You are on page 1of 8

Journal of Membrane Science 376 (2011) 225232

Contents lists available at ScienceDirect

Journal of Membrane Science


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

Alkali doped polyvinyl alcohol/multi-walled carbon nano-tube electrolyte for


direct methanol alkaline fuel cell
Wen-Han Pan a , S. Jessie Lue a,c, , Chia-Ming Chang b,c , Ying-Ling Liu b,c,
a
b
c

Department of Chemical and Materials Engineering, Chang Gung University, Kwei-shan, Taoyuan 333, Taiwan
Department of Chemical Engineering, Chung Yuan University, Chungli, Taoyuan 320, Taiwan
R&D Center for Membrane Technology, Chung Yuan University, Chungli, Taoyuan 320, Taiwan

a r t i c l e

i n f o

Article history:
Received 31 January 2011
Received in revised form 13 April 2011
Accepted 15 April 2011
Available online 22 April 2011
Keywords:
Direct methanol alkaline fuel cell (DMAFC)
Cell performance
Methanol permeability
Conductivity
Nano-composite

a b s t r a c t
A novel route to functionalize polyvinyl alcohol (PVA) onto multi-walled carbon nano-tube (MWCNT)
is reported in this work. FTIR, XPS, Raman spectroscopy, and TGA data conrmed PVA grafting onto
the MWCNT. The grafted PVA content was estimated to be 25% in the PVAfunctionalized MWCNT. A
nano-composite consisting of PVA and 0.05% functionalized CNT was successfully prepared using a solution casting method. Water solubility and diffusivity were enhanced in the CNT-containing membranes.
The ionic conductivity of the potassium hydroxide (KOH)-doped PVA/CNT membrane was improved by
adding the functionalized CNT, which might be ascribed to the ionic channels provided by the CNT. The
methanol permeability was suppressed in the CNT-containing sample. The alkali-doped electrolytes were
applied in direct methanol alkaline fuel cells. An open-circuit potential and a peak power density of 0.86 V
and 39 mW cm2 were obtained using a 2 M methanol fuel in 6 M KOH at 60 C with the PVA/CNT/KOH
electrolyte, signicantly higher than those without CNT incorporation.
2011 Elsevier B.V. All rights reserved.

1. Introduction
Fuel cells have attracted much attraction as an alternative power
supply in many applications. Direct methanol fuel cells (DMFC) are
suitable for portable electronic devices due to many advantages,
including high energy density, compact design, and room temperature start-up. Most DMFCs use a proton-exchange membrane, such
as Naon from Dupont [1,2], as the membrane electrolyte. This
peruorosulfonic acid membrane suffers from methanol cross-over
[35], resulting in detrimental effects: cathode catalyst poisoning and reduced cell voltage due to the mixed over-potential of
the unwanted methanol oxidation due to methanol transfer from
anode to cathode. Researchers have recently paid more attention
to developing hydroxide conducting polymer electrolytes for direct
methanol alkaline fuel cells (DMAFCs) [610].
In a DMAFC operation, hydroxide ions are generated on the
cathode and consumed on the anode [11]. The hydroxide ions are
transported from the cathode to anode and oppose the methanol

Corresponding author at: Department of Chemical and Materials Engineering,


Chang Gung University, 259 Wen-hwa First Road, Kwei-shan, Taoyuan 333, Taiwan.
Tel.: +886 3 2118800x5489; fax: +886 3 2118700.
Corresponding author at: Department of Chemical Engineering, Chung Yuan
University, Chungli, Taoyuan 320, Taiwan. Tel.: +886 3 2654130;
fax: +886 3 2654199.
E-mail addresses: jessie@mail.cgu.edu.tw (S.J. Lue), ylliu@cycu.edu.tw (Y.-L. Liu).
0376-7388/$ see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.memsci.2011.04.026

diffusion direction. The methanol oxidation rate is more favorable


in an alkaline solution than in an acidic one [1113]. Less expensive
cathode catalysts are available to reduce the cell cost from using
precious metals [1416]. These advantages are the main motivation in the development of hydroxide-conducting membranes.
Although one might be concerned that carbonate salt might have
formed on the anode of a DMAFC and this weak acid might have
reduced the alkalinity of the methanol/KOH solution, we found that
the K2 CO3 formation during 100-h of continuous operation in an
alkaline direct methanol fuel cell with recycling anode feed did not
affect the cell performance because the amount of produced carbonate was negligible and the potassium salt was soluble in the
aqueous solution [9].
Although anion-exchange membranes have been adopted in
DMAFC application, the cell performance is not as high as those
with alkali-doped electrolytes [610]. Polyvinyl alcohol (PVA) is
often used as a base material for alkali doping, owing to its inexpensiveness, hydrophilicity, and good lm forming property. The
abundant hydroxyl groups provide good compatibility and uptake
substantial amounts of alkali aqueous solution. Many modication
methods, including cross-linking [1719], copolymerization [20],
and the addition of inorganic llers [9,2123] are used to fabricate
PVA with enhanced mechanical strength. In addition, we found the
incorporation of nano-fumed silica particles into the PVA matrix
enhanced the stability in water due to the physical cross-linkage
mechanism, which suppressed the polymer crystal unfolding and
the membrane dissolution in water [24]. The ionic conductivities of

226

W.-H. Pan et al. / Journal of Membrane Science 376 (2011) 225232

the alkali-doped PVA/ller composites were enhanced as compared


with the alkali-doped PVA electrolyte [9,10] due to the enlarged
free volume size present in the polymer matrix caused by the
nano-ller [25]. This enlarged free volume size was big enough
for water molecules and hydroxide ion to permeate but hindered
larger methanol transport, thus reducing the methanol cross-over
rate [9,10].
In the last few years, carbon nano-tubes (CNT) have received
much attention due to their excellent mechanical, electrical,
thermal, and magnetic properties [26]. Some researchers found
advantages of polymer and CNT composites. Thomassin et al. [27]
reported benecial effects by dispersing 12% multiwalled carbon nano-tubes (MWCNT) into Naon membranes. The Youngs
modulus of the composite membranes was increased while the
methanol permeability was decreased without adverse impact on
the ionic conductivity. Joo et al. [28] found that the mechanical properties were improved by adding functionalized CNT into
sulfonated poly(arylene sulfone) (sPAS) matrix. The DMFC performance using the CNT-containing sPAS was signicantly improved
compared with the neat sPAS lm, owing to the former higher
ionic conductivity and decreased methanol permeability. Liu et al.
[29] demonstrated that blending Naonfunctionalized MWCNT
could improve the proton-exchange hydrogen fuel cell power density by 50% compared with pristine Naon. Recently Liu et al. [30]
reported that the incorporation of Naon and polybenzimidazole (PBI)functionalized MWCNT into PBI matrix also enhanced
the hydrogen fuel cell performance. The peak power density was
increased by 1332% as compared with the electrolyte without the
functionalized CNT addition [30].
In the present research, we investigate the DMAFC performance
using potassium hydroxide (KOH)-doped PVA/MWCNT electrolyte.
The CNT surface modication and the PVA/CNT composite preparation are reported. The enhanced cell voltage and power density
in fuel cells using the PVA/CNT/KOH electrolyte are correlated to
the membrane characteristics.
2. Experimental
2.1. Materials
Multi-walled carbon nano-tubes (MWCNT), with average diameters of 1050 nm and lengths of 125 m were received from
the Carbon Nano-tube Co., Ltd., Incheon, Korea. The purity of the
received MWCNT is 93%. MWCNT was washed with dimethylsulfoxide prior to use. Polyvinyl alcohol (PVA, average molecular
weight of 89,00098,000, more than 99% hydrolyzed), and potassium hydroxide (KOH) were obtained from SigmaAldrich. Methyl
alcohol (HPLC grade, 99.9%) was from Acros Organics, Geel,
Belgium. Gas diffusion electrodes with 5 mg cm2 PtRu alloy (1:1)
black for the anode and 5 mg cm2 Pt black for the cathode were
purchased from E-tek, Somerset, NJ, USA. Pure water with resistivity of 18 M cm was produced using a Millipore water purier (Elix
5/Milli-Q Gradient system, Millipore Corp., Bedford, MA).
2.2. Preparation of PVAfunctionalized MWCNTs
PVAfunctionalized MWCNT was prepared using an ozonemediated method as reported previously [31,32]. One gram of PVA
was dissolved in 30 mL water at 70 C, placed in a 100 mL onenecked ask with a stirrer. The PVA solution was purged with
ozone gas for 15 min. The ozone gas ow rate was 6 L min1 and the
ozone concentration was 28 g m3 . The solution was then purged
with argon gas for 15 min to remove the free peroxide groups. Half
a gram of CNT was added in the PVA solution quickly. The mixture was stirred at 80 C for 3 h to graft the PVA polymer onto the

MWCNT. After the reaction was completed, the functionalized CNT


and the solution were separated using a centrifuge and washed
with hot water until the solution was clear. The PVAfunctionalized
CNT was collected using ltration and then dried over night. The
MWCNT and PVA reaction scheme is shown in Fig. 1.

2.3. PVA/CNT composite preparation


The PVAfunctionalized CNT (0.0075 mg) was suspended in
10 mL of deionized (DI) water under ultrasonication at room temperature. Fifteen grams of PVA were dissolved in 50 mL of DI water
at room temperature to form a polymer solution. The CNT suspension was mixed into the PVA solution. Another 75 mL of DI
water was used to rinse the beaker wall and remove the remaining
CNT. The rinse solution was then combined with the CNT and PVA
solution. The solution (total of 135 mL) was heated to 90 C under
ultrasonication and stirring (250 rpm) for 6 h to obtain a homogeneous solution. The solution was cast on a glass with an application
knife (model 3580, Elcometer Instrument Ltd., Edge Lane, England).
A uniform thickness lm was obtained after being dried in vacuum
at 60 C for 6 h. The resulting PVA composites consisted of 0.05%
(weight basis) of PVAfunctionalized CNT, unless stated otherwise.
The thickness of the dried composite membrane was 130180 m.
The PVA and PVA/CNT composites were immersed in KOH solution
of various concentrations for at least 24 h. These KOH-doped PVA
and PVA composite were referred as PVA/KOH and PVA/CNT/KOH.

2.4. Characterization
Fourier transform infrared (FTIR) spectra of the functionalized CNT, PVA and PVA/CNT composite membrane were obtained
through the attenuated total reectance method using an FTIR
(Perkin Elmer Spectrum One, Perkin Elmer Corp., Norwalk, CT,
USA) equipped with a multiple internal reectance apparatus and a
ZnSe prism as an internal reection element. Raman spectra of the
pristine and PVAfunctionalized CNT were obtained using a Renishaw InVia Raman spectrometer (3D Nanometer Scale Raman PL
Microspectrometer, Tokyo Instruments, Inc., Tokyo, Japan) employing a HeNe laser of 1 mW radiating on the sample operating
at 632.8 nm. Thermo-gravimetric analysis (TGA) was performed
with an instrument from the Thermal Analysis Incorporation (TATGA Q-500, TA Instrument, New Castle, DE, USA) under a nitrogen
atmosphere at a heating rate of 10 C min1 . X-ray photoelectron
spectroscopy (XPS) analysis was conducted with a VG Microtech
MT-500 ESCA (Thermo Fisher Scientic Inc, Walthan, MA, USA)
using an MgK line as the radiation source. High-resolution transmission electron microscopy (HRTEM) was conducted with a
JEOL JEM-2010 HR-TEM (JEOL Ltd, Tokyo, Japan). The membrane
micro-structure was observed using eld emission scanning electron microscope (FESEM, model S-4800, Hitachi High-Technologies
Corp., Tokyo, Japan) after being freeze-fractured in liquid nitrogen and then sputtered with Pt. An X-ray diffraction (XRD, model
D5005D, Siemens AG, Munich, Germany) measurement was performed on PVA and PVA/CNT composite membranes to examine
their crystallinity characteristics. The X-ray radiation was gener from an anode operating
ated using Cu K (wavelength 1.54056 A)
at 40 kV and 40 mA. The scanning rate was 0.5 s1 with a 0.02
resolution. The XRD was recorded over the angles 1550 .

2.5. Water uptake and diffusivity


The membrane water uptake was determined by measuring the
difference between the dry weight (Wo , in g) and total weight (Wtt ,

W.-H. Pan et al. / Journal of Membrane Science 376 (2011) 225232

227

Fig. 1. Reaction mechanism of PVA grafting onto MWCNT.

in g) after immersion in 25 C DI water. The solvent uptake (M) was


calculated from the following equation:
M=

Wtt Wo
Wo

(1)

The water diffusion coefcient was calculated on the transient


regime using the limiting slope method [33]. The diffusion coefcient was tted for a relative uptake data set as a function of
time:

Mt
ln 1
M

ln

8
2

2 Dt
=
2

ln
(2)

where Mt and M are the solvent uptake (in g g1 ) at time t and at


equilibrium, respectively, D is the diffusion coefcient and is the
membrane thickness.
2.6. Methanol permeability measurement
Methanol permeability measurements for 1 M methanol at 30 C
were carried out on a PVA membrane and PVA/CNT composite.
A side-by-side diffusion cell consisting of two-compartment glass
reservoir (source and receiving reservoirs) was used to test the permeability. The source reservoir was lled with a methanol aqueous
solution of volume VA and the receiving reservoir was lled with
DI water of volume VB . These two reservoirs were separated by
the membrane under test. The permeation cell was maintained at
30 C. The methanol concentration transported through the membranes was determined by sampling a small amount of the solution
from the receiving compartment at time intervals for concentration measurement using a gas chromatograph (HP 4890A, Agilent
Technologies Co. Ltd., St. Louis, MO, USA). The analytical column
was a fused silica capillary column (30 m 0.32 mm, Supel-QTM
PLOT, Supelco, St. Louis, MO, USA). The methanol permeability was
determined from the methanol concentrations at various elapsed
times using the equation described in our previous paper [34]. The
following two equations are for the methanol mass conservation:
Flux =

VA dCA
VB dCB
=
A dt
A dt

(3)

where A is the effective membrane area, t is the time elapsed,


and CA and CB are the methanol concentrations in the source and
receiving reservoirs, respectively. In this experiment, VA equaled
VB for calculation simplicity. Taking mass balance on methanol by
assuming insignicant volume changes in both reservoirs during
the permeation experiment and negligible methanol sorbed inside
the membrane yields Eq. (4).
CA,o = CA + CB

(4)

where CA,o is the initial methanol concentration placed in the source


reservoir.
In addition, Ficks law holds true for a methanol diffusion process
through the membrane:
Flux = D

C m CBm
K(CA CB )
dC m
=D A
=D
L
L
dx

where D is methanol diffusion coefcient, Cm is methanol


concentration inside the membrane, x is the distance along transmembrane direction, L the membrane thickness, K the partition
constant relating methanol concentration inside membrane to that
in aqueous solution, and subscripts A and B represent the interface
at the donor and receiving reservoirs, respectively.
Combining Eqs. (3)(5) by eliminating CA and integrating both
sides of Eq. (3) result in the following equation:

(5)

CA,o
2ADK
(t to )
=
CA,o 2CB
LV

(6)

where permeability P equals DK and can be obtained as a slope by


plotting ln(CA,o /(CA,o 2CB )) vs. 2At/LV.
2.7. Ionic conductivity measurements
The PVA and PVA/CNT membranes were immersed in a 2 or 6 M
KOH solution for at least 24 h before conductivity measurement.
A potentiostat (Autolab, PGSTAT-30, Eco Chemie B.V., Utrecht,
Netherlands) was used to measure the alternative current (AC)
impedance of the KOH-doped electrolytes. The electrolyte was
sandwiched between two stainless steel electrodes, each with a
surface area of 1.33 cm2 , in a spring-loaded glass holder [19]. This
apparatus was maintained in a chamber with controlled relative
humidity and temperature. The relative humidity of the chamber
was kept at about 99% and the testing temperature ranged from
30 C to 60 C. The sample was scanned from 100 kHz to 100 Hz with
oscillating amplitude of 5 mV. The bulk resistance was determined
from the Nyquist plot of the KOH-doped electrolytes according to
the procedure described in the literature [9,35]. The conductivity
(, in S cm1 ) of the KOH-doped electrolyte was calculated using
the following equation:
=

l
Rb A

(7)

where l is the thickness of the electrolyte (cm), Rb is the bulk resistance (), and A is the contact area of the stainless steel electrodes
(cm2 ) [9].
2.8. Cell performance measurements
The PVA/KOH and PVA/CNT/KOH membranes were sandwiched
between anode and cathode gas diffusion electrodes to obtain
membrane electrode assemblies (MEA). The gas diffusion electrodes (combining catalysts on diffusion layers) were commercial
products and purchased from E-tek. The cathode catalyst was high
precious (HP) Pt black on Vulcan XC-72R carbon nano-particles.
The anode catalyst was HP Pt:Ru alloy (1:1) on Vulcan XC-72R.
The gas diffusion layer was a hydrophobicity-controlled microporous layer on carbon cloth. The catalyst loading was 5 mg cm2
for the anode and cathode. The effective area of the MEA was
5 cm2 . The current density (I) and potential (V) of the DMAFC were
recorded on an electrical load (PLZ164WA electrochemical system, Kikusui Electronics Corporation, Tokyo, Japan) at a scan rate
of 5 mA s1 . The power density was calculated as the product of

228

W.-H. Pan et al. / Journal of Membrane Science 376 (2011) 225232

Fig. 2. FTIR spectrum of PVA and PVAfunctionalized MWCNT.

cell voltage and current density. The power densitycurrent density (PI) curves were plotted to determine the maximum power
density (Pmax ).
3. Results and discussion
3.1. Characterization of PVAfunctionalized MWCNT
PVA was chemically incorporated into MWCNT bundles using
an ozone-mediated process. The obtained PVAfunctionalized
MWCNT was characterized for the functional groups present. Fig. 2
shows the FTIR spectra of the PVA and the PVAfunctionalized
MWCNT. PVA exhibited major absorption peaks of CO and OH
groups at about 1150 and 3300 cm1 , respectively. These peaks
were also observed in the FTIR spectrum of PVAfunctionalized
MWCNT. This demonstrates that the PVAfunctionalized MWCNT
possesses PVA chains and the PVA was successfully grafted
onto the MWCNT. The absorption peak at about 1605 cm1 in
the PVAfunctionalized MWCNT spectrum was attributed to the
absorption of C O groups, which were generated from the oxidation reaction in the ozone treatment. The PVAfunctionalized
MWCNT chemical structure was also further characterized with
XPS (Fig. 3). There was a signicant oxygen signal and high intensity
on the PVAfunctionalized MWCNT in the wide-scan XPS spectrum
(Fig. 3(a)). The C1s core-level spectrum of the PVAfunctionalized
MWCNT could be de-convoluted into 4 peaks of CH, O CC, CO,
and C O species at binding energy of 285.0, 285.5, 286.5, and
290.0 eV, respectively (Fig. 3(b)). The CH, O CC, CO, and C O
bonds originated from the grafted PVA. The C O groups could
arise from the un-hydrolyzed repeating units of PVA chains and
the oxidation reaction during the ozone treatment. These results
support the successful incorporation of PVA chains into MWCNT
bundles.
The changes in the MWCNT chemical structures due to PVA
functionalization were monitored using Raman spectroscopy. As
shown in Fig. 4, the pristine MWCNT shows a tangential band
(G band) at about 1572 cm1 and a disorder band (D Band) at
around 1324 cm1 . The PVAfunctionalized MWCNT exhibits similar absorption peaks compared to pristine MWCNT, demonstrating
the presence of MWCNT structures. In the PVA functionalization
treatment, ozone-treatment on the PVA chains generated radicals
that reacted with the MWCNT. Some sp2 -hybrid carbons in the
MWCNT were converted into sp3 -hybrid in the reaction. As a result,
PVAfunctionalized of MWCNT demonstrated an increase in the
D- to G-band intensity ratios (ID /IG ) from 1.15 (for the pristine
MWCNT) to 4.41 (for the functionalized MWCNT). This phenom-

Fig. 3. XPS characterization of PVAfunctionalized MWCNT: (a) wide scan spectrum


and (b) C1s core-level spectrum.

ena has been reported in the polyurethane functionalized MWCNT


[36].
Fig. 5 shows TGA thermograms of the PVA, MWCNT and functionalized MWCNT. The pristine MWCNT did not show obvious
weight loss, whereas the weight loss of the PVAfunctionalized
MWCNT was 20 wt% at 600 C, which could be attributed to the
thermal degradation of PVA chains wrapped on MWCNT. As the

Fig. 4. Raman spectra of pristine MWCNT (blue line) and PVAfunctionalized


MWCNT (red line). (For interpretation of the references to color in this gure legend,
the reader is referred to the web version of this article.)

W.-H. Pan et al. / Journal of Membrane Science 376 (2011) 225232

229

Table 1
Water uptakes, diffusion coefcients, and methanol permeabilities in PVA and
PVA/CNT composites at 25 C.
Composite

Water uptake
(g g1 )

Water diffusion
coefcient
(107 cm2 s1 )

Methanol
permeability
(107 cm2 s1 )

PVA
PVA/CNT

2.81
2.94

2.14
3.09

3.57
2.99

3.2. Characterization of the PVA/CNT composite

Fig. 5. TGA analysis in nitrogen of PVA (green line), pristine MWCNT (blue line) and
PVAfunctionalized MWCNT (red line). (For interpretation of the references to color
in this gure legend, the reader is referred to the web version of this article.)

pure PVA showed a residue of about 20 wt% in the TGA analysis, the PVA weight fraction of the PVAfunctionalized MWCNT
could be calculated to be about 25 wt%. It is noteworthy that
the weight loss of the functionalized MWCNT occurred at higher
temperatures than those for the pristine PVA. The chemical
linkages between the PVA and the MWCNT could enhance the
thermal stability of the grafted PVA chains. Moreover, ozonization and the subsequent addition reaction of PVA chains might
result in a cross-linked structure and increase the thermal stability of the PVA chains. Fig. 6 shows the HRTEM micrographs of
PVAfunctionalized MWCNT. Compared with the pristine MWCNT,
the amorphous polymer layers covering the outer bundles of the
MWCNT were observed on the functionalized MWCNT. The polymer layer thickness was about 3 nm. The TGA and HRTEM results
further conrm the successful preparation of PVAfunctionalized
MWCNT.

FE-SEM micrographs of the PVA and PVA/CNT composites are


shown in Fig. 7. Both the PVA and the PVA/CNT surfaces were
smooth and little difference was found on their surface morphology. From the cross-sectional views of the PVA/CNT composites the
CNT agglomerate was observed occasionally (Fig. 7), which might
be due to van der Waals force between the CNT bundles.
The X-ray diffraction measurement was examined to study the
crystalline of the PVA and PVA/CNT membranes. Fig. 8 shows their
diffraction patterns at 2 between 15 and 50 patterns. It is clear
that the pure PVA shows a remarkable peak for an orthorhombic
lattice centered at a 2 of 19.9 [24], indicating its semi-crystalline
Both samples
nature. The d-spacing was calculated to be 4.45 A.
demonstrated similar crystalline diffraction patterns but the peak
intensity of the PVA/CNT membranes was slightly lower than that
for the PVA lm. It implies that the PVA/CNT crystallinity was
reduced compared with that of the PVA.
3.3. Water sorption and diffusion
Fig. 9 shows the typical sorption uptake history for PVA and
PVA/CNT membranes. The water solubility in the PVA/CNT composites was slightly higher than that in the PVA lm (2.94 g g1
vs. 2.81 g g1 ). The water diffusion coefcients were tted from
the sorption history data using Eq. (2) and the water diffusion coefcient was improved with increasing the CNT content
(3.09 107 cm2 s1 for PVA/CNT and 2.14 107 cm2 s1 for PVA,
Table 1). There are two possible reasons for this diffusivity increase.

Fig. 6. HR-TEM images of pristine MWCNT (upper) and PVAfunctionalized MWCNT (lower) at 40 k, 100 k, and 300 k magnications.

230

W.-H. Pan et al. / Journal of Membrane Science 376 (2011) 225232

Fig. 7. Field emission scanning electron micrographs of surface and cross-sectional views on PVA and PVA/CNT composite.

The polymer crystallinity may be reduced at the CNT presence [25]


and more amorphous region in the PVA favored the water diffusion. Moreover, the embedded CNT had abundant grafted PVA
chains, which may serve as a special hydrophilic channel for water
diffusion.

3.4. Methanol permeability


To reduce methanol crossover, methanol permeability
must be overcome. It was found that the methanol permeability of the PVA/CNT composite was slightly reduced
to 2.99 107 cm2 s1 as compared with the pristine PVA,
which had a methanol permeability of 3.57 107 cm2 s1
[10]. This reduction in methanol permeability may be
ascribed to the lower compatibility of CNT and methanol
and the hindered pathway under the presence of CNT
[28].

Fig. 8. X-ray diffraction patterns of PVA and PVA/CNT membranes.

3.5. Ionic conductivity of KOH-doped PVA and composite


The resistance data of the KOH-doped PVA and PVA/CNT composite at various temperatures were measured using the AC
impedance analyzer. Fig. 10 shows the Nyquist plot for PVA and
PVA/CNT electrolytes with 2 M KOH at 30 C. From the x-intercepts
of the curves in Fig. 10, it was clear that PVA/CNT exhibited a lower
resistance than PVA, which might be caused by ionic channels
provided by the incorporated functionalized CNT [37]. The functionalized MWCNT bundles were covered by a layer of PVA (Fig. 6)
and full of hydroxyl groups on the CNT surface. These hydroxyl
groups not only absorbed water molecules (as shown in the water
uptake in Table 1), but also took up KOH species. The hydroxide ions
could transfer through these hydroxyl groups and the conductivity
was enhanced in the PVA/CNT/KOH electrolyte. Table 2 summarizes
the conductivity data for PVA and PVA/CNT electrolyte doped with
2 M and 6 M KOH at 3060 C. At an elevated temperature the conductivity increased due to higher ionic mobility. The PVA/CNT/KOH
had higher conductivity than PVA/KOH. The 6 M alkali concentra-

Fig. 9. Water sorption history for PVA and PVA/CNT composite membranes.

W.-H. Pan et al. / Journal of Membrane Science 376 (2011) 225232

Fig. 10. Nyquist plots of PVA and PVA/CNT electrolytes doped with 2 M KOH at
30 C. The small insert shows the high frequency range data, which are extracted to
obtained resistance and conductivity values.

231

Fig. 12. DMAFC voltage (left axis) and power density (right axis) as a function of current density at 30 and 60 C using PVA/CNT/KOH electrolyte (anode: 2 M methanol
in 6 M KOH with a ow rate of 5 mL min1 , cathode: humidied oxygen with a ow
rate of 100 mL min1 ).

Table 2
Ionic conductivity (in 102 S cm1 ) of PVA and PVA/CNT doped with 2 M and 6 M
KOH.
Alkali concentration

2 M KOH
2 M KOH
6 M KOH
6 M KOH

Membrane

PVA
PVA/CNT
PVA
PVA/CNT

Temperature ( C)
30

40

50

60

5.25
6.97
5.64
7.13

7.02
7.91
7.28
8.38

8.43
8.61
9.01
9.97

10.26
10.24
10.88
11.76

tion also enhanced the electrolyte conductivity than 2 M. This is


similar to the results for PVA/fumed silica KOH-doped composites
[9].
3.6. DMAFC performance
Fig. 11 shows the polarization curves for the DMAFC using the
PVA/KOH and PVA/CNT/KOH electrolytes with 2 M methanol in
6 M KOH at 30 C and 60 C. The PVA/CNT/KOH electrolyte outperformed the PVA/KOH electrolyte at both temperatures. The
activation over-potential drop was signicantly improved in the
low current density region for the CNT-containing sample. In the
medium current density region, the ohmic over-potential was more

severe in the cell using PVA/KOH electrolyte. This can be correlated to the lower conductivity in the PVA/KOH than that for the
PVA/CNT/KOH (Table 2).
The higher temperature resulted in a higher cell voltage at the
same current density level. This is caused by the higher catalytic
reaction kinetics at both electrodes and the higher conductivity.
Fig. 12 shows the power density values of DMAFC using PVA/KOH
and PVA/CNT/KOH electrolytes at 30 and 60 C. The peak power
density reached 39 mW cm2 with PVA/CNT/KOH electrolyte at
60 C. This value is signicantly higher than most literature data. We
reported on a similar peak power density employing PVA/FS/KOH
as the electrolyte. However, that FS load was 20%, much higher than
the CNT content of 0.05% used in this study. The addition of a very
small amount of the functionalized CNT can signicantly improve
the ionic conductivity and the fuel cell performance.

4. Conclusion
A novel route to functionalize PVA onto the multiwall CNT
is reported in this work. FTIR, XPS, Raman spectroscopy, and
TGA data conrmed the polymer grafting onto the MWCNT.
The grafted PVA content was estimated to be 25% in the
PVAfunctionalized MWCNT. A PVA composite consisting of 0.05%
functionalized MWCNT was successfully prepared using a solutioncasting method. Water solubility and diffusivity were enhanced in
the CNT-containing membranes as compared with the pristine PVA.
The ionic conductivity of the KOH-doped membrane was improved
by adding the functionalized CNT, which might be ascribed to the
ionic channels provided by the CNT. The methanol permeability
was suppressed in the CNT-containing sample. The alkali-doped
electrolytes were applied in direct methanol alkaline fuel cells.
An open-circuit potential and a peak power density of 0.86 V and
39 mW cm2 were obtained using a 2 M methanol fuel in 6 M KOH at
60 C with the PVA/CNT/KOH electrolyte, signicantly higher than
those without CNT incorporation.

Acknowledgements
Fig. 11. Effect of CNT addition in PVA on DMAFC performance: voltage (left axis)
and power density (right axis) as a function of current density at 30 C (anode: 2 M
methanol in 6 M KOH with a ow rate of 5 mL min1 , cathode: humidied oxygen
with a ow rate of 100 mL min1 ).

We thank the National Science Council of Taiwan for its nancial support (NSC 98-2221-E-182-034). Valuable inputs from the
referees are greatly acknowledged.

232

W.-H. Pan et al. / Journal of Membrane Science 376 (2011) 225232

References
[1] C. Xu, A. Faghri, X.L. Li, Development of a high performance passive vapor-feed
DMFC fed with neat methanol, J. Electrochem. Soc. 157 (2010) B1109B1117.
[2] B.C.H. Steele, A. Heinzel, Materials for fuel-cell technologies, Nature 414 (2001)
345352.
[3] X. Wei, M.Z. Yates, Control of Naon/poly(vinylidene uoride-cohexauoropropylene) composite membrane microstructure to improve
performance in direct methanol fuel cells, J. Electrochem. Soc. 157 (2010)
B522B528.
[4] W. Li, A. Manthiram, M.D. Guiver, B. Liu, High performance direct methanol
fuel cells based on acidbase blend membranes containing benzotriazole, Electrochem. Commun. 12 (2010) 607610.
[5] S.H. Seo, C.S. Lee, A study on the overall efciency of direct methanol fuel cell
by methanol crossover current, Appl. Energy 87 (2010) 25972604.
[6] C.C. Yang, Synthesis and characterization of the cross-linked PVA/TiO2 composite polymer membrane for alkaline DMFC, J. Membr. Sci. 288 (2007) 5160.
[7] C.C. Yang, C.T. Lin, S.J. Chiu, Preparation of the PVA/HAP composite polymer
membrane for alkaline DMFC application, Desalination 233 (2008) 137146.
[8] C.C. Yang, S.J. Chiu, W.C. Chien, S.S. Chiu, Quaternized poly(vinyl alcohol)/alumina composite polymer membranes for alkaline direct methanol fuel
cells, J. Power Sources 195 (2010) 22122219.
[9] S.J. Lue, W.-T. Wang, K.P.O. Mahesh, C.C. Yang, Enhanced performance of a direct
methanol alkaline fuel cell (DMAFC) using a polyvinyl alcohol/fumed silica/KOH
electrolyte, J. Power Sources 195 (2010) 79917999.
[10] S.J. Lue, K.P.O. Mahesh, W.T. Wang, J.Y. Chen, C.C. Yang, Permeant transport properties and cell performance of potassium hydroxide doped
poly(vinyl alcohol)/fumed silica nano-composites, J. Membr. Sci. 367 (2011)
256264.
[11] A.V. Tripkovic, K.D. Popovic, B.N. Grgur, B. Blizanac, P.N. Ross, N.M. Markovic,
Methanol electro-oxidation on supported Pt and PtRu catalysts in acid and
alkaline solutions, Electrochem. Acta 47 (2002) 37073714.
[12] A.V. Tripkovic, K.D. Popovic, J.D. Lovic, V.M. Jovanovic, A. Kowal, Methanol
oxidation at platinum electrodes in alkaline solution: comparison between
supported catalysts and model systems, J. Electroanal. Chem. 572 (2004)
119128.
[13] J. Prabhuram, R. Manoharan, Investigation of methanol oxidation on unsupported platinum electrodes in strong alkali and strong acid, J. Power Sources
74 (1998) 5461.
[14] J. Kim, T. Momma, T. Osaka, Cell performance of PdSn catalyst in passive direct
methanol alkaline fuel cell using anion exchange membrane, J. Power Sources
189 (2009) 9991002.
[15] H. Bunazawa, Y. Yamazaki, Ultrasonic synthesis and evaluation of nonplatinum catalysts for alkaline direct methanol fuel cells, J. Power Sources 190
(2009) 210215.
[16] E. Antolini, E.R. Gonzalez, Alkaline direct alcohol fuel cells, J. Power Sources 195
(2010) 34313450.
[17] J. Fu, J. Qiao, X. Wang, J. Ma, T. Okada, Alkali doped poly(vinyl alcohol) for
potential fuel cell applications, Synth. Met. 160 (2010) 193199.
[18] D.S. Kim, H.B. Park, J.W. Rhim, Y. Moo Lee, Preparation and characterization of
crosslinked PVA/SiO2 hybrid membranes containing sulfonic acid groups for
direct methanol fuel cell applications, J. Membr. Sci. 240 (2004) 3748.
[19] A.K. Sahu, G. Selvarani, S. Pitchumani, P. Sridhar, A.K. Shukla, N. Narayanan, A.
Banerjee, N. Chandrakumar, PVA-PSSA membrane with interpenetrating networks and its methanol crossover mitigating effect in DMFCs, J. Electrochem.
Soc. 155 (2008) B686B695.

[20] S.C. Jana, S. Maiti, S. Biswas, Graft copolymerization of acrylonitrile onto


poly(vinyl alcohol) in presence of air using ceric ammonium nitrate-natural
gums, J. Appl. Polym. Sci. 78 (2000) 15861590.
[21] S.D. Bhat, A.K. Sahu, A. Jalajakshi, S. Pitchumani, P. Sridhar, C. George, A. Banerjee, N. Chandrakumar, A.K. Shukla, PVASSAHPA mixed-matrix-membrane
electrolytes for DMFCs, J. Electrochem. Soc. 157 (2010) B1403B1412.
[22] C.C. Yang, Y.J. Lee, J.M. Yang, Direct methanol fuel cell (DMFC) based on
PVA/MMT composite polymer membranes, J. Power Sources 188 (2009) 3037.
[23] Y. Zhang, Z. Cui, C. Liu, W. Xing, J. Zhang, Implantation of Naon ionomer
into polyvinyl alcohol/chitosan composites to form novel proton-conducting
membranes for direct methanol fuel cells, J. Power Sources 194 (2009) 730736.
[24] S.J. Lue, J.Y. Chen, J.M. Yang, Crystallinity and stability of poly(vinyl alcohol)fumed silica mixed matrix membranes, J. Macromol. Sci. Part B: Phys. 47 (2008)
3951.
[25] S.J. Lue, D.T. Lee, J.Y. Chen, C.H. Chiu, C.C. Hu, Y.C. Jean, J.Y. Lai, Diffusivity enhancement of water vapor in poly(vinyl alcohol)-fumed silica
nano-composite membranes: Correlation with polymer crystallinity and freevolume properties, J. Membr. Sci. 325 (2008) 831839.
[26] P.M. Ajayan, Nanotubes from carbon, Chem. Rev. 99 (1999) 17871799.
[27] J.-M. Thomassin, J. Kollar, G. Caldarella, A. Germain, R. Jrme, C. Detrembleur,
Benecial effect of carbon nanotubes on the performances of Naon membranes in fuel cell applications, J. Membr. Sci. 303 (2007) 252257.
[28] S.H. Joo, C. Pak, E.A. Kim, Y.H. Lee, H. Chang, D. Seung, Y.S. Choi, J.B. Park, T.K. Kim,
Functionalized carbon nanotube-poly(arylene sulfone) composite membranes
for direct methanol fuel cells with enhanced performance, J. Power Sources 180
(2008) 6370.
[29] Y.L. Liu, Y.H. Su, C.M. Chang, Suryani, D.M. Wang, J.Y. Lai, Preparation and applications of Naonfunctionalized multiwalled carbon nanotubes for proton
exchange membrane fuel cells, J. Mater. Chem. 20 (2010) 44094416.
[30] Suryani, C.-M. Chang, Y.L. Liu, Y.M. Lee, Polybenzimidazole membranes
modied with polyelectrolytefunctionalized multiwalled carbon nanotubes for proton exchange membrane fuel cells, J. Mater. Chem. (2011),
doi:10.1039/c1jm10439j.
[31] Y.L. Liu, W.H. Chen, Y.H. Chang, Preparation and properties of chitosan/carbon
nanotube nanocomposites using poly(styrene sulfonic acid)-modied CNTs,
Carbohydr. Polym. 76 (2009) 232238.
[32] C.M. Chang, Y.L. Liu, Functionalization of multi-walled carbon nanotubes with
non-reactive polymers through an ozone-mediated process for the preparation
of a wide range of high performance polymer/carbon nanotube composites,
Carbon 48 (2010) 12891297.
[33] S.J. Lue, S.F. Wang, L.D. Wang, W.W. Chen, K.M. Du, S.Y. Wu, Diffusion of multicomponent vapors in a poly(dimethylsiloxane)membrane, Desalination 233
(2008) 277285.
[34] S.J. Lue, T.S. Shih, T.C. Wei, Plasma modication on a Naon membrane for direct
methanol fuel cell applications, Korean J. Chem. Eng. 23 (2006) 441446.
[35] Y. Gao, G.P. Robertson, M.D. Guiver, X. Jian, Synthesis and characterization of
sulfonated poly(phthalazinone ether ketone) for proton exchange membrane
materials, J. Polym. Sci. Part A: Polym. Chem. 41 (2003) 497507.
[36] C. Gao, Y.Z. Jin, H. Kong, R.L.D. Whitby, S.F.A. Acquah, G.Y. Chen, H. Qian,
A. Hartschuh, S.R.P. Silva, S. Henley, P. Fearon, H.W. Kroto, D.R.M. Walton,
Polyureafunctionalized multiwalled carbon nanotubes: Synthesis, morphology, and Raman spectroscopy, J. Phys. Chem. B 109 (2005) 1192511932.
[37] Y.H. Su, Y.L. Liu, D.M. Wang, J.Y. Lai, M.D. Guiver, B. Liu, Increases in the proton conductivity and selectivity of proton exchange membranes for direct
methanol fuel cells by formation of nanocomposites having proton conducting
channels, J. Power Sources 194 (2009) 206213.

You might also like