvious peak was detected in the CNT/
in-situ combination method offers a novel strategy for merging PVDF spectra, which is consistent with the FTIR results of a pure CNT
the advantageous properties of different materials. paper (Supporting information Fig. S1) and other reported FTIR results
for CNT [44,45]. The FTIR results well matched the XPS analyses,
3.4. Chemical composition of membrane surfaces demonstrating the complete composition difference between the
PVDF and CNT/PVDF membranes.
The CNT/PVDF membrane surface possessed a completely
different chemical composition due to the presence of CNT. The 3.5. Anti-fouling performance
elemental compositions of the membrane surfaces were analyzed
by XPS. As presented in Fig. 7A, four main energy peaks, repre- Antifouling performance of the pristine PVDF and CNT/PVDF
senting fluorine (F), oxygen (O), nitrogen (N), and carbon (C), membranes was evaluated through a three-cycle-filtration test
42 S. Wang et al. / Journal of Membrane Science 491 (2015) 37–44

Fig. 7. X-ray photoelectron spectroscopy (XPS) analyses of the pristine PVDF and CNT/PVDF membrane surfaces. (A) XPS survey scan of the pristine membrane (black) and
CNT/PVDF membrane (red). (B) Atomic concentration of carbon (C), fluorine (F), oxygen (O), and nitrogen (N) at the surface of the two different membranes. (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 8. Fourier transform infrared spectroscopy (FTIR) analyses of the pristine PVDF
and CNT/PVDF membranes.

using the synthetic foulant solution (Table 1). Different electric

fields corresponding with different voltages (0 V, 1 V DC, pulsed
0 V/1 V DC, 2 V DC, and pulsed 0 V/2 V DC) were applied between
the membrane surface and the titanium foil during the filtration to
evaluate the effect of applied electric fields on fouling mitigation.
For the 0 V condition (i.e., power off, Fig. 9A), compared with
the pristine PVDF membrane, the CNT/PVDF membrane main-
tained a relatively lower TMP, suggesting a slower fouling devel-
opment. This could be attributed to the low surface energy of the
CNT layer, which weakened the adhesion of foulants on the CNT/
PVDF surface. However, as the filtration went on, the pre-absorbed
foulants kept contributing to the increase of the surface energy,
thereby facilitating the foulant accumulation. Correspondingly, the
TMP for the CNT/PVDF membrane rose to the same level with that
for the PVDF membrane in the following two filtration cycles,
indicating an unsustainable antifouling performance.
For both the 1 V DC (see Supporting information Fig. S2) and
2 V DC conditions (Fig. 9B), the CNT/PVDF membrane always Fig. 9. Variations of transmembrane pressure (TMP) for the pristine PVDF and CNT/
maintained a lower TMP than both the pristine PVDF membrane PVDF membranes in different electrically-enhanced fouling filtrations: (A) No
electric field applied; (B) 2 V DC; and (C) 0 V/2 V DC, pulsed with 5 s intervals.
and the CNT/PVDF membrane under the 0 V condition. This
improvement in antifouling was directly related with the applied
electric field. With the CNT/PVDF membrane serving as a cathode, antifouling performance. Therefore, the membranes were further
we hypothesize that the foulants in the feed solution were driven tested in a pulsed condition.
by the force of electric field away from the membrane, thereby For the 0 V/1 V pulsed (Fig. S2) and 0 V/2 V pulsed conditions
significantly decreasing the tendency of foulant absorption. Con- (Fig. 9C), the CNT/PVDF membrane achieved further improve-
sequently, the conductive CNT/PVDF membrane obtained a ments in antifouling as compared with its performance in the
remarkable antifouling ability in applied electric field. However, corresponding DC conditions. The CNT/PVDF membrane always
there is reasonable concern about neutralization by cations which kept a lower TMP during the three cycles of filtration, indicating
might dissipate the effect of applied electric field on antifouling. remarkable fouling resistance. This result supported the above
Correspondingly, a temporary pause to allow the release of speculation about cation neutralization. As the foulants were
absorbed cations might give rise to a better and/or sustainable normally negatively charged, they would move toward the anode
 S. Wang et al. / Journal of Membrane Science 491 (2015) 37–44 43

while the cations toward the cathode membrane. The cations compared with the pristine PVDF membrane. Moreover, the CNT/
would be continuously absorbed on the charged membrane sur- PVDF membrane performed better in antifouling in the 0 V/2 V
face when the electric field exists, resulting in an electric neu- pulsed electric field than in the constant 2 V filed. This in-situ
tralization of the membrane thus receded electric field. Therefore, combination method provides a simple strategy for fabrication of
a suspension of the applied electric field could offer an opportu- functional membranes, suggesting a promising future for practical
nity for cation release, thereby recovering the receded repulsive applications. More efforts should be made to further improve its
force. Meanwhile, the released ions would contribute to the practical application value from several aspects, such as CNT-
increase of ionic strength, which might also favor the recovery of dosage reduction, regulation of membrane-rejection, and explora-
the applied electric field. With such an intermittent charging tion for novel conductive materials.
operation, the antifouling performance could be maintained for a
long period.
Additionally, the antifouling-performance improvement for the Acknowledgments
2 V scale is much more significant than that for the 1 V scale. This
could be related to the additional effect of water electrolysis that This research was financially supported by the Key Program of
caused by the 2 V voltage (standard potential difference for water the National Natural Science Foundation of China (No. 51238004),
electrolysis is 1.23 V). Water electrolysis could generate a con- and Program for Changjiang Scholars and Innovative Research
siderable amount of bubbles (Supporting information Fig. S3) Team in University (No. IRT1152).
which hindered the foulant adsorption thereby mitigating mem-
brane fouling.
Moreover, after membrane cleaning in the first filtration cycle, Appendix A. Supporting information
the difference of the TMP-curve slopes between the pristine and
CNT/PVDF membranes decreased for all electric-field conditions, Supplementary data associated with this article can be found in
suggesting a weakened resistance for initial fouling for the CNT/ the online version at http://dx.doi.org/10.1016/j.memsci.2015.05.014.
PVDF membrane. This could be due to foulant accumulation on the
internal and/or external membrane surfaces after the first filtra- References
tion cycle, which counteracted the effect of the intrinsic mem-
brane properties on antifouling thus leading to similar TMP slopes. [1] E. Drioli, F. Macedonio, Membrane engineering for water engineering, Ind.
However, owing to the repulsion effect of the applied electric field, Eng. Chem. Res. 51 (2012) 10051–10056.
[2] W.F. Blatt, M.P. Feinberg, H.B. Hopfenberg, C.A. Saravis, Protein solutions:
the CNT/PVDF membrane still maintained a much lower TMP in
concentration by a rapid method, Sci. Technol. Weld. Join. 150 (1965) 224–226.
the following filtration processes. The relatively slower increasing [3] A.L. Lim, R. Bai, Membrane fouling and cleaning in microfiltration of activated
rate of the TMP for the CNT/PVDF membrane also resulted in lower sludge wastewater, J. Membr. Sci. 216 (2003) 279–290.
final TMP for the CNT/PVDF membrane than the PVDF membrane [4] L. Lin, K.C. Rhee, S.S. Koseoglu, Bench-scale membrane degumming of crude
vegetable oil: process optimization, J. Membr. Sci. 134 (1997) 101–108.
with the 2 V pulsed condition, as different increasing rates would [5] N. Capelle, P. Moulin, F. Charbit, R. Gallo, Purification of heterocyclic drug
attribute to a different final TMP within the same filtration period derivatives from concentrated saline solution by nanofiltration, J. Membr. Sci.
(1 h). 196 (2002) 125–141.
[6] P. Le-Clech, V. Chen, T.A.G. Fane, Fouling in membrane bioreactors used in
Additionally, for all the three conditions, no obvious difference wastewater treatment, J. Membr. Sci. 284 (2006) 17–53.
was observed for the filtration performance of the pristine PVDF [7] J. Zhang, H.C. Chua, J. Zhou, A.G. Fane, Factors affecting the membrane
membrane, indicating that the enhanced antifouling property was performance in submerged membrane bioreactors, J. Membr. Sci. 284 (2006)
54–66.
due to the conductive surface rather than the titanium ring [8] M. Kumar, M. Ulbricht, Novel antifouling positively charged hybrid ultrafiltra-
electrode or other factors. For all the filtration tests, no foulant tion membranes for protein separation based on blends of carboxylated
accumulation was observed on the Ti foil. This might be attributed carbon nanotubes and aminated poly(arylene ether sulfone), J. Membr. Sci.
448 (2013) 62–73.
to its highly hydrophilic nature and smooth surface morphology. [9] N. Yamato, K. Kimura, T. Miyoshi, Y. Watanabe, Difference in membrane
Also, no detachment of the CNT layer was observed in either fouling in membrane bioreactors (MBRs) caused by membrane polymer
the crossflow filtration tests or membrane-flushing processes, materials, J. Membr. Sci. 280 (2006) 911–919.
[10] D. Rana, T. Matsuura, Surface modification for antifouling membranes, Chem.
indicating good membrane strength and durability. As demon-
Rev. 110 (2010) 2448–2471.
strated in Section 2.1, the CNT paper was covered on the liquid film [11] S.P. Hong, T.H. Bae, T.M. Tak, S. Hongb, A. Randall, Fouling control in activated
of the casting solution before polymer precipitation. The binding sludge submerged hollow fiber membrane bioreactors, Desalination 143
between the CNT paper and PVDF layer was ensured by the (2002) 219–228.
[12] S.P. Moulik, F.C. Cooper, M. Bier, Forced-flow electrophoretic filtration of clay
intensive and tight physical mergence, which could be consider- suspensions-filtration in an electric field, J. Colloid Interface Sci. 24 (1967)
ably stable in filtration operations. 427–432.
[13] J.H. Masliyah, S. Bhattacharjee, Electrokinetic and Colloid Transport Phenom-
ena, John Wiley & Sons, New York, 2006.
[14] S.N. Jagannadh, H.S. Muralidhara, Electrokinetics methods to control mem-
4. Conclusions brane fouling, Ind. Eng. Chem. Res. 35 (1996) 1133–1140.
[15] J.P. Chen, C.Z. Yang, J.H. Zhou, X.Y. Wang, Study of the influence of the electric
field on membrane flux of a new type of membrane bioreactor, Chem. Eng. J.
A conductive dual-layer CNT/PVDF membrane was successfully 128 (2007) 177–180.
fabricated through a novel in-situ combination method. A CNT [16] K. Bani-Melhem, M. Elektorowicz, Development of a novel submerged
layer was incorporated on the casting solution film before polymer membrane electro-bioreactor (SMEBR): performance for fouling reduction,
Environ. Sci. Technol. 44 (2010) 3298–3304.
precipitation, thereby ensuring a natural and tight in-situ mer- [17] S.S. Madaeni, S. Molaeipour, Investigation of filtration capability of conductive
gence between the conductive layer and supporting polymers. The composite membrane in separation of protein from water, Ionics 16 (2010)
incorporation of the CNT layer endowed the membrane with 75–80.
[18] A. Holder, J. Weik, J. Hinrichs, A study of fouling during long-term fractiona-
greater electrical conductivity, higher permeability, and comple-
tion of functional peptides by means of cross-flow ultrafiltration and cross-
tely different surface chemical composition. The conductive CNT flow electro membrane filtration, J. Membr. Sci. 446 (2013) 440–448.
layer allowed the CNT/PVDF membrane to function as a cathode, [19] Q.Y. Zhang, C.D. Vecitis, Conductive CNT–PVDF membrane for capacitive
and facilitate the generation of different applied electric fields for organic fouling reduction, J. Membr. Sci. 459 (2014) 143–156.
[20] A.V. Dudchenko, J. Rolf, K. Russell, W. Duan, D. Jassby, Organic fouling
antifouling. With the assistance of applied electric field, the CNT/ inhibition on electrically conducting carbon nanotube–polyvinyl alcohol
PVDF membrane obtained much better antifouling ability as composite ultrafiltration membranes, J. Membr. Sci. 468 (2014) 1–10.
44 S. Wang et al. / Journal of Membrane Science 491 (2015) 37–44

[21] C.-F. de Lannoy, D. Jassby, K. Gloe, A.D. Gordon, M.R. Wiesner, Aquatic [35] K.H. Oshima, T.T. EvansStrickfaden, A.K. Highsmith, E.W. Ades, The use of a
biofouling prevention by electrically charged nanocomposite polymer thin microporous polyvinylidene fluoride (PVDF) membrane filter to separate
film membranes, Environ. Sci. Technol. 47 (2013) 2760–2768. contaminating viral particles from biologically important proteins, Biologicals
[22] R. Saranya, G. Arthanareeswaran, D.D. Dionysiou, Treatment of paper mill 24 (1996) 137–145.
effluent using polyethersulfone/functionalised multiwalled carbon nanotubes [36] S. Liang, Y. Kang, A. Tiraferri, E.P. Giannelis, X. Huang, M. Elimelech, Highly
based nanocomposite membranes, Chem. Eng. J. 236 (2014) 369–377. hydrophilic polyvinylidene fluoride (PVDF) ultrafiltration membranes via
[23] L.F. Liu, J.D. Liu, G. Bo, F.L. Yang, J. Crittenden, Y.S. Chen, Conductive and postfabrication grafting of surface-tailored silica nanoparticles, ACS Appl.
hydrophilic polypyrrole modified membrane cathodes and fouling reduction Mater. Interfaces 5 (2013) 6694–6703.
in MBR, J. Membr. Sci. 429 (2013) 252–258. [37] S. Liang, G.G. Qi, K. Xiao, J.Y. Sun, E.P. Giannelis, X. Huang, M. Elimelech,
[24] J.W. Lee, J. Jung, Y.H. Cho, S.K. Yadav, K.Y. Baek, H.B. Park, S.M. Hong, C.M. Koo, Organic fouling behavior of superhydrophilic polyvinylidene fluoride (PVDF)
Fouling-tolerant nanofibrous polymer membranes for water treatment, ACS ultrafiltration membranes functionalized with surface-tailored nanoparticles:
Appl. Mater. Interfaces 6 (2014) 14600–14607. implications for organic fouling in membrane bioreactors, J. Membr. Sci. 463
[25] Y. Baek, C. Kim, D.K. Seo, T. Kim, J.S. Lee, Y.H. Kim, K.H. Ahn, S.S. Bae, S.C. Lee, (2014) 94–101.
J. Lim, K. Lee, J. Yoon, High performance and antifouling vertically aligned [38] B. Tansel, W.Y. Bao, I.N. Tansel, Characterization of fouling kinetics in ultra-
carbon nanotube membrane for water purification, J. Membr. Sci. 460 (2014) filtration systems by resistances in series model, Desalination 129 (2000)
171–177. 7–14.
[26] S. Iijima, Helical microtubules of graphitic carbon, Nature 354 (1991) 56–58. [39] K. Xiao, Y. Xu, S. Liang, T. Lei, J.Y. Sun, X.H. Wen, H.X. Zhang, C.S. Chen,
[27] B. Sakintuna, Y. Yurum, Templated porous carbons: a review article, Ind. Eng. X. Huang, Engineering application of membrane bioreactor for wastewater
Chem. Res. 44 (2005) 2893–2902. treatment in China: current state and future prospect, Front. Environ. Sci. Eng.
[28] E.M. Hotze, T. Phenrat, G.V. Lowry, Nanoparticle aggregation: challenges to 8 (2014) 805–819.
understanding transport and reactivity in the environment, J. Environ. Qual. [40] Y.X. Shen, K. Xiao, P. Liang, J.Y. Sun, S.J. Sai, X. Huang, Characterization of
39 (2010) 1909. soluble microbial products in 10 large-scale membrane bioreactors for
[29] T. Forati, M. Atai, A.M. Rashidi, M. Imani, A. Behnamghader, Physical and municipal wastewater treatment in China, J. Membr. Sci. 415–416 (2012)
mechanical properties of graphene oxide/polyethersulfone nanocomposites, 336–345.
Polym. Adv. Technol. 25 (2014) 322–328. [41] X. Huang, K. Xiao, Y.X. Shen, Recent advances in membrane bioreactor
[30] C.D. Vecitis, M.H. Schnoor, M.S. Rahaman, J.D. Schiffman, M. Elimelech, technology for wastewater treatment in China, Front. Environ. Sci. Eng. China
Electrochemical multiwalled carbon nanotube filter for viral and bacterial 4 (2010) 245–271.
removal and inactivation, Environ. Sci. Technol. 45 (2011) 3672–3679. [42] C.F. de Lannoy, D. Jassby, D.D. Davis, M.R. Wiesner, A highly electrically
[31] G. Gao, C.D. Vecitis, Electrochemical carbon nanotube filter oxidative perfor- conductive polymer-multiwalled carbon nanotube nanocomposite membrane,
mance as a function of surface chemistry, Environ. Sci. Technol. 45 (2011) J. Membr. Sci. 415 (2012) 718–724.
9726–9734. [43] H.S. Uhm, H. Yong Cheol, S. Dong Hun, C. Soon Cheon, Surface transformation
[32] W.Y. Duan, A. Dudchenko, E. Mende, C. Flyer, X.B. Zhu, D. Jassby, Electro- of carbon nanotube powder into super-hydrophobic and measurement of
chemical mineral scale prevention and removal on electrically conducting wettability, Chem. Phys. Lett. 427 (2006) 390–393.
carbon nanotube–polyamide reverse osmosis membranes, Environ. Sci.— [44] N.M. Meman, M. Pourkhalil, A. Rashidi, B. ZareNezhad, Synthesis, character-
Process. Impacts 16 (2014) 1300–1308. ization and operation of a functionalized multi-walled CNT supported MnOx
[33] L. Fu, N.A. Hashim, Y. Liu, M.R.M. Abed, K. Li, Progress in the production and nanocatalyst for deep oxidative desulfurization of sour petroleum fractions,
modification of PVDF membranes, J. Membr. Sci. 375 (2011) 1–27. J. Ind. Eng. Chem. 20 (2014) 4054–4058.
[34] S. Munari, A. Bottino, G. Capannelli, Casting and performance of polyvinyli- [45] M. Raja, Surface modification of carbon nanotubes with combined UV and
dene fluoride based membranes, J. Membr. Sci. 16 (1983) 181–193. ozone treatments, Fuller. Nanotub. Carbon Nanostruct. 23 (2015) 11–16.