Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
Preparation and Characterization of a Novel HydrophilicPVDF/PVA UF Membrane Modified by CarboxylatedMultiwalled Carbon Nanotubes
Gui-E Chen,1 Sun-Jie Xu,1 Zhen-Liang Xu,2 Wei-Wei Zhu,1 Qiong Wu,1 Wei-Guang Sun1
1 School of Chemical and Environmental Engineering, Shanghai Institute of Technology, 100 Haiquan Road,Shanghai 201418, China2 State Key Laboratory of Chemical Engineering, Membrane Science and Engineering R&D Lab, ChemicalEngineering Research Center, East China University of Science and Technology, 130 Meilong Road,Shanghai 200237, China
Polyvinylidene fluoride (PVDF)/polyvinyl alcohol (PVA) ultra-filtration (UF) membranes were prepared via a phase inver-sion method employing the modification of carboxylatedmultiwalled carbon nanotubes (MWCNTs-COOH). Variouscontents of MWCNTs-COOHs (0.00–0.15 wt%, weight ofcasting solution) were added into PVDF/PVA/dimethyl sulf-oxide systems for the fabrication of the plate UF mem-brane. Fourier transform infrared spectroscopy spectraidentified the successful introduction of carboxyl throughthe C@O peak at 1730 cm21. Scanning electron micros-copy images exhibited the external surface and the asym-metric morphology with the appearance of a sponge-likeinner structure. Atomic force microscopy analysis deter-mined the roughness values and rougher topography. Thehydrophilicity of the composite membrane containing 0.09wt% of MWCNTs-COOHs improved the most. This samplehas the highest pure water flux, approximately doubled(126.6 L m22 h21) compared to the PVDF/PVA membrane(68.6 L m22 h21), an enhanced bovine serum albumin fluxrecovery rate, showing an increase of 17%, and the bestfouling resistance ability. Meanwhile, the porosity anddynamic contactangle also indicate the enhancement ofmembrane hydrophilicity. Dextran (DEX) 600k rejectionreached 91.0%. Break strength, elongation at break, andYoung’s modulus also had improvements of 60%, 215.5%,and 56.7%, respectively, when the MWCNTs-COOH contentwas 0.12 wt%. POLYM. ENG. SCI., 56:955–967, 2016. VC 2016Society of Plastics Engineers
INTRODUCTION
Over the last dozen years, polymeric membranes have been
widespread in engineering applications and scientific usage [1],
such as the polyamide (PA) membrane for seawater treatment
and ethanol purification [2, 3], the polycarbonate (PC)
membrane for gas separation [4], the polyacrylonitrile (PAN)
membrane for polluted natural rubber effluent treatment [5], and
other polymeric membrane materials such as polytetrafluoroeth-
ylene (PTFE), polysulfone (PSF), and polyethersulfone (PES)
[6–8]. Polyvinylidene fluoride (PVDF) is regarded as a typical
example of a polymeric membrane material with outstanding
characteristics, such as being acid-proof, alkali-proof, corrosion-
resistant, and having excellent mechanical properties. Due to its
low molecular surface energy causing its strong hydrophobicity,
almost all studies have been carried out using either physical or
chemical modification methods to enlarge the pure water flux,
promote the surface anti-fouling capacity of the PVDF mem-
brane, and above all, enhance its hydrophilicity [9–12].
A simple, effective, and low-cost method is to blend quanti-
fied materials, which have hydrophilic functional groups such as
carboxyl, hydroxyl, and sulfonyl, into the casting solution to
prepare the composite membrane [13].
Polyvinyl alcohol (PVA) is a familiar polymer, which is non-
toxic, innocuous, and degradable, making it safe for the environ-
ment. Moreover, it has particular oil resistance, abrasive
resistance, solvent resistance, especially for hydrocarbon com-
pounds, and a remarkable membrane-forming ability [14, 15].
Since it was first discovered in 1924 by Hemalm, PVA has been
widely employed in food, pharmaceuticals, paper making, con-
struction, agriculture, and polymer materials [16, 17]. In the
structure of PVA, the molecular unit arrangement is linear with
a hydroxyl group on each molecular unit, so it can easily form
hydrogen bonds. Many researchers utilize PVA in diverse condi-
tions to prepare versatile membranes [18–21]. Zhang et al. [22]
used PVA as a hydrophilic additive in a PVDF/PES blend mem-
brane; their results indicated that when PVA content reached 0.3
wt% of the casting solution, the prepared membrane had the
best porosity, the lowest water contact angle of membrane sur-
face and improved pure water flux. Moreover, the extended Der-
jaguin–Landau–Verwey–Overbeek (XDLVO) theory also proved
the feasibility that adding PVA to the blend membrane system
can improve membrane surface hydrophilicity. Li et al. [23]
used a dip-coating method to immerse an alkali-treated PVDF
hollow fiber membrane into a self-prepared PVA-TiO2 disper-
sion and used glutaraldehyde (GA) as cross-linker to fabricate a
PVA/PVDF-TiO2 hollow fiber membrane. The results showed
that the presence of PVA induced no rejection effect for salts.
When the content of TiO2 was increased to 1 g L21, the modi-
fied membrane reached the highest performance, exhibiting an
outstanding antifouling property and thermal stability. In addi-
tion, it showed excellent separation efficiency in dye wastewater
Correspondence to: G.E. Chen; e-mail: [email protected]
Contract grant sponsor: National Natural Science Foundation of China; con-
tract grant numbers: 51372153 and 51572176; contract grant sponsor: Shang-
hai Union Program; contract grant number: LM201249; contract grant
sponsor: Key Technology R&D Program of Shanghai Committee of Science
and Technology in China; contract grant number: 14231201503; contract
grant sponsor: the 2013 Year Special Project of the Development and Indus-
trialization of New Materials of National Development and Reform Commis-
sion in China; contract grant number: 20132548; contract grant sponsor: Key
Technology R&D Program of Jiangsu Committee of Science and Technol-
ogy in China; contract grant number: BE2013031.
DOI 10.1002/pen.24325
Published online in Wiley Online Library (wileyonlinelibrary.com).
VC 2016 Society of Plastics Engineers
POLYMER ENGINEERING AND SCIENCE—2016
treatment. Li et al. [24] blended PVDF with various ratios of
PVA. The best ratio of PVDF:PVA is 80:20, for which the pre-
pared composite hollow fiber membrane showed a sharp
decrease of the water contact angle, and the hydrophilicity of
membrane surface increased extensively. Furthermore, they dis-
covered that the PVDF/PVA blends were incompatible, causing
the formation of interface microvoids in the membrane structure,
which benefit the membrane properties.
With the rapid development of nanomaterials, carbon nano-
materials have become a significant research focus in recent
years [25]. Carbon materials include a large family of different
species with various properties. Commonly, people are familiar
with graphene, graphene oxide (GO), carbon nanotubes, fuller-
enes and their derivatives, which are also classified into single-
walled or multiwalled or single-layer or multilayer in terms of
their structure [26]. Zhao et al. [27] added different contents of
GO and PVP into a PVDF membrane, followed by optimization
through the Taguchi design. This study determined that the best
content of PVDF is 12 wt%, GO is 3 wt%, and PVP is 5 wt%,
producing membranes showing the best properties. The intro-
duction of GO changed the characterization of the membrane
surface and improved the hydrophilicity, antifouling ability, and
mechanical properties. Wu et al. [28] used PSF as matrix mem-
brane, blended with a self-made “sandwich” structure SiO2-GO,
which was a derivative of graphene oxide. The SiO2-GO par-
ticles were well fixed into the membrane with a regular distribu-
tion. Analysis indicated that with an SiO2-GO addition of 0.3
wt%, the performance of the hybrid membrane surpassed that of
SiO2/PSF and GO/PSF membranes overall. The pure water flux
reached a peak value, which was nearly double compared with
the PSF membrane. Moreover, the hybrid membrane maintains
rejection above 98% to egg albumin. Carboxylated multiwalled
carbon nanotubes (MWCNTs-COOH) are the functionalized
derivative of carbon nanotubes. After some carbon atoms on the
nanotube wall have been carboxylated, properties of the carbox-
ylic MWCNTs are enhanced compared to pristine ones, includ-
ing mechanical performance, thermal stability, conductivity,
solubility, and hydrophilicity [29]. In addition, this process is
FIG. 1. (a) Particle size distribution and (b) SEM image of MWCNTs-COOH. [Color figure can be viewed in the
online issue, which is available at wileyonlinelibrary.com.]
956 POLYMER ENGINEERING AND SCIENCE—2016 DOI 10.1002/pen
simple to achieve [30]. Mahendra et al. [31] used multiblock
copolymers APAES and PAES with different contents of
MWCNTs-COOH to prepare a polymer ultrafiltration membrane
with the properties of porosity, hydrophilicity, a positive charge,
and an asymmetrical structure. They achieved a maximum pure
water flux and the best permeation performance when the
MWCNTs-COOH content was 2% of the polymer weight. Wang
et al. [32] modified polymethylmethacrylate (PMMA) and poly-
urethane (PU) matrix membranes with P-MWCNTs and
MWCNTs-NH2, respectively. The results implied that both pris-
tine MWCNTs and functionalized MWCNTs could improve
membrane performance effectively, but the functionalized
MWCNTs have better results than the pristine MWCNTs. Thus,
in this study, functionalized MWCNTs were adopted. While car-
bon nanomaterials are widely used in membrane modification,
the utilization of MWCNTs-COOH together with a hydrophilic
polymer to modify the membrane system is seldom reported.
In this study, a PVDF ultrafiltration (UF) membrane with
PVA and MWCNTs-COOH is prepared by the phase inversion
method [33–37]. The prepared plate UF membranes are ana-
lyzed through scanning electron microscopy (SEM), Fourier
transform infrared spectroscopy (FT-IR), pure water flux,
dynamic contact angle, porosity, polyethylene glycol (PEG) and
dextran (DEX) rejection, bovine serum albumin (BSA), flux
recovery rate (FRR%), and mechanical properties to illustrate
modified UF membrane capacity [38–40].
EXPERIMENTAL
Materials
Polyvinylidene fluoride (PVDF, SolefVR
6010) powder was
obtained from Solvay Advanced Polymers, L.L.C (Alpharetta
GA, USA). Polyvinyl alcohol (PVA, 99.8� moL/mol21, RG),
and dimethyl sulfoxide (DMSO, purity� 99.0%, AR) used as
the solvent were purchased from Shanghai Titan Scientific Co.,
Ltd. Polyethylene glycols in four molecular weights (PEG, MW
2 kg mol21, MW 4 kg mol21, MW 6 kg mol21, MW 10 kg
mol21) were purchased from Sinopharm Chemical Reagent Co.,
Ltd. Dextran (DEX, MW 600 kg mol21) was supplied by
Sigma-Aldrich Co., Ltd. The bovine serum albumin (BSA, MW
67 kg mol21) used in the flux recovery test was purchased from
Shanghai Lianguan Biochemical Reagent Company. All the
materials and reagents were without any chemical treatment
before use.
Carboxylated multiwalled carbon nanotubes, defined as
“MWCNTs-COOH” in this study, with 2.0 wt% of “ACOOH”
were purchased from Soochow Hengqiu Graphene Technology
Co., Ltd. The MWCNTs-COOH have an average surface area of
250–300 m2 g21 and an average length of 10–30 lm with outer
diameters and inner diameters in the ranges of approximately
20–30 and 5–10 nm, respectively. A laser particle size analyzer
(BT-9300Z) was used to measure the detailed information,
which includes the particle size distribution. As shown in Fig.
1a, the particle size of MWCNTs-COOH is under 30 lm, which
corresponds to the given length. Additionally, the median parti-
cle diameter obtained from the analyzer, which is 7.55 lm, sig-
nifies that 50% of the MWCNTs-COOH particles were larger
than 7.55 lm, and the other half were smaller than 7.55 lm. In
other words, this indicates significant dispersibility and less
agglomeration. Figure 1b shows the SEM image of MWCNTs-
COOH; the small size and long cylindrical morphology are ben-
eficial to the membrane performances [41, 42].
Preparation of Modified Composite Membranes
The PVDF/PVA/MWCNTs-COOH blend UF membrane was
prepared by the phase inversion method. The contents of the
membrane material and other reagents are listed in Table 1, in
which “MC” is the designation of this series of membrane,
“9010” represents the mass ratio of PVDF and PVA, and the
other three figures indicate the contents of MWCNTs-COOH in
the casting solution. For example, MC 9010 003 expresses the
composite membrane that contains 0.03 wt% MWCNTs-COOH
in the casting solution, and the mass ratio of PVDF to PVA is
90:10.
To briefly describe the preparation method, different amounts
of MWCNTs-COOHs and a constant 82 g DMSO were first
placed into a 250 mL conical flask. Shaking and ultrasonic treat-
ment for 10 min ensured that the MWCNTs-COOH were fully
dispersed. Then, polymer membrane materials were added.
PVDF powder and PVA grains were dried in the oven before
use for 12 h at 908C and 608C, respectively. The polymer con-
tent was set at 18 wt% in which PVDF:PVA is 90:10 by weight.
Then, strong mechanical stirring was carried out at 988C for
12 h so that the PVA could dissolve entirely, and a homogene-
ous casting solution was obtained. After complete degassing in
the oven at 608C for 12 h, a glass rod with fine copper wires
was used to transfer a film onto a glass plate. The glass plate
with a film of casting solution was immersed into deionized
water immediately. When peeled off from the glass plate, a
200-lm-thick membrane was successfully produced and then
stored in a container. The deionized water was changed twice a
day for 1 week to ensure the removal of residual solvent.
Finally, resultant membranes were stored in the container with
deionized water until needed.
TABLE 1. Compositions of different PVDF/PVA composite UF membranes.
Membrane no.
Membrane materials (wt%)MWCNTs-COOH
(wt%)
DMSO
(wt%) Coagulation bathPVDF PVA
MC 9010 000 16.2 1.8 0.00 82.0 Deionized water
MC 9010 003 16.2 1.8 0.03 82.0
MC 9010 006 16.2 1.8 0.06 82.0
MC 9010 009 16.2 1.8 0.09 82.0
MC 9010 012 16.2 1.8 0.12 82.0
MC 9010 015 16.2 1.8 0.15 82.0
DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—2016 957
Characterization of Modified Membranes
A scanning electron microscope (SEM, Hitachi S-3400N,
Japan) was used to observe the external surface and cross-
sectional structure of the modified membrane. Before observation,
the membrane was fractured in liquid nitrogen and fixed onto the
sample platform. Both surface and cross-sectional samples were
sprayed to form a thin platinum layer to make the samples con-
ductive. FT-IR (Thermo Nicolet iZ10, USA) was carried out to
investigate the functional groups and chemical composition of the
modified binary membrane, particularly to identify the introduc-
tion of carboxyl in the membrane. The wavenumber range of
4000–500 cm21 was observed. Atomic force microscopy (AFM,
Veeco, NanoScope IIIa Multimode AFM, USA) was carried out
to reveal the external surface topography and roughness of the
composite UF membranes using tapping mode. The membranes
were cut into small pieces and adhered to the microslide, and
then fixed onto the sample stage. The scanning area was selected
as 5 3 5 lm. Three-dimensional diagrams with average rough-
ness (Ra), root mean square roughness (Rms), and average maxi-
mum height (Rz) values were reported.
The hydrophilicity and permeation performance of the modi-
fied membranes were clarified though pure water flux, bovine
serum albumin (BSA, Mw 5 67,000) flux recovery rate (FRR%),
and the rejection of PEGs and DEX 600k. These events were
tested through self-made membrane permeation performance
equipment with an effective area of 38.5 cm2. All events were
tested at room temperature and under a pressure of 0.1 MPa.
After preloading the resultant modified membrane with
deionized water for 30 min until steady flux, the pure water flux
was obtained. In general, the pure water flux was defined as
Jw5V
A3t(1)
where Jw is the pure water flux (L m22 h21), V is the volume
of permeated water (L), A is the effective area of the equipment
(m2), and t is the time (h).
Subsequently, the BSA FRR% was tested. In other words,
this value reflects the cleaning effect of the membrane. The
higher the FRR% is, the higher the hydrophilicity of the mem-
brane is, which indicates a membrane with a low fouling degree
to a large extent. All the BSA solutions used in this study were
quantified BSA directly dissolved in deionized water to prepare
a 0.3 g L21 BSA aqueous solution (pH 5 7.0). The polluted flux
(Jp) was obtained by preloading the same modified membrane
used above with the 0.3 g L21 BSA aqueous solution for 30
min to make the membrane completely polluted. Then, the pol-
luted membrane was rinsed with sufficient deionized water and
immersed into the deionized water for 30 min, followed by
another wash. Then, the membrane was preloaded with deion-
ized water until the flux was steady. Recover flux (Jr) was
obtained, and the FRR% was calculated though the equation:
FRR%5Jr
Jw3100% (2)
where FRR% is the flux recovery rate (%) and Jr is the recover
flux (L).
The reversible fouling ratio (Rr%) and irreversible fouling
ratio (Rir%), which describe the concentration polarization
induced fouling and adsorption of proteins and organic pollutant
induced fouling, respectively, were also calculated from
Rr%5Jr2Jp
Jw
� �3100% (3)
Rir%5Jw2Jr
Jw
� �3100% (4)
where Rr% is the reversible fouling rate (%), Rir% is the irrevers-
ible fouling ratio (%), and Jp is the polluted flux (L m22 h21).
The sum of the reversible fouling rate and irreversible foul-
ing ratio described by the total fouling ratio (Rt%) was also cal-
culated to clarify the total flux loss through
Rt%5 12Jp
Jw
� �3100% (5)
Each test was carried out three times, and the average value was
recorded.
The same equipment was used to test the rejection of PEG in
different molecular weights and DEX 600k through another wet
membrane. PEG (1 g L21) or DEX (0.5 g L21) aqueous solution
were fed and permeated through the wet membrane under the
pressure of 0.1 MPa for 30 min; the permeation solution and
feed solution were collected, followed by a total organic carbon
(TOC) analyzer (TNM-1, Shimadzu, Japan) test. The concentra-
tions of the permeation solution and feed solution were obtained
and recorded as CP and CF, respectively. Thus, the rejection was
calculated as follows:
R%5CF2CP
CF3100% (6)
where R is the rejection (%), CF is the feed solution concentra-
tion (mg L21), and CP is the permeation solution concentration
(mg L21).
After that, the pore size distribution and molecular weight
cut-off (MWCO) were also obtained by a series of PEG rejec-
tion results though calculation on the log-normal distribution
function using Matlab software [43]. The log-normal distribution
function was defined as
f dð Þ51= ln r½ �dffiffiffiffiffiffi2pp� �
exp 2 1=2ð Þ ln d=l½ �=ln r½ �ð Þ2� �
(7)
After antifouling measurements, dynamic contact angle studies
were also carried out to clarify the membrane surface hydrophi-
licity. When a drop of pure water touched different membrane
surfaces, the rate of change in the contact angle was found to be
strongly related to hydrophilicity. Better hydrophilicity induces
faster decaying speeds [44]. Here, a contact angle analyzer
(JC2000D, Shanghai Zhongcheng Digital Technology Apparatus
Co., Ltd., China) with a sample injector was applied for this
test. A droplet, approximately 2 lL, was placed onto the modi-
fied membrane top surface, and images were collected with a
frequency of 4 s each until the droplet did not change. The
droplet was measured with the software, and the values were
recorded. To minimize the error, each membrane was tested five
times, and its average value was calculated.
Porosity is also an important parameter that can help explain-
ing some permeation and mechanical phenomena of the
958 POLYMER ENGINEERING AND SCIENCE—2016 DOI 10.1002/pen
composite membrane. The porosity was obtained by the gravi-
metric method. The samples were cut to fixed sizes and weighed
when both wet and dry. Calculations were made according to
Ref. 45:
e5Ww2Wd
qwAl(8)
where Ww is the weight of the wet membrane (g), Wd is the
weight of the dry membrane (g), qw is the density of water
(0.998 g cm23), A is the effective area of the binary membranes
(cm2), and l is the thickness of the composite membranes (cm).
Each membrane was measured five times, and the average value
was reported.
Break strength, elongation at a break, and Young’s Modulus
are the three most important parameters of membrane mechani-
cal properties. Tests were carried out by an electronic mechani-
cal property analyzer (QJ210A, Shanghai Qingji Instrument
Technology Co., Ltd. China) with two tongs in the vertical posi-
tion. The membranes were cut into strips and fixed onto the
tongs. The test started at a speed of 50 mm�min21; after the
series of tests, the reported values were averaged and recorded.
RESULTS AND DISCUSSION
FT-IR of PVDF/PVA Membrane and Modified PVDF/PVAMembranes
Figure 2 shows the spectra of the unmodified PVDF/PVA
membrane (MC 9010 000) and PVDF/PVA membranes modi-
fied with various MWCNTs-COOH contents (0.03 wt%, 0.06
wt%, 0.09 wt%, 0.12 wt%, 0.15 wt%, weight of casting solu-
tion). Obviously, the wide bands over 3100–3600 cm21 corre-
spond to the OAH band. This could be ascribed to the
introduction of PVA, but the absorbencies are very weak, which
may be because of the water solubility of PVA. During the
membrane forming process, a small amount of PVA attached to
PVDF precipitated through the phase inversion process, while
the majority of the PVA exchanged deionized water with
DMSO. In other words, the PVA in the casting solution worked
as the pore-forming agent. For the absorbance range of 2,920–
2,853 cm21, the ACH2A asymmetric stretching vibration is
present in both PVDF and PVA. In the comparison of those
modified membranes, all spectra show the stretching vibration
of C@O at 1,730 cm21, which becomes stronger. This phenom-
enon could be explained by observing that as the MWCNTs-
COOH content increases, the concentration of C@O in the
membrane also increases. This is strong evidence for the
enhancement of the surface hydrophilicity of modified mem-
branes. Meanwhile, the adsorption peak at 3,100–3,600 cm21
corresponding to OAH hardly changed, which implies that the
introduction of MWCNTs-COOH has no interaction with PVA
in the composite membranes. Next, the strong peak at
1,070 cm21 belongs to the CAF band, which is exclusive to
PVDF, and those peaks in the range of 840–1,400 cm21 demon-
strate that the polymeric crystal species is in the b-phase.
Morphology of the Modified PVDF/PVA Membrane
Figure 3 displays the SEM images of the PVDF/PVA mem-
brane and the modified PVDF/PVA membranes. The letter “A”
on the image panels refers to the external surface of each
composite membrane, while the letter “B” refers to the cross-
section of each composite membrane. Distinctly, all membranes
have a traditional asymmetric structure consisting of a compact
skin layer and a variational sublayer, which could be further
divided into the “upper part” finger-like structure and the “under
part” sponge-like structure. Compared to the unmodified PVDF/
PVA membrane (MC 9010 000), all modified PVDF/PVA mem-
branes show much denser “under part” structures, and with
increasing content of MWCNTs-COOH in the membrane
matrix, the “under part” becomes denser still. When the
MWCNTs-COOH content surpasses 0.12 wt%, the “under part”
is almost transformed into the sponge-like structure. This may
be because the presence of MWCNTs-COOH acted to densify
the “reticular-structure” induced by PVA in the membrane. This
effect endowed the modified membranes with greater mechani-
cal properties, which were also proven in later section. From the
external surface morphology, we can identify the appearance of
a bit macrovoid structure. This might result in the increase of
the pure water flux, which is attributed to the synergistic effect
of the MWCNTs-COOH and the different mutual solubilities
between PVDF and PVA. Quantified MWCNTs-COOH in the
casting solution can accelerate the exchange velocity between
the solvent DMSO and the nonsolvent deionized water during
the phase inversion process due to its hydrophilicity, which
might result in the formation of macrovoid structures. Other-
wise, the incompatibility between PVDF and PVA would cause
their phase separation, which is also conducive to this process.
Figure 3g Membrane “MC 9010 009 A” shows the most porous
structure, with no macrovoid structures. Moreover, the exchange
velocity was controlled at the most appropriate rate for this con-
tent level, and this kind of morphology provides the membrane
with the highest level of performance.
AFM Analysis of Different PVDF/PVA Composite UF Membranes
Figure 4 and Table 2 list the three-dimensional diagrams of
the top surface of the composite UF membrane and its rough-
ness values (Ra, Rms, and Rz). All modified PVDF/PVA compos-
ite membranes show rougher topography than the pristine
FIG. 2. FT-IR spectra of PVDF/PVA membrane and modified PVDF/PVA
membranes with various contents of MWCNTs-COOH. [Color figure can be
viewed in the online issue, which is available at wileyonlinelibrary.com.]
DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—2016 959
PVDF/PVA membrane. In terms of the material properties,
rougher surface topography corresponds to greater hydrophilicity
[46]. On the other hand, average roughness (Ra), root mean
square roughness (Rms), and average maximum height (Rz) val-
ues were also larger than the values for the pristine PVDF/PVA
membrane. This is especially true for MC 9010 009, for which
the values of Ra, Rms, and Rz are the largest, reflecting the
lumpy membrane surface roughness with enlarged effective
filtration area and the best hydrophilicity of all the modified
PVDF/PVA membranes. Unfortunately, after the MWCNTs-
COOH content surpassed 0.12 wt%, the MWCNTs-COOH con-
tent in the casting solution could not move to the interface
between the solvent and nonsolvent adequately during the phase
inversion process due to the high concentration. Thus, the
hydrophilic functional groups fixed on the top surface of the
membranes “MC 9010 012” and “MC 9010 015” were
FIG. 3. SEM images of different PVDF/PVA composite UF membranes: (a) external surface; (b) cross-section.
960 POLYMER ENGINEERING AND SCIENCE—2016 DOI 10.1002/pen
marginally less than the membrane “MC 9010 009” and resulted
in the reduction of surface roughness. Furthermore, deeper
“valleys” and steeper “peaks” were observed from all of the
composite PVDF/PVA membranes containing MWCNTs-
COOH. MC 9010 009 is the most obvious one, which may due
to the concurrence of PVA and MWCNTs-COOH. This can be
explained by two reasons. First, PVA is a crystalline polymer;
during the demixing process, the “peaks” grow as the PVA crys-
tal grows. Second, the peculiarity of the strong hydrophilicity of
this material can accelerate the demixing speed, resulting in the
formation of “valleys” and “peaks” in which the morphology
may enhance the membrane hydrophilicity and permeability.
Pure Water Flux
Pure water flux is measured to evaluate the permeation ability
of the modified composite membranes. As shown in Fig. 5 and
Table 3, the unmodified PVDF/PVA membrane was also tested as
a comparison. The hydrophobic characteristics of PVDF restrict
the pure water flux at a low level. After blending, the pure water
flux increased evidently. With an increase of the MWCNTs-
COOH content, the pure water flux also grows. When the
MWCNTs-COOH content reaches 0.09 wt%, the pure water flux
of the modified membranes is at its maximum (126.6 L m22 h21),
which is almost double compared to the unmodified membranes
(68.6 L m22 h21). Schematic diagram of introduction of
MWCNTs-COOH in the PVDF/PVA UF membrane is placed in
Fig. 6. The credible explanation for this is the synergistic effect of
MWCNTs-COOH and PVA. On one hand, with the addition of
PVA, macrovoid structure exists due to its hydrophilicity, and the
pore channels of the inner structure become wider, which can
increase the opportunity for water molecule inflow. These results
are in accordance with the SEM analysis (Fig. 3). On the other
hand, due to the immobilization of MWCNTs-COOH into the
membrane matrix, the abundant carboxyls disperse on the surface
of the modified membranes uniformly, resulting in the enhance-
ment of membrane surface hydrophilicity. Additionally, the cylin-
drical structure of MWCNTs-COOH may supply tunnels for water
molecules, making it easier to pass through. Otherwise, from the
AFM analysis, the largest effective filtration area also contributes
significantly to the flux. However, the pure water flux did not keep
the trend of monotonically increasing after the content of
MWCNTs-COOH exceeded 0.09 wt%; it exhibited a small decline
but still retained at a satisfactory value. Because of the solubility
of MWCNTs-COOH in the solvent DMSO, when the concentra-
tion of MWCNTs-COOH in the casting solution is high, some
small portion of the MWCNTs-COOH may agglomerate due to
the intermolecular interactions, which may somewhat influence the
performances of these membranes. Additionally, the high value of
porosity (82.4%) measured through the gravimetric method is
shown in Fig. 5. The variation tendencies are almost the same as
the pure water flux, which can also illustrate the effects of the
usage of MWCNTs-COOH.
Antifouling Ability
The antifouling ability of modified PVDF/PVA membranes
was estimated through the BSA FRR%. The BSA aqueous solu-
tion was prepared by directly dissolving quantified BSA in
deionized water to prepare the 0.3 g L21 BSA aqueous solution
at a pH value of 7.0. It is the rate of recovery flux (Jr) and pure
water flux (Jw) in which a higher FRR% corresponds to a better
cleaning effect of the membrane. The results are presented in
Fig. 5. The modified composite membrane, which has an
FIG. 3. Continued.
DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—2016 961
optimal antifouling ability when the MWCNTs-COOH content
in the casting solution is 0.09 wt%, has an FRR% of 81.2%.
Compared to MC 9010 000, it has an improvement of approxi-
mately 17%. This may be attributed to the introduction of
MWCNTs-COOH with sufficient carboxyls; the membrane sur-
face with plenty of carboxyls can obstruct most pollutants out-
side the compact skin layer. In other aspects, with the
enhancement of surface hydrophilicity, the pollutant is easy to
FIG. 4. AFM three-dimensional diagrams of different PVDF/PVA composite UF membranes. [Color figure can be
viewed in the online issue, which is available at wileyonlinelibrary.com.]
TABLE 2. Roughness values of different PVDF/PVA composite UF membranes.
Membrane no. MC 9010 000 MC 9010 003 MC 9010 006 MC 9010 009 MC 9010 012 MC 9010 015
Ra (nm) 6.8 7.9 13.9 15.2 14.0 14.3
Rms (nm) 9.0 11.5 17.7 19.4 19.1 18.4
Rz (nm) 39.7 41.8 62.0 64.9 60.6 60.6
962 POLYMER ENGINEERING AND SCIENCE—2016 DOI 10.1002/pen
wash out using deionized water from the membrane surface,
especially for a protein pollutant. For further investigation, as
shown in Fig. 7, the fouling resistance ratios change identically
with the FRR%. MC 9010 009 has the highest reversible fouling
ratio and the lowest total fouling ratio, representing the best
membrane surface antifouling ability among the modified com-
posite membranes [47]. Moreover, the MWCNTs-COOH modi-
fied composite membranes have higher reversible ratios and
lower irreversible fouling ratios than the pure membrane, except
MC 9010 012, for which the reversible ratio is 1.7% lower, but
its total fouling ratio was decreased by 5.8%. This may be
induced by the increase of membrane surface roughness; an
enhanced hydrophilic surface can defend against a majority of
pollutants, but a small part of them can be trapped in the
“valleys” and wrapped on the “peaks” of the membrane surface.
Therefore, the PVDF/PVA membrane modified with 0.09 wt%
of MWCNTs-COOH must represent the optimal loading. Addi-
tionally, the variation tendency of the FRR% and fouling resist-
ance ratios are consistent with the phenomenon reflected by the
membrane surface dynamic contact angle in the next section.
Dynamic Contact Angle
The dynamic contact angle is an important parameter for
assessing the surface hydrophilicity of a membrane. Generally,
researchers consider a lower contact angle value desirable.
The dynamic contact angle variation trend in 120 s is listed in
Fig. 8. An unmodified PVDF/PVA membrane has the highest
contact angle value, next to it is the modified PVDF/PVA
FIG. 5. The relationship between MWCNTs-COOH content, pure water
flux, recover flux, flux recovery rate, and porosity. [Color figure can be
viewed in the online issue, which is available at wileyonlinelibrary.com.]
TABLE 3. Permeability performances of different UF membranes.
Membrane no.
Pure water flux
(L�m22�h21)
BSA Polluted
flux (L�m22�h21)
Recovery flux
(L�m22�h21)
Flux recover
rate (%)
Reversible
fouling
rate (%)
Irreversible
fouling
ratio (%)
Total
fouling
ratio (%)
MC 9010 000 68.6 6 2.9 28.2 6 1.2 47.5 6 1.7 69.4 6 4.8 28.2 6 4.6 30.6 6 4.8 58.8 6 2.5
MC 9010 003 91.6 6 9.2 42.7 6 4.7 70.1 6 9.0 76.4 6 4.5 29.7 6 8.4 23.6 6 4.5 53.3 6 3.8
MC 9010 006 105.0 6 7.3 51.2 6 2.8 84.3 6 9.0 80.2 6 4.6 31.4 6 6.2 19.8 6 4.6 51.2 6 1.7
MC 9010 009 127.0 6 7.4 62.1 6 2.9 103.0 6 2.7 81.2 6 3.8 32.2 6 1.4 18.8 6 3.8 51.0 6 3.0
MC 9010 012 103.0 6 7.5 48.2 6 3.3 75.6 6 4.6 73.5 6 2.6 26.5 6 6.1 26.5 6 2.6 53.0 6 5.0
MC 9010 015 97.5 6 8.1 45.5 6 4.5 74.0 6 8.0 75.8 6 2.1 29.1 6 2.8 24.2 6 2.1 53.3 6 1.8
FIG. 6. Schematic diagram of introduction of MWCNTs-COOH in the PVDF/PVA UF membrane. [Color figure
can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—2016 963
membrane with only 0.03 wt% additive, which is designated
MC 9010 003. Membrane MC 9010 006 and MC 9010 015 are
superior to MC 9010 003 but inferior to MC 9010 009, which
has the smallest contact angle value. From these curves, we can
find that for the contact angles of the PVDF/PVA membrane
modified with different MWCNTs-COOH contents, the contact
angle decreased in varying degrees. This indicates that the intro-
duction of MWCNTs-COOH could improve the surface hydro-
philicity of composite membranes. Furthermore, with the
increase of MWCNTs-COOH content, the contact angle value
decreases at first, followed by a small increase, similar to the
other performance parameters. When the MWCNTs-COOH con-
tent is 0.09 wt%, the contact angle of the unmodified PVDF/
PVA membrane decreases from 60.18 to 38.98 at the period of
the 30th second, reaches a minimum and then becomes steady,
which reflects the optimal situation of surface hydrophilicity.
The mechanism could be interpreted as follows: MWCNTs-
COOH is hydrophilic, and during the membrane preparation
process, the abundant carboxyl was fixed into the membrane
and surface. At the same time, as the MWCNTs-COOH content
increases, it leads to an increase of surface roughness, and the
higher surface roughness facilitates the enhancement of mem-
brane surface hydrophilicity.
FIG. 7. Fouling resistance ratio of PVDF/PVA membrane and modified
PVDF/PVA membranes with various contents of MWCNTs-COOH. [Color
figure can be viewed in the online issue, which is available at wileyonlineli-
brary.com.]
FIG. 8. Dynamic contact angles of PVDF/PVA membrane and modified
PVDF/PVA membranes with various contents of MWCNTs-COOH. [Color
figure can be viewed in the online issue, which is available at wileyonlineli-
brary.com.]
FIG. 9. The relationship between MWCNTs-COOH content, pure water
flux, and DEX 600k rejection.
FIG. 10. (a) Cumulative pore size distribution curves and (b) probability
density function curves of various modified membranes. [Color figure can
be viewed in the online issue, which is available at wileyonlinelibrary.com.]
964 POLYMER ENGINEERING AND SCIENCE—2016 DOI 10.1002/pen
Pore Size, Pore Size Distribution, and MWCO of VariousMembranes
The liquid–liquid displacement porosimetry (LLDP) method
was used by many researchers to acquire pore size distributions
and molecular weight cut-off (MWCO) values due to its precise,
accurate, and reliable nature [48–50]. However, in this study, a
simple and efficient method was used to investigate the pores of
the composite membranes [51–53]. Figure 9 clarifies the rela-
tionship between MWCNTs-COOH content, pure water flux,
and DEX 600k rejection. Figure 10 displays the cumulative pore
size distribution curves and probability density function curves
of various binary membranes. The rejection data of PEGs,
which are used for the curve fitting, and other parameters are
shown in Table 4. From the figures, we can clearly determine
the larger mean effective pore size of membrane MC 9010 000,
which was not modified by MWCNTs-COOH. Compared to the
unmodified PVDF/PVA membrane, all of those modified PVDF/
PVA membranes possess smaller pore sizes and narrow pore
size distributions. These phenomena may be attributed to the
introduction of MWCNTs-COOH into the membrane matrix.
With the increase of the MWCNTs-COOH content, the mean
effective pore size first decreases then increases. This trend is
similar to porosity and the rejection of PEGs and DEX 600k,
but it is a reverse relationship. Unusually, in terms of the rejec-
tion of DEX 600k and pure water flux, membrane MC 9010
009 has the highest values. This can be explained in terms of
the smallest mean effective pore size at the highest quantities
and the auxiliary hydrophilic MWCNTs-COOH fixed on the
membrane contributing a great deal to the high pure water flux
and rejection values. Otherwise, according to the DEX rejection,
the modified membranes exhibit differences in MWCO between
measured values and computation values. This discrepancy can
be explained as follows: the introduction of MWCNTs-COOH
into the casting solution influenced the membrane formation
process, and it accelerated the exchange velocity between the
solvent and nonsolvent. Moreover, the strong solvent DMSO
also can maintain the speed at a high level, which may result in
the production of large-sized pores on the membrane surface,
but it makes little contribution to the flux at extremely small
amount. Therefore, the emerging pores offer a favorable channel
for the small solute to pass through, but it is just the size to pre-
vent the passage of the DEX 600k solute.
Mechanical Properties
The data of the break strength, elongation at the break, and
Young’s modulus are shown in Table 5. The obtained data indi-
cate that with the addition of MWCNTs-COOH, the three
mechanical property parameters of the modified PVDF/PVA
membranes all improved. As the MWCNTs-COOH content in
the casting solution increases, the three parameters increase until
the loadings reach to 0.12 wt%. This phenomenon disagreed
with the speculation in terms of permeation properties. The fol-
lowing reasons may account for this. In one aspect, with the
introduction of one-dimensional MWCNTs-COOH with long
and rigid structures, the PVDF chain wrapped with the well-
dispersed MWCNTs-COOH induced the enhancement of
mechanical properties. With the increase of the MWCNTs-
COOH content, the PVDF chain wrapped with more MWCNTs-
COOH continuously enhances the mechanical properties until
0.12 wt%. In addition, the mechanical properties lost by the
membrane containing 0.15 wt% was caused by the agglomera-
tion of superfluous MWCNTs-COOH. Before the turning point,
the cross-sectional structure of membrane becomes much more
compact with the presence of increasing MWCNTs-COOH con-
tent, which induces the membrane sublayer morphology to
transform gradually from finger-like into sponge-like, accompa-
nied by the disappearance of the net-like structure resulting
from the presence of PVA. Thus, the superior mechanical prop-
erties of the modified composite membrane including break
strength, elongation at break, and Young’s modulus have
enhancements of 60%, 215.5%, and 56.7%, respectively, com-
pared to membrane MC 9010 000. Nevertheless, membrane MC
9010 009 still reveals considerable mechanical properties versus
membrane MC 9010 012.
CONCLUSIONS
PVDF/PVA/MWCNTs-COOH modified UF membranes with
various contents of MWCNTs-COOH (0.00 wt%, 0.03 wt%,
0.06 wt%, 0.09 wt%, 0.12 wt%, 0.15 wt%, weight of casting
solution) were prepared via the phase inversion method.
TABLE 4. Porosity, pore size, molecular weight cut-off (MWCO), and rejections values.
Membrane no. MC 9010 000 MC 9010 003 MC 9010 006 MC 9010 009 MC 9010 012 MC 9010 015
Porosity (%) 80.0 6 1.6 80.6 6 3.9 81.6 6 3.1 82.4 6 1.4 79.4 6 3.8 80.4 6 3.1
l (nm) 9.7 5.4 5.0 3.2 3.6 4.3
r 1.9 1.5 1.8 2.0 2.1 1.8
MWCO/Da 116,000 22,700 32,000 17,300 25,300 25,400
PEG 2k rejection (%) 15.2 6 0.9 15.7 6 1.7 24.6 6 1.1 37.5 6 1.3 36.3 6 2.5 29.6 6 1.7
PEG 4k rejection (%) 23.1 6 1.7 27.5 6 2.1 41.1 6 1.5 56.8 6 0.7 50.5 6 1.1 39.4 6 0.7
PEG 6k rejection (%) 23.6 6 0.6 43.9 6 1.3 45.1 6 0.3 65.6 6 0.9 59.0 6 1.0 53.0 6 0.6
PEG 10k rejection (%) 30.4 6 0.6 43.0 6 0.9 47.7 6 0.9 68.9 6 0.8 64.4 6 0.4 60.2 6 0.4
DEX 600k rejection (%) 66.8 6 0.6 86.2 6 0.6 85.6 6 1.0 91.0 6 0.6 86.1 6 0.7 89.7 6 0.7
TABLE 5. Mechanical properties of various membranes with different con-
tents of MWCNTs-COOH.
Membrane no.
Break
strength (MPa)
Elongation at
break (%)
Young’s
modulus (MPa)
MC 9010 000 1.30 6 0.07 9.7 6 0.8 30.7 6 2.9
MC 9010 003 1.58 6 0.03 12.8 6 0.7 34.7 6 1.5
MC 9010 006 1.78 6 0.08 14.2 6 2.3 36.5 6 3.2
MC 9010 009 1.86 6 0.09 17.3 6 2.7 48.0 6 2.0
MC 9010 012 2.08 6 0.07 30.6 6 3.9 48.1 6 4.0
MC 9010 015 2.06 6 0.12 27.9 6 8.3 45.2 6 6.7
DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—2016 965
FT-IR spectra show the absorbency of C@O at 1,730 cm21
and prove the successful introduction of carboxyl groups. The
weak peaks of the OAH band over 3,100–3,600 cm21 are
ascribed to the introduction of PVA, implying the pore-forming
agent function of PVA in the membrane forming process, which
has contributions to the hydrophilicity and high flux. SEM
images demonstrate that all composite membranes have a tradi-
tional asymmetric structure constituted by a compact skin-layer
and a variational sublayer, which could be divided into the
“upper part” finger-like structure and the “under part” sponge-
like structure. MC 9010 009 displayed a perfect porous struc-
ture, and MC 9010 012 displayed the best sponge-like structure.
AFM three-dimensional diagrams and roughness values identify
that the roughest and most hydrophilic external surface belongs
to MC 9010 009, which exhibits the best surface hydrophilicity
and permeation performances.
MC 9010 009 displays the best permeation performances and
surface structure. The pure water flux was approximately
doubled (126.6 L m22 h21) compared to MC 9010 000 (68.6
L m22 h21), which can be identified by high porosity (82.4%).
Otherwise, the rejection tests of PEGs and DEX illuminate the
smallest mean effective pore size and MWCO of 3.2 nm and
17300 Da, respectively, which can obstruct DEX 600k up to
91.0%. The BSA flux recovery rate and fouling resistance tests
clarify the remarkable effect of the introduction of PVA and
MWCNTs-COOH. The FRR% of MC 9010 009 increased by
17% compared to MC 9010 000 and exhibited the highest
reversible fouling ratio and the lowest total fouling ratio. In
addition, the dynamic contact angle showed a decreasing trend,
with that of MC 9010 009 being the most obvious.
MC 9010 012 exhibits impressive mechanical properties of
which break strength, elongation at break, and Young’s modulus
have improved by 60%, 215.5%, and 56.7%, respectively, com-
pared to MC 9010 000. These properties benefited from the denser
structure formed by the addition of PVA and MWCNTs-COOH.
REFERENCES
1. J. Yin and B.L. Deng, J. Membr. Sci., 479, 256 (2015).
2. J.G. Gai, X.L. Gong, X. Zhang, W.L. Kang, and W.W. Wang,
Polym. Eng. Sci., 55, 466 (2015).
3. A. Marjani, M. Mohammadi, R. Pelalak, and S. Moradi, Polym.
Eng. Sci., 54, 961 (2014).
4. Y.J. Fu, J.T. Chen, C.C. Chen, K.S. Liao, C.C. Hu, K.R. Lee,
and J.Y. Lai, Polym. Eng. Sci., 53, 1623 (2013).
5. N.A.M. Nazri, W.J. Lau, A.F. Ismail, T. Matsurra, D.
Veerasamy, and N. Hilal, Desalination, 358, 49 (2015).
6. S.Y. Park, J.W. Chung, and S.Y. Kwak, J. Membr. Sci., 491, 1
(2015).
7. M.L. M�endez, A.I. Romero, V.B. Rajal, E.F. Castro, J.I. Calvo,
L. Palacio, and A. Hern�andez, Polym. Eng. Sci., 54, 1211
(2014).
8. S.M. Mousavi, E. Saljoughi, Z. Ghasemipour, and S.A.
Hosseini, Polym. Eng. Sci., 52, 2196 (2012).
9. J.H. Li, X.S. Shao, Q. Zhou, M.Z. Li, and Q.Q. Zhang, Appl.Surf. Sci., 265, 663 (2013).
10. L. Yan and J. Wang, Desalination, 281, 455 (2011).
11. Y.J. Wu, Q.L. Huang, C.F. Xiao, K.K. Chen, X.F. Li, and N.N.
Li, Desalination, 353, 118 (2014).
12. M.H. Razzaghi, A. Safekordi, M. Tavakolmoghadam, F.
Rekabdar, and M. Hemmati, J. Membr. Sci., 470, 547 (2014).
13. G.D. Kang and Y.M. Cao, J. Membr. Sci., 463, 145 (2014).
14. C.C. Yang and G.M. Wu, Mater. Chem. Phys., 114, 948
(2009).
15. M. Li, X. Xue, D. Wang, Y. Lu, Z.H. Wu, and H.Z. Zou,
Desalination, 329, 50 (2013).
16. H.L. Lin and S.H. Wang, J. Membr. Sci., 452, 253 (2014).
17. R.A. Azmi, P.S. Goh, A.F. Ismail, W.J. Lau, B.C. Ng, N.H.
Othman, A.M. Noor, and M.S.A. Yusoff, J. Food Eng., 166,
165 (2015).
18. N.X. Wang, S.L. Ji, J. Li, R. Zhang, and G.J. Zhang, J. Membr.Sci., 455, 113 (2014).
19. F.B. Peng, F.S. Pan, H.L. Sun, L.Y. Lu, and Z.Y. Jiang, J.Membr. Sci., 300, 13 (2007).
20. A. Mahmood, S. Bano, S.G. Kum, and K.H. Lee, J. Membr.Sci., 415-416, 360 (2012).
21. X.L. Zhang, C.F. Xiao, X.Y. Hu, X.F. Li, and R. Wang, Polym.Eng. Sci., 54, 276 (2014).
22. J. Zhang, Q.Y. Wang, Z.W. Wang, C.W. Zhu, and Z.C. Wu, J.Membr. Sci., 466, 293 (2014).
23. X.P. Li, Y.B. Chen, X.Y. Hu, Y.F. Zhang, and L.J. Hu, J.Membr. Sci., 471, 118 (2014).
24. N.N. Li, C.F. Xiao, S.L. An, and X.Y. Hu, Desalination, 250,
530 (2010).
25. D. Mattia, H. Leese, and K.P. Lee, J. Membr. Sci., 475, 266
(2015).
26. R. Das, S.B. Abd Hamid, M.E. Ali, A.F. Ismail,
M.S.M. Annuar, and S. Ramakrishna, Desalination, 354, 160
(2014).
27. C.Q. Zhao, X.C. Xu, J. Chen, and F.L. Yang, Desalination,
334, 17 (2014).
28. H.Q. Wu, B.B. Tang, and P.Y. Wu, J. Membr. Sci., 451, 94
(2014).
29. X. Zhang, W.Z. Lang, H.P. Xu, X. Yan, Y.J. Guo, and L.F.
Chu, J. Membr. Sci., 469, 458 (2014).
30. B. Zhao, J. Wang, Z.J. Li, P. Liu, D. Chen, and Y.F. Zhang,
Mater. Lett., 62, 4380 (2008).
31. M. Kumar and M. Ulbricht, J. Membr. Sci., 448, 62 (2013).
32. T. Wang, J.N. Shen, L.G. Wu, and B. van der Bruggen, J.Membr. Sci., 466, 338 (2014).
33. Y.F. Zhao, Z.W. Xu, M.J. Shan, C.Y. Min, B.M. Zhou, Y.L.
Li, B.D. Li, L.S. Liu, and X.M. Qian, Sep. Sci. Technol., 103,
78 (2013).
34. C.Q. Zhao, X.C. Xu, J. Chen, and F.L. Yang, J. Environ.Chem. Eng., 1, 349 (2013).
35. Q.Y. Zhang and C.D. Vecitis, J. Membr. Sci., 459, 143 (2014).
36. Z.H. Wang, H.R. Yu, J.F. Xia, F.F. Zhang, F. Li, Y.Z. Xia, and
Y.H. Li, Desalination, 299, 50 (2012).
37. T.L.S. Silva, S. Morales-Torres, J.L. Figueiredo, and A.M.T.
Silva, Desalination, 357, 233 (2015).
38. P.Y. Zhang, Z.L. Xu, X.H. Ma, H. Yang, W.Z. Wu, Y.M. Wei,
and Y.D. Liu, J. Membr. Sci., 486, 119 (2015).
39. F.C. Chiu and S.C. Yeh, Polym. Test., 486, 119 (2015).
40. C. Chen, L. Tang, B.C. Liu, X. Zhang, J. Crittenden, K.L.
Chen, and Y.S. Chen, J. Membr. Sci., 487, 1 (2015).
41. S. Yesil and G. Bayram, Polym. Eng. Sci., 51, 1286 (2011).
966 POLYMER ENGINEERING AND SCIENCE—2016 DOI 10.1002/pen
42. T.H. Nama, K. Goto, Y. Yamaguchi, E.V.A. Premalal, Y.
Shimamura, Y. Inoue, K. Naito, and S. Ogihara, Compos. A,
76, 289 (2015).
43. Q. Yang, T.S. Chung, and Y.E. Santoso, J. Membr. Sci., 290,
153 (2007).
44. Y.F. Zhao, L.P. Zhu, Z. Yi, B.K. Zhu, and Y.Y. Xu, J. Membr.Sci., 440, 40 (2013).
45. S.S. Yuan, J. Wang, X.J. Wang, S.R. Long, G. Zhang, and J.
Yang, Polym. Eng. Sci., 55, 2829 (2015).
46. Y.J. Tang, L.J. Wang, Z.L. Xu, Y.M. Wei, and H. Yang, J.Membr. Sci., 502, 106 (2016).
47. X. Shen, X.B. Yin, Y.P. Zhao, and L. Chen, Polym. Eng. Sci.,55, 1367 (2015).
48. P. Carretero, S. Molina, A. Lozanoa, J. Abajo, J.I. Calvo, P.
Pr�adanos, L. Palacio, and A. Hern�andez, Desalination, 327, 14
(2013).
49. R.I. Peinador, J.I. Calvo, P. Pr�adanos, L. Palacio, and A.
Hern�andez, J. Membr. Sci., 348, 238 (2010).
50. J.I. Calvo, R.I. Peinador, P. Pr�adanos, L. Palacio, A. Bottino, G.
Capannelli, and A. Hern�andez, Desalination, 268, 174 (2011).
51. X. Shen, Y.P. Zhao, L. Chen, X. Feng, D. Yang, Q. Zhang, and
D. Su, Polym. Eng. Sci., 53, 571 (2013).
52. P.Y. Zhang, Z.L. Xu, H. Yang, Y.M. Wei, W.Z. Wu, and D.G.
Chen, Desalination, 319, 47 (2013).
53. L.B. Zhao, M. Liu, Z.L. Xu, Y.M. Wei, and M.X. Xu, Chem.Eng. Sci., 137, 131 (2015).
DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—2016 967
本文献由“学霸图书馆-文献云下载”收集自网络,仅供学习交流使用。
学霸图书馆(www.xuebalib.com)是一个“整合众多图书馆数据库资源,
提供一站式文献检索和下载服务”的24 小时在线不限IP
图书馆。
图书馆致力于便利、促进学习与科研,提供最强文献下载服务。
图书馆导航:
图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具